1

f

f

f

f

i

j

/

Y

<

<

UJ

r
z

S

I

f

*

(

4

J
I

r

i

t
I

r

L
L

J

4

(

/^/

, /'

/

VOLUME 64

f

OF THE

1977

mm

m

Published by the Missouri Botanical Garden Press, St. Louis, Missouri 63110

© Missouri Botanical Garden 1978

CONTENTS

\

184

694

330

141

Adams, Robert P. Chemosystematics — Analysis of Populational Differen-
tiation and Varia])ility of Ancestral and Recent Populations of Junip-

erus ashei

AsHTON, P. S. A Contribution of Rain Forest Research to Evolutionary
Theory .

Austin, Daniel F. Realignment of the Species Placed in Exogonium

( Convolvulaceae )

AvERETT, John E. Taxonomic Notes and New Combinations in Leucophy-

salis ( Solanaceae )

AvERETT, John E, Chemosystematics: The Twenty-third Systematics Sym-
posium

Broome, C. Rose, (see Stibolt, Virginia M., C. Rose Broome & James L.

Reveal)

Carlquist, Sherwin. Wood Anatomy of Onagraceae: Additional Species

and Concepts 627

D'Arcy, William G. (see Gentry, Johnnie L., Jr. & William G. D'Arcy) 376

Davidse, Gerrit. New Taxa and Combinations in the Genus Lasiacis

145

373

( Gramineae )

374

Davidse, Gerrit. (see Goldblatt, Peter & Gerrit Davidse) 121

Dietrich, Werner. The South American Species of Oenothera Sect. Oe-

425

340

nothera {Raimannia, Renneria; Onagraceae)

Dunn, David B. & William E. Harmon. The Ltipinus montanus Complex

of Mexico and Central America

Eyde, Richard H. Reproductive Structures and Evolution in Ludwigia

Onagraceae. I. Androecium, Placentation, Merism 644

Fairbrotiiers, David E. Perspectives in Plant Serotaxonomy 147

221

138

321

Feeny, Paul. Defensive Ecology of the Cruciferae

Gentry, Alwyn H. A New Jacaramla (Bignoniaceae) from Ecuador and

Peru.

Gentry, Alwyn H. A New Species of Bignoniaceae from Madagascar .— 139

Gentry, Alwyn H. New Records of Apocynaceae for Panama and the

Choc6 „

Gentoy, Alwyn H. New Species of Gihsoniothamnus ( Scrophulariaceae/

Bienoniaceae) and Toumefortia (Boraginaccae) from Eastern Panama
and the Choco

Gentry, Alwyn IT. Studies in Bignoniaceae 25: New Species and Com-
binations in South American Bignoniaceae

Gentry, Johnnie L., Jr. & William G. D'Arcy. Solanum armentalis: A

New Species from Costa Rica

133

311

376

GoLDBLATT, Peter. Chromosome Number in PiUamiu (Iridaceae) 136

GoLDBLATT, Peter. Herhertiu (Iridaceae) Reinstated as a Valid Generic

Name

379

GoLDBLATT, Peter. Systematlcs of Moraea (Iridaceae) in Tropical Africa 243

GoLDBLATT, Peter & Gerrit Davidse. Cliromosome Numbers in Legumes 121

Harmon, William E. (see Dunn, David B. & William E. Harmon) 340

HocH, Peter H. & Peter H. Raven. New Combinations in Epilohium

( Onagraceae ) 236

Huckins, Gharles Albert. In Chromosome Numbers of Plianerogams. 7 142

Janzen, Daniel H. Promising Directions of Study in Tropical Animal-
Plant Interactions _ yng

Jordan, Carl F. & Ernesto Medina. Ecosystem Research in the Tropics 737

Mary

Tlie Applications of Molecular Evolution to System-

atics: Rates, Regulation, and the Role of Natural Selection I8i

King, Robert M. & Harold Robinson. Gmnjania dovidsei and Ilehecli-

n'lum gentrijl, New Species from Northern South America (Eupato-
rieae — Asteraceae)

Mabry, Tom J. The Order Centrospermae 210

366

M

18

Medina, Ernesto, (see Jordan, Carl F. & Ernesto Medina) 737

Gottlieb, L. D. Electrophoretic Evidence and Plant Systematics 161

Parnell, Dennis R. (see Raven, Peter II. & Dennis R. Parnell) 642

Prance, Ghillean T. Plant Inventory of the Tropics: Where Do We

Stand?

659

Ramirez B., William. A New Classification of Ficus 296

Raven, Peter H. A New Species of Lopezia (Onagraceae) from Sinaloa,

Mexico 638

Ram:n, Peter H. Perspectives in Tropical Botany: Concluding Remarks 746

Raven, Peter H. (see Hoch, Peter H. & Peter H. Raven) 136

Raven, Peter H. (see Seavey, S. R., Robert E. Magill & Peter H. Raven) 18

Raven, Peter II. (see Tomlinson, P. B. & Peter II. Raven) 657

Raven, Peter H. & Dennis R. Parnell. Reinterpretation of the Type of

Godetia hottae Sx:)ach (Onagraceae) 642

Reveal, James L. (see Stibolt, Virginia M., C. Rose Broome & James L.

Reveal )

373

Robinson, Harold, (see King, Robert M. & Harold Robinson) 366

Rourke, John & Delbert Wiens. Convergent Evolution in South African

and Australian Protcaceae and its Possible Bearing on Pollination by
Nonf lying Mammals 1

Seavey, S. R., Robert E. Magill & Peter H, Raven. Evolution of Seed

Size, Shape, and Surface Architecture in the Tribe Epilobieae (Ona-
graceae) 18

Spellman, David L. Four Species of Asclepiadaceae New to Panama 129

Stibolt, Virginia M., C. Rose Broome & James L, Reveal. Ahms mari-

tima Muhl ex Nutt., not Alnus metoporina Furlow 373

Straley, Gerald B. Systematics of Oenothera Sect. Kneiffia (Onagraceae) 381

ToMLiNSON, P. B. Plant Morphology and Anatomy in the Tropics — The

Need for Integrated Approaches 685

ToMLiNSON, P. B. & Peter H. Raven. Perspectives in Tropical Botany:

Introduction 657

Towner, Howard F. The Biosystematics of Calylophus (Onagraceae) __ 48

Turner, B. L. Chemosystematics and its Effect upon the Traditionalist __ 235

WiENS, Delbert. (see Rourke, John & Delbert Wiens) 1

WiTiiERSPOON, John T. New Taxa and Combinations in Eragrostis (Poa-

ceae)

324

WuNDERLiN, Richard P, A New Species of Bauhinia (Legvmiinosae) from

Peru :

371

OF THE

immki

VOLUME 64

1977

i

NUMBER 1

^

MAIN ENTRANCE, MISSOURI BOTANICAL GARDEh

CONTENTS

Convergent Floral Evolution in South African and Australian Proteaceae

and its Possible Bearing on Pollination by Nonflying Mammals John
Rourke ir Delbert Wiens — -

1

18

Evolution of Seed Size, Shape, and Surface Architecture in the Tribe Epilo-
bieae (Onagraceae) Steven R. Seavey, Robert E. Magill b- Peter

H. Raven

The Biosystematics of Calylophus (Onagraceae) Howard F. Towner 48

Chromosome Numbers in Legumes Peter Goldblatt b- Gerrit Davidse 121

Four Species of Asclepiadaceae New to Panama David L. Spelbnan 129

New Species of Gibsoniothamnus (Scrophulariaceae/Bignoniaceae) and

Tournefortia (Boraginaceae) from Eastern Panama and the Choco
Alwyn H. Gentry

133

NOTES

New Combinations in Epilobhim (Onagraceae) Peter C. Iloch ir Peter
H. Raven — — -—

Chromosome Number in Pillansia (Iridaceae) Peter Goldblatt ..„

A New Jacaranda (Bignoniaceae) from Ecuador and Peru Alwyn H.
Gentry

136
136

138

Phyllarthron bilabiatum: A New Species of Bignoniaceae from Mada-
gascar Alwyn H, Gentry

139

Taxonomic Notes and New Combinations in Leucophysalis (Solanaceae)
John £. Averett - -

Chromosome Numbers of Phanerogams. 7.

141

142

VOLUME 64

1977

NUMBER 1

OF THE

The Annals contains papers, primarily in systematic botany,
contributed from the Missouri Botanical Garden. Papers originating
outside the Garden will also be accepted. Authors should write the
editor for information concerning arrangements for publishing in

the Annals.

Editorial Committee

Gerrit Davidse, Editor-in-Chief
Missouri Botanical Garden

W. G. D'Arcy, Editor — Flora of Panama

Missouri Botanical Garden

John D, D^vyer

Missouri Botanical Garden & St. Louis University

Peter Goldblatt
Missouri Botanical Garden

Published four times a year by the Missouri Botanical Garden Press,

St. Louis, Missouri 63110.

For subscription information contact the Business Office of the Annals,
P.O. Box 368, 1041 New Hampshire, Lawrence, Kansas 66044.

Subscription price is $40 per volume U.S., Canada, and Mexico,

$45 all other countries. Four issues per volume.

Second class postage paid at Lawrence, Kansas 66044

© Missouri Botanical Garden 1977

OF THE

VOLUME 64

1977

NUMBER 1

CONVERGENT FLORAL EVOLUTION IN SOUTH AFRICAN

AND AUSTRALIAN PROTEACEAE AND ITS POSSIBLE
BEARING ON POLLINATION BY NONFLYING MAMMALS

John Rouhkk^ and DELiiEux Wiens-

AUSTHACT

Striking convergent evolution for a liitlden ( cryptic ), ground flowering ( gcoflorous )
habit in distantly related, low shrubby Australian and South African Proteaceae is interpreted
as an adaptation for joollination l)y nonf lying tnanuiials. The cryptic, gcoflorous habit is
especially well developed in species groups of Drtjandra in southwestern Australia and Piotea
in the Cape region of South Africa. Consideralile circumstantial evidence exists in both re-
gions for pollination by mouselike, often arboreal marsupials in Dryandra and true rodents
in Frotea. Evidence from inflorescence structure suggests the cryptic, gcoflorous habit is de-
rived from bird-pollinated species, possil)ly in response to fires common in the sclerophyllous
communities where these genera grow. A number of floral characteristics and the occurrence
in Australia of mouselike marsupials adapted to a nectar (and pollen?) diet suggests that a
class of flowers has e\'olved for pollination by nonflying mammals. This postulated floral class
possibly also extends to other Australian arboreal proteaceous and also myrtaceous genera,
but in South Africa is probably restricted to Pmtca.

Pollination by nonflying mammals is largely ignored or given little credence
in current treatments of pollination ecology (Faegri & van der Pijl, 1971; Proctor
& Yeo, 1972). There is, however, good reason for this; all the available evidence
relating to tliis phenomenon is either eirenmstantial, inferential, or anecdotal.

Nonetheless, field observations in Australia and South Africa and a subse-
quent search of the literature have led us to believe that true rodents and mar-
supials may, in fact, be the normal pollinators of several southern hemisphere
proteaceous genera. Fmthermore, various floral characteristics in these genera
and tlie special adaptations for nectar feeding in some of the putative pollinators
suggest structural coadaptations by both flowers and apparent pollinators. Al-
though plans are underway to conduct definitive studies, no unequivocal evidence
can be presented at this time for regular pollination by nonflying mammals, and

^ Kirstenbosch Botanic Garden, Newlands, Cape Province, South Africa 7700.
- Department of Biology, University of Utah, Salt Lake City, Utali 84112, U.S.A

Ann. Missouni Bor. Caiux 61: 1-17. 1977.

2 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

that is not the intent of tliis paper. We hope, however, that our comments will
help to reopen and stimulate research in this fascinating area of pollinatiou bi-
ology pioneered by Porsch (1934, 1935, 1936a, 1936b) and subsequently neglected
for 40 years. The purpose of tliis paper is fourfold: (1) to elucidate our observa-
tions and ideas on inferred pollination by nonflying maminals in the South African
and Australian Proteaceae, (2) to point out the striking convergent evolution of

flowering habits between southwestern /

Af

i<

of the Cape region, (3) to review some of the rather scattered and fragmentary
literature on the subject, and (4) to evaluate the evidence for the existence of a
class of flowxns adapted to pollination by nonflying mammals.

References to the subject of pollination by nonflying mammals usually men-
tion the arboreal Australian marsupials which apparently feed on nectar (e.g.,
the honey possum, Tarsipcs spenccrae) and to introduced rats suspected of pol-
linating a climbing pandan {Freycinetia arhorea Gaudich.) in Hawaii. Faegri &
van der Pijl (1971) and Proctor & Yeo (1972) furthermore state that no flowers
appear to be adapted for pollination by nonflying mammals, although Faegri and
van der Pijl mention the classic papers by Porsch (1934, 1935, 1936a, 1936b) in
which he builds a case for floral adaptations to pollination by nonflying mam-
mals in several Australian genera. Grant (1950) mentions, without comment, a
marsupial" pollinated flower class based on the proteaceous genus DnjamJriL

In addition to rodents and mouselike marsupials, some primates may also be
regular pollinators. For example, according to Coe & Isaac (1965) the baobab
{ Adamonia di<i,itata L.) is pollinated in East Africa by the lesser bush baby (a
lorisid primate). This characteristic African tree is generally considered to be
bat pollinated. Petter (1962) mentions that several arboreal, mouselike lemurs
{Lemur, Varcciay llapalemur^ Microcchus) visit flowers seeking nectar and are
generally attracted by sweet liquids in captivity. More recently Sussman & Tat-
tersall (1976, and personal communication) demonstrate that Lemur mongoz
mongoz is apparently an important pollinator of introduced kapok (Ceiha pen-
tcmdra Gaertn.) in Madagascar. F. L. Carpenter (personal communication) has
data from Australia indicating that some species of Banksia are pollinated almost
entirely by nonflying mammals, including an indigenous rat {Rattus fuscipes)
and various marsupials.

It is not our intent to evaluate the entire literature here. There are, however,
numerous instances of xarious mammals bein^ observed on or around flowers

(Porsch, 1934), but the nature of their activities are, in fact, virtually unknown.
As Faegri & van der Pijl (1971) point out with respect to pollination by mm-
flying mammals "much research remains to be done to establish relationship be-
tween possible regular pollinators and the blossoms in which they w^ork."

Flohal Chai^actkhistics and Convkrgent Evolution of Photeaceae

PUTATIVKLY POLLINATEI) liY NoNKlAlNC; MaMMALS

The most obvious Proteaceae are trees and large shrubs, e.g., Grevillea and
Banksia in Australia, and Protea in South Africa. Less known, however, is the

occurrence of species groups on both these continents with inflorescences at or
near ground level (geoflorous) and typically obscured from external view by

1977] ROURKE & WIENS— NONFLYIXG MAMMAL POLLINATION

3

overlying foliage (cryptic). The taxonomic distribution of these cryptic, geo-
florous species is limited principally to two distinct sections of Protea [Ilypo-
cephalae and Microgeantheae, sensu Phillips (1912)] and some additional species
of uncertain sectional classification in the Cape region of South Africa; in south-
western Australia, however, this flowering habit is associated with at least five
genera {Banksia, Conospermum, Drijamlra, Isopogon, and Petrophile) but is best
developed in Drtjandra [series Aphragmia and Niveae, sensu Benthani (1870)]
and to a somewhat lesser extent in Banksia.

In these equivalent infrageneric groupings in Protea and Dryandra the growth
habit is low, tufted, and often rhizomatous. The flowers occur in heads, usually
at ground level, or occasionally up to 30 cm high, but in either case the heads are
typically deeply hidden within the foliage of the dense and widely spreading
branch systems. The heads are generally visible only if the branches are forcibly
parted and the base of the plant carefully examined (Figs. 1-6). The flowers
are surrounded by a prominent series of overlapping bracts forming a cup-shaped
involucre. The bracts vary in color through various shades of brown and are
often flushed with different dull reddish tints. An inflorescence contains per-
haps 100-200 flowers, but the large spikes of Banksia bear several thousand in-
dividual flowers. Many of the species produce copious amounts of nectar and
the heads often emit a distinctive, "nutty'' or "yeasty" odor. In the cryptic, geo-
florous Cape species of Protea the basal portions of the bracts and flowers, par-
ticularly the styles, are also markedly succulent. Excellent illustrations of Protea
flowers (but not necessarily the growth habits) can be seen in Rousseau (1970)
for South African proteas and in Erickson et ah (1973) for Australian genera.

Dryandra, as in most western Australian Protcaceae, develops no obvious suc-
culence in the inflorescence or flowers. In general, the geoflorous habit, the cryp-
tic positioning of the inflorescences, and the gross (though superficial) morpho-
logical similarities of the heads suggest strong convergent evolutionary tendencies.
In fact, from a distance one would be hard pressed to distinguish between some
species of Dryandra and Protea even though these genera represent the end points
of evolution in two subfamilies of the Protcaceae, Grevilleoideae and Proteoidcae,
respectively, and occur on widely separated continents (Figs. 1-6).

EvmENCE FOR Rodent Pollination in South African Proteas
Field observations over a period of years of the cryptic, geoflorous species of

consi

these species (Table 1), but is especially obvious in P. ,mhuUfoIia. The specific
rodent activities associated with this species are: (1) freshly chewed involucral
bracts and styles during and just prior to anthesis (Fig. 7); (2) clearly demarcated
networks of heavily used runways linking different plants within populations, and
which often intertwine around flowering and old fruiting heads; and (3) occa-
sional burrows at the base of the plants.

The runways and burrows are related to activities of the Cape striped field
mouse (Rhabdomys pumilio pumilio). On various occasions and in different
populations this animal (which is diurnal) was observed on runways between

4

ANXALS OF THE MISSOURI BOTANICAL GAllDEN

[VoT,. 64

1977]

KOUKKE & WIENS— NONFLYING MAMMAL POLLINATION

5

(

Tahlk 1, The Suiitli African species of Vrotca in sections Ihjpoccphalnc and Mino-
iccuithcac. \\\ species gronnci flowering or near grormd flowering.

Hijpoccplialac

M ic r one CD i i hcac

P. suhidifolia (Salish. ex l\ araulos (L.) Reicharc?

Knight) Uourke"
P. amplcxicaulis R. Br.
P. decurrcns PhiUips
P. humlflora Andrews'*

P. an^tistata R. Br.
P. aspcra Phillips"
P. corduia Thunl).
P. foliosa Ronrke
P, glancophijUa SalisI).
P. hitonsa Rourke"
P. lacvis Tliiuil).
P. lorca R. Br/^

P. ))i()utau(i E. Mev. ex Xh'isn.
P. restionifoJia (Salisb. ex

Knight) R> croft'
P. recohda Buck ex Meisn.
P. scahra R. Br."
P. sc(d)ruisc\da Phillips
P. scolpcndrium R. Br.
P. scorzoncrifolia Salisb. ex

Knight
P. sxdfiuca Phillips''
P. voiltsiac Ronrke"

" Spfcies ii) wliicli evidt^ice uf rotk'iil activities has ]>een obstrvi'd on flowers.
'* Iiitlusioii in tins taxonomic settioji (lufstionahle.

flowerinj^ plants of F. suJmlif^

to the

///,

further demon-

strated when it was live-trapped utilizinj^ fresli (lowering heads of this species
as bait. In this instance traditional rodent l)aits snch as peanut l)ntter were in-
effectual in capturing this ain'iual.

The Cape striped field mouse apparently also visits the flowering heads of
Protea nana (Berg.) Thunb., a low shrubby species witli pendulous heads of un-
certain pollination type and not a member of the geoflorous seetions. The soft
floral parts of P, nana were chewed In the same manner as P. sulndifoVui and a
Cape striped field mouse was trapped at this plant within 24 h of the first noted
rodent activity.

The fleshy involucral bracts and styles of the cryptic, geoflorous proteas show
widespread (^vidence of being chewed. In one population of P. siihuVijoJia 17
plants bearing 49 inflorescences with open flowers were obser\ ed in an area of
approximately 30 m-; 20 heads, or 40%, showed extensive c^vidence of chewed
bracts and styles (Cape striped field mice were common in the area). The con-
sistent occurrence of chewed bracts and styles in the cryptic, geoflorous proteas
suggests that they function as food bodies; supporting this idea is the sweetness
(at least to the human palate) of these structures. In the larg(% bird-pollinated
proteas the bracts not only lack succulence Init are markedl)' acrid, apparently
containing high concentrations of taimin. In spite of their sweet, fleshy nature
and apparent lack of tannin, all the bracts and mature styles of any head are
rarely eaten. That the inflorescences are not completely destroyed suggests the
presence of secondary compounds which might limit the amount of feeding
as proposed by Freeland & TanzcMi (1974).

Figures 1-4. — 1. General aspect of Protea sulndifolia jnst beginning to flower (near
llennanus, Cape Prov., Sonth Africa). — 2. Cryptic, geoflorous inflorescences of P. suhidifolia
(same plant as ahove). — 3. General aspect of Drijandra tcnuifolia in full flower (Stirling
Range, West Australia). ^4. Cryptic, geoflorous inflorescences of D. tcnuifolia (same plant
as ahove).

Q ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 6i

Sweet, fleshy l)racts are also associated witli pollination in Freycinelia (Pan-
danaeeae) wliere they reportedly attract introduced rats in Hawaii (Degcncr,
1945) and in other areas of the Pacific (li. C. Stone, personal communication).
Freycinctia insi^nis Blume, an Asian bat-pollinated species, apparently utilizes
only odor and fleshy bracts as attracting devices (Proctor & Yeo, 1972).

The sense of smell is well developed in rodents, and in view of the hidden
nature of the inflorescences, odor must be the primaiy attracting mechanism re-
gardless of the pollinator. Furthermore, at the time of flowering in P. suhtdifolia
(late winter), the fleshy bracts and styles constitute one of the best sources of
soft palatable vegetable matter in the local plant community. Thus flowering
might correlate with the low point in the food cycle of rodents. The Cape striped
field mouse is apparently fond of soft vegetable matter, sometimes becoming a
nuisance in vegetable gardens (Roberts, 1951). As further evidence of this
dietary habit, newly harvested shoots of another proteaceous shrub, Leucodemlron
modestum Williams, were obserxed along a typical runway and entrance to a
Cape striped field mouse burrow. If the chewing activity in the flowering heads
of the Cape species of geoflorous proteas is due to the Cape striped field mouse,
or a similar-sized animal, pollen would surely accumulate about the head of the
animal and should theoretically be capable of transfer to nearby plants. The fur
of mammals should provide an excellent surface for pollen accumulation. This
is demonstrated by the presence of pollen on the head of a nectar-feeding Aus-
tralian marsupial, the sugar glider (Petaurus hreviceps) (Breeden & Breeden,
1970: inside back cover). An interesting description of pollen accumulation on
the Australian honey possum {Tarsipes spencerac) feeding on Proteaeeae is also
given by Vose ( 1972 ) .

Because of its known association with Protca (especially P. stihtili folia), the
Cape striped field mouse is perhaps the best possibility for a mammal pollinator
of the cryptic, geoflorous species of Protea. Ilow^cver, other rodents in the Cape
fauna should also be examined for possible activities relating to pollination. Dr.
J. Jarvis of Cape Town University (personal communication) suggests especially
the following animals: Dendromus mdanotis (climbing mouse), Leg^iada rninu-
toidcs (dwarf mouse), Otomys irroratus (vlei otomys), and Acomys stihspinosus
(Cape spiny mouse). None of these animals, however, appear to have any special

adaptations for nectar or pollen feeding.

A single case of interspecific hybridization (P. rcstionifolia X P. humiflora)
is known among the cryptic, geoflorous species of Cape Protea. That such a cross
occurs is proof that pollen can be transferred between these species. Further-
more, evidence of rodent activity is known in both parental species of the cross.
The Proteaeeae are apparently adapted for outcrossing and thus require a mecha-
nistn for pollen transfer. The family is apparently either protandrous (Rao, 1971)
or self-incompatible (Horn, 1962). Pollen dispersal ultimately occurs from a
specialized region of the style apex known as the pollen presenter (Rourke, 1969).
This is so close to the slitlike stigmatic surface that mechanisms to prevent
autogamy must be present or selling would be the rule and pollination unneces-
sary.

1977] ROUHKE & WIENS— NONFLYING MAMMAL POLLINATION 7

Evidence for Marsupial and Rodent Pollination in Australian Proteaceae

Field observations of the inflorescences of Dryandra temiifolia R. Br. in sonth-
vvestern Australia also showed evidence of nianunal activities similar to those men-
tioned for Protea suhulifoUa from the Cape region of South Africa; chewed heads
were particularly common. The inflorescences were also odoriferous and the
scent was surprisingly similar to the 'yeastlike'' odors prevalent in the cryptic,
geoflorous species of African Protea, Copious nectar was not detected, but our
observations were made in mid-afternoon when nectar content was possibly low.
Nectar production in the Australian cryptic, geoflorous Proteaceae may be
largely nocturnal to coincide with increased animal activity at that time ( Mor-
combe, 1968). Porsch (1935) repeatedly mentions high nectar production in
Dryandra nivea R, Br., which he obsei-ved under cultivation. He also noted noc-
turnal anthesis and an odor of ''sour milk" or "caraway liquor" in this species. Dr.
Alex George (personal communication) has also seen apparent mammal activity
in the inflorescences of the cryptic, geoflorous species of Banksia where chewing
and disturbance of the flowers appeared to be similar to our observations in South
Africa. F, L. Carpenter (personal communication) also has interesting evidence
that Banksia species in eastern Australia are largely pollinated by nonflying mam-
mals. She correlates nonflying mammal pollination in Banksia with the occur-
rence of stiff inflexed styles (illustrated in Baglin et al., 1972) which apparently
exclude foraging birds. Porsch (1935) suggested this as a feature of marsupial-
pollinated banksias; he also proposed that the "basket"-like inflorescences in some
dryandras were adapted to accommodate the heads of various marsupials (Fig. 8).
The situation in Australia, however, is probably more complex than in South
Africa. For example, many of the large, shrubby and even arboreal Proteaceae
(and also Myrtaceae) are also visited by nonflying mammals in addition to the
cryptic, geoflorous species, yet the latter appear to be better adapted for polli-
nation by nonflying mammals. Most workers probably consider these nongroimd
flowering species to be bird pollinated (e.g., Carlquist, 1974). Admittedly many
of the floral characteristics of genera such as Banksia do suggest bird pollination.
Yet some traits clearly do not. For example, Baglhi et al. (1972) state that all
Banksia inflorescences are odoriferous, yet odor is not associated with ornithophily.
Additionally, Morcombe (1968) reports that in Banksia nectar secretion is pro-
lific at night, a condition hardly adapted to pollination by diurnal flower birds.
Morcombe suggests that the great abundance of nocturnal insects are attracted to
Banksia inflorescences by the copious nectar, and these in turn are what entices
nonflying mammals to the flowers. Considering the highly specialized adai^ta-
tions of an animal such as the honey possum (see following discussion) for a nec-
tar (and pollen?) diet, it seems unlikely that insects would be the prime at-
tractaut, at least for this animal. However, animals such as the southwestern
bush rat (a true rodent) might well be attracted by insects. But this would
hardly explain why nectar secretion is abundant at night, since insects are highly
unhkely pollinators of these flowers. Typically, nectar secretion is synchronized
temporally for visitation by the established pollinators coadapted to that par-
ticular flower ( Faegri & van der Pijl, 1971).

8

ANNALS OF THE MISSOUlU lUVrAMCAL GAHDKN

[Vol. 64

1977] ROUBKE & W lEXS — NONFLYIXG MAMMAL TOLLINATION Q

Derivation of Chyptic, Geoflorous Species from Ornithophilous

Prototypes

Various sunbirds and the Cape sugarbird are the typical polhnators of the
well-known shrubby proteas in the Cape region with large terminal inflorescences
(Fig. 9). However, the cryptic, geoflorous positioning of the inflorescences in
sections Hypocephalae and Micw^i^eantlieae must preclude bird pollination,
since birds are attracted to flowers visually (Raven, 1972) and odor is not a
characteristic of bird-pollinated flowers. In fact, sunbird or sugarbird visits to
the cryptic, geoflorous proteas would violate established behavioral patterns in
these birds. They are not generally known to frequent the ground, or to explore
the dense interior of low shrubs which do not have exposed, colorful flowers.

Further evidence that birds are unlikely pollinators of the cryptic, geoflorous
proteas is based on observations of P. nana. As previously mentioned, this species
has pendulous, relatively small (ca. 3 cm wide), dark reddish heads. Initially
one might assume that they were bird pollinated. However, a number of flower-
ing plants of P. nana were observed in an area densely populated by the orange-
breasted sunbird, Nectarinia violacea and the Cape sugarbird {Fromerops cafer)
which were feeding freely on several proteaceous shrubs with large terminal
heads, and some ericas; however, no birds were observed on P. nana. Because the
heads are pendulous, flower-visiting birds would probably have to hover to ob-
tain nectar. Sunbirds are capable of hovering (Skead, 1967) but unlike hum-
mingbirds, they hover clumsily. Normally they feed while clasping branches.
Nonetheless, in an area with a high density of nectar-feeding birds, a great va-
riety of flowers are normally visited in addition to the preferred species. If sun-
birds and sugarbirds showed interest in the flowers of P. nana, at least rare visits to
these plants would be expected. If nectar-seeking birds are not attracted to P.
nana, whose inflorescences are visually conspicuous but otherwise generally re-
semble the cryptic, geoflorous species, it is still more difficult to believe that
birds pollinate the latter group. In fact, the great majority of these South African
Protcaceae with dark reddish bracts and mostly pendulous flowers might well be

pollinated by nonflying mammals.

The cryptic, geoflorous proteas do retain the copious nectar supply typical

of bird flowers. They differ from bird flowers, however, by (1) bearing their
flowers at or near ground level in a hidden position, (2) emitting a strong "yeast-
like" odor, (3) possessing much shorter flowers (ca. 1.5 cm high), and heads of
smaller diameter (ca. 4-6 cm wide), and (4) the dull purplish brown coloration
of the heads as opposed to the bright, vivid red and/or yellow inflorescences of
the bird-pollinated species.

Essentially the same arguments apply to the situation in Australia, except

Figures .5-8. — 5. Inflorescences of Protea suhvlifolia at anthesis (same plant as Figs.
1-2). — 6. Inflorescences of Dryandra tenuifolia at anthesis (same plant as Figs. 3-4). — 7. P.
suhnUfolia showing chewed snccnlent bracts and styles (left) and intact inflorescence (right)
(near Papies \'lei, Cape Prov., South Africa). — 8. Dryandra sp., note inflexcd styles forming
a **basket"-shaped inflorescence (near Perth, West Australia).

10

ANNALS OF THE MISSOURI liOTANICAL GARDKN

[Vol. 64

1977] HOURKE & WIENS— NONFLYING MAMMAL POLLINATION H

that here the predominant flower birds are tlie honey eaters (Melix^hagidae).
Likewise in many of tlie critical genera such as Banksia^ some species are ap-
parently adapted for bird polhnation and others for pollination by nonflying
mammals.

Although we believe the cr>^ptic, gcoflorous Proteaccac are not pollinated
by birds, the system nonetheless appears to have evolved from a bird-pollinated
prototype. In addition to some general aspects of the inflorescence and the
copious supply of nectar, the nature of the branching patterns associated witli
inflorescence development also supports the derived nature of the cryptic, gco-
florous Proteaceae. The large, shrubby bird-pollinated proteas have terminal
inflorescences, whereas the inflorescences in the geoflorous species are largely
axillary. These axillary inflorescences are almost certainly derived from terminal
inflorescences by progressive stem reduction. Members of the section Pinifoliae
(which includes P, nana) possess intermediate forms in which the stem bearing
the inflorescence is greatly shortened. Complete reduction of this stem would
give rise to the almost sessile, apparent axillary inflorescences characteristic of
P. stihiiJifolia and other highly reduced types that occur in the geoflorous sections.
Thus morphological evidence supports the proposition that the geoflorous species
are derived types originating from bird-pollinated groups. Both L. A. S. Johnson
and A. George (personal communications) support the notion that the cryptic,
geoflorous Australian Proteaceae are also derived types.

Another interesting example of the apparently deriv(xl nature of the cryptic
habit occurs in Protea recondita Buck ex Meisn. where floral crypsis is accom-
plished through an entirely different mechanism than geoflory. This species is
a low shrub (up to perhaps 1 m high) with terminal inflorescences positioned
similarly to the bird-pollinated proteas. The bases of the heads, however, are
encircled by a cluster of unusually large, vertically oriented leaves (bracts).
These bracts enfold the entire inflorescence (rather like a cabbage!) and, in
effect, obscure the head from external view during anthesis (Figs. 10-11).

If the cryptic, geoflorous species of Proteaceae are adapted for pollination
by nonflying mammals and were derived from bird-pollinated prototypes, what
selective forces might have shifted the system in this direction? The ecological
community in which these plants occur provides a possible explanation. Both
the Cape region and southwestern Australia are essentially sclerophyllous, fire-
adapted shrub communities. In fact, the general aspect of the two communities
is remarkably similar, even to the characteristic brownish cast of the vegetation.
Furthermore, both regions are extraordinarily rich floristically. Only tropical
rainforests are apparently richer in plant species diversity. Both floras are also

Figures 9-12. — 9. Protea comparta, a typical hird-pollinatcd spccios at anthesis (near
Papies Vlei, Cape Prov., South Africa). — 10. P. recondita, shoot with hidden terminal in-
florescence at anthesis ( Kirstenbosch Botanic Garden, Cape Prov., South Africa). — 11. P. re-
condita shoot with terminal inflorescence exposed behind the large, vertical bracts (same plant
as Fig. 10). — 12. (left) Bank^ia media (from a kodachrome by Alex George, West Australia)

inflorescence at anthesis, each rounded point represents one flower, total number of flowers

estimated at 4,400. (right) Banksia inflorescence with 34 mature fruits.

J2

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Tabll 2. AMstraliaii iiiannnals known to visit flttwcrs to ol)tain nectar and/or pollen.

Animal (all marsupials, except Rattus fw^cipcs)

References

AcroJxUcs py^nuu'us (pi.t^my glider)
Antcchinu:^ a}}icalLs (dihhler)

A. flavipcs (>el!o\v-ft)Oted anteehinns)
Burramrjs parvus (mountain pigmy pos-
sum)
Ccrrarfctits conriiiuns (southwestern

pigmy possum)
C )ianus (eastern pigmy possum)
Pclaurus ausfralis (fluffy or yellow-hel-

lied glider)
P. l)r('viccps (sugar glider)
P. tiorfolrcusis (squirrel glider)

Breed(>n & Breeden (H372), Carlquist (1965)
Morcoml)e (1908)

Breeden & Breeden (1972)

P. Cook (personal eoniuumieation)

A'^ose (personal communication), Hide (1970)

Baglin el al. (1972), Breeden & Breeden (1972)
Breeden & Breeden (1972)

Breeden & Breeden (1970, 1972), Sleumer (1955)
Breeden & Breeden (1972)

Phascogalc tapvatafa (iudu or WdmhvU' Breeden & Breeden (MJ'J'^j

ger )

Rattus fttscipcs (southern hush-rat)
Tarsi})c.s spcnccrac (honey mouse)

Moreoml)(> (19()S)

M(neom1)e (1908), Clanert (1958), Vose (1971,
1972), Ride (1970).

characterized by nutritionally depauperate soils (Wild, 1968; Loveless, 1961).
Insofar as the evolution of the cryptic, geoflorous habit is concerned, however,
we believe the fiie-adapted nature of these coiunuiuities is most important.

One way plants can survive burning is to develop rhizomaty. This condition
should sti'ongly promote ground flowering. Many, but not all, of the cryptic,
geoflorous Proteaccae are rhizomatous. Hence, if rhizomaty has survival value
and if there were pollinator competition between bird and nonflying mammals
for floral resources, fire and the concomitant development of tlie rhizomatous
habit could have shifted the selective advantage toward nonflying mammal pol-
lination. It is also possible, however, that nonflying mammals, as a result of their
generally more aggressive behavior, may ha\'e simply out-competed birds as
pollinators and hence shifted the selective balance in this way.

A Flohal Class Adai'ikd kou Pollinaiiox i^y Nonflyixc Mammals

AND EvmKNCK FOU CoKVOLUriON

I

Although experimental data are lacking, we believe sufficient circumstantial
evidence is a\ ailable to identify a class of flowers in Proteaccae adapted for pol-
lination by nonflying mammals. Furthermore, this class of flowers has evolved
independently at least twice (Africa and Australia) and at least one animal, the
so-called honey possum {Tarsipes spcnccrac) has probably coevolved with this
floral class in Australia. Furthermore, we believe other animals, particularly
some of those in Table 2, may also ha\'e coevolved with these proteaceous plants.

The fundamental characteristics we believe might distinguish this floral class
include: (1) Inflorescences as the basic units of attraction; generally they arc

cup-shaped heads (spikes in Banksia). (2) Heads t)pically hidden deep within

the foliage, often at or near ground level; if exposed (as in Banksia) then with

(a) structural modifications, such as stiff incurved styles or (b) nocturnal

1977] ROURKE & WIENS— NONFLYING MAMMAL POLLINATION ]^3

rhythms of nectar prockiction and/or anthesis to prechicle successful nectar forag-
ing by birds. (3) Heads about 2-8 cm wide with perhaps 100-200 flowers (several
thousand in Banksia), and strongly attached to stems, (4) Heads producing a
copious nectar supply; in some proteas also possessing apparent food bodies in the
form of soft, fleshy bracts and styles acting as complementary attractants. (5)
Heads odoriferous; we characterize these as "nutty" or "yeasty" in Protea; Porsch
(1935) suggests "sour milk" and "caraway Hquor" for Dryandra. (6) Heads with
reddish brown to purphsh bracts, individual flowers mostly whitish. (7) Temporal
spacing of anthesis in the inflorescence, thereby limiting the number of simul-
taneously open flowers in the head to no more than several of the outer whorls.

Most of these characteristics were discussed previously and need no further
elaboration. The most obvious feature of this putative floral class is its basic
resemblance to bat-pollinated flowers (cf. Faegri & van der Fiji, 1971). The pri-
mary differences are the cryptic, geoflorous habits and the compound inflorescence
as the attracting unit. There are apparently also structural modifications of the
styles in the Australian species to discourage bird foraging. The basic similari-
ties to bat flowers, however, should not be surprising since the apparent pol-
linators are all small mammals with perhaps generally similar energetic require-
ments and sensory systems [Faegri & van der Fiji (1971) point out that echo
location is only poorly developed in the flower-feeding bats, Megachiroptera].

One of the strongest lines of indirect evidence supporting the idea of a class
of flowers pollinated by nonflying mammals in South Africa and Australia re-
volves about the convergent nature of the floral characteristics in these two sub-
families of Froteaceae. If one examines the pollination syndrome of any flower
class, they have essentially the same general features over the entire world. Thus
convergent evolution for floral structure and habit is a necessary product of any
widespread pollination system.

The variations in floral habit and structure among the Froteaceae putatively
pollinated by nonflying mammals therefore reflect differences In (1) modes of
locomotion to the flowers (i.e., terrestrial or arboreal movements as opposed to
flying) and (2) foraging behavior. The great mobiHty of the nonflying mam-
mals around flowers and the highly developed chewing apparatus (particularly
among generalized feeding rodents) would probably make an attracting unit con-
sisting of a single, large flower nonadaptive because of tlie destructive nature
of these animals. The Froteaceae have apparently compensated for the highly
destructive activities of these apparent pollinators by increasing the number of
reproductive units far beyond what is necessary to maintain successful repro-
ductive levels. In the South African cryptic, geoflorous species seed set is con-
sistently low, usually below 5%. The same is true in the corresponding Australian
genera (A. George, personal connnunication). In Banksia, the number of flowers
per spike probably exceeds 4,000, yet the mature fruits are so large that it would
be a physical impossibility for more than perhaps 50 to develop (Fig. 12). In
addition to maximizing flower production, the flowering patterns in the heads
are staggered temporally so tliat only several outer whorls are in anthesis simul-
taneously. If all the flowers opened concurrently, and in view of their sweetness

14

AXXALS OF TIIK MISSOURI B0TA\K;AL GAHDEX [Vol. 64

at aiithesis, tlic entire inflorescence might be more easily destroyed by attracting

a large number of pollinators.

If generalist feeding, nonflying mammals are potentially so aggressive and
opportunistic in exploiting food sources, one might ask why they rarely disturb
the typically bird-poUinated South African proteas. First, the bird-pollinated
species are not odoriferous and the terminal inflorescences are typically borne
high above ground level ( Fig. 9 ) ; thus most such inflorescences probably escape
detection. Another aspect, however, is the presence in some species (especially
section Speciosac) of a thick layer of trichomes over the top of the heads. Such
a dense layer of trichomes might serve to discourage mammals from chewing to
the base of the heads where the nectar is located and thus act as a kind of "mam-
mal guard." Birds, of course, easil>' probe through this layer with their long bills.
The heads are also closely surrounded by stiff leathery bracts, which also have
considerable trichome development along their margins, where chewing is most
apt to be initiated. Excellent illustrations of these phenomena are found in Rous-
seau (1970). Finally, the acrid taste of these bracts, as opposed to the sweetness
of the bracts and styles in the cryptic, geoflorous species, might also be important.

Many observations of nonflying mannnals on Proteaceae (and also Myrtaceae)
are reported in the popular natural history literature of Austraha, e.g., Serventy
& Raymond ( 1974 ) and Russell ( 1974 ) ( see also references in Table 2 ) . In fact,
so prevalent are these observations that many Australian biologists take this pol-
lination system essentially for granted (Moreombe, 1968; Johnson & Briggs, 1975).
To our knowledge, however, no definitive data to estabhsh this relationship has
yet been published. Of the nonflying mammals listed in Table 2 as potential
pollinators, the honey possum or noolbenger {Tampes spencerae), appears to be
the best known and apparently the most highly specialized for nectar (and pol-
len?) feeding. This amazing animal was studied in captivity by Glauert (1958)
and Vose (1972, 1973), who hiclude illustrations. Because no comprehensive re-
view of its spectacular adaptations for nectar ( and pollen? ) feeding is evidently
available, a brief resume of these characteristics taken from the sources quoted

1965)

6-^

The

elongated, tapering snout composes two-thirds of the head. The ears are set far
back on the head and the nose is grooved. These features no doubt allow the

honey possum to probe deeply into flowers. The tail is longer than both the

8-10

For grasping; both characteristics being excellent modifica-
tions for the arboreal habit. But it is in the mouth where the most fascinating
adaptations for nectar (and pollen?) feeding exist.

The tongue is extensible to twice its normal length, tapered, .slightly serrated
on the margin and brushed at the tip (Fig. 2 in Vose, 1972). It is exserted
through a funnellike structure at the tip of the tapering snout where the lips are
modified into flanges. The palate is characterized by ridges which apparently
remove accumulated nectar (and pollen?) from the tongue when it is retracted.
The jaws are much reduced and dentition rudimentary. Only the upper canines
and lower incisors are developed and these appear to function largely in orienting

l'J77] ROURKE & WIENS— NONFLYING MANfMAL POLLINATION 3^5

the tongue during retraction. There is no caecum, such a digesting organ for
solid food apparently being superfluous in an animal adapted to a nectar diet.
Additional structural and especially physiological adaptations for nectar (and
pollen?) feeding will no doubt be discovered when more extensive studies are
conducted. The evidence that Tarsipes is adapted to a diet derived from flowers
(and probably occasional insects) is overwhelming. As a corollary, the conclu-
sion that Tarsipes has coevolved as a pollinating agent with various proteaceous
(and myrtaceous) genera is inescapable.

Sleumer (1955) states that in both northeastern Australia and southeastern
New Guinea, the sugar glider (Petaurus hreviceps), along with several flower
birds, are always associated with flowering Banksia denfata and the myrtaceous
genera Melaleuca and Eucalyptus. According to Sleumer, the sugar glider sucks
nectar with a "wormshaped" tongue, suggesting possible anatomical adaptations
for a nectar diet.

Although reference has so far only been made to the cryptic, geoflorous pro-
teas, other Australian proteaceous and myrtaceous genera such as Eucalyptus and
Melaleuca are known, or suspected, to be visited by various nonflying mammals.
For example, Tarsipes reportedly feeds on Ilakea and Beaufortia (Morcombe,
1968). \'ose (1972) also lists species of Callistemon and Grevillea from which
Tarsipes will extract nectar in captivity. Additional reports of nectar sources for
flower-visiting marsupials include Dryamlra (Glauert, 1958) and Angophora
(Porsch, 1934).

If the honey possum and possibly also other small arboreal marsupials have
apparently coevolved in Australia, why has coevolution between Proteaceae and
true rodents not occurred in South Africa? One obvious reason is that the proteas
in South Africa ostensibly pollinated by nonflying manmials do not flower
throughout the year. The flowering period for these plant groups is limited
primarily to late winter or early spring, as previously mentioned. Thus coevolu-
tion is impossi])]e because the flowers do not provide a constant food source
for these animals which are active throughout the year. Furthermore, it is likely
that plants can adapt relatively easily to a generalized feeder, such as many
rodents, and that pollination can be reasonably well assured by offering high
rewards and reducing competition with other food sources in the connnunity
by flowering at the low point in the food cycle.

Potential Pollination by Nonflying Mammals in Oihfr Plant Groups

con

Proteaceae by nonflying mammals in South Africa and Australia because we ob-
served many of these species and genera in the field. In any overall consideration
of the phenomenon, however, other plant groups should not be overlooked. If
pollination by small, arboreal marsupials occurs in Australian Proteaceae, it prob-
ably also occurs in Myrtaceae. The mouselike lemurs on Madagascar (Sussman &
Tattersall, 1976) which take nectar from introduced kapok must be adapted
for visiting similar indigenous flowers as well. Porsch (1935) mentions Mada-
gascan Symphonia (Guttiferae) as a possible flower adapted for pollination by

jg ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 64

noiiflying maninials. lie also discusses otlier families, e.g., Bombacaccae and
Lecythidaceac, wliieli also might have adaptations for pollination by various
nonflying mammals. Porsch's observations merit careful reconsideration, and,
especially, critical field studies to test his hypotheses.

LrrKRATUHE ClIKI)

n.uiLix, IX, B. MuLLiNs iy F. IkruLEY. 1972. A Treasury of Australian Wiklflowcrs. Ure

Smith Pty., Ltd., Sydney.
Br.NTHAM, C. 1870. Proleaceae. In Flora Aiistralirnsis. Xo], 5: 564-565.
Bi^EKDEx, S. & K. BuEEDEN. 1970. Living Marsupials. William Collins (Australia) Ltd.,

Sydney.

1972, Australia's South East. A Natural History of Australia: 2. Taplin^or rubl.

Co., Inc. New York.
Caulquist, S. 1965. Island Life. Natural History Press, Garden City, New York.

. 1974. Island Biology. Columbia Uni\ . Tress, New York.

CoK, M. J. & F. M. Isaac. 1965. Pollination of the baobab {Adamonia digiUita L.) by the

'lesser bush baby {Gahi^o crassicaudatus E. Geoffroy). E. African Wildlife J. 3: 123-124,
Dt:(;E\i:H, O. 1945. Plants of Hawaii National Park Illustrative of Plants and Customs of the

Soutli Seas. Edwards Bros., Inc., Ann Arbor, Miehi,v.'an.
Ericksox, R., a. S. Georce, N. G. Marchant ik M K. Morcombe. 1973. Flowers and

Plants of Western Australia. A. II. & A. W. Reed ( Pt> . ) Ltd., Perth.
Fae(;ri, K. & L. van ukk Pijl. 1971. Principles of Pollination Ecology. Ed. 2. Pergamon

Press, Ltd., London.
Fuekland, W. J. & D. H. Janzex. 1974. Strategies in hcrbivory by mammals: the role of

plant sccondar>' compounds. Amer. Naturalist 108: 269-289.
Geauert, L. 1958. The hone)- mouse. Austral. Mus. \hig. 12: 284-285.
Grant, V. 1950. The protection of the o\ules in flowering plants. Evolution 4: 179-201.
HoiiN, \v. 1962. BnHHling research on S(mth African plants: II. Fertility of Proleaceae.

j! S. African Bot. 28: 259-268.
Johnson, L. A. S. & B. G. Buj(;gs. 1975. On the Proteaceae— the e\olution and classification

of a southern family. Bot. J. Linn. Soc. 70: 83-182.
Loveless, A. R. 1961. The nutritional interpretation of selerophylly based on differences in

the chemical composition of sclerophyllous and mesoph>tic lea\'es. Ami. Bot. (London),

n.s.,25: 168-184.

MoHcoMLiE, M. K. 1968. Australia's Western Wildflowers. Landfall Press, Perth.
PirrrEH, J. ]. 1962. Ecological and behavioral studies of Madagascar lemurs in the field.

Aim. New York Acad. Sci. 102(2): 267-281.
PiiiLLH^s, E. P. 1912. Prt)teaceae. In Flora Gapcnsis. Vo]. 5: 560-501.
PonscH, O. 1934. Siiugetiere als Blumenausbeuter und die Frago der Saugetierblume. I.

Biol. Gen. 10: 657-685.

— . 1935. Siiugetiere als Blumenausbeuter und die 1- rage der Siiugetierblume. II. Biol.

Gen. 11: 171-188.

. 1936a. Saugetiere als Blumenausbeuter und die Frage der Siiugetierblume. III.

Biol. Gen. 12: 1-21.

1936b. Sllugetierblumen. Forsch. & Fortsehr. 12: 207.

PnocTon, M. & P. Yeo. 1972. The Pollination of Flowers. Taplinger Publ. Co., New York.
Rao, G. V. 1971. Proteaceae. Botanical Monograph 6. Council of Scientific and Industrial

Research, New Delhi.
Raven', p. II. 1972. Why are bird-visited flowers predominently red? Evolution 26: 674.
Rn^E, W. D. L. 1970. A Guide to the Native Mammals of Australia. Oxford Univ. Press,

Melbourne.
RoHEHTs, A. 1951. In The Nhnumals of South Africa, R. Bigalke, V. FitzSimons, & D. E.
Malan (editt)rs). Published by the trustees of the manuuals of South Africa book fmid.

Ilafner Publishing Co., N.Y.
RounKE, J, P. 1969. Taxonomic studies on Soro(C})]iaJtts and Spotalla. J. S. African Bot.

Suppl. 7: 1-124.
Rousseau, F. 1970. The Proteaceae of Soutli Africa. Purnell & Sons (S.A.) Ply., Ltd., Cape

Town.
Russell, F. 1974. Marsupial mysteries and mar\(4s. Audubon 76: 14-35.

1977] ROURKE & WIENS— NONFLYING MAMMAL POLLINATION ^7

Serventy, v. & R. Raymond. 1974. The honey possum. Austral. Wildlife Heritage. 4(48):

1505-1509.

Skeai), C. J. 1967. The Sunbirds of Southern Afriea, Also the Sugarbirds, the White-eyes
and the Spotted Creeper. South African Bird Book Fund, A. A. Balkema, Cape Town.

Sleumeh, II. 1955. Proteaeeae. In Flora Malesiana, C. C. G. J. van Steenis (editor). Ser. 1,
vol. 5(2): 147-206.

SussMAx, R. W. & I. Tattersall. 1976. Cycles of activity, group composition, and diet of
lemfiv mongoz mongoz Linnaeus 1776 in Madagascar. Folia Primatologia 26: 270-283.

VosE, H. 1972. Some observations on a honey possum (Tarsipes spencerae) in captivity.
W. Austral. Naturalist 12; 61-67.

. 1973. Feeding habits t)f the western Australian honey possum, Tarsipes spencerae.

J. Mammal. 54: 245-247.
Wild, H. 1968. Phytogeography in South Central Africa. Kirkia 6; 197-222.

EVOLUTION OF SEED SIZE, SHAPE, AND SURFACE
ARCHITECTURE IN THE TRIBE EPILOBIEAE

(ONACxRACEAE)^

Stkven ]\. Skavey-, Rohkht E. Magill^ and Pi:tI':r 11. Raven^

AllSTRACT

The seeds of move than half of the appioxiiiiately 210 speeies of Onagracoac trihe
Epilohieae were examined with the scanning electron microscope. The six species of Boisduvalia
ha\'e irre^uhn-I)' angnlar-fnsifonn seeds with con\'ex, flat, irregularly polygonal surface cells
in two species and an irregularly striated reticulum fornu^l hy the unevenly joining walls of
the STU'face cells in tlu* four others. They are similar to one another and sharply distinct from
those of KpHohinuu although the relationship between the genera is undoubtedly close. The
scvds of Epilohium fall in seven groups: ( 1) large, oboxoid seeds with a more or less prominent
micropylar constriction, in three small sections of generalized xerophytes and in one species,
E. rigkhtni, of sect. Epilohium; (2) smaller papillose seeds in over half of the oth(M* specii^s;
(3) fo\eolate seeds, independently evolved in many species; (4) olxnoid-patelliform seeds
in four Australasian species; (5) irregularly reticulate seeds in one subsection of Epilohium
sect. Cham(icncrio)i\ (fi) ridged seeds in a phylogenetically coherent group of North American
origin; (7) finely papillose, distinctive seeds in sect. Croswsti^ma. More or less prominent
chala/al beaks have e\()l\ed in some species. From xeroplntic ancestors, Epilohium has evolved
a highly successful group of niesophytes in sect. Epilohium that have achieved worldwide dis-
tribution. This trend seems to ha\e been accompanied by an increase in seed number and a
concomitant decrease in seed size.

The well-iuarkecl tribe Epilohieae, one of six that make up the family
Onagraceae, includes some 200 species of Epih)hium^ of worldvvicle distribution;
and six of Boisduvalia, five of western North America, with one common to Argen-
tina, and one additional species restricted to western South America. The western
North American 7AiuscJincna, often recognized as distinct from Epilolmnn^ is
based on a red-flowered, bird-pollinated species of one of the constituent groups
of Epilohium. 7Aiuschncria lias accordingly been reduced to the status of a section
of Epilohium (Raven, 1976). Of the six sections of Epilolmim, two, with a total

of three species, consist of annuals and are restricted to western North America;

two others, with a total of four species, are generalized xeropliytic perennials

of w^cstern North America; one, Chamaeuerion, includes seven species of Eurasia,

two of which extend to North America; and the remaining one, sect. Epilohiumy

consists of some 185 species, found on every continent except Antarctica, but

especially well represented at high altitudes and high latitudes.

The surface scidpturing of the seeds of Epiloljium has long been employed

as an important taxonomie character (cf. llaussknecht, 1884; Samuelsson, 1923,

* Tbe antbors are grateful to the U.S. National Scic^nce Foundation for a series of j^rants
to Peter II. leaven which made these investigations possible. The generous assistance afforded
Ihiven by Mr. W. S. Bertand, head of the Electron Microscope Section, and Mrs. Lesley A.
Donaldson, of the Ph>sics and Enpineerhi^ Laboratory, D.S.I.H., Lower Ilutt, New Zealand,
in carrying out preliminary SEN! studies of the seeds of some 75 species of Epilohium in
]^J(i9-70 is also gratefully acknowledged.

^ Department of Biology, Lewis & Clark College, Portland, Oregon 97219.

^ Botanical Hes(n\rch Listitnte, Private Bag XlOl, Pretoria, 0001, South Africa.

* Missouri Botanical Garden, 2345 Tower Grove Avenue, St. Louis, Missoiu'i G3110.

Ann. Missouni Bot. Gakd. (It: 18-47. 1977.

n>77] SEAVEY ET AL.— EPILOBIEAE SEEDS

19

1965

recent years with the scanning electron microscope (SEM). The regional or
more limited studies that have been concerned with the scanning electron mi-
croscopy of the seeds of Epilohhim are the following: Berggrcn (1974), Dcnford
& Karas (1974), Skvortsov & Rnsanovitch (1973, 1974), Raven & Raven (1976),
and Seavey et al. (1977). To a certain extent, these studies have provided back-
ground for the present more comprehensive effort.

In addition to the surface variations seen with the SEM, seeds of Epilol)i(vie
vary considerably in size and shape. It is the puipose of our report to evaluate
these seed characters against a background of other lines of systematic investiga-
tion, including cytology, moiphology, and biogeography. Included in this analy-
sis are more than half of the approximately 210 species of Epilobieae, represent-
ing the entire range of diversity; four of the seven species of sect. Chaitiaoieriou;
and all species of the reinaining sections of EpiloJnxim, as well as all six species of
Boisduvalia.

The seeds of all but one of the species of EpiJolnum possess a tuft of trichomes,
the coma, on their chalazal (distal) end, whereas none of the species of Boisdu-
valia have a coma. Although some variation exists in the color and relative
strength of attachment of the coma (cf. Raven & Raven, 1976), no diflercnces
in its ultrastructure have been detected and this feature of seed anatomy is not
dealt witli in the present report.

Materials and Methods

Seeds were collected from recendy grown garden plants or from herbarium
sheets. The coma was removed in most cases, but left intact if removal dama^^ed
the chalazal end of the seed. The seeds were then mounted on aluminum stuhs
with double-sticky tape, coated with gold/platinum in a Technics Hummer I and
examined with a Cambridge Stereoscan Mark 2A scanning electron microscope,
operated at an accelerating voltage of 20KV, at the Department of Pathology,
Medical School, ^Vashingt()n University. About 75 other samples were photo-
graphed in the same way in the Electron Microscope Section of the Physics and
Engineering Laboratory, D.S.I.R., Lower Hutt, New Zealand, and the results
of examining these photographs are likewise incorporated into the present paper.

Eour to six seeds of each sample were mounted and inspected. Little varia-
tion within samples was evident, and the seed photographed was in all cases
judged to be typical for the sample. Photographs were taken at approximately
60 X, 240 X, and 1200 X. The higher magnification photographs were taken near
the center of the seed, but since little variation is usually evident over the sur-
face of this seed, this practice was not strictly followed in all cases.

A list of specimens from which seeds were taken for the illustrations (Ti'^s.
1-210) in this paper is presented in Table 1. Voucher numbers are given in the
legends of Figs. 1-210 only when more than one collection is listed in Table 1.
Bars at the top of each plate indicate, respectively, 0.5 mm, 125 fim, and 25 uiu.

20 ANNALS OF THE MISSOURI BOTANICAL GAHOEN [Vol. 64

Tahlk 1. \^onchor iiifoniiation for species of BDisduvnIia and Epilohhini illustrated in
this paper. M and C numbers refer to garden plauliug lunnbers, used also on herbarium
Nonehers; R nuuibers are Raven eolleetions. All vouchers are deposited at the Missouri Botanical
Garden (MO), unless otlieruise indicated.

Boisduvalia rlcisto^attid Curran. U.S.A., Calif., Yolo Co., Cninipton 9222.

B. dcusiflom (Lindl.) S. Wats. U.S.A., Calif., Fresno Co., near Anberry, S^envey in 1974,

MI3().

B, glahcUa (Nutt.) Walp. U.S.A, Calif., Solano Co., near Eluiira, Crawptou 9219.

ii. macrmitlui Heller. l^S.A., Calif., Modoc Co., near Lookout, Scavctj in 1974^ Ml 37.

B. dricfa (A. Cray) Greene. U.S.A., Calif., Fresno Co., near Auberry, Scavcy in 1974, MI 33.

B. suhuhla (Ruiz & Pavon) Raimann. Cliile, Nuble, Cheese b Watson 4405, M158.

K})ih)])iitni aJpesire (Jae<i.) Krock. Switzerland, Canton Vaiid, seed from Conservatoire et

Jardin botaniciues, Geneve, 1972-1788, A/3/.
E. aisinifoliiun \\\\. U.S.S.R., Transearpathia, ?.kvortsov in 1968 (DS), R26179,
£. ojiinrense Hansskn. Japan, Pref. Yamanashi, Ha\akawa-cho, Deg\irhi in 1974, MHO.
E. ana<^aUidifolium Lam. U.S.S.R., V.. Chukotka, Iskat(^n Range, Kozhevnihov in 1972, M71S.
E. an^tistifoliuni L. subsp. eireuniva^uni Mosquin. U.S.A., Calif., Onion Valley, Ttcisselniann

5H25 (CAS).
E, atlantieitiH Litard. & Maire. Spain, Sierra Xe\ada, R2()166.
E. ])ehr!n<:ianuni TIausskn. Alaska, Kiska Quad., Buldir Is., Dick 414, M294.
E. hillardierianunt Srr. subsp. hiUardierlannni. Ke\w Z(»aland, Nelson, Bioekie CIIR199322,

M4.

E. eanuni (Greene) Raven subsp. ^arrettii (A. Nels. ) Raxcu. U.S.A., Utah, Washington Co.,

Zion N<itl. Park, Seavey in 1974, GSG3,
E. rannni subsp. eatntni. U.S.A., Calif., Santa Barbara Co., Santa Cruz Is., seed from Rancho

Santa Ana Botanic Garden, G844.
E. eanuni subsp. septentvionaJe (K(*ck) Ra\en. U.S.A., Calif., Humboldt Co., Trinit\ River,

Traey 5974.
E. eanum subsp. latifolitnn (Keck) Ra\i'n, U.S.A., Calif., Lake Co., Snow Mt., Seavey in

1974, CS07.

E. ehilense TIausskn. Chile, Prov. Cautin, near Uakc Icalma, Zollner 7868, M108.

E. ciUatnm Raf. U.S.A., Oregon, Josephine Co., near O'Brien, Seavey 1119.

E. eoloratum IMchler. Seeds from CopenhagcMi Botanical Gardens, A/.S(S'.

E. davuricum Fisch. CaiKida, Yukon Terr., Porsild 306 (UBC); U.S.S.R., S.K. Chukchi Pe-
ninsula, 1'r/r/.vce 6 Raszhivin in 1972,

E. dentietdatuni Ruiz ik Pa\on. Peru, l\uupa to Yamobamba, ca. 70 km F. of Trujillo, Conrad
2715^ M83.

E. dodonaei Vill. U.S.S.R., no xouchcr.

E. dnriaei Cay ex Godron. Spain, Puerto Wntaua, 0\ iedo, Merxndtller 6 Grau 21360, R26258.

E. ex(dtatnin Drew. U.S.A., Calif., Siskivou Co., Seavey in 1971, M409; Washington, Clallam

Co., Seavey 111 L M559.
E. jauriei II. Lev. Japan, Tottori Pref., YanuDnoto in 1970, M265.

E. foliosuni (Nutt. ex Tnrv. ik A. Gray) Snksd. U.S.A., Oregon, Douglas C:o., Raven 19089.
E. f^latienni Phil, ('hile, Prow Curico, Dc^pt. Curieo, Marfieorena, Matthei b Rodriguez 1, M68.
E. f^ttnnianinn Hausskn. Australia, N.S.W., New Fnglaud Natl. Park, Raven b Englehorn

25853,
E, hirsuluni L. U.S.S.R., F. Kazakhstan, Altai Mts., Beliauina in 1969 (DS), M58.
E. hirtuni Saumelsson, Pcmu, 20 km W. of Areciuipa, Averett 1004, M34L
E. Jiorneniannii Reichcnb. s. lat. U.S.S.R., E. Chukchi Peninsula, Yutisev b Sytin in 197L
E. konnirovia)iiini II. Lev. New Zealand, near Mt. Cook, Raven ir Engelhorn. CUR-199430

(MO).
E. latifolium L. U.S.S.R., Dist. Barguzin. Makeeva in 1971.

E. leiophyUnm Hausskn. Afghanistan, S. of Unai Pass, Breckle A2717, G352.
E.hiteiim Pnrsh. Alaska, Juneau, S/ir/nniV/r/ /ri 7.S.9/ (Gil).
E. niinutnni Lindl. ex Lebm. U.S.A., Calif., Plumas Co., llotvell 51156.

1977] SEAVEY ET AL.— EPILOBIEAE SEEDS

21

Table 1. (continued)

E. nesophilum Fern. Canada, St. Phillips, Newfoundland, Olscn in 1074, M121.
E. ncvadensc Munz. U.S.A., Nevada, Clark Co., Charleston Mts., Seavetj in 1974, G8()8.
E. nivium T.S. Brandegee. U.S.A., Calif., Lake Co., Snow Mts., Seavey in 1974^ GH6G.
E. nutans Schmidt. Czechoslovakia, W. Bohemia, Krai in 1967^ G23,

E. ohcordatum A. Gray subsp. ohcordatnm, U.S.A., Calif., Mono Co., Oneida Lake, Mc-
MiUan59-l (CAS),

E. ohcordaUnn subsp. siskiijoucnse Munz. U.S.A., Calif., Siskiyou Co., Sacramento River head-
waters, Pringle 1882 (NY).

E. ohscurum Schreb. England, Siurey, Raven 26092, M2.

E. oreganum Greene. U.S.A., Oregon, Josephine Co., near O'Brien, Seavetj 1117.

E, oregonense Ilausskn. U.S.A., Calif., Nh^ino Co., Rock Creek, Seavey in 1970, G350.

E. paliistre L. U.S.S.R., Chukchi Peninsula, Yitrtsev ir Raszliivin in 1972.

E. panicnJatum Nutt. ex Torr. & A. Gray. U.S.A., Oregon, Josephine Co., near O'Brien, Seavey

in 1975, M554.

E. cf. paueiflornm Samuelsson. Chile, on the pass of Loleo, ZoUner 6245, MHO.

E. pietum Petrie. New ZealaiKl, Tasnian \''alley, Raven ir Wilson 25617.

E. platy.stigmatosum C. B. Robinson. Taiwan, Ilualien Co., K. S. llsu 1714, G657.

E. pylaieanum Fern. Canada, Newfoundland, St. Stephen's, Ohen in 1974, M127.

E. ptjrrieholophuni Fr. & Sav. Japan, Nikko, seeds from University of Tokyo Botanical Garden,
'joehi, M73.

E, rigidum Hansskn. U.S.A., Oregon, Josephine Co., Seavey 1116; Calif., Del Noile Co., Curtis

1 (RSA).

E. scalar e Fern. Canada, Newfoundland, Highlands of St. John, Fernald 6- Long 28728 (Gil).

E. shirouniense Matsum. h Nakai. Japan, Pref. Yanianashi, Degtichi in 1974, M81.

E. stereophyllum Fresen. Kenya, Al)erdare Natl. Park, Raven 26164, M78,

E. stevenii Boiss. Iran, Azerbaijan Prov., Terme in 197 E

E. strictiun Muhl. Canada, Ottawa, Ontario, Carleton Co., Frankton in 1974, M65.

E. suffrutieosuui Nutt. U.S.A., Wyoming, Teton Co., Teton Natl. Park, Raven 26464.

E. treleasianuni II. Lev. British Columbia, near Bouff, Price 1900 (GH).

E. sp. Argentina, Estancia Moat, Tierra del Fuego, Moore 1686, M254; hke E. cunningliainii
Hausskn. except for pubescence.

Obserxatioxs

I. Epilohium

The results are pre.sented according to the sections recognized by Raven
( 1976 ) .

Sect. CordijJophonun (3 species):

Epilohiiim nevademc (Figs. 1-3), E. nivium (Figs. 7-9), and E. suffniticosian
(Figs. 4-6), all liave relatively large seeds (2.0-2.7 mm) with a prominent con-

striction toward the micropylar end. The seeds of E. nevadense are obovoid,
those of E. suffruticosum clavate, and those of E. nivium broadly obovoid. The
cells at the point of attachment of the coma form a distinct, although small, neck

region at the chalazal end.

The surface cells of E. nevadense (Fig. 3) are unique in shape. The center
of each cell is occupied by a thick, crater-shaped, apparently collapsed tangential
wall. The surface cells of the two other species are entirely convex, giving a cob-
blestone appearance at high magnification (Figs. 6, 9). With respect to seed
characters, E. suffrutieosuui resembles E. nevadense more than either resembles
E. nivium,

00

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol.. 61

Fk;uul:s 1-12. Scanning electron micrographs of seeds of Epilobiuin, sects. Cordylopho-
rttm and Xcwlohimu. — 1-3. E. rwvadcusc (Coydylophoruni). — 4-6. E. suffruticostnn (Conhj-
loj)honini),—7-Q, E. nivitiDi (CordyJopJionnn). — 10-12. E. pauictdatum (Xcrolobhun).

1977] SEAVEY ET AL.— EPILOBIEAE SEEDS 23

Sect. Xcrolobium ( 1 species) :

The large, broadly obovoid seeds of E. panicuhtum (Figs. 10-12) also have
a prominent constriction toward the micropylar end. The tangential walls of
the surface cells have a centrally situated convex prominence (Fig. 12). The
seeds of this species are similar to those of E. nivium. The neck region is incon-
spicuous.

Sect. Zauschneria ( 1 species) :

The seeds of four of the six subspecies of Epilohium caniun are presented in
Figs. 13-24. Resembling those of the preceding section, these seeds are largc^ and
broadly obovoid; and, like all of the species of the preceding two sections, they
have a constriction toward the micropylar end. The neck region is inconspicuous,
but, as in those of the preceding sections, distinct, especially when viewed from
a low angle (e.g., E. canum subsp. canum. Fig. 199). Tlie tangential walls of the
surface cells are entirely convex with little radial wall evident, except for subsp.
canum (Figs. 16-18), in which the tangential walls are convex but sunken within
prominent radial walls.

Sect. Chauiaenerion (7 species) :

The seeds of this section are small (1.0-1.8 mm) and relatively narrower than
those of the preceding three sections (Figs. 25-36). The neck region is evident
in dorsal view and is composed of irregularly constricted chalazal end cells
(e.g., E. latifolium, Fig. 209). The surface cells of the three species of subsect.
Leiostylae, E. angustifolium (Figs. 25-27), £. compcrmum Hausskn., and E.
latifolium (Figs. 28-30), lack convex tangential walls, and the major feature of
the surface is the irregularly polygonal reticulum formed by the radial walls

)]iu}n we (examined. E,

(Figs. 27, 30).

i

31-33

I and E. dodomei (Figs. 34-36) are of approximately the
same size and shape, but the surface tangential walls appear as raised, irregu-
larly compressed areas in the center of prominent regularly polygonal radial
walls.

Sect. Crossostigmu (2 species) ;

These small, obovoid seeds, blunt at both ends, are unique in the tribe in their
very finely papillose surface cells (Seavey et ah, 1977). The surface cells of E.
minutum (Figs. 193-195) are concave and very finely papillose over the tan-
gential walls as well as the prominent reticulate radial walls (Fig. 195), whereas
the surface in E. foliosum (Figs. 196-198) has isolated convex, smooth tangential
walls surrounded by finely papillose, elevated radial walls. The neck region in
both species is inconspicuous.

Sect. EpiJohiwn (ca. 185 species) :

Extensive variation exists in the size, shape, and surface architecture of the
species of this large section. Representatives of the species with the largest
seeds, E. rigidiim (Figs. 37-39, 40-42), may possess a prominent (Fig. 37) or
relatively obscure (Fig. 40) constriction toward the micropylar end, resembling
those of sect. Cordylophorum. The surface cells of this species are convex over
most of their surface, and the radial walls are mostly obscured (Figs. 39, 42).

24

AXNAI.S OF THE MISSOURI ROTANTCAL GAHOKN

[Vol.. 64

Figures 13-24.

13-15. E. canum siihsp.

Scanning electron microp;raphs of seeds of Kpilohium sect. Ziiuschneria.

E, cauum subsp. cami7u. — 19-21. E. camnn

[garret til.

16-18.

subsp. scpti'utriojiale. — 22-24. E. canum subsp. latifoJhnn,

1977]

SEAVKY KI AL.— EPILOHII'IAE SEEDS

25

LIES 25—36. Scanning electron micrographs of seeds of Epilohiu)}i sect. CJiamacneriou.
E. (Ui^ustifoliuDL — 28-30. E. latifolium. — 31-33. E. stcvcuiL — 34-36. E. dodonaci.
The first two species belong to sul:»seet. Leiostijlae, the second two to subscet. Rosviarinifolium.

Z.i .

26

AXXALS OF TMK MISSOl'RI BOTANICAL GARDEN

IVoL. 64

Fi(;uHEs 37-48. Scanning electron nu'crographs of seeds of Epilohhnii
—37-39. E. rv^idum, Scavcy 1116.— 40-42, E. riguluw^ Curtis 1. — 43-45
subsp. ohcordaliDii. — Jf> — 18. E. ohconlatmti sul)sp. siskitjoucti.se.

sect. Epilohitim,
E. ohcordatum

1977]

SKAVKY ET AL.— EPILOHIKAK SEEDS

27

Fic;uBEs 49-60. Scanning clrctron micrographs of seeds of Epilobiuni sect. E})ih)l)iiun.—
49-51. E. sJiiroutiicnsc, — 52-54. E. faurici. — 55-57. E. ])laty.stigmatosit))i. — 58-60. £. stereo
plujUiim.

2S

ANNALS OK TIIK MISSOIUI BOTANICAL GAHDKN

[Vol. 64

Fi(;uKKs 61-72. Scanning i^lcctmn niicroKraplis of South Anioricun species of Epilobhnn

sect. Epilohiiun.

— 70-72. E, sp.

61-63.

E. cf. pauciflorum.—di'QQ, E. dcnticulatum. — 67-69.

E. glaucum.

1977]

SEAVEY ET AL.— EPILOBIEAE SEEDS

29

FiGUHEs 73-84. Scanning electron micrographs of seeds of Epilohium sect. Epilobiuni.
73-75. E. liiriinn. — 76-78. E. orcgoncnse. — 79-81. E. coloratxim. — 82-84. E. leiophyllun).

This combination of characters is nniqvie in sect. Epilohium. Tlie seeds of E,
ohcordatiun (Figs. 43-45, 46-48) are similar in surface relief (esp. Fig. 45),
but they are smaller and lack a prominent constriction at the micropylar end.
The remaining species of sect. EpiJo])iurn have seeds which are smaller, lack

30 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 61

a luicropylar constriction, antl are characterized by surface cells which, although
often convex, alw ays haxe clearly evident radial walls.

The range in size of tliis remaining group of species extends from approxi-
mately 0.5 mm long for E, denticuhifum (Figs. 64-66), E. komarovhinuni (Figs.
121-123), and E, pictum (Figs. 124-126) to approximately 1.5 mm for E. cf.
paucifUmim (Figs. 61-63), E. pylairanum (Figs. 88-fX)), and E. strictum (Figs.
85-87). Tlie range in shape of this group of species extends from the obovoid
seeds of E. liirsiilum (Figs. 112-114) and E. obscurum (Figs. 109-111) to tlie
narrowly obovoid seeds of E. strictum (Figs. 8.5-S7) and E. shiwunwnse (Figs.
49-51). Ill overall shape, E. gunniamim (Figs. 118-120) stands out as having
an extcMided, flat ventral surface which extends beyond the outline of tlie main

body of the seed.

In many species the cells in the region of attachment of the coma proliferate
to varying degrees, thus forming a distinct neck. This neck is clearl) evident in
dorsal view in E. alsiuifoIhn}i (Figs. 145-147), E. anugaUidijolium (Figs. 148-
150), E. alhmtirum (Figs. 127-129), E. chilcmc (Figs. 172-174), E. ciliatum
(I'igs. 169-171), E. (Icnticuhlum (Figs. 64-66), E. hirtum (Figs. 73-75), E.
horncmannii (Figs. 136-138), E, oreganum (Figs. 160-162), E. oregonemc
(Figs. 76-78), E. scahrc (Figs. 154-156), and E. shiroumense (Figs. 49-51).
This neck may exceed 0.2 mm in E. davuricum (Figs. 94-96) and E. pijhieanum
(Figs. 88-90). Although the neck may not be seen as a distinct region in some
species when viewed from directly above with the seed lyiug flat, it can be dis-
cerned if the seed is tilted (e.g., E. watsonii. Figs. 157, 202).

The neck region may be a moderately thick extension of the chalazal end cells
(e.g., E. strictum, Fig. 203; E. exaUatum, Fig. 205) or it may be relatively thin, in
which case it is displaced toward the ventral side (e.g., E. ana gallidi folium, Fig.
204; E. orcgoncme. Fig. 206; E. scalarc. Fig. 207). In .species with the longest
necks, the neck cells appear individually elongated (e.g., E. davuricum^ Fig. 208).
The indi\idual trichomes of the coma are usually inserted at the very apex of the
neck, although they are occasicmally inserted over a broader area (e.g., E. palustrc^
Fig. 210). If the wvcV is relatively thin, as in E. cxaltatum (Fig. 205) or E. aat-

souii (Fig. 202), it is sometimes pellucid.

There are three t)pes of surface cells in sect. Epilohium: papillose (Group
I; Berggren, 1974), cells with a com^ex portion centrall)' located on the tangential
wall, \'anously shaped and isolated froiu its neighboring cells by a prominent
radial wall reticulum; fovcolate (Denford & Karas, 1974; Group III; Berggren,
1974), cells lacking au)- prominent feature other than the raised, regularly poly-
gonal radial walls; and ridged (Group II; Berggren, 1974), cells with a centrally
located coin-ex portion which is laterally compressed and fused with the raised
portions of neighboring cells, forming longitudinal ridges along the length of the

see

d.

M

)st species of sect. Epilohium have papillose seeds. The central papilla of
the surface cells of many species is irregularly compressed into a multisidcd
prominence as in E. alpcstrc (Figs. 103-105), E. amurcnse (Figs. 97-99), £.
oJjscuruni (Figs. 109-111), E. sliirouuicnsc (Figs. 49-51), and E. stereophyllum

1977]

SEAVEY ET AL.^KPILOBIEAE SEEDS

31

FiGUHKs 8.5-96. Scanning election micrographs of seeds of Epilobluni sect. Epilolnuni.
— 85-87. E. stnrftiw.—HH-9{). E. ptjlawmwm. — 91-93. E, daruiictnu, rorsild 306'.— 9 1-96.
E. duvuriiinn, Ynrfscv & Raszliiviu 'ni 1972.

32

ANNALS OF THE MISSOURI BOTANICAL GAR]:)EN

[Vol. 64

'.'S".' :S

Figures 97-108. Scanning electron micrographs of seeds of Eurasian species of Kpilohitmi

sect. Epilohium, — 97-99. E. amiircnse, — 100-102. E. pyrrirholoplium. — 103-105. E. alpcstie.
— 10(^-108. E. (hiriari.

1977]

SEAVEY ET AL.— EPILOBIEAE SEEDS

33

FiGURKs It)f)-120. Scanning (electron micrographs of seeds of Epilohiinu sect. Epilohhiui

109-111. E. o]}s('tini))i. — 1 12-114. E. hirsutuni. — 115-117. E. Inllardienanuni sulisp

InUanhcriuninn. — 1 l8-]2(). E. ^unnianum.

34

AXXALS OF THE MISSOURI B0TAXK:AL GARDEX

[Vol. 64

FicuHKs 121-132. Scanning clrctron niiciograplls of srctls of
12L-123. K. kontaiuvianinn.— 12 i-l2G. K. pirt urn.— 127-]2^).

Kpllohhnn sect.
E. athnitirmn

Epilohitnn,
—130-132.

1977]

SEAVEY ET AL.— EPILOBIEAE SEEDS

35

FuiUitLs 133-144. Scanning electron micrographs of seeds of Epiloljiuni sect. EpiJoI)iutn.
133-135. £. hehiinr^iannni,— 136-138, E. horncmannii s. lat. — 139-141. E. Juicum. — 142-

144. h\ trclcasiauum.

36

ANNALS OK TIIK MISSOUHI BOTANICAL GAHDEN

[Vol. 64

FiGUUKs 145-156. Scanning electron luieroj^raphs of seeds
115-117. E. ahinif()lium.—l4H-15{). K. (inagaUhlifoUum,-

of Epilohituu sect. Epilohitnn
-151-153. E. t}cso])JiiIu}u.—

15 l-15fi. E. scalarc.

1977]

SEAVEY ET AL.— EPILOBIEAE SEEDS

37

Fk;ui?e.s 157-168. Scanning electron micrographs of seeds of North' American species
t)f Ei)iJoJ)ium sect. Epilohium. — 157-159. E, waisonii. — 160-162. E. orcgaimm. — 163-165.
E. cxaltatum, A/-^Oy.— 166-168. E. cxaUatum, M559,

38

ANNALS Ol-' THE MISSOrHI 1U)TAM(:AL CAHDEN

[Vol.. 64

FinunKs 109-1T4. Scanning olectron niicrograplis of svvds of Epilohium sect. E})i}()])iuin,
-ira-171. E. rihiituni.— 112-17 \. E. chile use.

(Figs. 58-60), whereas in otliers it remains more regularly domeshapedj as in
E. (Icniictilatutn (Figs. 64-66), £. jauvici (Figs. 52-54), E. ghucum (Figs. 67-69),
and E. pijrricholophum (Figs. 100-102). In some species, the papilla is marked

by radial lines which appear as fine ridges, as in £. orcgonense (Figs. 76-78), jE,
pylaicannni (Figs. 88-90), some samples of £. hilhrdicrianum (Figs. 115-117),
E, chvuricum (Figs. 91-93), £. hUsulum (Figs. 112-114), and E. ncsoplulum
(Figs. 151-153). The papilla of some species, such as E. Icioplujllum (Figs.
82-84) and E. duriaci (Figs. 106-108), is collapsed in the center. The surface
cells are usually arranged in irregular rows running longitudinally the length of
the seed, but in some species, such as E. amurcme (Figs. 97-99), E. fauriei (Figs.
52-54), and E. ohscurum (Figs. 109-111), the rows are distinctly regularly ar-
ranged.

In fo\'eolate seeds the regularly polygonal reticulum formed by the radial
walls is the most prominent feature of the seed surface. This radial wall reticulum
IS relatively low on the seeds of E. gunniantnn (Figs. 118-120), and E, homarovia-
num (Figs. 121-123), but more pronounced on the seeds of E. ahinifolium (Figs.

146-147), E. amig(dlidifo]ium (Figs. 148-150), E. atlanticum (Figs. 127-129),
E. hcJnin<:^i(i)U(m (Figs. 133-135), E. h(}ii}C}}innuii (Figs. 136-138), E. lutcum
(Figs. 139-141), £, nutam (Figs. 130-132), and E. pictum (Figs. 124-126).

Ridged sc^eds, marl<ed by longitudinal rows ol laterally compressed, fused
ridges, are foimd principally in North American species. Among these arc £.
cilidfum (Figs. 169-171), E. cxaJtalum (Figs. 163-165), E. orcganum (Figs.
160-162), and E. icalsonii (Figs. 157-159). The South American E. chilense

1977]

SJ^AVEY ET AL.— EPiL()BII';AK SEEOS

39

Figures 175-186. Scanning electron micrographs of seeds of the foin- species of Bois-
duvalia sect. Boisdtivalia. — 175-177. B. dcnsiflora. — 17(S-1S(). B, mocnnitha. — 181-183. B

stricta. — 184-lSCl B. suhulata.

40

AXNALS OF TllK MISSOURI HOTAMCAL GARDEN

L\'()i.. 01

Fu.uiit^ 1S7-19S. Scanning electron niicro^raplis of seeds of the species of Boisduvalia
sect. Cuvrania and EpiJohhun sect. Cwsso.sti^nuL — 187-189. Boisduvalia clelsio^ama. — 190-
192. B,g/(//H'//</.— 193-195. Epilohium minutum.^h)e-m8. E. foUosum.

1977]

SEAVEY ET AL.— EPILOBIEAE SEEDS

41

Figures 199-210. Scanning election micrographs showing variation in chalazal end
of seeds of Epilobiuai, including the development of more or less pronounced beaks. All laxa
are incm1:>ers of sect. Epilobium except as indicated in parentheses following their names. —
199. Epilohium camim subsp. canum {Zauschneria) . — 200. £. rigicbim, Ctirfis 1, — 201. E.
obcordatum subsp. ohcordattim. — 202. E. watsonii. ^203. E. strictum. — 204. E. auanallidi-
folium. — 205. £. exaltainm. — 206. £. oregonense. — 207. E. scalarc. — 208. E, davuricuni,
YurtscD 6- Ra.szhivin in 1972. — 209. E. angustifolium. — 210. E. palustre.

42 ANNALS OF TUK MISSOURI BOTANICAL GARDEN [Vol. 64

(Figs. 172-174) likewise has ridged seeds. The seeds of liybrids between species
with ridged and papillose seeds are intermediate in varions ways and have been
illustrated by Skvortsov & Rnsanovich ( 1973).

II. BomJuvaUa

The seeds of Boisduvalia are distinct from those of Epilohhim. Tlieir shape,
which is irregularly angular-fusiform, is unique, as is the structure of their sur-
face cells. The relatively broad seeds of B. demiflora (Figs. 175-177) and B.
suhuJata (Figs. 184-196) have surface cells which are slightly raised, flat,
and irregularly polygonal. The seeds of B. macmntha (Figs. 178-180), B,
striata (Figs. 181-183), and the narrower seeds of B. cleistogama (Figs. 187-
189) and B. glabella (Figs. 190-192), have surfaces that are covered by an ir-
regularly striated reticulum composed of the unevenly joining radial walls of

the surface cells.

Discussion

A number of distinct seed types occur in the genera Boisduvalia and Epilo-
hiitm, as will be discussed below. In both genera, as in other Onagraceae, the
surface architecture is made up of the regularly repeated structures of individual
surface cells. Tlie characteristic forms of tliese surface cells are the result of dif-
ferential tliickening in their walls, as thin sections of the seeds of Epilohiiim (Ky-
tcivuori, 1972; Denford & Karas, 1974) demonstrate; seed coat patterns in many
other genera are comparable, for example in Cordylanthus (Chuang & Heckard,
1972), Melastomataceae ( Whiff in & Tomb, 1972), and Mentzelia (Hill, 1976), as
well as other inxestigations reviewed by Brisson & Peterson (1976).

There is a major contrast in Epilobieae between the angular-prismatic seeds
of Boisduvalia, which lack a coma, and the regularly obovoid, flattened ones of
Epilohium, which liave a coma in all taxa except in a few in which it has been
lost. In addition, the surface cells in Boisduvalia are irregular and thus unlike
any found in Epilohium, Those of B. densiflora (n= 10) and B. stdndata (n
19) are low, flat, and irregularly polygonal, whereas those of B. macrantha (n
10) and its probable aneuploid derivative B, striata (n = 9; Raven & Moore,
1965) are concave and have radial waills that are longitudinally striate and ir-
regularly thickened. These four species comprise sect. Boisduvalia (Raven,
1976); the remaining two species, which comprise sect. Currania (n = 15), have
seeds that reseml)le those of B. maarantha and B. striata, although they are
smaller.

In Epilohium, it is convenient to recognize seven distinct types of seeds, and
these will be discussed in turn.

1. Large, obovoid, constricted. In sects. Cordylophorum, Xerolobium, and
Zausahneria, the seeds are obovoid, large, and more or less prominently con-
stricted toward the micropylar end. Their surface is made up of cells which are
smooth and convex, with obscure lateral walls (Figs. 1-24). The relatively
prominent side walls in our preparation of E. aanum subsp. canum probably are
associated with shrivelling. The five species that comprise these three sections are

1977J SEAVKV ET AL.— KriLOIilEAE SEEDS

43

relictual (Raven, 1976), and tliis seed type is almost ccrtainl)' primitive for Kpilo-
Imun. In seet. Epilolxhan, with some 185 species and a worldwide distribution,
it occurs only in E. rigidum (Figs. 37-39, 40-42), a large-flowered, xeropliytie
species of the Siskiyou Mountains of northwestern California and adjacent Ore-
gon. The Siskiyou Mountains are a well known relict area (Whittaker, 1960),
and E. rigidum very likely resembles the basic stock from which the remaining
species of sect. EpHohhnn have diverged. As in E. nevademe, E. nivitdti, and E.
panicuJatum, the sid)tending l)raets in E. rigidtini are fused to their pedicel in
all but the lowermost flowers. The phylogenetic significance of this observation
remains to be determined.

2. Papillose. Most species of sect. EpiJohium have seeds that are smaller,
pai^illose, obovoid to narrowly obo\'oid, and lack a micropylar constnction. The
lateral walls of their surface cells are prominent. These include E. ohcordcduui
(Figs. 43-45, 46-48), the species most closely related to E. rigidum and, like it,
a large-fk)wered xerophyte of the western United States. In addition, the four
species of sect. Chamaenerion subsect. Rosnuirinifoliuni, exemplified by £.
dodonaci (Figs. 34-36) and E. stevenii (Figs. 31-33), have this seed type. It
seems clearly to have e\'olved from the first type and to have gi\'en rise in turn
to all of the other more specialized seed types within the genus, with the probal)le
exception of that found in sect. Crossostigma. The tribe Epi1obi(*ae seems to
ha\'e consisted initially of xerophytes, from which the more widespread and
numerous mesophytic and hydrophytic species were derixed. Perhaps an evo-
lutionary trend toward more numerous, smaller seeds accompanied the exploita-
tion of such habitats.

Seeds of this t\pe characterize more than a hundred species, including the
European E. alpcstre, E. collinnm^ E. duriaei, E. Jiirsutuin, E. lanceolatuni, E.
montanum, E. nervosum, E. oJ)scurum, E. parvijlorum, E. rosciun, and E. tetrago-
iiuni, as well as the circumboreal E. daviiricum and E. pcdustre (Sk\'ortsov &
Rusanovitch, 1974; Rerggren, 1974; this paper). Papillose seeds also occur in the
Asian E. amureme, (Figs. 97-99), E. fauriei (Figs. 52-54), E. phtijstigmulosum
(Figs. 55-57) and E. pijrricholophum (Figs. 100-102); the South American E.
dcnticidatum (Figs. 64-66), E. glaucum (Figs. 67-69), E. hirlum (Figs, 73-75),
E. cf. pauciflorum (Figs. 61-63), and one unidentified sj)ecies (South American
species # 1, Figs. 70-72); the African E. .stereophyUtun (Figs. 58-60); the Austra-
lasian E. hiliardicrianum (Figs. 1L5-117); and the North American E. coJorattun
(Figs. 79-81), E. orcgoncnse (Figs. 76-78), E. pijhicanum (Figs. 88-90), and
E. strictuiii (Figs. 85-87), as well as most populations of E. saxhnontautnn
Hausskn. Although onl\- one Australasian species of this first t)pe of surface
stmcture is illustrated in this paper, most species of this region have seeds of
this type (Raven & Raven, 1976).

Many of the microstructural details of papillose seeds are strikingly constant
from one sample to another. The convex portion of the surface cells of E. Jiirsn-
tum (Figs. 112-114), for example, are characterized by spirally arranged radial
lines, as illustrated by SkAortsov & Rusanovitch (1974; fig. IC). Similarly, the
surface cells of E. ucsophilum (Figs. 151-153), and the closely related E. pijJaie-
aniim (Figs. 88-90) and E. davuricum (Figs. 91-93) have characteristically

44 ANXALS OF TUK MISSOUHI BOTAXICAL GARDEX [Vol. 64

lowcomrx tangential walls with radial lines as illustrated in Skvortsov & Rnsano-
vitcli (1974: figs. 2E, F, and fig. 2G, respectively), although c()nsideral)le vari-
ability in surface structure within these two species is also evident in both studies
(e.g., E, (lavuricum, Figs. 91-93, 94-96).

Nearly all of the species with papillose seeds in sect. Epilobium, wvW over a
hundred, have the BB chromosome arrangement (Seavey & Baven, 1976; un-
published). Among the exceptions are the Asian E. fauriei, E. phtijsllgmutosum,
and E.. shiroinncnse, as wa^ll as most populations of the circumboreal L\ horncman-
nll s. lat. and the North American E. cJavatum Trel., all of which ha\ e the CC
arrangement. In addition, some of the species with the AA chromosome arrange-
ment, including the South American species of the group Dcnticulata (Samuels-
son, 1923, 1930) and the North American E. ghhcrrimum Barbey and E. hrcvisltj'
him Barbey, as wa^ll as E. saximontanum, also liave papillose seeds. Both the AA
and CC chromosome arrangements differ from BB by one reciprocal transloca-

tion, and we beliexe that each may be indt^pendently derixed from BJi. On the
basis of the evidence presented here, it appears that the common ancestor of each
of tliesc groups had papillose seeds.

Most species of Ilaussknecht's (1(SS4) group Palusfrifornies, including E.
(lavimnim (Figs. 91-93, 94-96, 208), E. palustrc (Fig. 210), and E. pylaiemium
(Figs. (S8-90) among those illustrated in the present study, have an exaggerated
beak at the chalazal end of the seed; althouiili others, such as E. strictum, w^hich

undoubtedly belongs to this gnmp, do not (Figs. 85-87, 203). The Palustn-
formes ha\'c the liB chromosome arrangement and papillose seeds in all taxa;
similar in these respects is E. oregonense (Figs. 76-78). EpiJolnum :^cahre, a
Newfoundland cMuhMuic that has been collected only once, has papillose seeds
and a pronu'nent beak also but does not resemble Palustriformcs in most respects.
Chalazal beaks are discussed in general on p. 30 and illustrated in Figs. 199-210.

3. Fo\'eolate. By flattening of the central papilla found in the surface cells
of the seeds of the prcx'cding group, foveolate seeds (Danford & Karas, 1974)
have evoKed. Such seeds, as viewaxl with a 20 X lens, have usually been de-
scribed as "smooth" in taxonomic papers on Epihlnum. Some species, including
the New^ Zealand E. ahinoides (different subspecies), have some populations
with papillose seeds and others with foveolate seeds. The same occurs in the
circumboreal E. liormnwinmi s. lat. and in the North American E. diwalum. It
appears, therefore, that the evolution of foxeolatc seeds has taken place repeatedly
within the genus.

In Australasia, w^here all species have tlie BB chromosome arrangement and
presun»ably evolved from a common ancestor (Raven & Raven, 1976), 31 species
have papillose seeds; 10, including E. komawviamun (Figs. 121-123) and E. pic-
fum (Figs. 124-126) have foveolate seeds; E. ahinoides, already mentioned, has
both papillose and foveolate seeds in different subspecies; and 4 species, includ-
ing E. grnmiamim (Figs. 118-120) have a distinctive seed type that will be dis-
cussed below. In almost e\ ery one of the 11 species in which foveolate seeds oc-
cur, these appear to ha\e evolved separately from papillose seeds (Raven &

Raven, 1976).

In North America, foveolate seeds occur in EpiJohium Juteiim, a rather iso-

1977] SEAVKY KT AL.— I-l'ILOBIEAK SKIDDS

45

latcd species, and in some populations of die £. ^landtdostDti complex (AA).
In addition, they are characteristic of some or all populations of several entities
in die CC <j;ronp: E, andgciUidijolium (Figs, 14S-L50), E, heJniu^ianum (Figs.
133-L35), £. clavalum, and E. horncmanmi s. lat. (Figs. 136-138). The seeds
of E, trcJcasianum may he papillose (Figs. 142-144) or foxeolate, owing to the
hybrid nature of E, trclcasianuDi, a series of populations of hybrid origin ])et\\een
E. hiteuni (seeds foveolate) and odier species (Seavey, in preparation). l\ipil-
lose seeds app(\u' to be dominant in a genetic sense over foveolate ones.

Among die sptx'ies diat occur in Europe, only species of die CC (Alp'niac
group and diree other species — E. nulans ( HB in part; Figs. 130-132), E. alhnili-
ciim (AA; Figs. 127-129), and E. alshiijolinm (AA; Figs. 145-147) — have foxe-
olate seeds. Judged from die widc^ morphological gap between the latter two
species and the diderent chromosome arrangement in the first, it is likcK' that
all evolved foveolate seeds independently. The fact diat all diree are species
of low stature that occur in alpine habitats, like the sp(^cies of grouj:) AlpUuic
here defined to compiise only CC species and therefore to exclude E, (dsnil-
foUum), suggests that some common selective force ma)' favor foveolate seeds
under such conditions. The third Furopean species with the AA chromosome
arrangement, E. (dpestrc (Figs. 103-105), has papillose seeds.

4, OboN'oid-patelliform. Four Australasian species — E. ginviiantnn (Figs.
lhS-12()), £. curtisiae Raven, E. uillisii Haven & Fngelhorn, and E. angtisltini

Clieesem.) Raven & Engelhorn — have distinctive seeds with a holhjw^ rin<j;
around their adaxial side, which are thus patelliform. The first three are prob-
ably closely related, but the occurrence of similar seeds in E. an<j^tisluni, ap-
parendy closely related to E. kouiarovianum, has not ])een explained satisfactorilx
(Raven & Raven, 1976). The seeds of E. uillisii and of man\- popuhitions of E.
^tninianuiii are finely papillose, those of the other taxa foveolate. Tlie oboxoid-
patelliforni seed type undoubtedly e\'()lved from the mor(^ frequent papillose t\pe
wn'thin Australasia.

5. Irregularly reticulate. Tlie three species of sect. Chanmencrion subsecl.
LciosfyJac have seeds in which die very thin radial walls of the epidermal cells
of the seed coat h)rm an irregularly polygonal reticulum (Figs. 2.5-30). The
four other species of sect. Cluunacncnon^ comprising subsect. RosifKiritufoliuni,
ha\e papillose seeds that resemble those of most species of sect. Epdohiuni in
size and shape. The close relationship of the tw^o subsections of sect. CJunnacuc-
rioti is beyond (juestion in that they share the following um'qne or highl}- unusual
characteristics for the geiurs: all leaves spirally arranged, flow^ers zygomorphic,
floral tube obsolete, pollen shed singly. This indicates uneciuivocally diat the
unusual seeds of subsect. Leiostylae, stressed by Skvortsov & Rusanovitch (1974)
and by Rrisson & Peterson (1976) as an argument h)r die generic distinctness of
sect. Cliamaencnon, represent instead an evolutionary .sp(x*ialization within this
group, odierw ise characterized by seeds similar to diose of many species of s(X't.
Epilolnum, Also implied by the seed morphology of subsect. Rosmarinijoliuni
is die divergence of sect. Chamacncrion from an ancestor that would be placed
widiin sect. Epilolnum long after die divergence of sects. Cordijhphorum, Xcwlo-
hium, and Zausclnwria from species such as E. rigiduni. Additional evidence for

45 ANNALS OF THE MISSOUlU HOTANICAL GAHDEN [Vol. 64

the close relationship of sect. Chamaenerion with sect. Epilohium is summarized

by Raven (1976).

6. Ridged. \Vitlun the North American group of species with the AA
chromosome arrangement (Seavey & Raven, 1976), there has originated a dis-
tinctive seed t\pe, described above, tliat doubtless delinu'ts a phylogenetically
coherent group of ta\a. Illustrated here are E. ciliatum (Figs. 169-171; extends
to Japan), £. exaltalum (Figs. 163-165), E. orcganum (Figs. 160-162), and E.
watsonii (Figs. 157-159). Epilohium ciliatum occurs as an adventive in Europe
and in Australasia (Raven & Raven, 1976), and its seeds have been illustrated
several times (as £. adenocaulon Ilausskn., Troughton & Donaldson, 1972: pis.
103-104; Skvortsov & Rusanovitcli, 1973, 1974; Rerggren, 1974; Raven & Raven,
1976), Virtually identical seeds occur in the South American E. chilcme (Figs.
172-174), this suggesting recent immigration to South America following tlic
origin of tlie group in North America. \Mtliin the Epilohium glandulo.sum com-
plc\\ fo\'eolate seeds also occur, but these have probably been derived from
ridged ones. The most primitix^e species of this group is apparently a local en-
demic of bogs in the Siskiyou Mountains, E. orcganum, which lias exscrted,
deeply 4-lobed stigmas.

Denford & Karas (1974) have interpreted the ridges in seeds of this type as
being formed of rows of flattened papillose cells flanked on both sides by fo\'eo-
late cells. Our SFM observations, however, show clearly that the ridges are in-
stead formed of the finlike central portions of individual surface cells, all such
cells on the abaxial side of the seed being similar. No fovcolate cells were ob-
scrxed on the abaxial surface of these seeds.

7. Finely papillose. The distinctixe seeds of sect. Crossostigma, described
on p. 23 and by Seavey et al. (1977), cannot easily be related to any other seed
type in the genus, and the relationships of the two annual species included in this
section are obscure.

LnKUATinit; Citkd

Bi:iu:(;iu:\, C ll)7t. Seed morphology of some Epilobiitni species in Seaiulana\ ia, Sxensk

Dot. Tidskr. 68: Kil-lOS.
Biussox, J. D. & R. L. Pktersox. 1976. A eritieal review of the use of scanning electron

microscopy in the study of tlu^ seed coat. Proc. Workshop on Pi. Sci. Applications of the

SEM IIT Res. Inst. 1976 (Pt. MI): 477-495.
CuuA.vr:, T. & L. R. IIec;kaud. 1972. Seed coat morphology in CordijlauUius (Scrophulari-

accac) and its taxononiic significance. Arner. J. Bot. 59; 258-265.
Denfoud, K. E. & I. Kauas. 1974. Some ol)senations on tlic seed coat structure witliin the

genus Epilohi\n)\. Kxperientia 30: 1 144-1147.
IlAussKNKCiiT, C. 1SS4. Monographic der Catding F.p\lo])iuni. Jena, Germany, viii + 318

PI), -f 23 pis.
IIu.L, R. J. 1976. Taxonomic and phylogenctic significance of seed coat microsculpturing

in Mmtzclia (Loasaceae) in Wyoming and adjacent western states. Brittonia 28: 86-112.
KY'rrtvuoiu, I. 1972. The Alpinac group of the genus Epilobhnn in northernmost Femio-

scandia. A morphological, taxononiical, and ecological study. Ami. Bot. Fenn. 9: 163-203.
MuNZ. P. A. 1965. Ouagraceae. North Amer. Fl. II. 5: 1-278.
Ran-kx, P. II. 1976 (1977). Generic and sectional delimitation in Onagraceae, tribe Epilo-

bieae. Ann. Missouri Bot. Card. 63: 326-340.
& D. M. Moom*:. 1965. A revision of Boisdnvalia (OnagraeeacO- Britloin'a 17:

238-254.

& T. E. Ra\k\. 1976. 4'lie genus E])ilohi\im (Onagraceae) in Australasia: A

19""] SKAVKV ET AL.— EPILOBIEAE SEEDS

47

systematic and evolutionary study. New Zealand Dept. Sci. Industr. Res. Bull. 216:
1-321.

Samuklsson, G. 1923. Re\'ision der sudamerikanischen Epilohiiun-Avtcn. Svensk Bot.
Tidskr. 17: 241-296, tab, II-V.

1930. Zur EpilohiuiU'Floni Siidanierikas. Svensk Bot. Tidskr. 24: 1-11.

Si-AVEY, S. R. & P. H. Raven. 1976. Chrouiosonial evolution in Epllohhnn sect. Epihhium
(Onagraceae). PL Syst. Evol. 24: 6-12.

Wright & P. H. Raven. 1977. A comparison of E})llo])unn minutuui and E.

foJiosMm. Madrono (in press).
Sk\()hts()v, a. K. & I. I. Rusanovitch. 1973. Hybrid cell structures of the seed coat of
Epilobium. Priroda (Moscow) 12: 89-93. [In Russian.]

& . 1974. Scanning electron microscopv of the seed-coat surface in E.pilohiuni

species. Bot. Not. 127: 392-401.

Tiu)U(;mt()n, J. & L. A. Donaldson. 1972. Probing Plant Structure. Chapman ^ Hall,
Loiulon. 116 pp.

WniFFiN, T. ^ A. S. ToMM. 1972. The systematic significance of seed morphology in the
neotropical capsular fruited Melastomataceae. Amer. J. Bot. 59: 411-422.

WmiTAKKH, R. H. 1960. Vegetation of the Siskivon" Mountains, Oregon and C:alifornia
Fcol. Monogr, 30: 279-338.

THE BTOSYSTEMATICS OF CALYLOFHUS (ONAGRACEAE)^

ITOWAIU) K. TOWNF.H-

AliSIHACT

The gunus Calylophtis (Onagracoao), a sogn'gate of Oenothera, was stiulifcl in reference
to sjstemalie relationships, I)ree(ling systems, iiolh'nation, and cytology. Six species are
recognized, Umv in s(^ct. Salpiu^ia: C, iuhicula, C, Jiarfwcfiii, C. tounuyi and C. lavmuIuUfolnis,
and two in sect. Cahjloplins'. C, bcthindicri (fornicrl) C drunimondianus) and C. scntdatus.
Several changes in nonienclatnre and rank arc made. Crosses performed among species
di-monstrated strong barriers to hybridization between the two sections of the genns and
slight to moderate barriers among species ^^'itln*n sections.

Populations of Ctdijlophus are distributed througli tlie Great Plains, tlu^ southwestern
United States, and northern Mexico. The various taxa occ-npy distinct habitats which range
from xeric sites in the Chihnahuan Desert to relati\ely nusic pine and pine-oak forests. In
most forms of the genus, the plants are suffrutesccnt perennials and occup\ calcareous soils.
One form is rcshictcd primaril) to gypsum soils.

Cytological in\estigations showed a remarkable degree of translocation heterozygosity in
natural populations of Ccdyloplnts. Translocations wcvv observed in all taxa, with 75% of 183
plants (excluding C. scrndatns) exhibiting hetcro/xgosity for at least one translocation.
Nunu^rous plants were hetcro/\gous for more tlian one translocation, and the mean inunber
per platit was 1.3. CcdyJopluis scntdatus is a complex structural hcteroz}gote and all indixiduals
oI)ser\ed were heterozygous for at least fixe or six translocations. Iljbridization experiments
with C. bcrhndicri suggested that C. scrruhitus maintains structural h) briditx with gametophytie

lelhals in pollen and eni1)r)0 sacs.

The basic cluomosonie number of the genus is x == 7, Tetraploidy was observed in
indix'idnals from 5 of 62 populations of C, Jiartwc^ii that were exanu'ned. A few plants of
several taxa possessed diminutixe chromosomes ranging from 1 to 11 in number. The most
freciuent ohserxalion was of a single dark-staining pair in addition to the normal complement.

Chromosome observations of hybrids shoxved profoimd intersectional dilferences in
structure, primarily froin translocations. I'ranslocation differences an^ also marked among
populations in sect. CalyJojdius, but are slight among the taxa of sect. SaJpiugia.

The breeding systems of CalylopJius are varied, xxitli C. scntdatus self-eompalible and
highly aut<igamous and the other species self-sterile. Cahjlophus hcrJandicri and C ttd)icula

have short floral tubes, strong ultraxiolet contrast patterns, niatinal anthesis, and arc visited
by a xarict) of diurnal insects. Anthesis of the remaining members of sect. Salpingia occurs in
the afternoon or evening. These plants possess long floral tubes, variable* ultraxiolet contrast
patterns, and are xisited by splu'ngid moths and crepuscular bees in numbers that x ary from
localitx to localitx'.

Biosxsteinatic studios of tlie gtums CaUjJophus were begun in 1967, sliortly
after it liad become apparent tliat tbis genus constituted a natiual group no less
distinct from Oenothera than from other genera of the tribe Onagreae (Ra\'en,
1964). This paper is based on those sttidies and on an exanu'nation of exttuisive

* This stud) xxas initiated as a doctoral dissertation at Stanford Unix-ersity (Toxvner, 1970b),
and subseqnentlx ampHficd. I would like to express great appreciation to Peter 11. Haxen, at
xvhose suggestion this research was begun. His adxice and generous assistance have been
inxaluable throughout the course of the study. He, D. E. Breedlox^e, and D. P. Gregory
prox'ided both material of Calylophtis and unpublished information on pollination. Sharon
Stexxarl gave me indispensable assistance in the field, laboratory, and greenhouse. Dan
Holmes and Judith Ljnch assisted me in the field. Steven and Ann Seavey were most helpful
in the maintenance of cultix'ated plants and the handling of herbarium material. John IT. Tliomas
directed the processing of herbarium loans. The illustrations of floxxers were draxvn by Julie

Spran/a.

The follow iug indixiduals kindly supplied Dr. Raven and me xvitli material of Cahjlophus

or xxith information regarding localities: L. C. Anderson, D. S. Correll, W. P. Emery, J. C.

Ann. MrssouHi Rcrr. Gauo. Gl: 48-120. 1977.

1977] rOW'SKH- CALYLOPUVS

49

herburiuni material. A preliminary publication stemming from tliis researcli
Towner & Raven, 1970) was bas(^d on a less complete study of herbarium ma-
terial and some of the taxonomic decisions made then are changed in the present
paper.

The genus Cahjiophm is distinguished from the other genera of the Onagreae
by the following suite of characteristics: a peltate stigma which may be discoid
or nearly sc^uare in shape, sometimes with 4 shallow, broad lobes; microsporog-
enous tissue divided into packets in the locules of the anthers; yellow flowers;
and a many-seeded capsule. In the opinion of Raven (1964), Cahjiophm is most
closel)' related to those genera of the Onagreae which share the feature of di-
\'ided microsporogenous tissue: i.e., Gaum, CUirkia, Ilelewgaura, and perhaps
Ilauya. The genus occurs over much of the CIreat Plains, extending into the
mountains of the Great Basin region and other parts of the Southwest, and also
reaching southward to the Mexican Plateau. The area of "[reatest di\'ersitv for
the genus includes West Texas, southern New Mexico, and north-central Mexico.
Poi:)ulations are usually colonial ar»d wideK^ scattered, often occurring in dis-
turbed areas. Habitats occupied by the species of Cahjlophus are typically some-
what xeiic plains or hills wath soil that is often calcareous. Plant associations in
which the various forms occur range from creosote bush scrub in the Chihuahuan
Desert to pine forests of several types.

The history of the genus Cahjhphws has involved several transfc^rs at the
generic lexel as \'arious authors have seen fit to separate the group from Oenothera
or to unite the two genera. Traditional treatments of the species here considiMed
have placed them in Oenothera subgenera Salpmgia and Cahjlophus or in the
respective genera Galpimia and Meriolix. Hafinesqne (1S19) was the first to
distinguish a species of what is now Cahjh)phus from Oenothera^ although his
publication, lacking a description of the genus, was invalid. The name Meriohx
was xalidated only later, by Endlicher, in 1840. Cahjlophus, the first generic
name oF legitimate publicaticm, was presented by Spach (1835a) and emended
by him to ''CahjU)phis' without justification in the same year (Spach, LS35b).
It thus has priority over Meriohx, Galpinsia^ and Salpingia at the generic le\'el.
Most of the nineteenth century treatments of the species of Cahjhiphus retained
them in Oenothera. In the late nineteenth century and until quite recently, the
generic names MerioUx and Galpinsia were frecjuently used for the various species
of Cahjhphus in treatments such as those of Ilellcr, Small, and Rydberg.

Fislicr, W, C. Jackson, W. M. Klein, R. L. MeCircgor, H. Marshall, T. Mos(iuin, D. R. Parni'll,
and the lalc L. C Shinncrs. Pollinating insects have been most graciously identified b\ W. F.
Hair, J. Hnrns, H. W Daly, G. Eickwort, and R. W. Thorp.

I am grateful to tlie curators of tlie folltjwing herbaria, from which loan material \\as
secured for the study, or which were \isited b> the aiitjior: ARIZ, CAN, CAS, COLO DAO

DS, F, CII, K, KAXU, KSC, LA, LL, MO, XEB, NMC, NY, OKL, OKLA, PII, POM, RM, RSa'
SMU, TFX, UC, l^XM, L'S, WTU.

Financial support during the course of this uork was supplied by Xational Institutes of
Health predoetoral and postdoctoral fellowships award(^d to me and by research grants to
Peter H. Raven from the National Science Foundation (1906-1970). Laboratory space and
greenhouse facilities were pro\ided by the Department of Biological Sciences, Stanford
Universit)'.

M^epartment of Biology, Loyohi Maryrnount l^ni\ersity, Lo>()la Boulevard at West 80th
Street, Los Ang(^]es, California 90045.

50

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Voi,. 64

During the nineteenth century the explorations and collecting of Nuttall,
Wright, Lindheinier, lk>rlandier, Ilartweg, Fendler, James, Cockerel!, and
Touniey provided t\pe material for most of the presently known taxa in the
genus. All taxa now considered valid, except Cahjiophus harhvegii subsp. mac-
cartii and C. tuhicuJa subsp. siii<^ulosus, were described and published between
1817 and 1898. Over this period and during the early twenHeth century a num-
ber of names were proposed for minor variants, especially by Leveille, Small, and
Nelson. These contributed additional confusion to the taxonomy of the genus,
whicli in any circumstance would have been difficult to intei-pret. The situation
was greatly improved with the publication of Munz's (1929) revision of these
species, a treatment which brought together the information available at that
time and gave relatixe order to the taxonomy of the group. Many superfluous
names were reduced to synonymy and for the first time decisions were based on
adequate herbarium material. Some of the species boundaries remained unclear
even then because of the complexity of variation in sect. Salpin<;ia, an erroneous
appraisal of the type of C. harticcg,u, and die absence of information on breeding

systems in the C. serruhtm group.

Ravens (1964) paper, which brought diis group of species together as the
genus Cahjiophus, was closely followed by the pubHcation of Simmers (1964).
Shinners presented a taxonomy similar to that used here, except that the facts
concerning breeding systems in C. serrulatns were not yet known, and that some
differences in the ranking of taxa exist between our treatments. Other differ-
ences include my recognition of C. hartwegii subsp. fendleri and a revised inter-
pretation of C. hartwegn subspp. hartwegii and fiUfoIius. In 1965 Munz pub-
lished a monograph of the North American Onagraceae (Munz, 1965) in which
the species of Cahjiophus were referred back to Oenothera and in which specific
status was granted without justification to several of the entities presented by
Munz as subspecies in 1929. Otherwise die taxonomy remained the same as in

the earlier publication.

A brief taxononiic treatment of this group by Towner & Raven (1970), the
basis of my account for the Manual of the Vascular PlarUs of Texas (Towner,
1970a), considered the species as belonging to Cahjiophus. The present work
generally retains die same bomidaries between taxa used in that paper. Excep-
tions include two changes in rank, the recognition of C. tubicula subsp. strigulo-
sus, the reduction of C. australis to synonymy with C. serrulatus, and tlie changes
in the names for the outcrossing members of sect. Cahjiophus. All of the fore-
going changes were found necessary after a more thorough study of herbariiun

material was completed.

///

tcrmincd to be well differentiated from C. hartwe<;ii and thus deserving of specific
rank. Cahjiophus serrulatus is here judged to be possibly of multiple origins,
and the series of populations once referred to as C. australis is now thought to be
distinguished from other such series only by its geographical separation from
the range of the rest of the species. Lastly, C. herlandieri was found to be the ap-
proi)riate name for the entity treated by Towner and Raven as C. drummomlianus.
Type material of Drummond's was determined by an examination of pollen fer-
tility to belong with the complex structural heterozygote C. serrulatus.

1977] TOWXEH— CALYLOP/ZL^S

51

Cytology

Mciotic chromosome configurations were determined to establish the frequen-
cies of structiu-al and numerical changes in natural populations of CaltjlopJius.
Further^ the cytology of experimental hybrids was observed in order to describe
chromosomal differentiation among taxa in the genus. Previous reports of cy-
tology in Calylophtis (as species of Oenothera) include those of Hagen (1950),
Lewis et al. (1958), Gregory & Klein (1960), and Kurabayashi et al. (1962).
These authors reported translocation heterozygosity, tetraploidy, and extra
cliromosomes. P. II. Raven (personal communication) later established the pres-
ence of complex structural heterozygosity in the genus. Those phenomena were
confirmed and their taxonomic distribution described in the present study. In-
version differences among populations were also found. Table 1 and the sys-
tematic accounts of tlie various taxa combine my findings and the results of
earlier investigations.

Translocation heterozygosity is extremely common in the genus and was
found in all taxa. CaJyIophus serruJ(itus\ with its system of complex structural
lieterozygosity, consistently forms rings of 12 or 14 chromosomes at meiotic
metaphase I. All other species form mutivalent associations regularly. Excepting
tetraploids and C. serrulatas, 130 of 183 plants (71%) had visible translocations,
and in this sample the mean number per plant was 1.3. The most frequent num-
ber of translocations per plant was 1, but individuals with 2 or 3 were com-
mon. In one observation from C. harticegii subsp. filifolius and two from C.
berlancUeri subsp. ])erj(indieri^ plants were heterozygous for 5 translocations.
Directed alternate disjunction seems to prevail since translocation heterozygosity
does not appreciably lower pollen fertility in natui'al populations. Heterozygosity
may well be maintained in populations by the action of heterosis, deleterious re-
cessive genes, or even balanced lethals, especially in taxa such as C. berlandieri
where translocation heterozygotes comprised 857^ of all plants examined. In its
high frequency of naturally-occurring ti'anslocations, Calijlophus resembles some
species of Gaura (Raven & Gregory, 1972b), Clarkia amoerui (Lehm.) Nels. &
Macbr. subsp. amoerui (Hakansson, 1942), and some populations of Clarkia un-
guicuhta hindl, (Lewis, 1953) among the Onagraceae.

Tetraploidy was observed sporadically in some subspecies of Calylophus
hartwegii, but nowhere else in the genus. Seven populations of C. hartwegii
subspp. hartwegii, puhescens, nuiccartii and intermediates among them were
tetraploid (/i— 14). These plants all showed a combination of bivalents and
multivalents in meiosis. No taxon was found to be wholl)^ tetraploid, and any
evolutionary significance is probably minor. Tetraploid individuals appear to
arise within populations from time to time.

Extra, diminutive chromosomes were found in several taxa from both sec-
tions of Calylophus, and could well occur in all of the taxa. They are small dark-
staining bodies that often appear to be heteropycnotic and tend to proceed
through first meiotic prophase and metaphase more rapidly than the normal
chromosomes. Pairing between diminutives was common, and possible associa-
tions with the larger chromosomes were seen occasionally. The most frequent ob-
servation consisted of a single dinu'nutive pair in addition to the normal comple-

52

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

3

a
O

c

o

_2

O

05

o

o

o

U

CQ

>

O

S

«5

o

O

>.

C3
O

S3

";3

^

1^

+

C

6

+

O

bc

o
bo

+

O

c

o

bc

+

CN

a i%

-^

+

o

Tf CO 50

4 #^

CT

2-C +

+ 1C

» r^

t- CO I-

to

en
CO

bjD

CO ^ Ol

8

a;

o
U .

o u

2

o

+

00

o

bo

I H^

+

o

bC

+

« rt

^

'T C TT

K 00 O

OC CO ^

O _Q

UuO

O

C
O

o

rt

K^^H

O

CO

bo '

V3

CO

o

bc

be

(M

+

+

C

CD

* #x

00

■ #\

S

o

bc

;^

00

^T3

CM

+

CD

O

bC

C

°co

+

- +

c , . ^

Mh VI .S

01 +

+ ::

^ bC

bO ^

CO

*C b£ bC

+

"^

-^

CM

• wy

o

g

3

■^^

rn +
o _

C/3 C

^bC

CM

+ +

-S o ^

•^ be o

.S bc

+ ^ .s

+^

fM —(

+ + +

C
CM

« r\

CO

*

C

mO

fM

• W\

bc c

c .5

-CT3 +

CM

bC bt bC

c G q

;C t-M -f

CM •— I '-t CM ^ ^ CM

+ + + + +^^+ + +

CM

+

CO

• f%

o bc o

bC.S b£,,

C cv. -f^ -^-

Q.TTCOiC'^'— iCMcOl-l^cOlOlO

Tf

iO

^

CO

I

00
fM

c:7>

^

CO

CO 1:^ I

O CO CM

b£.

CO ^1

00 05

o

CO

00

o^

5

hi

CO

CO
<M

be

e

o

o

X PQ

H

o
U

o

13
U

o

dU^U

« c« OJ

o

o

u

a;

OffiKU^

s

o
o

JO

o

z: (D c

QJ O

O

o

o

v^

o

o

CO
<M

Co
3

d

en

CD

o

U

o

£

C/3

8

rt

o

o

bC

+

t

« r\

'T fM

So

be be

fM a.

+ +

CO

'o

3 .5

c

0)

1977]

TOWKEU—CALYLOrilUS

53

03

1-1

CD

o

o

o

U

->%

^

en

+

CO

^ ^ ^
o o ^

tJ} b/) tjD

■S -S -S

'C 'C '^

0-. 0-. CM

+ + +

IC lO CO

05

o

-5

o

o

iO

+

o

o

c

+

00

+

* ^\

^ +
t* ^

^ ^
C ^

CO ^

hJ Pc;

CD

-^

p::

^ ^

+

CO

o g

-^

ccf^

^

CO

CO O

c

tJ5

PQ

O °

"^ ^ -^

O mh o

c^ ^ ;. o ^ ^

c c c ^ fl -^

^ Jh ^ "^ \^ 0-.

* r\

^g>

o

cr^

Dl CO

+ o

^ be

CO

o

bO
C

0-.

CNl

+C++++++++ ++

pc;

bC

+ +

'— — ' ^^ --^ 1^-^ ■— ' --^ ^^ ^ ^^ ^^ ^^ l^-i H-i ^-1 HH

t-^t-incO'-Hiocoicicco Tfiot^cNt^ ^^

CO
CO

CO

2

o

t^ t^ t^

;^ CO CO CD

00 CO
CO
CO

V4 t^

^
^

c

s

CO b^ OvJ

CO CO b-

CO

CO

^

o

S5

^

-5

;^ ;i<i ^i

lil

O

O

O

bO

O

2^ CO

CO
CO ^

CO ^

eg

o

tZ3

H

O g :-

-^ v^ X

O cfl <^

C PQ PQ

o
U

o

c;

c3

O

U .^ S. 6

o

pq pq WO W

o
U

o

n3

U

o

03

c

u

O

U

^ ^

«3

c

o

o

^

^

c

^

;/5

o

5

+

bO

r-!

+

CO

o

eg

+

o
bo

+

ic

^

<M

o

U

(D

C CO

o a

'^

o

bC

+

CO

• rv

'i^

- So

bC bi!) bCi tx bD

c c c

)H ^ 1-H

ei

+ + + + +

in CO in io io

CO

CO

CO

Cyn"^ CO 0> ^

Cij ^ lO T-i CM

K Cs| C\l i^ i^H

. o

O (J

p « U U o

^ > a S
• J5 c« ^

8

o

O

o

gUOOc;5 H

54

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

o

c4

t

O

S

8

a
o

0)

o
U

>s

o

O

h4

^

0-.

+ +

CO

\o

» r\

* r\

TT

CD

W5

C

S

^ O

bO

o

bo

s

c

+ + +^

■ #t

ic

-^ l^ lO 1>

05^10
b- 00 00 00

O
S Th

^5

t

o

+

o

be

+

o ..

bc"^

bC

«

t> l> lO t-

CO
CO

I

I

00
CO

§ CO b.

a:

o

u

o

52 tu

o

c

o

c«

O
00

CO

on

bJD

O

I

o

0)

O

CD

O

bc

+

o

^ .^. ^ ."^ ^

o c o

bC bT bC bC U)

9 .S .S .S ■£
c 'C *n *n 'C

cq

oq

+ + + + +

C CM CO lO CO lO

b^

CO

CO

CM

00 boot) <N

c

3

CO

CO
CO

ED

O

U 6

o
U

o

O

58

bfl

c

■ ^^

;-<

+

*

W)

^

C

1 ^^

M

»^H

C

n3

• ^H

f^^

^^

C

J

*^H

' *

n3

o

+

+

o o

eg

+

+

CO

4 f\

If)

S C

•5 O ,

CO I

CT5
»0

05

>^ Oti
O CO

^^

o

>

^

en

3

be

3

be

3
1

O

o

s

o

be

a

+

lO

* #%

a

^

^

^

+

CO

+

o

be

o

be

I B^

o

bO

c

+ +

Tf

lO

+

+

lO c -

T bC^ (^

cfl fl ..V C

•^ ^ C/; P

c ;

(N

-r T a ,

c t- I- I- t- t- i>

00 O

CO CO

CO ^

be '-^ "^

fN

Oi CO

(M Oi ^ ^

K3

in

be

;3

O

bo

g<U

Q

1977]

TOWNKR— CALY LOP nUS

55

a

o

>

o

a;

o

<

o

>.

^

o

bC

+ -^

en

a

o

x<

^

10 +

• ^

t- s: 1-- in

o

CO

CO
CO

CO

^ 10

CO "^

Q

^

^
^

O

CQ CQ CQ

o

m

3

CM

o
be

c

+ +

I— I

^ CO ^

• »N

fl + o

■3 - bO

t-

in

» r^

* r\

t-

i- 1

CO

10

' — <

CO

00
CO V5

a -CO

a
•^

o
SI

"3 p

•^5

s

H

H

v>

CM

o

b€

ty:

+

IC

« r^

03

fM

^ o

^ on

bi) be

CM

+

+ •'

CM

bf

6

+ +

t-- IC CO I- I

CO

CO

05

;::!

05

^ >t^ ^ <M CO H

S CO ^ CO CO S

-5 ^

^

.'^^

■U c
^- U .2 U

X t;^ ^ C/^ >

0)
G

^ ^ f^

O

U

N

[yi
C

cd

a*
o

00

3

v:i

bo

O

rf^

be

CM ^ o
+ be bJj

- .3 -S

01

CO

• t-,

+ +

C no CO

I-

K --H 10

^ -CD O

OvI CO ^

00

CO

CO

o

be

g

+

o
be

+

Ol

* #\

^

o
be

C 10

^ 00

O

CO

o

be

+

O

n^

o
be

o
be

+

01 =: 10

10

CO

o

be in

2

CVJ

CO

bl,

--i

•

u

.

u

C U

-. G 0) J-"

TO ^ 0^ qj

na
chise

Santa C

ew Mexi

Chaves

C/1

^ § g

CD '"-' '^

63 6 6

03

S; c

pq UU

<

2

H

56

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

n3

8

W

<

o

c

O

O

X

O
O

CO

^ >

o

•S P^

cq +
+ S

CD .- _

o
c

I ^H^

+

CO

* ^

^

Pj o . o
•S bc ti to

.£ 03 .S

Tt^

+ + M +

O

bo

^lO

' +

Tf IC C lO 1- I- lO

^ ^ iQ

CO £^ >f

CO 2 <^

CM "^ <M

Si

tlD

tit CM

^ S^ ^ o>

N c »^

s ^ c

3 » =:

00
00

1 ^

O

U

U

G O _
001:1-;

— a> o ^

o
U

S

i; CL. clh H :>

c
o

o

CD

bc

O

c
3

c>q
o

o>

J

o

be

+

ic

•>

2

-'^

CO

s

.S2 t

3 2

^

o

^

o
U

i-i

c

^

o

be

+

10

CO

10 ip

CO "^

CM H^

:^

o

-Sri J^

o--

+

05 P

CO

Si o

US

^

ffi

O

o

C*5

CO

^'^

fe c

be

c«

(J 3

a;

s

o

3

u

a;

o

O
>

o

o

U

u

0)

o
>

^

C/3

CM

C

bO
C

I-

+ +

C I

Id

00

CO

■^ JO

CO

5

o
2U

o
U

o

o
o

£, o

a;

O

Un o

t/3

'ft

o

bc

I ^pkH

+

10

CM

O

U

M)77]

TOW KER^CALY LOP IIUS

57

■y^

^
O

03

o

c

o

o

in

I

+

CO

o

+

^O c

"^ U)

c

ti . -r

^^^

T ,

^ s +

t ^ ,^

b£^ o

c^ '"'"• X

•r C ^

r^ ^-£

^-^ ^^

-[- *C cv.

- ^ o^

* ^1

l^ lO CO

o

CO

<:

o
U

a;

U

V

Co I—.

c3

O

ro

Crs

^

o

5

CO

o

o o

S -S

-w

o o

CO

CO CO

£3

_2;
O

V5

5
o

CM

Tf ^

O

o c

c c ^

0) :; oj o

b

c

■

CO

05

u

^

-fj 5

<c

ol

'JH

:y^

-aJ

c

"^ o "^ "^

o , o c

c r c c

^ -H oi CO

CO 00 CO CO

O Tt

c
U

o

U

u

o

U

cq

o

c

o

+

■ rv

-^ -^ -^ ^ ^ ^

wJ^ *-J^ wJ_,

£ .E .£

cj a; a;

o

bi

CD

O C

bJC

rr*

^

ZJ

r^ r-.

o

be

G

I ^H^J

c3 ?i

^' a a

ca

a

c

c

O

bfj

in

bjo

ooocot^ocoioccm

^

o

U

o
in

a;

a>

UQ

o
o

c
U

•G S

C

U

o

C

U

-^

'. >j

-u

MO hJ

bfj

o

O

o
U

a;

bc _ ::;

o •£ rt

P: G hJ

O

fN

O

bC

C

w

1-

o ^
bC +

S -

• r^

-^

^

S c
a bi:

a>

G

a;

c

CO

CO

^ CO

cc;

o
U

C

H

58

AWALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

a;

I

a

8

o

CO

Si

o

3

PQ

o

I

U

>.

o

oi

a

•c

CO

S

+

-^

« n

TT

Tf

^

03

CM

+

bc

c

V5

cd

rt _l_

o «

0)

t^ 00 CO ^O ^ '^
CO 00 CO I- In. CO

cd

f

o

bc

V5

o

o

:3i

JO

be

O

CD

O

bx)
•S

+

-^

• PI

Tj^ TT

O C

if- ^
bJj bc

s ^
+ +

CO CO

CD t-

CI) o

leron Co.

hran Co.
:son Co.
agorda Co
Patricio C
ling
3rara Co.

•

o
U

Can

izon
Coc

^

o

5

"3

o

&5

Co
O

bo

Co

d

3

O

o

Wl

Csl

bc

G

+

CO

c^

a.

O

o
>

2

en

G

O

3

o
a

c:i

3
3

a

O

0)

^

O

VJ

G

rf

^ ^

^

O

bO

G

I ^1^

O

c

I M^^

o o
be be

G ,C

o

bO

C

C/5

t

+ +\.++ +

G

10 t^ 10 c ic »o

IC I- I

» T

«

CO

^

IJO

:3

CO

W5

CO

-S

(M

•5^

10

r^

CO

3

CO

be

CO t^ 00

3

CO

CO

c

CO ^

bC ^

o .

o o

S £i S a

>< pq ex, D

US

n3

Si

I

PC^

CO

.9

o

O

C

PQ

C
13 —

O

»o

5^

en

G
O

II

IS

4 ^

00

'^5

5 « CJ
p (u *-

f #k

E

O

O w n

1977] TOWNKW—C ALYLOPIIL'S

59

meat. Niinihers of extra, diminutive chromosomes ranged from 1 to 11, but tlie
numbers were liiglily variable witliin populations and seemed to vary even
among separate determinations from a single plant. The extra, diminutixe
chromosomes of Calylophus differed from those reported in Gaura (Cregor)- &
Klein, 1960) and in Oenothera (Cleland, 1951, 1967; Cleland & Hyde, 1963) in
often being heteropyenotic; pairing of the diminutive, extra chromosomes was
frequent in all tliree genera of Onagraceae. The extra, diminutive chromosomes
reported by Ostergren (1947) in Anthoxanthum were heteropyenotic like those
in Calylophus. Cleland (1951, 1967; Cleland & Hyde, 1963) has hyptjthesized
that the extra, dimitiutive chromosomes in Oenothera hookeri Torr. & A. Gray,
may have been derived following h)'bndization between this species of Oenothera
and an entity belonging to another group of the genus. This appears doubtful
since it has not been demonstrated that the chromosomes of the different groups
of Oenothera differ significantlv in size, or that their differences are or would

be maintained in hybrids. Certainly, it would be veiy difficult to construct an
analogous hypothesis for similar chromosomes in CahjJophus and Ganra.

In Calylophns, supernumerar\ chromosomes of normal moipholog)^ were
found only in one population of C ])erhinclieri subsp. herJandieri {Towner 140),
in which plants had one to t\\o extra chromosomes. These resembled super-
numeraries as found in Clarkia (Lewis 1951; Ilakansson, 1949), Camissonia
(Raven, 1962), Gaura (Gregory & Klein, 1960; Raven & Gregory, 1972b), and
Gayophytum (Lewis etal., 1958).

Inversion differences were encouiitered only in some experimental hybrids
between C. ])erlandieri subspp. hcrJandicri and pinifolius, and also in some
crosses between C. hartwe<^ii and C. hivandtdifohus. No evidence of inx^rsion
lieterozygosity in natural populations was obtained, but it could well occur as
an infrequent event.

Chromosomal determinations from experimental hybrids indicated that the

taxa within sect. Salpin<j^ia ha\'e become differentiated b)' no more than 3 re-
ciprocal translocations and, rarely, b\' an inversion. Among C. tuhicida and all

subspecies of C. hartwe^ii, crosses showed cither complete homolog)' or 1 to
2 translocation differences. Cahjlophus lavandtdifohus differed from the above

by 2 to 3 translocations and an inversion. The two su1)species of C. herlandieri,
on the other hand, have become strongly differentiated, based on the current
evidence. Hybrids between them were heterozygous for 2-6 translocations and
sometimes for 1 inversion. Moreowr, geographically separated populations of
C. herJandieri subsp. herJandieri sho\\Td differences of the same magnitude.
Reciprocal crosses of C serrtdatus and C. J)erJandieri produced hj'brids with
3-6 translocation differences. Lastly, the few determinations from intersectional
hybrids between C. ttdncuJa and C. J)erJandieri demonstrated differences of at
least 5 translocations. In the hybrid plants meiotic chromosome pairing was
variable and poor, while anaphase movement was irregular. The high degree
of sterility seen in these and other intersectional hybrids was probably derived
from such chromosomal causes. In contrast, hybrid pollen fertility was moderate
to high for most intrasectional crosses. Lower hybrid fertility in certain intra-

60

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. (54

>.

s

o

a;

o

a;

n

d

S5.

Cfl

3

PC

§

O
C

3

o

lo ^ »c u:)

O ^ GO o
CT) CO ^ C^

^ vtH lr^ 'n*

CO »o

CO

CO

lo

05

I

;5j

:5i r^

Q> <U c^ ^

^ § -^ -^

O ^ ^ ^

^ ^ ^ ^

^ c ^ ^

c^ CQ c^ c^ ^: y:

|u UU

CO
05

CSJ

tic

o

5

-5 '^

-5-?)^

o

5

UU

^^-i;

Q

CM

1^ w

O

CO

o ^ U

-to-t:

o c

-J,

Co

GO 00 CO
C^ "^ K

05 CO o)

O '^ ^

CO

CO
05

^ ??

CJ ^

§^^

CM

^ E-^ (J (J c^

CO

-^-5

kJ

o

U 00

00

5

8p

o
o

0)

>1

o

JO c3

UN

N

O

o
c

u

Q

si)

qj

o
U

a;

c

03

'5 6

•nu

o o
UU

o

U

o

<

in

c
U

U; ;/: N

■^ '^^ ^t:^

o
U

o

^6

en

03

o
U

W5 03

r 6

2

C

U

a;

E

o

;5 ^.

1^

(LI

c

U
o

o

c

>

^ o
6U

X

H

«

o
U

03

X

c

U

>.

^ ^

•= P

« CD

C 0!

X

a;

o
U

03

W5

X

6

U

03

c

11>

O

m

o
U

o
U

t/3

X

CD

S S 6

U

03

c

g

C

en

d

^

;=

CO

00

C

CO

a.
d

in

51

CO

;2

^

t^

^

<i>

2

t

-fc

t3

a

2

U

U

I

o u

2

3

Oh
en

d

CO

3

3 ^
O 3

CO

CL*
0)

I

o

u.

s

to

a
o

a

ID

S

8

"T

I/)

CO

u

c/1
O

6

V W V ^ V

o o

1977] TOWyilLK^CALYLOPHUS

61

sectional crosses involving C. hvandulifoVius and some involving C. hcrhndicri
was correlated with cytological differences between the parental plants.

Floral Biology and Pollination

Information on floral biology is biused on the field work of D. P. Gregory
(1964 and personal commnnication) and P. II. Raven (personal communication),
and on my own field work and study of cultivated plants. Collections of flower
visiting hksects were made, and these are to be deposited at the California Insect
Survey, Berkeley. Determination of ultraviolet reflection and absorption pat-
terns was carried out by photography with black and white film under near-
ultraviolet illumination. All taxa of Cahjlophus were tested for self-incomx^ati-
bility by making repeated attempts at self-pollinatioTL Table 2 lists those
populations which were tested.

The breeding systems of Cahjlophus are of tln'ce basic types. That of C. scr-
ruhtiis is based on complex structural heterozygosity. In this .species the flowers
are highly autogamous, often self-pollinating before anthesis (Fig. 6). For this
system iiKsect visitation and pollen transfer are uimecessary, and are in fact un-
connnon in C. serrulatus. The other two types of systems involve self-sterility
and insect pollmation.

In sect. Saipingki, with the exception of C. tuhicula, flowers are adapted for
vespertine pollination by hawkmoths. They have narrow floral tubes measuring
25-50 mm in length (Figs. 1-3), sweet-scented nectar, large exserted stigmas,

vx^spertine or afternoon anthesis, and, except for C. toumeiji, strongly ultra\n'olet-
reflective areas on the distal portions of the petals (Figs. 7, 10). These reflective
areas on the petals contrast markedly with small ultraviolet-absorpti\^e regions
which are usually present in the center of the flower. Populations of the tubular-
flowered taxa experience vespertine and nocturnal visitation by hawkmoths,
especially the abundant and effective pollen vector llijhs Uneata (Fabr.) {Cc-
lerio Uneata). Some taxa and populations within taxa of C. hartwegii tend to
have mid-afternoon anthesis and large- to moderate-sized ultraviolet-absorptive
areas in the center of the flower. These seem to be secondary adaptations for
bee pollinatioiL Many populations of taxa in sect. Salpingia are visited in the
afternoon, early evening, and even morning by halictid, anthophorid, and otlier
bees. Some of these are oligolectic for the Onagraceae and probably contribute
to pollination. The late-opening subspecies of C. hartwegii (Figs. 9, 10), alon^^
with C. lavanduIifoUiis and C. toumeyi, have ultraviolet patterns less apj)ropriate
for bee pollination. The last species has no strongly u]tra\iolet-reflective areas on
the petals (Figs. 11, 12). In these taxa bee visitation is restricted to the early eve-
ning and is probably less effective in pollination.

Cahjlophus tubicula and C. herlandieri are pollinated by matinal and diurnal
insects. Their flowers open in the early morning, have short funnclform tubes
(Figs. 4, 5), and display large ultraviolet-absorptive regions in their centers (Figs.
13, 14). The petals are highly ultraviolet-reflective distally. Based on the available
evidence, the primary insect visitors to C. tubicula seem to be morning-active
halictid bees. Other potential pollinators include hawkmoths, which visit the
flowers lightly at about sunrise.

62

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Figure 1. Calijloplms toumciji in the Chiricahiia Mountains, Cochise Co., Arizona
( Towner 107),

A great N'uricty of insects eome to the flowers of C. herlandieri, which appears
to possess a generalized pollination system. Beetles, skippers, small butterflies,
occasional hawkmotlis, and a wide variety of bees have been observed gathering
pollen or nectar. Each of these groups may contribute to pollination, with its
relative importance varying with locality.

When present, ultraviolet patterns on the flowers serve to direct insects to
the anthers and nectar. The pattern typical of Cahjiophus consists of absorptive
regions at the base of the petals, at the mouth of the floral tube, and on the stigma
and anthers (Figs. 8, 10, 14). Cahjiophus toumciji differs in having moderate
ultraviolet absorption over the entire flower (Fig. 12). In Onagraeeae, ultra-
violet absorption results from the presence of one or more flavonoids \\ith ab-
soiption maxima in the near-ultraviolet (Dement & Raven, 1973, 1974). As in
many other yellow-flowered Onagraeeae, the ultraviolet-absorptive portions of the
petals of Cahjloplius hortwe<in and C. scrrulatus, and presumably those of the
other species as well, contain isosalipurposide, a chalcone with an absorption
maximum at 365 m/x (Dement & Raven, 1973, 1974, and personal com-
munication). In addition, C. luirtwe^ii has an accompanying flavonol, myrecetin-
3-glucx)side or galactoside, and C. serrulatus has an unidentified compound that
resembles an aurone, a class of flavonoid that is frequently associated with chal-
cones in Asteraceao (W. Dement, personal communication). All of tliese
flavonoids are absent from the idtraviolet-reflective portions of the petals. Carot-
enoids absorbing maximall)' at 400-470 m^ are found throughout the petals.

1977]

TOWNER— CALYL0i7/U.S

63

®

X.

N

■^

''^■.^rr^

©

a -

FiGUHKs 2-6. Longitudinal sections of flowers of Cahjlophus. — 2. C. Jiarltccf^ii su1)sp.
liaHwegiL — 3. C tounicyi, — 4. C. tuhinda subsp. txihicuJa, — 5. C. hcrhmdicri sul)sp. pinifoJius.
— 6. C scrrulatus. All X 1.1.

Tlierefore, insects witli trichromic vision would see the petal ajiox color as "bee
yellow" and the petal base, anthers, and stigma as "bee pmple."

It is inicertain whether these ultraviolet patterns would be as effective for
hawknioth pollination after sunset as they are for dim'nal insects in sunlight.

The small size or absence of ultraviolet absorbing areas in the evening-opening
CalyJopJuis perhaps signifies that such pattens are not useful for insect orienta-
tion at night. Alternati\ ely the minimization of ultraviolet contrast patterns ma)'
ha\x^ evolved to lessen pollen and nectar removal by diurnal insects inc^ffieient
in pollinating the vespertine Cahjlophus. This could be especially trne in the

64

ANNALS OF THE MISSOURI BOTAXICAT. GARDEN

[Vol. 64

Fic;uhl:s 7-12. riowcrs of Calyloplius under fluuiescent and ullra\ iolcl illuinination.
Collections refer lo seed sources. — 7. C. batiwcgii subsp. hanwegii (Durango, Mexico, BrcccUovc
1 1305 A); fhiorrscent illumination, emasculated. — 8. Same, with ultraviolet illumination. — 9.
C. Junttvcgii subsp. fcudlcri (Presidio Co., Texas, Jackson iu 1964); fluorescent illuuu'nation. —
10. Sauie, with u]lra\iolet illuruiuation. — 11. C. toumciji (Cochise Co., Arizona, Towner 107);
fluorescent illuminatitjn. — 12. Same, with ultrax iolct illuuu'nation.

case of C. toumciji, which lias flowers contrasting only slightly with its vegetative
parts and the baekgronnd in the nltra\'i()let region of the spectrum. Mazokhin-
Porslinyakov (1969) states that c()nsideral)le ultraviolet light is present at night,

especially in moonlight. However, the moths investigated liave shown low sensi-
tivities to ultraviolet light, and become functionally colorblind at light levels
below 0.05 lux. Their maximum scMisitivity to light is in the yellow to green

1977]

TOWSEH—CALYLOrilUS

65

FiGUiiEs 13-14. Flowers of CalylopJius bedandicri subsp. bcnIancUeri (Ilaitloy Co., Texas,
Robe lis 35) under fluorescent and iikrav iolet illunnnation. — 13. Fluorescent illumination. —
14. Ultraviolet illumination.

region of tlie spectrum. If hawkmoths follow this pattern, they could locate
Cahjiopluis flowers merely by their scent and r(^llectivity at wavelengths greater
than 480 lu/x, eliminating any need for ultraviolet reflection. With the moderate
light levels present at dusk both bees and moths could probably utilize the idtra-
\'i()let contrast patterns for orientation.

Piiyl(k;kxetic RKLAnoxsiui'S

The species of Cahjlophws fall into two clearly recognizable but related groups.
Those with keeled sepals and two sets of stamens of unequal length are included
in sect. CaUjlophus and those with plane sepals and subequal stamens are as-
signed to sect. SalpingicL No intermediates between these sections have been
found, and the species of the two groups are intersterile while being relati\'ely
infertile among themselves. Fifty-one of 101 iiitersectional crosses j'ielded some
seed, with the amount set ranging from 5 to 81% of normal^ but germination was
poor (1-3%) and the few weak hybrids that grew to maturity were largely stenle

(0-10

%

pollen fertility). Members of sect. Salpingia are most likel)' primitive

in comparison with those of sect. CaJyJopluis^ as they have a clui»iped perennial
habit and, except for C tuJnaila, are primarily pollinated by hawkmoths. Tn
growth fonri, tlie members of sc^ct. SaJpimiia r(^semble the niorc^ gencM-alizc^d forms
of Gaum (Raven & Gregory, 1972a), a genus which is also primarily moth-

pollinated and closely related to Cahjiopluis,

Among tlie hawkmoth-poUinated members of sect. Salpinfj^ia, C. Javamhili-
folius seems to be the most distinct, ha\'ing a more caespitosc* habit than the
othc^r forms and a broad geographical distribution which includc^s numerous
relict populations. CahjJophus iinnneiji appears to be most closely related to the
southern subspecies of C. Jmrtwc^ii, especially C. Iun'twc<i,ii subsp. hartue^iU
with which it sliarc\s long free sepal tips, prefc^rc^nce for montane habitats, and
strigulose pubesccMice. Within C. hartue^ii a pattern of reticulate rc^lationships
is evident and rc^lative affinities are difficult to assess. Cahjlophus harlwc^ii
subsp. maccartii. occup\'ing tht^ Tc^xas coastal plain, is likeh' a rc^ccMit dc^ri\-ati\'e of

QQ ANNALS OF THE MISSOURI BOTANICAL CAHDEN [Vol. 64

C. Jiurtwa^ii subsp. harUvegii, as these two forms share several cliaractcrs and
arc geographically adjacent. Calylophus hartwegii subsp. puhescem^ C\ hart-
wc'^ii subsp. fcnilleri, and C. liartwcgii subsp. fiJifoliiis seem to constitute a series
of related forms occupying slightly different ecological zones in the plains of the
southwestern United States. The subspecies of C. hartwegii are connected by
zones of intergradation, although not all of the possibilities for geographical con-
tact and gene exchange are realized. The short-lived perennial C, tuhicula is
unique in sect. Salpingia, having several characteristics of bee-pollinated llowers.
It is likely a specialized deri\'ative of relatively recent origin.

Sect. Calijlophns is apparently more specialized in having an annual or short-
lived perennial ha]>it, and perhaps also in its adaptations to diurnal pollination.
Of the two species in the section, C, bcrlandicri is clearly the more generalized,
having a self-incompatibility system, and chromosomes which form bivalents or
small rings at meiotic metaphase I. Cuhjlophus serrulatus^ a highly autogamous
complex structural heterozygote, is a specialized offshoot of either the former
species or a common ancestor. No clear knowledge of the exact ancestry of C.
serrulatus is a\ ailable, although hypothetical lineages have been worked out for
some of the other complex heterozygotes in the Onagraceae. It may be that this
species developed through the independent origin of complex structural heterozy-
gosity in different geographic regions. This is suggested by the fact that neighbor-

<

ing populations of C. serrulatus and C. ])crlatulieri often tend to resemble one
another phenetically, a condition which could be attributable to multiple an-
cestry for C. serrulatus^ to introgressive genetic exchange, or to parallel responses

by the two species to local selectiv^e regimes. Calylophus serrulatus is highly suc-
cessful and widespread in the Great Plains from Canada to New Mexico and
also occurs in Arizona and Mexico. Calylophus herlamlieri is largely restricted
to Texas.

A likely but speculative evolutionary sequence could have begun with the
primary di\ergence of the ancestors of the two sections. Accompanying this
split was the de\clopmcnt of morning anthesis and other characteristics of day-
pollinated flowers in sect. Calylophus, The moth-pollinated perennial ancestors
of sect. Salpingia probably evolved more slowly, but gave rise to C tuhicula.
Calylophus serrulatus presumably developed very recently from C. herlamlieri
or its ancestors. The course of evolution within sect. Salpingia seems to have

depended upon ecogeographical differentiation among the various forms, while
being little dependent on cytologieal transformations. Cyclic climatic fluctuations
in Pleistocene and Recent times have undoubtedly contributed to the taxonomic
differentiation within sect. Salpingia, judging from the confusing patterns of over-
lapping geographical ranges, isolated populations, and introgression that are
now in evidence among its taxa.

Divergence within sect. Calylophus has been marked by much greater cyto-
logieal change than in sect. Salpingia. Several chromosomal rearrangements
involving translocations were requir(xl for the derivation of C. serrulatus from
pair-forming outcrossers. Similar chromosomal rciDatterning has also occurred
within C. herlamlieri^ and characterizes at k^ast three series of populations. In
addition, the two races of C. herlamlieri ax^pc^u" to be ecologically differentiated,

1977] TOWNER— CALVLO/'/Zt'S

67

and remain well separated in regions where there exists a diseontinuity Ix^tween
their respective lia1)itats.

Taxonomy
Calylophus Spach, Hist. Nat. Veg. Phan. 4: 349. 1835.

'^Cahjhphis' Spach, Noiiv. Ann, Mus. Hist. Xat. 4: 337. 1835.

Meriolix Raf. vx Encll, Gen. PL 1190. June 1840: Raf., Amum". Montlilv Mag. <^ Crit. Rev. 4:

192. 1819, num. nucl.; Raf., J. Phys, Chini. Hist. Nat. Arts 89; 259. 1819, nom. nnd.

TYPE: Meriolix scrrulala (Nutt.) Walp. = CaJijlophiis scnulalws (Nutt.) Raven.
Salpin^ia (Torr. & A. Gray) Raiiu. in En^4er & Prautl, Xatiirl. Pflanzenfani. 3(7): 217. 1893.

Based on Oenothera subg. Salpiugia Torr. & A. Gra>, Fl. N. Amer. 1: 501. 1840. TYrE:

Oenothera lavandulaefolia Torr. & A. C;ra>' = Calijloplius lavanduUfoVms (Torr. & A.

Gray) Ra\en.
Calpinsia Rritton, Mem. Torrey Dot. Club 5: 236. 1894. Rased on Oenothera subg. SaJpingia

Torr. & A. Cray, Fl. N. Anier. 1: 501. 1840. type: Galphisia liarttregii (Benth.) Brilton
Cahjlophti.s liartwegii (Bentb.) Raven.

Herbaceous to suffrutescent perennials, rarely annual, from a wood}' eandex,
flowering in the first year. Stems nearly prostrate or decumbent to erect, with
grey to pinkish brown epidermis, tliis sometimes exfoliating. No basal rosette,
leaves cauline, more or less sessile, alternate, entire to spinuose-serrate, tlie up-
per leaves more or less uniforui in size, the lowermost often soniewdiat larwr;

stipules absent. Flower 4-merous, aetinomorphic, borne in the axils of the npp<M-

leaves, opening in the early morning or from inidaftcrnoon to near sunset, with

the stigma receptive and anthers shedding pollen simidtaneously upon anthesis

or soon afterwards, wilting in V- to 2 da\s; buds erect in inflorescence. Floral

tube w^ell developed and prolonged be\ond the oxary, deciduous after anthesis.

Sepals greenish N^ellow, often with purple or red n^u-kings, reflexed separately.

Petals yellow, in some species becoming red, orange, or purple upon wilting,

reflexed in anthesis. St}le yellow; stigma }'ello\\^ to yellow green, occasional!}'

blue black in one species, peltate, discoid to n(\irly s<|uare, sometimes obscurel}'

and shallow4y 4-lobed. Stan^ens 8, yellow; antliers narrowly elb'ptic to lineai',

versatile, the sporogenous tissue divided into packets witliin each locule; poJIen

yellowy shed singly. Capside many seeded, sessile, c}'lindrical, and often nar-
row^ed at each end, obtusely 4-angled, longitudinalK dehiscent, persisting on the

stem after dehiscence. Seeds in 2 rows in each of the 4 locnles. Basic chromosome
number, x — 7. Fi\'e of the six species are self-incompatible.

TYPE species: Cahjlophus nuttallii Spach — C scrruJatu.s (Nutt.) Raxen.

In the accounts which follow^, the taxa will be gionped according to then
phenetic affinities. Specimens cited were selected to represent tlie ranges of
morphological variation and geographical occurrence. Where possible, I ga\e
preference to recent collections and those with numl)ers of duplicates. M\- own
collecticms are deposited in the Dudley Herbarium of Stanford University (DS)
with duplicates to be distributed.

Key to Sectkjxs

a. Sepals plane, lacking a keeled midrib; stamens subeqnal Section I. S(ilpi)igia

aa. Sepals witb eonspicuonsK keeled midrib; stamens biseriate, tlie cpisepalons filaiuents

about twice as long as the epipetalous filaments .. Section II. CahjIopJms

53 ANNALS OF THE MISSOURI BOTANICAL CAHDEN [Vol. 64

Key to Species

Section I. Sal}}}ugia
a. Floral tiil)e fnnnclfonn in tlir upper hvo-thirds or more, or less than 15 mm long;

flowers opening near sunrise - 4. C. iuhicuia

aa. Floral lube fumielform in upper half or less, 15-55 mm long; flowers opening near

sunset.

b. With conspicuous axillary fascicles of small leaves, these up to 30 mm long;
sul)ulate sepal tips 2-9(-12) nun long; capsule thin walled and dehiscent only in
the distal portion; montane distribution, nortinvestern Mexico to southeastern

Arizona — - — - 3. C. touwnji

hi). With or without axillary fascicles of leaves or if i^resent, these less than 20 mm

long; subulate sepal tips 0.5-6 mm long; capsule thicker walled, dehiscent along

its entire length.

c. Plants low, fre<iuently caespitose, mostly 0.4-2 dm high; densely gray-

strigulose; sepal tips short, 0.3-3 mm long 2. C. lavamluUfoUiis

CO. Plants not caespitose, mostI>' taller and more openly branched, 0.4^ dm
high; variously pubescent or glabrous; if strigulose, sepal tips 2-6 mm
long __ 1. C. hmiwegii

Section 11. Cahjlophus
a. Flowers small, the petals mostly 5-12 nun long; stigma positioned near tlie apex of the
- floral tube or slightly beyond, within the circle of anthers; 30-80% of pollen grains

aborted 6. C. scrrulatus

aa. Flowers usually larger, the petals mostl>- 9-25 nuu long; stigma well exserted, usually

to the end of the ci^isepalous anthers or beyond; 85-100% of pollen grains fertile

____ _. 5. C. herhnulieri

Section I. Salpingia (Toir. & A. Cray) Towiicrj comb, nov.

Oenothera subgen. Salpingia Torr. & A. Ciray, Fl. N. Amer. 1: 501. 1840. Salpingia (Torr. &
A. Cray) Raim. in Engler & Prantl, NatUrl. Pflanzenfam. 3(7): 217. 1893. Galpinsia

Britton, Mem. Torrey Bot. Club 5: 236. 1894.

4-6

pubescent, strigulose, or with spreadiug tricliomes. Leaves 0.3-5 cm long, entire
to serrulate. Inflorescence sparse to dense; buds terete. Flowers opening in af-
ternoon, evening, or morning. Floral tube terete, tubular and gradually expanded

tlirough its entire length, or tubular proximally and funnelform distally, 5-70
mm long. Sepals plane. Petals suborbicular to rhomboidal or squarish. Stamens
nearly equal in length. Capsule promptly dehiscent upon drying, usually not
curved.

TYPK SPKCIKS: CahjlopJuis lavandulif alius (Torr. & A. Gray) Raven.

1. Calylophus hartwegii (Benth.) Raven, Brittonia 16: 286. 1964.

Oenothera haHwegi Benth., Pi. Ilartw. 5. 1839.

Herbaceous to suffrutcsccnt perennial arising from a woody caudex; stems
one to many, sparingly to densely branched above, nearly prostrate to erect,
0.4-5 dm high, strigulose, glandular-pubescent, glabrous, or with spreading hairs,
more densely j)ubescent above. Leaves sparsely to densely distributed along the
stem, sessile or indistinctly petiolate, spreading to ascending, sometimes rcflexed,
linear or filiform to ovate or oblaneeolate, 3-50 mm long, 0.4-12 mm wide, usually
not much reduced up to the stem, variously pubescent or glabrous, the tip acute

1977] TOWyER—CALYLOPIIUS

69

to obhise, the base acute-attenuate to truncate-clasping, the margin entire to ser-
rate, frequently undulate; fascicles of small leaves 1-15 mm long sometimes
present in the nonfloriferous axils; lowest stem leaves sometimes wider than
ahovCj frequently obovate to spatulate, to 65 mm long. Inflorescence lax, with
rarely more than one flower at a time fresh on a stem, variously pul)escent or
glabrous; buds terete. Floral tube terete, tubular in the lower one-half or more,
gradually expanded distally, 16-50 (-60) mm long, 4-20 mm wide at the throat
of pressed specimens, variously pubescent or glabrous without^ the inner sin-face
glabrous distally, sometimes minutely pubescent at the base, frequently fading
to pink or purple on wilting. Sepals 7-28 mm long, 2-10 mm wide, with sul)ulat(^
free tips 0.5-6 mm long, plane, variously pubescent or glabrous, pale yellow-
green, frequently with purple spotting or marginal stripes, fading as with the
floral tube. Petals suborbicular to rhomboidal, 10-35 mm long, similar in width,
highly ultraviolet-reflective, with basal ultraviolet-absorptive spot of varying size,
sometimes absent, frequently turning pinkish or purplish upon wilting. Stamens
subequal; filaments 4-13 mm long, glabrous; anthers 5-13 mm long. Style 25-65

(-75) mm long, usually exceeding the stamens, minutely pubescent below;
stigma flat to slightly revolute, squarish, 1.5-6 mm broad; ovary 4-30 mm long,
1-3 mm wide, variously pubescent or glabrous. Capsule 6-40 mm long, 2-4 mm
wide, moderately thick-walled, completely dehiscent, straight or slightly curved;
seeds 1-2.5 mm long, obovoid, rounded or sharply angled, truncate at the apex.
Self-incompatible. Gametic chromosome numbers, n = 7, 14.

type: Mexico, acuascalientes : Aguascalientes, 1S37, TJwodor IIartice<j^ 10
(K, holotype; P, isotype).

Distribution: Local and colonial to abundant and widespread on rocky,
sandy, gypsum, or limestone soils in arid to relatively mesic open areas, in south-
eastern Colorado, southwestern Kansas, western Oklahoma, Texas (except east-
ern part), New Mexico, southeastern and east-central Arizona, and in Mexico
from Chihuahua, northern Coahuila, and northwestern Tamaulipas south to
Aguascalientes. From ca. 30 to ca. 2,500 m elevation. Flowers Febniary to Oc-
tober.

As treated here, CahjJophus harticegii includes five intcrgrading subspecies.
The species is distributed over much of the Southwest and northern Mexico, oc-
cupying relatively dr)^ plains and mountain regions.

Long, slender floral tubes and vespertine anthesis characterize all subspecies
of C. hartwegii, suggesting a basic adaptation to hawkmoth pollination. Varia-
tion among the subspecies in exact time of anthesis, in ultraviolet reflection pat-
terns, and in flower size was observed, and is likely due to modal differences
in pollination systems. Principal flower visitors included halictid and anthophorid
bees and hawkmoths, but with differing proportions of these insects visiting
the various populations observed. The mean length of the floral tube in this
species and in C lavandtiUfoUus and C. toumeyi is about 30-40 mm. Anthers
and stigma are well exserted beyond the tube, but tend to block the entrance to
it. Short-tongued hawkmoths, such as the abundant species Ilyles lineata

'^Q AxVNALS OF THE MISSOUHI BOTAMCAL GARDEN [Vol. 64

extend

tube to

obtain nectar. In doing this they pick up quantities of pollen on the head and
thorax and can serve as effective agents of pollination. All subspecies of C.
luirtwcgii have been tested at least once for self-incompatibility. In a brief
check, each of 22 plants examined was found to be self-sterile.

con

ration, 56 individuals from 42 populations, including some from each subspecies,
showed translocation heterozygosity. The mean number of heterozygosities per
plant was 1.0 for the species as a whole. Plants with as many as 5 translocation
heterozygosities were found in seed obtained from natural populations, but
most individuals had only 1 or 2, displaying 1 or 2 rings of 4 chromosomes or a

ring of 6 at meiotic metaphase I.

CahjJophus harttrc^ii is a predominantly diploid species, although occasional
tetraploids and plants with extra chromosomes were found. Sixteen plants from
nine populations, including all subspecies but C hartwegii subsp. puhesccns, had
extra diminutive chromosomes. Five of the 62 populations from which chromo-
some number determinations have been made contained tetraploid indix'iduals.
No population was found to be comprised of both tetraploid and diploid plants,
although the low inunber of duplicate counts from tetraploid populations leaves
this possi])ility open. One plant intermediate between subspp. puhesccns and
liartwegii (Totcncr 3) appeared to be triploid, further increasing this likelihood.
Tetraploids were found in C. hartweg^ii subspp. puhescens, harticegii, and mac-

carta, all predominantly diploid taxa.

Difficulties of interpreting the variation in this species have arisen from a
number of causes. The entities in sect. Salpin<;ia tend to have extensive geo-
graphical ranges wliich overlap and interdigitate. Minor ecological differences
often separate the taxa locally. The subspecies of C. liartwegii are distinguished
moiphologically from one another by a few rather slight differences such as
pubescence and leaf shape. Tlie pattern of variation within the section is reticu-
late, with few characters varying concordantly. Broad areas of introgression
exist between certain of the taxa, while others show greater discontinuities in
areas of contact. All forms retained in C. harticcgii are connected either directly
or indirectly by intergradation. The center of the distribution of this species in
western Texas and nortlunn Mexico is characterized by an abimdancc of forms
occurring in near proximity and by a complex display of variation.

Absence of any significant lunnber of intermediates in the appropriate regions
indicates that C. hartwegii, C, hvan(hdifoJius\ and C. toumeyi should be dis-
tinguished specifically. The earlier decision to combine them (Towner & Raven,
1970) was intended to emphasize the overall unity of this group, but a subsequent
thorough study of herbarium material failed to bring forth evidence of intergrada-
tion among the three species.

Populations of C. hartwegii, as mentioned above, occasionally occur sympatri-
cally with C. hcrlamlicri, C, scrrulatiis, and C. tuhicuJa. They also frequently

^lius. and onlv rarelv form hvbrids with that

///

species

HJ77] TOWNER— CALYLOPH US

71

Key to SuiisPEc;iEs

a. Ovary (and usually stems and leaf nuirgins) with spreading triclionies; lea\es (execpt
lowest) al)rnptly narrowed to truncate or sli^^litly clasping at the base; widespread

le. subsp. j)uhcsccn.s

aa. Plant witliont spreading trichonies; ]<'a\'es gradually narrowed at the base or extremely
narrow throughout.
b. Plant glabrous or nearly so; flowers ox^ening from one hour before to one hour after

sunset; widespread „_ Id. subsp. fciidlcn

bb. Plant, especially on ovary and upper stems, with some form of pubescence; flowxTs
usually opening 2-5 hours before sunset, occasionally later.

c. Ovary and stems with sliort glandular pubescence; leaves glabrous to
glandular-pubescent, rarely sparst^ly strigulose, filiform to narrowly lanceo-
late; g>psum or limestone flats, central New^ Mexico to northern Mexico
Ic. subsp. filifolins

cc. Ovary, stem, and leaves usualK* strigulose, or if glandular-pubescent, the

the leaves narrow^ly lanceolate to lanecolate or oblanceolate.

d. Leaves mostly 4.5 to 11 times as l(mg as wide, usually with crinkled-

undulatc margins; plant sparsely strigulose or occasionally minutely

glandular-pubescent; low^ plains from southeastern Texas to nordicrn

Mexico lb. subsp. maccariii

dd. Leaves mostly 9 to 35 times as long as wide, usually not crinkled; plant
sparsely to densely strigulose; high plains and mountains from southern
Trans-Pecos Texas to Aguasealientes ..__ la. STibsp. hartwc^ii

la. Calylophus hartwegii (Bentb.) Raven subsp. hartwegii; Towner in Correll
and Johnston, Man. Vase. PI. Texas 112L 1970.— Fig. 2.

Salpin^ia hadwc^ii (Benth.) Raini. iu Fngler ^' Prantl., Natiirl. Pflanzenfam. 3(7): 217. 1893.
Gal pinsia hartwegii (Benth.) Britton, Mem, Torrey Bot. Club 5: 236. 1894. Oenothera
hanwegii Benth. var. ly})ica Munz, Anier. J. Bot. IG: 706. 1929, pro parte. Calylopliu.s
hartwegii var. hartwegii; Shinncrs, Sida 1; 342. 19G4, pro parte. Oenothera Jiartwegii
\'ar. hartwegii; Munz, N. Amer. FL, ser, 2, 5: 139. 1965, pro parte.

Oenothera greggii A. Gray var. pringk^i Munz, Amer. J. Bot. 16: 711. 1929, pro parte. O.
pringlei (Munz) Munz, N. Amer. Kb, ser. 2, 5: 138. 1965, pro parte, type: Bachimba
Canyon, Chihuahua, Mexico, 27 March 1885, C. G. Vringle 224 (CH).

Oenothera lavandulaefolia Torr. & A. Cray var. iypiea Munz sensu Munz, Amer. J. Bot. 16:
704. 1929, pro parte. O. hivanduU folia var. lacandulifoUa sensu Munz, N. Amer. M., ser. 2,
5: 138. 1965, pro parte.

Stems several to many, sparingly branelied above, deemnbent to somewliat
ascending, or plant tufted, 0.5-3 dm liigh; plant grayish-strigulose tluougliout,
more densely so on the ovary and inn()r(^seence than elsewhere. Leaves dense
on stems, more or less ascending, linear to narrowly lanceolate, 10-35 mm long,
0.5-4 mm wide, the tip acute, the base acute-attenuate, the margin entire to
shallowly and sparsely serrulate, occasionally undulate; fascicles of small leaves
2-15 mm long usually present in the axils. Floral tube ( IS-) 30-50 (-60) mm long,

4-13 mm wide at the throat, sometimes with purple longitudinal bands, often
fading pui*plish. Sepals 8-20 mm long, 3-7 mm wide, with free tips (l-)2-6
mm long, frequently with purple marginal stripes. Petals scjuarish or rhomboidal,
13-30 mm long, frequently fading to a purple or pink color, with basal ultra-
violet-absoiptive spot absent or present and of small to moderate size. FilauuMits
5-10 mm long; anthers 5-9 mm long. Style 30-65(-75) mm long, glabrous;
stigma 2-5 mm broad; ovary 5-12 mm long, 1-2 mm wide. Capsule 10-25 mm
long, 2-4 mm wide; seeds 1-2.5 nnn long. Self-incompatible. Gametic chromo-
some numbers, n = 7, 14.

72

AWALS OF THE MISSOURI BOTANICAL GARDEN

LVoL. 64

Fiouiu-: 15. Distributions of CahjJophus hadwegii subsp. Jiartwc^ii (clots), C h(iriicc<iii
subsp. maccartii (triangles), and intergrades between the two subspecies (open circles).

1977] TOWNER— CALVLOPHL/S

73

Distribution (Fig. 15): Mostly on rocky or gravelly soil, sometimes lime-
stone, in nigged canyons in the northern part of the range, to high plains and
moimtains, reaching pine forest at the southern limits of the range, from Brewster
and southern Hudspeth cos., Texas, south through central Chihuahua, Coahuila,
western Nuevo Leon, and eastern Durango to central and southeastern Zacatecas,
Aguascalientes, and southwestern San Luis Potosi. Elevational distribution from
ca. 900 m (near Solis, Brewster Co., Texas) to at least 2,300 m (6 mi N of Zacate-
cas, Zacatecas, Mexico) . Flowers February to October.

Representative specimens examined:

United States, tfxas: Brewster Co.: Nr. Solis, just N of Mariscal Canyon, Big Bend
National Park, Johnston 6- Correll 24568 (LL). Head of Fresno Canyon, Big Bend Ranch,
Coirell ir RoUins 23672 (LL). Near ^farath()n, Yoting 174 (MO, TEX). Hudspelli Co.:
Panther Hill-Fox Hill area of tlie central Malone Mts., Waterfall 5S30 (GH, NY). Val Verde
Co.(?): Pecos R., Thurhcr 123 (NY). County unknown: Coyote Mt., West Texas, Ilavard
37704 (CAN).

Mexico. ciiiHUAinJA: 12 mi W of General Trias, Breedlovc 15741 (DS). 5 nn' S of
Hidalgo del Parral, Brecdiove 15745 (DS). L3 mi E of Hidalgo del Parral, Brcedlove 14305
(DS). Santa Eulalia Mts., Rose 11693 (US). Santa Eulalia Hills, Wilkinson 4601 (F, NY).
Ca. 30 mi NW of Chihuahua, Lee 50 ( F, TEX). Vicinity of Chihuahua, Palmer 59 (F, MO,
NY, US). Alberto, SE of Chihuahua, Pcnnell 18627 (NY, PH, US). 11 mi N of Parral,
Waterfall 12514 (OKLA). Chihuahua, Le Sueur 130 (F, UC). Between Parral and Villa
Ocampo, Weber <b- Charette 11710 (COLO). Gallego Springs (between Carrizal and
Chihuahua), Wislizenius in 1846 (MO). Near Chihuahua, Pringle in 1886 (F, MO, NY, US).
13 mi N of Ciudad Chihuahua, Brecdiove 15736 (DS). 24 mi W and 1 S of Chihuahua,
Stucssy 1020 (TEX). 23 mi W of Chihualiua, White 2470 (ARIZ). Near Chihuahua,
Le Sueur 811=: 58 (F, MO, SMU, TEX, US), coahuila: Muzquiz, Palm Canyon, Marsh
1002 (F, OKLA, TEX). Muzquiz, Marsh 1134 (F, OKLA, SMU, TEX). Cerro de los Adjoles,
Jermtj 147 (US). 11 km NE of Jimulco, Stanford et al 9 (ARIZ, DS, MO, NY, UC, WTU).
24 km NW of Fraile, Stanford et al. 402 (ARIZ, DS, MO, NY, WTU). Sierra Mojada Mts.,
Jones 233 (MO, POM, US). Sierra Mojada, Jones in 1891 (POM). Near Patos (now Gen-
eral Cepeda), Gregg 723 (MO). Near Bnena Vista, SW of Saltillo, Gregg 387 (MO). Cienega
Grande (just NE of Parras), Gregg 492 (MO). Allende, Marsh 1776 (F, TEX). Siema de la
Paila, Purpus 4977 (F, US). Sierra de Parras, Purpiis in 1905 (UC). Saltillo, Arsene 6510
(US). Paso del Diamante, near Saltillo, Munz 15034 (MO, POM). 30 mi W of Saltillo,
Wislizenius 298 (MO). Saltillo (?), Palmer 344 (US). Saltillo, Palmer 337 (US). 9 mi S
of Saltillo, Straw ir Forman 1337 (RSA). Ca. 20 mi E of Saltillo, MeVaugh 12301 (RSA).
Saltillo, Fisher 32 (F, UC). nuevo ueon: 12 mi N of Sabinas Hidalgo, Heard ir Barklet/
14535 (TEX). 4 mi S of Caleana, McGregor et al. 65 (DS, KANU, SMU). DunANCo: 46
mi N of La Zarca, Straw 6- Forman 1525 (RSA). 15 km NE of Guadalupe Victoria, Ilenrickson
1762 (DS). 77 mi S of Parral, Wiens 3464 (COLO, DS). 6 mi NE of Hidalgo del Parral,
Breedlote 5947 (DS). 21 mi N of La Zarca, Brecdiove 14305A (DS). 71 mi NE of Dmango,
Wateiiall 13336a (OKLA, RSA, SMU). 3-6 mi W of La Zarca, Straw h Forman 1717
(RSA, UC). Ca. 54 mi S of Villa Matamoros, Chihuahua, Straw <Lr Forman 2058 (RSA, UC).
ZACATECAS: On route 49, 2 km N of Route 45, Cruden 1238 (DS). Near Concepcion del Oro,
Palmer 271 (F, MO, NY, UC, US). Gypsum flats. Sierra Hermosa, Shreve 8594 (ARIZ).
2 mi SE of Sonibreto (Souibrerete?), Waterfall 13799 (OKLA, RSA, SMU). 7 mi S of
Fresnillo, Breedlovc 15485 (DS). 9 mi N of Fresnillo, Brecdiove 15486 (DS). 2 mi SW
of Sain Alto, Brcedlove 5952 (DS). 6 mi N of Zacatecas, Brecdiove 15481 (DS). 21 mi N
of Sombrerete, Breedlovc 14338 (DS). 27 mi N of Fresnillo, Brecdiove 14344 (DS). Con-
cepcion del Oro, Pennell 17416 (PH). "Gravelly soil," Purpus in 1903 (UC). Ca. 22 mi NE
of Zacatecas, Straw ir Forman 1492 (RSA). 55 mi W of Zacatecas, Reveal et al. 2660 (DS).
SAX LUIS POTOsi: 8 km NE of Laguna Seca, on km 20 of road from San Luis Potosi to
Antiguo Morelos, Rzcdoivski 6325A (RSA). Charcas, Whiting 898 (ARIZ). Ca. 5 km
NNE of Matehuala, Rzedowski 9186 (DS). 44 mi NW of Sau Luis Potosi on road from
Zacatecas, Brcedlove 5954 (DS). 13 mi NW of San Luis Potosi on road to Zacatecas,
Breedlovc 14346 (DS). 27 mi NW of San Luis Potosi along road to Zacatecas, Brcedlove
15473 (DS). Charcas, Ltmdell 5125 (ARIZ, DS, F, POM, UC, US), aguascalientes: 9 mi
E of Aguascalientes, McVaugh 16680 (RSA, TEX). jalisco(?): Lake Chapala (locality
almost certainly in error) Lcmnion ij- Lemmon in 1905 (UC).

74 ANNALS OF THE MISSCKTRI BOTANICAL GARDEN [Vol. 64

CaJtjIophus hartwegii siibsp. hartwegii occupies a diversity of liubitats, in-
cluding desert scrub, thorn scrub, and pine forest. Typical material for this sub-
species comes primarily from montane areas of north-central Mexico, while col-
lections from lower altitudes and more northern localities exhibit evidence of
hybridization with other taxa. Poor sampling in the remoter areas of northern
Mexico has left details of the geographical range unclear. For example, the oc-
currence of this subspecies throughout most of northern Coahnila and eastern
Chihuahua seems likely, but is not yet established.

The limits of the variation in this subspecies do not correspond closely to
those set by Munz (1929) for Oenothera hartiie<i}i var. typica. The strigose
pubescence of Hartweg's type actually excludes it from Munz's description, and
tlie bulk of the present taxon would also not be included. Oenothera greggii var.
pringlei was named to accommodate tlie strigose-canescent forms, but since
Hartweg's original type represents that group of populations var. pringlei must
be reduced to synonomy. On the basis of leaf width, I have placed some elements
of Munz's var. typica in C. hartwegii subsp. maccartii. In addition, a porti(m of
Munz*s var. typica which included glandular-pubescent individuals is joined
with C. hartwegii subsp. filifoJius, as is part of Shinner's C harttoegii var. hart-
wegii. Lastly, some collections witli glabrous leaves and stems placed in var.
typica by Munz and in var. hartwegii by Shinners clearly belong with C. hart-
wegii subsp. fendleri as it is here constituted. In all previous treatments, the
couiposition of var. hartwegii has been extremely heterogeneous. Clear identifi-
cation of the nature of Hartweg's type and the recognition of larger geographical
assemblages have rendered the variation pattern for C. hartwegii less confusing,
especially with regard to subsp. hartwegii. Finally, a few plants assigned to
Oenothera hvamhilae folia by Munz clearly belong with C hudwegii subsp.
hartwegii on the basis of their long sepal tips, narrow leaves, and southern dis-
tribution.

Considerable variation in leaf width, stature, and degree of pubescence exists
in this subspecies. Especially pubescent plants with small leaves occur in Chi-
huahua and northern Durango. High montane plants tend to be shorter and
more tufted, while the stems of low altitude plants are longer and more erect.
Broader leaves occur in the latter populations, especially where they intergrade
with C. hartwegii subsp. maccartii in northeastern Mexico and in w^cstern Texas,
an extensive zone occupied by intermediates between subspp. hartwegii and
puhesccns.

In one field study, D. E. Brcedlovc (personal communication) found anthesis
in a population of this fonn to occur by I to 1% hours after sunset (San Luis Po-
tosi, BreeiUove 15473), However, no insects visited the flowers during the
period of his observations. In the state of Chihuahua, Mexico I observed several
bees (Agapostemon, Apis) and small butterflies on freshly opened flowers in the

midafternoon ( Towner 247), Anthesis in Chihuahuan populations occurred
from AV-i to 2 hours before sunset. Greenhouse-cultivated plants from a wide
range of Mexican localities showed great variation in opening times, extending
from 4^j liours before sunset to sunset. This variation in anthesis times may be
a result of latitudinal or seasonal differences in photoperiod or a product of adapta-

1977] TOWXER— CA/.y/.()i^//rS

75

tion to locally difforiiig insect faunas. Regional ecological differentiation may
well have occurred within this subspecies since the southern forms from pine
forests seem to open much later than those inhabiting scrub and grasslands in
the plains and hills of Chihuahua. Ilawkmoths are undoubtedly regular e\ening
visitors to all populations of subsp. harfwe<i,iL Ultraviolet-absorbing areas on the
petals can be nearly absent, present along the basal portion of the veins, or present
as a small basal spot (Fig. 8). This suggests that some populations, i.e., those
with spots absent, may be exclusively moth-pollinated while those with spots
and early anthesis may be visited by bees active in the late afternoon.

Chromosomal variation in this subspecies includes polyploidy, translocation
heterozygosity, and extra chromosomes. Two of 11 populations, 1 from Aguas-
calicntes and 1 from Zacatecas, yielded tetraploid counts. Translocation heterozy-
gosity was found in 7 of 17 diploid plants examined and in 6 of 9 populations
from which meiotic determinations were obtained. The mean frequency of
translocations per plant, 0.4, was the lowest for any taxon in the genus. Four
plants of this subspecies had 1 to 4 extra dinu'nutive chromosomes. Examination
of hybrids indicated that one population of C. Iiartwegii subsp. liarticcfiii differed
from the other taxa in sect. Salpin<i^ia by one or two translocations.

lutrogression of C hartwegii subsp. harticcgii with other taxa ai^pears to oc-
cur widely, especially with subspp. piihcscens and maccartii. Intermediates be-
tween subspp. puhcscens and hartwegii exist in southern West Texas, southeastern
Arizona, and probably in northern Chihuahua and Coahuila. Regions of inter-
mediate altitude in northeastern Mexico and along the upper Rio Grande River
in southern Texas contain populations varying on a continuum between subspp.
maccartii and hartwegii. As mentioned on p. 99, C tiihicula subsp. strigtdosiis may
represent a stabilized derivative of introgression between C. tiihicida subsp.
tubicuJa and C. hartwegii subsp. hartwegii. Lastly, plants with narrow leaves,
but lacking dense strigose pubescence, occur near Saltillo in southern Nuevo
Leon and may represent introgressants with C. hartwegii subsp. filifolius, which
occurs to the south of that area, or alternati\ ely they may be independent nar-
row-leaved derivatives of the species.

No examples of sympatry without hybridization have been documented for
C. hartwegii subsp, liartwegii and other taxa. The recent discovery of C. hivan-
dnlif alius in southern Nuevo Leon opens the possibility that it might come into
contact with C. hartwegii subsp. liartwegii. Those two taxa proved somewhat
intersterile in laboratory crosses, and might not be expected to hybridize exten-
sively in the field.

lb. Calylophus hartwegii (Benth.) Raven subsp. maccartii (ShinncMs) Towner
& Raven, Madrono 20: 243. 1970.

Calylophus hartwegii (Bcntli.) Raven var. maccarlii Sliinners, Sida l\ 343. 196-1.

Oenothera greggii A. Gray var. pringlei Munz scnsii Mmiz, Ainer. J. Bot. 16: 711. 1929, pro

parte. O. priugJei (Munz) Munz sensu Munz, N. Amer. Fl., scr. 2, 5: 138. 1965, pro

parte.

Stems several to many, sparingly branched above, nearly prostrate to ascend-
ing, 0.5-5 dm high; plants glandnlar-pubeseent or minutely strigulose. Leaves

76 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

sparse to dense on stems, spreading to more or less ascending, narrowly lanceolate

6-35 mm lonii. 1-6

glabrous to sparsely strignlose or glandular-pubescent, the tip acute, the base
acute-attenuate, the margin subentire to serrulate, usually undulate or undulate-
crinkled; axillary leaves present, to 15 mm long. Inflorescence sparsely to

d

17-45 mm long. 5-12 mm wide at the throat. Se-

5-7 mm wide, with free tios 1-6

witl

1

10-30

frequently fading puiple or pinkish, with a conspicuous, large basal ultraviolet-
absorptive spot. Filaments 6-12 mm long; anthers 5-9 mm long. Style 25-60 mm
long, glabrous abo\'e to minutely pubescent basally; stigma 2-4 mm broad; ovary
5-15 mm long, 1.5-2 mm wide. Capsule 10-22 mm long, 2-3 mm wide; seeds
1-2 mm long. Self-incompatible. Gametic chromosome numbers, n = 7, 14.

TKXAS

Grande, in mesquitc savannah, 24 March 1963, Rosa Ena Benavides 91 (SMU,

holotype; TEX, isotype).

Distribution (Fig. 15): Common semiarid grassy flats, in sandy to gravelly
soil, often of limestone, frequently with Prosopis glanduJosa, Opuntia, Acacia,
Larrca divaricata, and Yucca, on the South Texas Plains and along the Rio Grande
from Val Verde, Kinney, Uvalde, and Milam cos., Texas, south to southeastern
Coahuila, central Nuev^o Leon and northwestern Tamaulipas. From clcx^ations

500

M a rch

Representative specimens examined:

United States, texas: Dimmit Co.: E of Carrizo Sprinps, Jones 28153 (MO, POM).
8 mi S of Catarina, McGregor 16774 (DS, KANU). Duval Co.: 10 mi SW of Benavides,
Garcia 113 (OKLA, SMU, TEX). 16 mi NE of Freer, Malacara ir Gutierrez 30 (LL, SMU).
San Dic^o, Tharp 6031 (TEX, US). 7 mi E of Freer, Rodriguez 104 (OKLA, SMU, TEX).
Goliad Co.: Goliad, Williams 110 (PII, TEX). Jim Hogg Co.: 2 mi N of Santa Elena, Rios ir
Cavazos 68 (LL). Jim Wells Co.: 23 mi N of Alice, Painter et al 14436 (LL, TEX). 8 mi \
of Alice, Bnmi ct al J3 (LL). 15 mi NW of Alice, Castillo 20 (DS, SMU). Kinney Co.:
Spofford, Trelcau in 1900 (MO). Ca. 20 mi NE of Brackettville, Strother 299 (SMU, TEX).
26.0 mi SE of Del Rio, Totvner 34 (DS). La Salle Co.: Encinal, Vdsquez 43 (DS). Live
Oak Co.: 11.5 mi S of George West, Cory 28531 (POM). 8 mi S George West, Flyr 353
(DS, SMU). Maverick Co.: 30 mi SW of Eagle Pass, Bruni 8 (LL, OKLA, SMU, TEX).
5 mi N of Eagle Pass, Rowell 8824 (LL, OKL, OKLA). Eagle Pass, Schott in 1852 (F).
San Patricio Co.: 4 mi NW of Matins, Raven 6- Gregory 19386 (DS). Uvalde Co.: 5 mi W
of Uvalde, Munz 15558 (POM). Sabinal, Jones 29563 (POM). Val Verde Co.: Near
Comstock city limits, Warnock & Turner 696 (SMU). N of Del Rio, Jones 28158 (MO, POM).
Devil's River, Earle ir Earle 441 (MO, NY, US). Ca. 20 mi NNW of Del Rio, McVaugh 8259
(DS, F, SMU, TEX). Ca. 23.5 mi NW of Del Rio, Towner 32 (DS). 3.4 mi SE of Del Rio,
Towner 33 (DS). Webb Co.: Minera, Reverchon 3558 (MO, US). 10 mi S of Laredo,
Cisneros 15 (LL, OKLA). Laredo, Crockett 6444 (LL, US). 11 mi S of Laredo, Rohles 14
(SMU). 8 mi NW of Laredo, Ramirez 45 (DS, SMU). 23 mi NW of Laredo, McCart 7270
(OKLA). 9.5 mi S of Laredo, Corxj 28118 (POM). Zapata Co.: Zapata, Perez 42 (DS).
Near Zapata, Wood 42 (TEX). 5 mi S of San Ignacio, Rodriguez 27 (SMU). 3 mi S of
Zapata, SdncJiez 85 (OKLA, TEX). Zapata, Guajardo 32 (LL, SMU). 2 mi SE of Zapata,
Gonzalez- Arroyo 92 (LL, OKLA, SMU).

Mexk:o. tamax^tlipas : Along the river road, 20 mi E of the International Highway,
Escahntc 55 (SMU, TEX, OKLA). 3 mi SW of Headquarters, Loreto Ranch, Crutchfield 6-
Johnston 5568A (TEX). 50 mi SE of Nuevo Laredo, Garcia ir Garcia 35 (DS, WTU). nuevo
LEON: 24 mi W of Monterrey, Waterfall 6 Wallis 13214, 13215 (RSA, SMU). Monterrey,

1977] TOWNER— CALYLOPi/(7S

77

Fisher 272 (MO, US). Rio Santa Catarina, Monterrey, Arscne 6306 (MO, US). 65 mi S of
Nuevo Laredo, Frye 6 Frije 2369 (DS, MO, NY, RM, RSA, SMU, UC, WTU). 9 mi S of
Nuevo Laredo, Frye 6 Frye 2390 (NY, RSA, UC, US, WTU). 12 mi N of Sabinas Hidalgo,
Barkley 6 Heard 14535 (F, MO, US). 17 mi NE of Sabinas Hidalgo, Rodri'^uez 70 (SMU,
TEX). 16 km W of Sabinas Hidalgo, Domiiifiucz & McCart 8255 (SMU, TEX). 45 mi S of
Nuevo Laredo, MeCart et al. 8133 (OKLA, SMU, TEX). Sabinal (?), Jones 29563 (MO, UC).
Monterrey, Dodge 158 (US). Between Monterrey and Rcynosa, along side road to San Juan,
Langimn 2870 (DS, PH). Monterrey, Edwards 6 Eaton in 1846 (NY). 50 mi S of Laredo,
Hess ir Hull 637 (OKL). Ca. 54 mi S of tlie U.S. border in Laredo, Towner 35 (DS). 39 mi
N of center of Monterrey, Towner 36 (DS). coahuila: 9 km S of Parras, Sierra Ncgra,
Stanford ct al. 158 (ARIZ, MO). Near Diaz (now Piedras Negras), Pringle 8304 (DS, F MO,
PH, POM, RM, RSA, UC, US). Ciudad de Porfirio Diaz, Canby 109 (US). Guad'alupe,
Aguirre 703 (RSA). Parras, Aguirrc & Reko 82 (NY). Ca. 48 mi N of Saltillo, Jackson 6722
(KANU). 25 km S of Piedras Negras, lUneliart 218 (OKL, OKLA, RSA). 13.4 mi S of
central Saltillo, Toioner 52, 53 (DS).

Closely related to Calijlophus ]iartice<^ii sub.sp. hartwegii, this subspecies oc-
curs at liigher latitudes and lower elevations. It is relatively common in dis-
turbed areas in the grasslands of southern Texas. It corresponds closely to var.
maccartii as treated by Shinners except that narrower-leaved plants are included
here. The broad leaves, which are frequendy oblanceolate, early afternoon anthe-
sis, and sparser, often glandular pubescence serve to distinguish this sub.species
from subsp. hartwegii. Only leaf dimensions serve adequately in separating
subspp. maccartii and filifolius. Considerable phenotypic variation occurs in
sub.sp. maccartii in terms of pubescence, leaf shape, stature, and nature of the
leaf margin. Leaf margins may be serrulate, undulate, or subentire.

Pollination was studied near Saltillo, Coahuila, Mexico {Towner 52, 53) at
a roadside population of C. hartwegii subsp. maccartii. Anthesis was not ob-
served, but had been completed by 2Vi hours before sunset. Most visitors to the
flowers were halictid bees, especially Evyheus and Agapostemon, some of them
possibly oHgoleges. These may have played some part in pollination in the late
afternoon and morning in spite of their small size. No large native bees, except
for a single Bomhus, and no hawkmoths were observed visiting flowers, but
their involvement cannot be discounted on the limited evidence available.

Greenhouse studies showed anthesis times occurring 3-5 hours before sunset

and flowers with large central ultraviolet-absorbing areas. Se]f-incompatil)iIity

was found in the 3 plants available for testing. This suggests that C. hartwegii
subsp. maccartii has perhaps secondarily shifted from hawkmoth pollination,
as indicated by its morphology, to bee pollination, as might be inferred from its
ultraviolet pattern and behavior.

Two of 10 populations showed tetraploidy, in addition to 1 population inter-
mediate between Cahjlophtis hartwegii subspp. fiUfolius and maccartii. One plant
from each of 2 populations had a single extra diminuti\'c chromosome. Half of
16 plants, representing 5 of 8 diploid populations, were heterozygous for trans-
locations. The mean frequency of translocation heterozygosities was 0.6 per
plant, and only one plant had as many as 2. Experimental hybrids between C.
hartwegii subsp. vuiccartii and other members of sect. Salpingia were heterozy-
gous for 1 or 2 reciprocal translocations.

Introgression occurs, as mentioned above, with C. hartwegii subsp. hartwegii
in southern Texas and northeastern Mexico. It also appears to ha\'e taken place

78

ANNALS OF THK MISSOUIU BOTANICAL GARDEN [Vol. 64

inde

tlicrc is difficulty in recognizing the sources of variation in this region. Two
further collections [36 mi W of Monterrey, Coahuila, Mexico, Towmr 39 (DS).
5 mi W of Marathon, Brewster Co., Texas, Warnock 60004 (TEX)] appear to be
intermediate between subspp. maccartii and fiJifolius, although this may not
necessarily have resulted from introgression. The only other taxon of CaJyIophus
occurring near C. Jmrtwegii subsp. moccartii is C. herlandieri subsp. herlomlieri.
The two arc essentially intersterile, but in the South Texas Plains have often
been mistaken for one another because there they tend to resemble each other
in leaf shape and character of the margin.

Ic. Calylopluis hartwegii (Bcnth.) Raven subsp. filifolius (Eastw.) Townrr
& Raven, Madroiio 20: 243. 1970.

Oenothera tuhieula A. Gray var. filifolia Eastw., Pioc. Calif. Acad. Sci., scr. 3, 1: 72. 1897.
GaJpiusia filifolia (Eastw.) Heller, Cat. N. Amer. PL, ed. 2. 8. lUOO. Ocnothcia hart-
wegii Bcnth. var. filifolia (Eastw.) AIuiiz, Aiiier. J. Bot. 16: 707. 1929. Calylophus
hartwegii (Bcnth.) Raven var. filifolius (Eastw.) Shinncrs, Sida 1: 345. 1905. Oenothera
hartwegii \ar. fendJeri (A. Cray) A. Cray subvar. filifolia (Eastw.) H. Lev., Monogr.

Onulh.335. 1908.
Oenothera liaiiivegii Benth. var. tijpica scnsii Munz, Amer. J. Bot. 16: 706. 1929, pro parte.
Calylophus hartwegii (Benth.) Raven var. hartwegii sensn Shinncrs, Sida 1: 342. 1964,
pro parte. Oenothera hartwegii var. hartwegii sensu Munz, N. Amer. Fl., .scr. 2, 5: 139.
1965, pro parte.

Stems several to many, moderately to densely branched above, decumbent
and spreading to somewhat ascending, 0.5-4 dm high; plant minutely glandular-
pubescent throughout, more densely so on the ovary and inflorescence,
infrequently sparsely strigulose on the ovary and leaves. Leaves moderately well-
spaced to dense on the stem, spreading to ascending, filiform to narrowly lanceo-
late, 3-45 nun long, 0.4-3 (-4) mm wide, the tip acute, the base acute-attenuate,
the margin entire to remotely serrulate, occasionally undulate; axillary leaves
present, to 10( + ) nmi long. Floral tube 16-50 mm long, 4-14 mm wide at the
throat, occasionally fading pinkish. Sepals 7-17 mm long, 3-7 mm wide, with
free tips 0.5-4 mm long, frequently with purple spotting and occasionally with a
purple marginal stripe. Petals suborbicular to somewhat rhomboidal, 12-23 mm
long, occasionally fading pinkish, with a basal ultraviolet-absorptive .spot of

moderate to large size. Filaments 6-13 mm long; anthers 6-11 mm long. Style
26-60 mm lonsi, dabrous above, glabrous or minutely pubescent basally; stigma

g, 1-2 mm wide. Capsule 7-22 mm long.
Self-incompatible. Gametic chromosome

^

1.5-4 mm broad; ox'ury 4
2-3 nun wide; seeds 1.2-
nunibci ., n — 7.

type: United States, new Mexico: White Sands, pr()1)ably in Otero Co.,

Angust 1896, T. D. A. Cockerdl (CAS).

Distribution (Fig. 16): Highly local, but often abundant, almost always on

semiarid gypsum flats, dunes, or outcrops, frequently with Larrea divaricata.
Yucca, or Juniperus, from Otero and Torrance cos., New Mexico south and east
through the Trans-Pecos and soutluMu Panhandle regions of Texas, thence north-
east to Cottle Co., Texas and southward from widely scattered localities in cen-

1 y77] TOWNER— CALYLOP/7(7,S

79

tral Chihuahua aud Coahuila. Occuriing from elevatious of ca. 600 in (7 mi
N of Spur, Dickens Co., Texas) to ca. 1,850 m (7.2 mi SE of Wiharcl, Torrance
Co., New Mexico). Flowers May to October.

Representati\x^ .specimens examined:

United States, new Mexico : Cha\es Co.: 20 mi S of Roswell, Eaile 6 Edilc 293 (MO,
NY, POM, RM, UC, US). 2 mi E of Bottomless Lakes State Park Heackinarters, IIcss 73
(WTU). 56.7 mi SE of Vaughn, Towner 125 (DS). 18.7 mi N of Roswell on U.S. 285, Touner
128 (DS). De Baca Co.: 22 mi S of Fort Sumner, Bmol<s & Stephens 25763 (DS). Dona Ana
Co.: Jornada Came Reserve, Wooton, no date (US). Eddv Co.: ,3 mi NW of Texas state line on
U.S. 62/180, Raicn & Gregory 19156 (DS). 6 mi SW of White's City, Mutiz 6 Gregory 23357
(POM, UC, WTU). 11.4 mi SW of White's City, Towner 22 (DS). Lea Co.: 55-60 mi E of
Roswell, Pfl/wt'r 62 (F). Lincoln Co.: White Mts., 5,400 ft, ^\'of;;on iSi (ARIZ, DS, CII MO
NY, POM, RM, UNM, UC, US). 22 mi N of Carrizozo, Brooks 6 StepJicns 25957 (DS). Otero
Co.: Round Mt., along Tularosa Creek in Sacramento Mts., Wooton in 1899 (ARIZ, DS, NMC,
NY, POM, RM, US). White Sands National Monument, Miinz 6- Gregory 23335 (POM)!
White Sands National Monument, 2 mi W of headquarters, Towner 11 (progeny, DS).
Torrance Co.: Near Willard, Wooton 2730 (COLO, DS, RM, US). 7.2 mi SE of Willard,
Towner 122 (DS). texas: Culberson Co.: 2 mi SE of U.S. 62/180 at New Mexico line'
MeVaugh 8164 (DS, CII, LL, SMU, TEX). 30 mi N of \'an Horn, Waterfall 4122 (CII)!
Dick-ens Co.: 7 mi N of Spur, Moss 19 (OKLA). Ector Co.: 1 mi E of jet. of Texas 185 and
U.S. 385, Gregory 424 (DAO, DS, 1\SA, UC). Caines Co.: 15 mi E of Seminole, Linidell 6
Lundell 16955 (LL). Howard Co.: Big Springs, Tracy 8306 ( F, GH, MO, NEB, NY, TEX,
US). Hudspeth Co.: Cypsum quarry E of Finley, Wrtfrr/w// ,5023 (CH, MO). Kent Co.: 2 mi
W of Clairemont on U.S. 380, Correll l~ Johnston 22107 (LL). Loving Co.: Along Salt
Creek near higlnvay 285, N of Orla, Correll & Correll 26016 (Mixed with C. hartwegii subsp.
puhescens, LL). Martin Co.: E of Stanton, Lundell i^ Lundell 16916 (LL). Midland Co.:
i mi E of Midland, Cory 42030 (POM, TEX). Nolan Co.: Sweetwater, Reverehon 1285
(F, MO). Ward Co.: 9.5 mi S of Monahans, Gregory 174 ( RSA, UC, WTU). Winkler Co.:
1 mi N of southern county line on highway 18, Irvmg 69 (SMU, TEX).

Me.xico. coahuila: Morillo, Saltillo, Lyonnet 3497 (US). Saltillo, Fisher in 1926 (DS).
6 mi xN of La Ventura, Johnston 7644 ir Shreve 8726 (ARIZ, UC, US), nuevo eeox on
coahuila: Vanegas-Saltillo road, alkali plain, Lundell 5721 (ARIZ, F, POM), chhiuahua:
13 mi S of Callegos, Breedlove 15734 (DS). zacatecas: Intersection of highways 49 and 45
Cruden 1238 (TEX). No locality; Mexico, Gregg 33 (MO).

Cahjiophu.s liartwegii sub.sp. filifulius, which is largely endemic to gypsum
soils, may include some convergent populations of independent origin. The
Texas and Coahuila i>opulations are widely separated, with no collections having
yet been obtained in the intervening region, a span of over 650 km.

I have retained here certain populations with broader leaf dimensions than
were included by Shinners or Munz. Plants from the type locality do not all have
filiform leaves, although collections from the White Sands area do include the
narrowest-leaved plants in the .species. Inclusion of the broader-leaved popula-
tions here simplifies the variation pattern in C. hartwegii sub.sp. hartwegii and
renders subsp. fiUfolius a major geographical race which is nonetheless phcnct-
ically discrete. In West Texas and southeastern New Mexico this taxou is abun-
dant on plains at about 600-1,200 m elevation. Some variability is shown by
this subspecies in terms of leaf width, petal .shape, the presence and distribution
of anthocyanins, and length of the free sepal tips. Pubescence is relatively uni-
form, with nearly all plants being glandular-pubescent, and only a few being
even minutely strigulose.

The polhnation studies of Cregory (1964; as Oenothera hartwegii) in Ecto
I Ward COS., Texas indicated that flowers of C. hartwegii sub.sp. filifoliiis wer^

r

80

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

FiGuiiE 16. DistribuLions of Cahjlophits hartwcgii subsp. pubcsccns (dots), C. hartiocgii
subsp. fcndJcri (open circles), and C. hartwcgii subsp. fiVifolius (triangles).

1977] TOWSER—CALYLOmUS

81

open well before sunset and were visited by hawkmoths in the evening and
sometimes by bees in the afternoon. Greenhouse studies showed anthesis to oc-
cur 3-6 hours before sunset. Ultraviolet-absorbing spots at the base of each petal
were fairly large, presenting conspicuous regions of high contrast which would
be visible to diurnal insects. Tests for self-incompatibility on 6 plants all proved
positive.

Cytogenetically, C. liarticegii subsp. filijoUus exhibits a great deal of \'aria-
tion, with plants averaging 1.7 translocation heterozygosities. Ten of 12 plants
from 8 populations were heterozygous. The maximum association of cliromo-
somes observed consisted of a ring of 12 and a bivalent, present in a plant grown
from seed collected in Winkler Co., Texas {Irving 69), As many as 11 extra
diminutive chromosomes were observed in plants from that population, with
some being possessed by each of the 4 plants examined. No other population
demonstrated extra chromosomes, and no tetraploid or higher counts were ob-
tained from this subspecies. One possible intermediate between C. harlwegii

subspp. maccart'ii and /////

(Towner

lifoli

/'

and C hartwegii subsp. puhescens, and one or two translocation differences from

the other taxa.

/////

subsp. fendleri in the southern Texas Panhandle and in New Mexico, but hybridi-
zation is limited by the altitudinal separation of these subspecies. Similarly, there
is limited hybridization with C. harticegii subsp. puhescem in the same regions.
Possible hybridization with C. harticegii subsp. 7naccartii and C. harticegii subsp.
harticegii was treated under those taxa. Sympatry without hybridization occurs
in New Mexico where C herlandieri and C. serrulatus occasionally come into

contact with this subspecies.

Id. Calylophus hartwegii (Benth.) Raven subsp. fendleri (A. Gray) Towner
ik Raven, Madrono 20: 243. 1970.

Oenothera fendleri A. Grav, Mem. Amer. Acad. Arts., n.s., 4: 45. 1849. O. harfwefi^ii Benth.
var. fendleri (A. Gray) A. Cray, PL Wright. 2: 58. 1853. Galpinsia hartwegi (Hontli.)
Britton (var.) fendleri (A. Gray) Small, Bull Torrey Bot. Club 23: 186. 1896. G. fendleri
(A. Gray) Heller. Cat. N. Amer. PL, ed. 2: 8. 1900.

Oenothera harttcegii Bentli. var. tij})ica sensu Munz, Amer. J. Bot. 16: 706. 1929, pro parte.
Cahjloplun harttcegii (Bentli.) Raven var. hartwegii sensu Shinners, Sida 1: 342. 1964, pro
parte. Oenothera hartwegii var. haiiwegii sensu Munz, N. Amer. FL, ser. 2, 5: 139. 1965,
pro parte.

Stems one to several, sparingly to moderately branched above, ascending to
more or less erect, 1.5-4 dm high; plant glabrous throughout, infrequently mi-
nutely and sparingly glandular-pubescent. Leaves sparse to dense on stems,
more or less ascending, linear to oblanceolate or lanceolate, 10-50 mm long,
1.5-10 mm wide, the tip acute, the base acute-attenuate to obtuse, infrequently
nearly clasping, the margin entire to subentire, infrequently undulate; axillary
leaves usually absent, to 10 mm long when present. Floral tube 30-50 mm long,

J^2 AWALS OK THE MISSOURI IIOTAXICAL CARDEN [Vol. 61

7-15 nun wide at th(^ throat, sometimes witli purple lines, frequently fading pur-
plish or orange. Sepals 9-28 miu long, 4-10 mm wide, with free tips 0.5-3 mm
long, occasionally with purple margins or spotting. Petals obovatc to somewhat
rhomboidal or s(iuarish, 10-30 mm long, usually fading purplish or reddish,
with a basal ultra\'iolet-absorptive spot small or absent. Filaments 5-12 mm
long; anthers .5-13 mm long. Style 40-75 mm long, glabrous above, minutely
pubescent basally; stigma 2-6 mm broad; ovary 7-20 mm long, 1-2 mm wide.
Capsule 10-40 mm long, 2-3 mm wide; seeds 1-1.5 mm long. Self-incompatible.

Gametic chromosome number, n ~ 7.

type: United States, new mexico: without specific locality, "prob-
ably Santa Fe," 1847, A. Fcndlcr 230 (Gil, lectotype; Gil, MO, NY, P, PII, US,

isolectotypes); cf. Mun/, Amer. J. Bot. 16: 708. 1929. The sheets assigned this

4

iuiinl)t>r are prolxibly a mixture of collections from the originally published lo-
calities, i.e., Santa Fe, on the Rio del Norte (Rio Grande), and from Rock Creek
eastward to the Cimarron River.

Distribution (Fig. 16); Uncommon on clay or gravelly soils, occasionally cal-
careous, from high plains with Prosopis <iJ(indulosa and Junipcrus, to montane
forests with Junipcrus, Finns cihiVis, and occasionally Finns pondeiosa, from Rar-
l)er and Morton cos., Kansas, south through western Oklahoma and widely scat-
tered sites in the Texas Panhandle to (^astern Chihuahua, central Trans-Pecos
Texas, central and western New Mexico, and east-central Arizona. Elevational
distril)ution from ca. 370 m (Agawam, Grady Co., Oklahoma) to 2,2(H) m (17
mi N of Alpine, Apache Co., Arizona). Flowers April to Octoi)er,

Ht'pit's('ntati\e .spc'C'inu'n.s cxaniincd:

UNriEi) States, kaxsas: Biiil)cr Co.: NW corner of county, Baker iu 1904 (\V).

C.ypsnni liills, Ilitcluork 689 (CII, \MC, NY, RM, US). 6 mi W of Medicine Lodge, Stci>hcn.s
11150 (KANU, OKLA). 7 mi W of Medicine Lodge, McGicfior 14243 (KANU, SMU, US).
7 mi SW of Medicine Lodge, McGregor 14472 (SMU, US). Sandy .soil S of Coats, BoiiiJi/ in
1936 (ARIZ, CAN, F, MO, OKL, OKLA, RM). 7 mi S of Sun City, McGregor 14019 (KANU).
Morton Co.: 4 mi W of Rolla, McGregor 12858 (KANU, SMU, US). Oklahoma: Blaine Co.:

5 mi NK of Watonga, Ste})]ieus 6 Brooks 20819 (KANU). Roman Nose State Park, Goodman

6 Waterfall 4185 (CII, OKL, OKLA). Crady Co.: 8 mi SW of Chickasha, Pearce 767 (SMU)-
(^-eer Co.: 7.7 mi S of Mangum, Towner 79 (DS). 2.6 mi S of Mangnm, Towner 85 (OS).
0.5 mi S and 1.3 mi W of Brinkman, Towner 86 (DS). Harmon Co.: 10 mi S of McO'icen,
Stephens ir Brooks 20758 (DS, KANU). Harper Co.: Near Buffalo, Stevens 535 (DS, CII,
MO, NY, OKL, OKLA, US). 17 mi E and 7 S of Buffalo, StepJiens 6 Brooks 21685 (DS,
KANU). Kiowa Co.: Snyder, Stevens 1198 (OKLA). Roger Mills Co.: 18 mi N of Cheyenne,
Waterfall 11897 (OKLA, SMU, TEX, US). Ca. 8 mi E of Strong City, Towner 156 (DS).
Woods Co.: .37 mi W of Al\a, Stratton 6384 (KANU, OKL). Woodward Co.: 24 mi N of
Mooreland, Brooks 6 Stephens 21658 (DS, KANU). texas: Hemphill Co.: Prairies N of
Canadian, Eggerl in 1901 (MO). Jeff Da\ is Co.: 14.8 mi N of Marfa cit> limits, Parncll
68-T-30 (DS). 8 mi S of Fort Da\is, Munz 6 Gregory 23384 (RSA, UC). 3.8 mi W of Fort
Davis. Gregory 134 (NY, RSA, WTU). Mesa S of Fort Davis, Andrews 63 (COLO, CH).
Limpia Cainon, Davis Mts., Bray in 1902 (TEX). Presidio Co.: 8 mi E of Marfa, Warnork
7916 (LL, SMU, TEX). 8 mi NW of Marfa, Jaekson in 1964 (progenv only, DS). 13.0 mi
NW of Marfa, Towner 25 (DS). 11.9 mi NW of Marfa, Towner 26 (DS). 10.4 mi NW of
Marfa, Towner 27 (DS). Randall Co.: Bottom of can>on, Palo Duro Canyon State Park,
Lundell ir Lundell 11442 ( LL, SMU). Wilbarger Co.: 1.5 mi S of Harrold, Whitehousc 9764

(ARIZ, SMU, UC, US). .NKw me.xico: Catron Co.: Mangas Creek, Rushy in 1880 (US).
Mangas Canvon, Greene in 1880 ( F, MO, NY, PII). Crant Co.: Vieinit\- of Cila, Maguire
11630 (DAO, NY, WTU). Lincoln Co. (?): Gallinas Mts., Wooton 274l' (VS). Otero Co.:
8 mi E of Mescalero, Parker 2556a (ARIZ, COLO). Above Mescalero, White Mts., Wooton in

U)771 TOW'Wn—CALYLOPHUS

83

1895 (US). ]^i() Arriba Co.: Arroyo de Agua, Gregory 588 (UC). Sandoval Co.: San Isidro,
Benedict 2311 (US). San Mi^nel Co.: Near Pecos, Standley 4952 (CH, MO, NMC, NY).
Santa Fc Co.: 19 mi W of Santa Fe, alon^ir tlie I^io Crande R., Heller ir Heller 3G22 (MO,
NY, US). Socorro Co.: 0.1 mi S of Ma^dalena, Towner 119 (DS). Torrance Co.: 3.1 mi NW
of Ccdarvalc, Toiaier 123 (DS). 2.6 mi W of WiUard, Towner 12L Valencia Co.: 8 mi
E of Ramah, Wooton in 1906 (NMC, NY, US), aiuzona: Apache Co.: 17 mi N of Alpine,
Iheedlove I429S (DS). 7.9 mi N of Alpine, Towner 110 (DS). 4.5 mi N of Nutrioso, Towner
III (DS). S.O mi N of Nntrioso, Towner 112 (DS). Coconino Co.: Walnut Canyon,
MacDoui^id in 1898 (ARIZ, F, CH, NY, 1^11, RM, UC, US). Flagstaff Cemetery, Demaree
42817 (AinZ, DS, OKLA, RSA, SMU). Navajo Co.: Near Heber, Parker et al 6832 (ARIZ,
F). 12 mi N of \Vlnteri\cr, Goodman 6 Iliteheoek 1298 (DS, F, MO, NY, PII, UC).

Mexico, ciihiuahua: U mi E of Highway Hi on road to new lake on Rio Conchos,
Powell et al 2032 (TEX).

A race with distinctive characters of distribution, morphology, and floral
Ix^liavior, CahjIopJius hartive^^ii subsp. fendlcri definitely merits recognition,
contrar\' to the opinion of Shinners (1964). I.ate anthesis, glabrous vegetative
parts, and distribution at relatively high altitudes or latitudes are strongly cor-
related in this form. In tlie northern part of its range, it occurs at intermediate
or low altitudes, but in Arizona and New Mexico it ranges up into the coniferous
forests. Leaf dimensions not being of critical importance for delimiting this
subspecies has allowed the inclusion of elements from Munz's Oenothera hart-
wegii var. harlwcgii. Thus considerable variation in leaf dimension is retained
in C. hartwegii subsp. fendlcri^ which attains a large and broad leaf size for the
species, especially in collections from the Creat Plains. Relatively narrow-leaved
forms from the mountains of Trans-Pecos Texas belong here, although they have
been traditionally placed with Oenothera harticegii var. hartwegii.

I1ie type series of Evyhiciis gaJpinsiae (Cockerell) was collected from Calij-
lophus hartwegii subsp. fendlcri near Pecos, New Mexico where the bees were
active at 7:30 in the evening (Cockerell, 1903). Pollination studies of Gregory
(1964) in Jeff Davis Co., Texas (as Oenothera hartwegii) suggested that anthesis
occurs near s\mset and that pollination is largely accomplished by hawkmoths.
My field observations in Grant Co., New Mexico {Towner 244) differed some-
w^hat from Gregory's in that numerous bees of the genera SpJieeodogasfra and
Evylaeus were active gathering pollen from this subspecies for about one hour
starting at sunset. As observed by Gregory, hawkmoths visited the flowers heavily
in the early evening. Infrecjuent visits by bees were seen in the morning. Field
obserxations at this site and elsew^here in New Mexico indicated that anthesis
occurs at about sunset, and flowers on cultivated plants opened within an hour
before or after sunsc^t. Photography under ultraviolet light showed plants to
have either no spots of absorption on the petals or only very small ones (Fig. 10).
Two i:)lants were checked and found to be self -incompatible.

No tetraploid indi\'iduals have been found in this subspecies. Sixteen of 23
plants from 12 of 15 populations were heterozygous for translocations, with a
mean ninnber of 0.9 heterozygosities per plant. The maximum number of
heterozygosities exhibited was 2, seen in 5 plants. Hybrids with other taxa in
sect. Salpingia behaved identically to those involving C. hartwegii subsp. jili-
joVms in terms of chromosome pairing. Foin- to 5 extra diminutive chromosomes

were observed in 3 plants, each from a different population.

Instances of hybridization with C. hartwegii subsp. pubeseem, particularly in

J54 ANNALS OF THE MISSOURI BOTANICAL GAUDKN [Vol.. 64

Oklahoma, seem relatively common, and the two subspecies are not separated by
large ecological differences. One population in Harmon Co., Oklahoma consisted
of both subspecies and intermediates. Cases of intergradation with other taxa
have been discussed in previous sections. Cahjiophus harticegii subsp. femUeri
occurs frequently with C. sernilatus in Oklahoma^ with no indication of hybridi-
zation, A population in Torrance Co., New Mexico was found growing with C.
lavandulij alius, and again no intermediates or putative hybrids were observed.

le. Calylophus hartwegii (Benth.) Ua\'cn subsp. pubescens (A. Gray) Towner
& Raven, Madrono 20: 243. 1970.

Oenothera grep,gii A. Cray \^ar. pubescens A. Cray, Pi. Wright. 1; 72. 1852, Cahjlo}}]itis

Jiartivegii (Benth.) Ra\en var. pubeseens (A. Cray) Shinncrs, Sida 1: 344. 1964.
Oenothera greggii A. Cray, Meui. Anier. Acad. Arts, n.s., 4: 46. 1849. Galpiusia greggii (A.

Gray) Snuill, Bull, torrey Bot. Chil) 23: 18fi. 1896. Oenothera harticegii Benth. \ar.

feudleri (A. Cray) A. Gray subvar. filifolia II. Lev. f. thijmifoUa II. Lev., Mono^r. Onoth.

335. 1908, based on MO isotype. O. ^rr^^^^n var. fiy/j/a/ Munz, Amer. J. Bot. 16: 709. 1929.

Gregg 591 (CII, holotypc; MO, NY, isotypcs).

NW

Oenothera lampasana Bnckley, Proc. Acad. Nat. Sci. Phihidelphia 1861: 454. 1961, Galpinsia

hmpasana (Buckley) Wooton & Standley, Contr. U.S. Natl. Herb. 16: 152. 1913.

Oenothera greggii A. Gray var. lampasana (Buckley) Mini/, Auier. J. Bot. 16: 710. 1929.

type: United States, Texas, Lampasas Co., prairies, 1860-1861, S. B. Buckley (PII).
Calpinsia interior Small, Fl. S.E. U.S. 845, 1335. 1903. type: United States, Nebraska, Cherry

Co., Fort Niobrara, 25 June 1888, T. E. Wilcox (NY). This locality is more than 300 nu*

N of the known ran^e of this subspecies and probably resulted from dispersal by accident

or human intent, or the label may have been switched.
Galpiusia eamporum Wooton & Standley, Contr. U.S. Natl. Herb. 16: 152. 1913, Oenothera

camporum (Wooton & Standley) Tidestrom in Tidestrom & Kittell, Fl. Ariz. & N. Mex.

278. 1941. type: United States, New Mexico, Lea Co., Knowles, 29 July 1890, E. O.

Wooton (US-564592, holotype; NMC, POM, US, isotypes).

Stems several, moderately branched above, decumbent to more or less erect,
1-5 dm high; plant usually covered throughout with long spreading trichomes,
most densely on the ovary, inflorescence, and upper stem, occasionally also with
short glandular or nonglandular trichomes. Leaves somewhat sparse to dense
on the stem, most commonly spreading to reflexed downward, sometimes more
or less ascenchng, very narrowly elliptic or narrowly lanceolate to ovate, 5-40
mm long, 1.5-12 mm wide, the tip acute, the base acute to truncate or subcordate
and clasping, the margin entire to sparsely serrulate, occasionally undulate-
crinkled; axillary leaves often absent or much reduced, occasionally to 15 mm

long. Floral tube 20-50 mm long, 4-20 mm wide at the throat, only rarely with
purple stripes, occasionally fading purplish. Petals ol)ovate to somewhat rhom-
boidal or squarish, 12-35 mm long, frequently fading pinkish or pmplish, with
a basal ultraviolet-absorptive spot of small or moderate size. Filaments 5-12
mm long; anthers 4-13 mm long. Style 25-70 mm long, glabrous aliove, minutely
pubescent basally; stigma 1.5-5 mm broad; ovary 5-30 mm long, 1-3 mm wide.
Capsule 6-35 mm long, 2-3 mm wide; seeds 1-1.7 mm long. Self-incoinpatible.
Gametic chromosome numbers, n = 7, 14.

TYPE: United States, texas: dry hills beyond the Pecos River, probably
from Pecos Co., August 1849, Charles Wright 199 (Gil, holotype; GH, NY, PII,

1977] TOWKEE—CALYLOrHUS

85

US, isotypes). The locality was calculated from the dates and account of
Wright's trip given by McKelvey (1955: 1067-1068).

Distribution (Fig. 16): Common and colonial in moderately dry open places,
plains, and hills, in sandy to gravelly soil, often of limestone or gypsum, fre-
quently with Prosopis glandulosa and Juniperus, from Baca Co. and eastern Las
Animas Co., Colorado and Morton and Meade cos., Kansas, to western Okla-
homa and the Texas Panhandle, throughout central and Trans-Pecos Texas, thence
west through eastern and southern New Mexico to central and southeastern Ari-
zona; also south very locally in central Coahuila and northeastern Durango. Ele-
vational distribution from 200 m ( 10.5 mi E of Weatherford, Parker Co., Texas )
to 2,100 m (2.4 mi NW of Corona, Torrance Co., New Mexico). Flowers March
to October.

Rpiiresentative specimens examined:

United States. Colorado: Baca Co.: 20 mi S of Pritcliett, Harrington 3325 (RSA).
27 mi S of Pritchett, Wcher 4608 (COLO, UC, WTU). Las Animas Co.: 7 mi S and 16 E of
Kim, Wcher 4387 (COLO). 4 mi W of Andrix, Rogers 4952 (COLO, US), kansas:
Clark Co.: 10 mi S of Ashland, Rydherg <Lr Imlcr 744 (NY). Meade Co.: 8 mi S and
7 E of Meade, Ilorr in 1957 (KANU). SE corner of county, above Wolf Canyon, Horr
3612 (KANU). Morton Co.: Stony hills, Hitchcock 166 (GH, MO, NMC, NY, POM, RM,
US). Point of Rocks, Hitchcock 634 (GH). On Cimarron R., N of Elkhart, Point of
Rt)ck, Rt/clherg ir Imlcr 942, 943 (MO, NEB, NY). Point of Rocks, 7 mi N and 4 W
of Elkhart, Stephens 11258 (KANU). No county: SW Kansas, Plank in 1886 (GH).
OKLAHOMA: Bcckham Co.: 8 mi N of Sayre, WiecJman in 1959 ( OKL, OKLA). 6 mi S of
Elk City, Eskew 1503 ( GH, KANU, OKL, OKLA). 1.7 mi W and 2.4 N of Elk City, Stratton
6835 (KANU, OKLA). Cimarron Co.: 16 mi SE of Kenton, Waterfall 7433 (OKL, OKLA,
TEX). 2 mi N of Kenton, Hopkins 6 van Valkcnhurgh 5754 (NY, RM, SMU). Custer Co.:
1 mi W and 0.3 S of Weatherford, Waterfall 442 (OKLA, POM). Canyon rims, Clinton,
Dcmaree 12466 (ARIZ, GH, MO, NY, OKL, PH, POM, SMU, US). 10 mi W of Clinton,
Miinz ir Gregory 23508 (RSA). Ellis Co.: Near Shattnck, Clifton 3174 (GH, OKLA).
Greer Co.: 2 mi S of Mangum, Rohhins 3038 (NY, OKL). 3 mi S of Mangum, Stc])hens
20812 (DS). 4.5 mi S of Mangum, Towner 82 (DS). Harmon Co.: Near HoUis, Stevens
1162 (DS, GH, MO, NY, OKL, OKLA, US). 13.5 mi W of Mangum, Waterfall 7174 (OKL,
OKLA). Jackson Co.: 3 mi N and 1 W of Eldorado, Waterfall 9008 (OKL, OKLA). Kiowa
Co.: 3 mi W of Gotebo, Goodman 6274 (OKL, RSA, UC). Roger Mills Co.: Red lands,
Engleman 417, 418 (OKL). 2.5 mi S of Cheyenne, Wiedcman 183 (OKL, OKLA). Texas Co.:
Goodwell, Butler 85 (OKLA). 5.5 mi E of Hardesty, Stephens 6 Brooks 21775 (DS). 7 uu'
NE of Texhoma, Waterfall 9123 (GH, OKL, OKLA). texas: Brewster Co.: Glass Mts.,

Tharp 3629 ( US ) . 4 mi S of Alpine, Miinz 6- Gregory 23395 ( RSA, UC ) . Hot Springs area,
Spcrry 1732 (GH). 3.3 mi W of Alpine, Towner 28 (DS). Brown Co.: 8.2 mi S of Brownwood,
Towner 63 (DS). Callahan Co.: Ca. 17 mi SE of Abilene, Henderson 64-53 (DS). Coke Co.:
5.2 mi SE of Bronte, Raven 6- Gregory 19279 (DS). Coleman Co.: 14.3 mi N of Coleman,
Towner 75 (DS). Comanche Co.: 2 mi E of Comanche, Deckmeier 17 (LL, SMU, TEX).
Concho Co.: 2.5 mi W of Eden, Munz & Gregory 23425 (RSA, UC, WTU). 2.8 mi N of
Eden, Raven 6- Gregory 19277 (DS). Culberson Co.: 27 mi SW of White's City, New Mexico,
Mimz 6 Gregory 23364 (RSA, UC). 10 and 12 mi N of Van Horn, Waterfall 4095 (ARIZ,

GH, MO, NY). Victoria Canyon, Sierra Diablo, Correll <b Rollins 23783 (LL). 25.6 mi SW
of White's City, New Mexico, Towner 23 (DS). Glasscock Co.: 3 mi E of Garden City,
Munz ir Gregory 23420 (RSA, UC, WTU). Hardeman Co.: 1 mi N of Quanah, Stepliens
20721 (DS). Irion Co.: 30 mi N of Barnhart, Raven ir Gregory 19211 (DS). Lampasas Co.:
Lampasas, Reverchon 1302 ( DS, F, MO, NY, PH, UC, US). Maverick Co.: Eagle Pass,
Havard s.n. (US). Mills Co.: 0.9 mi N of center of Goldthwaite, Towner 70 (DS). Pecos
Co.: 25 mi NW of Sanderson, Munz ir Gregory 23405 (RSA, UC, WTU). Potter Co.: 3.1 mi
N of U.S. 66 on Farm Road 1719, Towner 93 (DS). 11.3 mi N and 3.9 W of central Amarillo,
Toicner 192 (DS). Presidio Co.: Ca. 35 mi S of Marfa, Bunton Flats, Warnock 46621 (RSA).
Ca. 3 mi SW of Marfa, Hinckley 706 (LL). 12 mi N of Shatter, Scuddy 396 (OKLA). Marfa,
Eggleston 17341 (US). Real Co.: Leakey, Pa/mer i074,9 (DS, PH, POM). Roberts Co.: 24 mi

gg ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. fil

S of Pcrnton, Afidcison 2988 (KSC). Taylor Co.: Camp BarkcU-y, Tohtcail 7065 (MO,
SMU, UC). Terrell Co.: 6.7 mi E of Sanderson, Raven <Lr Gregory 'l9202 (DS). 6.3 mi E
of Sandenson, Gregory 275 (DAO, RSA, UC). 9.6 mi W of Dr>den, Parks ct al 305 (LL,
SMU). Uvalde Co.: Sabinal, Palmer 11514a (MO). By Sabimil R., McKelvey IS79 (Oil,
POM). Uvalde, Dohic in 1930 (TEX). Val Verde Co.: Puinpville tunioff, WaruocJ; 1 1313
(LL, SMU). Ca. 5 mi W of Langtry, Wanwck 6 Cawcrou 9937 ( LL, SMU). Ward Co.:
Near Monahan.s, Wheeler in 1938 (LL). Wheeler Co.: Ca. 0.5 mi W of Shamrock, Towner 88
(DS). Comitie.s iiiikno\\'n: Near Mt. Carinel, Bio Crande, Parry 369 (NY, PII). On the Bio
Craiide, WrigJit in 1848 (MO). Between Austin and Stephensville, Kagan in 1966 (TEX,
biochemical voucher), new mexico: Cha\es Co.: Ca. 5 mi N of Roswell, Towner 126 (DS).
13.5 mi W of Hope, Towner 13 (DS). Ca. 8 mi E of Elk, Towner 12 (DS). Dc Baca Co.:
N side of Fort Sumner, Skinners 20922 (SMU). Dona Ana Co.: Organ Mts., Wootnn in 1900
(NMC, POM, BM, US). Eddy Co.: Near Three Forks of Bocky Arroyo, Cuadalupe Mts.,
Wilken 1734 (PII, US). L5 mi ENE of headquarters, Carlsbad Caverns National Park, Dole
74 (UC). Tonction of Delaware Creek and Pecos B., Pope in 1835 (CII). Memorial Hospital,
N end of Carlsbad, Mnnz b- Gregory 23355 (RSA, UC). Lea Co.: 1-11 mi N of Hobbs,
Pearcc 2569 (ARIZ). 60 mi E of Roswell, Palmer 66 (F). Lincoln Co.: Ca. 15 mi W of
Roswell, Dunn 8700 (RSA). 10 mi E of Capitan, lliteheock et al. 4201 (DS, UC, \V1"U).
Hando (Hondo?) Hill, Wooton in 1904, 1906 (NMC). Otero Co.: 9 mi NE of Alamogordo,
Munz 6 Gregory 23337 (RSA, UC, WTU). Qua>- Co.: Ca. 9 mi W of Tucumcari, Towner 94
(DS). 8 mi SW of Tucumcari, Skinners 21062 (SMU). 8 mi S of San Juan, Stepkens ir
Brooks 25573 (DS). Roosevelt Co.: Portales Springs, Maiiin 784 (WTU). Near Catisey,
Wooton in 1909 (NMC). Sierra Co.: Berendo Creek, Melealfe 1574 (F, CH, MO, NMC.
NY, POM, UC, US). Torrance Co.: 2.4 mi NW of Corona, Towner 124 (DS). Union Co.:
Ca. 4 nu' N of Moses, York 6 Roclgcrs 147, 149 (SMU, TEX), aiu/ona: Cochi.se Co.: 15 mi E
of Bernardino, Benson 10284 (ARIZ, POM, UC). 6 mi NW of Chiricahua, Gould 6 Ptiltz
3155 (ARIZ, GH, UC). 6 mi W of entrance to Chiricahua National Monument, Gregory 408,
411 (DAO, DS, RSA, UC, WTU). 3 mi E of Dos Cabczas, Maguire 11152 (13AO, Cl'l, NY,
UC, WTU). Mescal (ca. 7 mi W of Ben.son), Tkornher 4312 (ABIZ, OKLA, SMU). Dragoon,
Trogstadl 1068 (ARIZ, NMC). Ca. 3 mi E of Cochise Stronghold, Dragoon Mts., Towner 161
(DS). Cila Co.: 1 mi N of Black B., San Carlos Indian Bcscrxation, Goodmcm t~ Ilitckcoeh
1287 (DS, F, MO, NY, PII, POM, UC). 1 mi N of Blackriver Road, Granfelt in 1960 (ARIZ).
Between Clobe and Coolcy (Coolidgc?), Nelson 10372 (DS, MO, NY, mi.xture with C.
laiandulifolius). Pima Co.: Redington, Coodding in 1935 (ARIZ). Tucson-Rcdingtou Road,
San Pedro Valley, Brass 14282a (CH, NY). Pinal Co.: Near Oracle, Santa Catalina Mts.,
Lewis 1079 (RSA, UC). Peppersaucc Canyon, Santa Catalina Mts., Darrow in 1937 (W).

Hills near Oracle, Harrison 6 Kearney 6673 (US). 7.7 to 7.9 mi SE of Oracle, Santa Catalina

Mis., Toicner 1, 3 (DS). Santa Cruz Co.: Mustang Mts., Pringle in 1884 ( F, (^,11, MO, NY,

POM, US). Near Sonoita, Harrison i* Kearney 5713 (ABIZ, US). Sonoita to I'lgiu, Peebles ir

Ftdton 11485 (ABIZ, US). 7.5 mi E of S(moita, Gregory 404, 405 (DAO, RSA, VC). 7.5 mi SE

of Sonoita on road to Canelo, Towner 105 (DS).

Mexico, co.uiuila; Santa Rosa Mts., Marsk 1338, 1491 (F, OKLA, SMU, TEX). 27 mi

E of Boquillas, Hcnrickson 11611b (TEX). 64 mi W of Cuatro Cienegas, Henriekson 7861

(TEX). 4.5 km E of Matrimonio Viejo, Joknslon 10895 (TEX).

Most abundant of tlie races of this species, Calyl()))hiis haritcc^ii su])sp.
puhc.sccnfi occurs widely in Texas and neighboring states. In central Texas, it is
the only member of sect Salpin^ia. Spreading trichoines and broad truncate-
based leaves are strongK- correlated in tills form, and are characteristic of the
central Texas populations and most otliers wliich are not affected l)y introgr(^s-
sion. The type of Oenothera gTeggii is tentatively included here, although the
plants are stunted and somewhat lacking in distinctix'e characters. They are not
typical of C. hartwegii subsp. ptiheseem and may represent hybrids or intro-
gressants with subsp. hartwegii, which also occurs in that region. For these rea-
sons, I have followed Shinners' ( 1964) decisions and taken up the epithet
''puhescens' for this taxon.

I^"^J TOWSEn— C ALT IXIPHVS

87

Cahjiophus hartweoii subsp. puhesccm is one of the most variable taxa in
CahjJophm, mucli of this perhaps stemming from the infkience of introgression.
The size of the flowers and of lea\'es and other \'egetati\x* parts all xary widelv.
Leaf margins may or may not 1)e crinkled. Pnbescence may consist wholly of
spreading hairs or of these combined with shorter glandular or nonglandular
pubescence. These characters appear to vary in response to genetic exchange
with subspp. filifoJius and harfice<^u.

Records of pollinators reported by Gregory (1964; as Oenothera ^reg^ii) in
Terrell Co., Texas and Cochise Co., Arizona, included hawkmoths in the evening
at both sites (Ilyles lineata^ Mamhica quimiucmaeuhta, Spliiux dolU) and bees
in the morning at the Cochise Co. site (Megaehile, Melissodes, BomJms), On a
different date at the same locality in Arizona, I observed numerous halictid bees
{Dialictus, Agapostemon, Evylaem) and some bumblebees (Bomhus) gathering
pollen in the late afternoon. In the evening, hawkmoth {Uijles Jincaia and Man-
duca quinquemacidafa) visitation was fre(ineut. My field and greenhouse ob-
servations showed a range in anthesis times extending from 2 hours before sun-
set to about sunset. Thus flowc^rs in this subspecies may not always be open
early enough for significant afternoon \'isitation by bees. Ultraviolet-absorbing
regions on the petals were found to be of small to moderate size. Each of
four plants that w^cre artificially pollinated was found to be self-incompatible.

Tetraploidy and translocation heterozygosity are present in C. Jiartwegii subsp.
puhescens, with tetraploidy occurring in one (Pecos Co., Texas, Munz ir Gregory
23405) of the LS populations which have been examined. A population inter-
mediate betw^een subsp. puheseens and subsp. hartwegii (Brew^ster Co., Texas,
Munz ir Gregory 23401) was also found to be tetraploid by Kurabayashi et al.
(1962). Interchange heterozygotes comprised 15 of 20 plants examined for mei-
otic configurations, occurring in 11 of 15 populations. An average of 1.2 trans-
location heterozygosities per plant was calculated for this form, with a maximum
of 3, the most frequent number being 1. No extra diminutive chromosomes have
yet been observed in C. hartwegii subsp. puheseens. Configurations o])tained

1

from hybrids with other members of sect. SaJpingia exhibited 0-2 translocatioi
heterozygosities.

Most instances of introgression ha\e been discussed in earli(^r sections. Two
possible cases of hybridization with C. lavandtilifolitis were discovered for this
taxon, perhaps the only subspeci(\s of C. hartwegii which hybridizes in nature^
with C. lavandidifoUiis, One was a sight record from Cuadalupe Co., New^
Mexico, where some plants with slight resemblances to C. lavandtilifolius were
found in two populations of C. Jiartwegii subsp. pubescens. The second was a
single plant intermediate betwx^en the same taxa [between Globe and CooIey(?),
Arizona, Nelson 10372, (RM)] observed in a mixed collection of the two typical
forms. Numerous cases of sympatry without hybridization were observed,
especially with C. serrulatus in Oklahoma, New^ Mexico, and the northern Texas
Panhandle, and with C. herlandieri in central Texas and the Panhandle. Other
instances included populations mixed with C. hvandulifolius in Culberson Co.,
Texas and wdth C. tuhieida in Cha\ es Co., New Mexico.

gJJ ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

2. Calylophus lavandulifolius (Ton. & A. Gray) Raven, Brittonia 16: 2S6.

1964.

Oenothera lavamhilac folia Torr. & A. Gray, Fl. N, Anier, 1: 501. 1840. O. harfwegii Brnth.
van Javandnhejoha (Torr. & A. Gray) S. Wats., Proc. Anier. Acad. Arts 8; 590. 1873.
Galpinsia lav amhdae folia (Torr. & A. Gray) Small, Fl. S.E. U.S. 845, 1335. 1903.
Oenothera hartwegi var. fendleri (A. Gray) A. Gray subvar. lavandtdae folia (Torr. &
A. Gray) II. Li'v., Mono^r. Onoth. 334. 1908. O. lavandtdaefolia var. typiea Muuz,
Aiucr. J. But. 16: 704. 1929. Cahjiophus hartwegii (Bonth.) Raven var. lavandtdaefolius
(Torr. & A. Cray) Shinners, Sida 1: 345. 1964. Oenothera lavanduUfolia \'ar.
lavandidifolia; Munz, N. Amcr. Fl., scr. 2, 5; 138. 1965. CahjJophis haiixcegii snlisp.
lavanduhfohus (Torr. & A. Gray) Towner & Raven, \L\drono 20: 243. 1970.

Oenothera lavandtdaefolia Torr. & Graj' var. gJaudulosa Munz, Amer. J. Bot. 16: 705, 1929.
GaJ})insia hwandidae folia (Torr. & A. Gray) Small var. glandulom (Munz) Moldenke,
rhytologia 2: 134. 1946. type: United States, Nevada, White Pine Go., Ely, 30 July
1923, M. E. Jones (POM).

Similar to Calylophus hartwegii. Stiff riitescent perennial from a stout woody
cauclcx, caespitosc, sometimes appearing more or less tufted; stems several to
many, moderately branehed, spreading-decumbent to more or less ascending,
0.4-2(-3) dm high; plant densely gray-strigulose throughout. Leaves dense on
the stem, sessile, usually ascending, linear to narrowly lanceolate or narrowly ob-
lanceolate, 6-50 mm long, 0.8-6 mm wide, the tip acute or obtuse, the base acute-
attenuate, the margin entire or nearly so, occasionally slightly undulate, in-
frequently revolute; small axillary leaves present, 2-10 mm long; lowest stem
leaves somewhat wider and more oblanceolate than above. Floral tube 25-60
mm long, 5-15 mm wide at the throat minutely strigulose or glandular-pubescent
witliout, sometimes with purple longitudinal lines and base, occasionally fading
pinkish upon wilting. Sepals 8-20 mm long, 3-8 mm wide, with free tips 0.3-3
mm long, usually with purple marginal stripes. Petals 12-28 mm long, similar in
vvidtli, usually fading pinkish to purplish, liighly ultraviolet-reflective, with a
small basal ultraviolet-absorptive spot, rarely medium-sized. Filaments 6-12
mm long; anthers 5-11 mm long. Style 30-75 mm long, glabrous alcove, minutely
pubej:cent below; stigma 2-5 mm broad; ovary 4-16 mm long, 1-2 mm wide.
Capsule 6-25 mm long, 1-3 mm wide; seeds 1.5-2.5 mm long. Self-incompatible.
Gametic chromosome nmnber, n — 7.

type: Unitkd States. On plains, probably along the South Platte River in

southwestern Nebraska or northeastern Colorado, June or July 1820, Edwin
James (PII). The approximate locality was taken from McKelvey (1955: 213-

219) and is at the eastern limit of the range of this species.

Distribution (Fig. 17): Local and sparse, on sandy and rocky, often cal-
careous soil, on high x^k^ins and in mountains, frequently with Juniperus, Finns
monophyUa or ri)ius ednlis, Cercocarpus, Artemisia tridentata^ occasionally in
lower zones wath Larrea divaricata or in higher zones with Pi7ms ponderosa, from
southern Fall River Co., South Dakota, southeastern Wyoming, and far western
Nebraska, through w^estern Kansas, Colorado, eastern and southern Utah, north-
w^estern Oklahoma, and the Texas Panhandle to Trans-Pecos Texas, central
Nuevo Leon, central New Mexico, central Arizona, and east-central and southern
Nevada. Occurring from elevations of ca, 600 m (2 mi W of Hays, Ellis Co.,

1977]

TOWNER— CAAVLOrHL/S

89

FicuBE 17. Distributions of Calylophus toumcyi (triangles), C. hvandulifoUus (dots),
C. tuhicula subsp. ttthicula (open circles), and C. tuhicula subsp. strigiilostis (sqtiarcs).

90 ANNAT.S OF TllK MISSOURI BOTAMCAL GAKDKN LVoi.'. 6l

Kansas) to ca. 2,750 m (Lcc Canyon, Spring Mts., Clark Co., Nevada). Flowers
April to August.

F

Rcprcst'iUativc spc'cimeiis oxamimHl:

Unitki) SiAiKs. SOUTH ixvKoiA: Fall Ri\cr Co.: Rocky clr> lid^cs, Over lS()7i (RM).
wYOMiNc:: Coshcn Co.: 4 mi N of Lu Crungc, Sfcphois 6 Brooks 22903 (DS). Laramie Co.:
IIart\illo. Nelson H32H (CII, MO, NY, I'S). 20 mi W of Pine Bluffs, Porter ij Porter 8153
(DS, I'OM, UC). SK vdgv of Pine Blnffs, Stephens I- Brooks 22H7S (DS). Platte Co.: Wlialen
Canyon, Nelson 526 (CII, MO, XY, US). Near Cnernsey, Porter 3557 (DS, SMU, TEX, UC,
WTU). NKiiHAsKA: B().\ Butte Co.: 5 lui E and 5 N of IIenn"njfforcl, Slei'Jiens i^ Brooks
24512 (KANU). Cha.se Co.: 10 mi N of Imperial, Broicn 1255 (NEB). Carden Co.: 2 mi S
of Lewelleu, Stephens ir Brooks 11519 (KANU). Morrill Co.: Angora, Pool in 1912 (MO,
NEB). 1 mi N of Bro;ul\vafer, Stephens ^ Brooks 13907 ( DS, KANU). Scott'.s Bluff Co.:
1.5 mi W and 5 S of Melbeta, Stephens 5iH4 (K.ANU). Colorado: Baca Co.: 9 mi S and
2 E of Walsh, Stephens b Brooks 21H2() (DS). Bent Co.: 1.5 mi SE of Las Animas, Stepliens
6 Brooks 22003 (DS). 4.5 mi W of Prowers, Stephens 6 Brooks 21989 (DS). Dolores Co.:
Just W of Northdale, Anderson 3138 (DS). Huerfano Co.: 19 mi NE of Walseuhurg, Steiiliens
ir Brooks 22229 (DS). Kit Carson Co.: 5 mi E of Flagler, Stephens 6 Brooks 22633 (DS).
Las Animas ("o.: 6 mi N and 4 E of Audrix, Stei>hens & Brooks 21900 (DS). Otero ('o.:
2 mi S of Manzanola, Stepliens 6 Brooks 22315 (DS). Sedgwick Co.: 1 mi S of Jnleshnrg.
Stephens 6- Brooks 24060 (KANU). k.\n.sa.s: Clark Co.: 8 mi N of Ashland, llorr E248
(COLO, DAO, F, CM, KANl\ LL, OKI., HM, SMU, TEX, UC). Ellis Co.: 2 mi \V of FLiys,
Bonthj 77 (ARL/, V, WO, NMC, OKL, PIL RM, US). Coxe Co.: 20 mi S and 3 E of Oaklev,
Lalhrop 3374 (KANU, S\fU). Meade Co.: 12 mi E of Meade, Uorr 3532 (KANU, TEX, US).
Morton Co.: 7 mi N and 4 W of Elkhart, Stephens 8877 (KANU). Scott Co.: ITorscthief
Can>()n, Scott County State Park, Fearing & Lalluini in 1950 (C.H, KANU). Trego Co.: 14 mi
S of Ogallah, McGrefior 17124 (KANT^). oklaiio.ma: Beaxer Co.: E edge of Elmwood,
Stei>hens 6- Brooks 21745 (DS). Harper Co.: 10 mi S of Buffalo, Goodman 2394 (MO, NY,
OKL, POM, UC, WTU). T(>xas Co.: 5.5 mi E of Hardesty, Stephens 6 Brooks 21776 (DS).

Woods Co.: Near Freedom, Stevens 252 (DS, OH, NY, OKL, SMU). texas: Brewster Co.:
Foothills of Class Mts., 7.7 mi NE of U.S. 90 cm U.S. 67, Towner 29 (DS). Culberson Co.:
25.6 mi SW of White's Cit> , E .side of Cuadirlupe Mts., Towner 24 (DS). Hudspeth Co.:
31 mi E of El Paso, Hueco Mts., Tharp 46071 (F, MO, TEX), new mexico: Colfax Co.:
Near Raton, Nelson 6 Nelson 4681 (DS, RM, US). Edd> Co.: Near Three F'orks of Rocky
Arroyo, Cuadalnpe Mts., Wilkens 1711 (PH). Socorro Co.: Hell Canjon, NLigdalena Mis.,
Ilerriek 274 (US). Tovranee Co.: 3.0 nu' NE of Duran, Raven 19129 (DS). 5.8 mi SW of
Duran, Raven 19133 (DS). Ca. 7.5 mi W of Willard on U.S. 60, Towner 120 (DS). aiu/.ona:
Coconino Co.: Rim of C;m_\()n Diablo, Two Cuns, Deniaree 44216 (AlUZ, PH, RSA, SMU).
E rim of C;m>()u Diablo, Two Cuns, Towner 114 (DS). 10 mi SE of Tnl)a City, Peebles 13363
(CII, US). 10.9 mi S of Bitter Springs, Mosquin 6- Muscpiin 4247 (DS). Moja\e Co.: Ca.
1 mi from rim of canyon, Toroweap ^'alley, MeCIintock 52-512 (ARIZ, NY). Navajo Co.:
45.0 mi NW of Concho, Towner 115 (DS). utah: l^uchcsnc Co.: Juniper zone, below Moon
Lake, Craham 6412 (MO, POM). Emerv Co.: 50 mi N of Hanksville, San Rafael Swell,
Cronqnisl 9201 (D.\0, 13S, NY, POM, WTU). Carfield Co.: Red Canyon, 10 uu' W of Br\fe

Cau>un, Preece 2480 (COLO, POM, SMU). Bryee Canxou. Goodman ir Iliteheoek 1566 (DS,

CH, PO.\l, RNL UC). 10 mi E of Escalaute, llolmm'n ^ Nielsen 7734 (DS, POM, RM, UC,
WTU). Millard Co.: Tunnel Springs, Desert Came Range, Cottam 8553 (ARIZ, POM). San
Juan Co.: Tuwa Caiuou, Natural Bridges National Montunent, Welsh ir Moore 2294 (NY).
Uinta Co.: Willow Crei-k, S of Ouray, Holmgren 1882 (KANU, WTU). Washington Co.:
10 mi N of the Bea\er l^am sunnnit of U.S. 91 and 5 mi NW of the highway, Wiens 3917
(WTU). xevaua: Clark Co.: Old Saw Mill site. Sheep Mts., 6,600 ft, Alexander 6 Kellogg
1757 (CH, UC, US, WTU). Rock)- ridge S of Deer Creek, Charleston (Spring) Mts., 2,670 m,
Clokeii ir Clokeij 7605 (ARIZ, CAN, COLO, DAO, DS, F, CH, KANU, MO, OKL, PII, PONL
RM, RSA. SMU, TEX, UC, US, WTU). 4.8 mi N of K>'le Canyon on road to Lee Canyon,
Charleston Mts., Towner 101 (DS). Lee Catnon, 0.7 mi W of jet. to Kyle Canvon, Charleston
Mts., Towner 104 (DS). Lincoln Co.: Panaca Valley and vicinity, Cenlni 131 (ARIZ, DS,
UC, US). White Pine Co.: 3 mi S of Ruth, Moore 346 (DS, POM). 5.1 u'li S of U.S. 50, on
eastern road to Hamilton, Raven 6 Solhrig 13550 (DS). 2 mi W of Ely, Anderson 2897 (KSC).
MEXICO. xuKvo LKON: 15 mi S of San Rolierto Junction on Mexico .57, Sanderson 291, 292
(TEX); Tnrner 6357 (TEX). 16 mi S of San Rol)erto Junction, Reveal et al. 2652 (DS).
Near summit of N.L. Highwa\ 60 W of ('.alcana, Sanderson 288, in part (TEX).

IHTTJ TOWSEn—CALYLOrnVS

91

The limits of this species are approximately those given by Munz (1929).
Some collections included here by Munz clearly belong with Cahjlophiis hart-
wcgii subsp. hdrficegii^ e.g., Zacatecas, gra\'elly soil, Purpns in 1903 (UC). The
two taxa are easily confused but differ in th(^ shorten' s(*pal tips, denser pubes-
cence, and leaves which are usualK' broader and obtuse-tipped in C. lavamJtili-
folius.

A species of broad distribution, C IcwanduUfoUus occurs for the most part
to the nordi and northwest of the range of C. Ji(irtwc<iii. Isolated collections
havc^ been made, however, throughout much of the range of the latter. Where
C. lavandiilifoUus occurs with C Jiartwcgii, it tends to occupy higher elevations
than any race of that species, except for C. Jidrticcgii subsp. fcndlcri. Plants of
C. lavandtilifolius are t}'picall>' slow -growing, small, and sparsely distributed, and
are relati\^ely inconspicuous when mixed widi populations of the more abundant
C harttce<i^ii.

Hybridization with other species of Calylophus appar(Mitly occurs only rar(*ly.
Evidence for gene exchange stems only from the two colleetioi^s mentioned
previously which were intermediate betwe(Mi C^ hivandidijolius and C. Jiartnc^ii
subsp. puhcscens. Instances of populations of C. hivandtdifolius contiguous with

those of other taxa have also been Tn(Mitioned above, and include contact with
C. tuhicida subsp. lidncuh, C. {uhicida subsp. sfrigidcsus, C. harticcgii subsp.
pid)csccm\ C. hart{ce<iu subsp. fcndlcr'u and C. serndatus.

Variation within C, JavanduJijoVius is not clearly correlated with geography,
and division into subspecies seems inad\isable. The glandular pubescence of
the floral tube* ,\\m\ calyx used by Munz to distinguish Ooiothcra lavandtdacfolia
var. iijandidosa does not appear to \ary in any meaningful pattern or in associa-
tion with any other character. Considerable variation within tlie species occurs
in leaf dimensions, leaf margin (undulate, revolute, or plane), width of the floral
tube, and in other lloral characters.

X'isitors to a population in Clark Co., Ne\^ada {Toicncr 195) eonsisted en-
tirely of hawkmoths (Ilijles Jincata^ Mcniduca), which were active between dusk
and dark. Anthophorid and halietid ( A<s.(i})ostci}ion) bees were seen on flowers
2 hours before sunset at a population in Torranee ('o.. New Mexieo (Tanner
120). Anthesis in the field and greenhouse ranged from 3 hom"s before sunset to

sunset. In the population in C^lark Co., Ne\ada the median time of anthesis was
V/j hours before sunset. Ultraviolet abs(jrptiou patterns on the petals tend to be
small in this species. The foregoing facts suggest that haw^kmoth pollination
predominates in C. lavandtdifo1it(s\ but that bees ma\- also play a role at certain
localities. Tw^o plants were self-pollinated and found to be self-sterile.

Cytological \^ariation in Ccdijlophus JacandulifoHns occurs in the form of
translocations and extra dinn'nutive chromosomes. Seven of 10 plants from 5 of
6 populations wvw interchange heterozygotes. An average of 1.2 heterozygosities
per plant was calculated. Three individuals from separate populations had extra
chromosomes, which consisted of 1 or 2 dimiimtive pairs. Configurations dis-
played by hybrids of C. lavandidifolius with other taxa in sect. S(dpingia showed
2-3 translocation differences and occasional in\'ersion differences between the

92 ANNALS OF TIIK MlSSOrUI BOTANICAL GARDEN [Vol. 64

parents, a greater cytological divergence than shown by other crosses within the

section.

3. CalylophiLS toumeyi (Small) Towner, comb. nov. — Fics. 1, 3.

Galpinsia totnucyi Small, Bull Torrey Bot. Club 25: 317. 1898. Oenothera harixicgu Bt-ntli.
\'ar. iounicyi (Small) Munz, Amer. J, B(it. 16: 708. 1929. O. toumciji (Small) Tidestrom,
Proc. Biol. Soc. Wash. 48: 41. 1935. CaUjlophus hartwc^ii (Benth.) Raven wiv. ioinncifi
(Small) Shinncis, Sida 1: 341, 1964. C haiiicc^ii .snhsp. tonmcifi (Small) Towner &
Raven, Madrono 20: 243. 1970.

Similar to Calylophus hartwegii. Snffrntescent perennial from a stont woody
candex; stems several, sparingly branched or unbranched above, ascending to
erect, 1.5-6( + ) dm liigh; plant subglal-)rous to minutely strignlosc throughout.
Leaves sparsely distributed on the stem, sessile, more or less spreading, narrowly
hinceolate, 10-35 nun long, 1-7 mm wide, tlie tip acute, the base acute-attenuate,
the margin entire to obscurely and sparsely serrulate, not undulate; conspicuous
fascicles of small leaves 2-25 mm long in nonfloriferous axils; lowest stem lea\'es
usually tending towards oblanceolate shape. Floral tube ( L5- ) 30-60 ( -70 ) mm
long, .5-14 mm wide at the tliroat, yellowish, fading orangish to brick red upon
wilting. Sepals 10-25 mm long, 3.5-6 mm wide, with free tips 2-9(-12) mm long,
colored as the floral tube. Petals 10-20 mm long, similar in width, intensely
lemon yellow, fading orangish to brick red, moderately ultraviolet-absorptive
throughout. Filaments 4-12 mm long; anthers 6-10 mm long. Style 35-70(-S0)
mm long, glabrous above, minutely pubc\scent below; stigma discoid to squarish,

1.5-4 mm broad; ovary 6-20 mm long, 1-2 mm wide. Capsule 10-50 mm long,
1.5-4 mm wide, thin walled, sometimes almost paper) , dehiscent only in the dis-
tal half; seeds 2-3 mm long. Self-incompatible. Gametic chromosome numlnM*,

n = 7.

TYiM-:: United States. Arizona: Cochise Co., Chiricahua Mountains, 25 July
1894, /. W. Tourney 197 (NY); Munz, Amer. J. Bot. 16: 708. 1929,

Distribution (Fig. 17): Local and uncommon on shaded rocky slopes or
disturbed areas in pine-oak forest, from the Santa Rita, Huachuca, and Chiricahua
mts. in Santa Cruz and Cochise cos., Arizona, and the Mogollon Mts. in southern
Catron Co., New Mexico, south through northeastern Sonora in the Sierra Madre
Occidental to west-central Chihuahua, From elexations of ca. 1,500 m (Stone
Cabin Can)^on, Huachuca Mts., Santa Cruz Co., Arizona) to 2,600 m (summit

of the Jose Mts., Sonora). Flowe
lations in Mexico as early as May.

J

Repn-siMitative specimens exainini'd:

UxiTEi> States. m:\v mexico: Catron Co.: On or near the West Fork of the Gila R.,
MoKollon Mts., Metcalfe 555 (ARIZ, CII, MO, NMC, US), .\rizoxa: Cochise Co.: Huachuca
Mts., JIanison 6 Kcarnvrj 5773 (US). Fort Huachuca, Wilcox hi 1892 (NY). Near Fort
Huachuca, Wilcox 253 (US). Huachnca Mts., 7000 ft, Jones in 1903 (DS, POM, US).
Huacliuca Mts., Tounwij in 1894 (CH, RM, US). Tanner's Canyon, Huachuca Mts., Gilhcii in
1892 (NY). Tanner's Canyon, Huachuca Mts., Lcmnion in 1882 (UC). Rainse>'*s Can>()n,
Huacluiea Mts., Goodding 786 ( RM, US). Miller Canjon, Huachuca Mts., Carter in 1936
(NMC). Carr Peak, Huachuca Mts., fiSOO ft, Benson 10500 (POM). Carr Peak, Huachuca
Mts., Goodding 222 (ARIZ, CH, NEB, NY, OKLA, RM). 5.6 mi up Carr Canyon road from
Arizona 92, Huachuca Mt.s., Towner 106 (DS). Reef Mine, Huachuca Mts., 7100 ft, Gould

UJ77I TOW \KH CALYLOPIiUS

93

1475 (ARIZ, UC). Garden Caii>()n, Iluacliuca Mts., Harrison i^ Kearney 5773 (ARIZ). Near
Fort Huachuca, lluachuca Mts., Lcmmon 2700 ( GH ) . Riicker Ganyoii, upper left fork,
Cliirieahua Mts., Blumcr 2025 (F). Sugar Loaf Mt,, Chiricahua Nits., Darrow in 1937
(ARIZ). N slope of Sugar Loaf, Ghirieahua National Monument, Clark 8280 (ARIZ). Sugarloaf
Trail just below tunnel, Cliirieahua National Moniunent, Towner 107 (DS). Near sninniit of
pass, Ghirieahua Mts., Gooddin^ 165-47 (ARIZ). Pine Ganyon, Ghiriealuia Mts., 6700 ft,
Blumcr 1610 (ARIZ, DS, F, MO, NLB, NMG, NY, RM, US). Ida Peak, along Telephone
Trail, Ghirieahua Mts., 8,000 ft, "Stone 517 (PII, RM). Pinery Can>'on, Ghirieahua Mts., 7,000
ft, Barr 64-353 (ARIZ). 12.4 mi W of jet. of Arizona 186 & 181, in Pinery Can) on, Ghirieahua
Mts., ea. 7,(){)0 ft, Toicner 164 (DS). 12.5 mi \V of jet. Arizona 186 ik 181, Towner 171. 1 mi
below Onion Saddle, E. side of GJiirieahua NUs., Kaiser 49-209 (ARIZ). Gr(^st Trail, Ghirieahua
Mts., 7,000 ft, Uernhrode 136 (ARIZ). Gut Saw Ganyon, Ghirieahua Mts., Cooddin^ 2339
(L'G). Wonderland of Roeks, Ghiriealuia Mts., Darrow in 1937 (GH, NY). Boiu'ta Gan\on,
('hirieahna Mts., Henderson in 1933 ( TEX). Outlaw- Ganyon, Ghirieahua Mts., Gooddin
2339 (RM). Pima Go. (?): Sabino Ganyon, C:atalina Mts., 3,000 ft, Jones in 1903 (MO,
speeimen inuuature; either not of tliis speeies or loeality incorrect). Santa Gruz Go.: Madera
Ganvon, Santa Rita Mts., 5,f)()() ft, Darrow 2614 (ARIZ). Madera Ganyon, Santa Rita Mts.,
Peebles et al 4545 (ARIZ, US). Stone Gal)in Ganvon, Santa Rita Mts., 5,000 ft, Tlwrnher in
1903 (ARIZ). Santa Rita Mts., 6,000-8,000 ft, Frin^^le in 1881 ( F, Gil). Santa Rita Mts.,
7,000 ft, Darrow t- Arnold in 1936 (MO, OKL). Lpper Madera Ganyon, Santa Rita Mts.,
Clark 12351 (GH, OKL). Alon^r trail from Momit Wriglitson to White House Ganvon, Santa
Rita Mts., 7,000 ft, Parker et al 5835 (ARIZ, \Y).

Mkxkx). soxoka: Puerto de los Aserrados, region of the Rio de Bavispe, White 3190
(ARIZ). Sunnnit of the Jose Mts., 8,600 ft, Meams 1606 (DS, US), Gananea, MurdocJi in
1914 (F). CHIHUAHUA: 48 mi W of Malachie on road to Ocampo, 8,400 ft, Wiens 3445 (GOLO,
DS). Mojaraehie (Maguaraehie?), KnoJjloeli 5094 (F). Mts. NW of Ghihuahua, Le Sueur in
1936 (MO, TKX, UG, US). San Jose de Pinal, Rio Mavo, 7,000 ft, Gentry 2587 (ARIZ, U, MO,
POM, UG, US). Mts. SW of Glmehuiehupa, Ilarfnian 712 (F, NY, UG, US). Mexican NW
railroad, km 85, Barlow in 1911 (F). Ridge between Rio Ghico and Rio Gahallo, Mexican NW
railroad, Goutinental Divide, Barloic in 1911 (F). Carretas, White 993 (ARIZ). L30 mi W
of Ghihuahua Gity, 8,500 ft, Russell in 1957 (UG). Ganon Huahnatan, 10 mi SF of Madera,
Muller 3428 (UC). Gnayauopa Ganyon, Sierra Madre, Jones in 1903 (POM). Salto de
Rahicora, I.e Sueur 1407 (F). W from Pearson (now Juan Mata Ortiz), Sierra Madre, Barlow
in 1911 (F)- Sierra Madre, Nelson 6088 (US). No .specific locality, Le Sueur 101 ( F, TFX).

This most distinct of tlie lai'<j;c-flo\vcrccl members of sect. Salnhiiiia is rcadilv
separated using any of several characters, including its exclusively montane dis-
tribution, fascicles of large axillary leaves, tall erect stems, long sepal tips,
partially deliiscent capsule, and imusual ultraviolet al)sorption pattern on the
petals. Apparently completely allopatric to the other taxa of sect. Salpingia.
Calijhphus toumeiji experiences no ciuTcnt genetic exchange with them. The
absence of any intermediate collectioirs and the number of characters which show
discontinuities from C. hariicegii indicate the validity of specific status for C.
toumeiji, which was not recognized in an earlier publication (Towner & Raven,
1970).

Distributed in the moim tains of southeastern Arizona, southwestern New
Mexico, and northeastern Mexico, this species is physically and ecologically
separated from other members of sect. Salpingia. The blooming period is un-
usual for tlie genus, occurring in late summer and early fall in Response to sum-
mer rainfall. Earlier flowering is prevented by the late, dry spring, characteristic
of luontane areas in this region.

Flower visitation to C. toutfieyi seems sporadic, as insects were seen only once
in significant numbers during several attempted studies. On that occasion, in
the Chiricahua Mts., of Arizona {Towner 238), bees of the genus Las\o<iJossum
were active gatliering pollen shortl) before dusk, and again after sunrise. Ilaw^k-

94 ANNALS Ol THE MISSOUHI BOTANICAL GARDEN [Vol. 64

moths were abundant visitors in the evening. Moi'phological characters and
anthesis times suggest that hawkmoths are the principal pollen vectors. The
floral tube attains a length of 70 mm in some specimens, the maximum seen in
Calylophus, and is perhaps a response to pollination by the genera of sphingids
with longer tongues (cf. Gregory, 1964). Moderately absorptive to ultraviolet
light over their entire area, the petals have no contrast pattern, nor do they
contrast with vegetative parts (Fig. 12). Anthesis occurs Vj to IMi hours before
sunset. Bees are therefore not likely to be regular and significant contributors
to pollination.

Self-incompatibility is probably typical for this species. Only one plant was
checked by self-pollination, and it was self-sterile. In one field study, a popula-
tion was observed to have set no seed in spite of having been in flower for several
weeks. The same population showed over 60% fertile capsules on the date of the

cited pollination study.

Three plants from two Arizona populations were examined cytologically and
proved to be heterozygous for two translocations apiece. Multivalent associa-
tions included a ring of 6 chromosomes in 1 plant and 2 rings of 4 chromosomes
in the 2 others. One of each type of configuration was found in a population
from the Chiricahua Mountains {Toicner 107), indicating that at least 3 trans-
location polymorphisms were present in the population. Chromosome determina-
tions from hybrids were not obtained because of the difficulty of crossing C.
toutrieyi and C. hartwegii and because of the scarcity of floral buds on those
hybrids which were produced.

4. Calylophus tubicula (A. Gray) Raven, Brittonia 16: 286. 1964.

Oenothera Uihicula A. Cray, PI. Wright. 1: 71. 1852.

Herbaceous or slightly suffrutescent short-lived perennial, arising from a
slender woody caudex; stems one to several, sparingly branched above, sub-
decumbent-ascending to nearly erect, 0.4-5.3 dm high; plant with short glandular
pubescence throughout, or with some parts minutely strigulose. Leaves ± dense,
subsessile, ascending, linear to ovate or obovate, 7-46 mm long, 0.7-11 mm wide,
the tip acute, sometimes obtuse in lowermost leaves, the base acute-attenuate,
the margin entire or sparsely and shallowly serrulate, occasionally shghtly un-
dulate; fascicles of small leaves 2-15 mm long in nonfloriferous axils; lowest stem
leaves more frequently oblanceolate than above. Inflorescence dense, with buds
crowxled near the stem apex; buds terete. Floral tube funnelform in upper one-
half or more, often tubular below, 5-25(-33) mm long, 3-10 mm wide at the
throat in pressed specimens, the inner surface glabrous above to densely pubes-
cent basally, yellow, sometimes fading pink, and more rarely, purj^lish. Sepals
3-13 mm long, 2-6 mm wide, with subulate free tips 0.5-2 mm long, plane, yellow,
rarely with purple marginal stripes, only infrequently fading pink or purplish.
Petals suborbicular to obovate-tnmcate, 5-20 (-25) mm long, similar in width,
infrequently fading pink to purplish, highly ultraviolet-reflective, with a large
basal ultraviolet-absoiptive spot. Stamens subequal; filaments 1-6 mm long,
glabrous to minutely pubescent; anthers 2-7 mm long, sparsely and minutely

I

l'J77] TOWNER— CALYLOW/C/.S

95

pubescent. Style 9-30 (-40) mm long, usually exceeding the stamens, glabrous
above, minutely pubescent below; stigma discoid to squarish, 1-2.5 mm broad;
ovary 4-11 mm long, 0.5-1.5 mm wide. Capsule 8-19 mm long, 1.5-2.5 mm wide,
moderately thin walled, completely dehiscent; seeds 1.0-1.4 mm long, obovoid,
angled, truncate at the apex. Self-incompatible. Gametic chromosome number,

n = 7.

I

type: United States, texas: prairies beyond the Pecos River, probably in
eastern Pecos Co., August 1849, Charles Wright 197 (in part = S2i; Gil).

Distribution (Fig. 17): Primarily on hmestone soils in arid lowlands, but
occasionally in montane areas, from Guadalupe Co., New Mexico, south to west-
ern Texas, thence northeast to Howard Co., Texas and south to northern Zacate-
cas, south -central Nuevo Leon, and southwestern Tamaulipas. From ca. 600-
1,800 m elevation. Flowers April to August.

Cytological relationships, interfertility, and a number of morphological char-
acters demonstrate a close affinity between Cahjlophus tuhicula and the other
members of sect. Salpingia. However, the short-tubed, fuuuelform flowers (Fig.

4) and morning anthesis of this species constitute a phenetic similarity to sect.

Cahjlophus, a relationship likely due to evolutionary convergence. Calylophus
tuhicula is distinguished from its closest relatives by those same characters. In
addition, individuals of this species are shorter-lived than in other fonns in the
section. The plants rarely have large taproots, and are difficult to maintain for
more than one or two years in cultivation.

Self-sterility was found in six plants from two populations of C. tuhicula
subsp. tuhicula, and no evidence of self -compatibility was seen in any cultivated
or wild plants of either subspecies. Stigmas were invariably well exserted and
no greenhouse specimens of either subspecies ever set seed spontaneously. An-

awn

of C. tuhicula subsp. tuhicula in Eddy Co., New Mexico and in cultivated rcp-
resentati\'es of both subspecies. Stamens, the stigma, and a large spot at the
base of each petal are ultraviolet-absorptive, contrasting markedly with die rest
of the petal surface, which is highly reflective.

Flower visitors at field sites {Towner 14, 15) of C. tuhicula subsp. tuhicula
consisted primarily of small halictid bees of several genera, especially Evylaeus,
Dialictus, and Agapostemon. These were most active from just before sunrise to
mid-morning, gathering both pollen and nectar. Infrequent visits by hawkmoths
(Hyles lineata) were observed shortly before sunrise. In some c(jloiiies, removal
of pollen was virtually complete by afternoon. Field data, anthesis times, and
flower morphology indicate that this species is predominantly bee-polliuated. A
strong inference can be made that C. tuhicula evolved from a moth -pollinated
ancestor resembling C. hartwegii. This is based on the relatively long floral tubes
in some C. tuhicula and on the close phenetic and cytogenetic relationship of the
two species.

Relative to other forms of Calylophus, C. tuhicula has a low level of chromo-
some heterozygosity. Eight of 14 plants examined had multivalents, and no
evidence of inversions, diminutive chroinosouies, or polyploidy was found. The

96

AN'XAI-S OF THE MISSOURI BOTANKIAL GARDKN [Vol. 64

average number of observable translocation heterozygosities per plant was 0.7,
as compared with 1.0 for the rest of sect. Salpingia and 1.9 for C. berhndieri.

Individuals from certain populations of C. tuhicula may approach or exceed
the minimum floral tube length seen in C. httrhcegii. In such cases, C, tuhicula
generally retains the wider funnelform shape of the tube. Other populations
show vegetative characteristics which suggest recent introgression with or deriva-
tion from C. hartwegU. These cases will be discussed under the subspecies.

Differences of pubescence, leaf shape, floral pigments, and ecological dis-
tribution distinguish the two subspecies listed below. The extremes of each
form are tiuite distinct, but individuals from several collections in Mexico cannot

ty

KkY to SUHSPECIKS

a. Clandular-pubescent througliout; leaves narrowly lanceolate to ovate; flowers rarely

fading reddish or purple _ - - - 4a. subsp. tubicula

aa. Minutely grey-strignlose on the ovary and upper stems; leaves linear to narrowly

lanceolate; flowers connnonly fading reddish or purple 4b. subsp, strigidosua

4a, Calylophus tubicula (A. Gray) Raven subsp. tubitula. — Fig. 4.

GaJpin^^ia tubicula (A. Cray) Small, Bull. Torrey Bot. Club 23: 186. 1896. Oenotlwra
hartwc^i Benth. var. tuhicula (A. Gray) H. Lev., Monogr. Onoth. 335. 1908.

Oenothera tubicula A. Ciray var. (fcmissa A. Gray, PI. Wright. 1: 71. 1852, type: United
States, Texas, Culberson Co., on the Guadalupe Mts., October 1849, Charles Wright 197, in
part = 13380 (GH, holotype; US, isotype). The colleeti(m probably came from Texas and
not New Mexico ( McKelvey, 1955; 1070).

Oenothera X serrulaloides II. Lev., Monogr. Onoth. 335. 1908. type: United States, Texas,
Pecos Co., valley of the Pecos and towards the Linipio, June 1851, Charles Wright 1077
(MO, holotype; GH, NY, PH, isotypes). Remark on type sheet: "mais hybride de

tubicula X serrulata?"
Galpinsia glandulifera A. Nels., Amer. J. Bot. 21: 575. 1934. type: United States, New

Mexico, Eddy Co., sandy hillsides, vicinity of Carlsbad Caverns, May 1930, Glailys Convis

36 (RM). Published erroneously as 37,
Galpinsia carlsbadiana A. Nels., Amer. J. Bot. 23: 269. 1936. type: United States, New

Mexico, Eddy Co., near the Caverns, Carlsbad National Park, 24 May 1931, Aven Nelson

11396 (RM, holotype; DS, NY, POM, isotypes).

With short glandular pubescence throughout. Leaves narrowly lanceolate to
ovate or obovate, 7-46 mm long, 0.7-11 mm wide, usually entire or nearly so.
Flowers rarely fading reddish or puiplish upon wilting. Self-incompatible.
Gametic chromosome number, n — 7.

Distribution (Fig. 17): Colonial, primarily on Hmestone soils, in flat arid
grasslands, often with Lcirrea cVwaricata and Yucca, from Guadalupe Co., New
Mexico, south in the western side of the Pecos River drainage to western Texas,
where occurring from Culberson Co. east to Howard Co., thence south through
Presidio, Brewster, and Terrell cos., and probably most of central Coahuila,
to northern Zacatecas, southwestern Nuevo Leon, and southwestern Tamaulipas.
Elevational distribution from ca. 600 m (10 mi E of Dry den, Terrell Co., Texas)
to ca. 1,400 m (between Santa Rosa and Vaughn, Guadalupe Co., New Mexico).
Flowers April to August.

H)771 TOW AKH— (.A/.y/.()/7/('S

97

Reprcsciitalivc .sxx'cimciis cxainiiR-d:

United States, nkw mkxico: Chaves Co.: Ca. 4 mi N of Roswell, Towner 127 (DS).
luldy Co.: 4 mi W uf Hope, Munz ir Gregory 23350 (RSA). Memorial Hospital, N end of
C^arlsbad, Midiz 6 Gregory 23353, 23354^ 23356 (USA, UC). O.G mi W of Hope, Toicncr 14
(DS). 1.7 mi NE of Hope, Towner 16 (DS). Ca. 4.5 mi S of Carlsbad, Towner 17 (progeny
only, DS). 7.6 mi NE of White\s City, Towner 18 (DS). Otero Co.: Ca. 5 mi W of Elk
(1 plant), Towner 108 (DS). texas: Bre\vst(M- Co.: 41 mi S of Alpine, Anderson 3030 (DS).
1^4ats j.ear Old Blue, Class Mis., Warnoek W524 ( DS, POM, TEX). 15 mi E of Marathon,
Munz 6 Gregory 23400 (KSA, UC). Ca. 10 mi E of Alpine, Sperry T1005 (UC, US). 6 mi S
of .\hirathon, Rollins 6 Chambers 2766 (DS, CH, POM, RM, UC). Culberson Co.: 3 mi SW
of New Mexico line on l^S. 180, Munz 6 Gregory 23360 (RSA). 9 mi E of Van Horn,
Waterfall 4162 (AHIZ, CH, MO, NY). Eetor Co.: W of Odessa, Lundell ^ Lmidetl 16D21
(LL). Jeff Davis Co.: 8 mi S of Fort Stt)ekton, Munz i> Gregory 22385 (RSA, UC). Peeos
Co.: "Mesa slope," 77^//; 43^731 (OKL, OKEA, RM, TEX, CC). Ca. 20 nn* W of Sanderson,
Warnoek 6 MeBryde 14004 (EL, TEX). 11 mi E of Fort Stockton, Warnoek 5164 (LL,
SMU). 30 mi XE of Fort Stockton, Owidjey 6 Ownhey 1625 (MO, POM, RM, RSA, UC).
i'residio Co.: Up to di\ ide between Long Draw and Capote Draw on road from Marfa to
Hm'dosa, Hineklcy (NV, SMU). Cleveland Ranch, near Chinati Mts., Uinekley (CH, NY).
5 jui N of .\[arfa, Munz 6 Gregory 23380 (RSA, UC). Reeves Co.: On U.S. 80, 3 mi E of
intersection \\ ith U.S. 290, Munz ir Gregory 23370, 23371, 23372 (RSA, UC, WTU). On
U.S. 80, 1 mi E of intersection with U.S. 290, X edge of the Da\'is Mts., Waterfall in 1043
(CH, MO, NY). Phihis W of Pecos, Tracy 6 Earle 144 (F, CH, MO, NER, NY, TEX).
4\Trell Co.: 10 mi E of Dryden, Parks et al 56 (TEX). Upton Co.: 20 mi SE of Crane,
Raven 6 Gregory 10240. Val Verde Co.: Pnmp\ille, Fisher 290 (US). Ward Co.: N of Pvote,
Lundell 6 Lundell 11370 (POM, SMU).

Mexico, zacatecas: 18 km W of Concepcion del Oro, Stanford et al 500 (DS, MO, NY).
SAN LUIS POTOsi(?): "Prov. de San Luis," Octoust 1050 (P).

Tills su1)species occurs primarily on arid calcareous flats and outcrops in
soudicastern New Mexico and w estern Texas. Its distribution in northern Mexico
is very poorly known, and may include much of Coahuila in addition to the
localities listed. In southern New Mexico, it occurs in the Pecos lliver \'alley and
on the plains to the west of it. Other taxa of Calijlophus supplant C. tuhicula on
the higher plains to the east of the river (the "Llano Estacado").

The broader leaves and glandular pubescence are the primary characters dif-
ferentiating C. tiibiciila sul)spp. tuhicula and stn^ulosus from one another. In-
trogression between the two seems to occur in Nuevo Leon, Tamaulipas, and
Coahuila. Intermediate specimens are cited under subsp. stri<i^ulosus.

Twelve plants froiu 9 populations have been examined duriug meiosis; 6 had
5 bivalents and a ring of 4 apiece, while 6 had 7 bivalents. Experimental hybrids
w^ere obtained with difficult}' between C. tuhicula subsp, tuhicula and C. her-
landicri subsp. hcrhnuVwri, These plants were weak and essentially sterile (0-1%
pollen fertility). They showed low^ chiasma frequency and gave evidence of at
least 6 major translocation differences betwx^en the forms. Crosses of C. tuhicula
subsp. tuhicula with other members of sect. Salpingia produced progeny wdiich
were structurally houio/Agous or heterozygous for 1 or 2 traiislocations. Such
crosses produced good seed set and reasonably fertile hybrids, although the hy-
1)nds frequently show(xliOoor germination and \iability.

Introgression Ix^tween this subspecies and C. hartwcgii sid)sp. fendlcri has

been imputed by Munz (1965). Actually, ccmsiderable differences in elevational

distribution separate C, harhccgii subsp. fendlcri and C. tuhicula subsp. tuhicula,

the former occurring from ca. L500 to 2,150 m elevation iu New Mexico. This

should make coutact between these taxa infrequent. In the Carlsbad Caverns

98 ANNALS OF THE MISSOUUI BOTANICAL GARDEN [Vol. G4

area, individuals with broad leaves and long floral tubes have been treated as a
distinct taxon, Galpinsia carhhadianu^ and as possible liybrids between the fonris
mentioned above. The latter possibility seems remote, since I have discovered no
records of C. hartivegii subsp. fencUeri within 160 km of the Caverns area. Taxo-
nomic recognition is unwarranted also for the reason that the long-tubed fonns
occur together with typical C. iubicula in many populations throughout the
range of subsp. tubicula. It is uncertain whether this pattern should be in-
terpreted as a result of present or past genetic exchange or as spontaneous vari-
ation within C. tuJnculcL Collections of C. tubicula subsp. tubicula showing long
floral tubes include the following: 4 mi N of Carlsbad Caverns, Eddy Co., New
Mexico, Porter 6- Porter 8986 (DS, RM; of all collections, tins is the most hybrid-
like, being very similar to C, hartwegii subsp. fcncUcri). 45 mi S of Pecos, Pecos
Co., Texas, Moore 6 Moore 21 (NY, SMU, UC). 1-5 mi NW of Notrees, Ector
Co., Texas, CoUim 82 (OKLA). Borrow pits N of Pecos, Ree\'es Co., Texas, Nel-
son 6 NeU'on 4989 (DS, MO, RM, TEX, UC, US).

In general, C. tubicula subsp. tubicula was not found growing together witl
other forms of Cahjlophus, perhaps because of its restriction to xeric sites at low
elevations. In Chaves Co., New Mexico, two individuals of C. hartice<^ii subsp.
ptihescem ( Towner 126) were found in a large population of C. tubicula ( Towner
127)^ but with no evidence of hybrids. Likewise, no hybrids were apparent in a
situation near the Class Mountains in Brewster Co., Texas, where C. tubicula^
C. lavanduUfolius, and C. hartwegii subsp. pubesccns were all discovered within
a 0.3 mile stretch of graded roadside. To the north of Rosw^ell, New Mexico, C.
tubicula was observed growing within a few miles of C. hartwegii subsp. filifolius
in apparently identical habitats. It can be inferred that these two taxa occasionally
come into contact. Their similarity in pubescence might well be a result of some
past genetic exchange, although I have seen no signs of current hybridization.

1

4b. Calylophus tiibitula subsp. strigulosus Towner, subsp. nov.

Diffcrt a subsp. tubicula ovario et caulibus superis minute strigulosis, foliis lincaribus \c'l
aiigustf lanceolatis, et floribus plerumque rul)esc(^ntibus vel purpurascentil>us.

Minutely grayish-strigulose on the ovary and upper stems, sometimes through-
out, glandular pubescence generally absent. Leaves linear to narrow^ly lanceolate,
l()-35 mm long, 0.8-3 mm wide, often shallowly serrulate. Flowers commonly
fading reddish to purple. Self-incompatible. Gametic chromosome number,

TYPE; Mt:xico. NUEVO LEON: Aloug Highway 60, 2 mi W of Caleana Jet., dry
rocky open range, 1,700 m, 5 July 1963, McGre<;or, Harms, Robinson, del Rosario,
6 Segal 119 (DS-504949, holotype; KANU, SMU, isotypes).

Distribution (Fig. 17): Uncommon in rocky open sites and canyons in rela-
tively dry montane areas, sometimes in pine forest; southernmost Coahuila, south-
central Nuevo Leon, and southeastern Tamaulipas. From ca. 1,500 to 2,300 m
elevation. Records of flowering include July and August.

Representative specimens examined:

Mexico, nuevo leox; 15 mi SW of Cialeana, Mueller iLr Mueller 46 i ( F, TEX), Hacienda

l^"^"] TOWNKW—CALYLOPHUS

99

Pal)lillo, Galeana, Taylor 68 (ARIZ, DS, MO, NY, TEX, UC). On canyon wall, 5,400 ft,
municipality of Galeana, CIkusc 7745 (ARIZ, F, MO, NY). Ca. 35 mi S of Galeana towards
Ascencion, Straiv ir Forman 1374 (RSA). Near sunnnit of Nuevo Leon Highway 60 W of
Galeana, Sanderson 287, 288 in part (TEX), coahuila; SE of Saltillo, Clark 6710 (MO).
Fraile, 59 km S of Saltillo, Stanford et al 242 (DS, MO, NY, UC). tamaulipas: 3 mi N of
Miquihuana, Stanford et al 2472 (DS, NY, RSA, WTU). Jamnave Valley, kelson 4461 (US).

The ecological distribution of Cahjlophus tuhicula subsp. strigulosus, which
occurs in luontaue areas of northeastern Mexico, often with pines, contrasts
shaiply with that of subsp. lu])icula, Tlie feature in common between the two
distributions is perhaps aridity, although tlie Mexican sites would appear to be
more mesic.

Several collections assigned to this subspecies show similarity to subsp.
tuhicuJa, Typically such resemblance consists of combinations of strigulose and
glandular pubesence or of glandular pubescence and leaves as found in subsp.
strifiidosiis. Of the collections cited above, intermediacy is shown by Clark 6710,
Nelson 4461, and Stanford et al 2472. Populations in the area of the type lo-
cality seem the most distinct from subsp. tu])icida.

Tw^o plants cultivated from seed had 3 bivalents and 2 rings of 4 chromo-
somes at meiosis, each thus being heterozygous for 2 translocations. These speci-
mens showed morning anthesis similar to that seen in subsp. tuhicula. No field
collections or observations were made.

Introgression between C Jiartwe^^ii subsp. liartwegii and C. tuhicula in north-
ern Mexico may have given C. tuhicula subsp. strigulosus its distinguishing
characteristics. Such hybrid origin may have occurred quite recently since the
two forms still occur in close proximity and in similar habitats in the Sierra Madre
Oriental. The strigulose pubescence, greater prominence of anthocyanins, and
narrow leaves of subsp. strigulosus in comparison to subsp. tuhicula all represent
points of similarity to the local populations of C. hartwegii subsp. hartwegii.
Of the few collections we have from this area, one set seems to be fully inter-
mediate with C harticcgii, ha\ing a floral tube length of 19-21 mm [15 mi SW
of Galeana, Nuevo Leon, Mueller ir Mueller 464 (F, TEX) ].

S(ction II. Calylophiis.

CaJijJophus Spach, Hist. Xat. Veg. Phan. 4; 349. 1835. Oenothera subgcn. Cahjlophus (Spach)
Torr. & A. Crav, Fl. N. Anicr. 1: 501. 1840.

Meriolix Raf. tw Endl., Ccu. Pi. 1190. June 1840; Raf., Anier. Montlilv Mag. & Crit. Rev. 4;
192. 1819, num. tnid. Raf., J. Phys. Chim. Hist. Nat. Arts 89: 259. 1819, iioni. nud.

llerbaeeous to suffruteseent perennials or annuals, 1-8 dm high, glabrous to
strigulose or strigulose-caneseent. Leaves 1-9 cm long, subentire to spinuose-
serratc. Inflorescence dense, with buds usually crowded at the stem apex; buds
squarish in cross-section. Flowers opening near sunrise. Floral tube funnelform
and somewhat s(|uarish in cross-section distally, tubular in proximal one-third to
one-half of length, 2-20 mm long. Sepals with keeled midribs. Petals suborbicular
to obcordate. Stamens biseriate, the episepalous filaments about twice as long
as the epipetalous filaments. Capsule often tardily dehiscent, sometimes sHghtly

recurve

d.

JQQ ANNAKS or TlIK MISSOIUU BOTANICAL GARDEN [Vol. G4

TYPE SPECIES: Cahjlophus scrnilatiis (Nutt.) Ruven.

5. Calylophus beilandieri Spacli, Ann. Sci. Nat. Bot., ser. 2, 4: 272. 1835.

llcrbaccons to sulfrutcsccnt perennial or annual arising from a woody caudex;
stems one to man}', simple to moderately branched, subdecumbent to erect, 1-8
dm high, glabrous to strigulose or strigulose-canescent, especially above. Leaves
sessile or indistinctly petiolate, soinetinies early deciduous below, spreading to
more or less ascending, linear to narrowly lanceolate or oblanceolate, often folded
lengthwise, 1-9 cm long, 0.1-0.9 cm wide, usually not much reduced up the stem,
the abaxial surface glabrous to strigulose-canescent, especially at the base, the
adaxial surface glabrous to sparsely strigulose, the tip acute, the base attenuate,
the margin subentire to spinuose-serrate; fascicles of small leaves to 20 mm long
often present in nontloriferous axils; lowest stem leaves narrowly oblanceolate
to oblanceolate or even spatulate. Inflorescence more or less compact, with buds
usually crowded at the stem apex and one to several flowers fresh at one time,
sparsely and minutely strigulose to densely strigulose-canc^scent; buds squarish
in cross-section. Floral tube funnelform and somewhat squarish in cross-section
distally, tubular in proximal one-third to one-half of length, 5-20 mm long, 3-14
mm wide at the throat in pressed specimens, subglabrous to strigulose-canescent
without, especially along the midribs, within glabrous distally and nnnutely
pubescent to strigulose l)asally, pale yellow green, sometimes l)lue black within,
rarely fading pinkish. Sepals 4-12 nun long, 2-7 mm wide, with subulate free
tips 0-4 nnn long, w ith raised or keeled midribs, subglabrous to strigulose-canes-
cent, pale yellow green, occasionally with red midribs and tips, only rarely fading
pinkish. Petals suborbicular to oboxate-truncate or obcordate, 6-25 mm long,
7-30 mm wide, occasionally

mm wide, occasicmally becoming orangish to purplish on wilting, highly
ultraviolet-reflective, but with large basal ultraviolet-absorptive spot. Stamens
biseriate; episepalous filaments 2-8 mm long, the epipetalous filaments 1-4 mm
long; antliers 2-7 mm long; pollen fertility normally 85-1009^. Style 9-30 mm
long, glabrous above and glabrous to minutely pubescent basally; stigma discoid
to squarish, 1-3 mm broad, sometimes blue black, generally exserted to the ends
of the anthers or beyond; ovary 5-20(-27) mm long, 0.5-1.5 mm wide, minutely
strigulose to strigulose-canescent. Capsule 10-35 mm long, 1-2 mm wide, hard

and thick walled, completely and often tardily dehiscent, sometimes slightly re-
cur\'ed; seeds 1-1.8 mm long, sharply angled, truncate at the apex. Self-incompat-
ible. Gametic chromosome number, n — 7.

type: United States, texas: on shore of Espiritu Santo Bay, probably in
present Calhoun Co., March or May 1829, Jean Louis Bcrlamlier 539 — 1919
(P, holotype; Gil, ?II, isotypes). This locality is given on the Gray Herbarium
sheet and by Spach (1835b: 338). The date was determined from information

given by McKclvcy ( 1955 ) ,

Distribution (Fig, 18): Open, moderately dry areas on a variety of well-
drained soils, frequently calcareous, in southeastern Colorado, southwestern
Kansas, western and central Oklahoma, eastern New Mexico, Texas (except in
the northeast), Louisiana, north-central Coahuila, northern Nuevo Leon, and

1977J

TtnVXER— CALYLOrZ/rS

101

Fk.lhk 18. I^istrihutions of Cahjlophus hcrlmidicn subsp. hciiandieri (dots), C. hcrJmidwri
subsp. p'lnif alius (squares), unci iiitorgrades between tbe two subspecies (open circles).

102 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

perhaps northern Tamaulipas. From sea level to ca. 1,200 (-1,800) m elevation.
Flowering March to September.

An outcrossing species, Calylophus herjandieri is characterized by self-in-
compatibility, large flowers, and the absence of permanent complex structural
hybridity. Incompatibility systems of the S-allele type were demonstrated by
Linder & Brun (1957) in plants of this species. I tested 24 plants from 6 popu-
lations and found all to be self-sterile. The yellow flowers are morning opening,
open throated, and have petals with conspicuous contrast patterns in the long-
wave ultraviolet region (Fig. 14). A variety of diurnal and matinal insects, in-
cluding skippers, small butterflies, bees of various families, and beetles, were
observed as flower visitors. These groups probably make varying contributions
to pollination at different localities, but no highly specific relationship appears
to have evolved with any particular type of insect.

Analysis of meiotic pairing in this species demonstrated a high frequency of
translocation heterozygosity. Of 68 plants analyzed from 45 populations, 56 or
fully 82% were heterozygous for 1 to 5 reciprocal translocations. Tlie most fre-
quent classes of chromosomal types were those with 1 or 2 interchanges, but
several plants had as many as 5 translocation heterozygosities. No permanent
structural hybridity was proven in this self-incompatible species, but the high
frequency of heterozygotcs implies that some developmental or selective effect
probably reduced the numbers of homozygotes.

As reported by Towner & Raven (1970), Berlandier's type possesses highly
fertile pollen, indicative of the pair-forming, outcrossing species in this group.
A later examination of the pollen from isotypes of C. (IrummomJianus revealed
that those plants were only half-fertile. Further study of the specimens and their
localities made it ob\ious that they were complex structural heterozygotcs, thus
belonging with C. serruhtus. This leaves C. herhndieri as the earliest available
name for the outcrossing species.

The taxonomic separation of C. herlandieri from C. serndatus on the basis
of differences in cytology and breeduig systems rationalized a formerly confusing
situation in regard to geographical variation. As previously inteiprcted, these
forms presented a complex pattern of countering trends in variation of floral and
vegetative parts. In my treatment, there is less conflict in variation patterns, al-
though the situation is still not simple. The two species exhibit parallel geo-
graphic variation, and both display a wide range of statures, leaf dimensions, and
flower sizes. The parallelisms are seen primarily in vegetarive characters, whereas
the floral characters generally differ between the two species.

Calylophus herhndieri is polytypic, with two well-differentiated morphologi-
cal races. Calylophus herlandieri subsp. pinifoUus, a central Texas subspecies,
intergrades with C. herlandieri subsp. herlandieri in southern and west-central
Texas and, to a lesser extent, in the boundary between the Edwards Plateau and

the coastal plain.

Populations of this species occasionally occur together with C. tuhicula, C
hartwefiii, and C. lavandtdifolius, but I have seen no evidence of interbreeding.
Numerous test crosses performed on those combinations only rarely produced

1977] TOWSEh—CAIALOrnVS

103

\'ial')le offspring, and tliese were completely pollen-sterile. Sympatric occur-
rcMiees with C, serrnjatus are infrecjuent, and will be discussed in the section on
tliat species.

Key to Suu.speciks

a. Stoms several to many, suhdecnnihent to ascending, 1—4 dm tall; leaves 1-4 cm long

— - 5a. su])sp. hcrJandieri

aa. Stems one to several, subereet to erect, 3-8 dm tall; leaves 2.5-9 cm long

5b. snbsp. phiif alius

5a. Clalylophus berlandieri Spach subsp. berlandieri.

Oenothera hcrJandieri (Spach) Stend., Norn. Bot., ed. 2. 2: 206. 1841. Meriolix herhnulieri

(Spacli) Walp., Repert. Bot. Syst. 2: 79. 1843. Calijlophus cIrn?nmondiant{s Spach snl:)sp.

berlandieri (Spach) Towner & Ra\en, Madrono 20: 243. 1970.
Oenothera serrulata Niitt. var. typica sensn Munz, Amer. J, Bot. 16: 712. 1929, pro parte.

CaUJophus serruhitiis (Nntt. ) Ra\cn sn1:»sp. serridatus scnsu Shinners, Sida 1: 338. 1964,

pro parte. Oenothera serndata subsp. serrulata sensu Munz, N. Amer. FL, ser. 2, 5; 111.

1965, pro x^arte.
Oenothera serndata Nutt. var. pinifolia Engelm. ex A. Gray sensu Munz, Amer. J. Bot. 16: 715.

1929, pro parte. O. serrulata sul)sp. pinifolia (Engelm. ex A. Gray) Munz sensu Munz,

X. Amer. Fl., ser. 2, 5; 141. 1965, pro parte.
Oeonotliera serrulata Nutt. \ar, druniniondii Torr. & A. Cray sensu Munz, Amer. J. Bot. 16: 714.

1929, pro parte. O. serrulata subsp. drummondii (Torr. & A. Gray) Munz sensu Munz,

N. Amer. Fl, ser. 2, 5: 142. 1965, pro parte.

Perennial; stems several to many, moderately branched, subdecumbent to
ascending, 1-4 dm high. Lea\'es more or less crowded, Hnear to narrowly hmceo-
late or oblanceolate, 1-4 cm long, 0.1-0.6 cm wide, the margin siibentire to ser-

rate, and occasionally somewhat undulate; lowest stem leaves frequently oblanceo-
late to spatulate. Sepals often with only slightly raised midribs, the free tips 0-2
mm long. Inside of floral tube and stigma yellowish, never black. Self-incom-
patible. Gametic chromosome lunnber, n — 7.

Distribution (Fig. IS): Common on grassy prairies, plains, or low hills on
sandy, gravelly, and limestone soils in relatively dry areas, frequently with Pro-
sopis, Qficrcti.S' havordii^ and Opxintia^ from western Las Animas Co., Colorado,

Seward, Meade, and possibly Reno cos., Kansas, south through eastern New
Mexico, the Texas Panhandle, and western Oklahoma to Culberson, Ward, and
Crane cos., Texas, thence southeast near the Pecos and Rio Crande rivers to
the Gulf Coast, becoming widespread on the Coastal Plain north to Milam Co.,
Texas; also occurring in th(^ Santa Rosa Mts. of northern Coahuila and in north-
ern Nuevo Leon. From sea level near the Texas coast to ca. 1,200 m (Rita Blanca
Lake, Hartley Co., Texas), with one record at ca. 1,800 m elevation (12 mi S of
Trinidad, Las Animas Co., Colorado). Flowers March to September.

Representati\e speeimens examined:

United States. Colorado: Las Animas Co.; 12 mi S of Trinidad, Brenekle 48184
(SMU). KANSAS: Meade Co.: Just S of spillway at Lake Larrabee, Meade Co. State Park,
Bare 24 (KANU), Reno Co.: Hutehinson, Smyth 40 (US). Seward Co.: 14 mi NE of Lil)eral,
Ste})he7is 11206 (KANU). new mexic;o: De Baea Co.: 35,5 mi S of Ft. Sumner, Towner 130
(DS). Eddy Co.: 6 mi SW oi \Vhiies CAty. Munz ir Cre^ortj 23359 (VC, ]\S A) . IH mi SW
of White's City, Towner 21 (DS). Eddy or Lea Co.: Shinneries E of Carlsbad, Goodding

IQ4 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 61

45SG (ARIZ). Leu Co.: S of Jal, Bamchy 14482 (DS). Quay Co.: PoittM-, Su^^s in 1942
(NMC). Roosevelt or Leu Co.: Between Tatum and Portales, Gooddinp. in 1037 (ARIZ).
OKLAHOMA: Bcavcr Co.: 15 mi SW of Beaver City, Stevens 366 (DS, NY, OKL, OKLA, SMU,
US). Custer Co.: 1 mi S and 1 W of Weatherford, Waterfall 1593 (ARIZ, NY). Harmon Co.:
LI mi W of Vinson, Towner 87 (DS). Harper Co.: Supply, Demaree 12392 (Gil, OKL,
POM, SMU). Jaeksoii Co.: N bank of Red R., SW of Eldorado, Towner 138 (DS). 8.8 mi W
of Elmer, Towner 139 (DS). Kiowa Co.: 17.8 mi N of Altiis, Towner 78 (DS). Ro^er Mills
Co.: Antelope Hills, Coodnuin 2618 (MO, NY, OKL, POM). Texas Co.: 11 mi E of
Hardesty, Stephens & Brooks 21758 (DS). Tillman Co.: 2.6 mi W of Tipton, Towner 140
(DS). TE.XAs: Aransas Co.: Aransas Bay, Berlandier 567 = 1957 (CII, MO, POM, RSA).
Armstrong Co.: 15 mi S of Claude (Palo Duro Canjon), Stephens ir Brooks 25469 (DS).
Hailey Co.: 2 mi S of Miileshoc, Ferris ir Diinean 3397 (DS, MO, NY). Bastrop Co.: Bastrop
Park-, Warnoek 101 (TEX). Bexar Co.: 20 mi S San Antonio, Metz 678 (NY, RM). Brooks
Co.: 13.4 mi N of IIr1)I)ron\ille, Towner 57 (DS). 11 mi S of Falfurrias, Totcner 60 (DS).
5 mi N of Falfurrias, Garcia 49 (OKLA, SMU, TEX). Caldwell Co.: W of LuliuK, Crockett 218
(LL). Calhoun Co.: Port O'Connor, Tharp in 1930 (TEX). 7.5 mi W and 5.6 mi N of Port
O'Connor, Towner 179 (DS). 2.4 mi E of Seadrift, Towner 180 (DS). Callalian Co.: Ca. 17
mi SE of Abilene, Henderson 64-52 (DS). Childress Co.: Estelline, Beverehon 4307 (CH,
MO, NY, POM, US). Concho Co.: 2.5 mi W of Eden, Munz 6 Gregory 23426 (RSA).
Crockett Co.: 25 mi W of Ozona, McVaufih 8209 (LL, TEX). Dickens Co.: On side of
eanvon, S of U.S. 82, T.uudell 12979 (LL, TEX, US). Donley Co.: Jericho, Demaree 12130
(DS, NY, OKL, PII, POM, TEX). Dmal Co.: 15 mi E of Hebbrouville, Sandoval 6 MeCarl
7982 (OKLA, TEX). Caines Co.: 3 mi S of Scagraves, Tliarp in 1041 (CII, SMU, TEX).
Carza Co.: 2.5 nn' E of Post, Raven 6 Gregory 19304 (DS). Glasscock Co.: 3 mi E of Garden
City, Mtinz I- Gregory 23422 (RSA, UC). Goliad Co.: Near Goliad, Williams 9 ( F, PII).
Harris Co.: Spring, Gentry 865 (RM). Hartley Co.; 10.4 mi N of Channing, Boherls 35 (DS).
Hemphill Co.: 7 mi ENE of Canadian, Delso 122 (DS). Hidalgo Co.: 25 nn' N of Edinburg,
Clover 811 (NY). Hutchinson Co.: Near Stinnett, McFarland 13 (OKL, RM). Irion Co.:
Barnhart, Warnoek T343 (TEX, US). Jackson Co.: 13.4 mi W Palacios, Toicncr 176 (DS).
Jim Hogg Co.: 9.7 mi E of IIcbbron^ille, Towner 56 (DS). Jim Wells Co.: 20 mi N of

Fremont, Cahrera 102 (TEX). Kenedv Co.: El Toro I., Tliarp 49123 (in part, possibly a
mixture with C. serndatus- F, MO, OKLA, PII, POM, TEX, UC, US). 7.5 mi S of Ri\iera,
Towner 186 (DS). 17.8 mi N of Raymondx ille, Towner 192 (DS). Kleberg Co.: 0.8 mi W
of Riviera, Towner 185 (DS). Kinney Co.: 9 mi W of Brackettville, McVaugh 7685 (DS, F,
SMU, TEX). La Salle Co.: Near Cotnlla, S;»«// 6 Ar/icm/ 2J,94i ( NY). Lee Co.: Giddings,
Hall 209 (F, GH, MO, NY, POM, US). Lipscomb Co.: 23.7 mi N of Canadian, Rowell 10414
(DS). Live Oak Co.: Mikeska, Owens ir Parks 2407 (MO). Lo\ing Co.: Between Mentone
and Wink, Warnoek in 1052 (LL, SMU). Lubbock Co.: N of Lubbock, Demaree 7528A
(DS, RSA, SMU, TEX). Motley Co.: 5.2 mi WSW of Matador, Shinners 18668 (OKLA,
SMU). Nueces Co.: Bishop, Eifrig in 1926 (POM). Ochiltree Co.: 11.1 mi SE of Perrvton,
Towner 158 (DS). Palo Pinto Co.: 19 mi W of Mineral Wells, Warren 24 (DS). Pecos Co.:
30 mi W of Sheffield, Munz 13290 ( DS, POM, UC). PottcT Co.: 1 mi N of Canadian R. on
IIigh\\ay 287, Jefferson ir Jefferson 2676 (DS, F, MO, N1<:B, NY, RM, SMU, UC, US, WTU).
11.3 mi x\ and 2.0 mi W of Amarillo, Towner 91 (DS). Randall Co.: Palo Duro State Park,
Cory 13036 (LL, SMU). Refugio Co.: 8.9 mi W of Refugio, MeCarl 6831 (SMU). San
Patricio Co.: Near Malhis, McKelvey 1726 (GH). Ta>l()r Co.: 3 mi S of Gamp Barkelev.
Tolstead 7096, 41983 (MO, NEB, NY, POM, SMU, UC). Terrell Co.: 13 mi S of Sheffield,
Webster 130 (TEX). Tom Green Co.: 7% mi S of Christoval, Cory 50569 (NY, SMU).
Travis Co.: Austin, Tharp in 1938 (SMU, UC). Uvalde Co.: 6 nn' SE of U\alde, Munz
13316 (POM). Val Verde Co.: Ca. 1.9 mi from Del Rio, Traverse 2162 (SMU, TEX). Victoria
Co.: Inez, Palmer 9137 (DS, US). Ward Co.: 3 mi ENE of Monahans, MeVaugh 8186 (DS,
GH, LL, TEX). Webb Co.: 7 mi N of Laredo, Dickey 129 (SMU, TEX). Wheeler Co.:
11.5 mi E of Shaimock, Rowell 10080 (DS, RSA). Wilbarger Co.: 20 mi N of Vernon,
Towner 77 (DS). Willacy Co.: 1% mi from .shore at Port Mansfield, Webster 6 Wilbur
3074 (SMU, US). Wilson Co.: 5 mi N of Stockdale, Munz & Gregory 23443 (RSA, UC).
Winkler Co.: 11 mi W of Kermit, Raven ir Gregory 19228 (DS). Wise Co.: 3 mi N of
Bridgeport, Whitehouse 15278a (SMU). Yoakum Co.: 4.7 nn' N of Bronco, Towner 135 (DS).
Zapata Co.: 5 mi SE of San Ygnacio, Flores ir Flores 147 (TEX). Counties not known: From
Bejar (San Antonio) to Austin, Berlandier 479 = 1829 (GH). From Matamoros to Goliad.

Berlandier 1048 = 2478 (GH, MO, PH).

^■>~~i TO\\\v.n—(.: ALT i.oriius

105

Mexico, coamuila: Santa Rosa Mts., MarsJi 1351 (F, OKT.A, SMU, TEX). Mvizqiiiz,
MarsJi 110 (F, OKLA, TEX). IlaciViulu Mariposa, Mcpo. of Nhizquiz, Wijud I- Mueller 2(il
(ARIZ, MO, NY, US). Sniiimif of La Cucsta ^fa](•^la Mts., Reveal el al. 25()4 (DS). xukvo
i.i:(')N-: Lanipazos, Ef/(a/n/,v 356' (F).

This rclati\cly .sliort-Icaxed and low-staturcd .sulxspccies occurs oxer an ex-

tensive range on the plains of Texas and adjacent states. It is connnon hi tlie
Texas Panhandle and along the Culf Coast, and it is also found locall)- in sandy
areas of W'est Texas. As I ha\e definc>d it, Calylophm herlandwri subsp. l)cr-
kindicri incorporates elements from each of the dnee subspecies of ^Xlcnolhcra
.sx'n-f//c/frt" recognized by Munz (1965: 141-142).

For instance, the extremely narrow-l(\ned plants formerly known as Ocnolhcra
scrndata sub.sp. pinifolia are clearly \'ariants which can actually be found along
with broader-Iea\ed plants in populations of either sulxspecies of C. herlandwri.
My treatment of these forms is similar lo (hat of Shinners (1964), who did not
recognize subsp. pinifolia. viewing it as nierelx the extreme in a wide range of
variation, the latter due to "spontaneous mutation." The narrow-leaved plants
are most frequentl)' found in arenis of highly calcareous soil, including gypsum,
and they occur in the more arid portions of the range of C. hcrlandicri. Thus
their presence may well be due not to "spontatieous mutations," but to edaphic
selection factors.

In the past, narro\\-lea\'ed indi\iduals of C. hcrlandicri sub.sp. hcrlandicri
were often assigned to Oenothera scrndata var. pinifolia. Examples are the fol-
lowing: 5 mi N of Stockdale, \\jlson Co., Texas, Munz ir Gregory 23443 (UC,
RSA). 2 mi S of Mul(>sh()e, l^ailey Co., Texas, Ferris 6 Duncan 3397 (DS, MO,

N^'). Narrow-leaved iudixiduals are slightly more frecpient in C. hcrlandicri
subsp. pinifolius, although they are not representarive of that taxon as a whole.

The distribution of C. hcrlandicri subsp. hcrlandicri is divided into a Coastal
I^laius section and a Central Plains section. These two series of populations are
connected only tcnuouslv, this through a narrow zone alou-j; the Pio Crande

J -' ' "■"'" o

southwest of the Edwards Plateau. The coastal region is less severe in climate
than is the central region, but both areas are semiarid. Considering die breadd

of these separate ranges, Hieir distinctness, and their climatic differences, one
might expect the two .series of populations to have dixerged morphologically,
flowever, this does not appear to have been the case, as diey show complete!)'
ONcrlapping ranges of xariation in all characters wdiich I have examined.

In .spite of die morphological similarity of the two series, meiotic pairing in

hybrids .suggests diat considerable cytological divergence max- hax-e taken place.
Nine hybrids betxxcxn progeny of Roivcll 10414 (Lipscomb Co., Texas) and
progeny of Bohart 6 Thorp 65092H-1 (Victoria Co., Texas) shoxxed L./KI pollen
stainability of approximately 40%. Meiotic determinations of 2 of the hybrids
.shoxved chahis of 14 chromosomes, indicating the presence of at least 6 reciprocal
translocation differences betxveen the parents. Similar data were obtained from
crosses of RoxvelTs collection xx'ith Towner 185 (Kleberg Co., Texas).

The frequency of translocation heteroz)'gosity in natural populations of Caly-
lopJius herlamlieri subs^^. hcrlandicri xvas found to be extremely high. Of 38

106 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. fl4

plants examined, including some grown from field-collected seed, only 6 formed
7 bivalents at meiotic metaphase. The remaining 847o showed multivalent forma-
tion in various degrees, indicating structural heterozygosity. The average num-
ber of translocations per plant was 2.0.

Flower visitation was observed at one site in Potter Co., Texas (Tow^icr 91),
Anthesis occurred shortly before sunrise, at which time several hawkmoths {IlyJes
Jincata) visited flowers for nectar. At sunrise and afterwards, skippers, small
butterflies, and bees of small to medium size, e.g., Evylaeus and Agapostemon,
collected nectar and pollen from the flowers. Pollination was perhaps most ef-
fective with tlie anthophorid bees (Melissodcs and Anthophora), which were few
in numl)er, however. Oligolectic halictids probably made some contribution,
since they were common and because their appearance coincided closely \\n'th
anthesis times.

Tntergradation occurs between C. hcrUindieri subspp. herhmVwri and pini-
foUus in several areas. Along the eastern escai-pment of the Edwards Plateau,
few natural intermediates are found, apparently because the respective habitats
of the two forms arc separated by a relatix'cly sharp geographical discontinuity.
On the southern and western sides of the Plateau, intermediate forms are much
more frequent. There the geographical changes are more gradual, and broad
zones of hybridization are evident. On the western side of the range of C. ])er-
landien subsp. pinifolitis, numerous plants intermediate in stature and leaf length
are found: 25 mi \V of Ozona, Crockett Co., Texas, McVaugh 8209 (LL, TEX).
3 mi S of Camp Barkeley, Taylor Co., Texas, Tohtead 7096, 41983 (MO, NEB,
NY, PO, SMU, UC). 3 mi E of Carden City, Glasscock Co., Texas, Munz 6
Gregory 23422 (RSA, UC). Similarly, collections intermediate between C. her-
landicri subsp. pinifoJius and the Rio Grande Valley populations of subsp. her-
hndicri are relati\'ely fre(iuent. Several examples are as follows: Ca. 1,9 mi from
Del Rio, Val Verde Co., Texas, Traverse 2162 (SMU, TEX). 9 mi W of Brackett-
ville, Kinney Co., Texas, McVaugh 7685 (DS, F, SMU, TEX). 6 mi SE of Uvalde,
Uvalde Co., Texas, Muuz 13316 (POM).

0\'er much of its distribution, C. Jjcrlandieri subsp. herlandieri occurs with
or near populations of species of sect. Salpingia, In most cases of sympatry a few
plants of one form are found scattered in or near a colony of the other, and only
rarely are both forms common at any locality. The taxa I observed growing
together with C herlandieri subsp. herlandieri were C. luiHxcegii subsp. filifolitis
in Dc Baca Co., New Mexico and C. harticegii subsp. ptdyescens in the Texas
Panhandle. Also occurring with C. herlandieri subsp. herlandieri in the Pan-
handle and in western Oklahoma, but more distinct ecologically, are C. harticegii
subsp. fendleri and C. lavajuhdif alius. In West Texas there is some likelihood
of contact between C. tuhicula and C. herlandieri subsp. herlandieri, although
their edaphic restrictions seem to reduce this possibility severely. Lastly, local
sympatry may also occur with C. harticegii subsp. maccartii in the lower Rio
Grande \^illey, since the edaphic and geographical ranges of the two forms over-
lap there.

1^""] TOW'SER^CAlALOPflUS

107

5b. Calylophus berlandieii Spach subsp. pinifoliiis (Engclm. ex A. Gray)
TowncT, comb. nov. — Fic. 5.

Oenothera sermhia Nutt. var. pinifolia Kngelm. ex A. Cray, Boston J. Nat. Hist, fi: 189. 1850;
Munz, Amer. J. Bot. 16: 715. 1929. Meriolix serruhiia (Nutt.) Raf. (var.) pinifolia
(En^din. ex A. Cray) Small, Bull. Torroy Bot. Cliil) 23: 187. 189G. Oenothera serrulafa
subsp. pinifolia (Engclm. ex A. Ciray) Munz, N. Amcr. FL, scr. 2, 5: 141. 1965.

Oenothera eapiUifoIia Scheclcs Linnaca 21: 576. 1848. Meriolix capiUifolia (Schcclr) Small,
FI. S.E. U.S. 846, 1335. 1903. type: United States, Texas, Comal Co., Xe\v Braimfels,
April (1846?), Ferdinand Roemer (not loeated).

Meriolix hillii Small, Fl. S.E. U.S. 846, 1335. 1903. type: ITnited States, Texas, Edwards
Co., Frio Water Hole, 30 June 1895, R. T. ///// (XV).

MerioUx meJano<^httis Rydh. ex Small, Fl. S.E. U.S. 846, 1335, 1903. Oenothera ser~
rulaia Nutt. var. maenhila H. Lev., Monogr. Ouotli. 336, 339. 1908 (lectotype: Hel-
ler 1600, MO), typk: United States, Texas, Kerr Co., Kerrville, 19-25 April 1891,
A. A. Heller 1600 (NY, liolotype; ARIZ, MO, NEB, NY, PlI, POM, RM, SMU, UC,
US, isotypes).

Oenothera serndata Nutt. \ar. dnnnmondii Torr. &• A. Cra\ sensu Munz, Amer. J. Bot. 16: 714.
1929, pro parte. O. serndata sulxsp. druniniondii (Tmr. & A. Cray) Munz sensu Munz,
N. Amer. Fl., ser. 2, 5: 142. 1965, pro parte.

Oenothera serrulata Nutt. var. drummondii Torr. & A. Cra> f. flava Munz sensu Munz, Amer.
J. Bot. 16: 714. 1929, for the most part, excluding' the t\pe.

Calylophus serrulatus (Nutt.) Ra\en \ ar. spi}i\dosus (Torr. ^A. Crav) Shinners sensu Shinners,
Sida 1: 339. 1964, pro parte.

Calylophus drumnwndianus Spaeh subsp. druinniondiauus sensu Towner in Correll <Sc [ohuston,
Man. Vase. PI. Texas 1123. 1970, all, except for the type.

Annual to short-lived ptM-ennial; .stems onc^ to several, simple or sparsely
branehed, suberect to erect, 3-8 dm high. Leaves well spaced, linear to narrowly
oblanceolate or narrowly lanceolate, 2.5-9 cm long, 0.2-0.9 cm wide, the mar-
gin remotely serrulate to .spinuose-serrate; lowest stem leaves narrowly ob-
lanceolate. Sepals with conspicuously keeled midri])s, with free tips 0..5-4 mm
long. Stigma and inside of floral tube frecpuuitly deep blue black in certain
populations. Self-incompatible. Gametic chromosome nimiber, n = 7.

typk: UxrrEi) States, texas: Comal Co., rocky prairies, New Braimfels,
Aiwl 1846, Ferdinand TAndheimer 37 ^ 394 (Cfl, holotype; DS, MO, NY, PII,
RSA, US, isotypes).

Distribution (Fig. 18): Common on prairii^s and in open places in oak savan-
na, on rocky, clay, or sand)- soils, often calcareous, froni Blaine and Lincoln cos.,
Oklahoma, south through a narrow portion of nordi-ceutral Texas to central
Texas, wdiere it is wn'dely distributed, especially on the Edwards Plateau; also
occin\s locally in w^estern and 5:outheru Louisiana. Elevational distribution from
near sea level (Sulfur, Calcasieu Parish, Louisiana) to ca. 900 m (Sonora, Sut-
ton Co., Texas). Flowxu's mostly from March to June.

Hepresentati\'e speeiniens exauu'ncHl:

Untted States, .nussouhi: Jaekson Co.: Slieffield ( introdueed ), Bush 32S ( K, (;H).
OKLAHOMA: Blafue Co.: 3 mi W of Creenfield, IIo})ki}is 6 van Valkeidjur^h 4131 (OKL).
Caddo Co.: Rim of DeviFs Canyon, Hopkins et al 309 ( DS, I\ MO. OKL, OKLA, RM, SMU,
UC, WTU). Canadian Co.: 2 mi W of El Reno, Munz 6 Cre^^ory 23505 (RSA, UC, WTU).
Cleveland Co.: Norman, Demaree 12473 (MO, \V, OKL, POM*, RM, SMU). Cotton Co.:
fi.9 mi E of Walters, Towner 141 (DS). 9 mi W of Conumehe, Toxcner 142 (DS). Custer Co.:
Ca. 1.5 mi W of Custer, Towner 154 (DS). Lineoln Co.: 3.2 mi S of Perkins, Towner 149B
(DS). Logan Co.: 15 mi W of Cuthrie, Raven 6 Gregory 19462 (DS). 5.4 mi W of Guthrie,
Towner 151 (DS). McClain Co.: Blanehard (Johnson's Pasture), Demaree 13094 (MO, NY,

^Q^ ANNALS OF THE MISSOURI BOTANICAL GARDEX [Vol. 64

OKL, PII, rOM, TEX, UC, US). Okliilioina Co.: R mi W and 6 N of Oklalioma City,
Wahrfall 1423 (OKL, TOM). Seminole Co.: SE of Konawa, Rohhius 245<S (OKE, VC).
Stephens Co.: S of Comanclie, WcitcifaU 3679 (NY, OKL). texas: Austin Co.: RclKille,
Fisher 3>S1G (F). liastrop Co.: S of Bastrop, LumlcU 6- Luiulcll J03.55 (ARTZ, LL, POM,

US). Rcll Co.: 5 mi S of Temple, Wolff 4037 (F). Bexar Co.: San Antonio, CJcmens ir
Clemens 686 (NY, Til, POM, RM). Blaneo Co.: 2 mi S of Blaneo, Thaii) cl at 17T195
(RM, SMU). 5.5 mi N of Jolni.son City, Towner 68 (DS). 32.7 mi NW of San Marcos,
Toicuer 66 (DS). Brown Co.: 8.2 mi S of Brownwood, Towner 72 (DS). Bro\\n\\ood,
Ewin<: 26 (LL, SMU, TEX). Burnet Co.: Burnet, Fisher 40065 (ARIZ, NEB, TEX). O.ke
Co.: 2 mi E of Robert Lee, Shinncr<{ 31773 (SMU). Coleman Co.: 11.3 mi N of Coleman,
Toicuer 74 (DS). Comal Co.: New Brannfels, Umlheimer 809 (ARIZ, DS, F, CTL MO, NMC,
NY, OKL, PIT, TEX, UC, US). 20.9 nn' NW of San Marcos, Towner 64 (DS). Comanclie Co.:
Round Tt)p M(., E^gert In 1900 (MO). Coryell Co.: 13 mi E of Gates\ille, Jackson 4 (LL,
SMU, TEX). Crockett Co.: 30 mi N of Tuno, Wamock 15289 ( LL, TEX). Dallas Co.:
White Rock Lake, Lundell 6 Ltindell S53L (DS, GH, LL, POM, SMU). Dinnnit Co.: Carrizo
Springs, Palmer 33732 (MO). Ellis Co.: IVa mi N of Midlothian, Cory 53327 (DS, KANU,
WW, SMU, UC). Erath Co.: 3.5 mi W of Stephensville, Could 5644 (RSA, SMU, VC).

I'avette Co.: La Grange, llanisch in 1935 (TEX). Frio Co.: nn" NE of Pearsall, Lundell
13628 (I,L, TEX). Gillespie Co.: Bear Mt., CorreJI & Correll 12763 ( LL, SMU). Hmnilton
Co.: Hauiilton, Tharp hi 1941 (GIT). Harris Co.: Houston, FisJier 3481 (F). Hays Co.:
5 mi W of San Marcos, Grcgorti 419 (DS, RSA, US). 5.1 mi NW of San \hucos, Towner 62
(DS). 7.5 mi NW of San Marcos, Toicuer 63 (DS). Hill Co.: 8.5 mi NE of liillshoro,
Siduners T2494 (SMU). Hood Co.: Grandbury, "Naples School" 7174 (US). Irion Co.: 30
mi N of Barnhart, Baccn ir Cre<ior\i 19210 (DS). Karnes Co.: 3 mi SE of Karnes City,
Johns(m 857 (RSA, TEX). Kendall Co.: 19 mi S of Fredcricksbm )^^ Munz 6 Cregonj 23438
(RSA, \\T\J). Kerr Co.: -1 mi SW of Kcrrville, Cory 51763 (DS, SMU). Kimble Co.: No
locality, Tluirp 43-738 (TEX, UC). Kinnc> Co.: Ca. 30 mi SE of Brackcttville, Strother 240
(DS, SMU). Lampasas Co.: 1 mi S of Lampasas, Whitehouse 15384 (SMU). 1 lano Co.:
No locality, Lundell 6 Lundell 9050 (DS, GIT, LL, POM, SMU). 1.7 mi N of Llano,
Toicm'r 6',9 (i:iS). Nk-dina Co.: Hondo, Tilshry in 1903 (PII). Nhnard Co.: 10.3 mi N of
Menard, Racen ir Gre<i.ory 19273 (DS). McClennan Co.: Between Waco and McGregor,
York 46074 (TEX, UC). McCnlloch Co.: 9 mi SE Brady, Munz ir Gregory 23431 (RSA, VC,
WTU). MeMullcnCo.: No locality, Sc/n///^ «4 (US). Mills Co.: Cokllhwaitc, Fr;t''/vnn ^-2i
(MO, PIT, TEX, UC). San Saba Co.: 4 mi W of Pontotoc, Jones 24 (LL, SMU). Sutton Co.:
Sonora, Tharp in 1931 (TEX). Tarrant Co.: Lake Como, Rulh 30 (F). Tra\ is Co.: 9 mi
W of Oak Hill, Lundell i^ Lundell 8898 (DS, GIT, LL, NY, POM, RM, SMU, UC). Uvalde
Co.: 20 nn" N of Uvalde, Craves 9 (RM, RSA). Val Verde Co.: Devil's R. N of Del Rio,
Vilshry in 1903 (PIT). Wa.shinglcm Co.: No h)eahty, Brackett in 1938 (GH). Williamson
Co.: No localitv, York 46193 (TEX), loui.sjana: Acadia Parish: Prairies near Crowley,
Small 6 Wherry 11741 (NY). Calcasieu Parish: Sulphur, Palmer 7719 (MO). St. Mary
I'arish: Near Berwick, Small ir Wlwrry in 1925 (NY).

As ^^'itl^ CahjIopJius licrhiiuJicri siil^sp. herlandicri, sul)sp. pinifoUm m-
co\i)ondcs portions of sevoial of tlic taxa recognized by earlier antiiors. Tt cor-
responds rather closely witb Oenolhcra scrrulata sulisp. dnimmoiuJii except for
tlic inclusion of narrow-leaved individuals and exclu.sion of complex betcrozygotes.
As mentioned earlier, tlie t\pes seen of C. (Irumniondiamis were erroneously as-
signed to tlie oulcrossing .species in a preliminary report (Towner & Raven, 1970).
Tlie types actually belong with the complex structural heterozygotc C. scrnilatm,
and tlie name is reduced to synonymy. The collection used as the basis for
Oenothera sernilata var. dniiumopdii f. fhiva was al.so found to belong with
C. sernddhis, ha\'ing tlie small llowers and half-sterile pollen indicative of com-
plex structural heterozygosity.

Ccdylophus ])erhnd'wri subsp. pinifolius is distributed primarily from central
Texas to central Oklahoma, inhabiting more mesic areas than does sub.sp. her-
landieri. Most t\picall\' it is found in calcareous, rocky soil in oak savanna. Colo-

1977 J low \I:K CALYIXU'Iii S

109

nics occur in opcMi or clisliiil)cd areas in that liahitat and m prairies. This sub-
species is tlie only ]nenil)er of sc^ct. Calijlopluis occurring on tlie Edwards Plateau
in Texas.

Earlier reference was made to the narrow-leaved variants occurring in tlie
tw(^ subspecies of C. hcrhnulicri. TIk^ presence of these forms in both ta\a pre-
N'cnts the use of leaf proportion as a diagnostic or key character. However^ C.
bcrhindicri subsp. pinijolius is distinctly longer leaved than subsp. bcrlandicvi^
and the use of this character togetlier w itli dilferences in stature permits an eas)'
diagnosis of most iudixiduals. Subspecies p'uiifolius gen(U'all) has a slender tap-
root and is diilicult to maintain for more tlian a year in the greeiiliouse, leading
to the inlerence that it is probabi)' a short-]i\ ed perennial or annual in the field.
This contrasts with subsp. bcrhindicri, which is distinctly perennial o\er most
ol its range.

Some representati\'(*s ol this subspi^eies posscvss purplish black stigmas and/ or
inner siuface of the Horal tnbes. This extreniclv inteiesting character is present
as a pol)niorphism in many populations, but is restricted to those in south-
central Texas. The dark-j)igmented forms are especially frequent in Bexar,
Blanco, Comal, Gillespie, Ihus, KcMidall, Kerr, and Tra\n's cos. Some examph^s
of the variants include the following collections: Type specimen of McrioJix
niclcinoglottis llxdb. ex Small. Type spechnen of Mcriolix hillii Small. 19 mi S
of L^rcdericksburg, Kendall ('o., Texas, Miuiz 6- Gregory 23438 (RSA, W'TU).
7.5 mi NW of San Marcos, Ihus Co., Texas, Towner 63 (13S; most plants with
black stigma, some with black h\panthium). 5.5 mi N of Jolmson City, Jilaneo
Co., Texas, Toicncr 6S (l)S; plants with black stigma onl\ ). U.S. 87 near NW
city linnTs of San Antonio, Hexar Co., Texas, Klein 1671-1674 (DS). New Jh*aun-
fels, Comal Co., ^JVxas, Lindhcimcr 809 (ARIZ, DS, F, (dl, MO, NMC, NY, OKL,
PII, Tlv\, UC, US). 4 mi SW of Kerrxille, K(M-r Co., Texas, Cory 51763 (DS,
SMU).

Plants from populations of (\ bcrhindicri subsp. pi)iifoHiis from Oklahoma
differ UK^dallv from those in Texas in having; shallower leaf serrations and morc^
strigose pubescence on tlie upp(M- stems. This is especially apparent in the north-
einmost populations, e.g., those* in Logan, Oklahoma, McClain, and Cleveland

COS. In regard to stature, leaf dimc^rsions, and inost othcM" characters, these plants
are identical to individuals from Texas populations of subsp. pinifolius. Some ol
the largest-flo\vered meml)(M's of the species occur in these Oklahoma popula-
tions. Exainples of plants Iroin tins area ar(^ as follows: 5.4 mi W oi Cuthrie, Lo-
gan Co., Oklalioma, Towner 151 (13S). Norman, Cleveland Co., Oklahoma,
Demaree 12767 (OKL, POM, SMU). Blanchard, McClain Co., Oklahoma, Dcnia-
ree 13094 (MO, NY, OKL, PII, POM, TLX, UC, US).

As in C bcrhindicri sul)sp. bcrhindicri, this subspecies has a great deal of
translocation hetcrozxgosity. Of the 30 plants which luue been examined, 24
or 80% had ring or chain nndti\'alents at meiotie metaphasc L Of these plants,
those ha\in<j; 1 or 2 heteroz\<j;osities were the most frefiuent t\i:)es. Hie axeraiic
number p(M' plant was 1 .6.

Floral beha\'ior has ])ccu ()bs<M-\'(^d in Biown Co., Texas (Tow)icr 72) and in

the greenhouse at Staniord, Calilornia. Anthesis tot)k place shorth' after sun-

jJQ AWAT.S OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

rise. Insect visitors to flowers observed by nie, and by P. II. Raven and D. P.
Gregory in Hays and Kyle cos., Texas (personal commnnication), included
small butterflies, flies, skippers [Atalopedc.'^' canipestris (Boisdu\'al)], a variety
of beetles, especially cantharids and Acmacodcra (Buprestidae), numerous small
bees {A^a paste man, Augochlorella^ Dialictus, Evylaeus, and Halictus), and a
few large and medium-sized bees {Api% Bombiis, MegachiJe, Xylocopa, an-
thophorids). It seems likely that a broad spectrum of insects serves as pollen
vectors. Skippers, medium-sized bees, and beetles, judged by their abundance,
size, and contact with the anthers and stigma, may be tlie primary pollinating
agents for most populations. The black floral tubes and stigmas, which occur
together with large ultraviok^t patterns, may complement those patterns in
facilitating the orientation of visiting insects. This would be especially important
for any species of insect which lacked the tricolor xision of bees and could not
discrinu'nate in the ultraviolet region of the* spectrum.

With no e

/

gcthcr with C>\ liartwegii subsp. puhcscens in nuich of its range in central and
west-central Texas. No other member of sect. Salp'uigia overlaps significantly
in distribution with subsp. pitiifoUus. Minor geographic sympatry exists in south-
ern Texas with C hartivcgii subsp. maccartii and in western Texas with C.
tuhicuh subsp. tuhiciiJa and C. hartwegii subsp. fiJifoUus. In these cases, local
sympatrx' is (juite infre<[uent or absent, probably as a result of habitat segregation.

6. Calylophus serrulatus (Nutt.) Raven, Brittonia 16: 286. 1964.— Fig. 6.

Oenothera scrruhta Nutt., Cen. N. Anier. Pi. 1: 246. 1818. Meriolix scrmhtn (Nutt.) Raf,
Awww Monthly Mag. & Crit. Rw. 4: 192. 1819. Cahjhphus nuttallii Spadi, Hist. Nat.
Vog- Pl»an. 4: 350. 1835. Oenothera sernihita \ar. niittallii (Spacli) Turr. & A. Gray,
Fl. N. AiutT. 1: 501. 1840. Meriolix serrulaia var. nutlaUii (Spach) Waip., Rcpert. Bot.
Syst. 2: 79. 1843. Oenothera scrrulata \ar. tijpica Munz, Aiuer. J. Bot. 16; 712. 1929.
CaUjlophus serrulatn.s ( Nutt.) Raven var. serrulatus; Sliinncrs, Sida 1: 338. 1964.
Oenothera serndata subsp. serruJafa; Munz, N. Amer. Fl., ser. 2, 5: 141. 1965.

Oenothera Jeueoearpa Coinien ex Lelun. in Hooker, Fl. Bor. Amer. 1: 210. 1833. O. serruJata
Nutt. \ar. (hughisii Torr. & A. Cray, Fl. N. Amer. 1: 502. 1840. Meriolix serndata (Nutt.)
Raf. var. dou^la.s.sii (Torr. & A. Gray) Walp., Report. Bot. Syst. 2: 79. 1843. type:
Canada, Saskatchewan, connnon on limestone rocks on Red and Assiniboine rivers, Aujijust
1827, David Dou<ilas (K, lectot>pc).

Cahjlophus drummondiana Spach, Ann. Sci. Nat. Bot., ser. 2, 4: 272. 1835. Oenothera serrulata

Nutt. var. dnintmondii Torr, 6c A. Gray, Fl. N. Amer. 1: 502. 1840. O. spachiana Steud.,
Nom. Bot., ed. 2. 2; 207. 1841. Meriolix serrulata (Nutt.) Raf. var. drummondii (Torr. &
A. Cray) Walp., Report. Bot. Syst. 2: 79. 1843. M, drummondiana (Spaeli) Snuill, Fl. S.E.
U.S. 846, 1335. 1903. Oenothera serrulata subsp. drummondii (Torr. & A. Cray) Munz.
N. Amer. Fl., ser. 2, 5: 142. 1965, pro parte. Cahjlophus drummondianus subsp, drum-
mondianus; Towner in Correll & Johnston, Man. Vase. Pi. Texas 1123. 1970. typk:
United States, Texas, along the Rio Brazos ou the Texas coastal plain, i^robably between
Bra/os Co. and the coast, 1833, Thonuis Drummond (P, holot\pe; CH, NY, isotypes, but
uot Drummond 111. 79 at PII).
Oenothera serrulata Nutt. \ar. spimdosa Torr. & A. CIray, Fl. N. Amer. 1: 502. 1840. Meriolix
serndata ( Nutt. ) Raf. \'ar. spi}i\dosa ( Torr. & A. Cray) Walp., Rcpert. Bot. Syst. 2:
79. 1843. M. si)i}iulosa (Torr. & A. Gray) Heller, Contr. Herb. Frankl. & Marsh. 1; 70.
1895. Cahjlophus serrulatus (Nutt.) Raven var. sphndosus (Torr. & A. Cray) Shiuuers,
Sida 1: 339. 1964. Tvri:: United States, Oklahoma, vicinity of the Red R., probably r>ear
pre.sent-day Choctaw Co., May-June 1819, Thonias Nuttall, Torrej- Herb. [NY, lectot\pc;
PH, NY, isolectotypcs, but not "Red River, Nuttall," (CH), or "Arkansas," (PH)]. Some
of NuttalTs Red Ri\er and Arkansas collections, such as the CH specimen cited, are aetualK'

1 ^)"" I TOWSKW^CALYLOPIIVS

111

Calylophus hcrlamlicri. The type slieet lias the Leaxenworth specimen cited b} Tone\ and

Cray mounted next to the Red River type of Nuttall.

Meriolix intcnncdia Rydb. ex Small, Fl. S.E. U.S. 846, 1335. 1903. tvpk: United States,

Missouri, Atcliison Co., Watson, 7 Jmie 1894, B. F. Bu.sh 321 (NY, holotype; MO, OKL,
isotypes).

Oenothera scnulata Nutt. \ar. intei^rifolia H. Lex., Monogr. Onotli. 337, 339. 1908. tyi>k:
Ignited States, probably from southeastern Colorado, 1945, third expechtion of John C.
Fremont, no. 47 (MO); Munz, Amer. J. Bot. 16: 713. 1929.

Oenothera semilata Nutt. var. drummondii Torr. ik A. Cray f. fhva Munz, Amer. J. Bot. 10:
714. 1929. TVPt:: United States, Texas, Walker Co., Huntsvill(% B. C, Tharp H66 (FOM-
32801; other collections cited in the protolo^ue are C. herhmdieri subsp. pinifoUus).

Meriohx ohUmceolata Rydb., Brittonia 1: 93. 1931. tvpk: United States, Kansas, Comanche
Co., along road 2 mi W of Coldwater, 8 July 1929, P. A. Ri/dher^ ir It hnler 737 {K\\
holotype; KANU, NEB, NY, isotypes),

Cahjh)j)Jius serndatus (Nutt.) Raven var. w/izo/urf/.v Shinners, Sida 1: 338. 1904. typk: United

States, Arizona, Navajo Co., dry sandy riverbank 4 mi upstrt^am from White Ri\cr, 25 June

1951, S. ;. Preeee, Jr. 6 B. L. Turner 2692 (SMU).
Cahjlophti.s australis Towner & Raven, Madrono 20: 243. 1970. typk: United States, Texas,

Cameron Co., Texas route 4, 2.8 mi W of end of road at Boca Chica, 29 May 1969,

Towner IS7 (DS-612434, holotype; KSA, TEX, I'S, isotypes).

Similar to Cahjiopliiis berlandieri. Herbaceous to suffrutescent pererinial
from a woody caudex; stems few to many, 1-6 dm higli. Floral tube 2-12 (-16
mm long, 3-12 mm wide, never blue l)lack within. Sepals 1..5-9 nun long, 2-6
mm wide. Petals 5-12(-20) mm long, 5-15(-20) mm wide. Episepalous fila-
ments l-5(-7) mm long, the epipetalous filaments 0.5-3 mm long; anthers 1.5-4
(-7) mm long; pollen grains 30-80% aborted. Style 2-15(-20) mm long; stigma
1-2 mm broad, not exserted beyond the anthers, and often in eontaet with the
anthers at the apex of the floral tube, never blue blaek; ovary 4-13 mm long.
Self-compatible and highly autogamous. Gametic chromosome number, n = 7,
with a ring of 14 chromosomes or a ring of 12 plus a bivalent at meiotic meta-
pluise T.

type: United States. Plains along the Misscmri River, probably just nortli of
the Platte River in eastern Nebraska, April or May 1811, Thomas Nuttall (PII).
From Bradbury's account of the expedition (McKelvey, 1955: lb5-118) and
because of the innnature appearance of the type, these dates seem nif)re likely
than June, the date of flowering given in the original description.

Distribution (Fig. 19): Common on plains, in grassy open areas in woods, or,
rarely, in mountains, usually on sandy or rocky soils, from southern Alberta,
southern Saskatchewan, and southern Manitoba to eastern New Mexico, the
Texas Panhandle, and the Gulf Coast of Texas, including eastern Montana, east-
ern Wyon)ing, eastern Colorado, North Dakota, South Dakota, Nebraska, Kan-
sas, western and central Oklahoma, western and southern Minnesota, Iowa, north-
western Missouri, and with outlying populations in southeastern Wisconsin,
northwestern peninsular Michigan, east-central Arizona, and west-central Chihua-
hua, Mexico. Elevational distribution from sea level along the Texas coast to 2,100
m (18 mi N of Rubio, Chihuahua). Flowers March to August.

Representative specimens exann'ned:

Canada, alberta: Near Peigan, Moss 871 ( DyVO, NY), saskaicuewan: Katepwa,
lUissell 54519 (DAO). Pilot Butte, Hart in 1939 ( DAO, UC). 13 nn" W of Saskatoon'
Shumovieh 3H (CAN, DAO, RM). Moose Jaw, Turner 48 (CH, NY, POM, RSA). mamtoha;

112

AWALS OF THE MISSOIMU BOTAMCAL CAKOEN

LVoL. 64

Ficriu-: 19. Distribulion of CahjlopJius scniilalus.

Brandon, Foivlcr in 1SH7 (1)A(), MO, NY, US). Horlon, Love 6 Love 60SJ (DAO, Gil).
Xk'lita, Sco^^an 9791 (CAN, Cll). Montli of tlic gu\\ppcll(^ R., Marotni ir Ilrrnot 72,S77

(F, CII, NY).

Um'iki) Siates. ,\k)N'ia\a: Billings Co.: 1.5 mi S of Mcdora, SteijJicns 6" Brooks 13486

1977] rO\y\KK—CALYI.OriIUS

113

(KANU). Powder Ri\er Co.: 15 mi \W of Boradns, Booth 2517 (KAXU, WTU). Prairie
Co.: 32 mi XW of Terry, Stephens I- Brooks 23691 (DS). Sheridan Co.: Westby, Larscn 213
{Gil, NY). Wheatland Co.: 12 mi S of Harlowtoii, Hitchcock 2429 (MO, POM, RSA).
xouTii DAKOTA: Bames Co.: Valley City, Steccns in 1934 ( F, RM, UC). Divide Co.: Alkaho,
Larsen S8 (CH, MO, PH). Dunn Co.: 10 mi NW of Killdeer, Stephens ir Bwuks 12734
(DS, KANU). Golden Valley Co.: 1 mi E and 1 S of Sentinel Butte, Stephens ir Brooks
23428 (DS). Slope Co.: 7 mi S of Amidon, Cuth'r 2615 (DS, MO, NY). Ward Co.: 9 mi
NW of Minot, Stephens 6 Brooks 12897 (DS, KANU). souni Dakota: Custer Co.: 11 mi S
of Pringle, Stephens ir Brooks 13798 (DS, KANU). Harding Co.: 6 mi N, 11 W, and 2 N of
Ludlow, Stei)hens 6 Brooks 13657 (DS, KANU). Lawrence Co.: 0.8 mi N of Spearfish,
Mos<iuhi 6 MiiUi^an 5160 (DS). Meade Co.: Near Fort Meade, Fonvood 130 (CAN, US).
Pennington Co.: Rapid City, Rijdher^ 708 ( F, CH, NEB, NY, US). Perkins Co.: 10 mi E and
1 S of Bison, Stephens 7999 (KANU, SMU). miwesota: Chippewa Co.: Montevideo,
Moyer in 1908 (NY, UC). Dakota Co.: 5.5 mi W of Hastings, Moore 15797 (DAO, CH).
Faribault Co.: Elmore, Pamnwl 596 (CH, MO, RM, US). Nicollet Co.: No locality, Ailon in
1891 (F, NY, POM, US). Ottertail Co.: Perham, Clumdonnet in 1910, 1911 (CH, RM).
'bellow Medicine Co.: S side of Granite halls, Moore 13071 (SMU, UC). wiscoxsix: Pepin
Co. {?): Lake Pepin, Ilale 1861 ( F, MO), wvominc: Conxerse Co.: 15 mi N of ]3ouglas,
Stepliens 6 Brooks 23965 (DS). Crook Co.: 5 mi NE of Hulett, Porter l~ Porter 9564 (CAN,
DS, KM, RSA, UC, WTU). Goshen Co.: 12 mi W of Lagrange, Stephens 6 Brooks 2294
(DS). Niobrara Co.: 10 mi N of Lusk, Mosquin 6 Mulligan 5142 (DS). Platte Co.: Cuernsev,
kelson 8268 (DS, F, GH, MO, NEB, NY, POM, RNL RSA, UC, US). 3 mi W of Guernsey,
Porter 4903 (COLO, DS, GH, MO, PH, liM, RSA, SMU, TEX, WTU). Nebraska: Adams Co.:
20 nn- W of Hastings, Mathias 308 (MO, POM). Chase Co.: 6 mi N of Imperial, Stephens ij
Brooks 11490 (KANU). Cherry Co.: \^icinity of Hackberry Lake, Dworak in 1912 (NEB).
Dawes Co.: Chadron State Park, Porter 6 Porter 8807 ( DS, RM, WTU). Garden Co.: 2 mi
S of Lewellen, Stephens & Brooks 11557 (KANU). Greeley Co.: 9 mi N of Greeley, McGregor

(DS,

/'

19344 (KANU). Hayes Co.: 14 mi W
KANU). Holt Co.: 12.5 mi S of Atkinson, Stepliens 15579 (KANU). Kearney Co.: Minden,
llapeman in 1930 (OKLA, TEX). Lincoln Co.: North Platte, Jones in 1925 (DS, POM).
Morrill Co.: 27 mi N of Broadwater, Stephens 6 Brooks 13888 (DS, KANU). Saunders Co.:
4 mi S of Valparaiso, Croat 2126, 2127 (KANU, MO). Sheridan Co.: 13.5 mi S of Hay
Springs, Stepliens 6 Brooks 13839 (DS, KANU). Webster Co.: S of Blue Hill, Tolstead
411245 (NEB, UC). iowa: Emmet Co.: No locality, Crafty, s.n. (DS, F, NY, PO, UC).
Fremont Co.: Hamburg, Bush 10313 (GH, MO, PH, POM). Lyon Co.: Gitchie NLim-tou
State Park, Thome 14234 (SMU, UC). Palo Alto Co.: Highland Township, Hayden 10081
(PH, UC, US). Wayne Co.: Cor>don City Reser\oir, can Bruggcn 2716 (UC). colohado:
Baca Co.: 23 mi S of Walsh, Stephens I- Brooks 21788 (DS). Cheyenne Co.: 17 mi N of Kit
Carson, Stepliens 6 Brooks 22662 (DS). Douglas Co.: Wolhnrst, Clokey 3827 (CAN, DS,
F, GH, MO, NY, PH, POM, RM, SMU, UC, US, WTU). Elbert Co.: 4 mi SW of Limon,
Ownhey 1301 (COLO, GH, MO, NY, RM, UC). Kiowa Co.: 1 mi E of Eads, Stephens l~
Brooks 22703 (13S). Kit Carson Co.: 5 mi E of Flagler, Stephens & Brooks 22641 (DS).
Logan Co.: E of Sterling, Mathias 333 (MO, POxM). Phillips Co.: 5 mi S of Hol>oke,
Stephens ir Brooks 24072 (DS). Prowers Co.: 20 mi S and 7 W of Holly, Stephens 6 Brooks
21931 (DS). Sedgwick Co.: 1 mi S of Jnlesburg, Stephens & Brooks 24059 (DS). Yuma
Co.: Wray, Eggk-ston in 1919 (F, MO, POM), kansas: Barber Co.: 3 mi W and 5 mi N of
Medicine Lodge, McGregor 14430 (KANU, SMU, US). Clark Co.: 8 mi S of Sitka, Rydherg
ir hnler 766 (KANU, MO, NY). Ellis Co.: 2 mi W of Hays, Bondy 97 (ARIZ, CAN, F, GH,
OKL, OKLA, PH, RM, SMU). Ellsworth Co.: 1 mi N and 1 W of Kanapolis, Fearing 6
Latham in 1950 (GH, KANU, TEX, US). Ford Co.: Ca. 5 mi WNW of Dodge City, Towner
159 (DS). Grant Co.: High upland prairies, Thompson 1 (CAN, F, NY, UC, US). Harvey
Co.: 8.5 mi E of Newton, Harms 1633 (SMU, UC). Kearney Co.: 5 mi E of Kendall, Rydherg
6 Imler 1059 (KANU, NY). Kingman Co.: 3 mi E of Kingman, Stephens 11127 (OKLA).
Meade Co.: 11 mi E of Meade, Hubert 3593 (KANU, OKLA). Montgomery Co.: 2 mi S of
Sycamore, McGregor 12839 (KANU, NY). Pottawatomie Co.: State Park No. 2, Marsh
1725 (KANU, SMU, US). Riley Co.: Stony hills, Norton 168 ( GH, MO, NMC, NY, RM, US).
12 mi N of Manhattan, Raven 6 Gregory 19483 (DS). Scott Co.: 10.6 mi N of Scott City,
Towner 160 (DS). Smith Co.: 2 mi W of Cedar, Horr E131 (COLO, F, GH, KANU, LL,
OKL, OKLA, l^M, SMU, UC, US). Wilson Co.: 3 mi NW of Neodesha, McGregor 4306
(GH, KANU, US). Missoum: Atchison Co.: No locaHty, Busli 10321 (GH, PH, POM).
Iron Co.: Deo Arc, Stnith 460 (F). OKLAHOMA: Alfalfa Co.: 3 mi N and 7.8 E of Cherokee,

^^^ ANNALS OF THE MISSOURI BOTANICAT, GARDEN [Vol. G1

Stratum 6371 (OKL, OKLA). Atoka Co.: No locality, IIui>khi.s cl al 1132 (OKL, RM, WTU).
Blaine Co.: Roman Nose State Park, Goodman 6 Waterfall 4180 (OKL, OKLA). 21.5 mi VV
of Kingfisher, Towner 153 (DS). Carter Co.: 4 mi N of Springer, Waterfall 705 (OKL, OKLA,
POM). 11.5 mi S of Davis, Towner 144 (DS). Ca. 7 mi E of Fox, Towner 143 (DS).
Cimarron Co.: 4 mi N of Kenton, Rogers 5697 (OKL, TEX, US). Comanche Co.: Roggy
Hollow Creek, Eskew 1743 (OKL, OKLA). Dewey Co.: W of \\c\, Goodman 2573 (CII,
MO, NY, OKL, POM, RM, WTU). Ellis Co.: Near Shattuek, Clifton 3155 (CII, NY, OKLA).
Creer Co.: 2 mi S of Mangum, Robbins 3035 (OKL, SMU, UC). 6 mi S of Mangum,
Towner 80 (DS). 2.7 mi S of Mangum, Toicner 83 (DS). Harmon Co.: 13.5 mi W of
Mangum, Waterfall 7766 (OKL, OKLA, TEX). Harper Co.: Near Buffalo, Stevens 536 (DS,
CII, NY, OKL, OKLA, SMU). Kingfisher Co.: 8 mi E of Okeene, Keltinp 250 (KANU, OKL,
UC). Lincoln Co.: 6.5 mi S of Perkins, Towner 148 (DS). 3.2 mi S of Perkins, Towner 149a
(DS). Logan Co.: 18.8 mi N of Cuthrie, Towner 150 (DS). Major Co.: Togo, Demaree 12370
(MO, POM). 1.7 mi NE of Orienta, Raven 6 Gregory 19471 (DS). Murray Co.: Da\ is,
Demaree 12508 (CII, MO, NY, OKL, PII). Turner Falls, Cory 59044 (OKLA, SMU). 6.2 ml
E and 2.3 N of Sulfur, Towner 146 (DS). 8.1 mi S of Davis, Towner 145 (DS). Oklahoma
Co.: 2 mi W of Wood, Waterfall 2772 (Cli, OKL, OKLA). Pontotoc Co.: 1.5 mi NE of
Lawrence, Robbins 2995 (OKL, SMU, UC, WTU). Roger Mills Co.: Ca. 8 mi E of Strong
City, Towner 155 (DS). Tulsa Co.: W of Tulsa, McKelvey 2508 (GH, POM). Woods Co.:
Between Cimarron and Waynoka, Coodrmn & Waterfall 4229 (COLO, KANU, OKL, OKLA).

TEXAS: Anderson Co.: Palestine, Palmer 13426 (MO). Andrews Co.: 21-23 mi NE of
Andrews, Correll 32786 (LL). Aransas Co.: 4.6 mi NE of Roekport, MeCart 5566 (TEX).
Copano Ray, sandy beach, Boguseh S-75 (US). 1.3 mi W of Copano Village, near .shore of
Copano Ray, Towner 182 (DS). Armstrong Co.: Ca. 10 mi NE of Wa\side, Rowell 5402a
(OKLA). Austin Co.: Colbert's Station, Industry, Slieldon 3575 (F). Bailey Co.: 5 mi NW
of Muleshoe, Correll 13105, 13106 (LL, SMU). Brazoria Co.: 11.6 mi S of San Luis Pass
Bridge on road to Surfside, Towner 173 (DS). Brazos Co.: College Station, Parks in 1946,
1947 (RSA, TEX). NW of Bryan, in prairie, Lundell if Lundell 11304 (POM, SMU).
3.5 mi S of College Station, McVaugh 6997 (F, LL, SMU). Calhoun Co.: Magnolia Beach,
Tharp in 1930 (TEX). 0.1 mi from shore at Magnolia Beach, Toicner 178 (DS). Cameron Co.;
Stover Point, Laguna Ata.scosa National Wildlife Refuge, Traverse 1125 (SMU, TEX). Boca
Chica, Lundell ir Lundell 8617 (DS, GH, LL, NY, POM, SMU). Brownsville, Fislier 41188
(ARIZ, NEB). 0.8 mi N of bridge from inainland on Padre I., Towner 189 (DS). Childress
Co.: 10 mi N of Childress, Correll 6 Johnston 16876 (LL). Cochran Co.: 14 mi N of Bronco,
Towner 137 (DS). 10.3 mi N of Bronco, Towner 136 (DS). Collin Co.: Near Piano, Lundell
6- Lundell 9313 (DS, CII, LL, POM, SMU). Cooke Co.: 5 mi N of Gainesville, Gould 6867
(MO, SMU, TEX, UC). Dallam Co.: 1 mi SE of Texline, York isr Rodgers 192 (SMU, TEX,
UC). Dallas Co.: Dallas, Reverchon 3563 (NY). Denton Co.: 5 mi N of Denton, Cory
.5735,9 (SMU). Eastland Co.: Yiaugcr, Robinson in 1931 (POM). Erath Co.: Steplien.sville
State Park, Uoisinglon in 1946 (TEX). Fannin Co.: Bonham, Milligan s.n. (NMC).
Galveston Co. (?): Galveston I., Bcehdolt in 1870 (PH). Gra>- Co.: McLean, Craig in 1934
(POM). Grayson Co.: Near Cunter, fan Meter 18 (SMU). Hale Co.: 7 mi S of Plainview,
Gould 7156 (SMU). Hall Co.: Memphis, Thames 7192 (TEX, US). Hardeman Co.: Acme,
Rtissel 88 (TEX). Hartley Co.: 3 mi SE of Dalhart, Cory 32667 (POM). Hood Co.: Near
Centi'r Mills, Blaekwell 26 (NY, SMU). Hutchinson Co.: Borger, Hope 4 (LL). Jack Co.;
2.5 mi NE of Jack.sboro, Ilennen 421 (SMU). Jackson Co.; 11.2 mi W of Palacios, Towner
175 (DS). Johnson Co.: S of Rio Vista, Lewis 4 (SMU). Kleberg Co.: Beach along Laguna
Madre, Laurcles Division of King Ranch, Johnston 53224.13 (TEX). Padre I., Cory 49120
(GH, LL, SMU). Madison Co.: 3 mi N of North Zulch, Morgan 39 (TEX). Matagorda Co.:
6.5 mi S of Matagorda, Towner 174 (DS). Montague Co.: 4 mi N of Noeona, Whitehouse
10070 (SMU). Montgomery Co.: Willis, Warner s.n. (MO). Nueces Co.: Corpus Christi,
Drusliel 8932 ( MO, NY, US ). Corpus Christi, Heller 1517 ( PII, US ). Parker Co.; Weathcrford,
Traey 7820 (F, GH, MO, NEB, NY, TEX, US). Parmer Co.: 7.8 mi NW of I'arwcll, Rowell
10023 (DS, RSA). Refugio Co.: 4.1 mi SE of Au.stwell, Towner 181 (I3S). Robertson Co.:
3-4 mi S of Hi-anie, Reeves 930 (POM). San Patricio Co.: 2 mi S of lugleside. Cutler 920
(OKL, WTU). 3.5 mi S of Ingleside, ca. 0.5 mi from Corpus Christi Bay, Towner 184 (DS).
Suiith Co.: Troupe, Reverchon 2744 (MO, US). Tarrant Co.: Sandy soils, Reverchon 913
(F, GH, KANU, NEB, NY, PII, UC). Van Zandt Co.: 3 mi E of Wills Point, Shinners 12381
(COLO, SMU). Walker Co.: \^icinity of Iluntsville, Dixon 569 (F, POM, RM, US). Wise
Co.; Near Park Springs, McCart 1633 (SMU, TEX), new mk\k:o; Chaves Co.: 8.3 nu* W of
Claprock, Towner 134 (DS). De Baca Co.: 11 mi S of Ft. Sunmer, Towner 131 (DS). Edd>

1 ^JV7] ro\y\Kn—CALYLornus

115

Co.: Lakewood, Wooton in 1909 (NMC, US). Cuadalupe Co.: Halfway Ix'twecn Anton
Chic'o and Santa Rosa, Arsime 6 BaneJict 1668] (POM, US). Lea Co.: Knowlcs, Wooton in
1909 (NMC, US). Roosevelt Co.: 5 mi NE of Porlales, Goochnan 6 Hitchcock 1119 (DS, F,
Gil, AIO, XY, ril, POM, KM, UC). 14,0 mi SW of Elida, Towner 133 (DS). 8.0 mi E of
Taliban, Toicncr 132 (DS). San Miguel Co.: Between Las Vegas and Romeroxille, Arsene ir
Benedict 15458 (POM, US). Union Co.: Perico, Bartlctt 227 (NMC). aiuzoxa: Craham
Co.: Willow Spring, Palmer 181 (CIl, US). Navajo Co.: 4 mi N of Cariizo, PulUi ir Phillips
1008 (ARIZ, UC). 4 mi SW of Show Low, Lehto 1072 (ARIZ). Forcstdale, 66 mi S of
Holbiook, Slough S3 (US).

Mexico, chihuahua: 18 nn* N of Rul)io, District of Cusihuiriachic, Shreve 79G0 (POM,
US). TAMAULIPAS: Coastal dunes near Rio Grande, Lc Sueur 328 (probably of diis species,

ARIZ, E, TEX).

This species, tlie earliest described, cultivated, and illustrated (Hooker, 1825)
taxou of Calijlophus, occurs widely over the North American Plains, and is the
most familiar member of tlie genus, although its breeding system has only re-
cently been described (Towner, 1970b). Permanent translocation heterozy-
gosity, half-sterility of pollen and ovules, self-compatibility, and small flowers
enhancing self-fertilization are present in C. scrndatus as part of the genetic
system of complex structural heteroz\'gosit)\ This type of breeding system is
frequent in the Ouagraceae, but in CalyJopJius is restricted to this species. Charac-
ters associated witli this breeding pattern are used as the primary basis for dis-
tinguishing C. serruUitus from C. hcrhuidieri,

Whetlier factor complexes exist and are maintained in C. serrulatus by ga-
uietic lethals was not determined. The presence of gametophytic half-sterility
mak(\s the involvement of gametic lethals appear likely, but does not exclude
mechanisms utilizing megaspore competition and/or zygotic lethals. Although
unlikeK', the obser\ed le\'els of pollen and o\'ule sterility may stem from random
chromosome disjunction and the resulting genetic deficiencies and duplications
in the gametes. There were no consistent reciprocal differences in pollen fer-
tility, morphology, or chroiuosomal configurations in hybrids between C. ser-
rulatus and C. berlandicri, results which would be expected if gametic lethals
regulated the transmission of factor complexes. A system utilizing self-sterility
alleles combined with egg lethals (sec Steiner, 1956, 1957) does not seem to be

operating in C, serrulatus. All reciprocal crosses between C serrulatus and the

self-incompatible species C. herJandicri produced ordy self-compatible hy])rids,
whereas some self-sterile progen\' would l)e predicted if C. serrulatus had re-
tained a functional self-steri]it\ allele.

The flowers of C. serrulatus are generally identical to those of C. berlandieri

except for their smaller size, relatively shorter filaments and style, and the po-
sition of the stigma. The placement of the stigma among or near the anthers
and early dehiscence of th(* anthers fre(|uently causes flowers to self-pollinate

before anthesis. Undisturbed llowers in tlie greenhouse showed a high level of

autogamous seed set. Hie sinn'larity of the flowers of the two species extends

to morning anthesis times and the ultra\'iolet-absorbing areas on the petals,

stigma, and stamens. Morning anthesis had been known from the first obseiva-

tion of the species by Xuttall (in Hooker, 1825), and was seen and reported again

b}' Stevens (1920), but this was apparently not know^n to Munz (1965) or Raven

(1964). Both mentioned only vespertine anthesis for Calijlophus. Insects do not

IIQ ANNALS OF TlIK MISSOURI BOTANICAL GARDEN [Vol.. 64

seem to visit C serrulatus frequently. Stevens (1920) observed that bees ignored
C. serrulatus, althougli they were active on nearby plants of Gciura and Oenothera.
In my own collecting at 31 colonies of C. serrulatus, done at all times of day, no
insects were observed visiting flowers. Halictid bees {Agapostemon, Evylaeus,
Dialictus) were observed during the mormng at one population in Major Co.,
Oklahoma (P. Raven, personal communication), however.

Meiotic configurations of 34 individuals from 28 populations consisted of a
definite or probable ring or chain of 14 chromosomes. Five plants had a ring
or chain of 12 chromosomes plus 1 bivalent. Seventeen plants from 5 popula-
tions were examined and found to be self-compatible. In Cahjiophus the corre-
lation of self-compatibility and complex structural heterozygosity was found
to be nearly perfect. Two exceptions were individuals of C herhnulieri having
a ring of 12 plus a bivalent at meiotic metaphase I.

With a broad range of phenetic variation which largely overlaps that of C.
herlandicri, C. serrulatus cannot be reliably diagnosed without the use of floral
characters or pollen fertility. Parallel geographical variation in vegetative parts
is such that near most areas of contact, the two species are quite similar. This
implies that C. serrulatus, almost certainly a derivative of C. herlandieri, is of
multiple origins, has experienced secondary local introgression from the parental
species or has responded in a similar fashion to natural selection. Much of this
vegetative variation in C. serrulatus follows a smooth east-west cline, whereas
the variation is more discontinuous in C. herlamlieri Thus the discontinuities
in the parent species are not reflected in the derivative.

Populations formerly assigned to C. australis, because they closely resemble
adjacent ^wpulations of C. herlamlieri, and by their relative geographical sepa-
ration from the bulk of C. serrulatus, may be an exception to the introgression
hypothesis. They could well have been independently derived from C her-
laiidieri. In this paper they are synonymized because no firm knowledge is avail-
able on the phylogeny of other populations of C. serrulatus. Thus, to prefereiitiall)'
recognize C. australis, with scant genetic or morphological support, is to ignore
possible polyphyly in the rest of C. serrulatus. Instead, C. serrulatus is best recog-
nized as a complex assemblage of populations having a common breeding sys-
tem. These populations encompass a broad morphological diversity, and some
of them may have been evolved separately from the bulk of the species.

Variation in C. serrulatus involves leaf size and shape, stature, pubescence,
and flower size. Flower size is variable throughout the geographical range. Some
of the largest-flowered forms occur near large-flowered populations of C. her-
lamlieri subsp. pinijolius in central Oklahoma. Although the vegetative charac-
ters are clinally distributed, most populations occurring west of approximately
98°W longitude arc comprised of well-branched, short-leaved, and relatively
low-statured plants. East of that line plants are generally less branched, taller
and more erect, long leaved, and more densely strigose-canescent.

A few representatives of the eastern form are the following: Davis, Murray
Co., Oklahoma, Demaree 12508 (Gil, MO, NY, OKL, PH). Elmore, Farribault
Co., Minnesota, Pauunel 596 (CII, MO, RM, US). Near Piano, Collin Co., Texas,
Lundell ir Lumlell 9313 (DS, CII, LL, POM, SMU). Examples of the western

1977] TOWNER— CALVLOPHL/S

117

phenotypes are: 1 mi S of Texline, Dallam Co., Texas, York 6 Rogers 192 (SMU,
TEX, UC). Wolhurst, Douglas Co., Colorado, Clokey 3827 (CAN, DS, F, GH,
MO, NY, PH, POiM, RM, SMU, UC, US, WTU). Moose Jaw, Saskatchewan,
Turner 48 (GH, NY, POM, RSA). Representatives of the coastal Texas popula-
tions formerly assigned to C, austraUs include all specimens cited above from
Brazos, Austin, Galveston, Brazoria, Matagorda, Jackson, Calhoun, Refugio,
Aransas, San Patricio, Nueces, Kleberg and Cameron cos.

Observed instances of direct contact between C. serrulatus and C. herlandieri
were rare. No hybrid zones or populations were identified unequivocally, and the
two species were locally allopatric. The only evidence obtained concerning
possible mixed populations or direct contact came from the following collections:
3.2 mi S of Perkins, Lincoln Co., Oklahoma, Towner 149 (DS), one short-styled
plant seen in a population of long-styled C herlandieri subsp. pinifoUus, one of
which had a ring of 12 chromosomes and one pair at meiosis. 11.1 mi S of Perry-
ton, Ochiltree Co., Texas, Towner 158 (DS), chromosome counts of 3 pairs +
ring of 4, and possible ring of 12 + 1 pair from the same population, thus perhaps
some individuals of C. serrulafws in a population of C. herlandieri subsp. ])er-
landieri. 11.2 mi W of Palacios, Jackson Co., Texas, Towner 175 (DS), this was a
population of C serrulatus which was only 2.2 mi E of a colony of C herlandieri
subsp. herlandieri {Towner 176)^ the two being identical except for pollen counts
and the longer style lengths in the latter population.

Sympatry of C. serrulatus with members of sect. Salpingia is frequent. In the
southern Great Plains, C. hartwe^i subsp. fendleri is often found near or adjacent
to populations of this species. Less commonly, C harttce^i subsp. puhescens
and C, lavandulifolius occur with C. serrulatus in the same region and at the
eastern base of the Rockv Mountains. Lastlv, in eastern and southeastern New
Mexico, C. hartwegii subsp. filifolius occasionally comes in contact with C. ser-
rulatus on calcareous plains east of the Pecos River. In none of these cases has
any evidence of introgression or hybridization been observed.

LrTERATUHK CrrEJ)

Cleland, R. E. 195L Extra, diminutive chromosomes in Oenothera. E\'ohition 5: 165-176.
. 1967. Further evidence hearing upon the origin of extra diminutive chromosomes

in Oenothera hooJceri. Evolution 21: 341-344.

— & B. B. Hyde. 1963. Evidence of relationship between extra diminutive chromo-

somes in geographically remote races of Oenothera. Amer. J. Bot. 50: 179-185.
Coc:kerell, T. D. A. 1903. Notes on the entomology of Pecos, New Mexico. Canad.
Entomol. 35: 342-343.

Dement, W. A. & P. H. Raven. 1973. Distribution of the chalcone, isosalipmposide, in

the Onagraceae. Phytochemistry 12: 807-808.
& . 1974, Pigments responsible for u]tra\iolet patterns in flowers of

Oenothera (Onagraceae), Nature 252: 705-706.
Endlicher, S. 1840. Cenera Plantarum Secundum Ordincs Naturales Disposita. Fr. Becl:

Universitatis Bibliopolam, Wien. [1836-1840 for the whole work.]
Ghegory, D. P. 1964. Hawkmoth pollination in the genus Oenothera (part 2). Aliso 5;

385-419.

& W. M. Klein. 1960. Investigations of mciotic chromosomes of six genera in the

Onagraceae. Aliso 4: 505-521.

IIagen, C. W., Jr. 1950. A contribution to the cytogenetics of the genus Oenothera with
special reference to certain forms from South America. In R. E. Cleland (editor), Studies
in Oenothera Cytogenetics and Phylogeny. Indiana Univ. Publ. Sci. Ser. 16: 305-348.

118

ANNALS OF THE MISSOUHl HOTANICAL GARDEN LVol. 64

Hakanssox, a, 1942. Zytologische Studien an Rassen und Rassenbastarden von Godciia

ivJiitucyi und vorwandten Arten. Acta Univ. Lund, n.s., 38: 1-69.
. 1949. Superninnerary chromosomes in Godctia vimhwa. Heredilas 35: 375-389.

Hooker, W. J. 1825. Oenothera scrrulata. Exotic Flora 2: pi, 140.

KunAUAYAsiii, M., II. Lewis & P. H. Ravkx. 1962. A comparative study of mitosis in the
Onagraceae. Amer. J. Bot. 49: 1003-1026.

Lewis, II. 1951. The origin of supernumerary chromosomes in natural populations of

Clarkia clegans. Evolution 5: 142-157.
. 1953. The mechanism of evolution in the genus Clarkia, Evolution 7: 1-20.

, P. IL Ravkn, C. S. Venkatesh & II. L. Wedhehg. 1958. Obscnations of meit>tic

chromosomes in the Onagraceae. Aliso 4: 73-86.
LiNDER, R. & J. Brun. 1957. L'incompatibilite dans Oenothera serrulata Nutt. Experiontia

13: 23-24.
Ma'/okhin-Porshnyakov, C. a. 1969. Insect Vision. Plenum Press, New York. 306 pp.

McKelvey, S. D. 1955. Botanical Exploration of the Trans-Mississippi West 1790-1850.

Arnold Arboretum of Harvard Univ., Jamaica Plain, Massachusetts.
MuNZ, P. A. 1929. Studies in Onagraceae IV. A revision of the subgenera Salphigia and

Cahjlophis of the genus Oenothera. Amer. J. Bol. 16: 702-715.

-. 1965. Onagraceae, N. Amer. Fl., ser. 2, 5: 1-278.

Ostergrex, G. 1947. Iletcrochromatic B-chromosomes in Anihoxauthtim. Ilereditas 33:

261-296.
Rafinesque, C. 1819. The genera of North Auierican plants and a catalogue of the

species to the year 1817. By Thomas Nuttall. (Review.) Amer. Monthly Mag. & Crit.

Rev. 4: 184-196.
Ravex, P. II. 1962. Tlie systematics of Oenothera subgenus Chylismia. Univ. Calif. Publ.

Bot. 34: 1-122.
, 1964. The generic subdivision of Onagraceae, tribe Onagreae. Brittonia 16:

276-288.

& D. P. CrRKGORY. 1972a. A revision of the genus Caura (Onagraceae). Mem. Torrey

Bot. Club 23: 1-96.
g; , 1972b. Observations of meiotic chromosomes in Gaura (Onagraceae).

Brittonia 24: 71-86.
Shinners, L. H. 1964. Calylophus {Oenothera in part: Onagraceae) in Texas. Sida 1:

337-354.
Spach, E. 1835a. Onagraires, Histoire Naturelle des \\'getaux. Phancrogauies 4: 335—116.

. 18351). Mouographia Onagrearuui. Nouv. Ann. Mus. Hist. Nat. 4: 320- lOS.

Steiner, E. 1956. New aspects of the balanced lethal mechanism in Oenothera, Genetics

41: 486-500.
. 1957, Further e\idence of an incompatibility' allele system in the couiplex-hetero-

zygotes of Oenothera. Amer. J. Bot. 44: 582-585.
Stevens, O. A. 1920. Notes on species of llalietus \isii iug evening flowers (Ih'm.).

Entomol. News 31: 35-44.

Towner, II. F. 1970a. Calylophus. Pp. 1121-1123, in D. S. Correll & M. C. Johnston,
Manual of the \'ascular Plants of Texas. Texas Researeli Fomidatiou, Rcnuer, Texas.

. 1970b. Evolution and breeding systems in Calylophus (Onagraceae). Ph.D.

dissertation, Stanford Univ., Stanford, California.

& P. II. Ravex. 1970. A ne\\ species and some new combinations in Calylofilius

(Onagraceae). Madrono 20; 241-245.

Index or Latix Na:

NtiiiilxMS in hoM fare refrr to descriptions or tianies listed in synon\niy; names in roman
are recogin'/ed as \alid; names in italic refer to s\'nonynis; nnnibers \\\[h asterisks (*) refer
to names ineidentalK' mentioned.

Aeaeia 70*

Acmaeodera 1 10*

A.uapostemon 74*, 77*, S7*, 91*, 95*, 106*,
110*, IKi*

Antliopliora 100*
Anlhoxantlnnn 59*

^:

Apis 74*, IK)

Artemisia

tri(l(Mitata 88*

Asleraeeae 02*

Atalopi'des

eamp(\stris I 10*

Aiij^^oehlorella 110*
15oml)us 77*, 87*, I 10*

Bnprestidae 110*

Cahjhphis 49*, 67

Calylophns 48*-52*, 59*-65*, 67, 87*, 91

65*, 70*-72*, 74*, 75, 77*-7S*,
81*, 100*, 110*
-snl)sp. pnbescens 51*, 55*, 00*,

05*, 70*-7l*, 74*-75*, 78*, 80*-
81*, S3*, 84, 80*-87*, 91*, 98*,
100*, 110*, 117*

subsp. ioMinctji 92

■\'ar. jilifolius 78

■\ar. hariwcm 71, 74*, 78, 81

94*-95*, 97*, 99, 115*-117*

* /v— *

fiS

sect. Calylophns 48*, 50*, 65*-fi

95*, 98*, 99, 109*

srtt. Salpiii^ia IS*, 50*, 59*, Gl*, 65*-

68, 70*, 75*, 77

*

I 1 *

81

SO*, 87

T*

91*, 93*, 95*, 90*, 97*, 100*, 110*,
117*
(itisticilis 50*.

Ill

*

Ix'ilaiKlicri 48
08*. 70*

.'*

1 10
, 50*-51*
81*. 87*.

, 01*-02*, 00*,
90*, 100, 102*,

105*, 109*, 111*, 115*-117*

snl)sp. iKM-landiVn 51*-52*, 59*, 00*,

, 103,

05*. 78*

97

101

102

105*-100*. 108*-109*. 117*

03 * ,

siil)sp. pinifolins 52*, 59*-00*,

101*-102*, 105*-i00*, 107, 108*- <^'aniiss()nia 59*
111*, I10*-117*

— \ar. lavandulacfoUus 88
— \ar. maccartii 75, 77*
— \ar. ])ul)esccns 84
— \ ar. totimctji 92

lavaiRlulifolius 48*, 50*, 56*, 59*-01*, 05*,

07*-70*, 75*, 81*, 87*, 88, 89*, 91*,

98*, 102*, 100*, 117*
mil t alia (iJ*, 110
scnulutiis 48*, 50*-51*, 57*, 59*-03*, 00*-

08*, 70*, 81*, 84*, 87*, 91*, 100*,

102*, 103*, 108*, 110, 112*, 115*-

I 17*

— sii]).sp. scnulaius 103

— \ar. arizonicns 111

— \ar. scrnilatus 110

— var. spiuidosu.'i 107, 110

toiiiiK-yi 48*, 50*, 58*, 00*-05*, 08*, 09*-
70*, 89*, 92, 93*, 94*

luhitula 48*, 59*, 01*, 05*-00*, 08*, 70*,
81*, 87*, 94, 95*-99*, 102*, 100*
— suhsp. slii«uk)siis 50*, 58*, 00*, 75*,
89*, 91*, 90*-97*, 98, 99*

— sul)sp. tiilHcula 58*, 00*, 63*, 75*,
89*, 91*, 95*, 96, 97*, 99*, 1 10*

Cclcrio

(

Vmrata 61*, 70*

.'4.'*

liiiniDitntdluiui 110

clrmnnioiKliamis 48*, 50*, 102*, 103, 108* Orcocarpiis 88

— siibsp. iKihnulicn 103* (.larkia 49*, 59*
— siihsp. (Inniii)ioiidiami.s 107, 110 amoena 51*

hailwi'^ii 48*, 50*-51*, 59*, 61*-62*, 05*-

4*, 87*-88*, 91*-

68, 09*-7()
* 1 02 *

~i\* "J

07='=,
96*,
siibsp. fciulK ri 50*, 53*, 00*, 64*, 60*,

71*, 74*, 80*, 81, 83*, 84*, 91*,

97*-98*, 100*
,snl)sp. filifolius 50*-5l*, 54*, 60*, 65*,

^ 74*-75*, 77*, 78, 79*-81*, 83*,

71*,

87*, 98*, 106*, 110*, 117*

suhsp. liaifw t'jrii 50*-51*, 54*

71. 72*.

00*, 03*-06*, 70'

79*. 81

75*
99 *

77*

* vr**

80*^87

50*,

74*-

91*,

• — sid)si^. amoena 51*
nn^nieulatu 51*

Dialictns 87*, 95*, 110*, 116*

Kvylaeus 77*, 83*, 87*, 95*, 100*, 110*, 110
^^alpinsia 83*

Calpimia 49*, 67, 68
caniponini 84

carlshadiana 96, 98*
fendlcri 81
filifolia 78
^landulfcm 96

iireanU 84

hariwciii

snbsp. Iava}i(lulifi)hus 88

snhsp. maecartii 50*-5l*, 55*, GO

\'ar. fcudlcii 81

hartxceaii 67*. 71

120

ANNALS OF THE MISSOITHI BOTANICAL GARDEN

[Vol. 64

interior 84
lampasana 84

lavaudulacfolia 88
— var. filati(hiIosa 88

toumeyi 92

tubicula 96
Gaura 49*, 51*, 59*, 65*, 116*
Cayophytum 59*
llalictus 110*
Ilyles

lineata 61*, 69*, 87*, 91*, 95*, 106

Ilauya 49*
Ileterogaura 49*

Juuipems 78*, 82*, 85*, 88*

divaricata 76*, 78*, 88*, 96*

Lasioglossiim 93*

Maiidnca 91*

quinquemaculata 87*

Megachile 87*, 110*
Mdissodes 87*, 106*
Mcriolix 49*. 67, 99

bcrhndicri 103

capillifolia 107

drurnmondiana 110

hiUii 107, 109*
intermedia 111
mchnoglottis 107, 109*
oblanceolaia 111
scrrulata 67*, 110
— var. douglasii 110

van drummondii 110

■var. nutt(d!ii 110

-var. pinifoJia 107

var. spinuJosa 110

spinidosa 110

OiKiKraceae 48*, 59*, 62*, 66*

tribe Ona^reac 48*, 49*

Oenothera 48*-51*, 59*, 116

siibg. Salpiugio 49*, 67*, 68

— subg. Cidylaphus 99

hcrlandieri 103

camporum 84

capiUifolia 107

fcndleri 81

grrggii 84, 86*, 87*

— var. lampasana 84

— var. }}uhescens 84

— var. pringlci 71*, 74*, 75

— var. t [I pic a 84

hartwegi

— var, fendleri

— siibvar. lavandtdacfoJia 88

— var. tubicula 96

hartwegii 68, 79*, 83*
var. fendleri 81
subvar. fdifolia 78
— f. dnjmifolia 84

var, filifoUa 78
■var. hartwegii 71, 78, 81, 83
■var. lavaudulacfolia 88
■var. toumeyi 92

^ i*

— var. typica 71, 74
hookeri 59*
lampasana 84
lavandulac folia 67*, 74*, 88

— var. glandtdosa 88, 91*

78, 81, 88

var. typica 71

lavanduhfolia

var. lavandulifolia 71, 74*, 88
leucocarpa 110
pringlei 71, 75
serrulata 105*, 110

siibsp. drummondii 103, 107, 108

110
siibsp. pnn7o/m 103, 105*, 107
subsp. scrrulata 103, 110
■var. douglasii 110
var. dru))nnondii 103, 107, 110

— f. //ata 107, 108*, 111

\'ar. intcgrifolia 111

\ar. maculata 107

•var. nuttallii 110

-var. pinifoUa 103, 105*, 107

var. K])inulosa 110
var. typica 103, 110
X serrtdatoidcs 96
spachiana 110
toumeyi 92
tubicula 94

— ^'ar. dcmissa 96

—var. /?7/7(?//fl 78
Opuntia 76*, 103*

Finns

ednlis 82*

nionopliylla 88

ponderosa 82*, 88*
Prosopis 103*

glandnlosa 76*, 82*, 85
Quercus

havardii 103*
Salpingia 67, 68

hartwegii 71
Sphecodogastra 83*
Sphinx

dolli 87*
Xylocopa 110*
Yucca 76*, 78*, 96*

CHROMOSOME NUMBERS IN LEGUMES^

PkTKH CrOLPHLATT- AM) C. DaVI])SK'^

AiiSTRAfrr

Chroniosoiiu' numbers are reported for 39 species representing^ 35 genera. Tlie fol-
lowing are new generic (and sp(^eific) reports: Arrocarpus fraxinifolius 2u — 24, Ambhj-
^onocarpus andongcnsis 2n = 28, Baihaca plurijuf^a 2n = 24, Bmsillctia mollh 2n = 24,
Kotschya (wsclujnomcnoidcs In — 40, K. africana 2n — 30, K. strigosa 2n =^ ea. 36, Maarkia
amurcHsi.s 2u = (18)20, ParanuicwJohium cocmlrum 2n — 24, Sc]uzi)lo})iuw paralujhum
2n — 26, SpJtacro})hysa salsida 2n = 16, Splicno.stylis inarfi'nmta 2n — 22, Stnmgylodon macro-
hotnj.s 2n — 28, TachigaJia panicidala 2n = 26, Tyhscma fassoglcnsis 2n = 52, Xandwccrcis
zambcziaca 2n — 26, and Virgdia orohoidcs 2n — ca. 54. New species reports are tlie following:
Cordyla africana 2n — 20, Dcsmodiuni harclayi 2n ~ 22, Dioclea virgata 2n ~ 22, Entada

pursaetha 2n = 28, Erydirina livingsioncana 2;i — 42, Mucuna shanci 2n = 22. and Sindorn
wallichii 2n — 24.

This paper is preliniinar>' to a general re\ie\v (P. 11. Raven & P. Goldblatt,
in preparation) of cytology in the Legnininosae. This work is benig nndertaken
in conjunetion with other systematic research in the family to be presented at
the Legnme Conference plamied for 197(S. Tlie majority of connts reported her(^
are for species and genera selected to fill some of the many gaps in the cytology
ol the fanH'l\\ Althongh there are a large number of chromosome connts in the
Legnininosa(% the family can still be said to be in general poorly know^n cy-
tologicall)' witli an estimated 189^ of tlie ca. 18,000 species having been investi-
gated (Handel, 1975). The present paper inchules chromosome connts for 39
species in 35 genera. Of these, 15 are first reports for genera.

MArKHlALS AND MkTIIODS

Root tips only w^ere used in this study, hence all connts arc mitotic. Root
tips were pretreated eitlic^r in cold water for 24 hours and stained using the Feul-
gen technique or were placed in 0.003 M hydroxyquinoline for 4-5 hours and
stained in lactopropionic orcein. Species studied are listed in Tal)le 1 and al-
uiost all plants have herbarium vouchers which are housed mainly at Missouri
Botanical Ciarden (MO), but also at New York botanical Garden (NY) or else-
where.

Discussion

The results are listed in Table 1 and are discussed bv snbfamilv and tribe.
The three traditional subfamiHes Papilionoideae, Caesalpinioideae, and Mimo-

^ We wcmld lik(^ to thank Dr. B. A. Krukoff for Ins energetic and nntirin^ eff(nls to ol)tain
seeds of Le^Minnnosae. Almost all samples nsed in this stndy were obtained either directly or
indirectly dirongh him. We would also like to thank the various individuals and institutions
too numerous to mention who have provided us with material for study.

-B. A. Krukoff Curator of African Botany, Missouri Botanical Garden, 2345 Tower Crove
Avenue, St. Louis, Missouri 63110.

'Botany Department, Missouri Botanical Garden, 2315 Tower Gro\e Avenue, St. Louis
Missouri 63110.

A\x. Missouiu BoT. Gaud. 64; 121-128. 1977.

222 AWALS OF THE MISSOURI BOTANICAL GARDEN [Voi„ 61

soideae* are recognized. Tribal sul:)divisions are those of Hutchinson (1964) for
Miniosoideae and rapilioiioideae, while Taubert's (1891-1894) treatment is used
for Caesalpinioideae as presented recently by Heywood (1971). Previously
published chromosome numbers referred to in this paper are takcMi from the
standard chromosome indices, particularly Darlington & Wylie (1955), Fedorov
(1969), and the annual Index to Plant Cluomosome Numbers, the most recent of
which summarizes reports for 1972 (Moore, 1974).

MIMOSOIUKAK

Adcnanlhcrcac, — The report of 2/i = 28 in Amhhjgonocarpns is a new generic
record. Tliis count accords witli ?i = 14 in Prosopis and otlier allied genera in the
tribe where x — 14 is probably basic. Low base numbers of x = 13 in Adcnanthcra,
Piptadcnia, Neptiniia, and others, and .v — 12 in Cidpocidijx and XijlUi are most
likely d(Mi\'ed. Tlie coimt for Entada pnrseatha is a new species record and in
accord witli n = 14 reported for several other species of this genus.

7n<iC(ie. — The count of 2n = 26 for Sanuinea smnun confirms a prcATOus
record for this species. A base number of x — 13 has been reported in several
genera of the tribe with the onl)' e\ce|)tion, CaJVicmdm, x — 8, standing out liere.
Counts for more genera in this tribe are needed before the significance of tlie
exceptional .v = 8 can be assessed.

CAESALPlNKHnKAE

Cacsidpinlcac. — Counts in this group for Cacsalpinia pulcherrinui (2m =24)
and DeJonix re<g\a (2m = 28) confirm previous reports for these species. Counts
for Acroearpus (2m = 24), Schizohhitun (2/i = 26), and Brasilettia (2/t = 24)
are new generic records. With .r = 14 probably basic in this alliance and with
X— 12 and 11 well established in several other genera, present counts arc con-
sistent with the cytological pattern of this tribe. The number 2n = 26 in Schizolo'
hittni is, liowe\'er a new number for Caesalpinieae.

Dctaricae. — The counts for Baikiaca, 2n = 24, and Tachhgcdia, 2n = 26, are
new generic records. Tlie number in Baikiaca is consistent with the majority of
cluomosome mnnbers in Detarieae where x— 12 is probably basic. Other num-
bers in the tribe are m = 10 and m = 8 reported in only a few^ genera. The record
of 2n = 26 for Tachh^aVui paniculafa is a new number for the tribe.

Tlie count for Sindora waUichii is a first report for this species. Other counts
in Sindora are n= 12 in S. siainen.sis, m= 10-12 in S. cochinehinensis, and n = 8
in S. supa. Unless Sindora is cytologically heteroploid, n = 12 would seem most
likely correct for the genus. Sindora supa should, however, be checked to A^erify
the nmisual /( = 8 reported by Atchison ( 1951).

Afzclia af

thi^

s

species

Amhcrsticac, — The report for Paranwcrolohiuni is a new generic record and
the number 2m = 24 is consistent with x— 12 in the seven other genera of Am-

herstieae known cytologically.

Cerccae. — The count for Tijhscma jassoglensis, a segregate of Baidiinia (al-

U)77|

GOLDHLATT & PAX IDSTv-CTIROMOSOME NUMBERS J\ LEGUMES

123

O

0^

a

rt

o

(U

0^

C

o

a;

c

a;

O

cd

(U o

t:

o

CI

Zl

<l)

O

O p
O ci3

- *

U

cfl

t/3

>.

p^

1^

^

^
■^

■^

1;

u

I-

o

c

o

a;

Q

iJ

O

in

O

O

to

UK

"S s

■S^ 5

-; y

^ 5

GO CO

1*

0;

u

SO

«

2 -

Q

G

G

* *
#

Us

C

4

O

c/5

ri

o

o

CO

OJ

5

CO

UJ

Oj

O -^

I— 1 oJ

U

i

t^

'^

o

'^

^

^

U

CO

^^

_>>o

F«(f

■

rt

p^

ci

*^

o

Ij

Ij

H

c/5

5

C

o

o

- -f

en q^

<M

<M

CI.

Ui
^

>

«

3

14

Co

Q

O
O

to

*
*

O

o

^^,

!ii

CC

"1^
t -

= ^

^

w^

2 5

:/: ^ "aJ

^ ^ ?5

V

y 2

-* V

ti

IT N

c

<y-v

t 00 CD
Ol OJ o\

N

cd

CO

0^

J

§ S

ffi 5

?X Q ;:;

5

o

r* - O

S 5 -L*

O Q c^

*

o

to

o

o>

2 t=

CO

^

Qj -r

«

^

.'^

:^

1-

0;

t/1
0;

C

E c:

s o

*-
^

w

V3

Q

>

S 2 O

O

rv-

r o

C

vi

^

<

OJ

3 -S^

^ 5

3.2

* *

*

o

o

c

*

^O

o

CO

o^

CO

to

r

^ ::

t: o

t^

OJ

a>

cti

«

a;

Oj

•i

"

'C

-r -r

-f CO

^ 1^

^

^ CI

rt

Ol CI

CI C!

S (^i

Ol lO

<u

^^^

^

Q

s

u

o

o

- CJ

^ y:

^

^ ^

Or

*

124

ANNALS OF THE MISSOURI R0TA\U:AL GARDEN

[Vol. 64

n3

o

9

CQ
<

«

o

u

o

>

c

^

x:

C

c

U

05

D

O

o

u

0;

Oh n

<

O

CQ
<

CO

e

I

o

Q

c

CM

o
k4

o

C

U

0;
>

a:

<:

Co

5

a

^

c

u
<

3

u

0;

0)

a

c)5

00

-1

^

S2

CO s

CO
CO

cc <

Cc

u

"o

c*?

o

rt

C

5 .O

O «

o c

c/5 en

00 00

in (N

cd

c
2

i

Q

^ CQ

CC J

^)

£f2

CQ t^

IS

s

o^

* *

1^

O
:5. ?5

cq

32

CO

CO

in

Q

a

cc;

. O

c

o

c

s

o

cd ^^

S CO

<

o
o

CQ

.5 ,^

c S

cd cd

Id ^

-13

cd

C
O

O

2

<

CD

00 00

00

c

^^3

a;
CQ

<3

5

Ph :=

cd

O ?=

O
C/5

^^

CD

u

>

0)

S

H

O

CQ

2;

t

^

C/5

<u

cd

t/5

cd"

o

U

T3

4 «^

=0

oc

2

cd

2

cd

t/3

Q

3
3

cd
cd^O

0;

OJ

Cd

a;

•S

^

<M

'S oi

(U

CM

Cd ^

^

U

CAJ

>

o

Cd

0)

J

o

2;

-5

:3

C

U

u

«

cd

<

in

CO

Q

cd

CL,

1

§

%

1

^

Si

0?

-^

.2

O

c

«>.

o

^

^

5

CO

r<wih

^

0.

C/^

in

*

«

lyvT]

OOLDBLAIT &- DAVIDSE— CHROMOSOME NUMBERS IX LEGUMES

125

c

a

c

c

J

o

fl

u

5^'

u

r^

o

;<

r^

O X

U

N M

ffi;;

(M CI

CM oi

iA

c5 §

*

W

O p

o

. =3

^

^ .

c

O

^

^

^ :-

00
CO

'•r.

\^

'^y

^>

<

Ln —

tL C

0^ <l)

a>

a>

05

!^ ^

tiD

-V

-4

:f W

4 «

rO

-4

N

o u

■^ f>3 OJ
GO T Ol

rM CO

PS

q;

O

o

Co

^

a

5^

si

^

-t.-::: ^

tq f4j

» *

?i.

en

*

GO

N

N

»--..

o

::0

fM
<M

QJ

W

^

y

<:

o

CO

"3 —
^ 9.

*-*j

Ah _r

N

r N

^4

>^

* K

C P

1; 0^

C O

5 O CD

a

:= >

ii oj

o .

O

o>

0^

0^

Z

4

■I

-^

o

ca

>

Oj

C3

Co

s

C3

0)

o

O

CO

^

PP

'«W

tin

U

of

P^

a;

pp

2

]26 ANNALS OF THE MISSOURI B0TANK:AL GARDEN [Vol. 64

though hichidcd m Buuhinia by some authors), is a new generic record. This
species appears to be tetraploid witli a 1)ase number of x = 13. Tlie number re-
ported here supports tlie exehision of T. fasso^ijetisls from Bauldtiia in which x
14 is clearly basic and only multiples of this have been recorded in the genus.
In the other relative of Buuhinia studied here, Piliostigma thonningii, we have
obtained 2n = 24, which supports an earlier report by Mangenot & Mangenot
(1962) and conflicts with Turner & Fearing s (1959) report of 2/i = 26 for this
species. The other species of Piliosligma known cytologically have 2/i = 28
(Miege, 1960; Sharina & Uaju, 1966), a number indicating closer affinity with
iiau/i//i/a perhaps than with P. tlionningii.

PArU.l()N()n>KAK

Swart zicac. — The count of 2n = 20 in Conhjla africana is a new species record
and supports a previous count in this genus. Only two of the nine genera in
Swartzieae are known cytologically, the other being Sitartzia with n = 8.

Suphorcac. — Reports for Maackia^ Vir<;iUa^ and Xanthoccrcis represent new
generic records, wliilc 2n ~ 18 for Sophora lonientosa confirms several previous
counts for this species. Although /i = 9 is the most frequent number in Sophorcae,
X = 14, .V = 13, .V = 11, X = 10, and .v = 8 are other base numbers that have also
been reported in se\eral genera. W^ith this range of chromosome numbers, the
counts of n — (9)10 in Maackia and n = 13 in Xanthocercis appear consistent
with the present circunjscription of the tribe. The number in Maackia could not
be more firmly established, and the possibility exists that a small pair of chromo-
somes may represent large satellites. Gixen the frequency of n — 9 in Sophoreae,
2n = ca. 54 in yirgilia orohoidcs appears to indicate hexaploidy in this species
and a base munber of .v = 9 for the genus.

Podalyricac. — Counts for the four Australian species studied here support
recent detailed cytological work for the Australian species of this tribe (Sands,
1975). Important numbers in Podalyricac are n — 9 and 8 with a few genera
also n — 7. As pointc^d out by Polhill (1976), a- 9 occurs in the South African
and north temperate Podalyricac, while /i - 9, as well as /i = 8 and 7, occur in
the Australian rcprcsentati\ es of the tribe.

Tcphro.sicae, — The count for Muudiilca scricca confirms previous records for

this species in which both u = 10 and 11 have been reported. This suggests tluil

the count of n— 10 may be erroneous, particularly as x — 11 also occurs in the
majority of species of the related genus Tephrosia. Lower counts of n = S in
Tephrosia (Sands, 1975) and n =^ 8 in the related SpJiinctospennuni may be cor-
related with the adxanced position of these species.

Seshanicae. — The report of In — 12 in Seshania seshan confirms earlier counts
for this species and is consistent with lunnerous reports of n = 6 in Seshania.
The only other reports in the tribe are /* = 8 for Cracca^ which sliould in fact be
excluded from Seshanicae according to R. M. Polhill (personal communication).

Colulcac. — ^The count of 2n— 16 in Sphaeroplujsa is a first report for this

genus. /

Id

cytologically ha\'e x = 8, presumabl)' basic for the alliance.

1^*'"] GOr.nBLATT .\ DAVIDSK- CIIUOMOSOME XUMHKHS l\ T.KCrMES ] 27

Dioclcae. — The report of 2n ~ 22 for Dioclca vivgata is a new speeies report,
wliile that for Canavalia maritima eonfirms a previous eouiit for this speeies.

22 in D. reflcxa (Mangenot & Mangenot,

1958, 1962) and 2/i = 24 in D. hoykinii (Nemee, 1910). Our second report of 2u
22 in Dioclea throws some doubt on Nemee's record, especially as .v = ll is
clearly basic in Camptoscma, Catwvalia, and Fuemria, the only other genera of
Diocleae for which chromosome numbers are known,

Erythruwae.— The count of 2/i = 42 for Erythrina Iwing^toncana is a first
record for this species, while the count of 2?i = 84 for E. amazonica confirms
prexious reports of tetraploidy in this species, n = 21 ])eiiig basic in Erythrina.

Counts of 2/i = 22 in Mucutia confirm previous reports in the genus, the
record for ^L shanei being new for this species. The report for Strongylodon is

a ucw generic record and a uvw chromosome number for this tribe which is
cytologically heterogeneous with numbers ranging from .y = 21, x — 14, x— 11,
and X == 9 (and possibly also x = 10),

Phascolcae, — The count of 2/i = 22 in Splicnoslylis is a first record for tins

genus. Tlie number accords well with other reports for the tribe, the majorit)' of
genera known cytologically liax'iug n - 11. Only a few genera stand out as differ-
ent, notably Endomallus with x = 8 and Psophocarpus where n— 11, 10, and 9
are reported. There are also a few^ reports of n = 10 in Vif;na and DoUchos which
re<niire investigation since tliey differ from n = 11 in most species of both genera.

Aesctiynomcneae. — Counts reported here for three species of Kotschya rep-
resent the first cytological records for this genus. Othei- counts in the tribe ar(^
n- 10 in Aeschynomcne and Brya, n— 12 (probably) in Ormocarpum^ and n
19 (again probably) in SinitJiia. Kotschya does mA appear to conform with the
previously reported uuml)ers for Aeschyfwnwne^ the three species reported here*
having /i = 15, /i - ca. 18, and /t = 20, respectively. More counts in Kolsrhyn
and in the tribe are needed to clarify the cytology of this aHiance.

Desuiodwae. — The count for Desmodium harchyi is the first report for tliis
speeies and is consistent with n — 11, the only reported number in Dcsniodiitni.

LrrKivvruRE CrrEi)

AiTiiisoM, i;. 1951. Stiidirs in tlie l.eguininosae. VI. Clirornosoiric mnnbtrs anions Iropiial
woody species. Aiuer. J. Bot. 38: 538-547.

Banukl, G. 1974. (^luoinosoine numbers and evolution in \}w. T.e^nnu'nosa(\ C^ar\ ()I()<na
27: 17-32. * '

Daiujnctox, C. D. & A. P. WYi.n:. 1955. Chroniosome Atlas of KlowcrinK Vl.uds, Canine
All(*n & Unwin, London.

Fkdohov, a. (editor). 1969. Chroniosome Nnmliers of iHowcrini,^ ]*lan(s. V. L. KoniaroN
Botanical Institute, Leningrad.

IIkywooi), V. II. 1971. The Le^nniinosae — A systematic purview. Pp. 1-29, in J. R.
IIarI)orne, D. Boulter & B. L. 'I'urner (editors), Chemotaxonomx of tlie LeKMnninosa<\
Academic Press, London.

IlLncuiNsov, J. 1964, Cienera of Flowering Plants. Vo]. 1. Clarendon Press, Oxford.

MA\(;i:\or, S. & C. Mavcexot. 1958. Deuxieme list de nombres ehromosomi(iu(\s non\i'an\

chcz diverses dicotyledones et monoeotvledones d'Africpie Oecidentah* liuli Jard Bot

Etat. 28: 315-329.

& • 1962. En(iuete sur les noml>r(\s c-hromosomifpies dans une collect ion

d'especes tropieal(\s. liev. C\[. Biol. Veg. 25: 411-447.

Mij'CK, J. 1960. Nombres c]n-omi)somi(iues de plantes d^Afriijuc Occidcntale. liev C\tol
Biul. \'eg. 21: 373-384.

128

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 61

MooHK, R. J, (editor). 1974. Index to plant chroniosonie numbers. 1972. Regnuni Veg.

91: 1-108.

Nemkc, B. 1910. Das Problem der Refriictimgsxorgange und andere 7.>t()l()gisehe Fragen.

Borntraeger, Berlin.
PoLlllLL, R. M. 1976. Genistoae (Adans.) Bendi. and related tribes ( Legmninosae). Bot.

Syst. 1: 143-3()8.
Sands, V. E. 1975. The eytoevolution of the Australian Papilionaeeae. Proe. Linn. Soc. New

South Wales 100: 118-155.
Suauma, a. K. & D. T. Raju. 19fi9. Strueture and behavior of ehromosomes in Bauhima

and allied genera. Cytologia 33: 411-42(>.
Tauuert, p. 1891-1894. Leguminosae. In A. Engler & K. Prantl (editors), Die natiirliehen

Pfan/enfamilien. Ill, 3: 70-388. Wilhelm Engelmann, Leipzig.
TuRNKH, B. L. & O. S. Fearing. 1959. Chromosome numbers in the Legunn'nosae. IL
Afriean speeies, ineluding phvletie interpretations. Amer. J. Bot. 46: 49-57.

FOUR SPECIES OF ASCLEPIADACEAE NEW TO PANAMA

David L, Spellman^

AlJSTHACT

FiscJiciia hrachijcahjx, Gonol()J)tis rothschiihii, Matclea pscucloharbata, and M. viridis arc
newly reported for the Panamanian flora. Matclea viridis (Moldenke) Spellnian is a new
eonibination based on Fischeria viridis Moldenke.

Recent collections of plants from poorly known areas of Veragnas and Darien
Province's in Panama have added three sx^ecies of Asclepiadaceae to the flora of
the conntry. Fnrther studies of the fanu'Iy have resulted in a new combination
for one of the three, as well as the addition of a fourth species to the flora which
was incorrectly assigned in the treatment of the family for the Flora of Panama
(Spellman, 1975).

1. Fistheria bradiycalyx L. O. Williams, Fieldiana: Bot. 32: 43. 1968. type:
Costa Rica, Austin Smith 1211 (F, OH, MO, NY).

Stems glandular puberulent and hispid, the longest trichomes brown, to 3 nmi
long. Leaves elliptic, apically acuminate, basally cordate, mostly 9-12 cm long,
4-6 cm wide, the upper surface scabrous, the lower surface softly hispid; petioles
glandular puberulent and hispid, mostly 3-5 cm long. Inflorescences racemose,

<
to

glandular puberulent and hispid-pilose throughout; peduncles 4.5-S cm long,
pedicels 3-4 cm long. Floicers 1.4-2 cm in diameter; calyx abaxially glandular
puberulent and hispid-pilose, adaxially glabrous or nearly so, the lobes lance-
ovate, 4.7-7 mm long, I. .5-2.5 mm wide; corolla light green, shallowly cam-

panulate, the lobes ovate, 6-7 nun long, 3.5-5.5 mm wide, strongly crispate near
the apex on one margin, the upper surface papillate in a median band, this over-
topped by long white trichomes of variable density, the lower surface appressed
brown pul)escent, the crispate margin ciliate; gynostegimn 2-2.5 mm high, in-
flated portions of the anthers suborbicular In sinface ontline; corona prominently

5-gonal to shallowly lobate, the surface striate-sulcate. Follicles obliquely el-
lipsoid, 19-24 em long, ca. 3-4 cm in diameter, the walls thick, sublignose, finely
striate, drying reticulate, sparsely pubescent; seeds compressed-ovate, 10-12
mm long, 6.5-7 mm wid(% the marginal wing irregularly and deeply toothed
apically (fruit described from Costa Rican material).

Occurring in partial shade of moist forest, this species is most frecjuently col-
lected at eleviitions from 1,000 to 1,400 m in Costa Rica and Panama.

Using the keys given in the Flora of Fanama^ this species is identified as
Fischeria columhiana Schlechter. The latter species is readily separated from

F. hraclujcahjx in having its pedicels and calyces puberulent but lacking longer

hispid trichoines. In addition, the surface of the corona of F. columhiana is vermi-

form-fimbriate rather than striate-sulcate.

^Missouri Botanical (iarden, 2ok5 Tower drove A\'enue, St. T.onis, Missouri 63110.

Axx. Mrssouiu Bot. CJaiid. 64: 129-132. 1977,

130

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

VERACiUAS: Rio riiiiicio Braso, 2.5 km beyond Agriculture School, Alto Piedra near Santa

Fe, 700-750 m, 24 July 1971, Croat 25533 (MO).

2. Matelea viridis (Moldenkc) Spellinan, comb. nov.

lischcria viiiclLs Moldcnke, I'hytologia 1: 13. 1933. type: Colombia, A. E. Lawrame 39G
(F, Cll, K, MO).

Stems densely glandular pubcrulent and sparsely pilose. Leaves elliptic to
slightly obovate, apically rounded and abruptly acuminate, the upper surface
softly scabrous, the lower surface hirsute; blades 8-12 cm long, 3-4 cm wide;
petioles glandiilar pubcrulent and pilose, 1-1.5 cm long. Infhreseences race-
mose, densely brown, glandular pubcrulent and pilose throughout; peduncles
3.5-4.5 cm long; pedicels 2.5-3 cm long. Flowers 2.5-3 cm in diameter; caK
densely glandular pubcrulent and pilose abaxially, glabrous adaxially, the

3-4

co-

rolhi greenish white with darker green reticulate "veins," shallowly campanulate-
rotate, deeply lobed, the lobes lance-ovate but folded lengthwise to appear ligu-
lar, 10.4-11 mm long, 5.5-6 mm wide, undulate-crispate on one or usually botli
margins, the upper surface glabrous except for a dense median band of minute
papillae, the lower surface pubcrulent; gynostegium 3-3.5 mm high, yellow green
and white; corona carnose, sharply 5-lobed, the lobes triangular, the surfaces
colliculate, the corona column smooth and appendaged, the appendages ± ligular,
situated above each lobe, curving up around the gynostegium-head. Follicles
ellipsoid, attenuate to the apex, smooth, glabrous, to 19.5 cm long, ca. 4.5 cm in
diameter; seeds spatulate, 8.6-9.3 mm long, 4.7-5 mm wide, the marginal wing
coarsely serrate in the apical half.

The collection cited below represents the third known collection of this spe-
cies. The other two localities are from the central cordillera of Colombia at ele-
vations from 1,200 to 2,000 m.

The species is excluded from Fischeria primarily on the basis of its lack of dor-
sally inflated anthers. It differs also in the more technical a.spects of its pollinia

and gynostegium morphology.

In die keys to the family given in the Flora of Panama, M. viridis cannot be
placed in the proper genus. \Vith some effort, it can be forced through the keys
to Fischeria. It differs from other species of Maielea in its ligiilar-appearing,
undulate-crispate corolla lobes.

XLHACUAs: Lower montane wet forest 6-7 km W of Santa Fe on new road past Agricultural
School, 2,900 ft, 17 Feb. 1974, Ncc 9791 (MO).

3. Matelea pseudobarbata (Pittier) Woods., Ann. Missouri Hot. Card. 28:
235. 1941.

Gotwlohtis pscudobarbatus Pittier, Contr. U.S. Natl. Herb. 13: 105. 1910. type: Costa Rica,

Brc7ics s.n. (K, US).

Slender vines. Brown, glandular pubcrulent throughout, stems, petioles, and
inflorescences also with hispid-pilose hairs to 3 mm long. Leaves ovate, apically
i-:inrMiii(>- li;is:illv conLitc. 6.5-8.5 ctu loiui, 3.5-4.5 cm wide, the basal sinus nar-

^^""1 SI'KLLMAN— I'ANAMAMAX ASCLEPIADACEAE J3J^

row, 1-1.5 cm deep, the upper surface puberulent with scattered liispid-pilose
hairs, tlie lower surface ghuidular puberuleut, tlie veins and veinlets l)econiing
conspicuously dark upon drying; petioles 2.5-3 cm long. Inflorescences condensed

racemes, appearing ± umbelliform, ca. I()-15-flowered, but with the flowers
opening in pairs; peduncles 2.8-3 em long; pedicels 1.2-1.7 cm long. Fhncen ca.
1 cm long at anthesis; calyx denseh' glandular x^nberulent and pilose abaxially,
glabrous adaxially, the lobes lanceolate to ovate-acuminate, 2.5-3 nnn long,
1-1.6 nnn wide; corolla ca. 1.2 cm in diameter when extended, rotate but the lobes
sharply reflexed at anthesis, the tube 0.6-0.7 mm long, die lobes elliptic to ovate,
obtuse, 4.7-5 nnn l(mg, 3.3-3.7 mm wide, the upper surface villous-pilose, rarely
glabrous or nearly so, the trichomes ca. 1.5 nun long, the lower surface glandular
puberulent except at the base of the lobes and near the margins, the margins
ciliate widi cilia 1-2 mm long; corona annular, purple black, carno.se, ca. 1 mm
high, the surface bullate-collieulate; g>nostegium slightly exceeding the corona,
die head obscurely 5-gonal. Follicles (from Costa Riean material) said to be
ellipsoid-attenuate, dark olive gr(HMi, shiny, armed with soft, blunt-tipped .spines
or tubercles, the innnature follicles 16.5 em long.

Matelea pseudoharlnda is known from elevations of 1,000 to 2,000 m in Costa
l^ica.

This .species is closely related to, and was mistakenlv reported as, M. pingui-
folia (Standley) Woods, in the I'loru of Panama, a species stricdy limited to the
lowlands. The rounded cordate leaves w itli inconspicuous veins and lack of long
trichomes in the inflorescence eharaeterize .)/. pinguifolia and clearly separate it

from M. paeudobarbata.

cuiHiyei: At opening to caii\(in to Banihito, 5,000 ft, 28 June 1969, Ti/.son 5860 (DUKE
FSU, MO).

4. Gonolobus rothscluihii Sehleehter, Hot. Jahrb. Syst. 60: 368. 1926. rveE;
Nicaragua, RothschiiJi 557 ( H, destro\cd, x^hoto MO).

luacluiia ItvlcwphijUu Hemsf, Biol. Centr. Ainer., But. 2: 230. 1881. tvi-k: XieaiaKiia Tate
171 (240) (K).

Stcnus puberulent and hispid-pilose, the longest trichomes to 3 nm-i long.
Leaves elliptic, apically rounded and abruptly acuminate, basally truncate to
shallowly cordate, mostly 12-15 cm long, 5.5-7 cm wide, the upper surface
strigose with hairs 1.5-2 mm long, the lower surface sparsely hirsute; petioles
puberulent and pilose, 2-3 cm long. Inflorescences contracted racemes, ap-
parently ca. 10-flowered but with usually no more than 2 open at one time;
peduncles 3-5 mm long, .sparsely hispid-pilose; pedicels 18-21 mm long, hispid-
pilose. Flowers 2..5-3 cm in diameter; calyx abaxially pilose, adaxially glabrous
but for a median line of pilose trichomes, the lobes linear-lanceolate, sharply
reflexed at anthesis, mostly ca. 8.5 mm long, 0.9-1.5 mm wide; corolla green be-
coming greenish bronze, rotate, deeply cut, the lobes linear-lanceolate, mostly 14
mm long, 3-3.5 nnn wide, die upper surface glabrous, the lower surface pilose;
gynostegium 1.4-1.7 nnn high, the anther aj^pendages cuneate, bilobed, die
lobes diverging to form a conca\e triangle, ca. 1 mm long, 1.5 mm wide, die lobes

132

ANNALS OF THE MISSOURI BOTANICAL GARDEN LVol. 64

curling in drying; faucal annulus only sliglitly raised and somcwiiat fleshy,
nearly obscured by the corona; corona remotely 5-gonal, fleshy, the margins
thickened, rugose; ovaries 3-ridged, glabrous. Follicles unknown.

This species is typically found at elevations from 1,200 to 1,800 m in Costa
Rica, Honduras, Nicaragua, and possibly Guatemala.

A distinctive species, Gonolohus rothschuhii is separated from all other Pana-
manian members of the genus by its very narrow corolla lobes and color, and
by the shape of its anther appendages.

DARiEx: Cerro Tacaiciina, south slope, ridge-lop forest well below summit, premoiitane
wet forest, 1,250-1,450 iii, 26 Jan. 1975, Gcntni ij Mori 13921 (MO).

LlTEUATURE Cn ED

Spellman, D. L. 1975. Asclepiadaeeac. In \\. E. Woodson, Jr. & R. W. Sehery, Flora of
Panama. Ann. Missouri Bot. Gard. 62: 103-156.

NEW SPECIES OF GIBSONIOTHAMNUS
SCROPHULARIACEAE/BIGNONIACEAE ) AND
rOURNEFORriA (BORAGINACEAE) FROM

EASTERN PANAMA AND THE CHOCO^

Alwyx II. Oknthy-

AlJSTHACT

Tliree new species are described from wet forest regions of eastern Panama and the Choeo
of Colombia — Gibsouiotliatntius alatus A. ('entr\, Gihsoniothammis mirificus A. Gentry, and
TournefoHia tacarcuncnsis A. Gentry & No\\'icke.

Gibsoniothamniis alatus A. Gentry, sp. nov.

Frutex epiph) ticus, Kanuili irregulariter teret(\s. Folia elliptiea, aenta vel acuminata,
cimcata, glabra praetor domatia ciliata. Floies singnlares, pedicellis glabris. Calyx late
alatus, ad instar stellac, alis ultra 1 cm longis. Ct)r()l]a (non vidi) alba.

Epiphytic shrub. IJranchlets irregularly terete to subangulate, very sparsely
pilose. Leaves elliptic, acute to acuminate, euneate at the base, chartaceous to
snbcoriaceoiis, glabrous above and below except for ciliate domatia in the axils
of the lower secondary ner\'es, gland dotted bcdow, the margin entire, veiy slightly
or not at all revolute, drying dark olive abo\e, light olive below^, the secondar}^
\Tins plane or slightly impressed abo\'e, prominulous to pronn'nent below. In-
florescence a single flower; pedicel glabrous, 1.5-2 cm long. Calyx very broadly
winged, glabrous except a few trichomes near the ends of the wings, almost
star shaped, ca. 6-7 mm long and wide without the wings, the wings each over
1 cm long, tapering to an acute point. Corolla (not seen) white. Pistil 23 mm
long, the o\ aiy globose, the stjle slender, LS mm long. Fruit white, covered by
the calyx.

Type: Panama, hariex: N slopes of Cerro Pirre, lower montane rain forest
(cloud forest), 700-950 m, 6 Apr. 1975, Mori 6 KaUunki 5449 (MO, holotype;
isotypes to be distributed) .

Additional collection (examined: Pantama. daiuen: Cerro Canipanicnto, S of Cerro Pirre,
elfin forest, 20-22 Mar. 1968, Duke 15657 (MO).

This species is utterly distinct in the genus because of its laterally winged,
star-shaped calyx. Its closest relative is G. ptcrocahjx A. Gentry but that species
has much narrower longitudinally oriented teeth.

Gibsoniothamniis mirificus A. Gentry, sp. nov.

Frntex epiphyticns. Rannili irregnlaritcr snbangnlati, pilosi. Folia ohoxato-elliptica,
ohtnsa, cnneata, conspicne pilosa. Flores singulares, pedicellis pilosis. Calyx cnpnlatns, pilosns,
\ aide 5-dentatus, dentihus linearihns, 2—2.5 cm longis. Corolla tnbulosa, rnbra.

^Support for this work was provided by NSF grant OIP75-18202.

"Missouri Botanical Garden, 2345 Tower Croxe Avenue, St. Loin's, Missouri 63110

\nv. Mlssouiu Bor. (Iaiu). (iJ: 133-135. 1977.

134 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Epipliytic shrill). Braiichlcts pilose, irregularly subangulate. Leaves obovate-
elliptic, obtuse to acutish at the a^iex, cuneate at the base, chartaceous to sub-
coriaceous, 3-9 cm long, 1.5-4 cm wide, conspicuously pilose with 1-2-mm-long
trichomes above and below, the margin entire, vcv}' slightly or not at all revolute,
drying brownish olive above, tannish yellow^ below, the secondary veins plane
above, prominent below; petiole densely pilose, ca. 5 mm long. Inflorescence a
single flower; pedicel conspicuously pilose, 0.7-1 cm long. Calyx cupular, pilose,
strikingly 5-toothed, 5-7 mm long and 4-6 nnn wide without the teeth, the 5
linear teeth exceeding the calyx by 2-2.5 cm, pilose, extended along the calyx as
lateral ridges. Corolla red, tubular, glal)r()iis, 3.5-4.3 cm long and 0.5-0.6 cm

wide, lobes 3 mm long, with ciliatc* margins. Fruits not se<Mi.

Type: Panama, colon: Santa Rita Ridge Road along trail from end of road
(10.6 km from Transisthmian Highway, 3 km beyond hydrographic station) to
Rio Indio, 380 m, 13 Apr. 1976, Croat 3429S (MO, holotype; PMA, isotype).

Additional collection examined: Panama, colon: Plant collected by II. Wiehler on
Santa Kitu Ridge, cultix'ated at Marie Selby Botanical Gardens, Sarasota, Florida, Dressier
s.n. (MO).

The striking calyx teeth of this species are by far the longest in the genns.
It is otluMwise similar to Costa Rican C e})iphyticus (Standi.) L. Wins, which
is also more or less pilose thronghont bnt has very nnich shorter calyx teeth, a
fasciculate sc\'eral-flowercd inflorescence, and more coriaceous leaves.

Tournefortia tacaicunensis A. Gentry & Nowicke, sp. nov.

Herba erecta. Folia angnste elliptica, acuta, enneata, siibsessilia, glabrcscentia. Inflores-
centia scorpioidea, floribiis sepalis lanceolatis, ca. 4 mm longis, corollae tubo 8-9 mm loiigo,
lobis ca. 1.5 mm longis.

Herb 0.2-0.5 m; stems glabrescent. T^eaves alternate, narrowly elliptic, acute,
cuneate at the base, entire, witli 4-7 pairs of strongly ascending secondary nerxeSj
8-28 cm long, 2-6.5 cm wide, glabrous above, glabrescent below, rather suc-
culent when fresh, membranous when dry, drying dark brown above, tannish
gray below; petiole essentially lacking. Inflorescence scorpioid, contracted, 3-4
cm long, terminal; pedicels mostly 1-2 mm long. Calyx of 5(6) free sepals, lanceo-
late, ca. 4 mm long, sparsely puberulous; corolla orangish to greenish cream,
sparsely pubemlous outside, the tube 8-9 mm long, the lobes ca. 1.5 mm long;
stamens 5(6), borne near the apex of the corolla tube, sessile, the anthers ca. 1.2
mm long; ovary ovoid, the style ca. 6 mm long, the stigma ccmical, 1 mm long.
Fruit not seen.

Type: Panama, datuen: Cerro Tacarcuna, W ridge, trail from summit camp
to waterfall E of camp, 1,550-1,700 m, lower montaiu^ wet forest life zone, herb
0.5 m, flowers greenish cream, tin^ning tannish, inflorescence scorpioid, stamens
equalling petal number, ovary 2-locular with axile placentation, 2 Feb. 1975,
Gentry ir Mori 14114 ( MO, holotype) .

Additional collection examined: Colomiua. c:iioc6: Slopes of Serrania del Daricn, E of
Unguia, preniontane \\et forest, ca. 1,300 m, herb 0.2 m, flowers orangish, 19 July 1976,
Gentry, Leon ^ Forero 16772 (COL, MO).

I977J SPELLMAX— PANAMAMAX ASCLKPIADACEAK ^35

This species seems remarkably distinct from all other members of the genus.
Tt is apparently the only clearly herbaceous species of Tournefortia (a com-
pletely unrelated species, T. sihirica L. is a wiiy herb but often segregated as
Mcsserschmidia) , Another unusual feature is a tendency to 6-parted flowers.

The pollen of T. tacarcunensis is of the type described as *Type II," 3-col-
porate, subprolate with expanded poles, psilate at the poles and verrucate at the
equator by Nowicke & Skvarla ( 1974 ) .

The closest relative of T, tacarcwieims may be T. ramoncmis Standi, of up-
land Costa Rica and Chiriqui Province— though omitted from the Flora of Pan-
ama treatment of the family (Nowicke, 1969). That species has generally similar
flowers which differ in the corolla lacing densely and rather strigosely pubescent
outside. The inflorescence of T. ranwnensls is also mucli more elongate and
the flowers are essentially sessile. Vegetatively T. tacarcunenms differs con-
spicuously in its glabrescent, narrowly elliptic leaves which arc narrow!)^ cuneatCj
essentially sessile, and have only about 6 pairs of secondary nerves. Another Costa
Rican relative is 1\ hrencsii Standi., w^hich agrees in ptxlicellate flowers and an
onb' sparsely puberulous corolla but has nnich longcM* (4 mm) corolla lobes and
a loni>;-i)etiol(xl leaf with 15 secondarv nerves on each side.

T.iTEHAruiiE CrrLi)

W)\viCKi:, J. W. 1969. Borapinacfiie. hi Robert E. Wc
of PaiKinui. Ann. Missouri Bot. Card. 56: 33-69.

— 6i J. J. Skvahla. 1974. A palynological irncvstigation of t1ic gcntis Toumcfortia
(Bora^nnaccae). Aiikm'. J. Bot. 61: 1021-1036.

NOTES

NEW COMBINATIONS IN EPILOBIUM (ONAGRACEAE)

In the course of preparation of a revision of the North American species of
Epilohium, we ha\T found the new combinations proposed in tliis paper to be
desirable. They will be fully discussed and justified in our subsequent publica-
tions, but are offered at this time, with minimal synonymy, in order to malce the
names available.

Epilobium ciliatum Raf., Med. Repos. II. 5: 361. 1808.
Epilohiiini ciliatum subsp. ciliatum.

E. adcnocauhm Hansskn., Oesterr. Bot. Z. 29: 119. 1879.

E. leptocorpinn Ilausskii. \ar. macinmii TitI., Animal Rep. Missouri Bol. Card. 2: 103. 1S91.
E. hracliycarpttm sensu Munz, Aliso 4: 489. 1960; N. AnuT. Fl., ser. 2, 5: 218. 1965, iioii
Presl 1831.

Epilobium ciliatum subsp. glandulosum (Lehm.) Hocb & Raven, eomb. nov.
Based on E. ^lamhdostim Lehm., Stirp. Pug. 2: 14. 1830.

E. horcalc Ilausskn., Monogr. Epil. 279. 1884.

E. cxaUatuJH sensu auct. mult.; non Drew, Bull. Torrey Bot. Club 16: 151. 1889.

Epilobium ciliatiun sul)sp. watsonii (Barbcy) Hoch & Raven, comb. no\'.
Based on E. watsonii Barbcy, in Brew. & S. Wats., Bot. Calif. 1: 219, 1876.

Epilobium hornemannii Reichcnb., Icon. Crit. 2: 73, fifi. 313,
Epilobium hornemannii subsp. hornemannii.

Epilobium hornemannii subsp. behringianum (Hausslcn.) Iloch & RaviMi,
comb. nov. Based on E. hchrbv^iumim Ilausskn., Monogr. Epil. 277. 1881.

Support from {\\v U. S. National Science Fonnclation to Peter Raven is gratefn1I\- ac]<noul-
edgcd.

— Peter C. Uoch and Feter 11. Raven^ Mmoiiri Botanical Garden, 2345 Tower
Grove Aventte, St. Louis, Missouri 63110.

CHROMOSOME NUMBER IN riLLANSlA (IRIDACEAE)

The cbroinosonu' number in the Soutli African monotypic genus PiUansia
was prc\iously reported (Goldblatt, 1971) as 2n — 44. New material obtained
subsequently proved without doubt to be 2;i — 40 (Fig. 1). The earlier record
was ()l)tained from paraffin sections of root tips and from antlier squashes where
22 bivalents were noted. Plants in tlie earlier study were all from Rooi Els,
Bettys Ray, Cape Prov., South Africa, [Goldblatt 471 (BOL)], and those used in
the present work were from Arieskraal, Cape Prov., South Africa, [Powrie s,n.

1977J

NOTES

137

PIT' ■■■

■w-

■4y ■

^^J -■_>-■!

_^_ 1- ■> -_^ :k jj_

V-T".

-3v .™- "■-rM=ft^- - - ^ ^-'^ - ■ ^^■^'^^ -

■^F' irih^ -^ lL-> > ■ . .V. '^ ^'-^^

^

f »

i'lcuRE 1. Mrtaphase cliroiiiosonics of Fillait.sia tcnij^IcDiaiiu stained witli laclo-propionic

fllToill; X 1,200.

(MO)] soiiu" distance away. A .squasli tcchniciuc was used with the new ma-
terial, and rout tips were treated as d(>seribed elsewhcM-e (Goldl^latt, 1976).

Pillansia is a inonotypie genus of Iridaceae belonging to the exclusixely Old
World and predomhiantly African subfamily Ixioideae. The only speci(\s P.
tcnipJentanii (Baker) L. Bolus is rare and occurs in a very localized area of the
southwestern Cap(> Province of South Africa. Although it is undoubtedly a mem-
ber of the Ixioideae, it is peculiar in this subfaniiK in scn eral respects and hence
oi particular interest. Most remarkable is its paniculate inflorescence, (^uite un-
like the spike, txpical of the subfamily. The branched panicle is beliexed to be

an ancestral conditi(Ju from which the spike was deri\ed, which suggests that
Pillansia is a primitive Ixioid. A second peculiarity of PiUamia is that the conns
are persistent, lasting several seasons instead of being annual as is usual in the
Ixioideae. Lewis (1954) suggested that this condition in Pillansia was possibly
transitional in the e\'olution of the conn from a rhizome. Thus Pillansia, primi-
tive in both its inflorescence and rootstock, is probably a significant e\'olutionary
link between the Ixoideae and less speciahzcd subfamilies of Iridaceae, or their

ancestors.

The chromosome number of 2)i

20 reported here confirms the polyploid con-

dition in the genus. The closest allies of Pillansia— Watsonia, Thcrcianlhiis, and
Micranflms which together comprise subtribe Watsoniinae (Goldblatt 1971) —
are in contrast basically diploid, with x - 10 in Thcrcianlhus and Micranthus and
9 in Watsonia. As it is unlikel\- that Pillansia is heteroploid, the earlier report
22 appears erroneous. Unfortunately, no more material is available at

.V

o

f n

present to investigate the situation more fnlly.

138

ANNALS OF THE MISSOURI BOTANICAL GARDLN [Vul. 64

LlTERATrKE CiTED

r.oi.nm.A-ri , P. 1971. Cytological and inorpholoj^ical studies in the soutlicrn African Iridaccac.

-160

1976. Chionuisonie number and its significance in Balis niaritiiiia (Hatacoar).

J. Arnold Arbor. 57: 526-530.
Lkwis, C. J. 1954. Some aspects of the inorpliology, phylogeny and taxouoiu) of tlie South

African Iridaccae. Ann. S. African Mus. 40: 15-113.

Peter GoUlhlatt, B. A. Knikoff Curator of African Botany, Missouri Botanical
Garden, 2345 Tower Grove Avenue, St. Louis, Missouri 63110.

A NEW J AC ARAN DA (BIGNONIACEAE) FROM

ECUADOR AND PERU

Jacaranda sparrci A. Gentry, sp. iiov.

Arbor. Folia pinnatini biconiposita, plerumque 13-pinnata, pinuis 13-21-fi)liolatis, foliolis
1-2 cm longis, 0.5-1 cm latis, apiculatis. Flor calyce fere palelliformi, 5-den(alo, corolla
tubulo-campanulata supra basim angustatam arcuatani, extus puberula, stauu'nodio exserto,
antlieris 1-thccatis, ovario pnberulo. Fnictus ignotus.

Tree; l)ranchlets subtetragonal, very minutely puberulous, with wliitisli Icnti-
cels. Leaves pinnately l)ic()mpouncl, u.sually with 13 pinnae, each pinna with a
sliglitly winged raeliis and 13-21 sessile, asymmetrically oblong leaflets, these
1-2 cm long and 0.,5-l cm wide, apiculate, glabrescent alx)ve, barbate at least
along the base of midvein below. Inflorescence an open terminal panicle, puberu-
lous. Flowers with the calyx almost patelliform, shallowly 5-dentate, ca. 2 mm
long and 5 mm wide, puberulous; corolla purplish blue, tnbular-campanulate
above a narrow neck \\'hich is conspicuously curved and enlarged toward the
base, 2.5-3 cm long, 1.1-1.3 cm wide at the mouth, the lobes small, less than 5
mm long, the whole tube pubendous outside, glabrous inside except at the stamen
insertion; stamens didynamous, the anthers 1-thecate, the second theca reduced
to a minute appendage, each theca 3-4 mm long, the staminode 2.5-3 cm long,
siibexserted, the middle third and apex glandular pubescent; ovar>' flattencnl-

ovatc, 2 nun long, 2 nnn wide, densely pid>erulous. Fruit not seen.

Tyi'k: Fcuadoh. lojA: Between Panamerican Highway and Zumbi on road
to Nfachala, km. 69, dry ciuebrada \egetati()n, 2100 m, 23 Sep. 1967, Sparre 18862
(MO, holotype).

Additional collecliou exauu'ncd: Feru. piuha: Ayabaca, Oct. 1S68, Titujmmuli 1252
(USM).

This species is exactl)' intermediate between /. acutifoUa II. & B. and /. nwnosi-
folia D. Don on the one hand and the /. caucana complex on the other. It has the
relatively large leaflets and pubescent ovaiy of /. caucana Pittier but the pubescent
corolla tube of /. mimosifolia. The cur\'ature and enlarged base of the corolla are
more pronounced than in /. acutifoUa but less so than in /. caucana. Neither of
these species has such reduced corolla lobes nor notably ex«?erted staminodes as

1977]

NOTES

139

/. sporrei. Jaca

dry inter-Andean valleys of Peru, while /

/. actitif'

inter-Andean Cauca and Magdalena valleys of Colombia north to Costa Rica.

Supported by NSF Cranl DEB75-2()325 AOI.

Alwijn II. Gentry, Missouri Botanical Garden, 2345 Tower Grove Avenue, St.
Louis, Missouri 63110.

PHYLLARTHRON BILABIATUM: A NEW SPECIES

OF BIGNONIACEAE FROM MADAGASCAR

Phyllarthron bilabiatum A. Centry, sp. nov.— Fic. 1.

A P. vmdagoficaricnsc foliis angustis iienatma indistiiuta, a P, Jiumblotiann cal\cc S-cnstato,
et ab aniliabus foliis verticillatis et calyce l)ilal)iato diffurt.

Large tree to 25 m tall and 0.7 ni d.b.h., the tnmk convoluted with deep \ erti-
cal fissures, the branchlets subterete to subtriangular, glabrous. Leaves verticil-
late in 3's, of 2 superposed articles; petiole ca. 1 cm long; basal article very nar-
rowly oblanceolate-oblong, cuneate to the base, rounded at the apex, 3.5-7 cm
long, 1.5-2.2 cm wide, the second article very narrowly elliptic or elliptic-oblong,

rounded at the base, obtuse to subacute or cmarginate at the apex, 2-7 cm long
1-2.6 cm wide; drying olive to gray above, brownish beneath, glabrous, coriaceous,
the margins strongly revolute, the secondary nerves very obscure, hardly or not
at all visible. Infloresence a .short terminal panicle, the lateral branches op-
posite, each with 1 or 3 flowers; bracts and bracteolcs minute, deciduous. C^aly:

campanulate, strongly bilabiate, 12-13 mm long, 7-9 mm wide, split over Vs its
length (ca. 5 mm), with 5 conspicuous longitudinal ridges, these terminating in
minute denticulations, glandular and drying with a varnished surface, otherwise
glabrous. Corolla (single mature corolla seen) magenta with the top of the
throat darker, the floor of the throat white with yellow ridges, tubular-infundibuli-
form, 4.6 cm long, ca. 1.5 cm wide at the mouth of the tube, the tube 2.6 cm long,
the lobes 1.2-1.5 cm long, pnberulous outside and on the lobes and floor of the
tube inside, the lobes also glandular-lepidote. Stamens included, the anther
thecae divaricate; pistil and disc not examined. Fnu't unknown.

Type: Madagascar. niEco-suAREZ: Tsaratanana Massif, trail up S ridge of
Maramokotro, Andohanisambirano, 2,000-2,500 m, montane cloud forest, 9 May
1974, Gentry 11612 [MO, holotype; P, TAN, Service Forestierc (Madagascar),
isotypes].

ThyUarthron hihhiatum is most closely related to P. mmlagascaricnse (Boj.)
K. Schnm. and P. humblotianum Perrier. Its strongly 5-ribl)ed calyx suggests the
former. Its narrow leaves with indistinct venation and revolute marjiins sunfiest

C)' "V^fi

the latter. Neither of these species has whorled leaves. The leaves of P. hiJahia-
tum are conspicuously decurrent so that its branchlets appear almost triangular

140

ANXALS OF THE MISSOURI BOTANICAL GAHPKN

[Vol. 64

C-

Fi<;uni, 1. Ilahil oi rliylhiiiJiwu hihhhitum A. Gentry (X"4(i). [Cni/n/ IIB12 (MO).]

in cross-section, The most noteworthy floral character of this species is a strongly
bilabiate calyx which is matched in the genus only by the very different P. mc^.(ip-

tcrtim Perricr.

lY'rrier de la Bathie (1938a, 1938b) noted the variability of juvenile leaves of
this genus and excluded them from consideration in his key and species descrip-
tions. I have followed suit in using only the mature foliage in the description of
the new species. However, a sterile collection from the type locality [Gentry
IIGIS (MO) described as a sterile treelet 2 m tall] certainly represents a juvenile
form of P. hilaJnaium. The leaves of this collection are larger and thinner than

1977]

NOTES

141

niaturc leavers and liave secondary venation approacliing Uiat of P. nuidagas-
caricmc, but their wliorled placement agrees witli P. hHal)iatuni. Tlie stem of
this plant is distinctixeh triangular from the strongly decurrcnt lca\(\s^ a charac-
ter which appears to distinguish juxcnile plants of P. hilahialutn from juvenile
forms of an\' otluM" species of the genus.

LrrKHATUUE CrrED

Pi:iuui:h dk la Baihik, U. 1938a. Lrs Bignoniacees d(^ la region Malgachc. Ann. Tnst.

Bot.-Cc'ol. Colon. Marseille, ser. 5, f): l-IOl, pis. 1-18.
. l*)o(Sl>. Bignoniaeee.s. hi II. Ihunbert (editor), Flore de Madagascar, 17S*' l^iniille.

91 pjx

— Alinju II. Gcnlnj^ Missouri Botanical Garden, 2315 Toiccr Grove Avenue, Si.
Louis^ Missouri 631 KX

TAXONONIIC NOTES AND NEW COMBINATIONS IN

LKUCOrilYSALIS (SOLANACEAE

Inuring the course of a re\'isionary study of Chamacsaracha (Averett, 1973),
several species \v(M'e encountered which, at one time or another, had been
assigned to hut arc clearly not a part of Chamacsaracha. .\h)st of these species
were relegated, either by me or previous workers, to Lcucophijsalis or the Asian
genus Phijsahastrum. In dealing with the misplaced species, the close relation-
ship of rhijsahaslnnn to the North American genus LcucopJujsalis became ap-
parent, but since the* species were removed from CJuanacsaracha, the question
of the two being congeneric was postponed (Averett, 1971). The data now at
hand indicate that the species of PhijsaUastruni are clearl\- related to and are best
treated as LeucophijsaJis. The latter treatment necessitatc\s s(^\-eral nomenclatural

changes. Since my revision of the genus wall not appear for sev(Mal months, it

seems adx'isabh^ to make the followinii" new combinations at tliis litne:

Leucophysalis kweichoiiense (Kuang & Lu) Averctt, comb. nov.

rinjsaluisiruui hwciclioiicuse Knang (^ Lu, Acta Ph>tota\. Sin. 10; o51. 1905. rvn:: (^Inna,
Kweiehon rr()\ ince, Keili, Mao^^'n, 750 in, Chang Yon^ticn 13^)6 (Sit, ]i()l()t\pe, not seen).

Leiicophysalis sinicum (Kuang & Lu) Averc^tt, comb. nov.

Physaliastnnn sininun Kuang & Tai, Aeta Thytotax. Sin. 10; 352. 19G5. t^j'l;: China, Sliansi
rrovinee, Wci-fjhig Usia (321 ( SII, liolotvpe, not seen).

Leiicophysalis yunnanense (Kuang & Lu) Av(Tctt, comb. nov.

rJnjsalicLstnun ijtnuuuicnsc Kuang & Lu, Acta Phytotax. 10; 3-18. 1905. tyi'K: CIn'n i, Vmman
Province, Sunning, 1800 ni, T. T. Yu 16767 (SH, ]u)]ol>pe, not seen).

Leiicophysalis jajKmica (Fr. & Sav.) Averctt, coml), now

Cluumicsariha japonica Fr. ^- Sav., Eniun. PI. Jap. 2; 15 1. 1879. typi:: Japan, Ito K(M"ske,
Tanaka, Saiaticr 2166 (not seen).

142 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

C. echinata Yatahe, Bot. Mag. (Tokyo) 5: 355. 1891. type: Not designated.
Phtjsaliastrum echinatum (Yatabe) Makino, Bot. Mag. (Tokyo) 28: 20, 1914.
P. japonicum (Fr. & Sav.) Honda, Bot. Mag. (Tokyo) 45: 139. 1931.
P. japonicum (Fr. & Sav.) Kitaniura, Acta Phytotax. Geobot. 6: 19. 1937.

Leucophysalis savatieii (Makino) Averett, comb. nov.

Chamaesarcha .savatieii Makino, lUustr. Fl. Jap. I. (11): 1. Oct. 9, 1891. lectotype: Japan,

Kegon, Nikko. Oct. 5, 1890, Mahino s.n. (MAK).
C. waiaiiabei Yatabe, Bot. Mag. (Tokyo) 5: 315. Oct 10, 1891.
Physaliaatrum savatieri (Makino) Makino, Bot. Mag. (Tokyo) 28: 22. 1914.

Leucopliysalis kimuiai (Makino) Averettj comb. nov.

Fhijaaliastruin hniurai Makino, J. Jap. Bot. 3: 37. 1926. type: Japan, Musaslii Province, Mt.

Takao, rare, 24 Oct. 1926, K. Kimura s.n. (not seen).
P. savatieri f. kimurai (Makino) Oliwi, Fl. Jap. 1020. 1956.

Nine species in total, two North American and seven Asian, are considered
to compose tlie genus Leucophysalis,

Litfratuhe CriED
AvEUETT, J. E. 1971. New combinations in tlie Solaneae ( Solanaceae ) and comments

regarding the taxonomy of Leucophysalis. Ann. Mis.souri Bot. Card. 57: 380-382.

1973. Biosystematic study of Chamaesarcha (Solanaceae). Rhodora 75: 325-365.

John E. Averett, Department of Biology, University of Missouri-St. Louis, St.
Louis, Missouri 63121.

CHROMOSOME NUMBERS OF PHANEROGAMS. 7'

Counts by Charles Albert Huckins, Missouri Botanical Garden, 2345 Tower
Grove Avenue, St. Louis, Missouri 63110.

llOSACEAE

Mahis haccata (L.) Borkliausen. 2n == 34. U.S.A. Washington, d.c: Cul-

107683

66042101 (BH).

Malus haccata var. himalaica (Maximowicz) Schneider. 2n = 34. U.S.A.
MASSACHUSETTS: Cultivated, Arnold Arboretum accession number A101-34-C,
Huckins 6Q050S13 {mi) .

Mahis fhnentina (Zuccagni) Schneider. 2n = 34. U.S.A. Washington, d.c:
Cultivated, U.S. National Arboretimi accession number 3361, Iluckim 67081802
(BII).

A/fi/i/v/M.vcYi (Rafinesque) Schneider. 2n = 34. U.S.A. new york: Cultivated,
Durand-Eastman Park accession number 77, Iluckins 69052703 (BII).

Mahis rockii Rehder. 2n = 51. U.S.A. Massachusetts: Cultivated, Arnold
Arboretum accession number A 83-84, Iluckins 70031902 (BII).

^ The previous number in this series appeared in Ann. Missouri Bot. Card. 62; 513. 1975.

1977]

NOTES

143

Mains sikkimensb (Wenzig) Kochne ex Schneider. 2n = 51. U.S.A. new
YORK: Cultivated, Durand-Eastman Park accession number 841, Htickim
69052105 (BH) .

Mains tschonoskii (Maximowicz) Schneider. 2n = 34. U.S.A. new yohk: Cul-

r

tivated, Cornell University Horticulture Dcx^artnicnt accession number URI-

12, Htickim 68041701 ( BII ) ,

Mains tmnnanensis (Fr^Anohf^t) ^rhunlAnv

In — 34. U.S.A. NEW york: Cul-

Huckim 69050601 (BH).

accession

T-'

sue of the Annals of the J
896, was published on 14 J

PREPARATION OF MANUSCRIPT

The Annals publishes original manuscripts in systematic botany and related fields
Authors are asked to follow the suggestions below in order to expedite eltZJdpuhlicSS^'
If an author feels that h,s manuscript presents special problems, he shoull wite ?he eSr
concernmg the best way to handle these before submitting the manuscript

Manuscripts must be typewritten
manuscript

footnotes: legends, tables Hsts of Tpedmens and ThYKography'^^^F"^^^^^^^^^^^

'^"m l^ ^^^i separately and placed at the end of the text Legends fJ^eac W?oup ?f f S

't^nrh^.T\f ■^A-'"''^ parapph using the style found in the%urrent isSe S S?e AnSS

Session oli'i'nwSLr *5' ™^'^T *^^ approximate places for illustrations and table !
Submission ot the ongmal and one carbon or xerographic copy of the manuscript is required
and the author should also retain a copy of the final, t>T)ed draft. '»"u:.i.ripc is requirea,

placed

and

Z<SST:°Z^6l^t.fte' '°°*"°*^ ™ ^"^'^ "'"'■ '"'"' -*or-s Ml moling address should
u^r^^'^'lS/f^T^l^^tfTLZ'tl']!':^^^^^^^^^ ?he abstract should succinctly

summarize
itself.

A brief Latin diagnosis for each new taxon is preferred to a complete Latin description
A complete description should be given in English. p ^ ^ i-<i"n uescnpnon.

The citation of specimens should be concise. Geographic names are nut in ordf^r of

locahty

ated

PpriodJ'^/r'^Sf ' ft °"^'?1 ^1^^^^!^^,^ fo"^ consistency and to make sure they are unambiguous.
Periods are used after all abbreviations except measures, compass directions, and herbarium

designations.

AU illustrative material should be mounted on stiff cardboard. If the originals are too
large to be conveniently mailed, photographic reductions should be submitted. The maximum

since they are not printed as "plates.

Sf «l^^"K^'^.'?'i'*>"°''il^ ^.^^* ^"^ ^"^ therefore, the height of an illustration must
not exceed about 1.5 its width. Figures are numherpd onn.ecutively with Arabic numerals,

„„fi, A„:4. c I'll"". j^ 1 *',"' — r'".:s "'"•"^ "^ uwne with a mechanical device or

with dry-transfer lettenng and never by hand. The amount of reduction should be noted on

the back of each illustration, together with the figure numbers, author's name and title of the
paper. Photographs should be sharp, glossy prints. Numbering should be applied directly to the
surface of the photograph. Several photographs may be assembled to form a composite block,
and each photograph should be numbered separately. The individual photographs should be
mounted with all edges flush. Line drawings are prepared with India ink and must never be
placed m the same block with photographs. Authors wishing to have original illustrations
returned must notify the editor when proofs are returned.

, ^ . :p—f-.v,. r --i-^-- with care. The author should check to make sure

each entry in the bibliography is referred to in the text and each reference to a paper in the
text IS entered in the bibliography. Citations of periodical literature should appear as follows :
Authors last name, initial(s). Year. Full title of article. Name of journal. Volume: inclusive
pagmation. Nanies of journals are abbreviated according to Botanico-Periodicum-Huntianum.
Citations ot books should appear as follows: Author's last name, initial(s). Year. Full title.
Edition, if applicable. Publisher, place of publication.

J - - „^., „ - ^ pS, format of keys, indentation of

synonyms, and format of Notes, consult the current issue of the Annals.

Authors will receive page proof. Since changes in proof are expensive, authors should not
make revisions at this stage. Authors will be charged for alterations other than corrections of
printer s errors.

Reorints may be purchased at cost plus postage. An order blank for reprints will be
enclosed with proofs. A reprint price list is available from the editor.

r> J^ J!},Vl^^^P^ should be addressed to the Editor, Gerrit Davidse, Missouri Botanical
Garden, 2345 Tower Grove Avenue, St. Louis, Missouri 63110.

Publications of the

imm/ki

ANIMALS

The Annals of the Missouri Botanical Garden contains con-
tributions mainly in plant systematics. The Annals appears four
times a year, and four numbers constitute a volume. Many back
issues are available. The prices below are strictly net; there is no
discount to agents.

Subscription price $40.00 per volume, U.S., Canada,

and Mexico
45.00 per volume, foreign

FLORA OF PAIMAIVIA

The Flora of Tanama treats the various famihes of indigenous
and naturahzed plants of the Republic of Panama and the Panama
Canal Zone. Each family is dealt with in a critical manner with
synonymy, descriptions, discussions, illustrations, and citation of
representative Panamanian collections for each species. The Flora

Annals of the Missouri Botanical Garden

as

contributions accumulate, but

parts are available separately.

INDEX

The Annals of the Missouri Botanical Garden are now de-
voted primarily to plant systematics, but early volumes contain
numerous papers on topics ranging from carbohydrate metabolism in
Azotohacter to pod and stem blight of soybean. Volumes 1-15 con-

of viral particles in volume 8.

M. Duggar published an early paper on the

by

dexed.

1914-1968 )

Annals are

the 769 titles contributed

$9.00

Place orders for all publications with:

Allen Press, Inc.

1041 New Hampshire Street

Lawrence, Kansas 66044

OF THE

VOLUME 64

I

1977

• '^

?'.»

>1

: r;i

f> fe

^1*.

*- _^

■ i _

-■■ .-^ ■.;:-^;.- ;■- ..:;*;

■ ■

lV\

■''>

*^\'*

( . . . -*

*':v'*'

: -i^ «

^ L

f.

I*

• ^* r^

*■

■r

\ *'

iV ^

"•^

-t I

X

\

' ■^

I V -^.

•■f».

*»k»

,-/

■^ "

- f

ii

vj

I'

A'

#^ I

t.

-^rj

■ '. ^ ' - *

- C.

i

NUMBER 2

— : ^

^

.r V

V'C,

I- .

\ •* ^»

J i.'

\*^

* 1

' I,

r _>- ■->

u -'

f ;.

.' t

'^

:y

r AJ» 1 1 lj

f *

'V

I'

U>'

#*

■^

*^.:

^ -l-f

J.

C_*'=--

ll

1 r

t>

r* ^

* ^»*

.* «

f t

\-

1 ^

^

(

V.

t

f

■ y

/^/

>:^^A^J.

y^i*.

I. -

I, 1

f

f- •

^ ■

Wj»

■f:

/'i

*T--

1

.--^

'/

t -*

\

^^

'^

)

'7^ \

^,

7

I.

*•

■*

^ 1 .

\

■rv,

T *q

* ^

y^

»x

^ »

J .

w *

4 *

t *

h d

L

CONTENTS

JOHN S. LEHMANN BUILDING, MISSOURI BOTANICAL GARDEN

JOHN S. LEHMA

JBARDEN UBRA«^

CHEMOSYSTEMATICS: The Twenty-third Systematics Symposium John

E. Averett

145

Perspectives in Plant Serotaxonomy David E. Fairhrothers

147

Electrophoretic Evidence and Plant Systematics L. D. Gottlieb 161

Molecul

Rates, Regula-

tion, and the Role of Natural Selection Mary-Claire King ,_... 181

Chemosystematics — Analysis of Populational Differentiation and Variability
of Ancestral and Recent Populations of Juniperus ashei Robert P.
Adams ... 184

The Order Centrospermae Tom }. Mabry

210

Defensive Ecology of the Cruciferae Paul Feeny

221

Chemosystematics and its Effect upon the Traditionalist B. L. Turner

235

(Contents continued on back cover)

VOLUME 64

1977
NUMBER 2

OF THE

/km

The Annals contains papers, primarily in systematic botany,

contributed from the Missouri Botanical Garden. Papers originating

outside the Garden will also be accepted. Authors should write the

editor for information concerning arrangements for publishing in
the Annals.

Edftorial Committee

Gehrit Davidse, Editor-in-Chief
Missouri Botanical Garden

W. G. D'Arcy, Editor— Flora of Panama

Missouri Botanical Garden

John D. Dwyer
Missouri Botanical Garden ir St, Louis University

Peter Goldblatt
Missouri Botanical Garden

Published four times a year by the Missouri Botanical Garden Press,

St. Louis, Missouri 63110.

or subscription information contact the Business Office of the Annals
P.O. Box 368, 1041 New Hampshire, Lawrence, Kansas 66044.

Subscription price is $40 per volume U.S., Canada, and Mexico,

$45 all other countries. Four issues per volume.

Second class postage paid at Lawrence, Kansas 66044

© Missouri Botanical Garden 1978

OF THE

mm

m

mu

VOLUME 64

1977

NUMBER 2

CHEMOSYSTEMATICS: THE TWENTY-THIRD

SYSTEMATICS SYMPOSIUM

John E. Averett^

dealing

The following seven papers were presented at the Missouri Botanical Gar-
den's Twenty-third Annual Systeniatics Symposium held 15-16 October 1976.
The symposium, sponsored in part by the National Science Foundation, was at-
tended by approximately 300 scientists. This year's topic w^as chemosystematics.

The studies of R. E. Alston and B. L. Turner in the early 1960s which utilized
flavonoid patterns in elucidating complex hybridization and their publication of
Biochemical Systeniatics in 1963 were instrumental in bringing chemistry to the
systematic community. In 1962 the first international conference on biochemical
systeniatics was held, and the proceedings were published the following year
in Chemical Plant Taxonomy (Swain, 1963). Taxonomic Biochemistry and
Serology, also proceedings of an international conference held in 1962, ap-
peared in 1964 (Leone, 1964). Through the 1960s numerous papers
with natural products appeared in the literature and by 1970 the field of bio-
chemical systeniatics or chemotaxonomy was well established.

In 1972 tlie International Union of Pure and Applied Chemistry held a sym-
posium in Strasbourg on "Chemistry in Evolution and Systeniatics" (Swain,
1973). In the following year the topic of the 25th Nobel Symposium was
^'Chemistry in Botanical Classification" (Bendz & Santesson, 1974). The latter
was an attempt to bring chemists and taxonomists together, in what V. II. Hey-
w^ood has referred to as the milikely marriage of an exact science with one less
estricted that ventures into every other discipline.

Early studies utilized compounds, and except for serology, largely secondaiy
compounds, in resoKing hybridization complexes or as "finger prints" for charac-
terizing species or, occasionally, higher taxa. However, concomitant w^ith the
development of the field, techni(iues and instrumentation necessary for the iso-
lation and structural characterization of the compounds were refined. Further,

I

^Department of Biology, University of Missouri-St. Louis, St. Louis, Missouri 63121
Moderator of the Twenty-third Annual Systeniatics S\Tnposiuni.

Ann. Missouiu Bor. Gaud. 64: 145-146.

]^46 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

information on biosynthetic pathways and genetics of the secondary compounds
accumulated. Now it is relatively easy to assay the complex constitution of a
plant. Structural and biosynthetic data greatly enhance the utility of the com-
pounds in phylogenetic interpretations. The extent to which this is true, as
well as the sophistication that has developed within the field, is well illustrated
in the following papers.

Following the format of the symposium, the first three papers discuss the
utility of macromolecules and next three papers discuss micromolecules. The
final paper on the effect of chemistry upon traditional taxonomists was presented
as an evening talk by B. L. Turner.

Literature Cited

Alston, R. E. & B. L. Turner. 1963. Biochemical Systematics. Prentice Hall, Inc., En-

glewood Cliffs, New Jersey.
Benuz, C. & J. Santesson (editors), 1974. Chemistry in Botanical Classification. Proc.

25th Nobel Symposium, 1973, Sweden. Academic Press, New York.
Leone, C. A. (editor), 1964. Taxonomic Biochemistry and Serology. Ronald Press, New

York.
Swain, T. (editor). 1963. Chemical Plant Taxonomy. Academic Press, New York.
(editor). 1973. Chemistry in Evolution and Systematics. Butterworth & Co., Ltd.,

London.

PERSPECTIVES IN PLANT SEROTAXONOMY

I

David E. Fairbrotiiers-

Ahsthact

The capacity to view recent data in proper relation to other information, or the ahility
to correctly jndge the significance of facts and ideas, requires a knowledge of botli the past
mistakes and the forward strides within a discipline. This paper is intended to help tlie
reader formulate perspectives concerning 65 years of plant serota.xonomic research. The
disco\ery that the innnune reaction was only relatively specific and that the degree of cross-
reactivity was essentially proportional to the degree of relationships ])etween organisms had
important implications for comparative systematic serology. It is the specific reactions,
between determinants and antideterminants, which provide a measurement of protein simi-
larities. The comparison of protein mixtures, rather than purified single proteins, has
dominated taxonomic research l)ecanse such an approacli provides serological overall similarity,
and thus a multicharacter comparison. Tlie "antisystematic** reactions have recently been
shown to result from variation in the systematic ranges of deternn'nants; and the absorption
( presaturation ) techni(iue for removing common determinants increases the accuracy of
serological placements. The following items were evaluated: antigenic preparations, ad-
juvants, injection procedures, single versus mixed protein extractions, kind of plant tissue
extracted, and the interference of secondary compounds. Comus canadensis and C. succica
were found to be serologically very similar. The tested species of the genus Comus were
divided into three distinct serological groupings. The serological data support the separation
of the Cornaceae and Nyssaceae; and the inclusion of Caniptolheca and Nys.sa in the Nys-
soideae, and Davidia in the Davidioideae, both of the family Nijssaceae, Nyssa hiflora and
N. sylvatica were serologically very sinu'lar; A^. o^cche and N. aquaiica were serologically
distinct from each other and from N. hiflora and N. sylvatica. Nyssa ogcchc was the most dis-
tinct species of the genus. Cowkia cotoneaster had very little serological similarity with any
of the tested species of the Cornales.

To have the capacity to view recent data in proper relation to otlier infor-
mation, or the ability to correctly judge the significance of facts and ideas, re-
quires a knowledge of past mistakes and the forward strides within a discipline.
It also requires a degree of knowledge of the individual components as well as
the total products resulting from the various component combinations. The
author hopes this manuscript will provide the reader with the necessary infor-
mation and literature citations which will allow the formulation of perspectives
concerning 65 years of plant serotaxonomic research.

The ^present age" of chemosystematics or chemotaxonomic publications com-
menced to appear in the early 195()'s. The oldest of the "present age" approaches
is serotaxonomy and the newest is amino acid sequencing (Cronquist, 1976).

The discovery of serological reactions in Austria via the occurrence of precipi-
tin reactions took place SO years ago (Kraus, 1897). This discovery provided a
new technique which was soon used to aid in the investigation of systematic
problems in animals, ^^^ithin two years after the discovery of the precipitin re-

^ I dedicate this puhlication to the memory of my friend and fellow phytoserologist, Dr
Josef Kloz, who died in his 55tli year on Octoher 22, 1976 in Czechoslovakia.
The research was snpported by NSF Grants CB-13202 and BMS 75-17805.

^ Department of Botany, Rutgers University, New Bnmswick, New Jersey 08903.

Ann. Missouiu Bot. Gauu. 64: 147-160.

i^o ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

action the technique was applied to comparative problems by the Frenchman
Bordet (1899). Tliis was followed by a series of extensive comparative studies
which were conducted with various animals by the Englishman Nuttall (1901,
1904). Thus biologists have known for 75 years that organisms may share anti-
genic material (substances capable of inducing the formation of antibodies
and able to react with the antibodies); and that when they share the same anti-
genic material in different proportions it is assumed that the organisms are re-
lated. At first it was believed that the immune reaction was absolutely specific,
that is, that an antiserum would react only with the antigen that stimulated its
production. However, Bordet (1899) in conducting research with birds reported
that the reaction was only relatively specific and that the degree of cross-reactivity
was essentially proportional to the degree of relationships between organisms.
It was this early discovery which had important implications for systematics and
started the pioneer investigations in the discipline of comparative animal sys-
tematic serology.

A large number of animal systematic serological publications have been re-
ported in the bibliography prepared by Leone (1968) and the book edited by
Wright (1974). The pioneering work in the United States essentially began witli
Boyden (1926), and he has continuously contributed to the field of animal
systematic serology for 50 years (Boyden, 1973; Wright, 1974). Approximately
550 plant taxa (cultivars through orders) have been included in approximately
160 systematic serological publications since 1950 ( Fairbrothers, 1969a; Fair-
brothers et al., 1975).

Serology is concerned with the interactions of antigens and antibodies and/
or antibodylike substances, the lectins. The term "serology" is often used synony-
mously with immunology. However, some biologists prefer "serology" because
"immunology'' has an implication that immunity is concerned in all reactions be-
tween antigens and antibodies (Boyden, 1948).

Principles of Sei^ology

A consideration of a few basic facts related to the biology of the immune
response is a valuable aid to understanding the methods and inteipretations of

systematic serological research.

The term "antigenic" is relative since the response is frequently a property
of the route of injection, method of preparation of the antigenic material, and
the individual experimental animal used. Thus it is important that details about
such items and procedures should always be included as a portion of the metliods
and materials section of publications. Experimental data have demonstrated
the following: (1) There is a certain minimal molecular weight below which
substances are not in themselves antigenic. Some of the lower molecular w^eight
substances can become antigenic by mixing them with other substances (ad-
juvants). (2) Size alone is not enough to guarantee that a molecule will be
antigenic. Immunochemists indicate that the specific action is in part due to
the rigidity of certain chemical structures (determinants) which are difficult to
distort or alter. (3) Usually a molecule must be foreign to an organism to be
immunogenic. (4) Substances must be soluble or be able to be broken down

1977] FAIRBHOTIIERS— PLANT SKROTAXONOMY 149

into soluble antigenic components before being capable of inducing antibody
formation. (5) Too much antigenic material may cause immunological paralysis
(i.e., cannot be immunized). There appears to ])e a balance between stimula-
tion and paralysis for each antigenic material. (6) Many proteins have been
found to be immunogenic, and the best known immunogens are the proteins
with molecular weights of 40,000 or more (Abramoff & La Via, 1970).

Of the several kinds of serological reactions, the precipitin reaction has been
used most frequently in plant comparative serological mvestigations. It is a rela-
tively simple reaction capable of being applied to the comparison of the soluble
protein antigenic material extracted from all kinds of plants. Microcomplement
fixation has proved valuable in animal systematics (Champion, et al., 1974),
but has had practically no application in comparable plant research. Serological
research can be conducted employing (juantitative precipitation, precipitin tech-
niques in solutions, or by various qualitative precipitation techniques in gels.
The serological cliaracteristics of proteins are linked with the primary structure
of the molecule. The reaction is concerned with points on the molecule (determi-
nants) which are capable of initiating the production of immuno-globulins only
in certain cells of animals (not plants). These immuno-globulins possess proper-
ties accounting for the bonding to the respective protein reaction position. Thus
the serological characteristics of the protein are found in the determinants, which
are restricted to certain positions of the molecules.

A fairly accurate estimation of the size of determinants has been obtained
from protein fractionation experiments designed to detect the smallest molecule
fraction still capable of an immunological response. Arnon & Sela (1969) and
others have demonstrated that the active antigenic regions of proteins are com-
posed of 10-20 amino acids.

Systematists and taxonomists are interested in the comparison of antigenic
determinants from various taxa. It is the specific reactions between determinants
(antigens) and antideterminants (antibodies) which are valuable because they
provide a means for the measurement of protein similarities.

Metiioix)lo(;y

When deciding the type of antigenic preparation, the process of denatura-
tion, which means structural changes with concomitant loss of })io]ogical proper-
ties, must be taken into account. Protein antigens are not equal in susceptibility
to denaturation, IIowc\xm*, all such changes result in some loss of original specific-

ity.

The use or nonuse of adjuvants (Freund's, in our experiments) to increase the
level of an immune response (imnuino-enhancement) is discussed in systematic
serological research. The puipose of an adjuvant is to heighten and prolong the
immune response; and the value of this additional material must be judged for
each antigenic material. This means its use or nonuse should be decided after
experimentation.

The effect of injection procedures on the systematic reaction range has been
tested by various experiments. One of the very early reports indicating that longer
immunization periods extend the reaction range was conducted with a 7.ea tiunjs

250 ANNALS OF THE MISSOURI BOTANICAL GARDEN IVni.. 64

antiserum (Magnus, 1908). Lake et al. (1914) using purified seed protein ma-
terial in contrast to crude seed extracts as used by Magnus also found that a
longer injection period extended the reaction range of the antisera. Several
recent publications, with both plant and animal antigens in the form of purified
or mixed reagents, indicate that the systematic reaction ranges of antisera are
extended by injections continued until a maximum reaction is reached (Boyden,
1971, 1973). Such experiments indicate also that long continued immunization
is likely to induce the formation of greater proportions of cross -reacting anti-

which will reduce the discriminating capacity of the antiserum. Antisera
I from several and long injection periods may reach a higher level of

bodies, w

derived J
"fidelity''

d

search (Moritz, 1964). Leone (1952) demonstrated that longer immunization
periods tend to cause a lower discriminating capacity.

Therefore, the number of injection series used should be stated so the reader
can make proper comparisons. The use of a combination of "short'* and "long"
injection series may produce the largest amount of data for comparative sero-
logical investigations. This should not be considered experimental "manipula-
tion" because it is merely using serological techniques to the fullest advantage
based upon our present knowledge of the immune response.

There are two main approaches to serological research: (1) comparison of
single proteins, or (2) comparison of protein mixtures. The second approach has
dominated taxonomic research (Moritz, 1964; Fairbrothers, 1968; Jensen, 1974a).
An example of the comparison of a single protein is best illustrated by the re-
search with phascolin obtained from Phaseohis vulgaris, and the ribulose-1,5
biphosphatc carboxylase ("fraction I") found in green plants. Data obtained
from such research indicates that the systematic ranges of determinants vary.
Some determinants are found throughout the plant kingdom, while others have
a very restricted distribution. In general, the low taxonomic yield from single
protein investigations has not justified the large preparation effort required
(Jensen, 1974a). Serological comparison of protein mixtures which were ex-
tracted from seeds, pollen, spores, tubers, or leaves is most common. The re-
sults provide a serological overall similarity and thus a multicharacter compari-
son. Researchers working with plant protein extracts also find tannins, saponins,
alkaloids, lipids, and/or polysaccharides, which may have to be inactivated, re-
duced, or eliminated by diverse extraction procedures.

When using a pure protein, only a very limited number of determinants can
be compared. In contrast, when using mixtures of proteins, data from many dif-
ferent determinants are tested, and thus the chances of being misled in terms of
serological correspondence are lessened.

U k ^»»

Antisystematic Reactions

A phenomenon which has been designated "antisystematic," "asystcmatic, or

"unexpected"

very

spective (Moritz & Rohn, 1956; Proline et al., 1961; Moritz, 1964). Jensen (1974a,
1974b) indicated that these terms are no longer used by the above authors be-
cause past usage assumed serological convergence, which has been shown not to

1977] FAIRBROTIIKRS— PLANT SEROTAXONOMY ]^5]^

be the causative agent for such responses. Moritz (1964) included several re-
ports which illustrated his designated ''antisystematic" reaction. He also indi-
cated why such reactions were specific serological reactions, and not some kind
of non-specific effect, since they disappeared when presaturation experiments
were conducted.

The practice of some researchers during the last several years of designating
cross-reactions with a wide systematic range as "antisystematic" should be dis-
continued. Such wide-range reactions are the result of certain determinants be-
ing widely distributed in the plant kingdom. In other words, they should be con-
sidered as reactions of determinants that are widespread and remain relatively

constant.

th

search with Fraction I Protein (ribuIose-1,5 biphosphate carboxylase) extracted
from the tissue of green plants. The serological partial identity detected between
wide-ranging taxonomic groups has been shown to be the result of parts of the
protein structure unaltered in the course of long periods of evolution (Sugiyama
et al, 1969; Jensen, 1974a, 1974b). Thus we now know that the systematic ranges
of determinants do vary, and sometimes protein molecules carry several de-
terminants which reveal partial serological identities.

The understanding of the above reactions is important because it has sho\Mi
the value of the presaturation (absorption) technique for removing common
determinants and leaving only those systems specific for each taxon compared,
thus providing both a more accurate serological placement and measure of the
relative similarity.

The use of various techniques to remove nonspecific reactants which react
with normal rabbit serum (NRS) has become standard practice in present-day
plant serological research. We now know such responses often come from sero-
logical reactions resulting from the presence of lectins. Lectins can be removed
by hemagglutination techniques and thus l)e prevented from interfering with
normal serological reactions (Lee & Fairbrothers, 1972).

Thus in recent years experiments have provided answers to some of the for-
mer peq^lexing problems associated with plant systematic serological research.
This has been very valuable and allowed the continued development of such
research. It also assures that a larger spectrum of species can be compared by
using the techniques which are now available, and our percentage of accuracy
in terms of serological placements, is continuing to increase.

History of Plant Serology

As with animals, serological techniques were used in plant systematics and
taxonomy soon after discH)very. Mez of the Botanical Institute, University of
Konigsberg, Germany, conducted such research with his students and colleagues
from 1911-1936 (Mez & Gohlke, 1914; Mez, 1922). The "Konigsberger Sero-
diagnostik Stammbaum" (phylogenetic tree) was the climax of ye-ars of research
(Mez & Ziegenspeck, 1926). This group was known as the Konigsberg Sero-
logical School, in contrast to the Berlin Serological School which was headed by
Professors Gilg and Schiirhoff, who conducted research duriiig the 192()'s (Gilg
& Schiirhoff, 1926). These two groups of researchers (schools) conducted a

]^52 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

literary feud wliich seriously jeopardized the credibility of plant systematic
serological research for essentially 25 years. The techniques employed in the
early research proved to have several serious flaws. Mez's technique, where he
produced an immune serum through the influence of antigen on serum and
eliminated the use of living animals as antisera producers, was a serological
'^disaster." The vast amount of data reported using this procedure was not valid

and has been disregarded.

It was Moritz (1928, 1964) and his colleagues at the University of Kiel,
Germany, who, working from 1928 until the present, revealed the value of sero-
logical research for plant systematics. Jensen, formerly from Kiel, has recently
organized another serological laboratory at the University of Cologne, Frohne
continues the systematic serological research at Kiel.

In the United States Chester published plant serological papers in the 1920*s
and 1930's and also prepared a comprehensive critique about plant systematic
serology (Chester, 1937). In 1947 Johnson (1951), with his students, started
the present United States trend toward plant systematic serological research.
It was lie who introduced me to the techniques in 1957 after I had joined the
faculty of Rutgers University.

In 1953 Urano (1955) started phytoserological investigations in Japan, and
S. Sakaguchi and S. Arai have continued this research (Fairbrothers, 1969a).
The Solanum serological research in Birmingham, England was started in 1955
(Cell et al., 1956). J. Hawkes (Birmingham) and his students (Lester and P.
M, Smith) have continued serological research to the present time. In Prague,
Czechoslovakia in the late 1950's the husband and wife team, Kloz and Klozova,
started and continued serological investigations of Vicia faha and other legumes
(Kloz et al., 1960). The year 1963 saw the start of another plant serological
research center headed by Vaughan in London, and he has continued his multi-
disciplinmy Brassica investigations to the present time (Vaughan & Waite, 1965;
Vaughan et al., 1976). Cristofolini (1968) has published several reports from
his botanical laboratory in Italy since beginning his plant serological research
in 1966. The most recently organized plant systematic serological laboratory is
under tlie leadership of Drs. Morozova, Chupov and Kutjavina at the Komarov
Botanical Institute, Leningrad, USSR (Fairbrothers, 1975).

Interpreting the Results

Research has demonstrated that extracts of seeds, pollen, leaves, tubers, and
spores of vascular plants can be used, if the required extraction procedures are
followed (Fairbrothers, 1969b, 1975; Fairbrothers et al., 1975). However, most
systematic serological research has included seed material, due to the relatively
high concentration of proteins, relative ease of collecting, and relative ease of
assuring comparable developmental stages. We are presently pursuing the fol-
lowing two new studies in our chemosystematic laboratory using pollen as the
source of protein material: (1) serological investigation of selected amentiferons
taxa with Frank Petersen, and (2) a serological investigation of the Corylaceae
(Betulaceae) with Friedrich Brunner.

1977] FAIRHROTHERS— rLAXT SEUOTAXONOMY ]^53

Pliytoscrological research has provided provocative and vahiable data for
use ill the classification of flowering plants. The numerous examples cited in tlie
evaluation of the contribution of serological data related to Cronquist's and
Taklitajan's systems of classification were shown to be significant (Fairbrothers
et ah, 1975). This publication indicates that such data have contriljuted in the
classification of the following orders, and the placement of families within these
orders: Capparales, Caryophyllales, Cornales, Dipsacales, Illicialcs, Lamiales,
Magnoliales, Nelumbonales, Nymphaeales, Papaverales, Polemoniales, Ranun-
culales, Rubiales, Scrophulariales, Typhales, and Umbellales. In addition to
these orders, significant contributions have also been published for species,
genera, and/or tribes belonging to the following families: Ammiaceae, Ber-
beridaceae, Brassicaceae, Caprifoliaceae, Cucurbitaceae, Chenopodiaceae, Cor-
naceae, Fabaceae, Lamiaceae, Magnoliaceae, Nelumbonaceae, Nymphaeaceae,
Nyssaceae, Papaveraceae, Poaceae, Ranunculaceae, Solanaccae, and Typhaceae.

The serologic and disc electrophoretic characterization and comparison of
the spore proteins extracted from Osmnnda cinnamomea^ O. cJaijtoniana^ and
O. regaJis illustrated that fern spores were suitable material for such analyses.
Osmunda cinnamomea and O. claijtoniana were shown to possess greater protein
affinities for each other than either had for O. re<^alis\ Osmunda re<i^(dis\ in gen-
eral/ had greater protein affinities for O. claytoniana than it had for O. c'//i-
namomea (Petersen & Fairbrothers, 1971). Stein & Thompson (1975) compared
the same three Osmunda species using DXA hybridization techniques and in-
dependently indicated the same relationships reported by Petersen & Fairbrothers
(1971). Miller (1967), based on anatomical characters of living and fossil speci-
mens, indicated that O. claytoniana and O. regaJis were more closely related,
while Hewits()n*s (1963) anatomical and morphological research indicated tluit
O. cinnamomea and O. claytoniana had a closer relationship wdth each other
than either had with O. regalis.

The serological investigation of intra- and interfamily relationships of the
Cornaceae and Nyssaceae has continued intermittently in our chemosystematics
laboratory for 15 years, and various experiments have been conducted as ap-
propriate and adequate plant materials became available. In our systematic
serological research it has been expedient to conduct several projects simultane-
ously because no experiments can be conducted until adequate and appropri-
ate materials are available for the extraction of proteins, and until antiscra to
perform the essential experiments have been raised.

Cornus canadensis and C. siiecica have been foimd serologically very similar
based on photronreflectometer tests, Ouchterlony plates, and absorped and non-
absorbed antisera. These two taxa have also been shown to be the most dis-
similar from other taxa placed in the genus Cornus (Fairbrothers, 1966a, 1966b,
1968). When data were evaluated from cytology, morphology, anatomy, geo-
graphical distribution, and the j)iitiitive hybrid (C. unalaschkensis) , the close
similarity between the two was also detected. I l)elieve all the data indicate that
the two named taxa are subspecies of one circumboreal species which is very
distinct from the other species of Cornus. If there is justification for dividing
the genus Cornus into distinct genera, then this species (or two species) w^ould

154

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

"3

Q

^

-t^

o

u:

2 '

a;

W5

O

CD

8

C

o

O

■5 ^

^ r

0)

i

^ o

•i ^

Q

ft

pa ■—

■

Sp:

;3

1-

(X

a:

Si I

^

«

P2

=3 00

CO

5 J.

o

S

<

^

I-

CO

CO

CO

in

00

o>

lO

05

CO

lO

o

lO

00 >0 fM CD

CD O^ 00

CD

^ CD

^ CD

CO

00

eg

IC t-- CD O CO

■^ t- O 00

o

CO
CO

o o

in

CO

00 tP

lO

»0 05

CO CD

05 ^
00 Gi

OIOOCMCDOCOOO

CO CD -^ CO CO (N <M

O

CO

^ -H CO

CO CO

05

IC lO fM lO

00

00 CO CD CD

CO CO CM CM CO

I

CO
05

CO "^ 00 O
00 l> lO Tt* IC

O CD

CO CO

CO CO t-

co cq

««

5D

3 ^

CO

2 «

S -5

O
^

5 -c ^ -.5

a

u

Si

o o o o o o u

:?: ^ ^ Q

o

c

O

2

a

a;

o

O

0)

c
o

o

■a

£

a

•c

a

0)

o

u

a

o

.S

B

§ ^

"S

>

5 s

en

a;

? .2

4^ rt D

v:tD ft.

e

Si

1977]

FAIRBROTIIERS -PLANT SKROTAXOXOMY

155

o

u

CO

a;

a;

c

O

S

a;

o

c

03

c3 .

S a
15 «

o -2

a!

c
<1

§

C
O

It

CD

^'ci

r— I

^

U

^
^

5 ci

h

H

^

H

/^

tJ

O

h

^

H

^

H

H

X

S
u

<

fM fM Ol CO CO

CO ^ eg

CM

rM

oi CM

CM ^ O

CM CM ei CO CO CO lO CO

el

O CM

CM CO C^l IC

r>\ oi

c-i CO -r CO

oi oi

CO -^ CM

Ol CM C^l CO

CO Tt<

f>J

r-i Ol CM

CM

Tt^ Cvl CO

CO n* CvJ CM C^l CM

CM

Ol CM

Ol -^

C^J CM TT

^ O

lO CO CM CM Cvl CM CM

CM

lO CM O

Co

«

^

a
o

^

^

"S

■JC S. :;

u o o

:< k ^ Q

aj

A

^^

s

-D

hi

4 ^4

C3

Ci.

■ A

^

;c

CJl

#^

c

TS

o

C

C/1

rt

c^

-Q

^

o

«a-i

O

0^

Cfl

-n

0)

£

^^^

^

:3

fl

■— s

^^

9^

^M

-s

5

u.

X

C3

0^

o

Ert

(A

>

%

c

H

•si

C5 -^

en C

I ^1^

1 =

oj

t/5

t3 o
:: c

(A

-^

00

-4

fl J r^

CD
Oi

CD

156

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 61

best qualify for such a designation, and would have to be given the generic name
of Chauuieperichjmenuni.

Serological data have also indicated that within the genus Conms there are
the following three distinct groupings: (1) C. florida, C. kousa, and C. nni-
talUi; (2) C. omomum, C. stohnifcro, and C. racemosa; and (3) C. camdensis
and C. suecica ( Fairbrothers & Johnson, 1964; Fairbrothers, 1966a, 1966b, 1968).
These serological groupings correspond to the Cornus subgenera designated by
Ferguson (1966), except he placed C. Jiousa in a subgenus distinct from
that containing C. jlorida and C. nuttallii. Newer serological data from our lab-
oratory based upon additional antisera and many more experiments continue
to support a tripartite taxonomic disposition of the taxa placed in the genus Cor-
nus (Tables 1-2). The use of nonflavonoid glucosides as taxonomic markers
in the genus Cornus was reported by Jensen et al. (1975). Their suggested ar-
rangement of subgenera based on the presence or absence of iridoids, phis the
type of iridoid constituents agree with the reported serological groupings and
would correspond to their ( A/B ) , ( C ) , and ( F/G/H ) designations.

The families Cornaceae and Nyssaceae were recognized by Dumortier in
1829. However, this separation into two families was not followed by most
taxonomists for over 100 years. Recently Melchior (1964), Cronquist (1968),
Thorne (1968, 1976), Takhtajan (1969), and Dahlgren (1975) have recognized
two families. In addition, in most recent classifications the genus Davidia has
been removed from the Nyssaceae and placed in the Davidiaceae (Melchior,
1964; Cronquist, 1968; Takhtajan, 1969; Dahlgren, 1975). Thorne (1968, 1976)
placed Davidia in the subfamily Davidioideae of Nyssaceae following W'agcrin
(1910). Harms (1898) was the first author to use the two subfamihes Davidi-
oideae and Nyssoideae, and he placed them both in the family Cornaceae.

The serological data support the separation of the Cornaceae and Nyssaceae,
and the grouping of Comptotheca, Davidia, and Nyssa within the Nyssaceae
(Tables 1-2). At present I believe the serological data best support the place-
ment of Davidia in the Davidioideae, and Camptotheca and Nyssa in the
Nyssoideae both of the family Nyssaceae (Fairbrothers & Johnson, 1964; Fair-
brothers, 1966a, 1966b, 1968; Tables 1-2).

Perdue et al. (1970) reported that tests with Camptotheca acuminata dem-
onstrated that crude extracts exhibited significant activity against l>'mphoid leu-
kemia. This comprehensive report discussed the relationships of Camptotheca
widiin the Nyssaceae, indicating that Camptotheca was closely related to Nyssa
and only remotely related to Davidia. Research done by Titman (1949) using
wood anatomy, Eydc (1963) using fruit structure and the fossil record, and
Sohma (1963) using pollen support the taxonomic conclusions of Perdue et al.
(1970). Our recent serological data also indicate that Camptotheca is more
similar to Nyssa than to Davidia, and that Nyssa is more similar to Davidia than
is Camptotheca (Tables 1-2).

Our serological data are also supported in part by the findings of Hohn &
Meinschcin (1976) based on the fatty acid composition of seeds. They indi-
cated that primitive Davidia and advanced Camptotheca were placed on each
side of Nyssa, which is intermediate. Thus all the data presented lend credence

1977] FAIRBROTHERS PLANT SEROTAXOXOMY

157

to the placement of CamptotJieca and Nyssa in tlie snbfamily Nyssoideae and
Davidia in tlie Davidioideac of the Nyssaceae.

Serological experiments with fonr species of Nyssa have included compari-
sons of Nysm aquatica, N. hiflora, N. v^eche, and N. sylvatica. The various
experiments indicated that N. hiflora and N. .sylvatica were serologically very
similar. Ny.ssa ogeche and N. aquatica were serologically distinct from each

■fi

Ogeche.

mi and N. sylvatica. Tlie data also showed that scro-
s more similar to N. hiflora and N. sylvatica than to N.
eche was more similar to N. aquatica than to N. hiflora
and N. sylvatica, Nyssa ogeche was the most distinct species of the fonr com-
pared ( Fairbrothers & Johnson, 1964; Fairbrothers, 1966a, 1966b, 1968; Tables
1-2). The serological data snpport the conclusions of both Eyde (1963) and
Sohma (1963), who reported close similarity between N. hiflora and N. syl-
vatica and treated them as two varieties of one species. The serological data
does not support the findings of Hohn & Meinschein (1976) based on seed oil
fatty acids. They indicated IV. hiflora and N. sylvatica to be chemically dis-
tinguishable species. The various researchers agree that N. ogeche is the most
distinct from the other three species of Nyssa. The serological data lias not
clearly indicated whether Camptotheca, Davidia^ or Nyssa has the greatest
similarity with Cornus. Tlie three genera are serologically relatively similar to
Cornus; how^ever, the data indicate that Camptotheca might have sliglitly more
similarity with Cornus than do Nyssa or Davidia (Tables 1-2).

The genus Corokia (6 species) is restricted to the South Pacific region, rang-
ing from northern New South Wales, Lord Howe Island, New Zealand, Chatham
Islands, and Rapa Island, a distance of 4,000 miles.

The serological data reveal very little similarity ]>etween Corokia cotonecistcr
and any species of the Cornaceae and Nyssaceae tested (Fairbrothers et al.,
1975; Tables 1-2).

Most researchers have indicated that this genus has little affinity with mem-
bers of the Cornaceae in which it is often placed. Some botanists have suggested

an affinity with the Saxifragaceae within the subfamily Escallonioideae (Fscal-
loniaceae) (Philipson, 1967; Smith, 1958). Eyde (1966, 1967) concluded that

1

it was unrelated to Cornus but was possibly linked with ArgopJiyllum. Kubitski
(1963) retained the genus in the Cornaceae. Hegnauer (1965) indicated that the
Cornaceae may be related to either the Saxifragaceae or Loganiaceae. Taklita-
jan (1969) excluded the genus Corokia from the Comales and placed it in the
Escalloniaceae (Saxifragales). Cronquist (1968) considered Corokia as a pos-
sible nonmissing link between the Cornaceae and Escalloniaceae, Grossularia-
ceae, or Saxifragaceae sensu lato. Both Cronquist's and Taklitajan's classifica-
tions reflect the serological data which indicate the distinctiveness of Corokia
from members of the Comales,

However, Bate-Smith et al. (1975) investigated the distribution of several
chemical compounds in the Cornales and concluded that Corokia possesses a
chemical pattern consistent with that of the Cornaceae.

The experimental investigation of taxa within the Cornales has inchcated
that the use of diverse discii)lines has x^rovided valuable data for helping to

]^58 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

understand the evolntionary development and relationships of the families and
genera in the order. This order has proven to have been an excellent one for
diverse chemosystematic researcli. Serological comparisons have provided signifi-
cant data for evalnation in the continuing investigation of diverse coraaccous,

sensii lato, taxa.

Taxonomy eventually must strive to bring together, summarize, and utilize

what is known about the organisms to be compared. Systematic serologists have
essentially learned to use the properties of one of the classes of proteins, gamma
globulins, in comparative studies. Systematic serology provides comparisons
which are relatively objective measurements; however, like all detected relation-
ships, they are relative and not absolute. Thus, as stated in the first paragraph of
this paper, I hope that the included information has helped the reader to formu-
late perspectives concerning plant serotaxonomic research.

Literature Cited

Adramoi-f, p. & M. F. La Via. 1970. Biology of the Liunune Response. McGraw-Hill,

New York.
AuxoN, R. & M. Sela. 1969. Antibodies to a uni(ine region in lysozyme provoked by a
synthetie antigen conjugate. Proc. Natl. Acad. U.S.A. 62: 163-170.

Bate-Smith, E. C, L K. Fercusox, K. Hutson, S. R. Jensex, B. J. Nielsex & T. Swaix.

1975. Phytochcmical interrelationships in the Cornaeeae. Biocheni. Syst. Ecol. 3: 79-89.
BouuET, J. 1899. Sur Tagglutination et la dissolution des globules rouges par le sennu

d'aninuxux injectes de sang defibrine. Ann. Inst. Pasteur 13: 225-250.
BoYDEN, A. 1926. The precipitin reaction in the study of animal relationships. Biol. Bull.

Ahu\ Biol. Lab. Woods Hole 50: 73-107.

. 1948. "Serology" and "immunology." Science 108; 635.

. 1971. Concerning the specificity of precipitin and other serological reactions.

Bull. Serol. Mus. 46: 1-7.

1973. Perspectives in Zoology. Pergamon Press, England.

Champiox, A. B., E. M. Prager, D. Wac:hteu & A. C. Wilsox. 1974. Microcomplement
fixation. Pp. 397-416, in C. A. Wright (editor), Biochemical and Imnumological Taxon-
omy of Animals. Academic Press, London.

Chester, K. S. 1937. A critique of plant serology. Quart. Rev. Biol. 12: 19-46, 165-190,

294-321.

CnisTOFOLixr, G. 1968. A serotaxonomic approach to the complex ''Salicornia herbacea'

L. Ciorn. Bot. Ital. 102: 555-556.
Cronquist, a. 1968. The Evolution and Classification of Flowering Plants. Houghton

Mifflin Co., Boston.

. 1976. The taxonomic significance of the structure of plant proteins: a classical

tax(momist's view. Brittonia28: 1-27.

Uaiilgrex, R. 1975. A s>'stem of classification of the angiosperms to be used to demon-
strate the distribution of characters. Bot. Not. 128: 119-147.

DuMonriER, B. C. 1829. Analyse des families des plantes, des principaux genres qui s'y

rattachent. Tournay, France.
Eyde, R. H. 1963. Morphological and paleobotanical studies of the Nyssaceae, I. J. Arnold

Arbor. 44: 1-52.
. 1966. Systeuiatic anatomy of the flower and fruit of Corokia. Amer. J. Bot. 53:

833-847.

— . 1967. Tlie peculiar gynoecial vasculature of Cornaeeae and its systematic signifi-

cance. Phytomorphology 17: 172-182.
FAmRROTHERS, D. E. 1966a. The comparison and interpretation of serological data in
plant systematics. Pp. 458-464, m Z. Landa (editor), Symposium on the Mutational

Process. Symp. CSAV Praha.
-. 1966b. Comparati\'e serological studies in plant systematics. Bull. Serol. Mus. 35:

2-6.

— . 1968. Chemosystematics with emphasis on systematic serology. Pp. 141-174, iu
V. 11. lleywood (t*ditor), Modern Methods in Plant TaxoiH)n»y. Academic Press, London.

19'^7] FAIRBROTHERS— PLANT SEROTAXOXOMY

159

1969a. Plant scrotaxonomic (serosystematic) literature, 1951-1968. Bull. Serol.

Mus. 35: 2-6.

— . 1969b. Comparisons of proteins obtained from diverse plant organs for chemo-

systematic researeli. Rev. Roumaine Biocliim. 6; 95-103.

— . 1975. Advanees in phytoserologieal research. Bull. Serol. Mus. 51: 8.

& M. A. JoHXsoN. 1964. Comparative serological studies within the families Cor-

naceae (Dogwood) and Nyssacoae (Sour Cum). Pp. 305-318, in C. A. Le(me (editor),
Taxonomie Biochemistry and Serology. Ronald Press, New York.

, T. J. Mabuy, R. L. Scogix & B, L. Turner. 1975. The bases of angiosperm phylog-

eny: chemotaxonomy. Ann. Missomi Bot. Card. 62: 76.5-800.

Feu(;usox, I. K. 1966. Notes on the nomenclature of Comus. T. Arnold Arbor 44- 100-
105.

Frohxe, D., O. Moritz & U. Jensen. 1961. Uber die Behandlung der unerwarteten Reak-
tionen in der serologisehen Verwandtschaftsforschung. Flora 150: 332-352.

Cell, P. G. H., S. T. C. Wiu(;Hi- & J. C. IIawkes. 1956. Lnmunological methods in plant
taxonomy. Nature 177: 573.

GiLc;, E. & P. H. ScHiJiiHOEF. 1926. Die Serodiagnostik in der botaniscli(^n Verwandtschafts-
forschung. Bot. Jahrb. Syst. 60: 439-150.

IIahms, H. 1898. Cornaccae. In A, Engler & K. Prantl (editors). Die natiirlichen Pflanzen-

familien. Ill, 8: 250-270. Wilhelm Kngelmann, Leipzig.
IIe(;nauer, R. 1965. Chemismus und systematische Stelhmg der Cornaccae. Pp. 235-246,

in D. Groger (editor), Beitriige znr Biocliemie nnd Physiologic von Xaturstoffen. C.

Fischer, Jena.

IIewitson, W. 1963. Comparative morphology of the Osmnndaceae. Ann. Missouri Bot
Card. 49: 57-93.

llonx, M. E. & W. G. Meinschein. 1976. Seed oil fatty acids: evolutionary significance in
the Nyssaceae and Cornaceae. Biochem. Syst. Ecol. 4; 193-199.

Jensen, S. R., A. Kjaer & B. J. Nielsen. 1975. Tlie genus Cormts: nonflavonoid
glucosides as taxonomie markers. Biochem. Syst. Ecol. 3: 75-78.

Jensen, U. 1974a. The interpretation of comparatixe serological results. Pp. 217-226, in
G. Bendz & J. Santesson (editors), Chemistry in Botanical Classification. Proc. 25th
Nobel Symposium, 1973, Sweden. Academic Press, New York.

. 1974b. Close relationships between Ranunculales and Rutales? Systematic con-
siderations in the liglit of new results of comparative serological research. Bull. Serol.
Mus. 50: 4-7.

Johnson, M. A. 1951. The precipitin reaction — a tool for determining relationship be-
tween organisms. Ne\\' Jersey Sci. Teacher 1 : 11-14.

Kloz, J., E. TuRKovA & E. Klozova. 1960. Serological investigation of tlie taxonomie
specificity of proteins in various organs in some taxa of the family Viciaceae. Biol Pi
2: 126-137.

Kraus, R. 1897. iJber specifische Reaktionen in keimfreien Filtraten aus Cholera, Tijpiius
und pestbouillon Culturen erzeugt durch homologes Serum. Wiener Klin. Wocheiischr
10: 736-738.

KuHiTSKi, K. 1963. Znr Kenntnis des )mih)cularen Cornaceen-Gynozeums (Cornaceen

Studien I). Ber. Deutsch. Bot. Ges. 77(S(mdernnmmer ) : 33-39.
Lake, G. C, T. B. Osborne & H. G. Wells. 1914. Ihe imnumological relationship of

hordein of barley and gliadin of wheat as shown by the complement fixation, passi\e

anaphylaxis and precipitin reactions. J. Infect. Dis. 14: 364-376.
Lee, D. W. & D. E. FAUiBHOTHEns. 1972. Taxonomie placement of the Typhales within

the Monocotyledons: preliminary serological investigation. Taxon 21: 39-44.

Leone, C. A. 1952. Effect of nudtiple injections of antigen upon the specificity of anti-
sera. J. Immunol. 69: 285-295.

-. 1968. The immunotaxonomic literature. Bull, Serol. Mus. 39: 1-23.

Magnus, W. 1908. Weitere Ergebnisse der Serum Diagnostik fiir die theoretische und
angewandte Botanik. Ber. Deutsch. Bot. Ges. 26a: 532-539.

Melchioh, a. (editor). 1964. A. Engler\s Syllabus der Pnanz(nifann-lien. Ed. 12. \^)I.
2. Gebriider Borntrager, Berlin.

Mez, C. 1922. Anleitung zu serodiagnostichcn Untcrsuchungen fiir Botaniker. Bot Arch
1: 177-200.

& K. GoiiLKE. 1914. rh>sio]ogischs>stematischt^ Untcrsuchungen uber die Ver-

waTidtscbaften der Angiospermen. Bcitr. Biol. Pflanzt-n 12: 155-180.

160

ANNALS OF TllE MISSOURI BOTANICAL GARDEN [Vol. 64

& II. ZiEfiENsi'ECK. 1926. Der Konigsbcrger serologischc Stammbaum. Bot. Arch.

13: 483-485. , , ^, .

MiLLEH, C. N. 1967. Evolution of the fern genus Osimmth. Contr. Mus. Paleonlol. Lni\.

Mid.igan21: 139-203.
MoiuTZ, (). 1928. Zur Kritik der Phytoserologie. Biol. Zentralbl. 48: 431-443.
L. 1904. Some special features of serobotanical work. Pp. 275-290, in C. A. Leone

(editor), Taxonomic Biochemistry and Serology. Ronald Press, New York.
& II. L. RoHN. 1956. Untersuchungen zum Problem der serologischen Fernreak-

tionen. Planta 47: 16— 16. i i -r-

NuTTALL, G. M. F. 1901. The new biological test for blood in relation to zoological classifi-
cation. Proc. Roy. Soc. London 69: 150-153.

1901. Blood Immunity and Blood Relationships. Cambridge Univ. Press, England.

Pkrdue, R. E., R. L. Smith, M. E. Wall, J. L. Hahtwell & B. J. AiuiOTT. 1970. Caiupto-
ihcca acuminata Decaisne (Xyssaceae) source of Camptothecin, an antileukemic alka-
loid. Techn. Bull. U.S.D.A. 1415: 1-26.

Fairuuothers. 1971. North American Osmnnda species: a sero-

logic and disc electrophoretic analysis of spore proteins. Amer. Midi. Naturalist 85:

437-457. , „ , 1 T « . c

PiiiLiPSOx, W. R. 1967. Griselinia Foist, fil.— anomaly or link. New Zealand J. liot. 5:

134-1*65. , ,^ .,

Smith, S. N. 1958. Cowkia A. Cunn.— an addition to the Australian genera of Saxitragaceae.

Proc. Roy. Soc. Queensland 69: 53-55. _

SoHMA, K. 1963. Pollen morphology of the Nyssaceae I. Nyssa and Camptolhcca. Sci.
nlr^ TAl,ol-„ Tnip. Univ., Ser. 4, Biol. 29: 389-392.

F Thompson. 1975. DNA hvbridizaticm and evolutionary relationships

W

888-890

Suc;iYAMA, T., C. MATsuMtrro & T. Akazawa. 1969. Structure and function of diloroplast
proteins VII. Ribulo.se-l,5-diphosphate carboxylase of ChlorcUa ellipsoidcu. Arch.

Biochem. Biophys. 129: 597-602. , , , ,

Takhtajan, a. 1969. Flowering Plants: Origin and Dispersal. Oliver & Boyd, Edinburgli.
Thohne, R. F. 1968. Synopsis of a putatively phylogenetic classification of the flowering

plants. Aliso 6: 57-66. e inm . x, r-
. 1976. A phylogenetic classification of the augiospermae. Pp. 35-106, m M. C

llecht, \V. C. Steere & B. Wallace (editors). Evolutionary Biology. Vol. 9. Plenum Press,

New York.
Titman, P. W. 1949. Studies in the wood anatomy of the family Nyssaceae. J. Elisha

Mitchell Sci. Soc. 65: 245-261.
UuANO, K. 1955. Preliminary investigations on diagnosis of strain-relationship m corn
plant !)>■ precipitive tests I. Precipitive reaction between inbred lines of corn. Proc. Crop

Sci. Soc. Japan 24: 4-7,
Vauchan, J. C. & A. Waite. 1965. Taxonomic imestigation of several Brassica speci
using serology and the separation of proteins by electrophoresis on acrylamide gels.

Nature 208: 704-705.

, A. J. MACLEOD & B. M. G. Jones (editors). 1976. The Biology and Chemistry of

the Cruciferae. Academic Press, London.

Wacerin, W. 1910. Nyssaceae. In A. Engler (editor). Das Pflazeureich IV. 220a,

WUK

Wilhelm Engelmann, Leipzig.

r). 1974. Biochemical and Immunological Taxonomy of Animals.

Academic Press, London.

ELECTROPHORETIC EVIDENCE AND PLANT

SYSTEMATICS

L. D. Gottlieb^

Abstract

The study of plienotypcs and their \ ariation often provides evidence for phylogenctic in-
ferences in plant systeniatics. Therefore, it is critical that the phenotypes analyzed reflect as
directly as possible the underlying genotypes. The equation between phenotype and genotype
is simpler and better understood for evidence obtained by electrophoresis of plant enzymes
than for most morpliological characters. This article discusses the advantages and limitations
of electrophoretic evidence to test hypotheses in plant systeniatics and evolution. It also sum-
marizes the results of a large number of studies which have utilized this evidence. Three gen-
eral observations from these studies are: (1). Conspecific plant populations are extremely
similar genetically as documented by their very high mean genetic identities, 0.95 ± 0.02.
This result suggests that one or a few populations often c(mstitute an adequate sample of a
species. (2). Congeneric plant species ha\e strikingly reduced mean genetic identities, O.fiT
± 0.07. However, certain pairs of annual plant species have genetic identities similar to those
of conspecific populations. In these cases, die species ha\c been shown to be related as
progenitor and derivative with the derivative being of recent origin. (3). The auKumt of
genetic variability within plant populations appears closely correlated with their breeding
system, witli outcrossing populations substantially more variable than inbreeding ones. The
article also describes a numl)er of actual and potential applications of electrophoresis in i^lant
systeniatics.

Evidence obtained by electrophoresis of enzymes has not been widely utilized
by plant systeniatists although it has dominated the research of many of their
zoological counterparts and population geneticists ( Manwell & Baker, 1970;
Lewontin, 1974; Nei, 1975; Ayala, 1976). This has meant that the strengths and
weaknesses of such evidence for solving systematic and evolutionary questions
in plant biology have not been sufficiently discussed. The present article is de-
signed to facilitate an efficient evaluation, and emphasizes the unique charac-
teristics of electrophoretic evidence, the requirements for its analysis, and actual
and potential applications in plant systematics and evolution.

Electiwpiiohktic Evidence: Advantages and Limitations

The systematist analyzes phenotypes and their variation and often uses this
evidence for phylogenctic inference. Such inferences recjuire that observed
phenotypes have a specifiable relationship to unobserved genotypes. The equa-
tion between phenotype and genotype is simpler and better understood for
electrophoretic evidence than it is for evidence obtained from morphological
characters or chromatographic comparisons of secondary metabolites. This fol-
lows from the colinearity of amino acid sequence and nucleotide sequence as
well as the specificity of enzyme catalysis. It also reflects the fact that electro-
phoretic evidence is used to answer a very different kind of question than has
usually been posed by systematists.

Morphological analysis answers a question such as: Are flower petals with

^ Department of Genetics, University of California, Davis, California 95616.
Ann. Missouri Box. Gaiu>. 64: 161-180.

■^Q2 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

lobed limbs present in taxon A and taxon B? Chromatographic

analysis asks, for example: Is apigenin present in both taxon A and taxon B?
In contrast, electrophoresis answers a qnestion of a different kind: Does gluta-
mate dehydrogenase have the same electrophoretic mobility (ies) in taxon A
and taxon B? Such a question probes the physical properties of particular en-
zymes or other proteins on the hypothesis that these properties reveal, to a large
degree, the record of accumulated mutations that have taken place in the gene
specifying the enzyme. And that when a number of enzymes are cK)nsidered
simultaneously, this record can be evaluated with more precision and objectivity
than highly complex morphological features.

The primary observed evidence in studies of electrophoretic variation in
natural populations is bands of color in a slab of starch or acrylamide gel. These

rms

de

field. The enzyme variants are separated because they have different electrostatic

through

positive

CO

ration, the enzymes are identified by a staining reaction based on their catalytic
activities. The combination of electrophoresis and staining specificity makes it

possible to distinguisli particular enzymes among hundreds that may be present

in a crude tissue extract.

The different molecular forms of an enzyme that catalyze the same reaction
are called isozymes if their polypeptide constituents are coded by more than one
gene locus (e.g., lactate dehydrogenase, human ADH). They are called allo-
zymes if their polypeptides are specified by different alleles at a single gene
locus; the majority of enzymes routinely studied in natural populations have

different allozymic forms.

Allozymes are the biochemical consequence of the substitution, deletion, or
addition of amino acids in the polypeptides which comprise the enzyme, and they
can be distinguished if these changes affect their electi^ophoretic migration. Since
the amino acid sequence of a polypeptide is colinear to the nucleotide secjucnce
of its coding structural gene locus, allozymes result from gene mutation. Thus,
an analysis of protein structure using electrophoresis is, to a first approxima-
tion, an analysis of a gene. It is precisely this simple relationship between the
bands of color on the gel and the nucleotide sequence of genes that makes pro-
tein electrophoresis a powerful analytic tool for systematics.

A major advantage of electrophoretic evidence is that colinearity assures that
systematic comparisons can be made between products of genes which are
homologous (have a common origin), thus, avoiding problems of convergence
and functional correlation often prevalent with morphological characters. An-
other important advantage is that electrophoretic evidence is precise and di-
rectly quantifiable in terms of the number and kinds of enzymes studied, per-
mitting the amount of genetic information utilized to be stated exactly. This
is seldom possible with morphological or other characters. A third significant
advantage is that comparisons are made with enzymes that are generally always

19"'^] GOTTLIEB ELECTROPHORESIS AND PLANT SYSTEMATICS

163

present (with the exception of those selectixely turned on or off during develop-
ment) and little influenced by environmental factors. This avoids the frequent
situation that occurs with both morphological and chromatographic data in which
the absence of a character in one taxon is interpreted as an indication of a less
close phylogenetic relationship between it and other taxa that display the charac-
ter even though independent evidence is lacking regarding the cause of the charac-
ter's absence. Another advantage is that problems of a priori character weightinj

r

do not occur with electrophoretic evidence because all enzymes examined are
accorded equal value in similarity matrices or other methods of evaluating di-
vergence.

The theoretical advantages of electrophoretic evidence, to be sure, are offset
by certain shortcomings but, fortunately, these are reasonably well defined. The
first problem is that a small number of enzymes is sampled and these may not
represent enzymes in general since they are most often involved in some aspect
of glycolysis, intermediary metabolism, or in the catalysis of certain general
types of bonds (esterases, phosphatases, peptidases). Lack of representativeness
is probably less serious for systematics, which has not confined itself to charac-
ters thought to be representative, than it is for genetic studies which attempt to
estimate the total amount and kind of genetic variation in different kinds of or-
ganisms. In any event, the enzymes that are examined comprise a sufficient!)'
large category to give meaningful information about many kinds of evolutionar)
changes.

A problem which is more critical for systematics is that, even for those en-
zymes examined, the redundancy of the genetic code means that only about 30

r

^v

of the substitutions of nucleotides are expected to result in the substitution of
amino acids that cause changes in electrophoretic mobility (Shaw, 1970). In
addition, allozymes that have identical mobilities do not necessarily have idf^iti-
cal amino acid secjuences. In fact, recent studies utilizing amino acid setjuenc-
ing (Boyer et ah, 1972), heat denaturation (Bernstein et ah, 1973; Singh et ah,
1974), and variation in gel pore size (Johnson, 1976) suggest that a single mo-
bility class on a gel may sometimes contain more tl^ni one enzyme. This requires
that more weight be given to evidence of electrophoretic difference than to evi-
dence of similarity. Additional biochemical tests, however, are available to de-
termine wliether allozymes with the same molality have different amino acid
sequences. All in all, electrophoretic evidence should be regarded as providing
an underestimate of the actual amoimt of genetic difference between taxa.

Electrophoretic evidence does not include any information on the number
of amino acid differences, or mutational steps, that cause differences in enzyme
mobilities. A difference in mobility can reflect a single nucleotide substitution
or numerous changes in nucleotide sequence. Thus, electroplioresis can demon-
strate tliat two taxa have different allozymes of phosphoglucoisomerase, but it
does not provide information about the amount of difference. This is likely to
be greater witli increasing phylogenetic distance.

Thus, the equation between phenot) pe and genotype remains acceptable even
when the limitations of electrophoretic e\'idence are considered Ijecause they can
be specified and the direction of bias is generally Known.

164

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Electropiioretic Evldence: A Framework for Analysis

When tissue extracts are subjected to electrophoresis in a starch gel, the pat-
tern of enzyme bands (number, spacing, and intensity) is an expression of the
particular enzyme system assayed and its mode of inheritance. For some en-
zymes (usually esterases, phosphatases, and peroxidases), single individuals dis-
play complex patterns with as many as 10 to 15 bands because they possess
numerous gene loci that code different molecular forms. In contrast, other en-
zymes are specified by a single structural gene and individuals might display
only a single band following electrophoresis. The number of polypeptide sub-
units of each enzyme and the allelic state of the coding gene (homozygous or
heterozygous) also determine the number of enzyme bands displayed. Thus,
for an enzyme composed of a single polypeptide, an individual heterozygous
at the coding gene displays two allozyme bands, but if the enzyme is dimeric
(composed of two polypeptides), three allozymes are displayed, and if it is
tetrameric, five are displayed (Fig. 1). In other cases, the polypeptide constit-
uents of an enzyme are coded by different nonallelic genes, for example, alco-
hol dehydrogenase in maize (Freeling & Schwartz, 1973) and in sunflower (Tor-
res, 1976), producing still additional variants.

The number of enzyme bands can be reduced if enzymes specified by dif-
ferent genes overlap on the gel because they have similar mobilities, or if an
individual is homozygous for a "null" allele ( an absence of activity which genetic
analysis demonstrates to be allelic to genes that specify active forms of the en-
zyme). Artifacts that might result from procedures of extraction or electrophore-
sis can also change band number. In addition to these biochemical factors, the
pattern of enzyme bands displayed by different individuals in a population is a
function of the amount of genetic variation for the enzyme system.

The presence of so many factors which influence the appearance of the electro-
piioretic phenotype means that the systematist must reject the temptation to com-
pare electropiioretic data from different taxa by direct inspection, i.e., simply
counting the number of bands with similar and dissimilar mobilities. Not only
would this approach be biochemically faulty, but it nullifies several of the im-
portant advantages of electrophoresis, particularly the inference of enzyme
homology, and the precise and quantifiable form of the evidence. The frequent
complexity of the electrophoretic phenotype means that, at least for the complex
systems that have a number of electrophoretically separable enzyme variants,

a genetic analysis is necessary.

Such analysis demonstrates which variant forms of an enzyme system are
specified by allelic genes and which by nonallehc genes; i.e., it distinguishes
allozymes and isozymes. In addition, in many cases, it rules out the possibility of
biochemical artifacts. Genetic analysis also leads directly to a quantitative speci-
fication of the electrophoretic data that usually is ordered as follows: the num-
ber of structural genes specifying tlie enzymes examined; the proportion of genes
that show variation or, as the geneticist says, are polymorphic in that they have
more than a single allele; the number of alleles per gene in the population; and
the mean proportion of genes which is heterozygous per individual.

The protocol of formal genetic analysis can often be simplified with electro-

1977]

GOTTLIEB— ELEC:TR()PnORESIS AND PLANT SYSTEMATICS

165

FiGUUE 1. Tliese diagrams present three examples of eleetrophoretic patterns in parents
and their hybrids to illustrate that differences between individuals in the mnnber of enzyme
bands often do not equal the number of their genetic differences. Case 1 presents a cross be-
tween two individuals, homozygous for different alleles at a gene coding an enzyme. The num-
ber of bands per individual in their Fi hybrids depends on the subunit structure of the enzyme.
Note that for a dimeric enzyme, a heterozygous individual differs from each of its par-
ents by two bands, but one allele; for a tetrameric enzyme, it differs from them by four barids,
but one allele. Case 2 illustrates the same point but uses a cross between individuals homozy-
gous for the same allele at one gene and homozygous for different alleles at a second gene.
The Fi phenotype depends on the subunit structure of the enzyme. In this example, poly-
peptides specified by the two genes do not have affinity for one another. Case 3 i^resents a
cross between two individuals whose enzyme phenotype is also determined by two gene loci.
In this ease, the enzymes are considered to be dimeric, and polypeptides specified by both
different alleles and different genes associate to form "hybrid" or heteromeric enzymes. The
cross is a test cross between a double hetcrozygote and a double homozygote. If the genes
assort independently, the four progeny classes are produced in ecjual numbers; if they are
linked, the parental phenotypes are more frequent than the recombinant ones. Note that the
double hetcrozygote differs from the double homozygote by six bands but only one allele at
each gene. The polypeptide structure of each enzyme band is given on the left and the geno-
type of each individual is given below.

CASE 1

CASE 2

Number polypeptides

1

X

>

Parents

m enzymes

2

4

Fi phenotypes

X

>

Parents

Fi phenotypes

CASE 3

la2b +

lala
lalb
Ibib
Ia2a
lb2a
lb2b
2a2a
2a2b
2b2b

X

>

la 2a
lb 2b

lb 2r

lb 2a
Parents

la 2a
lb 2b

lb 2a
lb 2a

lb 2a
lb 2b

Progeny phenotypes

la 2a
lb 2a

phoretic data because enzyme bands, with few exceptions, show codontinant in-
heritance and segregate as single MendeHan factors (see reviews by Scandalios,
1969, 1974; Jacobs, 1975). Tliis makes it possible to utilize progeny tests to re-
place formal crosses b(^tween individuals with different phenotypes, examine Fi
progeny phenotypes, and analyze phenotypic segregation patterns iu the F;-

IQQ ANNALS OK THE MISSOURI BOTANICAL GARDEN [Vol. 64

generation. In the progeny tests, individuals are grown from open-pollinated
seeds collected on single plants in nature. Since these individuals all have one
parent in common, they necessarily have in common its alleles. For many enzyme
systems, study of the segregation pattern of the different variants in such prog-
enies can indicate which of them are specified by allelic genes and which by
different gene loci (Brown et al., 1975). However, for very complex systems in
which several polymorphic genes and overlaps in the migration of different en-
zymes are involved, formal analysis is still required. Figure 1 presents examples
of such analysis.

lieforc the introduction of the clectrophoretic technique, the study of genetic
variation in natural populations was unsatisfactory because it depended on the
identification and enumeration of rare recessive mutants that, when homozygous,
yielded visible morphological changes. The genetic basis of many of these charac-
ters is sunple and clearly demonstrable, but they constitute only a very small
proportion of the genetic variation in populations. The vast majority of pheno-
typic characters are apparently controlled by many genes each of which may have
different individual effects. These so-called quantitative characters are also often
strongly influenced by environmental variation. The result is that the contribution
of individual genes of this type cannot be ascertained, and the extent to which
they vary one from another in different individuals is undetectable. Thus, one
studied either those rare characters controlled by one or two genes with major
effects or the much more common characters controlled by many genes that are
neither individually identifiable nor distinguishable from environmental in-
fluences. In consequence, the traditional methods of studying genetic variation
were stymied by the impossibility of equating phenotypes with genotypes.

The clectrophoretic procedure avoids most of these problems. In addition,
it identifies genes which do not vary on the basis that their enzyme products do
not vary in their clectrophoretic mobility (within the limits mentioned in the
previous section), making it possible to determine the proportion of genes that
show variation. This advantage has been considered the "cornerstone" of the
method (Ilubby & Lewontin, 1966) because previously it was not possible to
equate lack of variation with monomoiphism at particular genes.

The determination of how many individuals and populations to sample before
a confident statement can be made regarding the amount of genetic divergence
between taxa is another important consideration in the use of clectrophoretic evi-
dence. A definition of the meaning of divergence greatly simplifies this prob-
lem. Thus, maximal divergence between two taxa at a gene locus means that they
have no alleles in common. Minimal divergence at a gene locus means that the
two taxa have similar complements of alleles in similar frequencies. Although
the two extremes are CK)nnected by a wide variety of intermediate situations,
the amount of divergence at a set of genes can often be fitted into a general pic-
ture. Thus, if a large number of genes is studied, about 30-50% of them are
likely to be monomorphic, another 30-40% will be moderately polymorphic (two
or three alleles), and the remaining genes will be highly polymoi'phic.

The distribution of genes in monomorphic and polymorphic categories per-
mits the systematist to decide, at the outset of a study, the amount of sampling

1977] GOTTLIEB— ELEC:TIU)PH0RES1S AND PLANT SYSTEMATICS J^gy

necessary to test a given hypothesis. For example, at a polymorpliic gene locus,
an allele can be considered connnon (moderate to high frequency) and wide-
spread, rare (less than 0.05) and widespread, common and local (one or two
populations), or rare and local. For certain systematic purposes, one might de-
cide that the first category is most relevant (it is the easiest to sample since al-
leles here have the highest probability of being included in a sample regardless
of strategy). In logistic terms, this means tliat relatively little effort need be ex-
pended to find one more allele which is likely to have low frequency, be local
in distribution, or both.

The number of individuals to sample per population is best viewed from the
standpoint of how many plants to examine in order to have a 95?f certainty of
observing all the alleles at a locus wliich have frequencies greater than 0.05. This
problem has been considered by Marshall & Brown (1975) who show that, even
in the inilikely case of 20 alleles with frequencies of 0.05 each, a random sample
of 120 gametes (60 individuals) will include, with 95% certainty, one copy of
each allele. In sum, decisions related to sampling can be neatly bracketed be-
cause electrophoretic evidence consists of discrete and precise units of informa-
tion.

A number of coefficients have been developed to summarize allele frequency
data into a single figure that might be used to assess the degree of genetic di-
vergence of taxa (Cavalli-Sforza & Edwards, 1967; Hedrick, 1971; Nei, 1972;
Rogers, 1972); however, all of them appear to provide similar estimates (Avisc,
1974). The data can also be used to construct dendrograms that cluster taxa ac-
cording to their similarities (references in Avise, 1974). Another approach, de-
scribed in the followivig section, that might be particularly useful for analysis of
conspecific populations, makes use of the presence or absence of alleles rather
than their frequencies (Gottlieb, 1975).

Elkctropiiohetic Evu)ence : Applications

Electrophoretic analysis of enzyme variation provides efficient, quantitative
estimates of the amount of genie variation within natural populations and the ex-
tent of genie divergence among populations. A very large number of electro-
phoretic stiidies have been made on animal species. The results appear remark-
ably consistent: (1) Single populations of both vertebrates and invertebrates
contain substantial genetic variability, and perhaps as much as 907c of the total
genetic information of their species (review in Powell, 1975; Selander, 1976);
(2) Conspecific populations have a very high degree of genie identity (Nei,
1972), often with a mean above 0.9(), on a scale of to 1. Their higli identity
reflects the fact that the same allele is usually fixed at monomoiphic genes (as
much as 80 to 907^ of the genes in vertebrates, for example), and, at the poly-
morphic genes, only the frequency of alleles differs (Avise, 1974); (3) Closely
related species are considerably more differentiated than conspecific popula-
tions, with a mean genetic identity around 0.50 to 0.60, which suggests that dif-
ferent species of animals are almost completely distinct in allelic composition
at about one-c^uarter to one-half of their genes (Ayala, 1975). Many of these
differences may have evolved since their origins as species because much higher

IQg ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

values of genetic identity are observed for pairs of species which apparently
originated in the Pleistocene (Nevo ct al., 1974; Avise et ah, 1975).

In contrast to the wealtli of data for animal species, the number of electro-
phoretic studies of plant species is extremely small (fewer than a dozen groups
of congeneric species have been examined for electrophoretic variati(m in a large
number of enzymes ) and, consequently, generalizations must still be considered
tentative. The paucity of studies with plants is unfortunate because, in many
ways, plants are better material than animals for electrophoretic investigations.
Thus, they are often easy to grow in high numbers; they need not be killed to
obtain a tissue sample so that individuals can be used for additional analysis;
progeny-testing to establish the genetic conti'ol of enzyme variants is straight-
forward; natural populations are often spatially and ecologically delimited, per-
mitting coordinated studies of ecological adaptations and amplitudes; phylogenies
are often known unambiguously, facilitating analysis of the consequences of
speciation; breeding systems are highly variable so that genetic consequences of
different amounts of inbreeding can be studied directly and correlated with
demographic inputs, etc.

The available electrophoretic studies with plants that are relevant to ques-
tions in systematics and evolution can be grouped into four major categories;
(1) genetic divergence among conspecific populations; (2) genetic divergence
among congeneric species, a subject which has also provided evidence of the
genetic and biochemical consequences of speciation; (3) enzyme expression in
diploid progenitors and polyploid derivatives; and (4) a heterogeneous group of
special-purpose studies (not reviewed here because of space Hmitations) deal-
ing with themes such as analysis of gene flow across species barriers (interspecific
hybridization) (Chu & Oka, 1970; Levin, 1975), consequences of unusual chromo-
somal pairing mechanisms in Oenothera (Levy & Levin, 1975), demographic
analysis (Schaal, 1975), the effect of breeding systems on the amount and ex-
pression of genetic variability (Allard, 1975; Allard & Kahler, 1971), and genetic
diversity and edaphic specialization (Babbel & Selander, 1974).

Electhophohetic Evidence: Conspecific Populations

Electrophoretic variation in enzymes has been examined in natural popula-
tions (at least two) of about 28 plant species (Table 1). However, in many re-
spects, the data is very uneven. For example, the number of enzyme systems
examined and the number of genes that code them in the different species vary

over a five-fold range. In addition, the choice of enzymes is diverse so that in
some studies the proportion of genes specifying enzymes that are frequently
highly polymorphic (esterases, phosphatases, peroxidases) is high, whereas in
other studies many additional enzymes are included that are involved in basic
metabolism (such as phosphoglucoisomerase, phosphoglucomutase, glutamate de-
hydrogenase, malate dehydrogenase, malic enzyme, glutamate oxaloacetate trans-
aminase). Further, the number of populations sampled per species varies widely.
The species themselves are nearly all annual but, in other respects, they are
heterogeneous including diploids and polyploids, obligate outcrossers and pre-

1977]

GOTTLIEB— ELECTROPHORESIS AND PLANT SYSTEMATICS

169

I

5 bC

o

o

ca

O CO

CM .a

a;

a

o

a;i

^

t/5

u
OJ

«

V5

0^

O

i H

>

cu cd

bC

03 ^

e

W5

1^

o

o

s

a

c

vSj

o

aD

.«

o

1- -r; -"

N

O

H

c

c

cd

03

a

a;

a o

< o ^

t^ ^ ^

4j

c

u

PC

o >.

02

S

1^

OJ

2^

O

ce-
ll -

^1

N P 3^

;2 ^

CIh

;2

OI fM

05 05

05

'S "H -? ^ c^ .^ :S 5

■^ "^ Tf'
t- I- I-

a^ (y> a^

rv r^

^

Oj Oj !> 0;

be bjj;i3 ;z3 ;r: ;n

bc b£' tj t; j-; t-:
^^ c o o c

UUOOOO

^ r\ r\

^ ^ ^

V QJ <U
^ ^^ "^

CCS
c^ c^ c3

0; o; 0^

c/^ c/^ c/^

0^ ^ o3

0) CJ Oj

QD CO
»- I-

Ot) 05

CO

05 in ID lo
I- t- I-

otJ Oi o:)

CO

52 o^

CO
O^ Oi

CO

G^ 05

c/5

lO

I-

G>

c/i v:

u.

cd cd

t-H

u

c c c

> >

J J J CD

> in

I-

05

cd

OiCCcCOCO

3^ 0^ qj

^ ^ ^

qj

Cd Cd 73
CQ (XI PQ

o

s

c5

^
o

^

>

3^

C P

Oj

>

;-H

^ ^

> ^

Ci; J J hJ J hJ

o j:^ ^ ^ ^ 0^ :<3

(L> O' a>

^ -^ "^ a: flj oj tu flj

'■ri -^ ^ iTi ■J'.

C C C C C

Cd

OOOccecXi^^x

lO 05

c o

05

05 O? I- -t I- CO

CJ5 05 05 05 05 05

* • • * • *

o o o o o o

GO in ni I- in i-

^ O^ O^ ':j: O^ !^

^_ ^L ^L ^_ ^_ ^_

00 05 05 r^ OC

05 05 05 05 05

^k ^- ^L ^^ ^^

O O O O O O

■ # * 4 4

o o o o o

c^ m t^

t^ in

rM rM

CO

CO

I

in CO

TT Tf

05

o o o o o

+ +
+ +

<

V vv

occoo;oofMio"t-<MinfMOc^oin3coinco

o CO CO p ^ o ^ o X ^. CO in o CO -^^ in CO CO rM CO

c4 CO CO oi CO CO oi c-i — i CO csi co (>i -h ^* —" ^ (>i oi ni

00 CD o in o
■^ od oi oi CM*

CM O
CM

CM CM rf CO 00

(X: CO CO CO CM t

+ « t * «

o o o o o

I

X OC t- O Tt< 00 o

X 00 m in ^ CM CO

in X in 05

CM CO CM CM

t- t- -^

in o o

o o o

CM

CM

in in X X

X X

t- I

CO CO l-- t^ X O O CO CO t^ l^ ^ Tt^ '^

^TTTfXOOOcOtO
XX^^^^CMCMCM^^CM

CO

^ rr in

CM CM CM m CM CM

xx-^cococococo

O5t^c0cM'^c0CM^cMco

* » # # * *

^ X X CD ^ 05 05
in i-H O (M CM

CM

in in CO CO CO CO in

(ff

ft

4

5 S "3

w:

a r '5

^

"S ^

5 i

u. :3 *:=;

Oi 1

a tli r^

rv

1^ ^

.^^

^^ 1-^ P^

/j

§ «^

<; -^o

!3

S

o

in

n

c^

c
Q

2

5 S

^^ .^ Cr

2 2

c 2

=3 *-

^

^

^

X
Qd

I ^^^^

Co

-4 C9

CO

3

c

3

Q 5

a

3 a

r ^

^ £ ■?:

n

u a

u u i: :r: ^kjkj'kj

'^

a

^

i 2 s

pf

.15

"^ ^,

o

a. -^
o 2

^ rT* ^

^ Q :^
- o o -.-^

2

^

CO

^ r

a

^

c?5

fjj

c>^ o6 ^ h< h^ ^ E^

c

as

X 5

<1>

tn

O

A
o

'Si
O

c

/^

« o ^

,E *-'^

o c o

n ^ ^

* *

*

YJO ANXALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

dominant selfers, unci recently evolved and ancient taxa. However, in spite of
these disparities, two results appear well established.

First, the amount of genetic variability within plant populations is closely
correlated with their breeding systems (Table 1). Thus, the mean number of
alleles per polymorphic gene averages 1.88 ± 0.12 for highly self-pollinating
species and 2.86 ± 0.24 for outcrossing ones. The mean proportion of genes that
is heterozygous per individual follows suit: 0.032 ± 0.013 for selfers and 0.133
± 0.026 for outcrossers. These results indicate that breeding system influences
not only the degree of genetic homozygosity but also the total amoui\t of vari-
ability that can be maintained in plant populations.

Second, conspecific plant populations are extremely similar genetically as
demonstrated by their very high mean genetic identity I = 0.95 ± 0.02, averaged
over all species (Table 1). This is an important result for systematics because
it suggests that clectrophoretic evidence from one or a few populations very
often constitutes an adequate sample of an entire species.

The very high degree of genetic similarity among conspecific populations
leads to the suggestion that should populations be discovered which have novel
alleles or distinct allele frequencies at more than a few gene loci, such popula-
tions are very likely to constitute distinct taxa and should be further examined
with this in mind. Such evidence has already been used to identify a subspecies
of Dwsophila willistoni (Ayala, 1973) and a new species of sea cucumber (Man-
well & Baker, 1963), It is not unlikely that clectrophoretic evidence will also be
similarly used to identify hitherto unrecognized plant species.

An approach to the comparison of conspecific populations that considers the
representativeness of single populations rather than their identity to one another
has also been proposed (Gottlieb, 1975). Designated the Complement Index,
it comp;ucs the number of n()nuni(|ue and nonubiquitous alleles (present in
more than one but not all populations) in each population with the total number
of such alleles identified in all the populations examined. Since the presence
of alleles increases the biochemical repertoire of a population, the number of
alleles constitutes a very good and easily obtained estimator of the relative po-
tential of different populations for adaptive evolutionary change. The Comple-
ment Index would be particularly useful for taxa in which populations contain
large numbers of different low frequency alleles, a situation that may be com-
mon in outcrossing plants. Thus, 11 populations of Stephanomeria exigua subsp.
carotifera all possessed the same gene at 6 monomorphic loci and 13 high fre-
quency alleles at 8 polymorphic ones, but different numbers of 25 other low
frequency alleles (Gottlieb, 1975), Calculation of the Complement Index showed
that the populations actually represented subsp. carotifera to very different de-
grees even though they had a mean genetic identity, / = 0.98 (Gottlieb, 1975).
The average population had only about half of all the genes identified in the
subspecies as a whole; nevertheless, one of them possessed every one of the non-
unique alleles. This population contained more of the genetic resources of subsp.
carotifera than any other and is the most likely to persist through environmental
fluctuations. In addition to identifying such populations (which, with cultivated
plants have obvious hnportance for germ plasm conservation), the Complement

1977] GOTTLIEB ELECTHOPllORESIS AND PLANT SYSTEMATICS yjl

Index could be used to compare the representativeness of p(;palations of dif-

ferent taxa.

Electhophohetic Evu)ence: Conc;enehic Species

The large amount of electrophoretic variation in natural populatious of plants
initially led systematists to presume that extensive surveys within and ])ctween
populations were required in order to use electrophoretic evidence for meaning-
ful systematic comparisons between species (Turner, 1969). This point of view
apparently took hold because genetic studies had only rarely been carried out
and therefore it was difficult, if not impossible, to make sense of the complex
banding patterns that were observed (many of the early studies unwittingly
utilized esterases and peroxidases which are the most difficult systems to inter-
pret). The absence of genetic data and the small number of populations that
had been sampled combined to give the impression that the polymoiphisms in-
herent in electrophoretic evidence lessened its value for systematics.

However, now it is realized that once polymorphisms are defined in genetic
terms so that "bands" can be equated with alleles and different gene loci, then
their presence actually incrc^ises the power of electrophoretic evidence for sys-
tematic studies since they reveal to a larn;e de<2;ree the accunnilated record of nu-

merous mutations that have become established in the coding genes. In addition,
extensive sampling of conspecific populations is often not necessary for species
comparisons because many of the alleles, particularly those with frerjuencies
above 0.20, as well as the genes at monomorphic loci, are now known to be present
in most, if not all, populations of a species. However, although the number of
populations sampled can be reduced, it remains important to increase the num-
ber of enzymes sampled. This w^ould tend to lessen the effect of biases that might
result from selecting only enzymes likely to be polymoiphic or those hmited to
any particular biochemical category.

About a dozen studies have been made Uiat provide evidence of the extent
of genetic divergence between congeneric species (Table 2). These studies show
that most pairs of species have strikingly reduced genetic identities; 7 = 0.67
± 0.07, averaged over all pairs of species examined. A number of species pairs,
however, have very high genetic identities, within the range of those characteris-
tic of conspecific populations. For three of these cases, Steplianomcria exi^ua
subsp. coronaria and "Malheurensis" (Gottlieb, 1973b, 1976), Clarkia hiloha
and C. lin^ulata (Gottlieb, 1974a), and Gaum lon^iflora and C demarcei (Gott-
lieb & Pilz, 1976), the species are known to be related as progenitor and deriva-
tive, respectively, with the derivative being of relatively recent origin.

The three cases represent the three possil)]e patliways of diploid speciation in
annual plants, defined in terms of the breeding systems: self -incompatible to
self-compatible, self-compatible to self-compatible, and self-incompatible to
self-incompatible, respectively. A fourth example of very high genetic similarity
between progeniU^r and derivative diploid species has recently been identified
in Lycopersicon (Rick et al, 1976). The lack of genetic divergence between the
members of each species pair indicates that, shortly after their origin, annual
plants, regardless of their breeding system, are still limited genetic versions of

172

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Table 2. Nei's (1972) moan genetic identity, J, between pairs of populations of con-
generic plant si^ecics.

Species

Clarkia hiloba X C. lingulaia
Clarkia rn])icunda X C franciscana
Caura lou^iflora X G. demarcci
Hijmcnopappus scahrosaeus X

H, ariemisiaejolius
Lnpinus tcxensis X L. stthcamosus
Oenothera strigosa X O. hiennis
Oenothera strigosa X O. parviflora
Oenothera biennis X O. parviflora
Phlox drummondii X P. euspidata
Stcphanomeria exigtia subsp. coronaria

X "Malhenrensis

»»

Ti(i^o))(y^on duJmis X T. porrifolitis
Tragopogon duhius X T. }>mtcnsis

Trafioi)ogon porrifolius X T. pratensis

I

0.88
0.28
0.99

0.90
0.35
0.97

0.54

0.55
0.67

0.94
0.50
0.62
0.53

Reference

Gottlieb, 1974a
Gottlieb, 1973a
Gottlieb & Pilz, 1976

Babbel & Selander, 1974
Babbel & Selander, 1974
Levy & Levin, 1975
Levy & Levin, 1975
Levy & Levin, 1975
Levin, 1975

Gottlieb, 1973b, 1976
Roose & Gottlieb, 1976
Roosc & Gottlieb, 1976
Roose & Gottlieb, 1976

their progenitors and posscs.s very few or no unique alleles insofar as their genomes
have been assayed. The species arc extracted from the parental repertoire of
phenotypic variation and genetic polymorphisms. Thus, the speciation process
does not seem to involve early reconstitution of the genome of the derivative
.species, even though it may possess certain unique moi-phological traits or other

features ( GottHeb, 1976 ) .

The other two examples of high genetic identity bet\veen species are not in-
consistent with this thinking. Thus, tlie similarity of Oenothera strigosa and O.
hiennis presumably reflects the fact that they have one genome in common (Levy
& Levin, 1975). And the high identity of the two species of Hymenopappm is
concordant with their very close phylogenetic relation.ship as judged by their high
overall morphological similarity; they were maintained as species because they
apparently do not hybridize despite extensive parapatric contact ( Turner, 1956 ) .
However, it is not clear if the results with annual plants will also characterize
perennial plant species. This is because perennials, especially long-lived ones,
appear to evolve gradually, rather than rapidly and abruptly, and they are more
likely to be reproductively isolated by ecological and pollination factors rather
than hybrid sterility resulting from chromosomal restructuring.

Electrophoretic evidence has also been used to show that species which ap-
pear similar may actually not be so. Thus, Clarkia franciscana, a highly self-
pollinating species thought to have evolved by rapid reorganization of chromo-
somes from tlie morphologically similar C. ruhicunda (Lewis & Raven, 1958),
is totally divergent from that species in a high proportion of its genes (those
coding six of the eight enzyme systems assayed) (Gottlieb, 1973a). In addition,
C. franciscana has a duplicated gene for alcohol dehydrogenase which further
distinguislies it from C. ruhicunda (Gottlieb, 1974b). Such marked genetic dif-
ferentiation requires that the phylogenetic separation of the two species oc-
curred much longer ago than had been presumed, and luakes its proposed mode
of origin quite uncertain. Therefore, a reasonable criterion to apply in cases like

1977] GOTTLIKB— ELECTROPHORESIS AND PLANT SYSTEMATICS yj^

this is that a species not be accepted as having originated recently from another
extant species if it is not electrophoretically highly similar to its putative parent
(Gottlieb, 1973a). This criterion has now been satisfied in the four examples de-
scribed above.

When additional electrophoretic studies are reported^ it may very well turn
out that such evidence reflects species divergence more sensitively than other
types of biochemical analysis. Thus, two-dimensional chromatography of cer-
tain flavonoids in Tragopogon cluhius, T, porrifoUus, and T. pratemis, failed to
distinguish a single component that was species-specific (Brehm & Ownbey,
1965), even though the morphological differences between these species are
"broad, shai-p and absolute" (Ownbey, 1950). But, electrophoretic analysis of
many enzymes in North American populations of the three Tragopogons re-
vealed very clearly that they were fixed for different alleles at about 40% of the
21 genes examined (Roose & Gottlieb, 1976). Other groups of plants have not
yet been studied so extensively both for electrophoretic variation in enzymes and
chromatographic variation in flavonoids and, therefore, it is not possible to know
if such a result will prove general. However, this is not implausible since changes
in the amino acid sequences of a large number of polypeptides which affect
electrophoretic mobilities of enzymes are more likely to reflect early stages of
genetic divergence than are changes in secondary metabolites such as flavonoids
which are products of enzyme-catalyzed biosyntlieses.

Electrophoretic evidence is also likc^ly to be useful in other systematic in-
vestigations which require knowledge of the extent of genetic similarity of closely
related diploid species. An attractive use will be to examine cases in which one
species appears to be a stabiHzed derivative of hybridization between two other
extant species such as LcLsthenia ])iirkei (Ornduff, 1976), Potentilla gJandulosa
subsp. hansenii (Clausen et al., 1940), Achillea rosea-allxi (Ehrendorfer, 1959),
and DelpJiinium gypsophikim (Lewis & Epling, 1959). Another use will be to
answer a novel systematic question having to do with the relative similarity of
species in different genera. A sample question might be: Are species of Clarkia
more similar to one another than species of Baptma? Once again the question
becomes plausible because of enzyme homology and because electrophoretic
evidence is composed of discrete, quantifiable units of information. The com-
parison of relative taxonomic distance in different genera might eventually lead
to the development of procedures to standardize certain taxonomic decisions.

A further application of electrophoretic evidence above the species level takes
advantage of its ability to distinguish species with different numbers of genes
specifying the same enzyme system. In cases where an enzyme is composed of
several polypeptides, the formation of "hybrid" enzymes by the association of
subunits coded by different gene loci provides strong evidence for their homology

and the origin of one of the genes through duplication. Clearcut examples of

such gene duplication have been documented for animal lactate dehydrogenase

(reviewed in Markert et al., 1975), hemoglobins (Ingram, 1961), and phospho-

glucoisomerase ( Avise & Kitto, 1973). Gene duplication is very probably a unique

event in the evolutionary history of organisms and, therefore, it can provide

evidence of the monophyletic origin of large groups of species and genera.

174

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

C L A R Id A

PG I

2 P G I GENES

PG I -1

P G I - 2 A/B

b/b

V^

GODETIA

£.

Primigenia

C. AMQ^

L

EUCHARIDIIM

PHAEOSTCm
C XANTIANA

Per I PETAsm

H. BILOBA

3 P G I GENES

PG I -1

a/a

a/b P
b/b

G I -2

PG I -3

Fk;. 2. Ei^ht diploid species in five of the seven diploid sections of Clarkia which have
been examined to date can he divided into two groups on the basis of the number of genes
they have specifying phosphoglucoisomerase subunits. The photograi^hs show typical electro-
phoretic phenotyi^es for diis enzyme system in the two groups. Details of the genetic analysis
are descrilx'd by Gottlieb (1977).

During studies of the genetic divergence of diploid species of Clarkia, I un-
covered two cases of apparent gene duplication (Gottlieb, 1974b, 1977). The

I ^^

one involving duplicated phosphoglucoisomerase (PGI) is particularly relevant
to systematic studies because it tests directly the taxonomic delimitation of sec-
tions within the genus proposed by Lewis & Lewis (1955). Genetic analysis of

PGI in Clarkia has shown that species either have two or three genes specifying
these enzymes (Gottlieb, 1976, 1977). To date, eight species in five of the
seven diploid sections of the genus have been examined: Clarkia ruhicunday
C amoemi, and C. franciscana in the presumed primitive section Primigenia
have two genes (Gottlieb, 1973a) as does C. wiUiamsonii in the Gocletia section
(Price, 1975). Section Phueostoma is represented by C. xantiana^ Peripetasma
by C. hiloha and C, dudleijana, and Eucharidium by C. coneintm, and all of
these species have three PGI genes (Gottlieb, 1977) (Fig. 2).

The duplication was originally recognized because individuals from species
with three PGI genes display more enzymes upon electrophoresis than those
from species with two PGI genes. The actual number of enzyme bands ob-
served depends on the allelic state of the coding genes and the affinity of poly-
peptides specified by the different alleles and the different gene loci. PGI is
composed of two polypeptides so that an individual heterozygous for different al-
leles at a single coding locus normally produces three enzymes by two-by-two
association of the two polypeptides (aa, bb, ab). In the Clarkias examined, the

1977] GOTTLIEB— ELECTROPHORESIS AND PLANT SVSTEMATICS 3^75

gene coding the most anodal PGI appears to be invariant since individuals with
either two or three PGI genes have always possessed a single fast PGI; poly-
peptides specified by this gene, called PCI-1, do not form a "hybrid" enzyme
with those specified by either of the other genes ( Fig. 2 ) .

Thus, the maximum number of PGI enzymes observed in the two-gene spe-
cies was four (the single fast PGI coded l:)y PGI-1 plus three enzymes in indi-
viduals heterozygous at PGI-2). However, as many as ten enzymes have been
observed in the three-gene species because polypeptides specified by the du-
plicated gene, called PGI-3, from "hyl)rid'^ enzymes with those specified by the
original PGI-2 gene. Thus, when both PGI-2 and PGI-3 are heterozygous, four
different polypeptides are made which aggregate to form nine distinguishable
enzymes (Fig. 1, case 3), and PGI-1 codes a tenth enzyme band.

The duplication is thought to have originated by the generation of a dupli-
cated chromosome segment in a progeny of a cross between individuals differing
for chromosomal rearrangements, possibly a partially overlapping reciprocal
translocation (reviewed in Burnham, 1962). Self-fertilization would make the
segment homozygous in a few generations. This mode of duplication is likely
in Clarkia because in this genus, species are self-compatible and differ by large
numbers of reciprocal translocations. Such duplications will not be linked;
recent genetic analysis (Gottlieb, 1977) in Clarkia xantiana has shown that
PGI-2 and PGI-3 assort independently, which is consistent with the proposed
duplication process.

The species with three PGI genes can be considered a monophyletic group
that traces back to an ancestor that l^ranched awav from the Primi<ienia-Godetia
stock. That two genes is the ancestral number is directly supported by the ob-
servation that Oenothera, the most closely related genus to Clarkia, also has two
genes for PGI (Levy et al., 1975) as does Gaura (Gottlieb & Pilz, 1976), another
genus in the same tri])e of the Onagraceae. The utilization of the number of
genes coding specific enzymes to classify groups of species into monophyletic
assemblages appears not to have a parallel in current systematic research. Gene
duplication provides a strict homology, al^sent with most morphological charac-
ters, because convergence in particular structural genes is highly improbable
since it would recjuire a very high numl^er of mutational changes to alter the cod-
ing properties of a different nonhomologous locus.

Electropiioretic Evidence: Polyi^loid Species

The ancestry of most allotetraploid species (tetraploids are used as an exam-
ple of polyploids) can, in principle, be traced back to a chromosome doubling in
a diploid individual which was produced b>- hybridization between differentially
adapted ^copulations. The initial allelic composition of the tetraploid plants is a
direct function of the degree of genetic di\ergence of the diploid progenitor
populations and is likely to be substantially greater if these represent spc^cies.
This follows because species have a very much higher prol^ability than con-
specific populations of possessing different alleles at their monomorphic gene
loci and nonoverlapping complements of alleles at polymorphic genes. After
the events of its origin, the courses of evolution in the tetraploid and its diploid

176

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

parents are independent which suggests that the more ancient the tetraploid, tlie
less hkely will it retain the alleles it inherited in an unmntated state, and, likewise,
alleles will continue to evolve in the diploids. Thus, the ability to identify the
diploid parents of a tetraploid species with electrophoretic evidence depends on
numerous factors having to do with the amount of divergence of the diploids at
the time of tetraploid origin as well as subsequent evolutionary events.

It follows that evidence brought to bear on the phylogeny of a tetraploid spe-
cies not be limited to a few enzymes or to a limited class of proteins, and that
particular attention be paid to specific protein homology and the mode of in-
heritance of the proteins examined in order that an absence or difference in
mobility can be interpreted in terms of genetic changes. This prescription has
been ignored in numerous studies of diploid and tetraploid plant species which
have employed only one or two enzyme systems or have sampled only seed pro-
teins and, consequently, many of these studies have uncertain value and are

not dealt with here.

In general, electrophoretic analysis has demonstrated that polyploid species
express, additively, enzymes present separately in their diploid parents. This
result has been reported, for example, in wheat (Hart, 1969; Mitra & Bhatia,
1971; Barber, 1970), cotton (Cherry et al., 1972), Nicotiana (Smith et al., 1970;
Reddy & Garber, 1971; Sheen, 1972), Phaseolus (Garber, 1974), Stephana mc da
(Gottlieb, 1973c), and Tragopogon (Roose & Gottlieb, 1976). If the duplicated
genes of polyploids specify different polypeptide subunits of multimeric en-
zymes (those that are composed of more than one polypeptide), additional "hy-
brid" enzymes are produced which are not expressed in a diploid parent if it
lacks both coding alleles. For many enzymes, the polyploid species is a "fixed
heterozygote" because all of its individuals express a multiple enzyme phenotype
that reflects their possession of different coding alleles inherited from the diploid
species. This multiple enzyme phenotype does not exhibit genetic segregation
because, at meiosis, chromosome homologues often pair preferentially so that
genes inherited from both diploid parents go to the same pole and each gamete
receives one copy of each of them. At fertilization, each gene is made homozy-
gous, but their presence in duplicate means that a heterozygous (multi-enzyme)
phenotype can be produced in the tetraploid. The multiplicity of enzymes in a
polyploid species may extend the range of environments in which normal de-
velopment can take place, and this is a reasonable hypothesis to account for
the frequent wider distribution of tetraploid species relative to the diploids in
^ ^ , , .,.3; Manwell & Baker, 1970; Gottlieb, 1976).

The general observation that the enzymes usually assessed by electrophore-
sis are expressed additively in polyploid species (the only apparent exception
is a study in wheat. Sing & Brewer, 1969) may reflect, to some extent, the rela-
tively recent origin of the tetraploids which have been examined. Although an-
cient polyploid complexes have not yet been studied by electrophoresis, many
recently evolved polyploids have considerable systematic and evolutionary signifi-
cance because they provide critical evidence regarding the initial genetic and bio-
chemical consequences of this type of genome doubling.

The most extensive comparison of enzyme variation in diploid and tetraploid

1977] GOTTLIKB— ELECTROPHORESIS AND PLANT SYSTEMATICS J77

species has been made in Tragopogon (Roose & Gottlieb, 1976). Tlie three dip-
loid species, T. duhhis, T. porrifolhis, and T. pratensis, were introdnced to
America from Europe during recent times. They are sharply delimited mor-
phologically without ov^erlap in a number of characters (Ownbey, 1950). In
southeastern \^^ashmgt()n and adjacent Idaho, Ownbey (1950) discovered that
interspecific hybridization betw^een them had given rise to two different tet-
raploid species: T. dubius and T. pornf alius were the parents of T. mirus, and
r. dubius and T. pratemis were the parents of T. miscellus. These two tetraploid
species represent the only unambiguous examples of the very recent natural
origin of polyploid species.

Electrophoretic evidence revealed that the three diploid species in North
America are completely divergei^t (monomorphic for different alleles) at about
40% of the 21 genes that were examined, a result fully concordant with their
morphological differentiation. Tlie tetraploids inherited both alleles at each of
these genes: T. mirus expresses an additive pattern for nine genes and T. mis-
cellus for seven genes, including five in common with T. mirus. In both tet-
raploids, the additive pattern includes novel hybrid enzymes not produced in
the diploid species. Each of the enzyme phenotypes in the tetraploids was fully
accounted for by simple additivity of the polypeptides specified by genes in-
herited from its respective diploid parents. Tlic observed patterns fully con-
firmed the ancestry of both tetraploids which was proposed by OwMibey (1950).
These results, as well as others, clearly indicate that electrophoretic analysis of
large numbers of enzymes can be an extremely useful and precise probe to iden-
tify diploid progenitors of polyploid species.

In summary, electrophoretic evidence can be used to test many different
types of hypotheses regarding genetic divergence that have already been gen-
erated in plant systematics. In addition, its unique perspective will probably
help to link questions previously considered to require only systematic or evo-
lutionary evidence to different sources of evidence in biochemistry and plant
development.

LiTERATURF ClTEI>

AuLAHi), R. W. 1975. Tlie mating system and microevolution. Genetics 79: 115-126.
& A. L. Kahleu. 1971. Allozyme p()l>nu:)ipliisms in plant populations. Stacller

Synip, 3: 9-24.
AvisE, J. C. 1974. Systematic value of electroplioretic data. Syst. Zool. 23: 465-481.
ik G. B. KiTTO. 1973. Pliosphoglueose isomerase gene duplication in tlie bon\- fishes:

an evolutionary history. Biochem. Genet. 8: 113-132.
, J. J. Smith & F. J. Ayala. 1975. Adapti\e difterentiation witli little genie change

between two native California minnows. E\"olution 29: 411-126.
Ayala, F. J. 1973. Two new subspecies of the Drosopliila iviJIistoiii group {Diptcra:

Drosophilidae). Pan-Pacific Entomol. 49: 273-279.
'. 1975. Genetic differentiation during the speciation process. E\()l. Biol. 8; 1-78.

Plenum Press, New York.

— . 1976. Molecidar Exolution. Sinauer Assoc, Inc., Sunderland, Massachusetts.

Babbel, G. R. & R. K. Selander. 1974. Genetic variability in edaphieally restricted and

widesi^read plant species. Evolution 28: 619-630.
Baker, J., J. M. Sxhth & C. Strobeck. 1975. Genetic polynu)rphism in the bladder cam-

pion, Silcne maritima. Biochem. Genet. 13: 393-410.
Barber, II. N. 1970. H>bridi/.ation and the e\-olution of plants. Taxon 19: 151-160.

jyg ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Bernsteix, S. C, L. H. Thhockmohton 8; J. L. IIuuby. 1973. Still more genetic vari-
ability in natural populations. Proc. Natl. Acad. U.S.A. 70: 3928-3931.

BoYEH, S. H., A. N. NoYEs, C. F. Timmons & R. A. Young. 1972. Primate hemoglobins:
polymorphisms and evolutionary patterns. J. Human Evol. 1: 515-543.

Brehm, B. & M. OwNBY. 1965. Variation in chromatographic patterns in the Tragopogon
dubius-pratensis-porrif alius complex (Compositae). Amer. J. Bot. 52: 811-818.

Bncnvx, A. H. D., A. C, Matheson & K. G. ELDmix;E. 1975. Estimation of the mating
system of Eurahfptus ohliqua L'Herit. by using allozyme polymorphisms. Austral. J. But.

23: 931-949.
BuRXHAM, C. 1962. Discussions in Cytogenetics. Burgess Publ. Co., Minneapolis.
Cavalli-Sforza, L. L. & A. W. F, Ein\ARDS. 1967. Phylogenetic analysis models and

estimation procedures. Evolution 21: 550-570.
Cherry, J. P., F. Katterman & J. Endrizzi. 1972. Seed esterases, leucine aminopep-

tidas(^s and catalases of species of the genus Gossypiuni, Theor. Appl. Genet. 42: 218-

226.
Chu, Y. & H. Oka. 1970. The genetic basis of crossing barriers between Onjza percnnis

subsp. harthii and its related taxa. Evolution 24: 135-144.
Clausen, J., D. D. Keck & W. M. ILesey. 1940. Experimental studies on the nature of

species. 1. Effect of varied envir(mments on western North American plants. Publ.

Carnegie Inst. Wasli. 520: 1-452.
Clegc, M. T. & R. W, Allard. 1972. Patterns of genetic differentiation in the slender

wild oat species, Avcna barhata. Proc. Natl. Acad. U.S.A. 69: 1820-1824.
EiiREXDORFER, F. 1959. Diffcreutiation-hybridization cycles and polyploidy in Achillea.

Cold Spring Harbor Symp. Quant. Biol. 24: 141-152.
Freet.ixc, M. & D. Schwartz. 1973. Genetic relationshix^s between multiple alcohol de-
hydrogenases of maize. Biochem. Genet. 8: 27-36.
Garrer, E. D. 1974. Enzymes as taxonomic and genetic tools in FhaHColus and Aspergil-

lus. Israel J. Med. Sci. 10: 268-277.
GoTTLiEii, L. D. 1973a. Enzyme differentiation and phylogeny in Clarkia franciscana^ C.

rubicunda and C. amoena. Evolution 27: 205-214.
. 1973b. Genetic differentiation, sympatric speciation and the origin of a diploid

species of Stephammieria. Amer. J. Bot. 60: 545-553.

— . 1973c. Genetic control of glutamate oxaloacetate transaminase in the diploid

plant Stephanomcria exigua and its allotetraploid derivative. Biochem. Genet. 9: 97-107.
— , 1974a. Genetic confirmation of the origin of Clarkia lingulata. Evolution 28:
244-250.

— . 1974b. Gene duplication and fixed heterozygosity for alcohol dehydrogenase in the

diploid plant Clarkia franciscana, Proc. Natl. Acad. U.S.A. 71: 1816-1818.

— , 1975. Allelic diversity in the outcrossing annual plant Stephanomeria exigua ssp.

carotifera (Compositae). Evolution 29: 213-225.

— , 1976. Biochemical consequences of speciation in plants. Pp. 123-140, in F. J.

Ayala (editor), Molecular Evolution. Sinauer Assoc, Inc., Sunderland, Massachusetts.
— . 1977. Evidence for duplication and divergence of the structural gene for phospho-

glucoisomerase in diploid species of Clarkia. Genetics 86: 289-307.

— & G. PiLZ. 1976, Genetic similarity between Gaura longijlora and its obligalelv

outcrossing derivative G. dcmareeL Syst. Bot. 1: in press.
Hart, G. 1969. Genetic control of alcohol dehydrogenase isozymes in Triticum dicoccum,
Biochem. Genet. 3: 617-625.

IIedrick, p. W. 1971. A new approach to measuring genetic similarity. Evolution 25:
276-280.

Hurry, J. L. & R, C. Lewoxtin. 1966. A molecular approach to the study of genie
heterozygosity in natural populations. I. T\\v number of alleles at different loci in
Drosophila j)setidoobscura. Genetics 54: 577-594.

IxcRAM, V. 1961. Gene evolution and the hemoglobins. Nature 189: 704.

Jacobs, M. 1975. Isozymes and a strategy for their utilization in plant genetics. I. Isozymes:
genetic and epigenetic control. Pp. 365-378, in L. Ledoux (editor), Genetic Manipula-
tions with Plant Material. Plenum Press, New York.

JoHxsox, G. B. 1976. Hidden alleles at the alpha-glycerophosphate dehydrogenase locus
in Colias butterflies. Genetics 83: 149-167.

Levin, D. 1975. Interspecific hybridization, heterozygosity and gene exchange in Phlox,
Evolution 29: 37-51.

1977] GOTTLIEB— ELECTKOPHORESIS AND TLAXT SYSTEMATICS yjQ

& W. L, Chepet. 1973. Genetic xariation in Lijcopodium luciduhun, a phylogenetic

relic. Evolntion 27: 622-632.

Levy, M. & D. A. Levix. 1975. Genie heterozygosity and variation in permanent trans-
location heterozygotes of the Oenothera biennis complex. Genetics 79; 493-512.

, E. E. Steineh & D. A. Levin. 1975. Allozynie genetics in permanent translocation

heterozygotes of the Oenothera biennis complex. Biochem. Genet. 13: 487-500.
Lewis, H. & G. Epling. 1959. Delphinium ^ijpsophihun^ a diploid species of hybrid origin.

Evolution 13: 511-525.
& M. E. Lewis. 1955. The genns Clarkia. Vmv. Calif. Pnbl. Bot. 20: 241-392.

& P. H. Raven. 1958. Rapid evolntion in Clarkia. Evolution 12: 319-336.

Lewontin, R. G. 1974. Tlie Genetic Basis of Evolutionary Change, Columbia VuW. Press,

New York.
Manwell, C. & C. M. A. Baker. 1963. A sibling species of sea cucumber discoxered by
starch gel electrophoresis. Comp. Biochem. Physiol. 10: 39-53.

& . 1970. Molecular Biology and the Origin of Species. Univ. Washington

Press, Seattle.

Markert, C. L., J. B. Shaklee & G. S. Wiutt. 1975. Evolntion of a gene. Science 189:
102-114.

Marshall, D. R. & A. II. D. Brown. 1975. Optinuun sampling strategies in genetic con-
servation. Pp. 53-70, in O. H. Frankel ik J. G, Ilawkes (editors). Crop Genetic Resources

for Today and Tomorrow. International Biological Programme, Vol. 2. Candjridge Univ.
Press, New York,

xMiTRA, R. & C. Bhatia. 1971. Isoenzymes and po]>ploidy. L Qualitati\'e and quantitative

isoenzyme studies in the Triticinae. Genet. Res. 18: 57-69.

Nei, M. 1972. Genetic distance between populations. yVmer. Naturalist 106: 283-292.

. 1975. Molecular Population Genetics and Evolution. North-Holland Publ. C^o.,

Amsterdam.
Nevo, E., Y. J. Kim, C. R. Shaw & C. S. Thaeleh. 1974. Genetic variation, selection and

speciation in Tliomonnjs talpoides pocket gophers. Evolution 28: 1-23.
Ohnduff, R. 1976. Speciation and oligogenic differentiaticm in Lasihenia (Compositae).

Syst Bot. 1: 91-96.
OwNBEY, M. 1950. Natural liybridization and amphiploidy in the genus Tra^opoiJ,o)i.

Amer. J. Bot. 37: 487-499.
Powell, J. R. 1975. Protein variation in natural populations of animals. Evol. Biol. 8:

79-119. Flenmn Press, New York.
PmcE, S. C. 1975. Genetic variability in ecologically marginal and central populations of

tlie diploid plant Clarkia willianisonii ( Onagraceae). M.S. thesis, Cahfornia State Univ.,

San Diego.
Reddy, M. & E. CiAUBER. 1971. Genetic studies of \ariant enzymes. III. Coniparati\e

electrophoretic studies of esterases and peroxidases for species, hybrids, and anipliii^loids

in the genus Nicotiana. Bot. Gaz. (Crawfordsville) 132: 158-166.
Rick, C. M. & J. F. Fobes. 1975. Allozymes of Galapagos tomatoes: x^olymorphism, geo

Wl-

graphic distribution, and affinities. E\{)lution 29: 443-457.

— , E. Kesicki, J. F. FoBEs & M. Holle. 1976. Cienetic and bios>stematic studies

on two new sibhng species of Lijcopersicon from interandean Peril. Theor. Appl. C^enet.
47: 55-68.

R(x;ehs, J, S. 1972. Measures of genetic similaritv and genetic distance. Univ. Texas Publ.
7213: 145-153.

Roose, M. L. & L. D. Gottlieb. 1976. Genetic and biochemical consequences of poly-
ploidy in Tragopo^on, E\()lution 30: 818-830.

Sc;andalios, J. G. 1969. Genetic control of nuiltiple molecular forms of enzymes in plants:
a review. Biochem. Genet. 3: 37-79.

. 1974. Isozymes in de\elopment and differentiation. Annual Rev. Pi. Physiol.

25: 225-258.

Schaal, B. a. 1975. Population structure and local differentiation in Liatris eiflindraeea.
Amer. Naturalist 109: 511-528.

Selaxder, R. K. 1976. Genie variation in natural populations. Pp. 21—45, in F. J. Ayala
(editor), Molecular Evolution. Sinauer Assoc, Inc., Sunderland, Massachusetts.

SiiAW% G. R. 1970, How^ many genes exoKe? Biochem. Genet. 4: 275-283.

Sheex, S. 1972. Isozymic evidence bearing on the origin of Nicotiana tabacum. Evolution

26: 143-154.

180

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Sing, C. F. & G. J. Brewer. 1969. Isozymes of a polyploid scries of wheat. Genetics 61:

391-398.

Singh, R. S., J. L. IIuhhy & R. C. Lewontin. 1974. Molecular heterosis for heat-sensitive

enzyme alleles. Proc. Natl. Acad. U.S.A. 71: 1808-1810.
Smith, IT. IL, D. E. Haxhll, E. A. Weaver & K. 11. Thompson. 1970. Multiple molecular

forms of peroxidases and esterases among Nicoiiana species and amphiploids. J. IlcrecUt)'

61: 203-212.
Torres, A. M. 1976. Dissociation-recombination of intergenic sunflower alcohol delndro-

genase isozymes and relative isozyme activities. Biochem. Genet. 14: 87-98.
TuiiNEH, n. L. 1956. A cytotaxonomic study of the genus UijmenopappHs (Compositae).

Rhodora 58: 161-308.
— — . 1969. Chcmosystematics: recent developments. Taxon 18: 134-151.

THE APPLICATIONS OF MOLECULAR EVOLUTION TO
SYSTEMATICS: RATES, REGULATION, AND THE

ROLE OF NATURAL SELECTION'

Mary- Claire King-

The dovelopinent of biocliemical metliods for comparing tlie amino acid se-
quences of homologous proteins from different species has provided a powerful
tool for investigations of evokition and systematics. Perhaps the most intriguing
(and most controversial) result of the comparative studies of proteins using
these metliods has been the discovery that sequences may change at nearly
constant rates (Wilson, Carlson & White, 1977). This is not to imply that dif-
ferent genes or proteins evolve at the same rate: rather, each class of proteins
has its own characteristic rate (Dickerson, 1971). (Serum albumin, for exam-
ple, has evolved more rapidly than cytochrome c\ but serum albumin has evolved
at approximately the same rate among all species of mammals tested, as has cyto-
chrome c.) Tlie degree of rate constancy has been the subject of intense debate,
but the most current evidence indicates that the variation in e\'olutionary rate
for a given protein is only about twice the variation expected for a totally
stochastic process such as radioactive decay (Fitch, 1976). Within these limits,
then, a given macromolecular sequence may be used as an evolutionary "clock."

The empirical discovery that molecules can be evolutionary "clocks" has been
applied to a variety of problems in evolution and systematics. Most frequently,
sequence data for a given protein from a number of species has be(Mi used to
reconstruct phylogenetic trees depicting the probable order of branching of tlie
lineages leading to modern species from a common ancestor. For example, Bouh
ter and his colleagues have reconstructed a possible phylogeny for the flowering
plants based on the cytochrome c, plastocyanin, and ferredoxin sequences of
representative species (Boulter, 1974). In addition, phylogenetic analysis of se-
quences of 58 and 16S ribosomal RNA from chloroplasts, bacteria, blue-green
algae, and cytoplasm of green plants has confirmed that chloroplasts evolved
from photosynthetic prokaryotes living as endosymbionts within the cytoplasius
of primitive heterotrophic plants (Margulis, 1970; Bonen & Doolittle, 1976;
Zablen et ah, 1975; Ilori, 1975).

Molecular phylogenies may also indicate evolutionary times of divergence il
the divergence time for at least one branching event in a tree can be accurately
estimated from paleontological or biogeographical evidence. The use of cyto-
chrome c sequences to estimate times of divergence for fk)weriug plants poses
a fascinating, and still unresolved dilemma. If the cytochromes c of plants are
evolving at the same rate as those of vertebrates, which have a unit evolutionary
period of about 20 million years, then the intraordinal divergence times for flower-

^I thank A. C. Wilson, S. C Carlson, and T. J. White for gcnerons contribution of ideas
and reference material, and A. Hinley for technical assistance.

" Department of Biomedical and Environmental Health Sciences, University of California,
Berkeley, California 94720.

Ann. Missouri Bot. Card. 64: 181-183,

]^g2 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

ing plants would be about 240 million years ago (Ramshaw et al., 1972; Boulter
et aL, 1972). Since the first clearly authentic fossils of flowering plants occur
about 130 million years ago (Sporne, 1971), an alternative interpretation of the
data is that morphologically plant cytochromes c have evolved twice as fast as
those of vertebrates (Cronquist, 1976). Workers in this field are now giving
serious attention to the possibility that the origin of flowering plants is more
ancient than is indicated by the available fossil evidence (Wilson, Carlson
& White, 1977). This situation may be analogous with the origin of mammals,
in that the group is very ancient, while adaptive radiation within the group is

more recent.

Tlie molecular evolutionary approach may also have revealed an important
mechanism for evolution at the organismal level. This discovery results from
the observed discrepancy between the evolution of macromoleculcs and the evo-
lution of organisms. The comparison of humans and chimpanzees at both the
macromolecular level and organismal levels indicates that the two species dif-
fer to an extent considered familial in morphology, behavior, and adaptive strat-
egy, while their protein sequences differ by less than one percent — a level of
difference characteristic of sibling species of Drosophilia or mammals (King &
Wilson, 1975), Major adaptive changes may thus be based on molecular events
other than sequence changes in structural genes. What sorts of events might
these be? Experimental studies of bacterial evolution have demonstrated that
major phenotypic changes — in the bacterial case the acquisition of a new meta-
bolic activity — depend on an increase in the effective concentration of a protein
which previously limited the rate of metabolism of a given substrate, rather
than on a (juaUtative change in the substrate specificity of any protein (Lerner
et al., 1964). These quantitative effects could be due to point mutations in regu-
latory genes or to chromosomal rearrangements such as duplications and trans-
locations (Wilson, 1975). The observation that rates of karyotypic change are
fastest in vertebrate groups with the most rapid phenotypic evolution may indi-
cate that major adaptive shifts in the evolution of multicellular organisms are
frequently associated with chromosomal rearrangements (Wilson et al., 1975).

The independence of the evolution of organisms and the evolution of their
structural genes may provide a new perspective for investigating the evolutionary
roles of natural selection versus random fixiation of selectively neutral alleles.

If tlie random fixation of neutral substitutions were principally responsible for
sequence evolution of genes and proteins, it would follow that the rate of se-
quence evolution would depend primarily on the mutation rate, which is as-
sumed to be constant with time. The "neutial" hypothesis is thus consistent with
the observation that sequence evolution depends on calendar time. In addi-
tion, the "neutral" hypothesis accounts for the observation that proteins can
differ greatly in sequence without differing appreciably in biological activity. At
the same time, it is unequivocably established that natural selection acts at the

level of the organism, and that this selective pressure varies greatly over time
and space. It is tempting to suggest a hypothesis consistent with — though in no
way proven by — each of these observations: that fixed substitutions in the vari-
able regions of structural genes have generally not been subject to selective

1977] KING— MOLECULAR EVOLUTION AND SYSTEMATICS ^§3

pressures, since these substitutions are largely irrelevant to the adaptive success
of the organism. Instead, natural selection at the level of the organism may be
reflected at the molecular level in both rapid elimination of deleterious muta-
tions in regulatory systems and fixation of occasional adaptive changes in loci
or patterns of genome organization controlUng tlie expression of structural genes.

LiTKUATUllK CiTKl)

BoxEX, L. & W. F. DooLiTTLi:. 1976, Partial setiiicnces of 16S rRNA and the phxlo^ciiy

of I)lue-green algae and chloroplasts. Nature 261: 669-673.
RouLTKn, D. 1974. The evolution of plant proteins with speeial reference to higher plant

cytochrome c. Curr. Advances Pi. Sci. Comment. Pi. Sci. 8: 1-16.
, J. A. M. Ramsiiaw, E. W. Thomi'son, M. Richardson & R. H. Ruown. 1972. A

phylogeny of higher plants based on the amnio acid sequences of cytochrome c and its

biological implications. Proc. Roy. Soc. London, Ser. B, Biol. Sci. 181: 441-455.
Choxquist, A. 1976. The taxonomic significance of the structure of plant proteins: a

classical taxonomist's view. Brittom'a 28: 1-27.
DicKEHsox, R. E. 1971. The structure of cvtochrome c and tiie rates of molecular evolution.

J. Molec. Evol. 1: 26-45.

Fitch, W. M. 1976. The molecular evohition of cytochrome c in eukaryotes. ]. Molec.
Evol. 8: 13-40.

HoRi, H. 1975. Evolution of 5S RNA. J. Molec. Evol. 7: 75-86.

King, M. C. & A. C. Wilson. 1975, Evolution at two levels in humans and chimpanzees.
Science 188: 107-116.

Lerneh, S. a., T. T. Wu & E. C. C. Lin. 1964. Evolution of a catabolic pathway in bac-
teria. Science 146: 1313-1315.

Mahgulis, L. 1970. Origin of Eukaryotic Cells. Yale Univ. Press, New Haven.

Ramsiiaw, J. A. M., D. L. Richardson, B. T. Meatyahd, R, H. Brown, M. Richardson, E.
W. Thompson & D. Bouli:er. 1972. The time of origin of the flowering plants de-
termined by using amino acid sequence data of cytochrome c. New Phytol. 71: 773-
779.

Sporne, K. R. 1971. The Mysterious Origin of Flowering Plants. Oxford Univ. Press, Ox-
ford.

Wilson, A. C. 1975. Evolutionary importance of gene regulation. Stadler Genet. Symp.
77: 117-134. Univ. of Missouri, Cohimhia.

, S. S. Carlson & T. J. Whitk. 1977. Biochemical evohition. Annual Rev. Biochem.

46: in press.

— , G. L. Bush, S. M. Case & M. C. Kinc;. 1975. Social structuring of mammalian

populations and rate of chromosomal evolution. Proc. Natl. Acad. II. S. A. 72: 5061-5065.

Zahlen, L. B., M. S. Kissil, C. R. Woese & D. E. Buetow. 1975. Phylogenetic origin of

the chloroplast and prokaryotic nature of its ribosomal RNA. Proc. Natl. Acad. U.S.A.
72: 241^-2422.

CHEMOSYSTEMATICS— ANALYSES OF POPULATIONAL

DIFFERENTIATION AND VARIABILITY OF ANCESTRAL

AND RECENT POPULATIONS OF JUNIPERUS ASHEl'

Robert P. Adams-

Ah SIR ACT

Three types of data were used to anal>'ze 28 natural populations of Jtinipcrus ashci:
16 morphological characters, 152 terpenoids, and 23 peroxidases. In this paper the peroxidase
electromorphs were treated as ordinary qualitati\'e chemical characters to examine the feasi-
bility of using isozymes for taxonomic purposes and as indicators of populational variability.
The data sets were subjected to various mnnerical analyses to examine regional trends, an-
cestral affinities, and variability within populations. Principal coordinate analysis was used
to resolve the major coordinates of variation from the similarity matrix of each data set.
Coordinate loadings were then contoured for the first three coordinates of each similarity
matrix to aid the visualization of the regional trends. The terpenoids and morphology showed
a series of uniform populati<ms from central Texas into the Ozarlcs with divergent popula-
tions on the south and west portions of the range, extending into northern Mexico. No re-
gional trends were apparent in the peroxidases and no corresponding modes of variation
were seen between the peroxidases and the otlier two data sets. Pleistocene x'egetation is
reviewed and migration paths are speculated upon. Advanced and primitive character states
, are discussed. The uniform body of /. ashei populations from central Texas to the Ozarlcs
appear to be advanced (recent), whereas the divergent populations seem to be more primitive
(ancestral). A method called differential similarities is introduced to analyze the clinal grada-
tion of /. ashci toward /. saliillensis in Mexico. Intrapopulational variability was analyzed
by use of the a\'erage similarity witliin populations and the coefficient of phenetic variation

(CPV). In general, the recent populations had high similarities and low variability, and the
ancestral populations had lower average similarities and higher CPVs with both the morphologi-
cal and terpenoid data. The pattern of variation in the peroxidases could not be generalized
upon, but appeared to be mosaic. Peroxidases did not appear useful in this analysis when
subjected to standard numerical analysis procedures. The evolution of /. ashei into its present
distribution appears to have had at least two phases composed of very uniform, recent migra-
tions and persistent, variable, relict populations perhaps extending close to the geographic
origin of this taxon in northern Mexico.

The use of ch(Miiical characters has gained widespread acceptance dnring the
past decade to the point that a graduate student thesis in systeniatics is now un-
usual if no chemical data are utilized. Because of the relative ease of use, flavo-
noids are widely utilized in systematic and evolutionary plant studies. The
early works on Asplenium (Smith & Levin, 1963), Lemnaceae (McClure &
Alston, 1966) and Baptisia (Alston & Turner, 1963) are classics, required read-
ing for chemosystematic students. Likewise, classic is the work on betalains by
Mabry and coworkers (summarized in this symposium). Whereas flavonoids
and betalains have been extensively used above the species level (probably due
to the qualitative nature of the methods), terpenoids, due to the quantitative
nature of gas/liquid chromatography, have been more widely used at or below

M thank Walter Kelley for field assistance and the use of the peroxidase data which he
diligently gathered. This research was supported by NSF Grants GB24320 and GB37315X.
Computer time was furnished by Colorado State University.

^ Science Research Center, Hardin-Sinnnons University, Box 1095, Gruver, Texas 79040.

Ann. \lissoum Box. Gard. 64: 184-209. 1977.

1977] AT:>AMS—JUNIPERVS ASHEl

185

the species level. The gymnosperms have been the focus of many studies of
populational differentiation which have uncovered clines (Flake et al., 1969,
1973), chemical races (Smith et al., 1969), hybridization (see von Rudloff, 1975
for an excellent review), and ancestral migrations (Adams, 1975a; Zavarin &
Snajberk, 1973). In the angiosperms, work on the monterpenes of Burscra
(Mooney & Emboden, 1968) demonstrated the use of these compounds in the
detection of clinal variation. Of course, the work on the Australian Euc(ily})tus
species has been of tremendous use in the classification of populations of these
taxa and is well known.

Although studies using terpenoid characters to analyze populational differ-
entiation are well known, the analysis of population variability in relation to im-
portant population biology (Questions such as the founder's effect, genetic drift,
the effects of small versus large populations and central versus peripheral sites
on variability have not been addressed. The relatively recent rise in the use of
isoenzyme data has rekindled an interest in the examination of these questions.
Gottlieb (at this symposium) has reviewed the literature on isozymes and their
use in systematics. Nevertheless, it seems in order to mention that the ''isoz\'me
bandwagon" has become the current fad before we have developed a very
thorough knowledge about the molecular basis of the electromorphs distinguished
on gels.

Before the widespread use of isozymes, the study of variability within popu-
lations seems to have stagnated with the exception of the numerical taxonomic
school (including morphometries). Gilmartin (1969a, 1969b, 1974, 1976) has
introduced a new idea called the coefficient of phentic variation (CF\^) to ex-
amine the combined effects of many characters on variability. The CPV is
merely the standard deviation of the mean similarity among a group of opera-
tional taxonomic units (OTUs) divided by that mean similarity. Whereas the
mean similarity of a group tells about the average affinities, the CPV shows how
homogeneous are the similarities of one group versus another group. Since the
CPV is normalized bv the mean similaritv, different character sets can be com-
pared as well as different levels of organization (i.e., population vs. species vs.

genus). To my knowledge, the CPVs have not been used to study population vari-
ability with the exception of the studies by Gilmartin. The purpose of this paper

is to examine population differentiation and variability in Juniperus ashei Ikich.
using three contrasting sets of characters: moiphological characters, volatile
terpenoids from leaves, and leaf peroxidases. The literature on /. ashei has been
reviewed by Adams & Turner ( 1970).

Juniperus ashei is a taxon of a rather restricted range, occurring on lime-
stone outcrops from northern Mexico to southern Missouri (Fig. 1). The Ed-
wards Plateau region of central Texas supports dense populations covering
thousands of acres, whereas the disjunct populations ( Lubbock-Post, Texarkana,

Arbuckle Mountains, Ozark Mountains, and northern Mexico) often have nearly
pure stands of /. ashei, but seldom cover such large areas. Being a fairK' con-
spicuous conifer tree, one can be relatively confident in the taxonomic distribu-
tion records which imply that there are few, if any, trees bet\veen the disjunct
populations and the Edwards Plateau populations. Thus, this would appear to

186

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

MISSOUR

300

KILOMETERS

SIERRA
CARMEN MTNS.

Fk;uhe 1. Distribution of Junipcnis ashci showing the 28 populations sampled for tliis
study. Tlie exact distribution of /. ashci in nortliern Mexico is not known and is indicated
giMUMally by a dashed line.

be an excellent taxon to test some of the hypotheses advanced by Ehrlich & Raven
(1969) in regard to gene flow versus selection m the maintenance of species.

Previous research (Adams & Turner, 1970; Adams, 1975a) has shown that
the terpenoids of this taxon exhibit a remarkably high similarity between central
Texas and the Ozarks (Fig. 2). However, many significant differences were
found between populations 12, 13, and 17 and the other popuhitions. One tree
of/, ashci (number 116 in Fig. 2) was discovered in Mexico and found to cluster
with the atypical populations (12, 13, 17). This, iilong with similar evidence in
/. pinchotti populations (Adams, 1975b) seemed to imply that relicit migrations
have been very important in the establishment of these patterns.

Evidence from rat middens and palynology in the southwestern United States
is considerable (King, 1973; Mehringer et al., 1970; Van Devender & King, 1971;
Wells, 1965, 1966, 1970; Wells & Berger, 1967; Whitehead, 1972; Wright, 1970)

1977]

AUAMS—JUNIPERUS ASIlEl

187

KILOMETERS

Figure 2. Contoured similarities based on 54 terpenoid cluiracters, F-1 weighted.
Notice the uniformity from central Texas to the Ozarks and the clustering of populations 12,
13 and 17 with tree 116 from nortliern Mexico (from Adams, 1975a).

that the Pleistocene ice advances pushed boreal and temperate species to lower
elevations and southward. The northern Chihuahuan desert was certainly in-
vaded by Jwiiperus (Wells, 1966) and crossed repeatedly. Even so, the data
presented by Adams (1975a) for the close similarities of populations 12, 13 and

J

This is due to

the use of only 1 tree (number 116) from northern Mexico and the fact tliat no
morphological data were used except in the largely preliminary study by Adams
& Turner (1970).

In this study I will remedy these shortcomings by reporting on 15 trees of
/. ashei from northern Mexico (population 25 in Fig. 1), as well as 4 additional
populations: Post (near Lubbock), 24; Pandale, 26; Texarkana, 27; SaHne
Creek, Oklahoma, 28 (sec Fig. 1). In addition, I report data on 16 morphological

188

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 61

Table 1. Sixteen morphological chiiracters and states scored for 15 trees from eaeli of
the 28 populations of /. ashci sampled (Fig. 1). Missing data was coded by a -1.0 for a flag
in statistical analysis.

Character

FDI
FCO

SPF

BLM

SEA

SER

WGA

WGR

WLM

wc;p

WRP

G/S
SLL
L/B

RAN

States (if applicable)

FEMALE CONE DIAMETER: avg. of up to 10 and not less than 4 (in mm).

FEMALE CONE COLOR: 1.0-4.0 (blue-yellow /brown).

SEEDS PER FEMALE CONE: avg. of up to 10 cones scored, not less than 4.

BLOOM ON CONE: 1.0-3.0 (none to very heavy coating).

SEED AREA: seed length X width, avg. of 10 seeds and not less than 4.

SEED WIDTH/LENGTH: a\g. of 10 seeds and not less than 4.

WHIP LEAF GLAND AREA: whip leaf gland width X length; avg. of 5 glands.

WHIP LEAF GLAND LENGTH/WIDTH: ratio, avg. of 5 glands.

WHIP LEAF MARGINS: 1.0-4.0 (smooth-heavy serration) avg. of 5 leaves.

WHIP GLANDS PROTRUSION: 1.0-3.0 (sunken-smoodi-protrudes), avg. of

5 glands.
WHIP GLANDS RUPTURED: 1.0-3.0 ( none-some-almost all), avg. of 5 ob-

scr\ations.

WHIP LEAF BLADE LENGTH/SHEATH LENGTH: avg. of 5 leaves.
WHIP LEAF CJLAND LENGTH/SHEATH LENGTH: avg. of 5 leaves.
SCALE LEAF LENGTH: avg. of 5 leaves.

SCALE LEAF L1:NGTH,'RRANCII WIDTH: Ratio of scale leaf length to the
width of the branch (twig) where that scale leaf was borne. Avg. of 5 niea-

sur»'ments.
BRANCHING ANGLE: Angle of branching of ultimate twig, avg. of 5 measure-
ments ( each to nearest 5 degrees ) .

characters, as well as peroxidases, from leaves. Finally, I compare these 3 sets of
characters both in regard to their use in the analysis of populational differentia-
tion and in the analysis of variability within popuhitions.

Materials and Methods

7

n

atiiral

range (Fig. 1). For the terpenoid and morphological characters, 15 trees were
sampled from populations 1 through 23 in December, 1970, and 15 trees were
sampled from populations 24 through 28 in December and January, 1974-1975, to
complete the sampling. The sampling methods are given in Adams & Turner
(1970), except that in 1974-1975, the foliage was generally frozen within a few
hours in the freezer of our field trailer. Voucher specimens are on file at Colo-
rado State University. All samples from each of the two sampling periods were

placed in a random sequence for distillation as advocated by Adams (1975c).
These procedures convert the temporal changes in foliage, oils, columns, etc. to
random variables. Therefore population differentiation patterns can be readily
separated from experimental procedural errors in the statistical analysis phase.
The volatile terpenoids were steam distilled for 2 h as outlined by Adams (1970)
and the extracts were kept at -20''C until analyzed by gas/liquid chromatography.
Separation was made on a 200 ft X 0.02 in. capillary column (wall coated

20 M

/

and Adams & Turner (1970). In-

dividual peaks were quantified with an electronic digital integrator and auto-
matically punched onto computer cards.

1977] ADAMS— /C/iV/PEfll'S ASIlEI

189

Sixteen morphological characters were scored as outlined in Table 1 for
15 specimens of 28 populations. Some fruit (female cones) and seed characters
were not scored (and were thus set to -1.0 as a flag) since not all trees sampled
had female cones.

One hundred and forty-two terpenoids were subjected to analysis of variance
(ANOVA) to detennine whicli characters showed significant differences among
populations. Fifty-nine terpenoids had F ratios greater than 1.0, a maximum
population average greater than 0.1% and were used to compute F-1 weighted
(Adams, 1975c) mean character differences (MCD or Manhattan metric) simi-
larity measures between populations (see Adams, 1972, for exact formulation).
This similarity matrix (28 X 28) was then used as input for principal coordinate
analysis (Cower, 1966, 1967; Williams et al., 1971) to factor the sinn"laritv matrix
into major coordinates of variation. The first 3 principal coordinates were used
to contour map populations as they were ordinated on each of the ortho";onal axes.

The 16 morphological characters were also analyzed by ANOVA and the
Student-Newman-Keuls (SNK) multiple range test was applied (P = 0.05) to
determine which populations were significantly different. Fifteen morphological
characters (FEMALE CONE COLOR was omittecl, F = 0.88) were used to
compute a similarity matrix which was then factored by principal coordinates.
The first 3 coordinates were contour mapped as outlined above.

For the peroxidase work, foHage of 30 plants (occasionally less, see Kelley,
1976) were sampled from 15 populations in N()\ ember-December, 1974, and
frozen in the field trailer within a few hours. This foliage was kept frozen until
extracted. The enzymes were extracted by grinding the foliage in liquid m'trogen
with alumina tlien adding an extraction buffer of 0.10 M trismaleate, pH 7.00
containing: 0.02 M sodium tetraborate; 0.25 M sodium ascorbate; 0.02 M sodium
meta-bisulfite; 0.02 M sodium diethldithiocarbamate (DIECA); 0.01 M ger-
manium dioxide; 107^ (v/v) dimethyl sulfoxide (DMSO) plus polyvinylpoly-
pyrrolidone (P\'PP), 10 gms/50 ml buffer. The complete instructions are lengthy
and the interested reader is referred to Kelley (1976) and Kelley & Adams
(1977a) for complete details. The peroxidases were concentrated and electro-
phoresed on acrylamide gels (discontinuous 4.5, 6, and 8% anodic, see Kelley,
1976) within 72 hours from the time of extraction. Although Kelley (1976) ana-
lyzed peroxidases, esterases, and an alcohol dehydrogenase, I am only using the
peroxidase data since it showed much of the same pattern of variability as the
other systems (Kelley, 1976). Peroxidases in Jiiniperus are little effected by sea-
sonal differences (Kelley & Adams 1977a), and peroxidases are generally very
stable (Kelley, 1976). Peroxidases were stained with o-tolidine/H.Oo (Deuna
& Alexander, 1975). An aggregate total of 23 peroxidase bands were found in
the 15 populations of /. ashei sampled. In cases where bands were very close
together on the gel, samples were corun to deternn"ne which electromorphs were
different. These bauds were each scored as 1.0 (present) or 0.0 (absent) for
each plant and then subjected to ANO\\\ to obtain some estimate of F ratios
for character weighting. Of course, ANOVA of (lualitative data has a tendency
to underestiuiate the F ratios, but this did pro\ide a crude method to obtain
relative character weights. Sixteen peroxidases had F greater than 1.0 and were

190

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

°/o), TERRENES

COORD.2(9%), TERRENES

COORD.1 (38%), MORPH.

COORD. 2(9%), MORPH.

COORD. 3(S°/o), TERRENES

COORD 3(8%)^ MORPH.

1977] ADAMS— /[/iV/P£flL/S ASHEI

191

not uniformly unique to one population. Similarity measures were computed
as outlined above, and the similarity matrix (15 X 15) was factored to obtain
principal coordinates. The first 2 coordinates were contour mapped for com-
parison of regional trends.

For the analysis of within populational varialjility, 3 sets of similarity measures
were calculated using all terpenoid, morphological, and peroxidase characters,
equally weighted. It appears that F weighting is not desirable when examinintr
intrapopulational variation. These analyses resulted in 3 kinds of similarity
matrices (terpenoid^morphological, and peroxidase) for each population. The
average similarity (Sr) was then computed for each population along with the
coefficient of phenetic variation (CPV = Sd-/Sr). The Sr's and CPVs were then
contour mapped to examine regional trends of intrapopulational variation.

Populational Differentiation

The principal trend in the terpenoid similarities is that of the differentiation
of populations 25, 26, 12, 13 and 17 from the rest of the populations (Fig. 3).
From these coordinate loadings one can see (Table 2) that 50% of the variation
in the similarities is mostly due to the divergent nature of populations 25, 26, 12,
13 and 17. The high negative loading of population 17 onto coordinate one in-
dicates that population 17 (New Braunfels, Texas) has considerable affinities
with the west Texas and Mexico plants. It is interesting to compare the major
trend of the terpenoids with that of tlie morphology (Fig. 3 vs. Fig. 4). This
major trend in the morphology accounts for 387r of the variation in similarities
and is practically identical to the major trend of the terpenoids. A couple of ex-
ceptions are that the Post population (24) seems more similar to the west Texas-
Mexico populations in the morphology, while the New Braunfels population
(17) is not quite as different from the central Texas pc^pulations in its morphology
as in its terpenoids. In both cases, from central Texas to the Ozarks a picture
of uniformity is presented. It might be noted that this compares very closely
with the contoured terpenoid phenogram in Fig. 2 (from Adams, 1975a). It
appears that the major coordinate of principal coordinates analysis is the dominant

FiGUHEs 3-8. — 3-4. Contoiiifd loatlinjis of principal coordinate 1 extracted from similar-
ity measures among populations, (see Ta])Ies 2-3).— 3. Tins pattern extracted 50% of the
variation from the terpenoid similarity matrix. Notice tliat this pattern is the principal pattern
previonsK shown (Fig. 2). Contours: 1 = -0.70; 7 = 0.18. — 1. This pattern accounted for
387r of the variation from the morpliological similarity matrix. The Post population (24)
shares some affinities to the west Texas-Mexico populations, l^opulation 17 seems a little less

di\ergent in its morphology than its terpenoids (Fig. 3). Contours; 1 = -0.64; 7 = 0.13.

5-6. Contoured principal coordinate 2. — 5. This trend (9%, terpenoids) seems to be due to
the divergence of populations 24, 27, and 28, plus sampling differences (see text). Contours:
1 = -0..37; 7 = 0.16. — 6. This trend {9%, morphological) seems to be due to procedural
differences in scoring the morphological characters (see text). Contours: 1 = -0..32; 7
0.16. — 7-8. Contoured principal coordinate 3. — 7. Di\ergence of the Post (24) population

is most e\ident along this coordinate (5%, terpinoids). Contours: 1 = -0.27; 7 = 0.28.

8. Note the strong di\ergencc between populations 27 and 28 (8%, moiphological ) Con-
tours: 1 = -0.22; 7 = 0.27.

192

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Tahi.k 2. Principal coordinate analysis of similarity matrices using terpenoids, morphol-

ogy

tions of ;. ashai. All eigenroots were extracted from each matrix until they failed to converge.
It is thought that when eigenroots begin to level off in values, additional roots represent only
random error \ ariancc.

Teiu'en()U)s (Srs based on 59 terpenoids, 28 X 28 matrix)
71% of variation extracted by 5 roots.

Eigenroots

% variation extracted

3.56 0.66 0.36 0.29 0.25

49.6

9.2

5.1

4.0

3.6

MonpiiOLOcY (Srs ba.sed on 15 morphological characters, 28 X 28 matrix)
72% of variation removed by 7 roots.

Kigenroots

% variati(m extracted

2.40 0.64 0.53 0.36 0.34 0.30 0.26

35.9

9.5

8.0

5.4

5.1

4.6

3.9

Pkhoxujases (Srs based on 16 peroxidases, 15 X 15 matrix)

\i3% of variation extracted by 10 roots.

Eigenroots

% variation extracted

1.68
30.1

0.85
15.2

0.62 0.53 0.35 0.30 0.27 0.21 0.19 0.18

11.2

9.5

6.3

5.4

4.8

3.8

3.5

3.2

theme of a single linkage phenogram (see Adam.s, 1975a). Thus, we see that the

typ

rphol

The second coordinate extracted from the terpenoid similarity matrix largely
separates the small island populations at Post (24), Texarkana (27), and Saline
Creek (28) from the rest of /. ashei (Fig. 5). These populations, along with
25 and 26, were collected and analyzed 4 years later (1974) than the other popu-
lations (1970 collections). Therefore part of these differences may be due to
sampling methods, seasonal variations, and different gas chromatographic con-
ditions. However, populations 25 and 26 seem to cluster well with populations
sampled in 1970, so this factor may be only a minor cause of this trend. It seems
that this small amount of variation (9%) is chiefly accounted for by the diver-
gence of these 3 small, isolated populations (24, 27, 28), along with a contribution
resulting from different sampling and analysis times. The second coordinate of
the morphological similarity matrix (Fig. 6) is clearly due to the fact that popu-
lations 24, 26, 26, 27, and 28 were sampled and analyzed in 1974 rather than with
the other populations (sampled and analyzed in 1970). It is felt that most of these
differences (approximately 97c of the variation in the similarity matrix) are due
to the fact that a diffc^rent technician measured the morphological characters of
populations 24, 25, 26, 27, and 28 (1975) than the other populations (1970-
1972). Even with close supervision and training, it is very difficult to get two

people to score morpholog

My experience

has been that comparisons between morphological data sets scored by com-
pletely different research projects is almost impossible. If we consider that the
eigenroots of about 5% may be mostly random noise (see below), then the 9% of
coordinate 2 is only about twice the experimental error but 25% the size of the

major trend.

The third coordinate does not appear to be very significant in the terpenoid

1977] a\^a\{S--]unipi:rus ashki

193

similarity matrix since only 5. 17^ of the variation was extracted and the eigenroots
have leveled off at this value (Table 2). Contouring of this coordinate (Fig. 7)
shows that most of the variation along tliis axis is due to population 24 at Post. This
population is on one of the most unusual sites that I have seen for /. ashei. It
is in a deep ravine, cut into the Pennian red clay, just east of the Llano Estacado.
The stand is occasionally mixed with /. pinchotii^ with /. ashei found in the more
mesic spots. This trend could represent a response to microhabitat selection or
environmentally induced plasticity. Transplant studies will probably be needed
to answer this question. Another trend is that the northern-most (including
Post) populations seem to be more heavily loaded onto this coordinate than
those populations in the central and southwestern portion of the range.

The third coordinate of the morphological similarity matrix extracted 8.0
of the variation and mijiht be sijjnificant as the 4th throuiih 7th roots sc^em to

o""" *^^ ^'tz)

have asymptoted to about 4 or 5%. The contour map of this coordinate (Fig.
8) shows a northwest-southeast trend across the populations, somewhat like that
in Fig. 7, except there is a decided split between the Texarkana population (27)
and those to the north and west. This population (27) is almost as atypical for
/. (islwi as the one at Post, Texas (24). At population 27, /. asliei is found on a
small (few acres?) limestone outcrop that is gently sloping and very moist ( 1,143-
1,270 mm of precipitation per year). It is a mixed stand with some /. virginiana.
Whether this pattern represents some small microhabitat selections or environ-
mentally induced plasticity in the morphology must await transplant studies for
additional information.

In any case, it is obvicms that the major trend in both the terpenoids and
morphology is the differentiation of populations 25, 26, 13, 12, and 17 from the
rest of the species.

Principal coordinate analysis of the similarity matrix based on peroxidases
(Tables 2-3) yielded (piite different results. A most notable difference being
that 10 eigenroots were extracted from a 15 X 15 matrix, whereas only 5 and 7
roots accoimted for most of the definable variation in the niTich larger (28 X
28) matrices of the terpenoids and morphology. This seems to indicate that the
peroxidases are varying in many different directions, whereas the terpenoids and
morphology seem to display much more directional or concurrent variation.
Another interesting facet is that the eigenroots of the terpenoid and morphological
similarity matrices ((uickly decreased to rather constant values after 2 and 3 roots,
whereas the roots of the peroxidases seem to tail out much farther. This seems
to imply a considerable amount of independence among tlie peroxidases. Exami-
nation of the first coordinate of the peroxidase similarity matrix (Fig. 9) re\^eals
a northeast-southwest pattern (remember that only the 15 populations marked
with an asterisk were analyzed for peroxidases). The Texarkana (27) and Junc-
tion (10) populations are most similar to each other, and the Ozark populations
(1, 2) are most similar to the north Texas (5, 7) and west Texas-Mexico popula-
tions (12, 25). This trend is imlike any other seen in either the moi"phological or
teipenoid data. The dix^ergence of the Texarkana population (27) from the
Ozark (1, 2) and north Texas (5, 7) populations would be easy to explain (if
one ignores the morphological and terpenoid data) as gc^ietic drift and/or

2^94 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol.. 64

Table 3. Coordinate loadings (principal coordinates 1, 2, 3 in each case) for popula-

tions ontt) coordinates. These coordinate loadings were used for generating the contour maps
in successi\'e figures. Note the close correspondence between c(M)rdinate 1 for the terpenoids
and niorpliology. Values in parenthesis indicate the aniotnit of the variation in the Sr matrix
accounted for by each of the coordinates.

TtM-penoids \h)rphology Peroxidases

CjI K>'2 C^;i v^i Vj2 v>:i v>L v>i; K^i

Population ( 50% ) ( 97c ) ( 5% ) ( 38% ) ( ^ ) ( 8% ) ( 30% ) ( 1 5% ) ( 1 1 90

1 0.11 0.18 0.13 0.16 0.06 0.15 0.52 0.06 O.Il

2 -0.02 0.20 0.08 0.13 0.13 0.10 0.32 -0.17 0.01

3 0.17 0.12 0.03 0.06 0.21 -0.15 —

4 0.15 0.11 0.04 0.13 0.06 -0.02 -0.16 0.21 -0.14

5 0.13 0.17 0.02 0.12 0.00 0.00 0.44 0.06 0.12

6 0.23 -0.08 -0.09 0.18 0.09 0.01 _ _ _

7 0.13 0.18 0.10 0.20 -0.02 0.11 0.38 -0.04 0.02

8 0.15 0.04 -0.14 0.20 0.09 0.15 _ _ _

9 0.11 0.18 0.01 0.17 -0.05 0.11 -0.27 -0.51 -0.05

10 0.21 -0.04 -0.14 0.13 0.03 0.10 -0.50 0.20 -0.20

11 0.11 0.15 0.06 -0.14 0.27 -0.14 _ _ _

12 -0.84 -0.05 -0.03 -0.71 0.28 0.20 0.31 0.16 -0.0 1

13 _().78 -0.03 -0.03 -0.49 0.18 -0.02 -0.06 -0.06 0.04

14 0.21 -0.14 -0.20 0.20 0.06 0.04

15 0.20 -0.05 -0.10 0.26 -0.04 0.10

16 0.18 0.06 0.02 0.18 -0.05 -0.04

17 -0.67 0.02 -0.00 0.23 0.05 -0.06 -0.33 -0.34 0.10

18 0.18 0.03 -0.08 0.14 -0.05 0.05 —

19 0.10 0.16 -0.03 0.09 0.00 -0.07

20 0.22 -0.23 -0.18 0.17 -0.05 0.03 _ _ _

21 0.26 -0.21 -0.13 0.17 0.01 -0.15 -0.02 -0.11 -0.28

22 0.19 0.07 -0.07 0.13 -0.07 -0.18

23 0.18 0.08 0.07 0.14 0.03 -0.17 _ _ _

24 0.22 -0.42 0.32 -0.14 -0.34 -0.27 -0.11 0.30 -0.42

25 -0.78 -0.07 -0.07 -0.71 -0.16 -0.06 0.18 0.17 0.15

26 -0.71 -0.05 0.08 -0.72 -0.26 0.13

27 0.20 -0.26 0.11 0.16 -0.36 0.31 -0.58 0.34 0.49

28 0.16 -0.13 0.19 0.02 -0.09 -0.26 -0.14 -0.28 0.07

founder's effect, but the peroxidase similarity to the Junction population (10)
rather stretches the point.

Coordinate two of the peroxidase similarity matrix shows (Fig. 10) high

loadings of population.s 27, 4, 24, 10, and 25. This coordinate seems to be a ran-
dom assortment of populations distributed across the range of /. ashei. Similar
variation (high similarities across disjunct populations and a random mosaic pat-
tern) has been previously observed in nonsignificant variation of individual
morphological characters (sec Adams & Turner, 1970, for several contoured

morphological characters). Coordinate three .shows another pattern of mosaic
variation and the interested reader is referred to Kelley & Adams (1977b) for
more detailed maps of peroxidases, esterases, and alcohol dehydrogenases.

How can we inteipret these conflicting results? One way to view geographical
variation is to consider the number of gene differences needed to produce the
observed changes. For morphological characters, Charles & Goodwin (1943)
have shown that in Solidago many morphological characters used in taxonomy

1977]

ADAMS— JUNIPERUS ASHEI

195

COORD.1(30^/o) PEROXIDASES

COORD. 2I15V0)

PEROXIDASES

FicuRES 9-10. — 9. Contoured loadings of principal coordinate 1, extracted from tlie
F-1 weighted peroxidase similarity measures among populations (see Tables 2-3). This co-
ordinate extracted 309^ of the \'ariation from the matrix. Only those 15 populations marked
with an asterisk were analyzed for the peroxidases. See text for discussion. Contours; 1
-0.51; 7 =: 0.43. — 10. Contoured principal coordinate 2 (15^ of the variation, peroxidas(^
similarities). No regional trends were uncovered in this or any of the successive coordinates
extracted. See text for discussion. Contours: 1

0.42;

0.28.

arc controlled by a niininium of 4, 5, and 6 j^cncs. Irving ik Adams (1973) in a
study of Iledeoma terpenoids found that those terpenoids were controlled by a
niininium of 1, 2, and 3, but up to 7, genes which agrees with the work on Piiuis
by Hanover (1966) and others. The peroxidase electromorphs isolated on gels
represent probably no more than 1 gene for each 2 bands in tlie composite. Sup-
pose we assume that the 15 morphological characters are each controlled on the
average by 5 genes, the 59 terpenoid characters each are controlled by 2 genes
(average), and the 16 peroxidase bands are each controlled by 1 independent
allele, witli 2 alleles (simple codominance) per gene. This means that the pat-
tern displayed by the morphological data sampled a )nininui)ii of 75 genes^ with
a minimum sample of 118 genes for tlie terpenoids, and a maximum sample of
8 genes for the peroxidases. Of course, we have ignored pleiotropy, epistatis, and
linkage, but we have no a priori knowledge that these factors are of differential
genetic importance in any of these 3 kinds of data. To obtain a random sample
of the genome, one w^ould have to favor the morphological and terpenoid data
on the basis of sample size alone. Together the morphology and terpenoids
{mi7}i)num of 193 genes) overshadow the peroxidase data {maximum of 8 genes).

Even so it is striking that no logical regional trends emerged from the peroxidases
(nor from the esterases or alcohol dehydrogenases, Kelley & Adams, 1977b).
The problems of homology may account for much of this random similarity be-
tween widely, disjunct populations (e.g., population 10 and 27, Fig. 9). Ho-
mology between the morphology of these populations (Table 1) is practicalb'
assured. The terpenoid variation is almost totally quantitative in this taxon, and

j^Qg ANNALS OF THE MISSOURI BOTANICAL GARDEN [Voi., 64

the resolution obtained witli capillary gas chromatography greatly increases the
probability that peaks from different populations of a quantitatively varying
species are in fact the same compound (although there is a small finite prob-
ability that different genes produce the same compound in different populations).
On the other hand, the peroxidases were often found to be qualitatively varying
between close, adjacent populations with a band being in very high frequencies
in one population and totally missing from our sample in another population.
The high similarities obtained in mosaic patterns (Figs. 9-10) are most readily
explained by lack of homology between peroxidase bands, although parallel
microselection could play an important role. As far as T know, there have been
no cases showing that electrophoretic mobility, per se, is under selection (that
is not to say that proteins bearing more positive or negative charges might not
be selected due to substrate affinity, etc. ) .

Four hypotheses have been advanced (Adams & Turner, 1970) to explain
the pattern of regional variation seen in the terpenoids and moiphology of J.
(isheL Two of these, sampling errors and parallel selection (in populations 17,
12, 13, 25, 26) have been pretty w^ell disposed of by Adams (1975a). The other
twH), predominately southerly winds during pollination (December-January)
and northward bird migration during the spring, and ancestral migration leav-
ing relict populations, deserve additional discussion. The prevailing wind dur-

from the south on

J

?

the Edwards Plateau (Arbingast et ah, 1967). Thus, one might expect pollen
to be generally blown northward. Coupled w^ith the northward migration of
Cedar waxwings and other birds that feed on /, ashei berries (female cones),
this would tend to isolate population 17 from breeding with adjacent popula-
tions (9, 16, 1(S). This would also help explain the north-south line of differenti-
ation between populations 12, 13 and 11, 14 (Figs. 3-4). Although these phe-
nomena help explain the persistence of the pattern, they do little to explain the
common patterns seen in populatitms 25, 26, 12, 13, and 17. Ancestral migrations
leaving relict populations could help explain these patterns. ,

Pleistocf.xe Patterns

Although there is considerable evidence of a continuous band of sclerophyl-
lous vegetation from central Texas into northern Mexico during the Tertiary
(Axelrod, 1975), I would like to focus on events of the Pleistocene, particularly
the last pluvial and interglacial periods. I have reconstructed parts of the \x^ge-
tation during the Wisconsin pluvial, 10,000-20,000 B.P., in Fig. 11. According
to King (1973) tlu^ western Missouri Ozarks were covered with boreal spruce
forest from about 25,000 to at least 13,0(X) B.P., with pine parkland preceding
the boreal spruce. Since the pine parkland and boreal spruce forest both ap-
pear to have been pushed southward from the north (Dillon, 1956), I ha\t' as-
sumed that the area south of the Ozarks may have been pine woodland or park-
land (also see Bryant, 1969). A pine-spruce woodland seems likely in the Llano
Estacado of northwest Texas (staked plains) according to Ilafsten (1961). Bryant
(1969) suggested that based on pollen profiles, the present Chihuahuan desert
area around Del Bio, Texas (430 m) was a pinyon woodland. W'ells (1966),

1977 J

ADAMS JUNIPEIWS ASIIEI

197

HYPOTHETICAL PLEISTOCENE PLUVIAL VEGETATION 10-15,000 bp

OZARK MTNS.

BOREAL SPRUCE
FOREST

PINE- SPRUCE

FORES T

I

ARBUCKLE MTNS

DAVIS MTNS.

V,

CHISOS MTNS

EDWARDS PLATEAU

P/A/ YON - J UN i PER
WOODL AND

A

SIERRA
DEL CARMAN

A S DEL BURRO
. A

M/XED DECIDUOUS
WOODLAND WITH
CONIFERS

100

200

300

KILOMETERS

CUATRO CIENEGAS

Fi(;uRE 1 1. Ihpothctical Pk'istocciK' plu\ iai xegctation, 1U,()U0-15,()()0 B.P. basrd on
poUni profiles ami rat niicklrii data lioiii the literature. See text for diseussion.

using data obtained from rat niiddcMis from tlie Big Herid Toxas region, con-

o

eluded tliat tlie life zones deseended about 800 ni for pinyon-juniper (/. pincJiotii
in that ease), allowing the advanee of pinyon-jimiper into most of the present
desert region between Big Bend and Del Rio. Another important faet has ])een
the reeent diseoxcry (D. 11. Riskind, pers. eomm.) of /. asliei, /. pinchotii, J.
flaccida, and J, scopulorum from tlie Sierranas del Burro (Fig. 11). Sinee typi-
cal /. pinchotii has been found (Adams, 1975b) just soutli of the Sierra del Car-
man (growing with /. ashei)^ it appears that tlie Sierranas del Burro may have
been an important refugium or island point in the pinyon-juniper woodland. A
mixed deciduous woodland with conifers is postulated in central Texas ( Biyant,
1969; based on a pollen protile).

]^g§ ANNALS OF THE MISSOURI nOTANIC:AL GARDEN [Vol. 64

Table 4. SNK tests for 15 morphological characters with F ratios greater than 1.0 in
ANOVA. SNK tests were rnn at P = 0.05, E,.,*-. = 1.58, Fo.,.i = 1.90 (df = 27/380). Any
two populations not iinderlined by a common line are significantly different. Poprila-
tioiis are listed in decreasing order of their means from largest to smallest. Ranges refer
to the maxinunn and mininumi means over all populations for that character.

FEMALE CONE DIAMETER (FDI), F = 9.3, no obs. for populations 24, 27, range
(8.91-6.4 mm)

15 5 14 6 3 23 4 7 11 21 16 2 10 19 8 22 28 20 1 18 9 13 12 17 26 25

SEEDS PER FEMALE CONE (SPF), F = 9.7, no obs. for populations 24, 27, rangt
( 1.G9-1.01 )

12 26 13 25 4 17 11 28 2 19 3 10 14 16 21 1 18 8 15 23 20 5 7 6 9 22

BLOOM ON FEMALE CONES (BLM), F = 1.2, no .significant differences

SEED AREA (SEA), F = 9.8, no obs. for populations 24, 27, range (27.1-13.6 mm')

20 16 21 14 5 15 4 23 6 22 8 7 10 3 9 19 2 28 11 18 1 17 13 12 25 26

SEED WIDTII/LENCTH (SER), F = 1.3, no significant differences
WHIP CLAND AREA (WGA), F 9.2, range (0.93-0.31 mnr)

12 25 11 13 3 26 23 17 5 6 28 14 7 20 16 21 4 18 1 22 24 15 19 10 8 2 9 27

WHIP LEAF GLAND LENGTH/WIDTH (WGR), F = 4.4, range (2.5-1.5)
13 25 17 26 5 12 114 1 19 8 2 7 14 3 27 22 16 18 20 6 9 10 23 15 28 24 21

WHIP LEAF MARGINS (WLM), F = 5.0, range (2.3-1.9)

24 28 25 22 23 26 16 13 21 27 5 17 9 6 19 18 20 14 4 3 15 10 7 1 12 8 11

WHIP LEAF GLANDS PROTRUSION (WGP), F = 2.6, range (3.00-2.87)
13 6 7 10 11 15 19 20 21 23 25 18 12 17 22 28 16 8 9 14 2 24 13 4 5 26 27

WHIP LEAF GLANDS RUPTURED (WRP), F = 2.7, range (1.08-1.00)
13 12 11 14 4 6 7 8 9 10 2 3 1 5 15 16 17 18 19 20 21 22 23 24 25 26 27 28

WHIP LEAF BLADE LENGTH/SHEATH LENGTH (B/S), F - 2.9, range (0.77-0.54)
24 25 2 26 9 12 11 1 27 28 14 15 16 10 6 5 13 19 22 20 21 4 7 3 8 18 23 17

WHIP LEAF GLAND LENGTH/SHEATH LENGTH (G/S), F = 16.7, range (0.41-0.22)
25 26 12 13 17 11 24 3 28 19 22 21 23 16 4 5 18 20 9 6 14 2 10 8 1 7 27 15

SCALE LEAF LENGTH (SLL), F = 4.1, range (1.74-1.43 mm)

28 27 24 25 19 7 18 16 1 22 15 26 14 21 20 4 5 23 11 13 9 17 12 3 2 8 6 10

SCALE LEAF LENGTH BRANCH WIDTH (L/B), F = 4.1, range (1.43-1.15)

9 27 17 15 19 18 5 16 7 20 1 14 8 22 23 4 21 25 28 11 3 2 13 6 10 24 12 26

BRANCHING ANC-LE (BAN), F i= 9.5, range (55.2-39.9 degrees)

12 26 25 10 24 13 27 20 7 17 5 18 11 28 19 16 9 4 23 8 15 6 14 22 21 1 2 3

19V7] ADAMS JUMFKRUS ASIIEI

199

Life zones were puslied southward and compressed during tlie Wisconsin
pluvial (Dillon, 1956), but how far they were extended into Mexico is not well
known. Additional rat midden and pollen profiles are needed in northern Mexico
and the Mexican plateau region. A study by Meyer (1973) in the Cuatro Ciene-
gas basin (Fig. 11) revealed no changes in the pollen profiles during the past
30,000 years. lie concluded that there was no evidence for pluvial nor hypsi-
thermal (Deevey & Fhnt, 1957) periods at Cuatro Cienegas during the time se-
quence studied. This agrees with Dillon (1956: 174) who shows a considerable
compression of life zones from Nebraska to south Texas but few differences past
northern Mexico. It seems likely that any generalized Mexican refugium must
have been in northern Mexico. One other point that seems relevant is that Wells
( 1966) mentioned that the pinyon found in the rat middens in the Big Bend area
contained consistently 2-needled fasicles suggesting that the pine involved may
not have been the predominately 3-needled Pinus cemhroides Zucc, but perhaps
Pinus cemhroides var. reniota E. L. Little. Pinus cemJ)roides var. remota now
persists on the Baleones escaipment of the Edwards Plateau (near population
14, Fig. 1) about 300 km to the east of the fossil site. I have recently examined
a herbarium specimen of /. ashei from eastern Brewster (bounty, Texas and ha\'e
indicated this location in Fig. 1 (dashed lined population, about 150 km west
of population 26). This new population is just north of Wells's (1966) Maravib
las Canyon rat midden site. Perhaf)s his juniper twigs should be reexanu'ned for
the presence of /. ashei. In any case, this western-most disjunct population of
/. ashei seems to be of the same relict nature (on preliminary morphological
examination) as populations 12, 13, 25, and 26.

Although I have previously considered populations 12, 13, 25, 26, and 17 to
be ancestral (Adams, 1975a), one might ask why these might not be advanced,
with the central Texas-Ozark populations being ancestral. Examination of Tal:)le
4 reveals the significant morphological differences between populations 12, 13,
17, 25, 26, and the other populations. The following characters show significant
differences: female cone diameter (smaller in 25, 26, etc.); seeds per cone
(generally more in 25, 26, etc.); seed area (smaller in 25, 26, etc.); w^hip leaf

gland area (larger in 25, 26, etc.); whip leaf gland length/width (more elongated
in 25, 26, etc.); whip leaf gland length/sheath length (larger in 25, 26, etc.);
and branching angle (larger in 25, 26, etc.). Reviewing the Sahina section of
Juniperus in North America, it seems that some of these character states are
rather unusual and are likely ad\'aneed (ratlier than primitix^e). Advanced charac-
ter states ( central Texas-Ozarks ) are : larger f(Mnale cones concurrent with
fewer seeds (just the opposite found in most of the junipers); whip leaf gland
area small (whip leaf gland area is generalh' large in jimipers where the glands
are visible); whip leaf gland length/width close to 1 or 1.5 (/. ashei is unique
in the genus, so far as is known, in having raised, round glands), the more elon-
gated glands (populations 25, 26, etc.,) are definitely the more primitive type;
and whip leaf gland length/sheath length (almost always large in Jtiniperus,
except the central Texas-Ozark /. asJiei). Ad\anced and primitive states are not
known for two characters: seed anni and blanching angle. Overall, the characters
expressed in central Texas and the 0/arks are generally unusual in occurrence

200

ANNALS or THE MISSOURI BOTANICAL GARDEN

[Vol. 64

7654 3

7 65 4 3 3

CHEM. SR, ASHEI-SALTILLENSIS MORPH.SR, ASHEI-SALTILLENSIS

Figures 12-13. — 12. The contoured F-I weighted inorpliological similarity of each popu-
lation of 7. ashci to /. saltillen,sis, collected near Saltillo, Mexico. Junipcrus saltillcnsis is
tliought to he closely related to the ancestral stock of /. av/ir/. Notice the clinal differentia-

tion from west Texas to the Mexico population (25). Contours: 1

0.16; 7

O.lfl— 13.

The contoured F-1 weighted terpenoid similarity of each population of /. ashei to /. saltillcn-
sis. The clinal trend seen with the morphology (Fig. 12) is steeper in the terpenoids, and

population iV is obviously more closely related to the ancestral stock of /. .saltillemis-J. ashci,
Gmtours: 1 — 0.17; 7 — 0.41.

in Junipcrus compared to the character states found in the southwest Texas-
Mexico populations. Further evidence regarding the ancestral nature of popu-
lations can be obtained by comparison of each population of /. ashei with its
presumed nearest ancestor (Zanoni & Adams, 1976), /. saltillensis Hall. Although
/. ashei probably did not descent from /. saltillensis, that taxon appears to bear
the closest morphological and teipenoid similarities to J, ashei of any in North
America. In Fig, 11 I have constructed differential similarities of each popula-
tion of /. ashei to a sample of 15 trees of /. saltillensis from near Saltillo, Mexico.
ANOVA was performed on 29 data sets (28 /. ashei populations and 1 /. saltiUen-
sis population) to determine a set of F-1 weights. Similarity measures were cal-
culated as outlined before, then each population of /. ashei was contour mapped

showing the change (differential) in similarity to /. saltillensis (the geographical
source of this taxon is not important for obtaining the similarities and is not
shown on the maps). This method of "differential similarity" should prove very
useful in the analysis of the interaction of two species across a geographical area.
Figure 12 is based on 15 morphological characters (female cone color omitted,
F — 0.88), F-1 weighted. Notice that the highest similarity to /. saltillensis is
from the Mexico population (25), followed by populations 26, 12, 13, and 17.
The knife edge break previously seen (Fig. 4) between populations 12, 13 and
11, 14 is quite widened in this analysis with a cline from populations 12 to 10.
The Post population (24) bears some similarity, but part of this similarity may
be due to environmental factors.

The terpenoids of /. ashei are interesting evolutionarily because there is a

19771 AnAWS— JUMP i.nvs AS///:/

201

greater shift toward the prcxlomiuance of a single e(;iiipouiicl (eaniphor, see \()ii
Rudloff, 1968; Adams & Turner, 1970) than in any other member of tlie genus.
In populations of central Texas camplior averages about 757^^ of the total oil (2
hr. extraction) whereas the di\'ergent populations ax'erage about 60/^. Juiiipcms
ashci has by far the simplest oil mixture of the North American jmupers, and
this seems to be an advanced character state of specialization. Tlie central Texas
populations are particularly low in the sesquiterpene oxygenated compounds
such as elemol, elemol-acetate, and a, ^ and y-endesmols. Larger (|uantitic\s of
these compoimds are the rule in tlie rest of the junipers and conifers in gcM»eral
(see von Rudloff, 1975).

Differential similarities, based on ANOVA (28 /. ashci populations plus 1
/. saltiUcnsis population) and using 68 terpenoids F-1 weighted, reveal (Fig.
13) a pattern almost identical to the differential similarities for the morphologi-
cal characters (Fig. 12). These similariti(\s indicate that the divergent populations
(25, 26, 12, 13, 17) bear a stronger affinity to /. saltillensis than the central
Texas-Ozark populations (lest the reader be suspicious ot mixed sampling in
population 25, etc., I should note that these di\'ergent populations clustered
strongly wath the central Texas type when an OTU of /. saltiUcnsis was added
to the matrix set, and intrapopulational cluster analysis of each of the 28 popu-
lations of /. asJici revealed no other taxa as would be the ease in mix(xl species
samples). Thus we see that in considering a fairly large set of characters (15
morphological and 68 terpenoids), the dominant theme is for the divergent popu-
lations to be progressively more similar to /. saltiUcnsis. It should be noted that
/. saUiUcnsis is not conspecific with /. ashci (Zanom* & Adams, 1975, 1976). In
fact, several characters found in /. saltilloisis (curved terminal wliips and beady
scale leaves) have not been found, even in the relict populations, in /. ashci. Al-
though relict hybridization could not be concliisi\x4y ruled out at present, it
seems unlikely since we ha\'e no direct evidence that the t\\'o taxa haw been
sympatric, and several distinguishing characters of /. saUillcnsis have not been
found in divergent J. ashci plants. It would appear that the most probable hy-
pothesis at present is that /. asJici and /. saUiUensis had a common ancestor (Ter-
tiary?) in the Sierra Madre Oriental. Junipcriis ashci differentiated and migrated
northeastward to the exposed limestone outcrops ( Fdwards Plateau, Ajbuckles,
Ozarks, etc.), whiU^ /. saltiUcnsis adapted to the drier, interior portion of the

Sierra Madre Oriental.

During the Pleistocene ice adx^ances, /. ashci may ha\'e become ivxtinct in
Missouri, Arkansas, Oklahoma, and most of central Texas as depicted in I'ig. 14.
During the same period, /. ashci probably expanded westward into the cinrent
Chihuahuan desert (Wells, 1966; Bryant, 1969), but not as far south as Cuatro
Cienegas (Meyer, 1973). Migration west of the Sierra del Carman was also pos-
sible since the species is cmrently found at the top of a pass (La Cuesta) just
south of the Sierra del Carman. Whether /. ashci could have crossed the high
plateau around Alpine and Marfa (1,500 m) is not known, but suitable liabitat
was probably available for colonization in the Presidio area. With this model,
populations of J. ashci would be forced to extinction in central Texas, Oklahoma,
Arkansas, and Missouri. The subsequent recolonization could then take place

202

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol.. 64

WISCONSIN DISTRIBUTION OF J. ASH El 10-15,000 bp

OZARK MTNS

A A><

DAVIS MTNS
CHlSOS MTNS

A

SIERRA
DEL CAR

ARBUCKLE MTNS

I

I

LOCJiL EXTINCTIONS
OF J. A SHE I

^^^

EDWARDS PLATEAU

BA L CONES

ESCARPMENT
ISURVIVAL (RELICT,

RETREAT TO PINYON
P EDULIS REMOTA ?h

Wf^^^UNlP£f^(^ ASHEI, PINCHOTII )

WOODLAND

too 200

300

KILOMETERS

CUATRO CIENEGAS

= REMNANT, ADAPTED TO MORE MESIC ENVIRONMENT

= ANCESTRAL TYPE POPULATIONS. LOW CAMPHOR TYPE

FifamE 14. Possible Wisconsin distribution of /. uf^hex, 1(),000-15,(M)() B.P. Following
the advancv of suhalpiiu' and montane species (Fig. 11), /. as\\ei popnlatiotis may have gone
extinct north of the Edwards Plateau. See text for discussion.

according to Fig. 15 over a very short period of time (hundreds of years?) from
some population in central Texas that may have gone through a selection "bot-
tleneck/' perliaps coupled with genetic drift. This "relict" population would have
had considerably more cainphor in the oil (as a plant defense?), more roundish

glands, larger female cones, fewer seeds (therefore a higher pulp to seed ratio
for bird dispersal), and a more lax foliage (smaller branching angle) which seems
to be associated with more mesic species. The rapid rccolonization of limestone
outcrops (Fig. 15) could then lead to a uniform taxon from central Texas through

1977]

ADAMS -;UA7/7:/U'S ASIIKI

203

POST GLACIAL MIGRATION AND DISTRIBUTION OF ^. ASHEI

DAVIS MTNS.

SIERRA
DEL CARMAN

CUATRO CIENEGAS

= REMNANT, ADAPTED TO MORE MESIC ENVIRONMENT DURING
THE PLUVAL PERIOD

= PRESENT DISTRIBUTION OF RELICT POPULATIONS

Fic;uuK 15. Possible post-glacial iniiiiation to attain tlir pifsciit distrihution of /. a.sJu'i.
Thv remnant (hi^h camphor type) adapted to a more mesic en\ ironment may have <iiiickly
expanded during the liypsithermal to reacli the present distribution (see Fig. 1). The dashed
line shows the phnial distribution of the ancestral t\pc (lower camplior) i^opulations (see
Fig. 14).

the Ozarks. Althougli this would explain tlie ()l)ser\'ed patterns, many nneer-
tainties remain. For instance, Dillon (1957) argues that tlie boreal forest ele-
ments were merely mixed with the present floral components in the southern
stat(\s. Graham (1973) feels that most central-southern conmiunities incorpor-
ated boreal elements (e.g., spruce) but retained the general cliaracter of the
original vegetation. If small pockets of /. ashei did persist during the full glacial,

204

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

5 4 J 2 2 3

AVG. SIMILARITY, TERPENOIDS

AVG. SIMILARITY, MORPHOLOGY

100 200

I 1 Jv

KILOMETERSX

CPV, TERPENOIDS

CPV, MORPHOLOGY

AVG. SIMILARITY, PEROXIDASES

CPV, PEROXIDASES

1977] ADAMS— JUNIPKRUS ASIIKI 205

one might find soniC evidence of this l);ised on intrapopulation variability, with
the smaller Pleistocene relictnal popnlations having less variability than larger
(sontli-central Texas) populations.

I\T1L\P()PULAT1()\AL \^\IUAlULlTy

Mean similarity (Sr) within each population (15 trees) for 152 terpenoid
characters (W = 1) shows high average similarities in the Ozarks and central
Texas (10, 11, 15) and low similarities at New lirannfels (17), Post (24), and
Mexico (25). The divergent popnlations (12, 13, 17, 25, 26) tend to be a little
less uniform, although Post (24) is also (juite variable. Populations that showed
the major trend of the terpenoids (Fig. 3) tended to be um'form. Examination
ol the homogeneity of the similarities was accomplished by computing the stan-
dard deviation of the mean sinu'larity and dixiding by the mean similarity of
that population for normalization (CPV). The most liomogeneous similarities are
in the O/arks (2, 3) and central Texas (10, 15, 7), whereas the least homogeneous
are New Hraunfels (17), lirady (20), and Post (24). The populations which
sliowed tlie highest similarities are the most honiogcMieons except for population
20 (Brady). The low similarities and lack of homogeneity at New Hraunfels
seems to be due to the interaction l)et\\een relict and modern genotxpc^s. One
might (juestiou if the population at Post (24) is hybridizing with sympatric /.
pincholii treses l)ut notice the close ordination of 24 with the central ^\^xas /.
ashci (Fig. 3). Exanu'uation ot the intrapopulation pluMiogram revealed no major
groups within any population.

Analysis of 16 morphological characters (W = 1) shows the highest similari-
ties in central Texas (7, 10, 18) and the Ozarks (1, 3), with lowest similarities
in the relict populations (12, 13, 26, 17, 25) and at Texarkana (27). Two small
island populations (27, 28) both show considerably lower similaritic^s in their
morphologx^ than the\' did in their terpenoids, whereas Post (24) is more medial
in its morphological similarities than with the terpenoids (Fig. 16). The CP\'

FicunKs 16-21. — 16. A\cM'agt' siniilarit) (Sr) within catli popiilatioii (15 trrrs) l)asrd
oil 152 equally wcM'^liti'd terpenoids. Most populations liad high internal similarities \\ ith
the exception of the ancestral populations (12, 13, 17, 25, 26) and the Post population (24).
Contours: 1 =i 0.78; 7 ^ 0.87. — 17. Contoured coefficient of phenetie variation (C'P\') ol
the terpenoid siunlarities. In g(^neral the populations witli hi^h intrapopulation similarities
\\ ere lionioj^eueous ( low CP\'s ) and \ ice versa, except for population 20 w Inch Itad lu,uh
similarities and was not so h()mogen(M)us (liij^h (^P\^). Contours: 1 ^ 0.30; 7 ^ 0.^1. —
18. Average sinu'larity within each population based on 16 eciually wi'iglited morpholot^ical
characters. Note that the ancestral populations are of generalK lowx^r internal similarities com-
pared to high similarities throughout central Texas. Tlie small populations at Texarkana and
northeastern Oklahoma are nu)rpliologically (juite variable, ('ontours: 1 = 0.85; 7 = 0.91.
— 19. C'ontoured coefficient of phenetie variation (CP\') of the average morpliological
similarities. The CP\' s(hmus higliK negatively correlated with the mean Sr except for x:)opn-
lations 19 and 20 which are not \ery liomogtMieous. Contonrs: 1 ^ 0.28; 7 — 0.60. — 20.
Contoured average sinnlarities of 23 e(iually weighted peroxidases. The Junction j^ioi^nlation
( 10) had the lov\est similarities along with Post (24) and the Arbuckles (4). (Contours:
1 = 0.54; 7 = 0.96. — 21. Contoured coefficient of phenetie variation of the peroxidase sinu*-
larities gives an almost idcmtical pattern as seen in the av(Mage sinnlarities (Fig. 20). Con-
tours: 1 - 0.13; 7 = 0.32.

906 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

gives a fairly similar pattern of homogeneity except for populations 19 and 20,
which, although very typical (Fig. 4) and of high average similarities, are not
very homogeneous. This same phenomenon was seen with the terpenoids (Figs.
3, 16-17) for populations 19 and 20. The Post (24) population is somewhat more
homogeneous in its morphology than its terpenoid's similarity. With the excep-
tion of populations 24 and 27, one notices that for each of these four statistics,
the populations generally present some trend of variability which is correlated
with either regional differentiation or proximity of one population to another

( the case for 19 and 20 ) .

All 23 peroxidase electromorphs were subjected to the computaticm of aver-
age similarities within population and CPVs, as with the morphology and ter-
penoids. Only the 15 populations marked with an asterisk in Figs. 20 and 21
have isoperoxidases analyzed. The average similarities within populations for
these 23 isozymes show the Ozark population (Fig. 20) to be quite similar
(0.97-O.SO), while the Junction, Texas population (10) has the lowest average
similarity (0.47). A surprising aspect of these average similarities is the low^
average similarity found in population 10 (Junction, Texas). It is interesting
that 3 different peripheral populations (1, 27, 24) show the whole range of vari-
ation from little to large amounts to intermediate variability. The CPV (Fig.
21) of these 23 peroxidases reveals that those p{)pulations that are highly similar
are generally most homogeneous and vice versa. A combined total of 43 iso-
enzymes has been analyzed by Kelley & Adams (1977b), and the results are
comparable to those shown in Figs. 20 and 21. However, the addition of 14 al-
cohol dehydrogenases and 4 esterases to the analysis seemed to have produced
a slightly less mosaic pattern in central Texas.

The pattern obtained from the isoenzymes is quite different from either the
morphology or terpenoids for in both of those analyses, population 10 appeared
to be (juite uniform and homogeneous and the relict populations consistently
displayed high to medium variability. It seems apparent that whatever vari-
ability the peroxidases are indicating, it is not directly related to variability in
the morphology nor terpenoids. Of course, it is possible that the variation seen
in peroxidases is below the le\'el of selection and merely represents "random
noise." Until more information is gathered on the selection value of various
electromorphs, we can only speculate.

The patterns of variability seem to give us a few clues as to whether the dis-
junct populations are of recent origin or relicts of the advanced high camphor
types. However, presently it is difficult to make generalizations about popula-
tional variability versus founder's effect, "bottlenecks," relictness, etc. iiince
different character sets give somewhat (to vastly) different answers, and one
could get the same obserx^ed pattern depending on time, microselection inten-
sity, or site variability.

Conclusion

Treating peroxidase data as qualitative taxonomic chemical characters did
not appear to be feasible. This is likely due to the lack of homology, intense
microhabitat selection or random (neutral) variations in the electromorplis.

1977] ADANrS— Jl/.V//'f:KC^S ASHEI

207

The use of these j)t*ro\idase electromorphs for the analysis of mtrapopiilatioii
variability could not ])e readily evaluated due to the mosaic pattern produced.
It appears that eheinosysteniatists will need more detailed l)i()chemical informa-
tion about the nature of isoenzymes, and their genetic control in ta\a to be studied.
The most probable center of origin for the modern (high camphor) popula-
tions of/. asJiei seems to be in central Texas, perhaps near Brady (20) or Iku--
net (19). These populations showed considerable variability (high CP\'s),
yet these populations are quite similar to the rest of the modern /. ashei popu-
lations. Northward migrations of birds during the spring carrying juniper seeds
could hax'e (re) colonized limestone outcrops in Arkansas, Oklahoma, and Mis-
souri in a span of a few hundred years. This could lead to the highly uniform
pattern observed in the moiphology and terpenoids from central Texas to the
Ozarks. Predominately southerly winds during pollination may have been iin-
portant in maintaining the north-south split in west Texas as well as the rehct
population at New Braunfels, Texas (17). However, it is possible that the popu-
lations persisted throughout the pluvial periods and failed to diverge due to
either a lack of variability, the relatively short time span involved, or intense se-
lection for the modern phenotx'pe.

Whether the modern populations of this taxon invaded the limestone outcrops
in the Tertiary or during the Pleistocene will probably not be known until some
pollen or macrofossil (rat midden) data has been analyze in the disjunct popu-
lations of Arkansas, Missouri, and Oklahoma. The differential similarity of /.
(isliei population to /. saltillensis from Saltillo, Mexico shows a clear trend of past

(Pleistocene) migration from northern Mexico. The northern Mexico Sierra

Madre Oriental seems a likely site for tlie origin of both /. ashei and /. sallil-

lensi.s, perhaps from a conunon ancestor.

This study presents additional evidence that selection may be more impor-
tant than gene flow (Ehrlich & Raven, 1969) in the maintenance of species. In

/. a^hei we have found that populations \\'ith disjunctions of 200-300 km, and
a trivial chance for gene exchange, were very similar to other populations cover-
ing 1,000 km of range (cf. Ozarks and central Texas populations). Yet popula-
tions which are in close (almost continuous) proximity have maintained either
ancestral or modern patterns in spite of potentially large amounts of gene flow.
(New Braunfels and populations to the north and west, and the relict/'modern
populations of west Texas) .

LriKHATuiu; Catk])

Adams, R. P. 1970. Seasonal variation of tcrp(Mioi(l constituents in natural populations of

Junipi'tus pinchotii Sudw. Pliytoclieinistr\' 9: 397-402.
— -, 1972. NnnuMical analwsis of some eonnnon errors in elienios\ stenuities. Brittonia

24: 9-21.

. 1975a. Gene flow versus selection pressure and ancestral differentiation in the

composition of species; analysis of popnlatioit ^■ariati()n of Jtoiipcnis ashei Buch. using ter-
penoid data. J. Molec. Evol. 5: 177-1(S5.
— . 1975b. Nnmerical-ehemosystematic studies of infraspceific variation in Juniperus

pinchotii Sudw. Bioeheni. Syst. Ecol. .3: 71-74.

1975e. Statistical cliaracter wciglitini^ and similarity stability. J^rittonia 27: 305

316.

208

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol, 64

— & B. L. TuHNEU. 1970. Chemosysteinatic and nunirrical studies of natural popula-
tions oi JiDiipcrus (ishci Buch. Taxoa 19: 728-751.
Alston, U. E. 6t B. L. Ttoxkh. 1903. BiocluMnical Systeniatics. Prentice-Hall, Inc., Engl*-

wood ("liffs. Now Jerso}'.
Ahuincas'i, S. a., L, (.. Kkxnameh & M. E. Bonixe. 1967. Atlas of Ttxas. Bureau of Busi-

lu'ss Research., Univ. of Texas Press, Airstin.

AxKi.Hon, D. I. 1975. Evolution and biogeography of Madrean-Tethyan sclerophyll vege-
tation. Ann. Missouri Bot. Card. 62: 280-334.

BiiVANT, \\ M., Ju. 1969, Late full-glacial and post-glacial pollen analysis of Texas sedi-
ments. Ph.D. thesis, Univ. of Texas, Austin. 168 pp.

ChiAiu.Ks, D. H. ^ H. 11. (iooDwix. 1943. An estimate of the uiininiuni number of genes dif-
ferentiating two species of golden-rod with respect to their morphological characters.
Anier. Naturalist 77: 53-69.

Dekvky, K. S. & R. F. Flint. 1957. Postglacial Iiypsithcrmal inter\al. Science 125: 182-

181.
Dlnna, 1). W. & M. B. Alexandkr. 1975. The isoperoxidases in Cucurhita pcpo L. In

C. L. Nhukert ((nlitor), Isoenzymes. Vol. 2: S.5 1-864. Academic Press, New York.

Dillon, L. S. 1956. Wisconsin climate and life zones in North America. Science 123:
167-176.

EiiULicii, P. ^ P. II. Raven. 1969. Differentiation of populations. Science 165: 1228-1332.

Flake, I^ II., E. von Ruuloff & B. L. Tuuneil 1969. Quantitative study of clinal vari-
ation in Jttiiipcrus viri^iiiiana using teri^euoid data. Proc. Natl. Acad. U.S.A. 62: 487-

494.

, & . 1973. Confirmation of a clinal patttnu of chenn'cal differenti-
ation in Jnui})('nts vir^iniana from terpenoid data obtained in successive years. In V. C.
Runcckles & T. j. Mabry (editors), Terpenoids; Structure, Biogenesis and Distribution.
Recent Ad\ances in Phytochcmistry. \'ol. 6: 215-228. Academic Press, New York.

(^iLMAiniN, A. J. 1969a. Tlie quantification of some plant-taxa circumscriptions. Amer. J.

Bot. 56: 654-663.
. 19691), Numerical phenetic samples of taxcmonn'c circumscriptions in the Bromelia-

ccae, Taxou IS: 378-392.

— , 1974. Variation within j^opulations and classification. Taxon 23: 523-536.

. 1976. Effect of changes in character-sets upon within-group phenetic distance.

Syst. Zool. 25: 129-136.
CowKH, J. C. 1966. Some distance properties of latent root and \ector methods used in

nmltixariate analysis. Biouictrika 53: 325-338.

— . 1967. Mnlti\ariate analyses and multidimensional geometry. Statistician 17: 13-28.
GuAMAM, A. 1973. History of the arborescent temperate element in the northern Latin

Auierican biota. Pp. 301-314, ;n A. Craham (editor), \'egetation and Vegetatitmal His-
tory of Northern Latin America. Els\ier Scientific Publishing Co., New York.

IlAKsrEN, U. 1961. Paleoecolog}' of the Llano Estacado. Mus. New Mexico Press, Santa
Fe, New Me\ic(».

1 Ianovku, J. W. 1966. Genetics of tcrpcnes 1. (knie control of monoterpene levels in
Piuiis nionticola Dougl. Heredity 21: 73-84.

luvixc, R. & R. P. Adams. 1973. Genetic and biosyndietic relationship of mtmoterix'ues.
hi \. C. Runeckles & T. J. Mabry (editors), Terpenoids: Structure, Biogenesis and
Distribution. Recent Advances in Ph\ tochemistrv. Vol. 6: Aead(Mnic Press, New York.

Kr.LLLY, \\. A. 1976. Isoenzyme studies in Junipcrns. PIlD. thesis, Colorado State Univ.,
Ft. Collins, Colorado.

& R. P. Auams. 1977a. Preparation of extracts froui juniper leaves for electropliore-

sis. Ph>tochemistr)- 16: 513-516.

— ^ -. 19771). Variation in isozvines from natural populations of Jnuij)enis, Rho-

dora (in i^r(\ss).

Kint;, J. E. 1973. Late Pleistocene palynologv' and biogeograi^hy of the western Missouri
Ozarks. lu'ol. Monogr. 43: 539-565.

McC^Li'UK, T. \V. & R. E. Alston. 1966. A chemotaxonoTuic studv of Lcninaccae. Amer. J.
Bot. 53: 819-859.

Mkhhincku, p. J., Jh., J. E. KiNf: & E. H. Lindsay. 1970. A record of Wisconsin-age \ cge-
tatiou atid faima from the O/arks of western Missouri. Pp. 173-184, in W. Dort, Jr. &
J. Kuox Jones, Jr. (editors), Pleistocene and Recent Environments of the Central Great
Plains, l^niv . of Kansas Press, Lawrence, Kansas.

1977] ADAMS^JUNIPEHVS ASHEI

209

Mkyer, E. R. 1973. Late quaternary paleoecology of the Cuatro Cienegas l)asin, Coahuila,

Mexico. Ecology 54: 982-995.
MooNEY, H. A. & W. A. Embodev. 1968. The relationslup of terpene composition, moiphol-

ogv, and distribution of populations of Bursera microphylla ( Burseraceae ) . Brittonia

20; 44-51.
Smith, D. M. & D. A. Levix. 1963. A chromatographic study of reticulate evolution in the

Appalachian Aspleniuni complex. Amer. J. Hot. 50: 952-958.
Smith, R. H., R. L. Peloquix & P. C. Passof. 1969. Local and regional variation in the

monoterpenes of ponderosa pine wood oleoresin. U.S.D.A. Forest Serv. Res. Pap. PSW-

56: 1-10.
Van- Devener, T. & J. E. Kixc. 1971. Late Pleistocene records in western Arizona. J.

Arizona Acad. Sci. 6: 240-244.
vox RuDLOFF, E. 1968. Gas-liquid chromatography of terpenes. Part XVI. The volatile

oil of the leaves of Jtmiperus ashei Buchholz. Canad. J. Cheni. 46: 679-683.
. 1975. Volatile leaf oil analysis in chemosystematic studies of North American coni-

fers. Biochem. Syst. Ecol. 2: 131-168.

Wells, P. V. 1965. Scarp woodlands, transported grasslands, soils, and concept of grass-
land in the Great Plains Region. Science 148: 246-249.

• . 1966. Late pleistocene vegetation and degree of pluvial cUmatic change in tlie

Ghihuahuan Desert. Science 153: 970-975.

— . 1970. Postglacial vegetational history of the Great Plains. Science 167: 1574

1582.

& R. Beiw;er. 1967. Late pleistocene history of coniferous woodland in the Mo-
have Desert. Science 155: 1640-1647.

Whitehead, O. R. 1972. Approaches to disjunct populations: the contribution of paly-

nology. Ann. Missouri Bot. Card. 59: 125-137.

Williams, W. T., M. B. Dale & G. N. Laxce. 1971. Two outstanding ordination prob-
lems. Austral J. Bot. 19: 251-258.

Whk;ht, II. E., Jii. 1970. Vegetational history of the Central Plains. Pp. 157-172, in
W. Dort, Jr. & J. Knox Jones, Jr. (editors), Pleistocene and Recent Environments of the
Central Great Plains. Univ. of Kansas Press, Lawrence, Kansas.

Zan'ONT, T. a. & R. P. Adams. 1975. The genus Juniperus (Cupressaceae) in Mexico and
Guatemala: Numerical and morphological analysis. Bol. Soc. Bot. Mexico 35: 69-92.

& . 1976. The genus Juniperus (Ciipressaceae) in Mexico and Guatemala:

numerical and chemosystematic analysis. Biochem. S>st. Ecol. (in press).
Zavarin, E. & K. SxAjHEUK. 1973. Variability of wood monoterpenoids from Finns aristata.

Biochem. Syst. Ecol. 1: 39-44.

THE ORDER CENTROSPERMAE'

Tom J. Mabhy-

Abstract

Perliaps no other order of flowering plants of its size is as well investigated morphologi-
cally, nltrastructurally, and chemically as is the Centrospennae. Tlie betalain pigment dis-
coveries of the early 1960s were followed by the moie recent discovery of nni^ine protein de-
positions in the sieve-element plastids of members of this order. These and otlier molecular
data, including DNA-RNA hybridization results, have permitted a circumscription of the order
based on 11 core families, including all 9 betalain families: Aizoaceae, Amaranthaceae, Basel-
laceae, Cactaceae, Chenopodiaceae, Didiereaceae, Nyctaginaceae, Phytolaccaceae, and Por-
tulacaceae, as well as two anthocyanin families: Caryophyllaceae and Molluginaceae. Several
smaller betalain taxa (including Giaekia, lialoplnjtum^ Hectorella, and Dtjsphania) which are
sometimes treated as independent families or as members of one of these 11 core families also
clearly belong to the order. Other families such as the Bataceae, G>'rostemonaceae, Vivaniaceae,
and Thchgonaceae are excluded from the Centrospermae. The betalain evolutionary line of cen-
trospermous families may have originated from a centrospermous ancestor which lost the ability
to produce anthocyanins and then subsequently gained the two or three steps required to
produce betalains. Pollen morphology of centrospermous taxa and the widespread occiirrence
of Cj photosynthesis in the Centrospermae are also discussed.

Since all the review papers from a symposium on the "Evolution of Centro-
spermous Families/' presented in July, 1975 during the Xllth International Bo-
tanical Congress, Leningrad, USSR, have now been published (Mabry & Behnkc,
1976a), a summary of our current views of the Centrospemiac (or Caryophyllales)
will suffice in this review. This account will emphasize the way our interpre-
tations of the order have been shaped by molecular data.

Since 1876 when Eichler (see Table 1) introduced the name for tlie order,
the Centrospermae have always contained a core of about 8-12 famihes. Eichler
(1876 and, in part, 1878) recognized most of what we now consider to be centro-
spermous families including, for example, the Cactaceae (in the 1876 treatment).
In the 100 years following Eichler's work, most systematists also included those
families now generally recognized on the basis of molecular data as belonging
to the order but often included additional ones (compare Tables 2, 3 and 4).

The molecular data which bear upon our current treatment of the order are suiu-
marized in the following sections.

Sieve-element PLAsxms

Of all the modern approaches for investigating the Centrospermae, none has,
in my opinion, contributed more to our understanding of the circumscription of
the order than the ultrastructural investigations of the sieve-element plastids.

^ Portions of the research described here were supported by NSF Grant DEB 76-09320,
NIII Grant IID-04488, and Robert A. Welch Foundation Grant F-130. Helpful discussions
with, and in some instances, unpublished data from Profs. Martin Ettlinger, H.-D. Bchnke,
Walter Brown, H. Musso, F. Ehrendorfer, and Jerry Brand are gratefully acknowledged.

^Department of Botany, The University of Texas at Austin, Austin, Texas 78712.

Ann. Missouri Bot. Gahd. 64: 210-220. 1977.

1977]

MABRY— CENTROSPERMAE

211

Taijle 1. Centrospermae (A. W. Eichlcr, 1876).

I. Order: Oleraceae

1. Polygonaceae

2. Nyctaginaceae*

3. Chenopodiaceae*

4. Amaranthaceae*

II. Order: Caryophyllinae

5. Car\'()phyllaceae

III. Order: Opuntinae

6. Pliytolaccaceac*

7. Portulacaceae*

8. Aizoaceae*

9. Cactaceae*

(? In this Order perhaps
Begoniaceae )

* Betalain families.

The discovery by Bchnke that the nine core betalain families (see Tal)le 4) as
well as the two core antliocyanin families (the Caryophyllaceae and Molliigin-
aceae) contain ringlike inclusions composed of proteinaceous filaments of a
type (Fig. 1) not found elsewhere in the angiosperms (for current reviews, see
Benhke, 1976a and in press) established that these eleven families represent
the core centrospermous families. Behnke (in press) defined the sieve-element
plastids which are unique to the Centrospermae as belonging to the P-III sub-
type. It is my current view that the presence of the P-III subtype sieve-clement
plastids in a taxon which classical data suggest might be centrospermous (see
Eckardt, 1976 for comments on centrospermous characters) establishes that it
belongs to the Centrospermae.

C4 Photosynthesis ix the Centrospermae

It is interesting that among the dicotyledons which have the C4 photosynthetic
pathway (the Kranz syndrome), 7 of the 11 families (Table 5) and about 857^^
of the genera have been reported among the Centrospermae (Walter Brown,
private communication). It is perhaps significant that the Phytohiccaceae, which
most workers consider to be the basal family of the order, exhi1)its only the C^

Table 2. Caryopliyllidae (CroiKiuist, 1968).

I.

Order Caryophyllales
Phvtolaccaceae*

1.

2

( iiiel. Acliatocarpaceae,
Barbeiiiaeeae, Gyro-

Agdestidaceae*,

stemonaceae, Peti\eriaceae*, Stegno-

*)

spermaceae
Nyctaginaceae*

3. Didiereaceae*

4. Cactaceae*

5. Aizoaceae* (incl. Mesem1)ryantheinaccae*,
Tetragoniaceae* )

6. Molluginaceae

7. Car\^()pliyllaceae ( incl. Illecebraceae )

8. Portulacaceae*

9. Basel laceae*

10. Chenopodiaceae (incl. Dysphaniaceae*,
Halophytaceae* )

1 1 . Amaranthaceae*

n. Order Batales
1. Bataceae

III. Order Polygonales
1. Polygonaceae

IV. Order Plumbaginales
1. Plunibaginaceae

* Betalain families.

212

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Table 3. Caryopliyllalos, Polygonalcs and Plumbaginales (Takhtajan, 1973).

27.

Ordnung. Caryophyllalcs
1. Pliytolaccaceae* (incl. Achatocar-

paccae, Agdestidaccae*, Bai-
bcuiacoac, Petiveriaceae*, Steg-
nosperniaceae,* excl. ( ? ) Rhah-
(hilcndron
Gyrosteiuonaceae
Bataceae

2.
3.
4. Nyctaginaccae*

10. Basellaceae*

1 1 . Didiereaceac*

12. Halophytaceae*

13. Ilectorellaceae*

14. Caryophyllaceae (incl. Illecehra-

ceae)

15. Vivianiaceae

16. Amaranthaceae*

17. Clicnopodiaccae* ( incl. Dysphania-

5.
6.

7.

Molluginaceae ( incl. Gisokiaccae* )

Aizoaceae*

Tctragoniaceae*

,*

)

28.

ceae

Ordnung. Polygonalcs
Polygonaceae

8. C'actaceae*

9. Portulacaccac*

29. Ordnung. Plumbaginales

Plumbaginaccac (incl. Armeriaceae)

* Betalain families.

pathway. Although the C4 pathway probably represents a derived condition
as taxa of tlie order radiated into xeric and other high hght intensity habitats, it
is hkely that some form of preadaptation for the Kranz syndrome exists througli-
out the order. This preadaptation has permitted repeated and independent evo-
kition of the syndrome at various times and in numerous genera in at least seven
fc^iilies of the order (Walter Brown, private communication). The only other
dicotyledonous families exhibiting the C4 pathway are the Boraginaceae {IleJio-
tropium in part), Compositae (7 genera), Euphorbiaceae [Chamaesijce) ^ and
Zygophyllaceae (3 genera).

It should also be noted that the CAM ( crassulacean acid metabolism) photo-
synthetic pathway occurs in members of the Cactaceae and Aizoaceae; this path-
vv^ay, although anatomically, functionally and phylogenetically distinct, does uti-
lize enzymes of the Kranz syndrome. Although CAM plants are photosynthetically

Table 4. Order Centrospemiae* or Caryophyllales (modified here from Mabry, 1976).
(Taxa with P-III subtype sieve-clejnent plastids.)

Suborder Cuenopodiixeae'' ( betalaix

FA^^LIEs)

Aizoaceae (incl. Tetragoniaceae and pos-
sibly Cist^kiaceae)
Amaranthaceae
Basellaceae
Cactaceae

CluMiopodiaceae (incl. Dysphaniaceae)
Didiereaceae
Halophytaceae

Nyctaginaccae

Pliytolaccaceae (incl. Achatocarpaceae, Ag-

destidaceae, Petiveriaceae, Stegnosperm-
aceae )
Portulacaceae ( incl. Hcctorclla )

Suborder Caryophyllixeae ( axthocyaxix

families)

Caryophyllaceae
Molluginaceae ( excl. Gisekia )

" Certain families which on occasion have been treated as members of the Centrospermae but
are now known to contain neither the Centrosperniae-specific sieve-element plastids ( subtyjje P-III ) (see
Behnke, 1976a, 1970b, in press) nor betalains are excluded from the order: Polyjionaceae, Plumbasinaceae,
Fouquieraceae ( Behnke, 1976b), Frankeniaceae ( Behnke, 19761)), Rhahdodendron (Behnke. 1976b; re-
mains to be anal>zed for pigments), Vivianiaceac (Behnke & Mabry, 1977), Thelijionaceae (Mabry et al.,
1975), and Bataceae and Gyrostemonaceae. So far neither anthocyanins nor betalains have been detected
in the latter two families both of which contain j^lucosinolates (see Goldblatt et al., 1976 for recent com-
ments <m the status of these two families).

** Whether or not such betalain taxa as Pctivem and Agdcstis (Behnke et al., 1974), Halojthyfum
(Hun/.iker et al., 1974), Gisekia (Mabry, Behnke & Eifert, 1976), Dysphania (Mabry & Behnke, 1976b)
and IlectorcUa ( \Iabr>% preliminary results) sh(mld each be treated as families in the suborder Chenopodiineae
or as members of one of the core betalain families is not resolved.

1977]

MABRY— CENTROSPERMAE

213

Tahle 5. Kranz and CAM Photosynthesis" in tlie Order CcntrosixMniae*" (data from
Walter Brown, private eonnniniication, 1976).

Suborder Chenopodiineae
( Betalain Families)

Aizoaceae
Sesuvieae
Cisekiaceae
Tetragoniaceae

Amaranthaceae

Basel laeeae

Cactaceae

Chenopodiaceae

Dysphaniaceae

Nyctaginaceae
Phytolaccaceac

Achatocarpaceae

Agdestidaceae

Petiveriaceae

Stcgnospermaceac

Portulacaceae

Kranz and CAM

Kranz

Kranz

C.

Kranz (11 genera)
C.! (3 genera)
CAM

Kranz (.31 of 67 gen-
era )

Kranz (3 genera)

C.
C.

G;
C.

C.

Kranz (1 genus)

Suborder Caryophyllineae
(Anthocyanin Families)

Caryophyllaceae

Kranz

T.ychnideae

C.

Polycari)eae

Kranz

Paronychieae

C.

Diantheae

C.

Alsineae

C;;

Sperguleae

c.

MoUuginaceae

Kranz

(1 genus)

(2 genera)

» Kranz z=: C^ photosynthesis; CAM = crassulactnin at id iiu'tabnlism pathway.

*> Not yet examined: Didiereaceae, Ilectorcllaceae and llalophv-taceae. (Added in proof: Didiercaceae
and llal(>ph>taceae are C.;.)

inefficient, they are efficient at conser\ing water (see, for example, Winter,
1974) since they, unHke C^ and C4 plants, have their stomata open only at night.
In contrast, Kranz plants have probably been selected for efficient photo-
synthesis since they, unlike Qj plants, have evolved an anatomy and enzymatic
system which permits them to provide the Calvin cycle with high levels of CO-
in an oxygen-deficient atmosphere. In the presence of low oxygen concentrations
the enzyme which fixes CO2 in the Calvin cycle, ribulose diphosphate carboxylase
oxygenase, is free to function strictly as a carboxylase; therefore under these con-
ditions and with high CO:> levels and high liglit intensities, CO2 fixation is ap-
parently maximized. Thus the reasons for selection of the Kranz syndrome in
the Centrospermae (and elsewhere) may be associated in part with evolution
in ha])itats of high light intensity, which, of course, includes many of those which
are xeric. Other factors such as salinity also may be important for the selection
of C4 and CAM photosynthesis. In any case, the widespread occurrence of the
Kranz syndrome in both betalain aiul anthocyanin centrospermous families and
its sporadic and hmited occurrence in other dicotyledons supports current treat-
ments of the Centrospermae (see Table 4) and provides additional evidence
that the Centrospermae is an old, independent evolutionary line in the angio-
sperms.

Pigment Dichotomy and DNA-RNA IIybruhzation Data

FOR Centrospermous Families

Pligher plants usually contain vacuolar red and yellow pigments which
are either anthocyanins or betalains (Fig. 2); however, so far as known, the two
types of pigments never occur together in the same plant or even in specie.*)

214

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

(a)

globular crystalloid

(b)

polygonal crystalloid

(c)
no crystalloid

Cactaceae
Didiereaceae

Tetragoniaceae
Aizoaceae

MoUuginaceae

Hectorellaceae
Halophytaceae
Basellaceae
Portulacaceae

Nyctaginaceae

Caryophyllaceae

Chenopodiace
Amaranthace

Limeum

P_ b _y_ t -0- i _a _c_ c _ a c e a e

Stegnosperma

/

FiciURE L The P-III subtype sieve element plastids wliich are characteristic for the Cen-
trospenuae always contain a ring-shaped btindle of protein filaments with either globular (a)
or polygonal (b) protein crystalloids or no crystalloid at all (c). I gratefully thank Prof. H.-D.

Behnke for this figure which illustrates the distribution of these three modifications of P-III
subtype in the Centrospermae.

of the same family. Of these two types, anthocyanins are much more wide-
spread, indeed, they account for most flower pigments in higher plants (for recent
reviews of anthocyanins, see Timberlakc & Bridle, 1975; Harborne, 1967). In the
1960s it became clear that most centrospermous families contained an entirely
new class of red and yellow pigments, designated in 1966 as the "betalains"

1968; for more current reviews, see Mabry, 1973, 1976;

Although we know today that

M

Mabry, Kimler & Chi
betalains also account for some of the orange and red pigments in many mush-
rooms, notably species of Anumita (Dopp & Musso, 1973, 1974; von Ardenne et
ah, 1974), these nitrogenous pigments have not yet been reported outside the
Centrospermae among angiosperms. It is this restricted distribution of betalains
as well as their being mutually exclusive with anthocyanins that makes them
interesting as phylogenetic markers.

1977]

MAB]{Y— CENTROSPERMAE

215

Red

Yellow

RO

HO

H

HOOC

HOOC

^max ^37

2

COOH

'^max ^"77

COOH

OH

'^max ^3^

OR

A

max

A77

Future 2. The visible absorption maxima of typical red and yellow betalains (top row ),
whieh are found only in phyletieally related eentrospermoiis families and mushrooms, are
similar to those for some red and yellow anthocyanins (bottom row). Anthoeyanins account
for pigments in most plants, including members of two centrospermous families, the Caryophyl-
laceae and Molluginaceae.

Biogenesis of Betalaixs

It is the biogenesis of betahimic acid and its snbseqnent condensation with
amino acids and amines which appears to be significant among angiosperms for
centrospermous plants. It now appears that the 4,5-extra chol cleavage of L-dopa
(Fig. 3) can lead to betalamic acid (Fischer & Dreiding, 1972; Impellizzeri &
Piattelli, 1972; Chang et al., 1974) in the Centrospermae and the mushrooms
(Musso, private communication) and to stizolobic acid in the antliocyanin-con-
taining Leguminosae (Ellis, 1976) and also mushrooms (Saito et ah, 1975, 1976;
Musso, private communication). Yet the conversion of the cleaved product to
betalamic acid and its conversion into other betalains is known for the Centro-
spermae and mushrooms only. Whether or not these different groups of organ-
isms utilize the same enzymes to synthesize betalains is not known.

PlIYEOGEXKTIC SIGNIFICANCE OF BetALAINS

Among angiosperms, betalains are known only for centrospermous families,
and we use this unique character to circumscribe the suborder Chenopodiineae,
order Centrospermae (see Table 4). Whether or not all systematists accept
this particular subordinal treatment is not important; it is, however, significant
that today the presence of betalains in a family of angiosperms has become
a key character used by all workers for its inclusion in the Centrospermae, Such
families as the Cactaceae and Didiereaceae were allied with the other betalain

216

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

CH

a- cleovage

^^a-cleovage

''b-cleavoge

HO
HO2C

b- cleovage

l^ HgN- -COgH

0^1*

HO2C

CH HgN CO2H

<

HO2C

'''^N CO2H

muscaflavin
( Amgnito ; Hygrocybe )

H02C>,^NH2

stizolobinic acid

( Mu CO no , Leguminosac)

stizolobic acid
( Mucono . Leguminosce)

betolamic acid

betalains
(Centrospermae; A monito )

Fic;uiiE 3. L-Dopa is known to undergo 2,3-1 (a-cleavage) or 4,5- (b-cleavage) extra
(liol cleavage in different organisms. Only in the Centrospermae and in mushrooms does L-dopa
lead via the b-cleavage pathway to betalamic acid, the precursor of all other betalains. I
thank Prof. H. Musso, Univ. of Karlsruhe, for private discussions which led to the scheme
presented here.

families with a greater degree of confidence once their pigments were recognized.
Of course, other data such as Jensen's (1965) serological results also firmly es-

cially the Cactaceae and Portulacaceae.

bet

rmous

I suppose over the years the most controversial question has focused upon
our 1963 (Mabry, Taylor & Turner, 1963) proposal that the Centrospermae be re-
served for the betalain families, with the closely related but anthocyanin-produc-
ing centrospernious families (e.g., Caryophyllaceae and Molluginaceae) being
placed in a close but distinct order, the Caryophyllales. Although we did not
then nor do we now concern ourselves with resolving the question of rank, the

h as the Caryophyllaceae into a
separate order "disturbed" a number of leading systematists. Therefore, our cur-
rent treatment (Table 4), which still maintains distinct taxonomic categories
(suborders) for the betalain and anthocyanin families, is more acceptable be-
cause all the families are recognized as being "centrospermous." While the phy-
logenetic importance of the betalains is emphasized in our current treatment,
the significance of other characters is recognized. It should be noted once more
that our current treatment has been sharply influenced by the occurrence of the
same P-III subtype sieve-element plastids in the Caryophyllaceae and Mollugina-
ceae as are found in the betalain families. Despite having our views shaped by
different kinds of data, the views of myself and many leading systematists are
converging towards a common interpretation of the Centrospennae.

1977]

MABRY— CENTROSPERMAE

217

Table 6. Pollen Morphology (Xowicke, 1975).

Taxa with Centrospermae-Specific Pollen:
Spinulose and Tubuliferous/Punctate Ektexine

Some Taxa with Noneentrospermons

Betalain Families
Aizoaceae
Aniaranthaccae
Bascllaeeae
Caetaceae

Chenopodiaeeae ( inel. D\ sphaniaceae )
Didiereaceae
Nyctaginaeeae
Phytolaccaceae
Portulacaceae
Antliocyanin Families

Caryophyllaccae
Molluginaceae

Poll

en

Achatoeaipaceae

Bataeeae

Gyrostenionaceae

Tlieligonaecae

Polygonaceae

A number of families sometimes treated as being centrospermous but wliicli
contain neither the P-III subtype sieve-element plastids nor betalains are now
excluded from the Centrospermae proper (see Table 4, footnote a). Although
the available molecular data do not suggest an alternative alignment for many
of these taxa, this is not the case for the Bataeeae and Gyrostemonaceae. DNA-
RNA hybridization data bear upon the relationship of the Bataeeae to the Centro-
spermae (Chang & Mabry, 1974). First, these data indicate that the betalain fami-
lies which were tested are closer to each other than to any other family and that
the Caryophyllaccae is the closest family to the betalain group. At the same time,
the Bataeeae were clearly separated from the Centrospermae on the basis of the
available DNA-RNA hybridization data. Moreover, it is significant that Prof.
Martin G. Ettlinger, University of Copenhagen, recently detected (unpublished
manuscript) benzylglucosinolate in Batis maritima L., a species previously re-
ported to contain thioglucosidase (Schraudolf et al., 1971). On the basis of these
and other data, Prof. Ettlinger allies the Bataeeae with other glucosinolate-
containing families (Ettlinger & Kjaer, 1968); furthermore, the isolation of an
isothiocyanate from Codonocarptis cotinifolius (Desf. ) F. Muell. (Bottomley
& White, 1950) indicates that the Gyrostemonaceae also belongs with these same
families. As noted in the next section, the studies of pollen structure not only
support a close relationship of Bataeeae to the Gyrostemonaceae but also dis-
tinguish these two families from those in the Centrospennae (Goldblatt et ah,
1976; Nowicke, 1975).

As a result of discussions with Prof. F. Ehrendorfer (Vienna) and from com-
ments in a recent paper of his (1976), I agree that we must consider the possi-
bility that the betalain families arose from an ancestral taxon which had lost the
ability to produce anthocyanins. Such a process would require that the ances-
tor had lost the one or two enzymatic steps required to convert dihydrofla\x)-
nols into anthocyanins and subsequently gained the two or three steps needed
to form betalamic acid from L-dopa and then condense this aldehyde with
various amines and amino acids to produce the red and yellow betalains.

21S ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Pollen Moiuuiology in the Centrospehmae

Recent investigations (Nowicke, 1975; Skvarla & Nowicke, 1976) of pollen
moipliology for centrospermous taxa (Table 6) also support a close relationship
of the betalain families with the Caryopliyllaceae and Molluginaceae in accord
with our treatment (Table 4). In her examination of 190 species from 16 fami-
lies by light and scanning electron microscopy, Nowicke (1975) found three
common pollen types, all with a spinulose and tubuliferous/punctate ektexine
among the betalain famiUes and the Caryophyllaceae and Molluginaceae. Two
additional minor related pollen types were detected in the Nyctaginaceae,

The pollen morphology of other taxa such as Achatocarpaceae, Bataceae,
Gyrostemonaceae, Polygonaceae, and Theligonaceae do not support their inclu-
sion in the Centrospermae. Of these, only the Achatocarpaceae would appear
to be centrospermous on the basis of having P-III subtype sieve-element plas-
tids; the pigment content of this taxon is not yet known.

Summary

It is our view that all the betalain families and the two anthocyanin families,

the Caryophyllaceae and Molluginaceae, a
spermous" ancestral taxon which evolved t
tids now characteristic of all these families.

u

ccntro-

btyp

h

ably preadapted for C^ photosynthesis and for a pollen morphology with spinulose
and tubiferous/ektexine. The ancestral taxon for the betalain families may have
arisen either from a taxon which had anthocyanins, then lost them and later
gained betalains, or from a taxon which had not previously contained either type
of pigment. In any case, the centrospermous evolutionary complex is now repre-
sented by eleven core families which in my opinion are best treated in one or-
der, the Centrospermae or Caryophyllales, which consists of a betalain-suborder,
the Chenopodiineae, and an anthocyanin-suborder, the Caryophyllineae (Table

4).

LiTERATURK ClTED

Ardenne, R. von, H. Dopp, IL Musso & W. Steglich. 1974. Isolation of nuiscaflavin
from Uijgrocijhc species (Agaricales) and its dihydroazcpine structure. Z. Naturf. 29C:
637-639,

Beiinke, II.-D. 1976a. Ultrastructurc of sieve-element plastids in Caryoph>I]a]es (Centro-
spermae), evidence for the delimitation and classification of the Order. PL Syst. EvoL 126:
31-54.

. 1976b. Sieve-element plastids of Fotujuieria^ Franhenia (Tamaricales), and Rhah'

doilcndwn (Rutaceae), taxa sometimes allied with Centrospermae ( Caryophyllales).

Taxon 25: 265-268.

— . In press. Transmission electron microscopy and systematics of flowering plants.

PI. Syst. Evol.

— & T. J. Mabry. 1977. S-Type sieve-element plastids and anthocyanins in Vixiania-

ceae; evidence against its inclusion into Centrospermae. PI. Syst. Evol. 126: 371-375.
— , C. Chanc;, I. J. EiFERT & T. J. Mabry. 1974. Betalains and P-type sieve-tube plas-

tids in Pctivcra and Af^dcstis (Phytolaccaceae). Taxon 23: 541-542.
BoTTONTT.EY, W. 6i D. E. White. 1950. The chemistry of Western Australian plants, Part
IL The essential oii oi Codonocar})tis cotinifolius (Desf.) F. Muell. J. Proe. Roy. Austral.
Chem. Inst 17: 31-32.

1977] MABRY— CENTROSPERMAE

219

Chang, C. & T. J. MAiiUv. 1974. The constitution of tlie Order Centrospcrmac: RNA-
DNA hybridization studies among betalain- and anthocyanin- producing famiHes. Biocliem.
Syst. 1: 185-190.

, L. KiMLER & T. J. Mabry. 1974. Biogenesis of betalamic acid. Phytocheniistry

13: 2771-2775.

Croxquist, a. 1968. The Evolution and Classification of Flowering Plants. Houghton
Mifflin Co., Boston.

Dopp, H. & H. Musso. 1973. Fliegenpilz Farbstoffe, II. Isolierung und Chroniophore der
Farbstoffe aus Amanita muscaria. Chem. Ber. 106: 3473-3482.

& . 1974. Chromatographic analysis of betalain pigments in toadstools and

higher plants. Z. Naturf. 29C: 640-642.

Eckarj:)T, T. 1976. Classical morphological features of centrospermous families. PI. Syst.
Evol. 126: 5-25.

Ehrendorfer, F. 1976. Closing remarks: systematics and evolution of centrospermous
families. Pi. Syst. Evol. 126: 99-105.

EicHLER, A. W. 1876. Syllabus der Vorlesungen ubcr Phanerogamenkunde. Schwer'che
Buchhandlung, Kiel.

. 1878. Bliithendiagramme. Zweiter Teil. W. Engelmann, Leipzig.

Ellis, B. E. 1976. Dopa ring-cleavage in the biogenesis of stizolobic acid in Mucuna
deeringiana. Phytocheniistry 15: 489-491.

Ettlinger, M. G. & A. KjAER. 1968. Sulfur compounds in plants. In T. J. Mabry, R. E.

Alston & V. C. Runeckles (editors), Recent Advances in Phytocheniistry. Vol. 1: 59-144.

Appleton-Century-Crofts, New York.
Fischer, N. & A. S. Dreiding. 1972. Biosynthesis of betalains. On tlie cleavage of the

aromatic ring during the enzymatic transformation of dopa into betalamic acid. Ilelv.

Chim. Acta55: 649-659.

Cc)LDBLATi\ P., J. W. NowicKE, T. J. Mahry & H.-D. Behnke. 1976. Cyrostemonaceae :
status and affinity. Bot. Not. 129: 201-206.

IIardorxe, J. B. 1967. Comparative Biochemistry of the Flavonoids. Academic Press,
London.

Hunziker, J. H., H.-D. Behnke, I. J. Eifert & T. J. Mahrv. 1974. IlalopJiytinn amcghinoi:
a betalain-containing and P-type sieve-tube plastid species. Taxon 23: 537-539.

Imrellizzeri, G. & M. Piatelli. 1972. Biosynthesis of betalains: formation of indicaxan-
thin in Opuntia ficus-indica fruits, Phytocheniistry 11: 2499-2502.

Jensen, U. 1965. Serologische Untersuchungen zur Fruge der systematischen Einordung
der Didiereaceae. Bot. Jahrb. Syst. 84: 233-253.

Marry, T. J. 1973. Is the Order Centrospermac monophyletic? Pp. 275-285, in G. Bendz
& J. Santesson (editors), Chemistry in Botanical Classification. Nobel Foundation, Stock-
holm and Academic Press, London.

1976. Pigment dichotomy and DNA-RNA hybridization data for centrospermous

families. Pi. Syst. Evol 126: 79-94.

— & H.-D. Behnke (editors). 1976a. Evolution of centrospermous families. Pi. Syst.
Evol. 126: 1-105.

— & H.-D. Behnke. 1976b. Betalains and P-type sieve-element plastids: The systema-
tic position of Dysphania R. Br. (Centrospenuae ). Taxon 25: 109-111.

— & A. S. Dreiding. 1968. The betalains. In T. J. Mabry, R. E. Alston & V. C.

Runeckles (editors), Recent Advances in Phytocheniistry. Vol. 1; 145-160. Appleton-
Century-Crofts, New York.

, H.-D. Behnke & I. J. Eifert. 1976. Betalains and P-type sieve-element plastids

inGisekiah. (Centrospermac). Taxon 25: 112-114.

, L. Kimler & C. Chang. 1972. The betalains: structure, function and biogenesis

and the plant order Centrospenuae. In V. C. Runeckles & T. C. Tso (editors), Recent
Advances in Phytocheniistry. Vol. 5: 104-134. Academic Press, New York.
— , A. Taylor & B. L. Turner. 1963. The betacyanins and their distribution. Phyto-
cheniistry 2: 61-64.

■

, I. J. Eifert, C. Chang, H. Marry, C. Kidu & H.-D. Behnke. 1975. Theligonaceae.

Pigment and ultrastructural evidence which exclude it from the order Centrospermac.
Biochem. Syst. Ecol. 3: 53-55.

NowicKE, J. W. 1975. Pollen morphology in the order Centrospenuae. Grana 15: 51-77
Piattelli, M. 1976. Betalains. In T. W. Goodwin (editor), Chemistry and Biochemistry
of Plant Pigments. Vol. 1; 560-596. Academic Press, New York.

220 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Saito, K., a. Komamine & S. Senoh. 1975. Biosynthesis of stizolobinic acid and stizolobic
acid in the etiolated seedlings of Stizolohium hassjoo, Z. Naturf. 30C: 659-662.

& . 1976, Further studies of the biosynthesis of stizolobinic acid

and stizolobic acid in etiolated seedlings of Stizolohium hassjoo. Z. Naturf. 31C: 15-17.

ScHRAUDOLF, H. B., B. Sc:hmiut & F. Weuehlixg. 1971. Das Vorkommen von **Myroslnase"

als Ilinweis auf die systematische Stellung der Baiidaccae. Expcrientia (Basel) 27: 1090-

1091.
Skvarla, J. J. & J. W. Nowicke. 1976. Ultrastructure of pollen exine in centrospernious

families. Pi. Syst. Evol. 126: 55-78.
Takhtajan, a. 1973. Evolution und Ausbreitung der Bliitenpflanzen. Gustav Fischer

Verlag, Stuttgart.
TiMBEHi.AKE, C. F. & P. Briule. 1975. The anthocyanins. Pp. 214-266, in J. B. Ilarbome,

T. J. Mabry & H. Mabry (editors), The Flavonoids. Chapman and Hall, London.
Winter, K. 1974. Evidence for the significance of crassulaccan acid metabolism as an

adaptive mechanism to water stress. Pi. Sci. Lett. 3: 279-281.

DEFENSIVE ECOLOGY OF THE CRUCIFERAE'

Paul Feeny-

Abstract

The glucosinolates (mustard oil glucosides), present in all crucifer species examined,
seem to provide a major line of chemical defense against bacteria, fnngi, insects, and mam-
mals. Circumstantial evidence suggests that other classes of secondary compounds, each re-
stricted to one or a few genera, represent a second line of chemical defense.

Survival of wild crucifers depends partly on escape from adapted enemies in time and
space. Discovery of crucifers by several enemy species is aided by behavioral responses to
glucosinolates or their breakdown products. Allylghicosinolate (sinigrin) in the lea\'es of
Thlaspi awense releases allylthiocyanate instead of the more typical allylisothiocyanate, wliicli
is used as a liost-finding attractant by several insect species. This change in secondary chemis-
try may reduce the rate of discovery of Thlaspi plants by crucifer-adapted enemies.

The defensive ecology of crucifers seems to t>pify that of herbaceous plants generally;
chemical resistance, in the form of small amounts of toxic compounds, combined with low
apparency to enemies which are adapted to the chenu"cal defenses. The importance of the
Cruciferae and other families of herbaceous plants as sources of food-plants for man may
result in large part from their relati\'el>' \o\\ concentrations of toxins. The mature foliage of
trees, shrubs, and grasses, by contrast, remains poor food for man, just as for other plant

enemies.

An important component of the defensive ecology of crucifers and other unapparent plants
seems to be chenn"cal diversity in space and time. Closer simulation of this diversity in fields
of agricultural crops may reduce the need for synthetic pesticides.

The family Cruciferae comprises approximately 400 genera and 3,000 species,
tbe vast majority of which are herbaceous (Vaughan et al., 1976). The greatest
number of species are found in temperate regions of the northern hemisphere,
especially in those with a Mediterranean type of climate. Tbe Irano-Turanian
region alone contains about 150 genera and 900 species and may well have been
tlie evolutionary cradle of the family, at least in the Old World (Hedge, 1976).
The family has colonized a great variety of habitats, including arctic and alpine
regions and some of the most climatically inhospitable deserts, though it is poorly
represented in the tropics ( Hedge, 1976) .

The family is the source of several economically important species and vari-
eties, especially the cole crops of the genus Brassica. Economic incentives have
stimulated extensive research on interactions between crucifers and their as-
sociated insects and pathogens. Understanding of the chemical aspects of these
interactions has been helped greatly by unusually thorough knowledge of the
family's chemistry (see Kjaer, 1976).

^ I thank the students in my general ecology class for compiling tlie list of human food-
plants, Karen Arms for improving my grammar, and David Bates for considerable help with
the preparation of the Appendix. Financial support was provided by research grant BMS-
7409868 from the National Science Foundation and Hatch grant NYC-L39U3.

■ Department of Entomology and Section of Ecology and Systematics, Cornell University,
Ithaca, New York 14853.

AxN. Missouri Bot. Gahd. 64: 221-234. 1977.

222 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Primary Chemical Defense

Tlie first characteristic line of chemical defense in crucifers is evidently pro-
vided by tlie ghicosinolates (mustard oil glncosides, thioglucosides). The oc-
currence of these compounds, more than 70 of which are known, is restricted al-

Resed

d

1960, 1976; Ettliniier & Kiaen 1968

hydrolyzed typically to yield volatile isothiocyanates (mustard oils) when plant

ucosi

poncnt of plants of the genus Brassica and breaks down to allylisothiocyanate:

N-O-SO3

CH2 ^ CII - Clio - C + ILO -^ CII > = CH - CH > - NCS + GKicose + IlSOr

\

S-Glucose

Allylisothiocyanate, released from allylghicosinolate, is largely responsible for
the odor of cooked cabbage (MacLeod, 1976). Hydrolysis of glucosinolates
is catalyzed by a group of enzymes (myrosinases) which are stored separately
within the plant tissues but which come into contact with their substrates when
the plant tissues are bruised or otlierwise damaged (Kjaer, 1976; Bjorkman,
1976). Storage of isothiocyanates in the form of glucosinolates may represent
an adaptation to avoid autotoxicity; isothiocyanates arc strongly phytotoxic
(Hooker et al., 1945; Bell & Muller, 1973).

Glucosinolates or their breakdown products are known to be powerful anti-
biotics (Virtanen, 1958, 1965) and to inhibit the growth of fungi (Walker et al.,
1937) and insects (Brown, 1951; Lichtenstein et al, 1964). The concentration
of allylglucosinolate in the foliage of Brassica nigra plants in Tompkins County,
New York, was found to be about 0.4% of fresh weiglit, depending somewhat on
habitat and leaf age (P. Feeny and L. Contardo, in preparation); at this concen-
tration the compound proved to be toxic to larvae of the black swallowtail but-
terfly, Papilio poltjxenes, which naturally feed on umbellifers but occur in the
same habitats as many crucifer species in the northeastern United States (Erick-
son & Feeny, 1974; P. Blau, P. Feeny and L. Contardo, in preparation). Glucx)-
sinolates or their liydrolysis products, when ingested in large quantities, are also
toxic to mammals; the effect seems to result, at least in part, from the effective-
ness of allylisothiocyanate as a tissue irritant (Kingsbury, 1964).

Glucosinolates in crucifers may play a role as allelopathic agents, inhibiting
the germination and growth of competing plants. Patches within the annual
grasslands of southern California are dominated by B. nigra, introduced from
Europe. Bell & Muller (1973) showed convincingly that the persistence of these
patches from year to year can be attributed to inhibition of the germination and
growth of other plants by compounds leached from B. nigra. They found that
allylisothiocyanate is a potent inhibitor of gemiination by seeds of several grasses
but ruled it out as the allelopathic agent because of its rapid loss of activity in
the soil. The unknown toxic compounds are water soluble and are leached from
dead B. nigra tissues of the previous season's growth by the first fall rains, which

1977] FEENY— DEFENSIVE ECOLOGY

223

also serve as the stimulus for germination by the seeds of most species (Bell &
Muller, 1973). I am not convinced that the authors have completely ruled out
the possibility that the allelopathic agent is allylglucosinolate, stored in dead
stems during the summer drought period and capable of releasing allylisothio-
cyanate over a period of time after being leached into the soil.

The available evidence thus suggests that the biological activity of the glu-
cosinolates is broad and supports the contention that predation, disease, and per-
haps competition are selective pressures which have wntributed to the evolution
and diversification of these compounds in the Cruciferae (see also Feeny, 1976).

The Crucifer Fauna

In spite of their content of glucosinolates, crucifers are attacked by an ex-
tensive array of insect species, several of which have become major pests of cul-
tivated cruciferous crops. Many of the insect species which attack crucifers
are "specialists" which rarely or never attack plants of other families; examples
include larvae of the familiar cabbage ])utterfly, Pieris rapae, the cabbage aphid,
Breviconjne hrassicae, and the cabbage flea beetles Phyllotreta cruciferae and
F. striolata (Root, 1973). Generalist insects which include crucifers among their
normal range of host-plants include the cabbage looper, Trichoplusia n/, and
the green peach aphid, Myzus persicae. Crucifers are subject also to attack by
an extensive array of fungi and bacteria (Westcott, 1971) and are probably eaten
in significant quantities by wild mammals. In view of the deleterious effects of
glucosinolates on organisms which do not normally attack crucifers we must
presume that the various species making up the typical fauna of natural crucifers
are somehow adapted to detoxify glucosinolates or otherwise avoid their harm-
ful effects. The actual mechanisms of detoxification by adapted enemies are
not yet known. Larval growth of the cabbage butterfly, P. rapae, on a wide
range of crucifer species and cultivated varieties was compared recently by
Slansky & Feeny (1977). Growth showed no obvious relationship to the varied
pattern of glucosinolates present in the test plants but was closely related to the

lity

Individual glucosinolates vary in their

toxicity to nonadapted insects (e.g. Brown, 1951); in crucifer-adapted insects,
however, the extent to which tolerance of one glucosinolate confers tolerance
of others needs to be examined in more detail.

When concentrations of allylglucosinolate in the leaves of collard plants,
Brassica oleracea, were artificially increased b}^ culturing to 20 times the tyjiical
level, growth of P. rapae larvae remained unaffected (P. Blau, P. Feeny and L.
Contardo, in preparation). This result suggests that glucosinolates represent
"qualitative" or "evolutionary" barriers to nonadapted insects: once overcome
by adaptation they have little or no toxic effect in spite of wide variation in con-
centration (Feeny, 1975, 1976). This is consistent with the finding by van Em-
den (1972) that relative growth rate of the cabbage aphid, B. hrassicae, was
correlated positively with the "total allylisothiocyanate" content of crucifer test-
plants. Glucosinolates stimulate feeding by this crucifer-restricted aphid but
they are evidently not toxic to it, at least at concentrations normally encountered
in the plants. Dosage-related toxicity of glucosinolates may have greater ecologi-

224

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

cal effects on crucifer-adapted bacteria and fungi than it seems to have on in-
sects which speciaHze on these plants.

The effects of gkicosinolates on generalist insects seem to be intermediate be-
tween their effects on criicifer-speciahsts and those on insects which do not natu-
rally attack crucifers. The southern armyworm, Spodoptera eridania, and peach
aphid, M. persicae, naturally attack crucifers as well as plants of many other
families. They must therefore be able to tolerate at least low levels of gkicos-
inolates. However, larvae of S. eridcinia have less tolerance for leaves artificially

boo

M

corr

negatively with the "total allylisothiocyanate" content of crucifer

leaves (vanEmden, 1972).

Escape from Adapted Enemies in Space and Time

Such is the ability of adapted enemies to damage and destroy crucifer plants,
once they have been discovered, that the survival of crucifers in nature must
surely be attributable in large measure to their ephemeral life histories which
provide a constantly changing pattern of geographical and phenological distribu-
tion. The habitats most favored by crucifers seem to be those in which periods
favorable for growth are severely limited by climatic variables such as rainfall
(e.g. chaparral, grassland, desert) and temperature (arctic and alpine habitats).
Typical crucifers must therefore be capable of rapid growth to maturity and
seed-set, and it is perhaps not surprising that so many species have evidently
been preadapted to exploit disturbed areas associated with human activities. Short
growth season, shifting pattern of geographic distribution, and association with
harsh and somewhat unpredictable climatic conditions are all characteristics
which are likely to favor escape by plants from their adapted enemies (Janzen,
1970; Rhoades & Gates, 1976; Feeny, 1976).

The importance of escape from adapted insect enemies to the ecology of
herbaceous plants was well illustrated by the history of introduced Klamath
weed, Hypericum perforatum^ in California (Huffaker & Kennett, 1959). In
1951 this plant infested more than 2 million acres of range land, covering up to
80% of the ground area in some places. Introduction from Europe of the Hj/-
perfcw m-adapted leaf beetles, Chrijsolina quadrigemina and C. hypericin reduced
the plant to less tlian 1% of its former abundance by 1959. Both plant and beetles
continued to persist at low densities, the plant surviving best in shadier habitats
where the beetles are less effective (Huffaker & Kennett, 1959). "It would seem
that the new low density of Hypericum perforatum is maintained at a level at
which interplant distance restricts epidemic development of the beetle by limit-
ing its opportunity to discover the isolated specific food plants" (Harper, 1969).
While no such dramatic examples are available, it seems, for cruciferous plants,
escape from discovery by adapted enemies is likely to be an important component

of their defensive ecology also.

Pimentel (1961) and Root (1973) have shown that populations of crucifer-
adapted specialists such as B. hrassicae and F. cruciferae reach higher densities
on collard plants grown in monoculture patches than on plants grown among di-

1977] FEENY— DEFENSIVE ECOLOGY

225

verse meadow vegetation. Root (1973) attributed these findings to "resource
concentration": herbivores are more Hkely to find and remain on hosts that are
growing in dense or nearly pure stands. An individual collard plant is more 'ap-
parent" (i.e., susceptible to discovery) when growing next to other collard plants
than w^hen growing among plants of other families (Feeny, 1976). Comparable
experiments by Smith (1976) showed that populations of B, hrassicae and other
crucifer-feeding species reached higher levels on Brussels sprout plants grown
on weed-free soil than on plants grown among weeds. Trapping experiments
showed that weed-free plants were more attractive to colonizing aphids, prob-
ably because a background of bare soil presents greater visual contrast tlian does
a background of weeds (Smith, 1976). Diversity of surrounding vegetation may
similarly permit wild crucifers in natural habitats to escape or reduce the risk
of discovery by searching insects (Feeny, 1976) .

Plants of the genus Dentaria differ from more typical crucifers in several re-
spects. They are perennial and form patches, often of substantial area, among
the ground vegetation of mature deciduous forests. The plants leaf out very
early in the spring and approach senescence by the time the forest canopy has
leafed out. Plants of D. diphylla were damaged heavily, after transplanting into
open field habitats, by the typical open-habitat crucifer flea beetles P. cruciferae
and P. striolata (Ilicks & Tahvanainen, 1974) and Dentaria leaves supported
better growth of P. rapae lan^ae than did those of any other crucifer tested
(Slansky & Feeny, 1977). Though subject to their own specialized enennes,
such as the butterfly Pieris virginiemis and the flea beetle PhyUotreta hipmtu-
lata, Dentaria species have probably benefited by their escape, in evolutionary
time, into a habitat which is atypical of crucifers and thus not frequented by
many of the typical crucifer-adapted enemies.

Plant-Finding Adaptations

Many crucifer-adapted insects have evolved behavioral responses to glucosi-
nolates or their breakdown products, thus permitting them to find their food-
plants more easily and to discriminate them from other vegetation. An early
example of such behavior was described by Verschaffelt (1911) who found that
larvae of P. hrassicae and P. rapae can be stimulated to feed on normally re-
jected plants by treating the plants with solutions of allylglucosinolate. A recent
review by Schoonhoven (1972) lists a dozen insect species which are known to
make use of these compounds as behavioral cues. There is even a crucifer-
adapted fungus, Plasmodiophora hrassicae, the spores of which are stimulated
to germinate by the presence of allylisothiocyanate (Hooker et ah, 1945). Jie-
havioral responses to glucosinolates or isothiocyanates by individuals of any one
insect species usually depend on concentration and may also vary from one
compound to another (e.g., Thorsteinson, 1953; Hicks, 1974; Finch & Skinner,
1974 ) .

The cmcifer-feeding flea beetles, PhyUotreta cruciferae and P. striolata, are
strongly attracted to traps containing solutions of allylisothiocyanate (Feeny et
al., 1970) and can also be induced to eat bean leaves, which they normally re-
ject, when these have been cultured in solutions of allyglucosinolate (Hicks,

226 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

1974). We have found recently that addition of vials containing solutions of
allylglucosinolate in mineral oil to 3-plant islands of Brassica nigra, planted among
diverse vegetation, greatly accelerated the rate of discovery of the plants by
these flea beetles (P. Feeny, J. Gaasch and L. Contardo, in preparation). This
finding not only confirms the effectiveness of allylisothiocyanate as a host-find-
ing attractant but also shows that leakage of such compounds, even in small
amounts, can be a liability to B. nigra plants since it increases their apparency
to adapted enemies.

Secondary Defense in Crucifers

Many crucifers are known to contain other secondary compounds in addition
to glucosinolates. The genera Erysimum and Cheiranthus, for example, contain
cardcnolides, the genus Iheris contains cucurbitacins, and plants of the genera
Lunaria and Capsclh contain alkaloids (Gheorghiu et al., 1959; Ilegnauer, 1964).
The genera Lepidium and Thlaspi contain atypical enzymes which break down
glucosinolates not to the typical isothiocyanates but to their correspondmg geo-
metrical isomers, the thiocyanates (Gmelin & Virtanen, 1959).

Many of these plants are avoided by crucifer-adapted insects or, if fed upon,
support unusually poor growth. Larvae of P. rapae^ for instance, grow poorly on
Thhispi arvetise, Lepidium virginicum, and Lunaria annua (Slansky & Feeny,
1977); they will refuse to eat leaves of Erysimum cheiranthoides and Capsella
hursa-pastoris (A. M. Shapiro, personal communication). Verschaffelt (1911)
found that C. hursa-pastoris was attacked only very slightly by larvae of P. rapae
and P. hrassicae-y E. perofskianum was also less preferred by these lai'vae relative
to most other crucifers offered to them. Plants of E, cheiranthoides, C. hursa-
pastoris, and Iheris amara are not eaten by P. cruciferae flea beetles (Feeny
ct al., 1970). Chew (1975) found that lai-vae of Pieris napi macdunnougJui in
Colorado refused to eat Erysimum asperum. Larvae of P. napi macdunmmghii
grew normally on ThJaspi montanum^ a native plant in Colorado, but they and
larvae of P. occidcntalis died after eating T. arvense, an introduced species. The
unusual resistance of plants of these genera to typical crucifer enemies may re-
sult from their content of atypical secondary compounds (see Verschaffelt, 1911).
Allylthiocyanate, for example, is known to be toxic to insects (Brown, 1951).
Such cK)mp()unds could have been evolved as a "second line of defense" in re-
sponse to enemies which have evolved mechanisms for tolerating glucosinolates
and their typical hydrolysis products. Diversification of secondary chemistry,
in other words, may permit escape from certain enemies in evolutionary time,
at least imtil further counteradaptations are evolved by the associated insects
or other enemies.

In addition to their possible toxic or growth-inhibitory effects, unusual sec-
ondary compounds may further benefit a plant species by reducing apparency
to adapted enemies. Hydrolysis of allylglucosinolate m leaves of Thlmjn arvetue
yields all>'lthiocyanate instead of the more typical allylisothiocyanate (Gmelin
& Virtanen, 1959; P. Feeny and L. Contardo, in preparation). Three-plant is-
lands of T. arvense were colonized by PhyUotreta flea beetles at a consideral)ly
slower rate than were nearby islands of Brassica nigra, perhaps because allyliso-

1977] FEENY— DEFENSIVE ECOLOGY

227

thiocyanate is au attractant to the beetles whereas allylthiocyanate is not. Coloni-
zation of T, arveme islands was accelerated by addition of vials containing solu-
tions of allylisothiocyanate (P. Feeny, J. Gaasch and L. Contardo, in preparation).
Cnicifers may derive additional protection from adapted enemies as a result
of association with plants of different chemistry. Tahvanainen & Root (1972)
have found that odors from tomato, Lycopemcon hjcopersicum {—esculent urn) ,

l

foil

?red with the ability of P.
The reduction of plant ap-

parency to enemies by neighboring plants of different species is an important
component of "associational resistance" (Tahvanainen & Root, 1972) — a phe-
nomenon frequently exploited by organic gardeners.

Chemical Defense and the Human Diet

The defensive ecology of crucifers seems to typify that of many ephemeral
herbaceous plants — plants which rely to a great extent on being hard to find
(unapparent) in natural habitats. Such plants seem to contain rather low con-
centrations of effective toxins. They probably benefit from a diversity of chemi-
cal defense in any one species and from association with other plants of different
chemistry (Rhoades & Cates, 1976; Feeny, 1976). Their defenses clearly differ
from those of the mature foliage of more persistent plants such as shrubs and
trees. Such plants are bound to be found by enemies and must correspondingly
be well defended; they often contain large amounts of general growth-inhil)it()ry
compounds, like tannins, resins and silica, which are resistant to simple counter-
adaptation. The foliage of apparent plants is usually tough and deficient in
nutrients and water when compared with that of most herbaceous plants
( Rhoades & Cates, 1976; Feeny, 1976 ) .

These differences in the defensive ecology of plants, depending upon their
apparency to enenu'es, seem to be reflected in human food preferences. One
hundred students in the general ecology course at Cornell University were asked
to list as many species of human food-plants as they could think of in 15 minutes.
Their total of 108 species, excluding plants used primarily as spices and drugs,
was then tabulated by plant growth form and by what part of the plant is eaten
(Fig. 1 and Appendix). Though undoubtedly a biased view of more general
patterns of plant consumption by man, this suivey revealed some interesting and
suggestive trends.

Most of the species listed are harvested only for their fruits or seeds, and of
these species most are trees (Fig. 1). The production of fleshy fruits is probal)ly
an adaptation for seed dispersal by vertebrate animals, including our primate
ancestors. Ripe fruits are adapted to be attractive to animals by their size, color,
and taste; fruit-eating behavior by the animals is probably reinforced by the
fact that fruits contain not only energy-rich carbohydrates and fats but also vita-
mins and mineral ions which are vital for the sundval of many vertebrate ani-
mals and not readily available from other natural sources (see McKey, 1975).
One can even speculate that our "sweet tooth/' now a conspicuous liability in
times of readily available sugar, represents a physiological adai^tation which

228

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Shrub

or

perennial

13

vine

Herb

16

16

Grass

f V

Root

Foliage Fruit / Seed

FuiUUE 1. Distri])ution of 108 species of Iniman food-plants according to plant ^^rowth

forni and part of plant eaten. Figures indicate number of food-plants listed in each catej^ory
(3 species listed twice). See Appendix for details.

stimulated our ancestors to seek out fruits with their high nutrient vahae (see
Yudkin, 1969).

A second striking pattern reflected in this survey is that plants whose roots
or leaves form part of the human diet are almost all herbs (Fig. 1). These are
the plants, including the ancestors of our cruciferous vegetables, which tend to
be ephemeral and unapparent in nature. The origins of cultivated plants from
herbaceous species were probably due to the concentrated food vakie of the
roots or tubers of many of these species and to the unique preadaptations of

1977] FEEXY— DEFENSIVE ECOLOGY

229

"weedy" plants to thrive in the disturbed habitats associated witli liunian habita-
tion (Ilawkes, 1969). Preferential consumption of herbaceous species may also
reflect the presence in trees and shrubs of extensive chemical and physical de-
fenses, evolved by trees and shrubs because they are relatively apparent to natural
enemies. Many of the drugs and spices used by man come from the foliage and
roots of trees and shrubs, though they are rarely consumed in large quantities.
By contrast we are presumably able to tolerate the comparatively low concen-
trations of defensive compoiuids in crucifers and other herbaceous plants both
because of the detoxication enzymes concentrated in the vertebrate liver (Free-
land & Janzen, 1974) and also, since the cultural evolution of the use of fire, be-
cause we further detoxify or remove many of these compounds by choking (Yud-
kin, 1969; Leopold & Ardrey, 1972). Only because they contain relatively small
concentrations of toxins can we consume such plants in large quantities.

Api>aiu-:ncy axd Agriculture

Tlie effectiveness of natural plant defenses is reduced by present agricultural
methods. When they are planted in monocultures, crop plants become more
apparent to natural enemies than are their ancestors in nature, yet they possess
chemical and physical defenses inappropriate for survival as apparent plants.
This is a major reason that substantial quantities of synthetic pesticides are
often required to prevent widespread devastation of crops.

It would undoubtedly be possible to modify crop varieties and agricultural
methods so as to mimic the defensive ecology of wild ancestral plants more
closely. Levels of natural defensive compounds could be maintained or restored
by selective plant breeding and emphasis placed on diversity of defense within
any particular crop species. Plant apparency could be reduced by such tradi-
tional techniques as crop rotation and inteiplanting of different crops or chemi-
cal varieties of any one crop. Apparency could be reduced further by eliminating
or modifying those plant chemicals which the more important plant enemies use
as behavioral attractants or feeding stinnilants.

Strategies to improve and diversify chemical resistance would be more ef-
fectiv^e if they were coordinated with strategies to reduce plant apparency. Just
as the evolution of resistance to a particular pesticide by an insect population
may result from extensive and repeated exposure to that compound, so also the
fewer the insects which find a particular plant variety, the less likely they are
to evolve methods of tolerating the plant's chemical defenses (Southwood, 1973).

A key component of the defensive ecology of crucifers and other unapparent
plants seems to be chemical diversity in space and time (Rhoades & Gates, 1976;
Futuyma, 1976; Feeny, 1976). The more closely we can simulate this diversity
in our fields of vegetable crops, the less dependent are we likely to become on
the use of synthetic pesticides to achieve a given level of agricultural production.

LitilRature CrrED

Bell, D. T. & C. H. Muller. 1973. Dominance of California annnal grasslands by Bras-

sica nigra, Ainer. Midi. Naturalist 90: 277-299.
BjoRKNfAN% R. 1976. Properties and function of plant myrosinases. Pp. 191-205, in

230 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

J. C Vau^lian, A. J. MacLeod & B. M. G. Jones (editors), The Biology and Clicniistry

of the Cruciferae. Academic Press, New York.
Brown, A. W. A. 1951. Insect Control by Chemicals. Wiley, New Yorlc.
CiiEW, F. S. 1975. Coevolution of pierid butterflies and their cruciferous foodplants. I.

The relative quality of available resources. Oecologia 20; 117-127.
Erickson, J, M. & P. Feeny. 1974. Sinigrin: A chemical barrier to the black swallowtail

huttevilyy Pap ilio poly xcnes. Ecology 55: 103-111.

Ettlincer, M. G. & A. KjAER. 1968. Sulfur compounds in plants. In T. J. Mabry (edi-
tor), Recent Advances in Phytocliemistry. Vol. 1: 59-144.

Feeny, p. 1975. Biochemical coevolution between plants and their insect herbivores. Pp.
3-19, in L. E. Gilbert & P. 11. Raven (editors), Coevolution of Animals and Plants.
Univ. of Texas Press, Austin.

-. 1976. Plant apparency and cliemical defense. In J. W. Wallace & R. L. Mansell

(editors), Recent Advances in Phytocliemistry. Vol 10: 1-40.

, K. L. Paauwe & N. J. Demonc. 1970. Flea beetles and mustard oils: Host plant

specificity of Phyllotreia cruciferae and P. striolata adults (Coleoptera: Chrysomclidae).

Ann. Entomol. Soc. Amer. 63: 832-841.
Flncii, S. & G. Skinner. 1974. Studies on the cabbage root fly: Evaluation of attractants.

Rep. Natl. Veg. Res, Stat. (Wellesbourne, Warwickshire, England) for 1973, pp. 84-85.
Fheelanp, W. J. & D. II. Janzex. 1974. Strategies in herbivory by mammals: the role

of plant secondary comp(mnds. Amer. Naturalist 108: 269-289.
FuTUYMA, D. J. 1976. Food plant specialization and environmental predictability in

Lepidoptera. Amer. Naturalist 110: 285-292.
Gheohgiiiu, a., a. Constantinescu & E. Ionescu-Matiu. 1959. Extragerea si separarea

alcaloizilor fluorcscenti din CapscUa bursa pastoris L. Stud. Cercet. Biochim. 2: 403—406.
Gmelin, R. & A. I. ViRTANEN. 1959. A new type of enzymatic cleavage of nnistard oil

glucosides. Formation of allylthiocyanate in Thlaspi arven,sc L. and benzylthiocyanate

in Lcpidium ruderale L. and Lepidium sativum L. Acta Chem. Scand. 13: 1474—1475.
Hari»er, J. L. 1969. The role of predation in vegetational diversity. In G. M, Woodwell

& II. II. Smitli (editors). Diversity and Stability in Ecological Systems. Brookhaven

Symposium in Biology. No. 22: 48-62.
IIawkes, J. G. 1969. The ecological background of plant domestication. Pp. 17-29, in

P. J. Ucko & G. W. Dimbleby (editors), The Domestication and Exploitation of Plants

and Animals. Duckworth, London.
IlKix;E, I. C. 1976. A systematic and geographical survey of the Old World Cruciferae.

Pp. 1-45, in J. G. Vaughan, A. J. MacLeod & B. M. G. Jones (editors). The Biology and

Chenn'stry of tlie Cruciferae. Academic Press, New York.
IIecnauer, R. 1964. Chemotaxonomie der Pflanzen. Band 3: Dicotyledoneae: Acantha-

— Cyrillaceae. Berkhiiuser Verlag, Basel.
Hicks, K. L. 1974. Mustard oil glucosides: feeding stimulants for adult cabbage flea

beetles, Phtfllotrcta cruciferae (Coleoptera: Chrysomclidae). Ann. Entomol. Soc. Amer,

67: 261-264.
— ' & J. O. Tahvanainen. 1974. Niche differentiation by crucifer-feeding flea beetles

( Coleoptera: Chrysomclidae ) . Amer. Midi. Naturalist 91 : 406-423.
Hooker, W. J., J. C. Walker & K. P. Link. 1945. Effects of two mustard oils on Plas-

modiophora hrassicac and their relation to resistance to clubroot. J. Agric. Res. 70: 63-78.
Huffaker, C. B. & C. E. Kennett. 1959. A ten year study of die vegetational changes

associated with biological control of Klamath weed. J. Range Managem. 12: 69-82.

Janzex, D. H. 1970. Herbivores and the number of tree species in tropical forests. Amer.

Naturalist 104: 501-528.
Kincshury, J. M. 1964. Poisonous Plants of the United States and Canada. Prentice-Hall,

Inc., Englewood Cliffs, New Jersey.

KjAER, A. 1960. Naturally derived /.sothiocyanates (mustard oils) and their parent gluco-
sides. Fortschr. Chem. Org. Naturstoffe 18: 122-176.

. 1976. Glucosinolates in the Cruciferae. Pp. 207-219, in J. G. Vaughan, A. J. Mac-
Leod & B. M. G. Jones (editors). The Biology and Chemistry of the Cruciferae. Aca-
demic Press, New York.

Leopold, A. C. & R. Ardrey. 1972. Toxic substances in plants and the food habits of early
man. Science 176: 512-514.

Lichtenstein, E. p., D. G. Morgan & C. H. Mueller. 1964. Naturally occurring insecti-
cides in cruciferous crops. J, Agric. Food Chem, 12: 158-161.

1977] FEKNY— DEFENSIVE ECOLOGY

231

MacLeod, A. J. 1976, Volatile flavour compounds of tlie Cruciferae. Pp. 307-330, in
J. G. Vau^^han, A. J. MacLeod & B. M. G. Jones (editors), The Biology and Cheniistry
of the Crueiferae. Academic Press, New York.

McKey, D. 1975. The ecology of coevolved seed dispersal systems. Pp. 159-191, in L.

E. Gilbert & P. IL Raven (editors), Coevohition of Animals and Plants. Univ. of Texas
Press, Austin.

PiMEMEL, D. 1961. Species diversity and insect population outbreaks. Ann. Entomol. Soc.
Amer. 54: 76-86.

RiiOAOES, D. F. & R. (;. Gates. 1976. Toward a general theory of plant antiherl)ivore
chemistry. In J. W. Wallace & R. L, Mansell (editors), Recent Advances in Phytochemis-
try. Vol. 10: 168-213.

Roor, R. B. 1973. Organization of a plant-artliropod association in simple and diverse
habitats: the fauna of collards {Brassica olcracea). Ecol. Monogr. 43: 95-124.

ScaooNHOvEN, L. M. 1972. Secondary plant substances and insects. In V. G. l^uneckles
& T, G. Tso (editors), Recent Advances in Phytochemistry. Vol. 5: 197-224.

Slansky, F., Jh. & P. Feeny. 1977. StabiHzation of the rate of nitrogen accunuilation by
lar\ae of the cabbage butterfly on wild and cultivated food-plants. Ecol. Monogr. 47:
209-228.

Smith, J. G. 1976. Infhience of crop background on aphids and other pli>tophagous in-
sects on Brussels sprouts. Ann. Appl. Biol. 83: 1-13.

South WOOD, T. R. E. 1973. The insect plant relatiousliip — an evolutionary perspective.
Pp. 3-30, in II. F. van Emden (editor), Insect Plant Relationships. Blackwell Scientific
Publications, Oxford.

Tahvaxaixex, J. O. & R. B. Root. 1972. The influence of \egetational diversity on the

population ecology of a specialized hcrbi\orc, PhijUotrcta crueiferae (Goleoptera:

Ghrysomelidae). Oecologia 10: 321-346.
Thohsteixsox, a, J. 1953. The chemotactic responses that determine host specificity in

an ohgophagous insect {PlutcUa maeuJipennis (Gint.) Lepidoptera). Ganad. J. Zool.

31: 52-72.
VAX Emdex, II. F. 1972. Aphids as pliytochemists. Pp. 25-43, in J. B. llarborne (editor),

Phytochemical Ecology. Acadenn'c Press, London.
VAU(;ifAx, J. G., A. J. MacLeod & B. M. C Jones (editors). 1976. The Biology and

Ghemistry of the Gruciferae. Academic Press, New York.
Vershaffelt, E. 1911. The cause detennining the selection of food in some herbivorous

insects. Proc. Acad. Sci. Amsterdam 13: 536-542.

ViHTANEX, A. L 1958. Antimikrobiell wirksame subslanzcn in kultiu-pflau/.en, Angew.
Ghem. 70: 544-552.

. 1965. Studies on organic sulphin- compounds and other labile substances in

plants. Phytochemistry 4: 207-228.

Walkeh, J. G., S. MoiiELL & H. II. Foster. 1937. Toxicity of nmstard oils and related sul-
fur compounds to certain fungi. Amer. J. Bot. 24: 536-541.

Westcott, G. 1971. Plant Disease Handbook. Van Nostrand Reinhold Co., New York.

YunKix, J. 1969. Archaeology and the nutritionist. Pp. 547-552, in P. J. Ucko & C W.
Dimbleby (editors), The Domestication and Exploitation of Plants and Animals. Duck-
worth, London.

232

ANNALS OF THE MISSOURI B0TANK:AL GARDEN

[Vol. 64

Appendix
Species of human food-plants listed by 100 students in the general ecology course (Bio.

Sci. 360, Fall 1976) at Cornell University, and categorized

a function of: (1) Growth

form of plant and (2) Part of plant eaten. Bulbs and tubers are inchided with roots; shoots,
stems and flower parts are included with foliage. Three species {Vitis vinifcray Beta vulgaris^
and Bra.ssica rapa) are listed twice, an. = annual, bien. = biennial, per. — perennial.

A. TREES

( i ) Root: No species listed,
(ii) Foliage: No species listed,
(iii) Fruit:

Mango

Mangijera indica

tree

Anacardiaceae

Pawpaw

Asimina triloba

small tree

Annonaccae

Papaya

Carica papaya

small tree

Caricaccae

Japanese persimmon

Diospyros kaki

tree

Ebenaceae

Avocado

Persea americana

tree

Lauraceae

Fig

Ficus carica

tree

Moraceae

Breadfruit

Artocarpus alfilis

tree

Moraceae

Banana, plantain

Mwsa acuminata and

tall per. herl)

Musaceae

Musa Xparadisiaea

tall per. herb

Musaceae

Conunon guava

Psidium guajava

small tree

Myrtaceae

Olive

Olca curopaca

tree

Oleaccac

Date

Phoenix daetylifera

tall palm

Palmaceae

Pomegranate

Punica granatum

small tree

Punicaceae

Quince

Cydonia ohlonga

small tree

Rosaceae

Apple

Mains putnila

tree

Rosaceae

Pear

Pyrus communis

tree

Rosaceae

Apricot

Prunus armcnica

small tree

Rosaceae

Sweet cherry

Prunus avium

tree

Rosaceae

Plum

Prunus domesiica

small tree

Rosaceae

Peach, nectarine

Prunus persica

small tree

Rosaceae

Sweet orange

Citrus sinensis

tree

Rutaccae

Grapefruit

Citrus Xparadi^i

small tree

Rutaceae

Nagami kumquat

Fortunella margarita

small tree

Rutaccae

(iv) Seed:

Cashew

Anacardium occidentale

tree

Anacardiaceae

Pistachio

Pistacia vera

small tree

Anacardiaceae

European filbert/hazelnut

Corylus avellana

small tree

Corylaceae

American filbert/hazelnut

Corylus americana

small tree

Corylaceae

Fm-opean chestnut

Castanea sativa

tree

Fagaceae

Beech

Fagus grandifolia

tree

Fagaceae

Pecan

Carya illinoetisis

tree

Juglandaceae

Hickory

Carya ovata and

tree

Juglandac(*ae

Carya laciniosa

tree

Juglandaceae

Butternut

Juglans cinerca

tree

Juglandaceae

English walnut

Juglans regia

tree

Juglandaceae

Brazil nut

Bertholletia excelsa

tree

Lecythidaceae

Coconut

Cocos nucifcra

tall palm

Palmaceae

Piiion

Pijws cemhroides

tree

Pinaccae

Almond

Prunus dtdcis

small tree

Rosaceae

B. SHRUBS AND PERENNIAL VINES

(i) Root:

Sweet potato (tuber)

Yam

Cassava 'manioc

Ground nut (tuber)

(ii) Foliage:
European grape

ipomoea oaiaias
Dioscorea spp.
ManiJwt escidenta
Apios americana

per. vme
per, vine
shrub
per, vine

i^onvoivuiaceae

Dioscorcaceae

Euphorbiaceac
Leguminosae

Vitis vinifera

per. vine

Vitaceae

1977]

FEEXY— DEFEXSIVE ECOLOGY

233

ArrLiNDix. ( contiinicd )

(iii) Fruit:
American elder
Huckleberry
Cranberry
Blueberry
Passion fruit/purple

granadilla

Red raspberry
Black raspberry
Loganberry, bo>senberry

Rose hip

American gooseberry

Red currant
Fox grape
European grape

Sanihficus canadensis
Gatjlussacia spp.
Vaccininm macrocarpon
Vacciniuni spp.
Passiflora edulis

Ruhus idacus
Ruhus occidcntalis
Ruhns ursinns
Rosa villosa
Rihes hiiicUum
Rihcs sativum
Vitis lahrusca
Vitis vinifcra

(iv) Seed: No species listed.

C. HERBACEOUS PLANTS

(i) Root:

Onion ( bulb )

Beet, sugar beet

Rutabaga

Turnip

Radish

Burdock

Jerusalem artichoke (tuber) Hcliatttlius tubewsus

Allititn cepa
Beta vulgaris
Brassica napus
Brassica rapa
Raphanus sativus
Arctitini lappa

Camass (bulb)
Potato (tuber)
Cattail
Carrot
Parsnip

(ii) Foliage:
Leek
Comfrey
Beet
Spinach
Cabbage, kale, etc.

Chinese cabbage

Water cress

Endive

Chicory

Artichoke ( flower bud and

scales )
Lettuce
Dandelion
Asparagus (young stem)

Rhubarb (leaf stalk)
Celery (leaf stalk)
Fennel ( leaf stalk )

(iii) Fndt:
Pineapple
Sunflower
Wateniielon
Melon
Cucumber

Camassia quaniash
Solatium tuberosum
Typha spp.
Dauctts carota
FaMtiiuica sativa

Allium ampcloprasum
Symphytum officinale
Beta vulgaris
S}}i}mcca oleracea
Brassica oleracea
Brassica rapa
Nastuiiitim officinale
Cichorium endivia
Cichorium intyhus
Cyjiara scolymus

Lactuca sativa
Taraxacu m officinale
Asparagus officinalis
Rheum rhaharharuni
Apium graveolens
Foeniculum vulgare

Anatuis comostis
Ilelianthus antiuus
Citrullus lanatus
Cucumis inelo
Cucumis sativus

Squash, pumpkin, zucchini Cucurhita spp.

shrub
shrub
shrub
shnd^

per. vine

shrub

shrub

shrub

shrub

shrub

shrub

per. vine

per. vine

per.

an./bien.

an./bien.

an./bien.

an./bien.

per.

per.

per.

per.

per.

an./bien.

bien.

bien.

per.

an. bien.

an./bien.

an./bien.

an./bien.

per.

an./bien.

per.

per.

an./bien.

per.

per.

per.

bien.

an. /per.

Strawberry

Green pepper, chili

Fragaria X ananassa
Capsicum annuum

per

4

an.

an.

vine

an.

vine

an.

vine

an.

vine

per,

1

an./

'per.

Caprifoliaccae

Ericaceae

Ericaceae

Ericaceae

Passifloraceae

Rosaceae

Rosaceae

Rosaceae

Rosaceae

Saxifragaceae

Saxifragaceae

Vitaceac

Vitaccae

yVmaryllidaceae

Chenopodiaceae

Cruciferae

Cruciferae

Cruciferae

Compositae

Compositae

Liliaceae

Solanaceae

Typhaceae

Umbelliferae

Umbelliferae

Amar>'llidaccae

Boraginaceae

Chenoi^odiaceae

Chenopodiaceae

Cruciferae

Cruciferae

Cruciferae

Compositae

Compositae

Compositae

Compositae

Compositae

Liliaceae

Polygonaceae

Umbelliferae

Umbelliferae

Bromeliaceae

Compositae

Cucurbitaceae

Cucmbitaceae

Cucurbitaceae

Cucmbitaceae

Rosaceae

Solanaceae

234

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Appendix. ( continued )

Tonuito
Eggplant

(iv) Seed:
Peanut
Soybean

Lentil
Lima bean
Kidney bean
Garden pea

D. GRASSES

Lycopersicon hjcopersicum an. /per.

Solanum melongena

Arachis htjpo^aea
Glycine max
Lens culinaris
Phaseohis limcnsis
Phaseohis vulgaris
Pisui)i sativtim

(i) Root: No species listed,
(ii) Foliage:

Bamboo (young shoots)
Sugar eane (stems)

Phyllostaehys spp. and

Bamhusa spp.
Saccharum officinarttm

( iii ) Fruit: No species listed,
(iv) Seed:

Oat

Barley

Rice

Broomcorn/millet

Rye

Sorgbum
Common wheat
Corn

Avena sativa
Hordeum vulgare

Oryza sativa
Panicum miUaccxim

Secale cerealc
Sorgluim hieolor
Triticum aestivum
Zea mays

an. /per.

an.

an.

an.

an. /per.

an.

an. vine

per.
per.
per.

an.

an.
an.
an.

an.
an.
an.
an.

Solanaeeae
Solanaceae

Leguminosae
Leguminosae
Leguminosae

Legmninosae
Leguminosae
Leguminosae

Gramineae
Gramineae
Gramineae

» 11 i »

Graminc

Graminc

Gramineae

Gramineae

Gramineae

Grann'ne

Grannnt

Graminc

<i£»

*'j/*

^ '1 £\

CHEMOSYSTEMATICS AND ITS EFFECT

UPON THE TRADITIONALIST'

B. L. TUUNER-

What is a traditionalist, taxonomically speaking? I suppose a traditionalist
mi^ht best he defined as a taxonomist trained as a pheneticist, practicing his
trade as a pheneticist, and constructing his classification using primarily phenctic
data. By this definition I am a traditionalist and consequently can claim to
answer, for myself^ the effect of chemosystematics upon my own traditional at-
titudes and outlooks. And this has been profound.

I say profound not because this new field has solved any large number of
critical problems in plant taxonomy, but because where it has been used with
skill and judgement, it has proved luuch more effective than phenetics in solving
the particular problems concerned. Indeed, without chemical data many of the
more intractable problems having to do with familial relationships among flower-
ing plants generally are likely to remain unresolved: there are simply too many
cooks and nearly all with varying tastes. Even if they all see the same phenetic
substances in the phyletic cabinet, they nonetheless are prone to come up with
different combinations of this or that ingredient (selected characters), with
varying amcmnts (intuitive weighting), to say nothing of the condition (basic
LQ.) or temperature (zealousness) of the oven (i.e., brain).

I suspect that most traditionalists, even some of the best, do not like to be
reminded that their approach is fraught with such variables, or that data de-
rived from some other discipline might prove superior to those from their own.

As an example, when the late Dr. Alston and I first showed the utility of
paper chromatography for resolving problems of natural hybridization in Bap-
tisidy an eminent, not so classical, plant systematist suggested that our documen-
tation of complex hybridization in this genus could have been accomplished with
equal clarity using selected morphological characters arranged upon Anderson-
type scatter diagrams. Needless to say this intellectual guffaw was issued by the
late Edgar Anderson, and the ironic part of all this is that Anderson himself was
the first to collect and call attention to the existence of hybrid swarms among
this group of plants (Anderson, in Larisey, 1940b), but he failed to perceive its
complexity, in spite of the fact that he collected his hybrid populations of Bap-
tisia in a region where the potential for trihybridization is not infrequent (Alston
& Turner, 1963). In fact, I seriously doubt that Anders(m, or any traditioiml sys-
tematist, including myself, would have been able to recognize, much less intuit,
trihybridization within this group, to say nothing of its documentation with rea-
sonable certainty using morphological characters.

Trihybridization, of course, is rather the exception in nature: most species
tend to comingle two at a time at any one site. But even then, lacking in situ
clues (for example, two parental taxa occurring together with their putative

' Supported, in part, by NSF Grant DEB 76-09320.

'Botany Department, The University of Texas, Anstin, Texas 7S712.

Ann. Missouri Bot. Gaud. 64: 235-242. 1977.

236 AXNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

hybrids, as liappens with Baptism upon occasion), two-way liybrids may be
difficult to detect, especially where these are quite distinct and relatively widely
distributed. Hence we find the southeastern taxon, Baptisia serenae Curtis, be-
ing recognized as a good species for over 125 years by a wide range of workers,
including such an outstanding traditional worker as Wilbur (1963). It is, how-
ever, an Fi hybrid of B. tinctoria (L.) Vent. X B. alba (L.) Vent. A more re-
markable Fi hybrid, also long-recognized as a good species by nearly all tradition-
alists, including the only recent monographer of the genus (Larisey, 1940a) is
B. microphyUa Nutt., this being the relatively rare hybrid between B. perfoliata
(L.) R, Br. and B. laticeohta (Walt.) Ell. While it is perhaps likely that these
very distinct Fi hybrids would have been detected if they were found growing
with their putative parents, reasonable verification, short of long and laborious
crossing experiments, would be difficult. If, however, their flavonoid profiles
were sufficiently different, putative Fi hybrids might be readily confirmed,
even from hybrids mounted on herbarium sheets up to 100 years old.

Speaking of the superiority of micromolecular data for the resohition of sys-
tematic problems, tlie most telling example of its efficacy is that involving the
detection of allopatric introgression. Edgar Anderson (1949) in an Epilogue to
his brilliantly conceived text, Intro<^ressive Hybridization^ made the following
statement:

How important is introgressive hyhridization? I do not know. One point seems fairly
certain: its importance is paradoxical. The more imperceptible introgression becomes, the
greater is its biological significance. It may be of the greatest fundamental importance when
by our present crude methods we can do no more than to demonstrate its existence. When, on
the odier hand, it leads to bizarre hybrid swarms, apparent even to the casual passer-by, it
may be of little general significance .... Only by the exact comparisons of populations can
we demonstrate die phenomenon .... Tlie wider spread of a few genes (if it exists) might
well be imperceptible even from a study of population averages, but it would be of tremen-
dous biological import .... Hence our paradox. Introgression is of the greater biological sig-
nificance, the less is the impact apparent to casual inspection.

In Other words, in well-differentiated, sympatric species such as Baptisia
where natural hybridization can be easily recognized and readily documented,
its biological impact on evolutionaiy processes is negligible. But in allopatric
situations where hybridization is very difficult to detect it is likely to be of the
greatest biological significance.

In spite of these reflections from the foremost proponent of introgressive
hybridization, few, if any, well-documented studies have been forthcoming on
allopatric introgression. In fact, the best documented case in the literature for
allopatric mtrogression is reportedly that involving Juniperus virginiana L. and
/. asJwi Buchh. (Anderson, 1953; Davis & Heywood, 1963). However, in a num-
ber of detailed studies, centered at The University of Texas (using 60 to 80 chem-
ical characters as detected by gas chromatography as well as morphological
characters which distinguish between the taxa), the existence of Fi hybrids or
their immediate derivatives could not be detected, even at sites where large
populations of both species grew intermixed, and in no instance could the exis-
tence of introgression be inferred from the data accumulated ( Flake, von Rud-
loff & Turner, 1969; Adams & Turner, 1970; Flake, von Rudloff & Turner, 1973).
In short, what was taken to be a very well-documented case study of allopatric

1977] TURNER— CHEMOSYSTEMATICS AND TRADITIONALISTS 237

introgrossion turned out to be a situation in which dinal intergradation in habital
features over a broad region occurred such that, siiperficiaUij, hybridization and
introgression might be inferred.

In liindsight, it now seems rather reasonable to liave viewed the case study
of introgression between /. virginiana and /. ashei with considerable doubt, for
the two species are readily distinguished by a number of morphological features
and are placed in different species — groups (sections) of the genus, and the
character used for such taxonomic segregation (cilia along the leaf margins)
does not, to our knowledge, segregate in putative hybrid swarms (i.e., the two
species can always be recognized by this feature, and others, as attested to by the
repeated correlation of this character with a plethora of chemical characters);
other experienced field workers such as D. S. Correll (pers. comm.) have also
had no difficulty in placing the plants concerned in one taxon or the other. In
fact, as already indicated, the moiphological variation found in /. virginiana is
clinal, i.e., the species has formed or is in the process of forming regional races
as a result of adaptational mechanisms arising out of its own gene pool, this be-
ing unrelated to the possible influx of genes from the largely allopatric /. a^^hei.
Our work has substantiated fully these suppositions (Flake & Turner, 1973).
Again, it is ironic that Hall, who was Anderson's student, should have docu-

mented introgressive hybridization where this was not occurring. We attribute
this to the plasticity of the morphological characters selected for its detection.
It was the absoluteness of the chemical data themselves which permitted reso-
lution of the problem.

What we were left with then was no well-documented case study of al-
lopatric introgression of a regional nature. Fortunately, however, there has been
a recent, carefully conceived, populational study of /. virginiami and /. scopulorum
Sarg. in the Missouri River Basin of the north central United States by Van
Haverbeke (1968a) which appears to be a situation involving allopatric intro-
gression of the type Anderson felt to be so important in evolutionary processes.
The study seems to be unusually well documented. Van Haverbeke made very
accurate records of the populational sites, including precise data on ten indi-
vidually marked trees which were selected for study at each site. These included
photographs and detailed field notes. In short, the /. virginiana-J , scopulorum
complex appeared to provide an ideal case study of allopatric introgression using

the chemonumerical methods that proved so effective in disproving the oc-
currence of this phenomenon in the /. ashei-J. virginiana "complex."

Van Haverbeke (1968a), through his study of these two taxa in the Missouri
River Basin, has stated that:

The entire Junipcrus population witliin the Basin is apparently of hybrid deri\ati()n with
neither of the extreme parental types hein^ found. There is a trend of inereasing h\l)rid index
values (also percentage germ plasm values) from southeast to northwest over the l^asin
from the reported range of /. virginiana to and into the reported range of /. scopulorum. This
condition may be the result of bilateral introgression between the two species. Ther(^ was,
however, a strong tendency toward bimodality within the population as demonstrated by the
presence of two distributions in each of the three hybrid indices. Tliis indicated the
presence of two different species — /. scopulorum and /. virginiana.

While Van Haverbeke (1968b) admits his data might be interpreted as

238 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

constituting evidence for introgression he, nevertheless, suggests, indeed cham-
pions, an alternate hypothesis:

As an alternative interpretation, it would seem that beeause of the greater diversity of
the junipers in western Nortli America, that /. virgiruana was at some time derived from this
area. It seems possible tliat with the inherent variabihty in the germ plasm ancestral to both
/. scopttloniin and /. virginiana, that prox:)agules could flourish in sites toward the east. This
could have initiated an eastward migration-propagule — which througli mutation and selection
eventually became what we now recognize as /. virginiana"

It should be noted that this latter evolutionary model is in direct conflict with
that proposed by us (Flake, von Rudloff & Turner, 1969, 1973) in which we
suggest that the Appalachian Region is the ancestral center for the origin of /.
virginiana and its various races. Hence, the question of introgression between
/. virginimiu and /. scopulomm is left open by Van Haverbeke's study.

Initial investigation of the terpenes of Juniperus scopulomm, unlike that of
/. ashei, showed tliat its volatile components were essentially those of /. virginiana,
differing only in their quantitative expression. Subsequent populational analysis
of the type employed in the /, ashei-]. virginiana studies showed that regional
intergradation of the chemical characters occurred across the Missouri River
Basin, much as found by Fassett ( 1944) and Van Havcrbeke (1968a) for morpho-
logical features.

Three models might be proposed to account for the variation found in this
region:

1. ANCESTRAL GENE POOL — Jwiipenis scopulorwu and /. virginiana may
have arisen from ancestral populations largely endemic to the Missouri Ri\^cr
Basin. Subsequent evolutionary divergence to the west and east, respectively,
might have occurred, leaving a residuum of genes common to each in the area
concerned.

2. ALLOi'ATHic INTROGRESSION — The variability is due to extensive gene
flow from /. scopulomm into /. virginiami as a result of hybridization and
backcrossing in pt^ripheral regions of contact and areas of sympatry.

3. MIGRATORY TAILINGS — The River Basin was an ancestral migratory
route through which /. scopulorum-Mke populations passed on their way to be-
coming what is now known in the eastern United States as /. virginiana. In Van
Haverbeke's words (1968b), *Thus, rather than being considered as an intro-
grcssive series, this juniper population [those of the Missouri River Basin] can
alternatively be interpreted as a divergent evolutionary series which has not
yet completely separated."

It should be emphasized that in the investigation by Van Haverbeke about
40 morphological characters were selected for measurement and numerical analy-
sis. These were obtained from some 700 trees from 72 sites scattered throughout

the River Basin area. In spite of this excellently conceived, carefully documented,
laborious study, the investigator was unable to decide, unequivocally, between
models 2 and 3; in fact, he believed that his data best fit the migratory tailings
model. (Model 1 was not tested, j)resumably because of its implausibility, con-
sidering the biogeographic history of the Basin region.)

Our own study (Flake, Urbatsch & Turner, 1978) also involved about 40
characters, all chemical. These were obtained from some 200 trees from 10 sites

1977] TURNER— CHEMOSYSTEMATICS AND TRADITIONALISTS £39

systematically selected at about 150-niiIe intervals in a southeast-northwest
transect across the Basin. In spite of the fewer populations sampled and the
smaller overall sample size, we conclude our data overwhelmingly suggest that
the variable River Basin populations are the result of allopatric introgression,
primarily in the direction of /. virginiana, much as Van Ilaverbeke thought niiglit

be the case, but the morphological characters which he used were not sufficicntl)
indicative to jirove decisive.

Macromolecular Approaches

If I were interested in obtaining the most meaningful arrangement of present-day angio-
sperm families, phylogenetically speaking, I would rather have available to me tlie primary
structure (anu'no acid sequence) of ten metaholicalhi important enzymes (such as cytoclnome
c) of all of the taxa which comprise these groups than ha\e a detailed listing of all of the
exomorphic features which characterize the groups (Turner, 1969).

The nature and proper taxonomic x:)osition of the Inpothetical past organisms that rei^rc-
sent the branch points in tlie scheme cannot be determined solely from the phylogenetic
relationships of modern species as deciphered from the anu'no acid sequences (CroiKiuist,
1976).

Protein sequencing and other molecular methods may, in fact, become in the near future
the most powerful tools for the study of phlogeny ( Ayala, 1976).

The amino acid sequence trees are obviously more compatible with some possible phy-
logenetic interpretations than others, ox there would be no point in making them at all. If we
assumed that they wx^re in all respects correct insofar as they go, they would place certain limits
on the general phylogenetic trees that could be seriously considered (Cronc^uist, 1976).

Though this ])e madness, yet there is method in't (Shakespeare, Hamlet, Scene II, Act 2).

. . . fossil evidence is highly in accord with an overwhelming mass of evidence from com-
parative morphology of living species that the Magnoliidae are the most primiti\e (i.e., least
modified) group of living angiosperms . . . (Croncjuist, 1976).

. . . the molecular tree indicates that present-day families represent relic groups wlu'ch
have for the most part had a long separate e\<)lutionary history. They do not support the sug-
gestion implicit in, for example, Cronc^uist's scheme . . . tliat the Magnoliidae gave rise to
the CaryophyUidae on the one hand, and to the Rosidae on the other, the latter, in turn, giv-
ing rise to the Asteridae ( Boulter, 1973).

Something is rotten in the state of Denmark (Shakespeare, Hamlet, Scene IV, Act I).

\\'ith rekitively few exceptions, the traditionalist might yawn at the seemingly
trivial impact of micromolecular data npon his various systematic models. But
he has not yet been able to treat with indifference the likely impact of macro-
molecular data upon his most treasured erection, the "Tree" to plant families.
As unrecognizable as this tree might l)e to the various workers concerned, any
reinsertion of branches or elevation of roots, using such cliemical data, is met
with alarming cries from this or that proponent. I refer specifically to the recent
paper by Cronquist (1976) entitled, "The Taxonomic Significance of the Struc-
ture of Plant Proteins: A Classical Taxonomist's View." This is a 27-page ram-
bling review covering the whole field of comparative cMizymology, the gist of

which is, because these data do not or have not supported my particular views,
there must be something wrong with the approach.

The approach is the same as that which has been applied to animals success-
fully, namely, the use of anu'no acids among the homologous proteins in different

240 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

organisms as an indicator of time of branching. And, strangely enongh, he ac-
cepts, in principle, the use of cytochrome c as a reasonable, but often unsteady,
clock for animals, yet rejects this as valid for plants. I quote;

Given the difference in evolutionary pattern between plants and animals, it sliould not
be surprising if tlic animal physiological system places stronger constraints on the acceptance
of amino acid substitutions in c>tochrome c than does the plant physiological system. It would
be entirely in harmony with the other differences in plant and animal evolution if tlie s
kinds of changes could be accepted by very different sorts of plants and if back mutations were
not notably counter-selective.

Cronquist focuses his attack largely upon the data from Boulter's laboratory
in Durham, England, which is the only group to sequence any significant num-
ber of plant proteins, namely plastocyanin and cytochrome c, Amino acid se-
quences from the latter, in particular, suggest that the familial tree is quite dif-
ferent from the one proposed by Cronquist (and, of course, that of Takhtajan, the
two being quite similar). This is disturbing: everyone should accept that the
Magiwliidae among the angiosperms is primitive to everything else. He does not
like the Caryophyllidae coming off as a first branch on the cytochrome c familial
tree. He does not like to think of the moi-phologically highly advanced Com-
positae represented as a very old isolated branch; everyone should know that it
is recent, going back to the Miocene-Oligocene boundary (in spite of the fact

very

lished pollen fossil data might push the family back to the Paleocene, if not ear-
lier ) .

adit

ubiquity of this macromolecule among organisms generally and cognizant of
its crucial role in the metabolic pathway of both plants and animals, should at-
tribute the discrepancies to poor or erratic functioning of this kind of clock,
rather than to the moiphological data, which, after all, has no face, no dial, no

rne

<£

Cronquist (1976: 5), while accepting tlie general premise that the cytochrome
c clock works for animals, nevertheless makes great gloat over the fact that the
amino acid sequence of rattlesnake cytochrome c is out-of-line with the position
of that organism in the phyletic tree. There follows a typical Cronquistian quote,
If the reported sequence for rattlesnake is correct, there seems to be no easy way
to explain it, short of conjuring up a vision of a lonesome cowboy on the lone
prairie, with none but a rattlesnake for company [referring to the seeming simi-
larity of its sequence to that of the genus Homo] J' I think that there are better
ways to explain that single discrepancy, even // the sequence is correct,

Cronquist presumably wants us to believe that an occasional unsteady tick (if

even that! ) in the animal world is sufficient reason to belie\T that this same clock

is largely unsteady in the plant world. In his desire to discredit such data, at least

that of cytochrome c, he likens this to the Age and Area concepts of Willis (ludi-
crous!), followed by the statement that:

Evolution of other characters in hotli plants and animals tends to undergo periods of rapid
radiation, interspersed with periods of more gradual change, and there is no a priori reason to
suppose that adaptively significant changes in amino acid sequence would proceed any dif-

ferentlv.

i977] TURNKH-CMKMOSVSTEMATICS AND TRADITIONALISTS £41

Of course, that's the point; there has been a sufficiently long record of plant
evolution so as to believe that the cytochronie c clock has some kind of accuracv
tast or slow upon occasion, it nonetheless seems to average out as stochastic over
time. Anyway, a generally erratic clock is better than no clock, giving tlie muddle

rph

most plant taxonomists work.

As a final denouement, in case he hasn't convinced the reader, Cronciuist adds
a neat punch paragrapli:

Tills discussion of the exolutionary clock may be something like heatinjr a dead horse, but
some people are still trying to ride the horse. If the horse is really dead it won't nu'nd the
beating.

One should perhaps remind Cronquist that, to judge from the recent articles
by Fitch (1976), Zuckerkandl (1976), King (this symposium), and articles in
press by yet other such w^orkers, the horse is alive, is being ridden (juite nicel\',
and perhaps doesn't deserve the beating being administered!

It would be unfair to conclude this address with the audience feeling that
Cronquist might be quite negative towards the application of cliemical data to
taxonomic problems. He is not, for he concludes, in hindsight, that:

I welcome the appearance of amino acid s(Mjuences as an additional tool for taxonomists
. . . . When we have the setiuences for sexeral proteins from members of a wide range of fanu'-
lies, inchiding critically important ones, \\t» can make good use of this powerful tool.

Let\s hope he means this; in the meantime he might wish to paraphrase
Shakespeare's King Richard the Third, "A dead horse, a dead liorse, My Kingdom
for a dead horse!"

LrrEHATUHK CrrKi)

Adams, R. P. & B. L. Tuuxkh. 1970. Chemosystematic and numerical studies of natural

populations o{ Junipenis ashei Bush. Taxon 19: 728-751.
Alstox, R. E. & B. L. TuHXEH. 1963. Natural hyhridization anum^ four species of Ba\)tisia

(Legunu'nosae). Amer. J. Bot. 50: 159-173.
AxDEUsoN, E. 1949. Intro^nessive Hyhridization. John Wih^y & Sons, New York.

-. 1953. Intro^ressive Hyhridization. Bioh Rev. 28: 280-307.

Ayala, E. J. 1976. Molecular genetics and evohition. Pp. 1-20, in F. J. Ayala (editor),

Molecular Evolution. Sinauer Assoc. Inc., Sunderland, Massachusetts.
BouuTEH, D. 1973. Amino acid sequences of cytochrome c and plastocyanins in phylo^^enetic

studies of higher plants. Syst. Zool. 22: 549-553.

Choxquist, a. 1976. The taxononu'c significance of the structure of plant proteins: a classi-
cal taxon(miist's view. Brittonia28: 1-27.

Davis, P. W, & V. H. Hkvwood. 1963. Principles of Angiosperm Taxomnny. Oliver and
Boyd, Edinburgh.

Eassett, N. C. 1944. Junipcrus vir^iniana, /. liorizontalis and /. sco])ulonini. Bull Torre\'
Bot. Club71: 475-483.

EiTCH, W. M. 1976. The inolccular evolution of cvtocluome c in eukarvotes. J. Molec.
Evol.8: 13-40.

Elakk, R. H. & B. L. TunxKK. 1973. Volatile constituents, especially teipenes, and their
utility and potential as taxonomic characters in populational studies. Nohel Svmp 25-
123-128,

, L. UuHATscH ik B. L. TuHXEH. 1978. Chemical documentation of ahopatric intro-

gression in junipcrus scoptdorum and /. virginiana. (In press.)

, E. vox RuDi.oKF & B. L. TuHXER. 1969. Quantitative study of clinical variation in

Junipcrus virginiana using terpenoid data. Proc. Natl, Acad. U.S.A. 64: 487-494.

— , & . 1973. Confirmation of a clinal pattern of chenn'cal differenti-

ation in Junipcrus vir<^iniana from terpenoid data obtained in successi\f vears. Hect^it
Adv. Phvtocheni. 6: 215-228.

242

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol,. 64

Laiusey, M. M. 1910a. A monograph of the genus Baptisia. Ann. Missouri Rot. Card. 27;

119-258.

19401). Analysis of a hybrid complex between Baptisia leucantha and Baptisia

viriiUs in Texas. Ainer. J. Bot. 27: 624-628.
Tui<>4KH, R. L. 1969. Chemosysk'inatics: recent developments. Taxon 18: 134-151.
Van IIavkubeke, D. F. 1968a. A popnlation analysis of ]tnuj)crus in the Missouri River

Basi!i. Univ. Nebraska Stud., N. S., 38: 1-82.

19681). A taxonomic analysis of Juniprrus in the central and northern great plains.

Troc. Sixth Centr. States Forest Tree Iniprov. Conf. 6: 48-52.
WiLiiun, R. I.. 1963. The It^guniinons phuits of North Carolina. North Carolina Agric. Exp.

Sta! Tech. Bull. 151: 1-294.
ZucKKHKANDL, E. 1976. Evolutionary processes and evolutionary noise at the molecular
level. II. A selectionist model for random fixations in proteins. J. Molee. Evol. 7: 269-
312.

SYSTEMATICS OF MORAEA (IRIDACEAE

IN TROPICAL AFRICA^

Peteh Goldblatt-

Ahstoact

Moraea, with 24 species, is well represented iii tropical Africa, although the center for
the genus is to the south in the winter rainfall region of southern Africa. Nine species are
described for the first time. (A/, callisia Goldhl., M. afro-orientale Coldbl,, M. iriiigcnsis
Goldhl., A/, inyangani Goldhl., M. angolensis Goldbl., A/, tanzanica Goldbl., A/, upcrnhaua
Goldbl., A/. brevifoUa CJoldbl., A/, halundana Goldbl.) Only two of the five subgenera of
Monica occur in tropical Africa, subgenus Viciissenxia and subgenus Gramliflora, the latter in
particular exhibiting considerable radiation in south central Africa. On the basis of linu'ted
knowledge of morphological and cytological variation patterns in tropical Africa, Monica is
believed to be of recent origin here. Details of ecology, c>tology, and evolution are elaboratt^d
and compared to patterns in southern Africa.

Moraea is a large genus of some 95 species, occurring throughout sub-Saharan
Africa. The genus comprises small to meditun-sized herbaceous geophytes. It
is found mainly in highland areas in the tropics, usually in well-watered grass-
land, but also in open woodland or in marshy places. In temperate southern Af-
rica, Moraea occurs at all elevations. Moraea is of some economic importance
as all species are to some degree toxic to stock, section Polyanthes particularly
so. Only a few species luive been pr()\'en toxic, but all should be assumed to be
until shown otherwise. The present revision treats only the tropical African
members of the genus and will complete my taxonomic study of Moraea, begun
in 1973 with a revision of Moraea in the summer rainfall region of southern Af-
rica, followed l)y a revision of Moraea in the winter rainfall region of southern
Africa (Goldblatt, 1973, 19761)).

This treatment includes as tropical Africa the area south of the Sahara and
north of the region covered by the Flora of Southern Africa [Namibia (South
West Africa), Botswana, Lesotho, Swaziland, and South Africa]. In this area,
extending from Rhodesia north to Nigeria and Etliiopia, there are 24 species of

Moraea, a larger number than previously recognized, but low in comparison to

the 76 species in southern Africa. Only five species occur in both tropical and
the southern African sxunmer rainfall region, while one species, M, spathulata,
occurs in the winter rainfall region of southern Africa as well. The present work
is the first modern comprehensive treatiiient of Moraea in tropical Africa and
the only treatment since Baker's (189(S) rexision for the Flora of Tropical Africa,

Relati()nshii>s

The affinities and relationships of Moraea are discussed in detail in my re-
vision of the species found in the winter rainfall region of southern Africa (Gold-

^This study was supported by graut BMS-74-189()5 from the U.S. National Science
Foundation. I thank Jean Pawek for proxitling nuitcri'al for cytological study and for her gen-
erous assistance during field work in \hilawi. I also thank Janet Klein for the illustrations and
my wife Margaret for assistance in the field and with the manuscript.

"^ B. A. Krukoff Curator of African Botany, Missouri Botanical harden, 2345 Tower Grove
Avenue, St. Louis, Missouri 63110.

Ann. Missouri Bor. Gaud. 64: 243-295. 1977.

244

ANNALS OF THE MISSOURI BOTANICAL OAHDKN [Vol. 64

])latt, 19761)) and also in a paper dealing with the cytology and snbgeneric class! -

ded

d

these two genera probably had a common origin from ancestors of the genns
DU'tcs. Moraea is, however, more closely related to a gronp of Sonth African
conn-bearing genera which, like Moraea, also liave a secondary bifacial leaf
than to Iris or Dictes. These genera include Ilomeria, Ilcxa^httis, Ctjnandriris,
Galaxia, and BarnardieUa which are placed with Moraea in Ilomerinae (C;old-
blatt, 1976a), one of the three snbtribes of Irideae, a predominantly Old World
tribe of subfamily Iridoideae.

Geochai'iiy of the Thopical African Species

Although Moraea occurs throughout sub-Saharan Africa, its present center
is clearly the winter rainfall region of southern Africa which extends along the
south and west coast of the Cape Province. Not only is this area richest with
54 species, 48 being endemic, but it is also the center of diversity in the genus.
All five subgenera occur here and three of these are essentially endemic. The
two largest subgenera Vieusseuxia and GrandifJora, also considered the more
advanced, extend well outside the winter rainfall region. In fact, the center for
subgenus Qrandiflora lies outside this area, in the mountains of southeast Africa,
extending from Natal to Malawi, Tanzania, and Zaire.

Subgenus Viem.seuxia, the second subgenus found in tropical Africa, com-
prises two sections. Section PoJijanthes, the less specialized, occurs from the
southern Cape to Ethiopia and is well represented in Tanzania, as well as in the
South African provinces of Natal and the Cape, The most primitive members of
section PoJyanthcs are believed to be those with several leaves and unlimited
branching, rather than single-leafed, few- or unbranched species; these include
M. polyanthos and M. pohjstacJiya from the Cape Province and M. carsonii from
Rhodesia-Zambia-Malawi. The center for this section thus may also lie in the
mountains of southeast Africa but perhaps more likely to the south along the
interface between the summer and winter rainfall regions. The second section
of subgenus Vieusseuxia, section Vieusseuxia, is clearly specialized (Cxjldblatt,
1976b) and does not occur outside southern Africa where it is concentrated in the

winter rainfall region.

lu tropical Africa, Moraea occurs in almost all highland areas above 1,200 ni

and concentrations of species are found mainly in isolated areas that are con-
siderably higher. Representation of the genus is notably poorer north of the
equator with only two species in Ethiopia, three in the Uganda-Sudan-Kenya
hi<'hlands, and only one in the mountains of eastern Nigeria and Cameroons.
Significant areas of concentration are all in the highlands of central tropical Af-
rica and se\'eral endemics occur in these isolated, and well-watered montane and
semimontane regions. The main centers of endemism are as follows:

1. Inyanga highlands of Rhodesia: 5 species, one endemic (M. inyan<:,ani).

2. Northern Malawi-eastern Zambia escarpment and Southern Highlands
of Tanzania: 8 species, 3 endemic (A/, tanzanica, M. callista, M. irin^cnsis).

1977] G()LI>BI.ATT— A/OfiAFA

245

3. North-central Zambia, southern and eastern Shaba: 14 species, 5 endemic
(A/, hrevifolia^ M. iinifoliata, M. halundana, M. iKwonei, M. upemhana).

Subgenus Vieusseuxia is predominantly eastern in distribution and thougli
it occurs from Ethiopia to Rhodesia, it does not extend anywhere in tropical
Africa west of longitude 24''E, and is thus absent from Angola and the Nigeria-
Cameroon highlands. Almost as wide ranging as subgenus Gramliflora in tropi-
cal Africa, subgenus Vieusseuxia is poor in species, with only 7 species north
of South Africa, and only 4 of these exclusive to tropical Africa. Significant
speciation in subgenus Vieus.scuxia occurs only in the Southern Highlands of
Tanzania where two very striking species, A/, callista and M. irin<^ensis, are en-
demic. Other species of subgenus yieussexixia are widespread, with M. thom-
sonii extending over almost the entire range of the subgenus in tropical Africa,
and found from Ethiopia in the nordi to Transkei in southern Africa.

Subgenus Grandiflora is well represented in tropical Africa by 16 species,
only two also foimd in South Africa where an additional 11 species occur. Tlie
only really widespread species of this subgenus in tropical Africa is M, schimperi
which extends over the whole range of Monica in tropical Africa. Other species
are considerably more restricted, and several species are known from single
collections or a very small area. Moraca iuij(in<^(ini is one example while several
species in the M. tanzanicci-M . unifoUata alliance of reduced species are also
very local, especially M. nnifoUata, M. balundara^ M. ])ovonei, all from Shaba,
Zaire, A/, hrcvifolia in northeastern Zambia, and A/, tanzanica in southwestern
Tanzania-northern Malawi. A very marked center of speciation for subgenus
Grandiflora is evident in this belt across south-central tropical Africa.

HisroRY

Tlie first collections of Moraca from tropical Africa were made as earl}' as
the 184()s when Schimper discovered A/, schimperi in Ethiopia. Schimper's
collections were first assigned to two separate species in the genus Utjmcno-
sti<ima. Knowledge was later extended by \\'elwitsch who collected A/, schimperi,
M. clavafa, and A/, tcxtilis in Angola in the years following 1853. Welwitsch's
Angolan Iridaceae were only descril)ed much later, in 1878, by Baker who actu-
ally admitted seven species from Angola including tliree now regarded as con-
specific, and referred to Ferraria rather than Moraca.

The important milestone in the knowledge of Moraca in tropical Africa is
J. G. Baker's (1898) treatment for Fhra of Tropical Africa. Baker recognized
16 species in his treatment. However, four of these are clearly tlie same species
of Ferraria, F. <i}utim)sa, one is today regarded as Dictes, while of the remaining
11 species A/, wehcitschii. A/, zamheziaca, and A/, divcrsifolia are conspecific
{A/, schimperi) and another two, Ai. textilis and A/, mechowii, are also regarded
here as the same species. A last species, M. ])eUa, was until recently associated
with the Cape endemic A/, angiista, although it is in fact unrelated. By
only seven of the pr(\sently recognized species were known, six were nam(*d and
one misidentified.

Subse(pie!»t studies on Moraea were carried out ind(*pendently by German,

246 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

liolgian, and liritish botanists who confined their interests largely to the colonies
of their respective conntries. In particular, de Wildeman, working on the flora
of what is now Zaire and Burundi, described six species, all new for the Congo,
but four of which have proved to be synonyms of earlier species from other parts
of tropical Africa. More recently, Moraea has received very little systematic at-
tention in tropical Africa. In fact, the recent treatment by Geerinck (1970) for
Zaire and Burundi is the only significant work in the last fifty years. Geerinck
recognized six species, one tentatively; and subsequently one more (Geerinck,
1972). Ceerinck's species concepts were far broader than those held by me and
I recognize several more species in the Zaire-Burundi area than Geerinck did.
In the present revision 9 of the 24 species are new to science. Knowledge of the
genus in tropical Africa is, however, far from satisfactory, and several species are
known from one locality while collections of some species are incomplete, lack-
ing conns or fruits or adequately preserved flowers.

Ecology

HAIUTAT

Monica occurs in two distinct types of habitat in tropical Africa, either in
dry situations in grassland and open woodland or in wet, marsh situations known
in southern Africa as vleis, and in south tropical Africa as dambos. Dambos may
be seasonal or permanent, but have the characteristic of being poorly drained
yet seldom deep, and usually have a rich vegetation including many grasses,
sedges, as well as geophytes and herbaceous dicots. The dry habitat may com-
prise more than a single niche for Moraea, but data available does not indicate
this, rather suggesting that species may be fairly tolerant of minor ecological
differences and thus growing equally well in open woodland or grassland of
various types. Most species, though not all, appear to be restricted to either one
or other of the two major habitats. Records suggest that A/, vcntricosa and
A/, textilis are exceptions, as they are listed as having been collected in wood-
land, grassland, or on the edges of marshes. Personal observation has shown that
M. schimperi also occurs from wet marshes to open, well-drained grassland, the
latter habitat being occupied only in areas of higher rainfall.

FLOWEIUNG TIME

Species of Moraea may be found in bloom almost throughout the year, though
few species flower in the dry season. Peak blooming time is mid wet season.
Each species, however, has its own fairly limited flowering period, and detailed
examination of flowering times suggests that this factor is of considerable sig-
nificance in the evolution of the genus in tropical Africa. In general, it appears
that only a few species, and only one in each subgenus, may bloom in a particu-
lar habitat in any given locality at one time. Species of Moraea tend to bloom
for about two months; thus in a particular area several species might occur, each
flowering at a different time. The significance of flowering times is elaborated
further in the following discussion on evolution.

1077] GOLDBLATT— A/O/iAEA

247

Evolution

The pattern of species tliat prevails in Monica in tropical Africa suggests that
the group is rapidly evolving into a limited num])er of spatial, temporal, and
ecological niches available to the genus. A very limited degree of important floral
differences suggests that there has not been sufficient time for fundamental
moiphological changes to become established in the flowers, so that selection
for different pollinators has not been accomplished.

There are three major groups which are dealt with separately in the follow-
ing discussion; the large-flowered M. spathulata-M . textilis series of subgenus
Grandi flora; the smaller flowered and moipliologically specialized M. tanzanica-
M. iinifoliata series of subgenus GrancUflora; and the small uniformly blue-
flowered subgeiuis Vieusseuxia.

1. The Moraea spathuJata-M , textilis series.

The series comprises vegetatively similar, large-flowered species in which
both flowering time and habitat differences have played the major role in their
evolution.

The dry habitat does not support species of subgenus Grandiflora in the dry
season, with the possible exception of the poorly known Septemb(M-blooming
M. upemhana. Early in the wet season, from December, A/, verdickii blooms in
Tanzania, Zambia, and Zaire. Moraea ventricosa follows in this region with a
blooming peak in March. Subsequently, at least in the east, in Malawi and
Tanzania, M, macrantha blooms from March to June, lasting into the beginning
of the dry season.

In the wet habitat, the wide-ranging M. schimperi blooms from September
to early December south of the ecjuator. It is replaced later in the season by a
variety of more localized species of the series dealt with in the following sectio)»,
including A/, hrevifolia in Zambia, M. imifoliata^ M. hovonei, and M. balundana
in Zaire. Later, towards the end of the wet season, M. hella blooms in Zambia,
Zaire, Malawi, and Tanzania.

Geographical isolation is of relatively minor significance in the evolution of
this series but has played a role as in M. ventricosa-M. textilis^ which occupy
similar habitat and temporal niches but Ikivc different geographical ranges. A
similar role for geographical isolation is seen for M. verdickii-M. spaihulata.

2. The Moraea tanzanica-M. nnifoliata series.

The species of this series are all relatixely small in size, have similar yellow
flowers, and exhibit distinctive vegetative modifications making them easy to
recognize. Unlike the previous series, all, with the exception of the poorly un-
derstood M. upemhana, occupy a similar habitat and flower at the same time.
Spatial isolation is the major factor of evolutionary significance here, with each
species relatively limited in range and none sympatric. Small, local populations
are characteristic in the series.

3. Subgenus Vieusseux'uL

Subgenus Vieusseuxia, also a group w^here speci(\s 1ia\e \'er\ similar flowers,
exhibits a similar pattern of evolution to subgenus Grandiflora with ecological
factors assuming dominance over a secondary geographic element. The dry

248 ANXALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

grasslaiul habitat supports Af. thonisonii in tlic later dry season; A/, carsonii in
the early to mid wet season at least in the south; A/, afro-orientale in similar situ-
ations north in Tanzania, Kenya and Uganda; and M. iringcmis locally. In the
wet habitat, A/, natalemis blooms in the wet season throughout south tropical
Africa. There are no late-blooming species and presumably taller grasses shade
out these smaller plants in late season.

The exception to the patterns described above is the very striking A/. caUisfa.
This species has a large and very different flower from its allies and, although
it presumably evolved in isolation, it is now distinguished by its floral peculiar-
ities and, by inference, has a distinctive series of pollinators.

Summarizing, two main modes of evolution appear to have been operative
in the tropical African species of Moraea. These are: (1) an ecological factor
(habitat and flowering time) and (2) a geographical (spatial) factor. Within
the two major groups, subgenera Grandijlora and yieusscuxia (excluding A/. caJ-
lista), floral differences are small and probably not of adaptive significance, so
that the species of each group do not appear to have evolved in response to pres-
sures from pollinators.

This situation is similar in the summer rainfall area of southern Africa, al-
though somewhat more complex, with more species and basic types, and an ele-
ment of floral evolution is evident in section Vieusseuxia. However, the evolu-
tionary pattern in the winter rainfall region is quite different. Flowers vary
considerably and floral adaptation is of the greatest significance. Evolution
for limited and specific ranges of pollinators consequently must ha\'C been im-
portant. Flowering time and habitat are minor as most species grow in essentially
similar situations and all are spring })looming. Soil differences assume great
importance as does the geographical factor. Related species are often isolated
from one another geographically or by distinct soil preferences. The evolutionary
patterns here, in fact, are typical of an area with a Mediterranean climate (Raven,

1973).

Taxonomic Chahactei^s

d

in my recent paper on Moraea in the winter rainfall region of southern Africa
and little need be added here. It will, however, be helpful if I summarize the

important features of the two subgenera that occur in tropical Africa.

(lifl

the genus,

though a few specialized species in tropical Africa — e.g. A/, clavata, A/, angolen-

f

Floral morphology is uni-

form apart from size and color, most species having yellow flowers, only A/.
schimperi^ A/, ventricosa. A/, macrantha and A/, textilis having blue flowers.
Outer tepals are outspread, the inner ± erect. All species are unbranched, and

have a single leaf, usually well developed and basal, but in a few species re-
duced and inserted well abo\^e ground level. All species in which fruits are
known have large flattened ± discoid seeds and large capsules. The subgenus
extends from the southern Cape in South Africa to Nigeria and Ethiopia but

19771

COLDHLATT- A/OnA/:.A

249

Table 1. Cliroinosomc numlHT in tropical African MoraccL An asterisk (*) indicates a
connt for Sonth African material only; new connts arc in hold type.

Species

Diploid nninher

M. cars()}\ii

M. clliotii

M, natdJcnsis { as M. crici-ro.senii )

M. tlionisonii (as M. stricta)

M. spalhtilata ( incl. subspp.)

M. uidcrauUui (as A/, textills)

M. ? vcntricosa (flowers not seen)

M. tanzanica

2m

12

12

12
1«

12*
12

12
12

12

12

If^*

21*, 3(^

Collection data or relertMict

Coldhlatt, iy76a

Lewis, 1966; Goldhlatt, 1976a.

Lewis, 1966.

Cluniplianiba, 1974.

Malawi, Zoniba Mt., Goldhlalt .s.n.

( no voncher).

Goldl)latt, 197L

Goldblatt, 1971, J976a.

Goldblatt, 1971; Chinipban)l)a, 197 b
Malawi, Zoniba Mt., Goldhlafi 425D
( MO). Malavv^i, Cln'kangawa-M/,n-
zu, Goldblatt 45W (MO). Malawi,
M/n/n, Marynionnt, Paack 5H33
(MO).

Goldblatt, 1976a.

Malawi, M/n/n, Paicck .53,%' (MO).
iNhdawi, Katnmbi-Nyika intersec-
tion, Goldblatt sji. (no \()ncber).

Bnrnndi, Mt. Bona, E. Biiiinnbura.
Goldblatt s.n. (no vonclier).

Nhilawi, Nyika Plateau, Jnni[)er for-
est road, Pawck 6660 (MO).

is notably absent from tlic sontlnvrstcrn Cape, an important center for tlie genus
as a wliole.

Subgenus Vicusseuxia, composed of two sections of wliicli only section Pohj-
(inthes occurs outside South Africa, comprises sn»all to fairly large plants all witli

very similar blue (to white) flowers, wliich in tropical Alrica are (juitc small

and in marked contrast to subgenus Grandiflora. Tepals of both whorls arc out-

.spread, the exception being M. calllsta from southwestern Tanzania which has
imusual, large blue and white flowers with reflexed tepals. Species var\' in leaf

number and position of insertion which ranges from basal to just below the in-
florescence in M, natalensis. Moraea carsonii has 2-3 leaves. A/, callistra has
tw^o, while other species typically ha\'e one leaf. Seeds of all species, where*
known, are small and angular and capsules arc small. The subgenus extends
from tlie southwestern Cape to Ethiopia but is absent from West Africa and An-
gola. Section Polyanthes is the more widespread, with section Vieusscitxia re-
stricted to Africa south of the Lintpopo.

Cytoi.ocy

Moraea is well known chromosomally (Goldblatt, 1976a), some 65% of the
95 species having been studied. It is least well known in tropical Africa and only
9 of the 24 species have now been coxmted. Counts Icjr the genus in tropical Af-
rica are summarized in Table 1, where sexeral new counts are presented.

250 ANNALS OF THK MISSOURI BOTANICAL GARDEN [Vol. G1

The tropical African species liave fairly unifonn karyotypes, with a base
number of x — 6. The chromosome complement in all species of subgenus

(lifl

12, with 4 large

telocentric pairs, and two ± acrocentric pairs. Satellites arc located on a telo-
centric pair (Goldblatt, 1976a: 13). Moraea schimperi is distinctive in having a
secondary constriction on the end of the long arm of an acrocentric pair, a feature

also noted by Chimphamba (1974).

In subgenus Vietisseuxia the chromosomes are acrocentric with the longest
pair almost metacentric (Goldblatt, 1976a: 13). Polyploidy has been reported in
both sections. In section PoIyantJies, M. thormonii is heteroploid with tetraploid
and hexaploid plants recorded in South Africa and an octoploid from Malawi.
Chimphamba (1974) has reported 2n = 12 in this species but determination
needs to be confirmc^d and the locality is unknown. Moraea cUiotii is also hetero-
ploid, with diploidy and tetraploidy reported in South African plants.

The cytological uniformity of Moraea in tropical Africa, especially in sub-
genus CramUflora, contrasts with a great degree of heterogeneity in South Af-
rican winter rainfall area species. This uniformity supports my suggestion based
on limited morphological variation that Moraea is of fairly recent origin in tropi-
cal Africa.

Taxoxomic Treatment-'^

Moraea Miller, Figs. PI. 159, tab, 238. 1758, as Morea and altered to Moraea by
Limiaeus, nom. cons, type: A/, vegeta L.

For complete synonymy and generic description see Goldblatt (1976b).

Distribution: Sub-Saharan Africa, in highland areas in the tropics, at all al-
titudes in temperate southern Africa; species concentrated in South Africa,
mainly in the southwestern winter rainfall region.

Key to the Specucs

1. Pnuluc'cd leaves 2 or more, well developed.

2. Sheathing portion of upper (cauline) leaf more than 1.3 em long,

3. Stem few to many branehetl; spathes 2.5-5 cm long ___ 1. A/, carsonii

"3/ Stem simple or 1 -branched; spathes 4-6 cm long __ 2. A/. ciiUista

2.' Sheathing portion of upper leaf 2-6 mm long 3. M. afro-orient ale

1.' Produced Knif solitary, occasionally reduced and ±bractlike or lacking at flowering
time.
4, Plants either leafh'ss at flowering time, or with a dry withered leaf attached;

blooming in the dry season.

5. Mowers yellow; outer tepals 2.7-4.5 cm long _ 21. M. upcmhana

5/ Flowers pale blue violet; outer tepals 1.5-2.2 cm long 7. M. thomsonii

'All major collections of African flora were consulted for this study, and unless other-
wise stated all t)'p(^ specimens were seen. Field work was carried out in Malawi and to a
limited degree in Rhodesia.

SpeeiuRMis examini'd are generally arranged in accordance with modern floristic projects
for particular areas. Modern place names are used, but old names where well known or
recently altered, are provided in parentheses.

Collection data or literature citation for chromosome numbers in the descriptions are
gi\ en in th(^ section on Cytology.

1977] OOLDIU.ATT --MORAEA

251

4/ Plants witli a green prodnccnl leaf at flowering time, usually well de\elnpecl but
oecasionally very short and inserted just below the infloreseenee.

6. Leaf inserted immediately under the infloreseenee (leaf sometimes braet-
like and almost entirely sheathing).

7. Flower blue; stem usually branelied 6. M. natalcn.sis

7/ Flowers yellow; stem simple.

8. Outer infloreseenee s^^athe only slightly shorter than the inner.

9. Leaf barely distinguishable from the spathes and not or only

shortly exeeeding th(Mn _.__ ___ 24. M. unijiAiaia

9/ Leaf several times longer than the spathes 22. M. bovonci

8/ Outer infloreseenee spathe less than half as long as the inner

23. A/. })alu)\d(ina

6.' Leaf inserted from the base to the upper part of the stem but not inunedi-
ately l)elow the infloreseenee.
10. Outer tepals L4— 3.5 em long.

IL Sheathing portion of braet leaves at least L5 em long.
12. Flowers blue; outer tepals L4-2.5(-3.() ) em long.

L3. Leaf inserted in the upper third of the stem - 6. M. tiatdlcti.sis
13/ Leaf inserted in the lower third of the stem .._. 5. A/, clliotii
12.' Flowers yellow; outer tepals (2.()-)2.5-3.5 em long.

14. Braet leaf solitarv — . 20. M. clavata

14.' Braet leaves 3 or 4.

15. Spathes 5-8 em long; braet leaves 5—8 em long __._

11. M. i)i\j(in<s,ani

15.' Spatlies 4.5-5.5 cm long; braet leaves 3.0-4.5 em

long ___ 12. A/, anaolcnsis

t->

11.' Sheathing portion of braet leaves 1-7 mm long.

J

16. 0\ary 3-5 mm long; anthers 4-5 mm long ,_._ 3. A/. afro-orienlaJc

16.' Ovary 7-13 mm long; anthers ±6 \nn\ long 4. M. irin^cnsis

10.' Outer tepals 3.8-10 em long.

17. Braet and spathes ibdry at flowering time and the flowers blue
and bloonung at the end of the dry season to early in the wet sea-
son ( Sep. -Dec. south of the eiiuator, Dee.-May north of the
e(iuator); leaf often shorter dum the stem, the prophylls dark
brown and pronu'nent ..__, _ _ 9. A/, schiniperi

17.' Bracts and spathes at least parti)' lierbaeeous at flowering time, or
if not, the fl(n\ers white to )ellow and not flowering at the end
of the dry sc^ason; leaf exceeding the stem, or if not, flowers white
to yellow.

18. Leaf inserted abov<' ground level, very short and rarely ex-
eeeding the spathes, the braet leaves usually solitary

. . . ... 19. A/, brcvifolia

18.' Leaf ih basal or if inserted well above the ground, then the

bract leaves 1 or 2, the leaves not exceeding the spathes or the
plants no more than 35 cm high.

19. Plants of Rhodesia, southern Mozambique (and South
Af riea ) .

20. Flowering in spring and early sunmier, Sep.-Xov.,
usually in moist habitats; 30-50 (-70) cm tall;
leaf ca. 5 mm wide 10. M. muddii

20.' Flowering in summer in open grassland; usually
more than 50 em tall; leaf Tisuall)' more than 1 em
wide 8. A/. spaihuJata

19.' Plants of Angola and oecurring north of Rhodesia and
southern Mozambique.

21. Plants 30-35 cm tall with 2 bract lea\es only and
the leaf only shortly or not exceeding the spathe — _
18. A/, tanzanica

21.' Plants usuallv more than 40 cu) tall, if less, then the
bract leaves more than 2 and the leaf much exceed-
ing the spathes.

252 ANNALS OF THE MISSOURI BOTANICAL GAHOEN [Vol. fl4

22. Bracts not ()\ criapping.
23. Flowers bluo.

24. Outvr tepals 5.5-8 cm long; corni tu-
nics of pale, reticulate fibers; plants of
Za'ire, Tanzania, Zainl>ia, Malawi

___ __ __._ 15. A/, macraniha

24.' Outer tepals 4.7-6 cm long; corm tu-
nics of wiry, dark fibers; plants of An-
gola - 17. A/, fcxtilh

23/ Flowers yellow.

25. Flowering Nov. to Feb. (Mar.), usu-
ally in open grassland; antlu'rs 10-14

mm long; outer tex^als 5-10 cm

_ 14. M. vcnlickii

25.' Flowering ( Feb. ) Mar. to July iu
damp situations; antbers 8-10 mm
long; outer tepals 4.5-5.5(-6.5) cm ...
p _____ „___ 13. Af. hclhi

22.' Bracts (nerlapping.
26. P^lowers blue.

27. Outer tepals more tlian 5.5 cm long _.„

____ „ 15. A/, macraniha

27.' Outer tepals less tban 5.5 cm long.
28. Antbers 8-10(-ll) mm long; in-
ner tepals 3.7-4.5 cm long;
plants of Zaire, Zambia, Burundi,

Tanzania 16. A/, ventricosa

28.' Anthers (10-) 10.5-15 mm long;
inner tepals 4.8-6.5 cm long;

plants of Angola 17. M. tcxiilis

26.' Flowers wliite to yellow.

29. Antbers 8-10 unn long; bract number

3_4(_5) 16. Af. ventricosa

29.' Antbers 10-15 mm long.

30. Outer tepals 5-10 cm long; bract
leaves 2-3(or 4 but tben tbe

tepals more tlian 5 cm) ___

__ 14. Af. verdickii

30.' Outer tepals 4.5-6.5 cm long;

bract leaves (4-)5-7 ._ 17. A/, tcxtilis

Subgnms \'iKUssEUXiA(de la Roche) Baker — Section Polyaxthes GoklM.

1. Moraea carsonii Baker, Bull. Misc. Inform. 1894: 391. 1894. typk: Zambia,
Mbala (Abercorn) district, "Fwambo/' Carson s.n. (K, holotype). — Fig. 1A.

M. Iionihlci Ue Wild., Coutr. FI. Katanga, Suppl. 4: 7. 1932. type; Zaire, Sbaba (Katanga),
near Kolwezi, Ilomblc 1024 ( BR, bolotype).

Plants small to medium in size, 20-40 cm high, usually bearing several
branches. Corm ca. 1.5 cm in diameter; tunics of dark brown, fine to medium
reticulate fibers. Prophylls dry and papery, pale brown, the upper often torn
somewhat distally, occasionally becoming fibrous. Leaves 2, occasionally 3, the
lower ±basal or inserted well below the branches, the upper cauline, canalic-
ulate, 2-5 mm wide, usually falcate, exceeding the infloresence; sheath of
upper leaf 1..5-2 cm long. SpatJics occasionally reddish, herbaceous below, be-
coming dry from the apex, the margins dry, pale brown, the apices dark brown,
attenuate; inner spathe 3.2-5.0 cm long, the outer about two-thirds the inner.

1977]

COLDBLATT A/OfiAEA

253

A

B

J" t <J_«i.'^

FiGUHK 1. MoKJca species. — A. A/, ('(trsotiii. — B. M. ajro-oricntale. (xO.5)

254 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

FJower l)lue witli yellow nectar guides; outer tepals 2.1-3 cm long, lanceolate
spreading, the limb slightly longer than the claw; inner tepals 2-2.8 cm long,
spreading. Filaments ca. 7.5 mm long, free in the upper third; anthers 4.5-5.5
mm long. Ovary 3.5-7 mm long; style branches 8 mm long, the crests 7-10 mm
long. Capsule ovoid, to 8 mm long; seeds small, angular. Chromosome number

2;i^l2.

Flowering time: December to February.

Distribution: Rhodesia, Zambia, Malawi, southern Zaire. — Fig. 2.

Its two or sometimes three leaves, indeterminate growth pattern with several
to many branches and small unspecialized flowers, place A/, carsonii in a primi-
tive position among the tropical African species of the genus. It clearly belongs
in subgenus Vieusseuxia section Pohjanthcs and is probably most closely allied
to the South African, southern Cape species. A/, pohjanthos^ which generally
has three leaves and similar flowers, and also to the widespread southern African
M. polystaehya, a taller species with much larger flowers, Moraea carsonii ap-
pears to represent a link between these two multi-leafed southern African species
and the several tropical and South African summer rainfall region species of
section Polyanthes^ most of which have a very similar flower but only a single
leaf, and are distinguished mainly by differences in growth form.

Moraea carsonii has a relatively wide range, occurring in the Zaire province
of Shaba (Katanga), in northern and eastern Zambia, and adjacent Malawi. A
somewhat different form is found to the south in the Inyanga highlands of Rho-
desia, which is g(Mierally more robust, and has larger flowers, giving it a resem-
blance to M. pohjstachya to which it is frequently referred. Not all plants from
Inyanga are ef^ually robust, and several collections cannot be distinguished from
plants occurring further north. For this reason the Inyanga populations are not
accorded taxonomic recognition here.

In Tanzania and northwards to the Sudan A/, carsonii is replaced in similar
habitats by A/, afro-orientale Goldbl., which has until now been included in A/.
carsonii (liurtt, 1938). Moraea afro-orientale differs in having a single basal
leaf inserted below ground, smaller flowers, and fewer branches which are often

clustered in umbellate fashion. Occasionally, particularly in southern Tanzania,
forms occur with two leaves but the upp(M' leaf has an unusually short sheath.

4

Malawi, nohiiiehn hegion: Mtwalo-Mziinba road, Paicck 3358 ( K, MAL). Vipya

hills S ()[ M/.u/ii, raaek 4421 (K). SW M/uzu, Patcck 8051 (MO, SRGH). 5 km S Mt.
Ilora, Riiniplii district, lUUard <b Btirtt 4456 ( K, MAL, SUCH). c:knthal hegion : Kongwc
Mt. near Dowa, Rohson 1654 ( BM, LISC, PRE, SRGH).

RiioDKsiA. EAsiKHN HEGION: Mt. Invaiigaiii, Plowes 2153 (PRE, SRGII); Duvulse,
Simon dr Pope 6536 (MO); NorUndh 6 Weimarck 4968 ( BM, BR, K, PRE, SAM); West
7010 (SRGH); Wild 4927 ( K, LISC, MO, PRE, SRGH). Near Bonda Mission, Wild 5478

(COI, K, MO, PRE, Sl^GH). Above Rhodes Hotel, Inyanga, WheUan b Davies 997
(SRGH). Nyaniaropa Forest Reserve, Tnvanga, Dale SKF204 ( K, SRGH); Dreae 28,
54 (SRGH); Wild 7495 (BR, K, LISC, PRE, SRGH, MO). Invanga Downs, Wild 5475 ( K,

MO).

Zahu:. SHABA: 10 km S of Lubninbashi ( Elizabethville), Schinitz 1286 (BR). Lubum-
bashi, Salesiens 1062 (BR). Fungunime, Sijmocnfi 14023 (BR, K).

Zamhia: centhai. hegion: Lusaka, Angus 1466 ( K, LISC, SRGH). Kafne River,
Allen 498 ( K, SRGH). Kiindalila Falls, Serenje di.striet, Strid 2900 (K). nohthehn heciox:
Mbala township, Sunane 986 (K). Near Nakatali, Rieluird.s 8023 (K). Kalambo Falls, Mljala

1977] GOLDBLATT— A/0RA/:A

255

(Abercorn), Richards 3933 (K), Sansia Falls, Mhala district, Ricfwrds 7436 (K). Mhesuma
ranch, Astle 11H5 ( K, SRGII). western hegion: Luanshya, Famhawe 1739 (BH, K, SHCII).
Mufulira, Famhawe 1 1749 (SRCII). Ndola, Famhatve6I2 (BR, K).

2. Moraea callista Goldbl., sp. iiov. type: Tanzania, Soutlicrn Highlands, El-
ton Plateau, Richards 7562 (K, holotype; BI^, isotype). — Fig. 3A.

Planta 30-70 cm alta. Connus i^notus. Folia duo, canaliculata, iiifcrius hasalc, caulis
simplex \el uniramo.sus. Spathac intlcxac, hcrhaccac, exterior 4—6 cm longa, interior 3—4.5
cm longa. Fh)rcs caeruleo-mal\'ini, tepala alhescentes distale; tepaJa exteriora 3-3.5 cm longa,
limhis 2-2.5 cm longis, reflexis; intcriora hreviora^ reflexa. FiJanwnta ad <S mm longa; autlierae
mm longae. Gcrmen 5-7 mm longum; rami styli ca. 8 mm longi; cristae 5-7 mm longae.

Plants solitary, simple or rarely l-branched, 30-70 cm higli. Conn not known.
ProphijUs membranous, the uppermost dry and fibrous towards the apex. Leaves

2, the lower basal and larger, the upper cauline, to 7 mm wide, canaliculate
with slightly thickened hyaline margins, as long as, or slightly exceeding the in-
florescence. Spathes often inflexed, herbaceous, with dry upper margins; outer
spathe 4-6 cm long, the inner ca. % the inner. Flowers blue mauve with the
tc^pals fading to white distally; otiter tepaJs 3-3.5 cm long, the limb 2-2.5 cm,
fully reflexed when open; inner lepals somewhat smaller, also reflexed. Fila-
ments to 8 mm long, united in the lower 5 mm; anthers 6 mm long. Ovary 5-7
mm long; style branches ca. 8 mm long, very broad and outspread, the crests
5-7 ni!U. Capsule and seeds imknown. Chromosome number not known.

Flowering time; January to February in the west, May in the east.

Distribution: Southwestern and eastern Tanzania in mountain grassland,
l,800-3,000m.— Fic. 2.

Moraea callista stands apart from the other species in section Polijanthes be-
cause its large striking flower with blue and white, fully reflexed tepals are (juite
distinct from the usual small, uniformly blue flowers with spreading tepals
of other spcxnes. It occurs in two wideh' separated areas, high mountain grass-
land in the Njombe area in the southwestern part of Tanzania, and to the east in
the Uluguru mountains where only one gathering has been made, flowx*ring at

J

the western

part of its range. The species is known from \'ery few collections and is all too
little understood.

Tan/ama. souiiiKiiN HIGHLANDS: EJton iMutcau, Procter 1646 ( EAH); Richards 7562
(BR, K). Mangale-Njomhe road, Richards 14216 (K). Mwakete, Njomhe district, Richards
7H23 (K). MOiUKiOHo distihct: Mzimihi, Scntsci 1707 (EAH, K, PRE).

3. Moraea afro-orientale Coldbl., sp. nov. tyim:: Uganda, Northern Province,
Mt. Debasien, IIedher<; 1953 (UPS, holotype; EAH, K, S, isotypes),— Fk;. IB.

Fhi)ita 15—40 cm alta, gracilis. Tiniicae cormi l)rnnneae, reticnlatae, tenuis. Folitnn
solitarium, basale, raro folio secundo caiilino, canaliculatum. Caulis ramosus, ferens hracteae
sole vel folio caulinum unum, vagina hrevissima. Spathac herbaceae, interior 2-3.5 cm k)nga,
exterior 1-2 mm hrevior. Florcs caerulei; tepala exteriora 1.8-2.6 cm longa, effusa; tepala
ititeriora 1.5-1.8 cm longa. Filamcnta ca. 5 mm longa; aiitherae ca. 4.5 mm longae. Germen
ad 8 mm longum; rami styli ca. 5 mm longi; cristae ad 5 mm longas.

Plants small, solitary, usually few branched, 15-40 cm high. Conn ca. 1,5
cm in dianu^ter; tunics brown, cancellate to reticulate, of medium to fine fibers.

256

ANNALS OF THE MISSOURI B0TAMC:AL GARDEN

[Vol. 64

FicuHK 2. Distribution of Monica carsonii, M. culUsta^ M. afro-ork'utalt\ and A/, irin-

^cnsis.

ProphyUs usually 3, membranous, entire, obtuse-truncate. Leaf usually soli-
tary, basal, inserted below the ground, the base covered by prophylls, canalic-
ulate, 2-5 mm wide, relatively short, as long as or slightly exceeding the in-
florescence; occasionally a second leaf developed at the first aerial node, usually

1977] GOLDBLATT— A/ORAZ^A

25

/

bractlike, but 2-10 cm long, occasionally also exceeding the inflorescence; canline
leaf sheath 3-6 mm long. Spathes subequal, herbaceous, the margins mem-
branous, rarely dry but not brown, the apices acute, brown tipped; inner spatlie
2-3.5 cm long, the outer 1-2 mm shorter. Flowers blue with orange nectar
guides; outer tepah 1.8-2.6 cm long, lanceolate, the limb slightly longer than the
claw, spreading; inner tepah 1.5-1.8 cm long. Filaments ca. 5 mm long, free in
the upper third; anthers ca. 4.5 mm long. Ovary 3-5 mm long; style branches
ca. 5 mm long, the crests to 5 mm. Capsule ovoid, to 8 mm long; seeds small,
angular. Chromosome number not known.

Flowering time; April to May north of the equator in Kenya, Uganda, and
Sudan; December to January south of the equator.

Distribution: Southern Sudan, western Kenya, Uganda, and central and
western Tanzania; in short grass and often in seasonally waterlogged ground,
blooming soon after the start of the rainy season. — -Fig. 2.

Moraea afro-orientale is closely allied to M. carsonii, from which it is prob-
ably derived, and it has until now been included in this species. It differs, how-
ever, in several respects and is distinctive in overall habit, with prominent mem-
branous prophylls, short stem and few% often clustered branches. Most forms
have only a single comparatively short and erect leaf, inserted below ground
level in contrast with the two long, flaccid leaves in M, carsonii. In tlie southern
part of its range, in central and western Tanzania, A/, afro-orientale often has
two leaves, but when this is the case, the upper, canline leaf is always distinctive
in having a very short sheath (2-5 mm), contrasting with the more usual longer
sheath (10-20 mm) in A/, earsonii and its other relatives. In the single-leafed

/

leaf is reduced to a bractlike

structure, often with a short free apex, and tliis bract also has a very short sheath.
Moraea afro-orientale shares this distinctive, very short leaf or bract sheath
with two other tropical African species. A/, irin^i^ensis^ which has a larger flower
and a long ovary which is included in the spathes, and A/, callista^ which has
two large leaves, and a large striking flower with blue and white, fully reflexed
tepals. Both these species occur rather locally in southwestern and central Tan-
zania on the southern extremity of the range of A/, afro-orientale (Fig. 2).

Kenya, nyanza province: Biingoma district, Kokicaro 134 (EAII). SE slopes of Mt.
Elgon, Padwa 16 (EAII, F, K, PRE, SRGH). nwr valley piu)vinc:e: Elgon and Trans
Nzoia, Twecdie 444 (K). Kitale-Siuun Mill road, BaUtf 2487 (K). Slopes above End(4)ess,
FoJhill 411 (BR, K, PRE). Mt. Elgon, Raijner 545 (K); Lugard 573 (K); Adamsou 513
( EAH, K ) . Endehess, Webster 9018 ( E AH ) .

Sudan, ec^uatohia : Mt. Lotuke, Didinga Mts., Jackson 1333 ( BM ) ; MacDonald 91
(BM); Myers 10952 (K).

Tanzania, central province: Singida-Dodoma, Feirarzl 5708 (EAH). soutkkrn
HIGHLANDS: IguMissi, Mbeya district, Procter 2284 (EAH, K). Rnalui National Park, Ma-
gangwe Hill, Bjornstad 2231 (EAH, K). western province; Moanzi, Sumbawanga dis-
trict, Vezey-FitzGerald 1378 (K, SRGH). Kuturia-Lukunga confluence, Ufipa district, Rich-
ards 10259 (BR, K). Above Malonje, near Sumbawanga, Richards 7210 (K).

Uganda, ruganda province: Mengo, Dradii 846 (EAH). eastern province: Butiro,
Liebenherg 846 (K). northern provinc:e: Karauioja district, Lodokcteininit, near Moroto,
Kerfoot 995,4947 (EAH, K). Mt. Debasien (Kadani), lledberg 1953 (EAH, K, S, UPS).
Moruita, Eggeling 5785 ( BR, K ) . Karauioja, Thomas 2836 ( BR, K ) .

258

ANNALS OK THE MISSOURI BOTANICAL GARDEN

[Vol. 64

*?>

FiciiHE 3. Morava specie's. — A. M. caUista. — B. A/, irin^cnsis. — C. M. nataJcmis

(XO.5)

4. Moraea iringensis Goldbl., sp. nov. type: Tanzania, Southern Higlilandsj
Kyiinl)ila-TaiKlala, Stolz 2362 (K, holotypc; BM, BR, PRE, isotypes.)— Fig.
3B.

4

PliUita L5-30 cm aha, Foli\im solitariuni, basale, usitatc panuii inflorcsccntiuiu brcvior,
canalifulatuin. Caitlis paiicirainosa fcreiis braclca una \aginis brevissinuis. Spathac herbaceae,
interior 3-4.5 cm longa, exterior L5-2.3 cm Vm^n. Flares caernlci; tcpala cxtcriora ca. 3 cm
longa, etlusa; intcriorn ca. 2 cm longa, effusa. FihiDicnta ca. 7 mm longa; anthcrac ad 6 mm

1977] GOLDBLATT- -MORAEA

259

longas. Gcrmcu 8-12 mm longinn, non exsertiim; rami styli ca. 1 Lin longi; cristac ad 8 mm
longas.

Plants solitary, (10-) 15-30 cm high, usually branched. Corm ca. 1 cm in
diameter; tunics of fine to medium reticulate fibers. Prophylls 2-3, entire, mem-
branous, often red-flushed, obtuse-truncate. Leaf solitary, basal, usually slightly
shorter than the inflorescence, canaliculate, to 6 mm wide, with slightly thickened
hyaline margins. Bract leaves to 3 cm long, sheatlnng at the base only, acuminate.
Spathes herbaceous, acuminate; inner spathe 3-4.5 cm long, the outer ± half the
inner, often with the upper part not sheathing. Flower blue; outer tepaJs ca. 3
cm long, lanceolate, the Hmb ca. 1.5 cm long, spreading to slightly reflexed; inner
tepals ca. 2 cni long, spreading. Filaments ca. 7 mm long, united in the lower
third; anthers 6 mm long. Ovary unusually long, S-12 mm, enclosed in spathes;
style branches ca. 1 cm long, the crests to 8 mm long. Capsule and seeds not
known. Chromosome number not known.

Flowering time: December to Jaiuiary.

Distribution: "Grassland and open Brachste^^ia woodland" in the southern
highlands of Tanzania, recorded between Sao Hill and Makumbako, l,S()()-2,2()()
m. — Fig. 2.

Moraea iringensis is closely allied to the more widespread A/, afro-orienfale
which occurs in western and central Tanzania, Uganda, western Kenya and the
southern Sudan. It differs in being more robust, with larger leaves and flowers,
and it is distincti\'e in having a very long ovary, (S~12 mm long, compared to
3-4 mm in A/, afro-orientale. In spite of its length, the ovary is almost entirely
enclosed in the spathes, in contrast to the exserted ovary found in all allied
species.

Tanzania, southkhv tiighlands: Sao Hill, Iringa district, Chambers 38 (EAH, K);

Cannichad 331 (EAII); Robertson 825 (EAII). Kyiinbila, Tundala, Stoh 2362 ( K, HM, \U\,

PRE). Mufindi, Bjornstad 577 (K). 20 km X of Sunji, Njomhr district, Siurtz 79 (DAl^,
EAII).

5. Moraea elliotii Baker, Handb. Irid. 58. 1892. type: South Africa, Transvaal,
marshes near Lake Chrissie, Scott-Elliot 1592 (K, holotype).

For complete synonymy see Goldblatt ( 1973).

Plants small to medium, reaching to 55 cm high, usually branched. Cor)n
1.5-2 cm in diameter; tunics of dark brown fairly coarse fibers often extending
upwards in a neck. Prophylls membranous, becoming dry from above. Leaf soli-
tary, terete or linear and canaliculate, inserted near the base or the lower part
of the stem, and much exceeding the inflorescence. Bract leaves 1-4, 2..5-6 cm
long, herbaceous or becoming dry and brown. Spathes herbaceous with dry,
light brown margins, attenuate; inner spathe 4-6 cm long, the outer ca. 1 cm
shorter. Flowers blue violet with orange yellow nectar guides; outer tepals 2-3
cm long, lanceolate, the limb to 1.5 cm long and 1.2 cm wide, spreading to re-
flexed at 45""; inner tepals 1.5-2.4 cm long, linear-lanceolate. Filaments ca. 6
mm long, joined in the lower half; anthers ca. 6 mm long. Style branches ca. 1
cm long, the crests to 0.5 cm long. Capsule ovoid, to 1.2 cm long; seeds small,
angled. Chromosome number 2n = 12, 24 (South African collections only).

260

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

l^'i(;i'HK 4. nistributicin of Moraca natalcnsis and M. eUiotii.

Flowering time: September to March.

Distribution: Eastern South Africa, Swaziland, and Malawi. — Fig. 4.

Moraca clliotii is fairly common m the grasslands of South Africa, especially
in well-watered highland areas. As defined earlier (Goldblatt, 1973), it includes
a wide range of forms, from the early, spring-blooming eastern Cape plants with
a ±basal canaliculate leaf, to summer-flowering plants from Natal, Transvaal,
and Swaziland, also witli a canaliculate leaf, but inserted somewhat above
ground level, to terete-leafed, usually late-flowering plants, mainly from the
eastern Transvaal A single collection from Malawi is identical with the latter,
and though such a gap in distribution from South Africa to Malawi is unex-
pected, the Mahiwian collection must be assigned to M. elliotii. The species is
clearly predominantly South African, and for further details of synonymy, vari-
ation and distribution, readers are referred to my earlier treatment of the species

(Goldblatt, 1973).

Moraca elliotii is closely allied to species such as M. carsonii from Central
Africa and M. poJyanthos from the southern Cape, and all three species have
a similar flower. It would seem more specialized in its solitary leaf and rela-
tively few branches than the multi-leafed and branched M. pohjanthos and the
2-leafed M. carsonii. The more reduced A/, natalcnsis, with its short leaf in-
serted \\x41 abo\T the ground and with a somewhat contracted inflorescence,
probably evoKed from multi-leafed ancestors via plants like Af. elliotii; and M.
natalcitsis is also seen as closely related.

Malawi, cen ihal hegion: Dedza, suinniit of Ciwan, Chongoni, Chapman 1 176

(SUCH).

1977] GOLDBLATT— MOfiAEA

261

6. Moraea natalensis Baker, Handb. Iiid. 56. 1892. types; Soutli Africa,
Natal, Sanderson 253 (K, lectotype; S, isolectotype ) ; Sutherland s.n. (K,
paratype) . — Fig. 3C.

M. erici-wsenii Fries, Wiss. Ergcb. Schwed.-RhoJ. -Kongo Exp., 1911-1912, Bot. Untcrsucli.

1: 234. 1916. type: Zambia, Kalambo, Fries 1345 (UPS, holotype; Z, isotype).
M. parviflora N.E. Brown, Trans. Roy. Soc. S. Africa 17: 346. 1929. type: South Africa,

Transvaal, Tomson's vlei, NyLstroom, Pole Evans 19668 (K, holotype; PRE, isotype).

Plants 15-45 cm high, inchiding the leaf, usually branched. Corm to 1.5 cm
in diameter; tunics of dark brown to black fibers. Prophylls membranous with
the upper becoming dry from the apex and often lacerated. Leaf inserted well
above ground, shortly below the inflorescence, canaliculate to terete, to 20 cm
long and shortly exceeding the inflorescence or reduced and about as long as
the spathes. Stem with the lowermost internode very long, and the produced
leaf inserted in the upper part. Bract leaves if present seldom exceeding 2.5 cm

and usually dry and light brown. Spathes herbaceous, becoming dry above, the
margins dry and pale brown, the apices attenuate; inner spathe 2-3.5 cm long,
the outer ca. 1 cm shorter. Flowers blue mauve with yellow nectar guides; outer
tepals 1.4-2 cm long, lanceolate, the limb 0.7-1.4 cm long, up to 1.0 cm wide, re-
flexed to 45°; inner tepals to 1.5 cm long, reflexed. Filaments ca. 5 mm long,
united in the lower half; anthers 4-5 mm long. Style branches ca. 6 mm long,
the crests to 5 mm long. Capsule ovoid-subglobose, 4.5-10 mm long; seeds small,
angular. Chromosome number 2n = 12.

Flowering time : November to January ( February ) .

Distribution: Zambia, Malawi, Rhodesia, Mozambique, South Africa; in
moist situations, often in vleis and dambos and in seasonal pools. — Fig. 4.

The most striking characteristic of M. natalensis is its leaf insertion which
is well above ground level at the top of the long lowermost internode. The leaf
itself is relatively short, no more than 20 cm long, but more often smaller and
sometimes not exceeding the inflorescence spathes. In some forms, notably in
the northern part of its range, in Zambia, the stem above the lower internode
is much contracted so that the leaf seems to be inserted almost immediately
under the inflorescence. In these forms the leaf is often at its shortest, appear-
ing almost bractlike and easily confused with the inflorescence spathes. The
type of M. erici-rosenii corresponds to this form. Most other collections from
Zambia, and Rhodesia, where the species is very common, resemble the typical
South African form with a less contracted stem above a longer leaf. The existence
of a whole range of forms from the extremely short-leafed, much-contracted type
represented by M. erici-rosenii to the relatively long-leafed and extended-
stemmed type make it necessary to reduce M. erici-rosenii to synonymy.

Moraea natalensis is one of the more widespread species in the genus and
occupies a similar ecological niche throughout its range. It occupies rather moist
depressions or vleis and dambos in open grassland and in exposed rocky situa-
tions which are seasonally moist, and is also found around rock pools. It oc-
curs at moderate altitudes of between 1,000 and 2,000 m, though it is also found
near the coast in South Africa, It is most closely related to the predominantly

252 ANNALS OF THE MISSOURI BOTANK^AL GARDEN [Vol. 64

South African species A/, elliotii, which has a single leaf inserted lower on the
stem, usually shortly above the ground.

Malawi, cexthal region: Near Tumanda Mission, Rohson 1093 ( BM, K, LISC,

PRE, SRCII). KasTingu, Jackson 2299 ( K, MAL, PRE, SRGH). Kasungu National Park,
llall-Mariin 1364 (SRGH). northern region: S. Vipya, Lwanjati Peak, Pawek 8905 (SRGII,
MO).

MozAMiiiguE. \iamc:a & sofala: 12 km from Vila Pery, Tone 6- Corrcia 131S0 (LISC).
ZAMRE'/iA: Mujeha, on Pilxme road, Quelimane district, Faulkner K152 (BR, K, PRE, S).

Rhodesia, central region: Marandcllas, Dehn 556 (SRGII); Rattray 829 (SRGH);
Day s.n. (SRGn-2129); Collins 10 (K, SRGH). Riisape, Hopkins SM. (SRGH-fiSSO); Munch
455 (K. SRGII). Salisbury, Eyles 1896 ( K, PRE, SAM), 3737 (SAM, SRGII), 6099 (BOL,
K, SRGH); Willou^hhy sju (SRGH). Surprise siding, Selnkwc, Taiflor s.n. (SRGII-2(S25).
Near Qw Que, Bingham 374 ( K, SRGII). Gwelo, Lovcridge 540 ( K, PRE, SRGH). Dar-
wendale, Cordon 193017 ( K, SRGH), southern region: Near Fort Victoria, Plowes 3154
( K, LISC, PRE, SRGH ). western region: Bulawayo, Brain s.n, ( SRGH). Matobo dis-
trict, A/i//rr 1998 (K, LISC, MO, PRE, SRGH), 2051 (MO, PRE, SRGH). Matopos, BorJe
58 (K); Carley 112 (K, SRGH); Eyics 1144 (ISRGH). Lochview, Cross 335 (SRGH).

Za'ire. shara; Lumbumbashi ( Elisabethville), Ilock s.n. (BR). Near Mukumbi, Iloff-
vwnn962 (BR).

Zambia, central region: Lusaka, King s.n. (K); Best 4, 10 (K), 82 (SRGH). north-
ern region: Mbala (Abercorn) district, Kalambo Falls, Richards 19295 (K). 25 km W of
Kasama, RohiuMm s.n. (K). southern region: Machipapa, Mazabnka district, White 6253
(K). Katomo, Batoka Plateau, Sykes 267 (K). Kaloma, Fanshatve 9179 (SRGH); Mitchell 17/29
(SRGH). Between Cliomo and Mouze, van Renshurg 3084 (BM, K, SRGH). Cht)mo, Astle
1841 (SRGH); Latcton 1185 (K, SRGII). western region: Kalenda dambo, Mwinilunga
district, Milne-Rcdhead 3597 (BR, K, PRE). Zambezi Rapids, Mwinilunga district, Richards
17146 (K, SRGH); Lewis 6224 (K, MO). Kitwe, dambo, Linley 35 (MO, SRGH). Ncbanga,
Ferrars.n. (SRGII-4800).

7. Moraea thomsonii Baker, Ilandb. Irid. 57. 1892; Bot. Mag. tab. 7976, 1904.
tyim:: Tanzania, "plateau north of Lake Nyassa/' Oct. 1880, Thomson s.n.
{ K, liolotype ) . — Fic. 5.

A/, siricta Baker, Vicrteljabrsscbr. Naturf. Ges. Ziiricb 49: 178. 1904. type: South Africa,
Transvaal, Sliilouvane, Junod 563 (Z, liolotype; K, LD, isotypes).

M. telUnii Chiov., Ann. Bot. (Rome) f): 138, 191L type: Ethiopia, Begemdir & Simen,
Debarck, Chiovenda 3007 (F, lectotype); several otiier types cited, all from Ethiopia.

M. curtisae Foster, Contr. Gray Herl). 127: 46. 1939. type: Kenya, Norok, Noyroscra, 60
km SE of Narok, Curtis 676 (GH, liolotype).

M. irita N. E. Brown, Trans. Roy. Soc. S. Africa 17: 347. 1929. type: South Africa, Trans-
vaal, Lydenburg, Wilms 1419 (K, holotype; P, PRE, isotypes).

M, parva N. E. Brown, Trans. Roy. Soc. S. Africa 17: 347. 1929. type: South Africa, Trans-
vaal, Woodbush, Moss 15564 (K, holotype).

M. rnossii N. E. Brown, Trans, Roy. Soc. S. Africa 17: 347. 1929. type: South Africa, Trans-
vaal, Johannesburg, Moss 15805 ( K, holotype; PRE, isotype)

Plants small, 12-30 cm high, branched, leafless when in flower. Corui 1-2 cm
in diameter; tnnics dark brown, of tough reticulate fibers. PwphijIIs dry and ir-
regularly broken. Leaf solitary, quite dry, or lacking at flowering time, occasion-
ally a new leaf emergent and then not attached to the flowering stem, slender,
terete, long and trailing. Branches short to ±sessile, subtended by dry bract
leaves. Spathes usually quite dry and papery at flowering, acuminate, becoming

lacerated with age; inner spathe 2.5-4 cm long, the outer 5-10 mm shorter.

Flower pale blue lilac with yellow orange nectar guides; outer tepah 1.5-2 cm

long, lanceolate, the hmb to 1 cm, spreading; inner tepah 1.4-1.7 cm long, erect,

becoming outspread. Filaments 4-5 mm long, free in the upper third; anthers

1977]

GOLDBLATT— A/ORA£A

263

J r ci^Lo.

I-ilM

Figure 5. Morphology and distribution of Monica tho}}isonii (xO.5).

4-5 mm long. Style branches 7 mm long, tlic crest vestigial or 3-5 mm long. Cap-

12,

sale ovoid, ca. 1 cm long; seech small, angnlar. Chromosome number 2n
24, 36, 48.

Flowering time: (August) September to November (December) south of
the equator, December to Jiuie (August) nortli of the equator, in the dry season
before the rains begin.

264 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Distribution: Widespread^ in dry grassland in the eastern half of Africa; ex-
tending from the summer rainfall areas of South Africa through Mozambique,
Rhodesia, Malawi, and eastern Zambia to Tanzania, Uganda, Kenya and Ethiopia.
Fig. 5.

Several species are reduced here to synonymy in M. thomsonii, notably 3/.
teUhiii from Ethiopia, M. curtisiae from Kenya, and M. stricta from South Af-
rica. These all differ to a small extent from M. thomsonii^ but they have its char-
acteristic growth habit and peculiar terete leaf, characters which are considered
more significant taxonomically than the minor floral differences that are perhaps
to be expected in such a widely distributed species.

The typical form of M. thomsonii whicli occurs in Malawi and adjacent Zam-
bia and Tanzania has short, almost vestigial style crests, but has relatively broad
tepals and style branches. Plants occurring further to the north and also to the
south have well-developed style crests. The northern forms, however, have the
broad tepals and style branches of the type, while the South African form, earlier
treated as A/, stricta by me (Goldblatt, 1973), has rather narrow tepals and style
branches. Plants occurring in Rhodesia are intermediate between the typical
and southern forms and usually have style crests.

Moraea thomsonii is one of the more widespread species in the genus, extend-
ing almost the length of the eastern half of Africa, from the eastern Cape Province
of South Africa to northern Ethiopia, It occurs in open grassland and typically
flowers towards the end of the dry season. The leaf and flowering stem are pro-
duced at different times, the long, terete leaf emerging after the first rains, and

attaining full size during the wet season.

dry

often becoming broken and decayed by the time the stem and, subsequently, the
flowers are produced. Owing to the widespread practice in Africa of burning
grasslands at the end of the dry season, the leaf of M. thomsonii is often charred
or is entirely destroyed, so that flowering specimens frequently lack leaves.

Moraea thomsonii is generally easy to recognize in the herbarium because
specimens usually lack leaves, or if these are present, it is obvious that the leaf
was dead when collected. Difficulty is sometimes experienced in deciding if the
leaf was in fact dry when gathered and then the terete nature of the leaf, the
short, often sessile branches and dry, almost transparent spathes are sufficient
for determination.

ErHU)i>iA. AHUsi: Mt. Chillalo, ^cott s.n. (K). begemdih & simkn: Debarrk, Ainhara,
Cliiovcmla 3007 (F). eritkea: Hamascn, Fiori 895 (F). Near Adi Nefas, Ilamasan, Pappi
4149 (F). Amba-Dcro, Pappi 3541 (F). Bclesa, Hamasen, Terraciano is- Pappi 343 (F).
At Taclcsan, Terraciano 6- Pappi 415 (F). harah: Mt. Achim, Bally 10054 (EAH, K).
siDAMo: 15 km SE of Mej^lidli, Burner 1827 ( F, K). Mogada forest, Mooncy 5469 (K).
Wadora, Mooncy 5629 (EAH, F, K). 15 km S of Adola, Bally 3137 (K). Agheremeiiam,
Gillctt 14549 (BR, EAH, F, K). Between Neglielli and Filtu, de Wilde 6675 (WAG).

Kenya, coast province: District aroimd Nyora, Rontledge 1908 ( K ). ceni^ral
PRoviNCK: Nairobi National Park, Verdeourt 3284 (K). Between Ngong and Kiknyu, van
Someren 1437 (EAH, K). Nanyuki, Watt s.n. (K). rift valley rrovince: Elgeyo, Bat-
tiscomhc 1184 (EAH, K). Kaptagat, Agnew ir Agnew 9031 (MO); Verdeourt 2148 (EAH,
K, PRE). Kapiyet, Daugli.sh 82 (K). Trans Nzoia, Tweedic 443 (K). Mt. Elgon, Lugard
548 (K). Kitalc*, Thorold 2752 (K). northern frontier: Near Kisima, Leakey 8544 (K).
nyanza ruoviNCE: Tinderet, Poviano s.n. (F). vSOUthern province: Near Kongoni River,
Vries b Fries 1538 (K, S, UPS). 60 km SE of Narok, Noyrosera, Curtis 676 (CH).

1977] GOLDBLATT— MORAEA

265

Malawi. NoirrHEHN hegkxv: Nyika National Park, Pawek 1390 (SRGII). Nyika Pla-
teau, Robson 231 (K, LISS, SRGII). central phovince: Between Dedza and Nelien,
Jackson 148 (K). Clintembwe, Kota Kota district, Brass 17579 (K, MO), soutiiehn hegion:
Mlanje Mt., Burtt Davy 21981 (K); Pawck 3795 (K). Zomba Mt., Salubeni 89 (MAL,
SRGFI). Mt. Malosa, W/i(/f^ 5,n. (K).

Mozambique. manic:a & sofala; Barue, Serre de Ghoa, Mendonga 301 (LISG). Be-
tween Skeleton Pass and Ghimaniinani Plateau, Grosvenor 225 ( LISC, SRGH ) . niassa :
Nainiamba, near Vila Cabral, Mendoii(;a 776 (LISG).

Rhodesia, eastern region: Chiinanimani Mts., Mujich 325, 326 (SRGII). The Cor-
ner, Chimanimani Mts., Wild 3350 (K, SRGII); Sturgeon s,n. (SRGH-3()671, K, LISG,

PRE); Chase 2973 (BM, MO, SRGII). Melsetter, Plowes 2460 (SRGH), 2800 (LISG,
SRGII). Banti North, Unitali, Chase 7828 (SRGH). Liyanga, Leach 8139 (K, SRGH).
Mare River, Inyanga, Wild 3859 ( K, MO, SRGII). Slopes of Rukotso, Inyanga, Phipps 734
(SRGII). S of Pun^rwe View, Methuen 309 ( K).

Sudan, equatoria: (reported as Uganda) Iniatong Mts., Eggeling 3558 (K).

Tanzania, central rrovince: Wotta, Mpwapwa district, Gane 133 (EAH, K). eastern
province: Morogoro, Schhehen 1237 (BR, K). noi^thern province; Ngorongoro, Tanner
3863 (K). Liliondo, Masailand, Foshrooke 20 (K); Gihhins s.n. (K). W slopes of Kiliaman-
jaro, Greenway 6704 (K, PRE). Hanang Mt., Burtt 4010 (K); Geilinger 1313 ( K, Z). Mpo-
lolo, Moshi district, Haarer 1478 (K). Oldeani Volcano, Burtt 4232 (K). southern hic;ii-
lands: Rnhndje-Lupembe, Schlieben 1237 (K, LISG, MO, PRE). Iringa district, Greenway
3429 (EAH, K). Mbeya Mt., Geilinger 2797 (K,Z). western province: Hem, Kasulu dis-
trict, Eggehng 6195 (BR, EAH, K). S of Sisaga, Kigonia district, Jefford ir Newbold 1770
(K). Keto Mt., 15 km from Zambia border, Richards 6186 (K).

Uganda. Without precise locality, BaUy 10748 (K). northern provinc;e: Mt. Debasien,
Eggeling2693 (K).

Zambia: eastern region: Lundazi, Nyika Plateau, Pawek 2854 (K, MAL). northern
REc;ioN: Lumi marsh, Kawimbi, Richards 6116 ( K).

Subgenus Grandiflora Goldbl.

8. Moraea spathulata (Li.) Klatt in Tb. Durand & Schinz, Consp. Fl. Afr. 5:
152. 1895.— Fig. 7A.

Iris spathxdata L.f., Suppl. Pi. 99. 1781. type: South Africa, Cape, Langkloof, \\'olwekraal,

Tliunberg s.n. (Herb. Thunber^ 1172, UPS, holotype).
I. spathacea Thunb., Diss. Iride no. 23. 1782, type: as for 7. spathulata L.f.
Moraea spathacea (Thunb.) Kcr, Bot. Mag. tab. 1103. 1808, non Thunb., 1787.

M, longispatha Klatt, Linnaea 34: 560. 1866. type: South Africa, Cape, Transkei, banks

of Kei River ( "Tambikuland"), EckJon ir ZeyJier Irid. 3 (MO, lectotype).
A/, spathulata subsp. transtaalensis Goldbl., Ann. Missouri Bot. Card. 60; 253. 1973. type:

South Africa, Transvaal, near Sabie, Goldblatt 610 ( BOL, holotype; MO, isotype).
A/, spathidata subsp. saxosa Goldbl., Ann. Missouri Bot. Gard. 60: 254. 1973. type: South

Africa, Transvaal, summit of Long Tom Pass, Goldblatt 612 (BOL, holot)'pc; PRE,

isotype).
M. spathulata subsp. autumnalis Goldl)!., Ann. Missouri Bot. Gard. 60: 254. 1973. type:

South Africa, Cape, Transkei, Nyameni Mouth, Port Edward district, Strey 8619 (PRE,

holotype; NH, isotype).

Plants large, 50-90 cm higb, solitary or in small clumps. Conn 1.5-2 cm in
diameter; tunics of brown, finely reticulated fibers. Prophijlls prominent, brown
to pale, firm in texture, brittle, dry, entire or irregularly broken, or frayed at the

apex. Leaf solitary, flat or canaliculate, to 1.5 cm wide, ^teni simple or occa-
sionally bearing one branch. Bract leaves 2-3, often dry and brown, to 15 cm
long, rarely overlapping. Spatlies herbaceous, or becoming dry and brown from
the apex, attenuate; inner spathe 10-14 cm long, outer ca. ^^ the inner. Flower
pale yellow; outer tepals 3.5-5.5 cm long, the limb 2-3.5 cm, spreading; inner
tepals 3-4 cm long, erect. Filaments 8-12 mm long, free in the upper third;

266

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Figure 6. Distribution of Moraea spathulata.

8-12 mm long. Ovary 2-3 cm long; style branches 1.2-1.8 cm long, the

crests to 1 cm long.

5^5

Chromosome

12 ( South African plants only ) .

number 2n

Flowering time: (November) December to March (early April).

Distribution: Open grassland in the Inyanga and Chimanimani highlands of
Rhodesia and Mozambique, also in South Africa, Lesotho, and Swaziland. — Fig.
6.

The tall, summer-flowering Moraea frequently collected in the Chimanimani
and Inyanga highlands of Rhodesia and Mozambique is clearly closely allied
to the A/, spathulata complex of South Africa. The plants from the Chimanimanis
in particular are virtually identical to tlie Transvaal and Swaziland forms which

1 previously referred to M. spatliulata subsp. saxosa Goldbl. (Goldblatt, 1973).
Since 1973 when I recognized AL spathulata as comprising four subspecies, I
have had the opportunity to observe living plants in several places and to see
more herbarium material. As a result, it has become clear that my attempt to sub-
divide M. spathulata was not satisfactory, and the proposed subspecies did not
accurately reflect the variation found in the species. The subspecies autumnalis
from the Transkei is merely a very early blooming coastal form and must be in-
cluded in the typical form. Moraea spathulata in the southern part of its range
in the Knysna district of the Cape Province blooms from July to September but
not unusually in June or even May, thus the March to May blooming subsp.
autumnalis is not particularly umisual. The remaining two subspecies, subsp.
tramvaalerms and subsp. saxosa, both from the Transvaal and Swaziland, dif-
fer from one another mainly in that the lower-altitude subsp. tramvaalemis

1977]

GOU:>nhATT—MORAEA

267

A

c

/,

n

fl

(XO.5).

FiGUUE 7. Monica species. — A. A/. sputhuhiUi. — li. A/, iuyangani, — C. M. schimperi

2«c ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

forms clumps, while the higher-iiltitiide subsp. saxosa appears to be predomi-
nantly solitary in habit and has a slightly longer ovary and capsule. However,
when the entire pattern of variation in M. spathuJata is considered, both the
tendency to solitary habit, and the longer ovary of subsp. saxosa and the features
of subsp. transmalensis now seem altogether too insignificant to make any
taxonomic recognition worthwhile. It seems far more useful to recognize M.
spathuhta as a single, complex, and variable entity.

North of the Limpopo River M. spathuhta thus extends as far as the Inyanga
region of Rhodesia-Mozambique. The largest forms occur in Inyanga, some
plants even bearing branches while rather dwarfed forms occur in the Chima-
nimanis. Differences between the two appear significant until the few collections
from the intervening highland areas in Mozambi(iue are considered, and these
prove intermediate in all characteristics.

Mozambique. m.\nica & sofala: Tsetscrra, Excll, Men(Ion<,-(i 6- Wild 332 (LISC),

Torre 6 Corrcia 15687 (LISC). Gorongosa Mts., summit, Tinley 2436 (SRGII). Si-rra Mcs-
samlni/i, Torre <Lr Correia 13302 (LISC). Between Mussapa River and frontier at Tendcra,
Torre 6 Correia 13159 (LISC). Barue, Sena de Choa, Torre 6 Correia 15485 (LISC).

Rhodesia, eastehx region : Melsetter, Hanmcr s.n. (SRGH-18354); Chase 1421 (BM,

K, SRGII). Melsetter distriet, Mutsarara farm, Crook 318 (SRGII). G1(>ndingwe Estate,
rlowcs 3477 (SRGII). Pork Pie Hill near Melsetter, Bamps, Symoens <b van den Berfihen 776
(BR, SRGH). Inyanga, Chase 4351 (BM, K). Inyangani Mt., Chase 8124 (BM, K, LISC,
PRE, SRGII); Plowes 2429 (SRGH); WheUan ir Davies 984 (K, SRGH); Fries, NroUndh &
Weinwrelc 4967 (BM, BR, PRE). Mt. Gungingurwe, Stapleford, Wild 5711 (K, MO, PRE,
SRGII). Bant Nortli, Umtali distriet, Wihl 4522 (K, LISC, MO, PRE, SRGH). Himalayas-
Engwa. Wihl 4456 ( K, LISC, SRGH). Vumba Mts. near Umtali, Obermeyer 2113 (PRE).

9. Moraea schimperi (Hochst.) Pic.-Serm, Webbia 7: 349. 1950.— Fig. 7C.

Uymeno stigma schimperi Hochst., Flora 27: 24. 1844. type: Ethiopia, Begemdir & Simen
"Ensehedcap," Schimper 1173 (B, holotype; BM, F, K, M, MO, P, S, isotypes).

U. tridentatum Hochst., Flora 27: 25, 1844. type: Ethiopia, Begemdir & Simen, Bainam,
Mt. Baehit, Schiwper 1296 (K, lectotype).

Vieusseuxia sehimperi (Hochst.) A. Rich., Tent, Fl. Abyss. 2: 305. 1850 (Hist. Nat. Bot.

V. 5).
V. tridentata (Hochst.) A. Rich., Tent. Fl. Abyss. 2: 305. 1850.
Iris diicrsifoJia Steud., m.s. (cf. A. Rich., Tent. Fl. Abyss. 2: 305. 1850).
Xiphion diversifolium Steud. ex Klatt, Linnaea 34: 572. 1866, nom. illeg. superfl. type: as for

Ilymenostignia schimperi Hochst.
Moraea diversifoliu (Steud. ex Klatt) Baker, J. Linn. Soc, Bot. 16: 130. 1877.
M. wehvitschii Baker, Trans. Linn. Soc. London, Bot., Scr. 2, 1: 270. 1878. type: Angola,

Iluila, Lopollo River, Welwitsch 1548 (BM, holotype; K, LISU, P, i.sotypes).
M. zamheziaea Baker, Fl. Trop. Afr. 7: 340. 1898. type: Zambia, Mangaja hills, Mclhr

s.n. (K, holotype).
loekii De Wild.. 1

11: 540. 1913. type: Zaire, Shaba,

between Buggege and Lukcmi, Hock s.n. (BR, holotype).

Plants 20-40 cm high, unbranchcd. Conn 1.5-2 cm in diameter; tunics brown,
fibrous. ProphyUs 3-5, large and prominent, entire, firm, usually dark brown.
Leaf solitary, basal, canaliculate to flattened, 9.6-15 mm wide, emerging as
flowering begins, but eventually much exceeding the inflorescence. Stem usually
short at flowering time but elongating in fruit. Bract leaf usually solitary, 10-15
cm long, becoming dry and brown. Spatlies usually dry and brown, (6-)7-10
(-12) cm long; outer .spathe about % the length of the inner. Flower blue purple
with yellow white nectar guides; outer tepals lanceolate, 4-6.5 cm long, the limb

1977] GOLDBLATT— ^A/OfiAEA

269

equal to or slightly exceeding the claw; inner tepals erect, 3.5^.5 cm long. Fila-
ments 9-15 mm long, united in the lower half; anthers 8-12 mm long. Ovary
1.5-2 cm long; style branches 1.5-2 cm long, the crests 1-2 cm long. Capsule
2.5-3.5 cm long; seeds flattened and ±discoid. Chromosome number 2n = 12.

Flowering time: From the end of the dry season into the early w^et season,
August to November (December) south of the equator; November to May, oc-
casionally until July in West Africa and Ethiopia.

Distribution: Widespread from Rhodesia north and west through Central
and East Africa, to Angola, Zambia, Malawi, Mozambique, Tanzania, Zaire,
Burundi, Sudan, Ethiopia, Cameroons and Nigeria; often in marshy or wet situ-
ations, occasionally in grassland. — Fig. 8.

Moraea sehimperi has been given a surprising number of names which seems
remarkable in the light of its singular morpliology and habit. This is probably
explained more by its great geographic range than by any morphological dif-
ferences, and in fact most of its synonyms have been applied to plants from
different areas of Africa. Moraea sehimperi extends from the Rhodesian-Mozam-
bique highlands in the south, west to Angola and north, through Zaire and
Burundi to southern Sudan and Ethiopia. It also occurs in the higher areas of
Nigeria and Cameroons, with a large break in the range over the Congo Basin.
Significantly it is also absent from Uganda and Kenya, where the highlands may
be of too recent origin for it to have become established.

Moraea sehimperi usually occurs near a permanent water source, either un-
derground or at the surface, in a vlei or dambo, or along a stream, but is also
found in well -drained grassland in areas of high rainfall. It flowers at the end
of the dry season and continues in bloom in the early wet season. Thus in the
southern tropics it blooms from September through October and into November
and early Decem])cr. On the equator and to the north, plants most often begin
flowering in December to February, and continue into March and May. The
morphology of M. sehimperi is most distinctive. The flowering stem appears be-
fore or as the new season's leaf emerges, and early in the season the bright blue
flowers are borne well above the leaf apex. The prophylls, bracts, and spathes
are large, and their usually rich brown color is alone sufficiently characteristic
to permit determination. As the season progresses, stem and leaf continue to
grow, and by the time the seeds are ripe the stem may be a meter high and the
leaf much longer. There is only a small degree of variation from this pattern,
notably in plant size, with forms from Nigeria-Cameroons and also some in
Ethiopia being more slender than the comparatively robust plants in the southern
part of its range.

Though clearly related to the M. spathulata complex, as can be seen from
the similarity of the bracts and prophylls, M, sehimperi is unusual in this al-
liance in having blue purple flowers. The treatment of M. sehimperi here
differs somewhat from that of Geerinck (1970) in that M. arnohJiana is not con-
sidered a synonym. The latter is a late-blooming species from Zaire and is as-
signed to synonymy under M. macrantha in this treatment.

Angola, bie: Cliiteni1)o, Barbosa 6- Moreno 12245 (LISC). cuanza sul: Murta ir
Silva 857 ( COI, LISC). iiuamho: Rio Ciumza, Mucosa, Montciw ir Muiia 1884 ( COI,

270

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Fic;uuE 8. Distrilnition of Moraea schimpcri.

LISC, SRGH). Huanibo (Nova Lisboa), da Silva 3188 ( LISC, PRE), huila: Missae Cato-
lica, Iliiila, Santos 62 (LISC). Lopollo River, Wclwitsch 1548 (BM, COI, K, LISU, P).
Near Vila Arture de Paiva, Leach ir Cannell 13861 (K, LISC). Lnhango (Sa de Baiu-ira),
Rio Caia, llenriques 6 Brites 1137 (K, LISC, LISU, PRE). Ilumpata, Memlcs 1816 (LISC).
Chcila Mts., GcmciJcr 13316, 13317 (LISC). malanje: Cossweilcr 922 (P).

BuHUN-ni. Biimri, Becquet 162 (BR, EAH, K, WAG); Reckmans 1041 (BR, EAII).

Near Kuniuyange, Lcwallc 6131 (BR, MO).

Cameroon, west: Bauer 22 (K). Bamenda, 4 km NE of Banibili, Bauer 154 (K).
Bftwtrn Bamenda and Santa, Kcaij sm. (FHI-28348, K, P). east: Nganha Mts., E. Ngoaun-

1977] GOLDBLATT AfOfiAEA

271

derc, Jacques-Felix m32 (P); de Wilde 4521 (WAG). E of Ngoannck'rc, Rai/nal 6 Rai/nal
12236 (P); de Wilde 4312 (WAG). Dschang, Mt. Banib(nitos, ViUicrs 672 (P); Jacques-
Felix 2707 (P); Saxer 88 ( K, WAG, Z). Dscliang, MeuriUon 765 (BR). Djuttitsa, Meuril-
lon 179 (BR), 74,5 (P), 1212 (BR, K, P). Mt. Santa, Jacques-Felix 2810 (P). Gotcl Mts.,
NNE Banyo, Letouzey 8592 (P).

Ethiopia, ahusi: Between Shashamane and Dodollo, de Wilde 6823 (\\'AG). Near
Ascla, de Wilde 6595 (MO, WAG). Near Kofole, Mocmey 7106 ( F, K, S). Between Bekoji
and Adaba, Gilbert 1150 A (K). begemdih & simen: Near Miehubbi, Fiehi-Sermolti 2674
(F, K). "Enschedcap," Sehimpcr 1173 (B, BM, F, K, M, MO, P, S). Sering, Schimpcr 1536,
1537 (P). harah: Gara Mnllata Mts., Gillctt 5332 (EAR, F, K, P, S); Burger 1706 (K),
2.933 (F, K); de Wilde 6335 (WAG), shoa: Negri 413 (F). Bet^^'een Addis Aba1)a and the
Bhie Nile, Buscalioni 957 (F). Mt. Zuguola, Busealioni 2158 (F). sidamo: S of Agere Selain,
de Wilde 6721 (WAG). SE of Agere Selani, de Wilde 10298 (MO, PRE, WAG), wolega:
Bari, Smeds 1241 ( K). N of Beiea, Mooneij 7729 ( EAH, K).

Malawi, nohthehx region: Chelinda Bridge, Nyika, Paicek 1407 (SRGII). Nyika
Plateau, Cottrell 3, 65 (SRGH); Rohson 326 ( BM, BR, K, LISG, SRGH). Mzu/.n, Mary-
mount, Puwek 5833 (MO, SRGH). Vipya Mts., Chapman 1702 ( K, LISC, SRGII); Garley
576 (SRC;H). central province: Dedza, Brass 17632 ( K, MO, SRGII); Kulunude 2 ( K,
SRGII). Chongoni Forest Reserve, Dedza district, Saluheni 829 (MAL). southern i'rovince:
Chambe Plateau, Mt. Mulanje, Neuman 6 Whilmorc 653 (BR, K, SRGII, WAG). Mt. Mulanje,
Forbes 67 (EAH); Greemcaij 6306 (EAH, J'RK). Lichenya Plateau, Mulanje Mts., Paicek
3793 ( M AL) . Linibe, off Cholo road, Moriarty 399 ( MAL ) .

Mozambique, massa: Amaramba near Mandiniba, Mendor\(;a 666 (LISC). tete:
Between Furancungo and Angonia, Torre 3362 (LISC). zambezia: Namuli hills. Last s.n.
(K). Near Gurue, Torre 6 Correia 15923 (LISC). Serra de Guiue, Mendon(;a 2229 (LISG).

Nigeria, northern region: \'oin, Jos Plateau, Dent Young 245 (K); McC Unlock 199
(K). Mambila Plateau, Tuley 1929 (K); Latilo I- Daramola s.n. (FIII-3438(), BR, K, S,
WAG); Wit, Ghile & Daramola 2076 (Fill); Hepper 1774 (BR, K, P); Gbile 6 Daramola
s.n. (Fl 11-62880, WAG). Between Nguroje and Maisaniari, Chapman 2640 (Fill). Vogel
Fcak, llepper 1504 (K, P).

Rhodesia, eastern region: Melsettcr, Markhanis Kloof, Crook M151 (K, LISC, MO,
PRE, SRGII). Inyanga downs, Drewe 18 (SRGH). central region: Salisbury, Eyles 1791
(K, PRE, SAM, SRGII). Cleveland Dam, Wild 3871 ( K, SRGH). xMarandellas, Rattray 562,
309 (BM, SRGII); Corby 196 (SRCIH).

Sudan, equatoria: Imatong Mts., Andrews 1895 (K); Johnsion 1492 (K); MacDonald
80 ( BM ) . Kippi, CJiipp 89 ( K ) .

Tanzania, southern highlands: Poroto Mts., Geilinger 2611 (K, Z); McGregor 3 ( K,
EAH). Between Cbunya and Mbeya, Burit 6225 (BM, BR, EAH, F, K). Chunya escarp-
ment, Richards 25795 (K). Njombe district, Richards 6585 (K); Gillett 17781 (EAH, K).
Itaka, Greenway 3651 (EAH, K, PRE). Mnfindi, Pagel-Wilkes 261 (EAII, MO). Elton
Plateau, Richards 18460 ( K, SRGH); Davies s.n. (K). southern province: 8 km W of Songea,
Milne-Redhead ir Taylor 8377 (K). northern province: Between Dongobesh and Mkulu,
Uaarer 11613 (K). western prontnce: Sumbawanga, Vcni Ren.sburg s.n. (EAH).

ZaThk. SHABA (Katanga): Care de Biano, ScJvnitz 4893 (BR, W^AG). Kipopo,
Schmitz 5495 (BR). Pare Nationale de I'Upemba, Robyns 3940 (BR). 70 km S of Jadot-
ville, Schmitz 4055 (BR). Lubudi, Kendo, de Witte 537 (BR, WAG, Z).

Zambia, northern region: Mbala (.Abercorn) district, Lumi River flats, Richards 5844
(BR, K, SRGH). Ndundu, Mbala district, Richards 13138 ( K, SRGH). Mansa (Fort Rosebery),
Fries 608 (UPS). N of Lake Shiwa Ngandu, Greenway 6 Trapnell 5709 (EAH). central
rec;ion: Serenje, Fanshawe 6727 (SRGII). eastern region; Nyika Plateau, Robinson 3012
(K, M, PRE, SRGH). western region: N of Kakeuia River, Mwinilunga district, Milne-
Redhead 965 (VRE) .

10. Moraea muddii N. E. Brown, Trans. Roy. Soc. S. Africa 27: 346. 1929.
tyi'e: South Africa, Transvaal, Mac Mac Creek, Sabie district, Mtichl s.n.
(K, holotype).

Flants l5-40(-70) cm liigli, solitary. Conn ca. 1.5 cm in diameter; tunics fine,
usually pale. FrophyUs, short, dry, pale brown, usually irregularly broken. Leaf
linear, canaliculate ca. 5 miii wide, exceeding the inflorescence. Stem un-

272

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

branched. Bract leaves 2-4, herbaceous, 8-12 cm long. Spathes herbaceous;
inner spathe 7-12 cm long, the outer ca. % the inner. Floicers yellow; outer
tepnls lanceolate, 3.5-5 cm long, the limb 2-3 cm long; inner tepals to 3.5 cm
long, erect. Filaments to 1 cm long, free in the upper third; anthers 8-11 mm
long. Ovary ca. 1.5 cm long; style branches to 1.3 cm long, the crests to 1 cm
long. Capsule ca. 2 cm long; seeds flattened and disc shaped. Chromosome

number not known.

Flowering time: Late August to November.

Distribution: Eastern mountains of southern Africa, from the eastern part
of the Cape Province to the Chimanimani Mountains of Rhodesia and Mozam-
bique; usually in wet situations in grassland. — Fig. 9.

The few collections of a rather short, large yellow-flowered Moraea from
the Chimanimani Mountains of Rliodesia and Mozambique match very closely
M. mudclii, a species previously recorded only from South Africa. The Chimani-
mani plants resemble M . muddi not only in form, but like it, are early bloom-
ing, and generally occur in moist habitats.

The range of M. muddii extends from the Chimanimanis south, along the
eastern escarpment of South Africa to the Hogsback in the eastern Cape. The
Rhodesia-Mozambique plants differ mainly in their rather early flowering, from
August to September, in contrast to October to November in South Africa, but
otherwise correspond very closely.

Moraea muddii is related to the M. spathulata complex, particularly to soli-
tar>^-growing species such as M. mog<i,ii, M. hiemalis, and M. carnea, amongst
others, and is distinguished by its shorter size, often smaller flower, and early
blooming habit.

Mo'/.AMHiyuE. mank:a & sofala: Near suininit of Chimanimani Mts., Grosvenor 195

ruse, SRGH, UPS, WAG).

Rhodesia, eastern region: Chimanimani Mts., Carley 177 (SRGIT); Lovericlgc A42

(SUCH). Bank of Bundi River, Melsottcr district, WhcUan 2153 (SRGH)

11. Moraea inyangani Goldbl., sp. nov. type: Rhodesia, vlei on Mt. Inyangani,
8,000 ft. Wild 5519 (SRGH, holotvpe; K, LISC, M, PRE, isotypes ) .— Fig.
7B.

Planfa parva, 15-30 cm alta. Tunicac cormi pallidae, tcnuissimae. Folium solitarium,
canaliculatuni, infloresccntium exccdens. Caulis simplex forens 3 (-4) bracteis. SpatJuif
herbaceao, intrriora 5-8 cm longa, exterior paiilo brevior. Flares luteis; tepala exteriora ca.
2.5 cm longa, liinlx:) ca. 1.5 cm longo, cffuso; interiora 1.5-2 cm, erecta. Filamenta 4 mm
longa; anthenie ca. 6 mm longae. Germen ca. 1.4 cm longum; rami styli 8-9 mm longi; cristac
4-8 mm longac.

Plants small 15-30 cm high. Corm to 1 cm in diameter; tmiics of fine, pale
fibers. Prophylls brown, papery, irregularly broken. Leaf solitary, canaliculate,
appearing terete with the margins tightly inrolled, to 3 mm wide and exceed-
ing the inflorescence. Stem unbranched. Bract leaves 3(^), herbaceous, 5-8
cm long. Spathes herbaceous with dry, brown acute apices; inner spathe 5-8
cm long, the outer only slightly shorter. Flowers pale yellow; outer tepals ca.
2.5 cm long, lanceolate, the limb ca. 1.5 cm long, spreading; iiiner tepals erect,
1.5-2 cm long. Filaments 4 mm long, free in the upper half; anthers ca. 6 mm

1977]

GOLDBLATT— MORAEA

273

10

20

M. muddii

H. inydngani

M. anyolensis

e 100 300 100 400 900 IDO MILES

3O0\i00 100 too 1000 NH

". '-'"J

Fu;uHE 9. Distribution ot Monica muddii^ Af. imjangani^ and M. angolcnsis.

long. Ovarij ca. 1.4 cm long; style branches 8-9 mm long, the crests 4-8 nnn
long. Capsule and seeds unknown. Chromosome number not known.

Flowering time : September to October ( also in April ) .

Distribution: Rhodesia, endemic on Mount Inyangani, Inyanga district, in
moist situations at high altitudes. — Fig. 9.

The distive, small-flowered Moraea inyang^ani appears to be a very local spe-
cies, having been recorded only from the higher altitudes of Mount Inyangani,
in the Inyanga Highlands of Rhodesia. It is most closely related to A/, muddii,
which occurs to the south on the Chimanimani mountains of Rhodesia and
Mozambique and in South Africa. Moraea muddii has a larger flower and a
canaliculate leaf in contrast to the subterete leaf of M. inyangani. Both species
appear to occupy the same habitat, high altitude grassland in moist situations,
and both flower early in spring.

Rhodesia, eastehn hec.iok: Inyanga district, Mt. Inyangani, Leech 8140 ( SRCH ) ;
Corhtj 808 ( K, SRGFI). Summit ridge of Inyangani, damp flush, Drummond ir Robson 5830
(K, PRE, SRGII). Inyangani, wet flush below summit, Boughey 264 (SRGH). Vlei on Mt.
Inyangani, Wild 5519 ( K, LISC, M, PRE, SRGII); Burrows 570 (SRGH).

12. Moraea angolensis Goldbl., sp. nov. type: Angola, Moxico, Rio Culiti,
Capello ir Ivens IS ( LISU, holotype ) .

Planta graciles. Folium solitarium. Cauli,s 35-40 cm alta, bracteis tria, ilia 4-4.5 cm longa.
Spatha 4-5 cm longa. Flores par\i; tcpcda exteriora ca. 3.5 cm longa, ungues ca. 2 cm; iepcda
interiora ca. 3 cm longa. Fihunenta ad 11 mm longa, libera ad ai^icem 2 mm; antherae 6 mm
longae. Germen 1-1.2 cm longum; rami styli ca. 8 mm longi; cristae ca. 8 mm longae.

Plants slender, 35-40 cm tall. Conn unknown. Leaf solitary, basal, ca. 2 mm
wide and much exceeding the inflorescence, channeled. Stem erect. Bract leaves

274 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

3, 4-4.5 cm long witli dry brown apices. Spathes 4-5 cm long, herbaceous, with
the upper 5 mm dry and brown; outer spathe about % as long as the inner. Flow-
ers ? yellow; outer tepals ca. 3.5 cm long, the claw 2 cm long; inner tepals ca. 3
cm long. Filaments to 11 mm long, free near the apex for 2 mm; anthers 6 mm
long. Ovarii 1-1,2 cm long; style branches ca. 8 mm long, the crests to 8 mm
long. Capsule and seech unknown. Chromosome number unknown.

Flowering time: August.

Distribution: Plains of southeastern Angola. — Fig. 9.

It is with some hesitation that I describe this species as it is based on two
very poorly preserved specimens. The holotype and only gathering was col-
lected by Capello and Ivens in August 1885 on their expedition across Africa
from Angola to Mozambique. Dr. L Melo of the University of Lisbon and Dr.
E. Mcndcs, Director of the Junta de Investigacoes do Ultramar, Lisbon, Portu-
gal have kindly aided me in localizing the place of collection, Rio Culiti, as soutli-
eastern Angola between latitude 14"50'S and 15°50'S and longitude 19^20't:
and 21''50' E. No other species of Moraea are known from anywhere near this
area. The plants are quite different from any other tropical African species, per-
haps most resembling M. mudclii and M. intjangani both from the highlands of
southeast Africa. The slender form, very short bracts and spathes, combined
vvitli a small flower make it easily distinguishable from these.

Angola: moxico, Rio Culiti, CapcJIo i^ Ivcns 18 ( LISU).

13, Moraea bella Harms, Bot. Jahrb. Syst. 28: 364. 1901. tyi'E: Tanzania,

Southern Highlands, Uhehe, Goetze 698 (B, holotype). — Fig. 10.

Plants solitary, slender, ( 30- ) 40-60 ( -70 ) cm high, very rarely branched.
Corm ca. 1.5 cm in diameter; tunics of pale medium to coarse reticulate fibers.
ProphyUs small, light brown, dry and irregularl)^ broken. Leaf solitary, canalic-

5-6

&

Stem slender. Bract leaves 2-4, widely spaced, 6-7 cm long. Spathes dry and
brown in the upper part or entirely .so; inner spathe 6-10 cm long, the outer
V:i to Ml the length of the inner. Flowers pale yellow, with a darker nectar
guide and often conspicuously veined and spotted; outer tepals 4.5-5.5(-6.5)
cm long, the limb 2.5-4 cm, spreading; inner tepals 3.5-5 (-6) cm long, erect.
Filaments 9-14 mm long, free in the upper third; anthers S-10 mm long. Ovary

5-2

Capsule

ovate-oblong, 2-3 cm long; seeds flattened, ± triangular to discoid. Chromo-
some number not known.

Flowering time: (Late February) March to June (July).

Distribution: Northern Mozambique, Tanzania, Malawi, Zambia, Zaire;

onl}^ in seasonally or permanently waterlogged habitats such as marshes, vleis,
dambos; flowering towards the end of the wet season and in the dry season.
Fic. 10.

Moraea hella can readily be recognized by the combination of several moipho-
logical features and its moist habitat and late flowering. It is a slender plant

1977]

GOLDBLATT— MO/MEA

275

LJ

1-1M1

Figure 10. Morphology and distrilnition of Moraca hcUa (xO.5).

with a long, narrow leaf,

2-4 comparatively short, widely spaced bract

leaves. The yellow flowers, which are sometimes characteristically speckled,
appear from late February through the end of the wet season and well into the
dry season in July. The peak flowering period for the species is in April, Moraea
hella is almost invariably found in very wet situations, usually in dambos but

276

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Fk;uhe 11. Distribution of Moraea verdickii and A/, macrantha.

on occasion simply in damp, poorly drained grassland which retains moisture
well after the rains are over.

Though described early this century, the name M. bella has been until now
overlooked. Plants belonging to this species have either been assigned to M.
angusta, a common and quite unrelated Cape species, or were included in other
species, as for example by Geerinck (1970) who included specimens of M. hella

1977] GOLDBLATT— AfOfiAEA

277

in his very broad concept of M. textilis. Moraea bella is closely related to M.
verdickii, a larger- flowered species with the same general range and to the M.
macrantha-ventricosa-textilis complex of south central Africa.

Malawi, northeun region: Mziniba, Benson 1195 (K), 1339 (BM). Mak-m-
bo, Mzimbii district, Pawck 8460 (K, MO, SRGH). Fort Hill, Whijte s.n. (K). Near
Fort Hill, Pole Evans 6 Erens 705 (BR, K, PRE, SRGH). Chelinda bridge, Runiphi district,
Patvek 2196 (K). central region: Mchinji district, Bmmmitt 10206 (K). Kasungu Game
Reserve, Bnimmitt 11603 (K); Ilill-Martin 1776 (PRE, SRGH). 22 km S of Kasmigu,

Moriarty 355 ( M AL ) .

Mozambique, niassa: yildCdhnil, Pedro ir Peclrofiao 3629 (EAH).

Tanzania, southern highlands: Iringa, Ltjnes 269 (K); Pedersen 990 (DSM); van
Rcnshurg 677 (K). Between Iringa and Dabaga, Eggcling 6122 (BR, EAH, K). Lunziia Ag-
ricultural Station, marsh, Richards 5173 (BR), western province: 65 km S of Sumbawanga,
Leach & Brimton 10072 ( K, LISC, SRGH). Ufipa district, Isopa, Robertson 641 (K, WAG).

Zaire, shaba: Near Kapona, Plateau des Muhila, Schmitz 1652 (BR). Manmgu-
Katomia, tan den Bromde 1 (BR). Jadotville, vallee de la Mulende, Dubois 1274 (K), 1074
(WAG). Kankela Valley, Manika Plateau, Malaisse 8564 (BR).

Zambia, eastern region: Chipata (Fort Jameson) Fanshawe 4501 (EAH, K, SRGH);
van Renshnrg 2111 (K, SRGH); Munch 459 ( K, LISC, SRGH). Lundazi, Famhawe 9291
(K, SRGH). Lukusuzi Game Reserve, Sayer 205 (SRGH). central region: Chakwenga
headwaters, E of Lusaka, Robinson 6473 ( K, SRGH). Kawambura, Fansliawe 3674 (K).
Between Undaunda and Rufunsa, Kornas 1519 (K). northern region: N of Mbala (Aber-
corn), Greenway 6219 ( K, EAH, PRE). "Fwambo," Nutt s.n. (K); Carson s.n. (K). Mbala
district, Kawimbc road, Richards 21407 ( K, MO); Sanane 1225 ( K, PRE); Robertson 618
(EAH, K); Richards 897 (K), 1367 (K). Kambole escarpment, Richards 9989 (K). Chisinga
Ranch near Luwingu, Astle 524 ( K, SRGH), 3017 (SRGH).

14. Moraea verdickii De Wild., Ann. Mus. Congo, Ser. 4, Bot. 1: 17. 1902.
type: Zaire, Shaba, Lukafu, Verdick 281 (BR, holotype).— Fig. 12A.

Plants large, 45-75 cm high, unbranched. Conn ca. 2 cm in diameter; tunics
coarse, of pale or occasionally dark fibers. Prophijlls pale to dark brown, dry,
broken and fibrous apically. Leaf linear, shortly exceeding the inflorescence,
canaliculate or ± flat, 4-12 mm wide, inserted near the base but often up to 10
cm above ground. Stem bearing 2-3 (-4) herbaceous bract leaves, usually widely
spaced but occasionally overlapping. Spathes- herbaceous, with a brown, acute
apex; inner spathe 9-15 cm long, the outer ca. % the inner. Flowers yellow;
outer tepals (5-) 6-7.5 (-10) cm long, the limb 3-5 cm long, spreading; inner
tepah erect, 4..5-7 cm long. Filaments 11-16 mm long, free in the upper third;
anthers (10-) 11-14 mm long. Ovary usually cxserted, 1.5-2 (-2.5) cm long;
style branches 1.7-2.5 cm long, the crests 1-1.7 cm long. Capsule and seeds not
known. Chromosome number not known.

Flowering time: Late November to February ( March) .

Distribution: Grasslands and open bush, occasionally in damper situations,
in eastern Angola, northern and western Zambia, Zaire, Tanzania, Malawi, and
central Mozambique. — Fig. 11.

Moraea verdickii is one of the taller species of Moraea and has particularly
large flowers. It occurs in a belt across south tropical Africa from eastern Angola
through southern Zaire and Zambia to Malawi, southwestern Tanzania and
Mozambique. Though recognized as early as 1902 by De Wikleman, it has in
the past been included in M. ventricosa in herbaria, and most recently was in-

278

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

(XO.5).

Fir.uHK 12. Monica species. — A, A/, vcrdickii. — B. M, vcntricosa. — C. M, textilis

1977] GOLDBLATT— AfORAEA

279

eluded with M. ventricosa in A/, textilis by Geerinck (1970) who held a very
broad concept of this species.

A detailed and critical comparison of a large number of specimens of M. ver-
dickii and its close allies, M. ventricosa, \L textilis, and also including the related
A/, maerantha was made in the course of this study. In my estimation the best
solution to the systematics of this group is to recognize all four species. Moraea
verdickii can ])e distinguished from its close allies by its large, consistently yel-
low flower, with the outer tepals ranging from (.5-)6-10 cm long and tlie anthers
(10-) 11-14 mm long; few bract leaves, generally 2 or 3; its habitat, grassland
and open bush; and its early flowering, from late November to February, oc-
casionally until March. Moraea maerantha has similar large flowers which are
always dark blue, has 4-5 bract leaves, and flowers later, from mid Fel)ruary to
July. Moraea ventricosa has smaller, blue or white flowers, 3-6 bract leaves,
often grows in wet situations, and flowers from February to May; while A/, tex-
tilis, with either yellow or blue flowers, generally has 5-7 bract leaves, and
reaches peak flowering in May.

AxGOLA. Moxico: Lusavo falls, Milnc-Rcdhcad 40SS (BR, K).

Malawi, centhal piuwinck: Dcdza, Jckc 79 ( K, SRGH). Chongoiii forest, Dcdza,
Banda 526 (SRGII); Siduhcni 1253 (K, MAL, SRCII). Dcdza mountain area, Adlard ir
CJiaprnanSGl (K, LISC, SRGII). Kirk Ran^c, Nchcu-Neno road, Robson lt04A (K, LISC,

SRGII).

M()ZA^n5IQUE. xiassa: 80 km from Mijjao dc Massangido to Vila Cabral, Corrcia 154

(LISC). Near Mtengida, Scddon 7 (K). tete: 8 km from Mlangeni, on Nehen-Dedza road,
Brummitt 8609 (K, SRGII). Kirk Range, Neh(Mi-Neno road, Rohson 1404 (K). Massangnlo,
Gomes 6 Sousa 1306 (K). zambezia: Grime ncdr Vila Quanquiero, Torre 5078 (LISC).

1'axzania. southerx liiGHLAxns: Iringa district, 25 km S of Dabaga, PoUiill 6 Paulo
1560 (BR, EAH, K, LISC, PRE). 50 km N of Mbeya, B(dhj 6 Carter 16475 (EAII, SRGH).
Ca. 120 km N of Mbeya on Itigi road, Boalcr 843 (K). southehx province: Matengo bills,
Songea, Milne-Redhead l- Taylor 9024 (LISC, PRE, SRGH).

Zaihe. shaha: SE Lnbnmbasbi ( Elisabetbville), Se/n/wYr 3722 (BR). Near Lnbnmbasbi,
Uirschherg 58 (PRE). Keyberg, Sehmitz 1230 (BR). Derubo, Qmnre 1008 (BR). Marie
Jose, Qiiarre 1524 (BR). Lnpakn River, Kassner 2466 (BR, K, Z). Nhu-nngu, Kisinde, Dubois
'l074 (BR, WAG). Mitwaba terr. River Kenia, de Witte 2470 ( K, SRGII).

Zambia, xortherx province : Kam1:)ole esearpment, Riehards 18881 ( K, UPS ). Item-
bwe Gorge, Ml)ala (Abereorn) distriet, Riehards 12055 (EAH, K, MO, SRGII). Itembvve
Gap, Riehards 18830 (K, SRGH). Mbala, Kalambo falls road, Riehards 17080 ( K, SRGII);
Bulloek s.n. (BR, K). Mpika, Fanshawe 1890 (BR, K, SRGII). 40 km N of Mpika, Wil-
liamson 1470 (K, SRGH). S of Ncbelenga, Kawambwa distriet, Williamson 1233 (SRGII).
Cbisbinga raneh, Astle 1395 (SRGH). western province: Matonebi farm, Mwim'lnnga

distriet, Milne-Redhead 3930 (BR, K, LISC, PRE).

15. Moraea maerantha Baker, Fl. Trop. Africa 7: 340. 1898. type: Malawi,
Northern Province, without precise locaHty, Whijte s.n. (K, holotypc).

M. arnoldiana De Wild., Ann. Mns. Congo, Scr. 4, Bot. 1: 16. 1902. tyve: Zaire, Slial)u,
Kasenga, Lukafu, Verdiek 606 (BR, holotype).

Plants solitaiy, unhranched, 50-70 cm high. Corni ca. 1.5 cm in diameter;
tunics of pale fine to medium fibers. Prophylls brown, soft textured and ir-
regularly broken to fibrous. Leaf solitaiy, linear, basal, canaliculate, to 7 mm
wide, much exceeding the inflorescence. Stem unbranched. Bract leaves 4-5,
herbaceous, usually overlapping, attenuate, 6-10 cm long. Spathes herbaceous,
with a dry acute apex; inner spathe 9-13 cm long, the outer ca. % the inner.

280 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. fi4

Flowers blue violet with pale nectar guides; outer tepah (5.7-) 6.5-8 cm long,
the limb ± equal or exceeding the claw, 3.0-4.5 cm long; intier tepah 5.5-7.5
cm long, lanceolate. Filaments 1.3-1.7 cm long, free in the upper half to third;
anthers 1.2-1.5 cm long. Ovary ca. 2 cm long, often enclosed in the spathes; style
branches ca. 2.5 cm long, the crests to 2 cm. Capsule ca. 3 cm long; seeds flat-
tened, ± triangular. Chromosome number 2n = 12.

Flowering time: (Mid February) March to early July.

Distribution: Open woodland or montane grassland in northern Malawi, east-
ern Zambia, southwestern Tanzania, and Zaire. — Fig. 11.

Moraea macrantha is closely related to A/, verdickii and M. textilis (see dis-
cussion under M. verdickii). It can be distinguished from M. verdickii by its
dark blue flowers, 3-5 bract leaves, and later flowering, from mid February to
July; and from M. textilis by its generally larger flowers, with tepals in the 5.7-8
cm range, and anthers 1.2-1.5 cm long.

Moraea macrantha is apparently common in northern Malawi and relatively
rare elsewhere, though it extends through southwestern Tanzania to the higher
areas of Shaba in Zaire. It occurs only in highland areas and flowers from mid-
to late-wet season.

Malawi, northkhn region: Vipya hills, Chapman H 640 (SRGII). Vipya plateaiK

Sahiheni 660 (K, LISC, PRE, SRGII). Vipya link road, Pawek 1060 ( K, MAL, SRGH)".

Vipya, Liiwawa, Chapman 1632 (K, SRGII). Mzuzu, Pawek 5396 (K, MO, SRGII), 5832

(MO, SRGII), 2448 (K, MAL). Mziniba district, Jackson 1283 (BM, BR, K, MAL). Rumphi

district, Pawek 3595 (K, MAL). Nyika Plateau, Pawek 6696 (MO, PRE). Chitipa, between

Misuku hills and Kalenje River, Pawek 5170 (K, MAL, MO, SRGII). 30 km S of Chikangawa,

Vipya Mts., Brummitt 10470 (K). Wenya, Benson 1325 (BM). Mafinga Mts. near Chisengo,
Chapman 1864 (SRGII).

Tanzania, southehn highlands: Ulanda, Ulambya, Leedal 599 (EAR). Kasehe,
Bundali, Leedal 439 (EAH).

Zaire, shaba: Liikafu, Mutoli-Tuli, Verdick sJi. (BR). Pare Nationale de rUpemba,
Pandeluru, de Witte 2611 (BR). Plateau des Kundulungu, Liwwski, Malaisse 6 Syrnocns
11621 (BR). Kibura Plateau, 25 km N of Mitwaba, Lisowski, Mahiisse i:r Symoens 13676
(BR). 5 km S of Poste de Katshupa, Lisowski, Malaisse 6- Symoens 5798 (BR).

Zambia, eastern trovince. Nyika Plateau, Richards 14388 ( K).

16. Moraea ventricosa Baker, Bull Misc. Inform. 1895: 13. 1895. type: Zambia,

Northern Province, "Fvvambo," Carson 37/1894 (K, holotype).— Fig. 12B.

M. bequacrtii De Wild., Repert. Spee. Nov. Regni Ve^. 11: 540. 1913. types: Zaire, Shaba
(Katanga), Lubumbashi, Bequacrt 316 (B, lectotype). Zaire, Tshisenda, Rin^oet 419
(BR. syntype); HomhU 360 (BR, syntype). Zaire, Lubumbashi, Ihnnhle 238 (BR, syn-

type ) .

(M

Corm ca, 1.5 cm in diameter:

tunics of pale medium to fine fibers. Prophylls pale to dark brown, broken and
becoming fibrous. Leaf exceeding the inflorescence, linear, canaliculate, 3-7
mm wide. Stem bearing 3-5(-6) overlapping, herbaceous bracts. Spathes her-
baceous; inner spathe 8-11.5(-13.5) cm long, the outer ca. % the inner. Flowers
comparatively small, blue purple or white to pale yellow; outer tepals 4-5(-5.5)
cm long, the limb equal or sHghtly shorter than the claw; inner tevals 3.7-4.5

cm long.

8-10(-ll)

long. Ovary ca. 2 cm long; style branches 1.4-1.7 cm long, the crests ca. 1

1977] GOLDBLATT— A/ORAEA

281

cm long. Capsule ovoid, to 3 cm long; seeds not known. Chromosome number
2n = 12.

Flowering time: Late February to early May (one record from September).

Distribution: Burundi, western and southern Tanzania, southeastern Zaire,
and northern and western Zambia; open woodland or grassland, often in moister
places along the margins of dambos, marshes, etc. — Fig. 13.

Moraea ventricosa as treated here includes both white (to pale yellow) and
blue, small-flowered plants. It is one of the few species of subgenus GramJiflora
where forms with different colored flowers are known. White-flowered plants
occur in the eastern part of its range in western Tanzania, Burundi, and north-
eastern Zambia, while blue-flowered plants predominate to the west, in Zaire
and northwestern Zambia, M. beqiiaertii from Zaire exemplifying the latter.
Though recorded from open woodland or grassland, Af. ventricosa is more
often found in moist situations as along dambo margins or poorly drained areas.

As discussed under M. verdickii, M. ventricosa is part of a complex of closely
allied species including M. macrantha and M. textilis. In fact, it appears most
closely related to the Angolan M. textilis w^hich generally has larger flowers,
with outer tepals in the 5-6 cm range and anthers 10-13.5 mm long. Specimens
of M. verdickii have in the past usually been included in M. ventricosa, but the
larger yellow flowers of the former, as well as its earlier blooming habit, should
be sufficient to prevent confusion.

Burundi. Gitega, Burasira, Lewalle 1728 (BR, K, WAG). Vicinity of Karuzi-Btireni,

van der Ben 1966 ( K ) .

Tanzania, southern highlands: Mtengulu, Kyiinbila district, Stolz 2625 (BM, BR,
K, PRE). Elton Plateau, S of Chimala, Milton 72 (EAH). Tunduma, Morcau ir Morcati
9807 (EAH). Tunduma-Snmbawanga, Lecdal 1076 (EAH). western province: Ufipa
district, vlei near Zambia border, Whellan 1212 ( K, MAL, SRGH). Near Chapota, Smii-
bawanga, Richards 8507 (K). Lake Kwela, Bullock 2650 (K). Kundi, Bullock s.n. (K).

Zaire, shaba: Lubumbashi ( Elisabethvillc ), Boitsfort, Robyns 1655 (BR, EAH, WAG).
Lubumbashi, Bequaert 316 (BR); Hirschhcrg 58 (K). Vallee de Lubumbashi, Quarre 3179
(BR, WAG). Kipopo, Thoen 4646 (BR); Schmitz 5835 (BR). Lopoto, Schmitz 4855 (BR).
Kafubu, Quarre 182 (BR, WAG). Vallee de Kapiri, HomhU 1238 (BR). Tsbinsinka, Manika
(Biano) Plateau, HomhU 1275 (BR).

Zambia, northern phovinc^e: Mbala (Abercorn) district, Liuni River Marsh, Richards

12278 (SRGH). Kawimbe, Richards 8329 (K), 19038 (K, MO). Kali, dambo, Richards
4976 (K). western province: Solwezi, Drummond 6 Rutherford-Smith 7031 (BR, K,
LISC, M, PRE, SRGH). Ndola, Fanshawe 969 (BR, EAH, K, SRGH). Kitwe, Fanshaive
10142 (SRGH); Mutimushi 267 (SRGH), 25-^6' (K). Mutufa River, Lawton 237 (K). Shi-
buchinga, Fanshawe 8388 (K, SRGH). Mwinilnnj^^a district, Marks 42 (K). Kabompo road,
25 km from Mwinilunga, Edwards 676 (LISC, PRE, SRGH).

17. Moraea textilis Baker, Trans. Linn. Soc, London Bot., Ser. 2, 1: 270. 1878.

type: Angola, Iluila, Lopollo River, flowering in April, Welwitsch 1549
(BM, holotype; COI, K, LISU, isotypes).— Fig. 12C.

M. mechowii Pax, Bot. Jahrb. Syst. 15: L5L 1893. typk: Angola, Gatala Canginga, flower-
ing in Feb., Teuscz in Exped. Mechow 557c ( B, holotype; BM, isotype).

Plants large, 40-80 cm high, unbranchecL Corni ca. 2 cm in diameter; tunics
of coarse, grey, wiry fibers. Prophylis rigid, j^ale, streaked with longitudinal
dark veins and irregularly broken. Leaf linear, canaliculate, exceeding the in-

282

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

»-n» l|

Fic;uHK 13. Distribution of Moraca vciitricosa and M. tcxtilis.

florescence, 6-8 mm wide. Stem bearing (3-) 5-7, often dry and usually over-
lapping, bract leaves, Spatlics herbaceous, becoming dry from above, acute, of-
ten becoming reddish; iimer spathe 8-15 cm long, the outer ca. % the inner.
Flowers blue purple or yellow, usually conspicuously streaked with purple;
outer tepah 4.7-6.3 cm long, the claw often slightly exceeding the limb; inner
tepah erect, 4.8-6.5 cm long. Filaments 1.5-2.2 cm long, free in the upper quar-

1 977] GOLDBLATT— A/ORAE A

283

tcr; anthers 1.0-1.35 cm long. Ovary ca. 2 cm long; style branches 1.5-2.5 cm
long, the crests to 1.5 cm long. Capsule ovoid, ca. 3 cm long; seeds flattened
and triangular to discoid. Chromosome number not known.

Flowering time: (January) April to June (July to August).

Distribution: Throughout the Angolan plateau above 1,200 m in grassland
or rocky sites, occasionally reported from damper situations. — Fig. 13.

A much more restricted concept of M. textilis is proposed here compared to
Gecrinck's (1970) treatment of this species which included M. ventricosa and
M, verdickii. As treated here, M. textilis occurs almost exclusively in Angola,
from the highlands in the southwest to the border of Zaire and Zambia, one col-
lection being known from western Zambia. Collections suggest that the species
is common in the provinces of Huila and Huambo, and relatively rare elsewhere,
though this may simply be a reflection of collecting activity being concentrated
in population centers. Moraea textilis is reported from grassland and rocky sites
at elevations above 1,200 m, though occasionally also from damp places. It has
a remarkably long flowering period from December to July, though most col-
lections are from the months of May and June, in the dry season,

Moraea textilis is most closely related to M. ventricosa^ a somewhat smaller
species occurring mainly in moist situations to the east, and it is also allied to M.
verdickii and M. macrantha as discussed under M. verdickii. The latter in-
variably has yellow flowers which are larger than those of A/, textilis, and it
blooms in the wet season, from November to February and March.

Angola, benguella: Higlilands brtweoii Cauda and Caconda, Hundt 822 (BM, Z).
Bailono district, Wellman 1906 (K). Caconda, Gosswcilcr 4206 (COI). cuaxzo sul: Ani-
boini, Gosswcilcr 996K (COI, K). huamho: Huambo (Nova Lisboa), da Silva 3604 (LISC).
Between Huambo and Quipeio, Exell 6- Mcndon(;a 1874 (BM). Near Caululo, Moreno 163
(COI, LISC). Cuima, Exell ir Mendoui'a 1937 (BM). Country of the Gangnellas and Am-
buellas, Gosstveiler 4136 (K). Sacaala forest, Mttrta 170 (COI). Mt. Moco, plateau above
village, Huntletj 3430 (PRE), huila: Ibxiue, Ucnriqucs 1020 (BM, COI, K, LISC, LISU,
SRGH). Leba, Pritchard 339 (BM, K, LISC). Near Lopollo River, Welwitseh 1549 (BM,
COI, K, LISU). Turdevala NW of Sa de Bandeira, Kers 3294 ( K, LISC, PRE, S); Barbosa
ir Moreno 10218 (COI). Lubango, Buraco do Bimbe, Mendes 3762 (LISC). Tchivinguiro,
Henriques 59 (LISC). lunda: R. Coxi, Kxell i^ Mendon^'a 1358 (BM). Xassengue Caiango,
near River Cuando, Gossweiler 11855 (COI). mdxkjO: Texeira de Sousa, GossweiJcr 12248,

12249 (BM, LISC).

Zambia, western province: Plains, Mwlnilunga district, Marks 42 (K).

18. Moraea tanzanica Goldbl., sp. nov. ^fype: Tanzania, Southern Highlands,
Kyimbila district, Stolz 2142 (K, holotype; BM, BR, Gil, PRE, Z, isotypes).
Fig. 14 a.

Flanta 20-35 cui alta. Tunicae coruii pallidas. Folium solitarium, canaliculatuni, 15-25
cm longiun raro spathas excedcntiuni, inscrtum hrcviter supra terrain. Caidis simplex, ferens
braetcis duabus. SpatJiae berbaceae, interior 9-10 cm longa, exterior 6—8 cm longa. Flores
luteis; tepaJa exteriora 4.5-5.5 cm longa, limbis 1-3.5 cm longis, expansis; hiteriora 3.5-4.5
cm longa. Filamenta 1.2-1.5 cm longa; anthcrae ca. I cm longae. Germen 1.7-2.2 cm
longum; rami styli 1.7-2 cm longi; cristac ca. 1.3 cm longae.

Plants solitary, inedinm, 20-35 cm high. Corm ca. 1.5 cm in diameter; tunics
of pale medium fibers. Prophylls dry-mcml)ranous, pale, irregularly torn above.
Leaf solitary, 15-25 cm long, seldom reaching beyond the midline of the spathes,

284

AXNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

B

i:

f!

^iVii

(X0,5).

Figure 14. Moraca species. — A, M. tanzanica. — B. A/, clavata. — C. M. brcvifolia.

canaliculate, inserted shortly above ground level. Stem simple. Bract leaves 2,
usually overlapping. Spathes herbaceous, attenuate; inner spathe 9-10 cm long,
the outer ca. % the inner. Flowers yellow; outer tepals 4.5-5,5 cm long, the limb
3-3.5 cm long, spreading; inner tepals erect, 3.5-4.5 cm long. Filaments 1,2-1.5

1977] GOLDBLATT— A/OiUEA

285

cm long, free in the upper half; anthers ca. 1 cm long. Ovary 1.7-2.2 cm long;
style branches 1,7-2 cm long, the crests ca. 1.3 cm long. Capsule cylindrical, to
3 cm long; seeds flattened, discoid. Chromosome number 2n = 12.

Flowering time: January to March.

Distribution: Northern Malawi and southern Tanzania, at high altitudes,
2,000-3,000 m; in open mountain grassland. — Fig. 15.

Moraea tanzaniea is related to the A/, textilis complex, and particularly to the
yellow-flowered M. verdickii. It is, however, a much smaller species, though
with a very large flower, and can always be distinguished from M. verdickii by its
consistently short leaf which seldom reaches beyond the midline of the spathes.
Moraea tanzaniea appears to stand at the beginning of a series of reduced species
in which the leaf is generally short and inserted above ground level, sometimes
at the base of the inflorescence. In this group the bract leaves are often con-
gested and may be reduced in number or even lacking, as in M. iinifoliata, the
most specialized species in this alliance.

The species most closely related to M. tanzaniea among the reduced species
are M. brevifoUa, which often has a shorter leaf and solitary bract leaf, and M.
clavata, an early-flowering species with much smaller flowers. Moraea tanzaniea
is confined to southwestern Tanzania where it has been collected mostly at high
altitudes in short grassland. It flowers from January to March, particularly to-
wards the end of summer. In contrast, M. brevifolia, found to the west in Zambia,
blooms in December to January and occurs only in marshy situations. Moraea
clavata, flowering even earlier in late spring to mid summer, is also confined to
moist situations and occurs from central Zambia to southern Angola.

Malawi, northern region: Nyikii Plateau, Katuinbi-Juniper forest turnoff, Pawek 6660

(MO); Phillips 1071A (K, MO, SRGH).

Tanzania, soutiikhn highlands: Kyiinhila district, Stolz 2142 (BM, BR, Gil, K, PRE,
Z). Elton Plateau, Richards 14128 (EAH, K). Rungwe Mt., Richards 14314 (K). Rungwe
caldera, Pocs 6505 (DAR), Mbeya Peak, Kcrfcwt 1628 (EAH). Kipengere Mts., Njoinbe
district, Richards 7574, 7617 (K). Ndunibi Forest Reserve, Semsei 1664 (EAII, K).

19. Moraea brevifolia Goldbl, sp. nov. tyi>e: Zambia, Northern Region, Lu-
mangwe Falls, Mporokoso district, Simon 6- Williamson 1483 (K, holotype;

LISC, SRGH, isot> pes ).—Fic. 14C.

Plantae 25-50 cm alta. Tunicae cormi pallidae, teniies. Folium solitariiim, breve, 5-20
cm longum, insertuni in pars inferior canlis. Caulis simplex, ferens bractea una, raro dua.
Spathae herbaceae, interior 7.5-12 cm longa, exterior 6-9 cm longa. Flores luteus; fepala ex-
teriora 5-7 cm longa, limbis 3-4 em longis, expansis; interiora 4-5 cm longa. Filamcnta 10-12
mm longa; antherae 9-16 mm longae. Gennen 1-1.3 cm longum; rami styli 1.2-2.5 cm
longi; cristae ad 1.5 cm longas.

Plants 25-50 cm high, solitary. Corm 1-1.5 cm in diameter; tnnics of fine
brown reticulate fibers, Prophylls dry-membranous, becoming lacerated above.
Leaf solitary, short, 5-20 cm long, with a free apex or entirely sheathing and
bractlike, inserted in the lower half of the stem, above ground level, canaliculate.
Stem simple. Braet leaf 1 (occasionally 2), 6-9 cm long. Spathes herbaceous,
with brown attenuate apices; inner spathe 7.5-12 cm long, the outer ca. yii-% the
innpr F/oi/;^r.v vellow: outer tevals 5-7 cm lontl. the limb 3-4 cm lonii, spread-

286

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Fk;uuk 15. Distribution of Monica ianzanica, M, brevifolia, and A/, clatata.

ing; inner tepals erect, 4-5 cm long. Filaments 10-12 mm long, free in the up-
per third; antJiers 9-16 mm long. Ovary 1-1.3 cm long; style branches 1.2-2.5
cm long, the crests to 1.5 cm long. Capsule and seeds unknown. Chromosome
number not known.

Flowering time: December to January.

Distribution: Zambia, central to nortlieastern Zambia; in wet places. — Fic. 15.

1977]

GOLDBLATT— A/ORAEA

287

Moraea hrevifolia is apparently rare and occurs only in marshy situations in
northern and western Zambia where it blooms in midsummer. It is somewhat
variable in leaf morphology, some plants having a very short, almost entirely
sheathing leaf, while in others the leaf is longer with the free portion reaching
the spatlies. The leaf is, however, distinctive in its point of insertion, well above
ground, at about the midline of the stem. Moraea hrevifolia is closely related
to the predominantly Angolan M. clavata, a generally smaller species, which con-
sistently has smaller flowers. Moraea hrevifolia may also be confused with M.
tanzaniea, which has similar flowers, but is vegetatively more robust, has a larger
broader leaf, and always has two bract leaves where one is the rule in M, lyrevi-
foJia. These two species also differ in habitat and time of flowering, M. tanzanica
being found in montane grassland and flowering later in the summer tlian A/.
hrevifolia,

Zanhua. northern province: Liiniangwe Falls, Mporokoso district, Simon ir William-
son T4<S3 (K, LISC, SRGII). NY'lu'lcngi-Luapuhi River road, Kawanibwa district, Richards
15476 (K). Kambole-Mbala road, Richards 11901 (K). Cliinakila-Loye flats, Mbala dis-
trict, Richards 19473 (K). western recjion: Mwinilnnga district, Milnc-Rcdhcad 3930
(BM, IJR, LISC). Plauis, Mwinilnnga district Marks 41 (K).

20. Moraea clavata Foster, Contr. Grav Herb. 114: 49. 1936. type: as for M.

gracilis Baker. — Fig. 14B.

M. gracilis Baker, Trans. Linn. Soc. London, Rot., Scr. 2, 1: 271. 1878, hoin. illcg. non M,
gracilis ( Licht. ex Rocni. 6i Schnlt.) Dicls, 1833. type: Angola, Hnila, near Lopollo
River, Wchcitsch 1545 (BM, lectot\pc; K, LISU, isolectotypes).

Plants small, often growing in eluinps, 15-30 (-35) cm high. Conn 1-1.5 cm
in diameter; tunics of fine, grey brown reticulate fibers. Prophylls short, mem-
1)ranous to herbaceous, usually entire at the apices. Leaf solitary, inserted mid-
way up the stem, usually 5-12 (-20) cm long, rarely exceeding the inflorescence.
Stem simple with an extended lower internode. Bract leaf 1, herbaceous, 3-4
cm long, above the leaf. Spathes herbaceous with dry upper margins; inner
spathe 4..5-6(-7) cm long, the outer ca. % the inner. Flowers yellow; outer tepals
2-3(-3.5) cm long, the limb 1-2 cm long, spreading; inner tepals 1.3-2.0 cm
long, (probably) erect. Filaments 5-6 mm long, united in the lower half; anthers
4-5 mm long. Ovary 5-8 mm long; style branches ca. 1 cm long, the crests to
1.0 cm long. Capsule ovoid-globose, 7-10 mm long; seeds small, angular to some-
what flattened. Chromosome nmnber not known.

Flowering time : October to January ( March ) .

Distribution: Angola and Zambia, in moist situations. — Fic. 15.

Moraea clavata was among the first species of Moraea known from tropical
Africa, having been collected in the 1850s by Welwitsch during his exploration
of Angola. It was first named A/, gracilis by Baker (in 1878) but this, a homonym,
was replaced with M. clavata by Foster in 1936. Only a few collections are
known, mainly from the southwestern highlands of Angola, but there are also
records from western and northern Zambia, in the Mwinilnnga and Serenje dis-
tricts, the former some 800 km and the latter 1,500 km to the east of the Angola
populations. It is a moisture loving species, found only in marshy places, dambos,

233 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

etc. and has a very long flowering period, from late spring in October to January,

witli one record in March.

It is a distinctive species, one of the smallest in subgenus Craruliflora, with
a ratlier short leaf inserted well above ground level, a single bract, and a very
small flower with an ovary 6-8 mm long. It is clearly related to M . brevifolia, a
taller and larger-flowered species which grows in a similar habitat in western
and northern Zambia. The difference in size between these two alone makes
confusion unlikely.

Angola, iiuila: Serra de Chela, Humpata, Gossweiler 13315 (LISC). Humpata Plateau
near Nhiiuc, Santos 76 (LISC). Huila, Antunes s,n, (LISU). Near the Lopollo River, Wel-
tvitsch 1545 (BM, COI, K, LISU). Without precise locality, Antunes 643 (LISC).

Zambia, centhal region; Serenje, Fanshawe 6731 (K, SRGH). Road hetween Mpika
and Kapiri-Mposhi, Serenje district, Richards 16869 (K), westerx i\egion: Mwinilunga
district, dambo NE of Dobc^a Bridge, Milne-Redhead 3620 (K).

21. Moraea upembana Goldbl., sp. nov. type: Zaire, Shaba, Pare Nationale

de rupeniba, banks of Lumana River, de Witte 2400 (BR, holotype). —
Fig. 16,

Planta parva, 10-20 cm alta, Tunicae cormi tenues, reticulatae. Folium ignotuni, absens
tempore florenti. Caulis simplex ferens 1-2 bracteis. Spatliae herbaceae, interior 3.5-7 cm
longa, exterior 3^ cm longa. Flores luteis; tepala exteriora 2.7-4.5 cm longa, limbis 1.5-2.5
cm longis, expansis; interiora 2-2.5 cm longa. Filamenta ad 7 mm longa; antherae ad 8 mm
longas. Germen 8-15 mm longum; rami styli ca. 10 mm longi; cristae ad 7 mm longas.

Plants small, 10-20 cm high, unbranched. Corm ca. 1 cm in diameter; tunics
of pale to dark, finely reticulate fibers extending upward in a neck. Prophylls
dry-membranous, entire or irregularly broken. Leaf unknown, absent at flower-
ing time. Stem bearing 1 or 2 herbaceous bract leaves. Spathes herbaceous,
with dry, pale brown acute apices; inner spathe 3.5-7 cm long, the outer 3-4 cm
long, ca. half the length of the inner. Flowers yellow; outer tepals 2.7-4.5 cm
long, the limb 1.5-2.5 cm long, spreading; iwier tepals probably erect, 2-2.5 cm
long. Filaments to 7 mm long; anthers to 8 mm long. Ovary 8-15 mm long;
style branches ca. 10 mm long, the crests to 7 mm long. Capsule and seeds un-
known. Chromosome numbers not known.

Flowering time: July.

Distribution: Zaire, Shaba, apparently very local. — Fig. 17.

Known only from the type collection, this rare plant was placed by Geerinck
(1970) in his treatment of Moraea in Zaire and Burundi in M. textiliSy according
to his very wide interpretation of this species. It has however a number of very
unusual features which make it appear to me quite distinct and perhaps not
even very closely allied to M. textilis. Firstly, M. upembana flowers in July, in
the dry season, is relatively short, standing up to 20 cm high, and is apparently
quite leafless at this time. In contrast, M. ventricosa and M. macrantha (also
included by Geerinck in Af . textilis) normally flower in March and April in Shaba
at the end of the wet season, and even those plants collected as late as May and
June are as tall and as large flowered as the collections from earlier in the season,
and do have a well-developed, green leaf. In other morphological features A/.
upembana also seems distinct, for it has very fine corm tunics, only 1 or 2 short

1977]

GOLDBLATT— A/ORAKA

289

'—^^*rj-ru^^yt^'^. ^^^ ^ ^^.«kAJP«JX*«iVV-'^

■JV^K-

-l\

^-^

-^-n .

1"*^

s

■*

■-^

Fk;uhk 16. T\pe coUeL-tioii of Mordca tipcmlycnuL

bract leaves, and is much smaller in all clnnacteristics than M. ventricosa or
M. macrantha^ both of whicli have coarser corm tunics and 3-5 large bract leaves.
Though the single collection known is inade(iuate for a complete understand-
ing of M. upemhana, it seems ])est associated with the group of morphologically

290

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

B

J L U-4l'^

FiGUHK 17. l)istril)uti()n of Monica upcnihana^ M, bovotici, A/, hahimlana^ and M, uni
joliatii witli niorpliology of — A, A/, hovonei.—^. M. bahnulana. — C. M. unifoliatn (xO.5)

reduced species of subgenus Grandijl
Angolan A/, chvata and its relative M

f

Further collec-

tions are needc^d to resolve the remaining (juestions about this species, including
its habitat, variation pattern, and even the details of its leaf and flower structure.

1977] GOLDBLATT — MORAEA

291

ZaThe. SHABA : Pare Nationale de I'Upfmba, Lumana River, de Witte 2400 (BR).

22. Moraea bovonei Chiov., Niiovo Giorii. Bot. Ital. 26: 72. 1919. type: Zaire,

Shaba, Ditiingula, Lake Mweru, Bovone 12 [F, lectotype; T (photograph
seen) isolectotype]. — Fig. 17A.

Plants slender, to 40 cm tall, with the leaf extending a further 40 cm. Corm
and tunics not known. ProphyJh papery, pale, becoming torn and irregular at
the apices. Leaf terete, ca. 40 cm long, inserted at the base of the inflorescence.
Stem with a long lower internode at tlie apex of which are 2 herbaceous bract
leaves, each ca. 3.5 cm long, and the inflorescence. Spathes herbaceous with
dry attenuate apices; inner spathe ca. 6 cm long, the outer sHghtly shorter.
Flower yellow; outer tepals to 2.6 cm long; inner tepals to 2.4 cm long. Filaments
ca. 8 mm long, united in the lower half; anthers 7 mm long. Ovanj 8-10 mm long;
style branches 1.2 cm long, the crests to 7 mm long. Capsule and seeds un-
known. Chromosome number unknown.

Flowering time : Summer ( Februaiy ) .

Distribution: Zaire, eastern Shaba, marshes of Lake Mweru. — Fig. 17.

Moraea hovonei, known from only a single gathering, is a most unusual spe-
cies. It has a very long slender stem, some 40 cm high, and the long leaf is in-
serted at the stem apex, immediately under the inflorescence. Unlike the re-
lated species, M. halundana and M. unifoliata, in which the leaf is also inserted
below the inflorescence, the leaf of M. hovonei is very long, approximately equal
in length to the stem. Moraea hovonei can be distinguished from its allies not
only by its long stem and leaf but also by the presence of two bract leaves at the
base of the spathes, whereas there is only one in M. halundana and none in A/.
unifoliata.

Zaihe. SHABA: Ditungiila, Lake Mweru, Bovoiic 12 ( F, T).

23. Moraea halundana Goldbl., sp. no v. type: Zaire, Shaba, 52 km SW of

Mutshatsha, Rohiason 6025 (K, holotype; M, SRGII, isotypes).— Fig. 17B.

Plaiita 15-40 cm alta. Tunicae cormi fiiscae, tcniics, reticiilatae. Folium solitariiiiii, in-
scrtuni ad base inflorescentiae, 6-11 cm longuiii, teres. Caiilis simplex, fcrens bractcam imum.
Spattia interior berbacea, 3-4.5 cm longa, exterior sicca, Iminnea, 1.5-2 cm longa. Flares
kitcis; tcpala cxteriora 3-5 cm longa, limbis 1.7-3 cm longis, expansis; tcpala intcriora 2-3 cm
longa. Filamcntu 8-10 mm longa; anthcrac 8-9 mm longae. Gennen ca. 8 mm longum; rami
styli 1.2-1.5 cm longi; cristae 5-10 nun longae.

Plants slender and unbranched, 15-40 cm high. Corm ca. 1.5 cm in diameter;
tunics of dark, finely reticulate fibers extending shortly upward in a neck.
ProphyUs membranous, but the upper one dry and becoming fibrous. Leaf 6-11
cm long, terete, inserted at the base of the inflorescence, and shortly exceeding
it. Stem slender with a very long lowermost internode. Braet leaf 1, short,
brown. Spathes unequal; inner spathe 3-4.5 cm long, herbaceous below, dry-
brown at the apex, the outer 1.5-2 cm long, entirely brown, less than half the
inner. Flower yellow; outer tepals 3-5 cm long, the Hmb 1.7-3 cm long, spread-
ing; inner tepals 2-3 cm long, evidently erect. Filaments 8-10 mm long, free

in the upper half; anthers 8-9 mm long. Ovary ca. 8 mm long; style branches

292

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

1.2-1.5 cni long, the crests 5-10 mm long. Capsule and seeds unknown. Chromo-
some number not known.

Flowering time: December.

Distribution: Zaire, southwestern Shaba (Katanga) in permanently wet

dambo. — Fio. 17.

Moraea haJumhina, known only from a single gathering in the southern part
of Zaire, close to the Angola and Zambia borders, is named after the Balunda
tribe who li\'e in this area. The .species is closely related to M. hovonei which
occurs to the east near Lake Mweru. Moraea haJundana differs from M. hovonei
mainly in its shorter leaf, single, dry, dark brown bract leaf, and unusually short,
brown outer spathe. Further collections are needed to assess the variation in the
species, and thus to better understand its relationship to the very similar M.
Inwonci and to M. iinifoliata, also closely related. However, in the absence of
more material, the unusual features of leaf, bracts, and .spathes necessitate speci-
fic recognition. The species was assigned to M. angusta (Thunb.) Ker by
Gcerinck (1972), but the resemblance to this South African plant from the Cape
region is only superficial. There seems no doubt that there is no close relation-
ship between the two, assigned by Goldblatt (1976a, 1976b) to different sub-
genera of Moraea.

Zaiiu.:. .sHAiiA, 52 km SW of Mutshatsha, near Zambian border, Rohinson 6025 (K, M,
SRGII).

24. Moraea unifoliata Foster, Contr. Gray Herb. 114: 48. 1936. tyi'e: as for

M. aphylla De Wild.— Fig. 17C.

M. ophylhi De Wild., Ann. Mus. Con^o, Ser. 4, Bot. 2: 21. 1913, honi. illejj. non M. apJnjUa
L.f., 1781. type: Zaire, Shaba (Katanga), Tenibwe, Hock s.n. ( BH, holotypc).

Plants 20-35 cm high, unbranched. Conn ca. 1.5 cm wide; tunics of pale
medium-fine reticulate fibers extending up in a neck. Prophylh papery, Hght
brown, often fibrous above. Produced leaf of flowering individuals 4-6.5 cm
long, seldom longer than the .spathes, inserted near the apex of the stem and
usually partly sheathing the inflorescence. Stem simple and lacking bract
leaves. Spathes 4-6 cm long, herbaceous, or becoming dry, attenuate; inner
spathe equal to or .shortly exceeding the outer. Flowers yellow; outer tepals
lanceolate, ca. 3.5 cm long, the claw to 1.5 cm long, spreading; inner tepals
2.5-3 cm long, erect. Filaments ca. 7.5 mm long, free in the upper part (half to
a quarter); anthers 7.5-9 mm long. Ovary 7-10 mm long; style branches to 1.5
cm long, the crests 5-9 mm long. Capsule ovoid-glabose, to 1 cm long, seeds
not known. Chromosome number not known.

Flowering time: Late November to January.

Distribution: Zaire, southern Shaba in moist situations along dambo mar-
gins or sponges. — Fic. 17.

Moraea unifoliata can immediately be recognized by its apparent lack of a
produced leaf. It does, however, have a leaf which is inserted at the base of the
inflorescence, but it is very short, and seldom extends more than 6 cm, thus not
exceeding the inflorescence .spathes. It is also distinctive in lacking bract leaves.

1977] GOLDBLATT— MOHAZ- A

293

a feature which separates it from its close allies, M. hovonei and M. hahindana
in both of which the leaf is also inserted at the base of the spathes. The slender
M. hovonei, however, has a very long leaf and two small bracts, while M, hahin-
dana has a shorter leaf and a single short, brown bract. These three species all
grow in similar sitnations, moist sponges or along the margins of marshes or
dambos, and bloom in summer. Moraea unifoUata is a fairly local species re-
corded from several sites, all in southern Shaba near Lumumbashi.

The singular habit of M. unifoUata^ with its very short produced leaf placed
immediately below the inflorescence and its complete lack of bracts, make the
species appear one of the most .specialized species of subgenus Grandijlora. It
appears as the end product of a series of reduced species which begins with A/.
tanzania and M. ]}revifoUa, and leads with progressive reduction in leaf size
and bract leaf number through M. chivata, AL halundana and M. hovonei.

Zaire, shaba (Katanga): MiikucMi, DctiUcux 175 (BR, WAG). Miikucii Kst., Srlmiitz
4250 (BR). Matiiitui River, 10 km SSE of Lnbuml)ashi ( Elisabethvillo), Schmitz 2371
(BR). Kapeluka, 16 km NE of Lul)mn1)aslii, Schmitz 4766 (BR, WAG). Kipopo, 25 km
NE of Liibumbaslii, Schmitz 8118 (BR, WAG). Marie-Josc, Quarw 1478 (BR, Gil).

IXCOMPLETELY UNDERSTOOD SpECIES

The following taxon, represented by two eollections, is problematie and
may represent an imdescribed speeies or a variant of a known one. Material is
inadequate at present to provide a satisfactory solution.

Moraea sp. 1

Two collections only, of a very tall, large yellow-flowered speeies were
made in the region of Lake Tanganyika. One, Rohyns 2192 was collected in
Zaire, between Pweto and Baudoinville, the otlier, Wallace 1302, was found in
Tanzania. The plants are about 1 m tall and have large flowers which suggests
that they may be forms of M. vcrdickii. They are, however, late flowering, in
May, and have between 4 and 7 large bracts whicli in the Wallace colkx-tion
are closely imbricate. Typically, M, vcrdickii blooms from December to Febru-
ary and rarely has more than three bracts, which do not overlap. The Robjns
collection resembles to some extent M. hella, though it is unusually robust for
this species and does not seem to fit in M. hclla as presently circumscribed.
Further material from this area is needed before a decision can be reached con-
cerning these specimens.

Excluded Species

1. Moraea revohita Wright

This species is excluded as the type material is tcratological. The specimen,
said to have been grown at Kew from corms collected in Angola, clearly belongs
to subgenus Grandijlora^ and the flowers resemble the pale form of A/, textilis.
Unique for the subgenus, however, are the several, linear, produced leaves. No
other species or specimen of subgenus Grandijlora has more than a single leaf
and the presence of more is almost certainly abnoniial

294 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

2. Several species of Moraea, described from Angola and Rhodesia, are now
recognized as belonging to the genus Ferraria. Some have been transferred to
Ferraria while others remain in Moraea, In spite of the multiplicity of names,
there is probably only one species of Ferraria in the whole area including Angola,
Zambia, Zaire, and Rhodesia. The most commonly accepted name for the species
is F. ^lutinosa (Baker) Rendle and the many synonyms originally placed in
Moraca are listed below. Ferraria glutinosa is distinguished from Moraea by
several features, the most readily observable being the equitant leaf, while the
branched axis with sticky exudate below the internode are features lacking in
the tropical African species of Moraea. It is further distinguished from Moraca
by its long-lived, tuberlike corms which apparently lack tunics.

Ferraria glutinosa (Baker) Rendle, Cat. Afr. PI Welw, 2: 27. 1899.

Moraca glutinosa Baker, Trans. Linn. Soc. London, Bot., Ser. 2, 1: 27L 1878. xYrE: Angola,
Iluila, near Lopollo River, Welwitsch 1543 (BM, holotype; LISU, isotype).

M. camlclahmm Baker, Trans. Linn. Soc. London, Bot., Ser. 2, 1: 27L 1878. type: Angola,
Hnila, Morro de Lox^ollo, Welwitsch 1544 (BM, liolotype; LISU, isotype).

M. andougciisis Baker, Trans. Linn. Soc. London, Bot., Ser. 2, 1: 271. 1878. type; Angola,
Pnngo Andongo, Welwitsch 1532 (BM, holotype; LISU, isotype).

M. sj)itliamaea Baker, Trans. Linn. Soc. London, Bot., Ser. 2, 1: 271. 1878. type: Angola,
Hnila, plains aronnd Ilumpata and Lopollo, Welwitsch 1547 (BM, liolotype; LISU, iso-
t>'pe ) .

M. aurautiaca Baker, Fl. Trop. Afr. 7 (addendnm); 575. 1898. type; Angola, Malange,
M echo w 303 (B, holotype).

M, kitamhcnsis Baker, FI. Trop. Afr. 7 (addendnm): 575. 1898. tyi^e: Angola, Bangala,
swamps at Knango, Buchncr 679 ( B, holotype).

M, randii Rendle, J. Bot. 36: 144. 1898. type; Rhodesia, Bnlawayo, Rand s.n, (BM, holo-
type ) .

Ferraria candrlahrum (Baker) Rendle, Cat. Afr. Pi. Wclw. 2: 27. 1899.

r. andongcnsi.s (Baker) Rendle, Cat. Afr, PI. Welw. 2: 27. 1899.

Moraca maJatigcnsis Baker, Bnll. Herb. Boissier, ser. 2, 1: 862. 1901. type: Angola,
Malange, Mcchow 386 (not seen).

(Other synonyms include Ferraria welwitschii Baker and F. hirschbcrgii Bolus.)

Literature Cited

Bakeii, J. G. 1878. Report on the Liliaceae, Iridaeeae, Hypoxidaceae and IlacMuodoraceae
of Welwitsch's Angolan herbarium. Trans. Linn. Soe. London, Bot., Ser. 2, 1: 245-274.
. 1898. Irideae. In Thiselton-Dyer W. T. (editor), Flora of Tropieal Afriea. Vol.

7. Reeve & Co., London.
BuHTT, B. L. 1938. Moraea carsonii. Bot. Mag. 161: tab. 9544.

CiiiMPHAMiiA, B. B. 1974. Karyotype analysis in Moraea and Dietcs. Cytologia 39: 525-

529.
Geerixck, D. 1970. Etude du genre Moraea Miller (Iridaeeae) an Congo et an Burundi.

Bull. Soe. Roy. Bot. Belgiqne 103: 143-156.
. 1972. Trois especes dTridaeees meeonnues au Zaire. Bull. Soe. Roy. Bot. Belgiqne

105: 5-8.

GoLDDLATT, P. 1971. Cytological and morphological studies in the southern African Iri-
daeeae. J. S. African Bot. 37: 317-460.

. 1973. * Contributions to the knowledge of Moraca (Iridaeeae) in the summer rain-

fall region of South Africa. Ann. Missouri Bot. Card. 60: 204-259.

— . 1976a. Evolution, cytology and subgeneric classification in Moraca ( Iridaeeae ) .

Ann. Missouri Bot. Card. 63; 1-23.

— . 1976b. The genus Moraea in the winter rainfall region of southern Africa. Ann.

Missouri Bot. Card. 63: 657-786.

1977] GOLDBLATT— A/OKAEA

295

Lkwis, W. H. 1966. ChroiuosoiiH's of two Monica ( IridLiceae) from soiitliern Africa. Sida
1: 381-382.

Ravex, p. 11. 1973. E\olution of Mediterranean floras. Pp. 213-224, in F. di Castri & H.

A. Mooney (editors), Mediterranean Type Ecosystems — Origin and Structure. Springer-
Verlag, Xew York.

A NEW CLASSIFICATION OF FICUS

William Ramirez BJ

AllSTHACT

The ta\;i of Ficus arv classified on tlie basis of the specificity and morpliology of tl^ieir
symbiotic wasps (Agaonidae), systems of pollination, and morphology and physiology of the
figs. The new classification is a modification of Corner's system with the following changes.
In su]')geniis Fictts, subsection Eriosijcea is elevated to sectional rank. Series Rivularcs and
PsciidopahiHic do not belong to the group of BJastophaga-poWiudicd figs and are transferred
to the new Cc*ro^(MY>7<^n-poIlinated complex of subgenus Syconionis, Two subsections, Scahrae
and Varinga, are recognized in section Sijcidium, and series Phacopilosac and subsection
Faleomorphc are recognized as sections. The subgenus Sycomorus is much expanded to in-
clude eight sections: Adcnos'pcnna, Ncomorphc, Frostratae, Fungcnics, Fscudopalmae, Rivu-
larcs, Sycocarpus, ami Sycomorus.

The object of this study is to group the taxa of Ficus into related groups
considering the specificity and morphology of their symbiotic agaonids, the dif-
ferent systems of pollination, as well as the moiphology and physiology of the
figs.

The last systematic arrangement of Ficus was made by Comer (1965) and
is summarized in Table 1. A parallel list of the pollinating agaonids (genera or
subgenera reported up to now for each fig taxon) is also included. The list of
agaonids was taken from Hill (1967) and modified by me. Parallel to the
groups of wasps there are columns showing the absence or presence of corbiculae
in the wasps (Ramirez, 1974).

The New Classification of FiCUS and its Pollinatoi^s

The proposed classification of Ficus is found in Table 3. Modifications are
extended only to the level of series.

SUBCEXUS LROSTIGMA

This group of figs remains as treated by Corner (1965) (Tabic 1).

Section VrostigmcL — The figs are inhabited by Bhstophaga (group E), which

are characterized by the presence of coxal and sternal corbiculae (as in Figs. 3
and 4 ) .

Section Leucogytie, — This section comprises two species. One of them (F.
tsieJa) is pollinated by ManieJJa delhiemls, with coxal and sternal corbiculae
( as in Figs. 3 and 4) .

Section Conosycea, — The species of this section are pollinated by several
groups of wasps. The only Blastophaga (B. amottiana and B. errata) known from
this group of figs have sternal corbiculae and coxal comics. Ceratosolen megarho-
palus (the Megarhopalus group) and the majority of Waterstoniella wasps arc^
characterized by only veiy rudimentary sternal corbiculae (Figs. 5-6); some

^ Facultad de Agrononiia, Universidad do Costa Rica, Costa Hica, America Central
AxN. Mkssouiu But. Caho. 64: 296-310. 1977.

1977] RAMI HI-:/— CLASSIFICATION OF FICUS £97

Watcrstoniella (e.g., W. snndaica and W. jacohsoni) do not have corbiculae
(Fig. 2). The other two groups of agaonids {Eupristitui and Parapristina)
found in section Commjcea have sternal and coxal corbiculae (as in Figs. 3 and

4).

Section StiJpnopJiylhim. — This section contains only Ficus elastica which is
polHnated by Blastophaga clavigera {Blast ophaga group B) a wasp with sternal
and presumably coxal corbiculae (Wiebcs, personal communication) .

Section Malvanthera. — This group is unique in that its anthers have two pol-
len sacs which dehisce with one crescentic or equatorial slit. The section is pol-
linated by Pleistodontes wasps. However, there are apparently several Pleis-

todontes groups pollinating the different groups of Malvanthera figs (personal
observation).

Pleistodontes imperialis is characterized by sternal and possibly coxal cor-
biculae while the other known Pleistodontes do not possess corbiculae at all.
Series Malvanthereae is pollinated by w^asps without corbiculae (P. hlandus, frog-
gatti, rieki, pJehejiis, and regalis). The only Pleistodontes (P. inmaturus) known

from series Cijclanthereae apparently doc\s not possess corbiculae. For more in-
formation on Pleistodontes wasps see Wiebes (1963b: 319, Table 1). It is
probable that the group Pleistodontes as well as its Ficus hosts, will have to be
reclassified when more is known of both groups.

Section Galoghjchia. — This group of figs resembles the last section in the in-
flexed, not interlocking, apical and internal bracts of the ostiolum (Corner 1959:
376), but it has normal anthers with four pollen sacs. It is pollinated by two
main groups of wasps: (a) those with only sternal corbiculae (Agaon, AUotrio-
zoon and Paragaon) and (b) those with sternal and coxal corbiculae {Alfon-
siella and Elisahethiella) .

Section Americana. — According to Corner (1959: 376) this section is
closely related to both sections Urostigma and Conosycca. It is pollinated by
Blastophaga wasps of the subgenus Pegoscapus (Ramirez, 1970) with coxal and
sternal corbiculae. However, P. carlosi and P. inariae (the pollinators of F,
tuerckheimii in Costa Rica, Mexico and Panama) do not possess coxal corbiculae
(Ramirez, 1970).

SUHCENUS PIIARMACOSYCEA

Corner (1959: 407) considered that the Old World section Oreosycea has the
same essential characters, and indeed, is with difficulty distinguished from New
World Pharniacosycea species. However, in the descriptions of the two sec-
tions we find very important differences, some of which are pointed out in Table
2.

The Old \\^orld species have in the past been referred to the subgenus Uro-
stigma where they are out of place, particularly in being independent trees and
not banyans or stranglers. The species from New Caledonia have never been
properly classified and they are the closest in several respects to the American
species. Corner (1959: 407) stated that he divides the subgenus Pharmacosycea
into two sections, maintaining the geographical distinction for convenience, but
that redefinition will be necessary when the American species are better known.

298

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

OJ

o

U

ri

o

05
C

a5J

H

CD

o

W5

CT3
0) o

vi

^

0)

be

o

bo

g ==

C

bt

bo

bO

o
S

o S

O

CO

O

o

03

o

CJ

o

c

QJ

u

O

c

ID
en

a

e

o
U

s

o

3

IS

o
U

a

o

o-.

v^

c

CD

bO

(A

o

o

cr;

C2
O

o

C/5

c

bc

++++++ ++ +

+ +

CL.

+ + +

+++++++++ ++ +

+ + +

+

++++++

+

+

+

+ + + +

cs Q a O

bjD b£ t^ ly)
a Q a a

be

Co

^
a

CQ 2Q CQ ^. CQ

3
O

bo

S/5

^ ^ r^

c^ 2 2

rt V* w

^ w 2

« .«

^

■>

!^

2

^

O O

^ 3 "3 Jr

c:>

b

bC

t3

Q <ii c^

^ ^ 5

S ^ S

UJ CQ ^ ^

=3

'5

^
Ck5

15

3

05

bC "tS

2 I

Gh CQ S;

>«

^

Co >( ^

;j ^ ^

^
o

o o

cr 5? i^

-2

<: Gh <: U4 ^

CJ
C

b£

-^ «

Q ^ N O

si

3

^ -^ O

!3 a. -^j

"3 Q ^

a.

g

QJ

Z 3i

§

^

^

^

^ z.

^1rr

■3

U Q c/3 a, CQ O S

:3 «

o

bjD

^

"3
-a

<

1977]

RAMfKEZ— CLASSIFICATIOX OF FICUS

299

o

pq
<

o

oo

c

<

C

C/5

o

CD

l=i
O

3

1

c

^

OJ

S;

bo

C

^ 1

G

C/5

+ + + +

+ + + +

+

+

+ + +

a Q «
^ ^ ^ to

2 C3 !3 a

— ^-* ^ Go

c' c '^

Co

c

^

Q =3 a

t^JD CiC GJD

^ Q G

.ii fe

?:.

;u

Q CQ CQ CQ ^

Co Co Co

v! ^ ^

S S 5

c

CD

2-

•s

Co

CO

o
o

CO

o

■^ ::r O y

^

^

>^

~ Co

CC Cis c/: U tij pg

^
^

o

CO

o

CO

Oh ^

«

^
^

io
C

o,

&3

+

«
tx

^

0>i

^

^

^

a
=1^

^>n

^

•^^ ^

< -5 -H

C J- ;^

CQ ^ h.

^ ^ ^ o

^
^

tg

-t:

^
^

o

u^

+

"3

>-o

CQ

^ Q a C
Oh f^ On =Q

C

Q

c o

CO

^

^

CO

s i 5

\3 ri ^

t£l

Co

5

^
^

fts

5
o::

o ^

.:2 -^2

Co

CC

300

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

IS

u
O

U

c

c
o

u
J
n
^

H

o

c^3

E

a;

Ln

c

<

I

'C

t/1

c

c

C/5

o

++++++

++++ +++++++

+

+

+

2

o

e*

^ « :3 a

S tX ^ t;C

6

a <3 ?3

o o o

53 ?i t? en

CQ

O U CQ CQ CQ

s s s

o fc S:

3 3

a. a.

I- C i-

CQ hj hj ►J kJ ►J

£
^

C K C C
:^ ^ 5^ ^

<w ^ "-^ "-^

■^ Crj Co Co

COCO

I

c

Co
O

o

^

^
O

a o a

U K w

^ <:j ^

§ §

c

<:i)
a

CW

O

o

p^ s^

c

a.

c

5.

Co

3

« "■ ^

^ O "iJ c

l§

tic

.2

c/^

^

a.

K

tie

o

,c

o

hS

a

a

>

ft^

6o

"S

C

V C ^

:5i^ ."S

U&->

a

5

.Co

1..

s?

^

a.

^» ■

.jo

^

o

3^

s

05

5

u.

a.

3

a

<

Q

ft^

3i
C

a

«

3

"a

o

£

^

^

^

^

1

'^

(a

^

^

Co

3

O

C/3

3

1977]

RAMIREZ— CLASSIFICATION OF FICUS

301

n3

P

c
o

U

w

pq
<

o
U

u

C

<

'5

o

I/}

C/1

V5

en

c

3

o

+ + + + +

+

CO CO

Q C

CO CO

c

^

CO

o

o

o

a O a =3 Q

K K V. k 1^

'^ ^ ^ ^ ^

u o o u u

CO ^

"3

^

CO

2^

^ Co

H 3

\1

=3

OsUua:^t£HU^

tii

a

3 3
O 3

3

CO

'o

&0

3

o

3i

Co
3

£

o

CD

CO

5

5

;u

c/:

lac

a;
bO

a

U

S3

O

3
,£3

I

o

I

13

CQ

I

c

<

^

302

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

1

3

2

4

5

7

6

8

FiciMiKs 1-8. — 1. Mcsosternuin without corbiculae of Blasfopliaga psencs, the pollinator
of tlu' edible fig. — 2. Mesostermiin of Blastopha^a (Watcr^lotiicUa) sundaica, a wasp without
corbiculae but with abundant bristles which are probably used to carry pollen. — 3. Front leg
of ManicUa dclliicnsis with coxal corbieula. — 4. Riglit side of inesostcrnum of Blastophaga
{Pegoscapus) cuininioisls showing corbieula and some pollen in place. — 5. Mesosternuni of
Blast o])]ia^a {Watcision'ieUa) sundaica widi incipient corbiculae. — 6. Mesosternuni of Cera-
tosolcn mc^arliopalus (tlie Megarhopalus group), right corbieula with some pollen. — 7. Meso-
sternum of Blastophaga javana {Blastophaga group B) with dexeloped open corbiculae. — 8.
Mcsosternum of Liporvhopahim m'mdanacnsis with closed corbiculae as in Ccratosolcn.

1977J

RAMIREZ CLASSIFICAIIOX OF FICUS

303

Table 2.
riiarmaccmjcea.

Presence

or absence of some characters in the two sections

of sul^genns

Section

I'igs
Single

Ostioluni Ovary with Pachycaxilous Anthers
Crateriforni a Red Mark Trees Nunierons

Pollen

Exposed

at Male Pliase

Pharniacosycca

+

+ + - +

+

Orcosycca

—

- - + -

—

Corner (1967: 40) noted that the new look bronglit into tlie subgenus
Pliarnuicosycea by the plants from New Caledonia is the brown liairiness, some-
times almost furriness, of twig, leaf, and fig, oonpled with the rosettes of large
leaves, the many-veined obovate lamina with cordate base and short petiole^ and
the large size of the fig. All of these characters are more or less primitive and
pachycaul signs in Fictis, Section PJiannacosycea in the New World does not
present all the pachycaulous characters mentioned by Corner (1967) for some
Old World Oreosijcea.

In order to explain the presence of pharmacosyceous figs in both the Old and
New World, Corner (1967: 41) postulated that there must have been a land
connection with tropical Africa such as is suggested by the great extension of the
4,000 mile line to the west of Peru. In 1967 he further stated that this con-
nection is demanded by other moraceous genera such as Antiaris, Antiaropsis,
SparatUisyce and Trophis, as well as by the monocotyledons Dianella., Ileliconia
and Spatliiphyllum in very diverse families.

Two hypotheses to explain the presence of PJiannacosycea in the Old and
New Worlds are: (a) Sections Phannacosycea and Orcosycca do not belong
to the same subgeneric taxon and their species are more or less similar because
of convergence. If this is true, each should be elevated to the subgeneric level,
forming biological units separated geographically and by their respective pol-
linators, New \\'orld Phannacosycea being the host of Tctrapus wasps (without
corbiculae) and Old World Orcosycca of Blastoplni^a {Blastophaga group F)

and of DoUchoris vasciilosae (both with coxal and sternal corbiculae). (b) Sec-
tions Phannacosycea and Orcosycca belong to the same subgeneric category,
but section Pharniacosycca migrated to the New World before the agaonids
evolved corbiculae. This line of thought would agree with the ideas of Corner
(1967: 53), although not demonstrating the particular land connection that he
postulated.

SUI5C;EXUS FlCUS

In the new classification the subsections Ficus and Eriosycca are elevated
to sectional rank as suggested ])y Corner (1959: 417). The series Rividarcs and
PscudopaJnicae are not considered to belong to the group of Blastophaga-]}o\-
linated figs and are transferred to the new Ceratoso/cn-pollinated complex (the
subgenus Sycomonis, Table 3). Corner (1969b: 326) stated that F. pseudopalma
and F. rivularis (two Philippine species) differ from the rest of section Ficus
and from each other markedly enough to require separate taxononn'c series (Table

304

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

Tabll: 3. Proposed classification of the genus Ficus considering the specificity and
morphology of its symbiotic agaonids, tlie different systems of pollination, as well as the
morphology and physiology of the figs; with a list of the agaonid pollinators (modified from
Hill, 1967) of each group, and the i)resenee or absence of eorbiculae.

Subgenus

Section

Urostii^t)ui Uwsiifj^nui

Lcucogyne
Conosijcca

Siilpno])liijUnm

Mahanthem

Galoclychia

Americana

Subsection

Conostjcca

Dictijoncuron

Bcnjamina

Agaonidae

Blastophaga
Group E

ManicUa

BUistophaga
Mcgarhopalus

Group
Eupristina
WatersfonicUa
WatcrstonicIIa

Walcrstoiiiella

Eupristina
Paraprisiina

Blastophaga
chivigcra
{—Blastophaga
Group G)

Plcistodontcs
Flcistoclontcs

Agaon
Alfonsiclla

Allotriozoon

FAisahcthiclla

Faragaon

Pcgoscapus

Corbiculae

Absent Sternal Coxal

+

+

+

+

+

+
+

+

+

+
+

+

+

+
+
+
+
+

+

+

+
+ ?

+

+
+

+ ?

+ ?

+

+

+

Fliarma-
cosijcca

Ficus

Orcosycca

P]iarniac(hsycea
Ficus

iViizocladus

Kalosyce

Sinosycidiuni''
Eriosycca

Sycidiuni

Fhacopilosae

Palcomorphe

Scahrac

Varini:a

Palcomorphe
Copiosae

Blastophaga
Group F

Dolichoris
Tetrapus

Blastophaga
Group A

Blastophaga
Group A

Blastophaga
Group A

Blastophaga
Group H

Blastophaga
Group B

Blastophaga
Group B

Blastophaga
Group C

Liporrhopalum

Blastophaga
Group D

+
+

+

+

+

+

+

+

+

+

+
+

+

-f

1977]

RAMIREZ— CLASSIFICATION OF FICUS

305

Table 3. (Continued)

Subgenus

Section

Neomorphe

Prostratae

Pugentes

Pseudopalmeae

Rivulares^

Sycomortis
Sycocarpus

Subsection

Sycomorus Adenosperma

Agaonidao

Ceratosolen
Ceratosolen
Ceratosolen
Ceratosolen
Ceratosolen

Ceratosolen
Ceratosolen

Corbiculae

Absent Sternal Co.xal

+
+
+
+
+
+

+
+

« Probably pollinated by a wasp of Blastophaga^ Group A.
** Probably pollinated by a Ceratosolen wasp.

1). Wiebes (1963a: 101, 104) indicated that the pollinator of F. pseudopalma
(C. bakeri) has aberrant characters for the genus Ceratosolen, but appears re-
lated to the C. ahnormis and C. armipes groups (pollinators of figs of section
Sycocarpus ) .

Sections Kalosyce and Rhizocladus, — These two sections are left in the taxo-
nomic position given them by Corner (1965). They form two well-defined groups
polhnated by Blastophaga (Blastophaga group A) wasps without corbiculae
(Fig. 1). The pollinators of these two groups of figs are quite similar to the
ones found with section Ficiis (Table 3). These two sections are associated by
their pollinators. Corner (1960: 3), however, suggested that sections Kalosyce
and Rhizocladus might be considered to form a fifth subgenus.

Section Sinosycidium. — This section is left in the same taxonomic position
given by Corner (1960: 24). It has a single species (F. tsiangii). Because of its
dispersed diandrous flowers and the slightly bifid stigmata of the female flowers,
I consider this section to be related to section Fictis (as in Table 3), although the
ramiflorous bracteate receptables are like those which occur in sections Sycidium,
Sycocarpus and Adenosperma according to Corner (1960: 24-25). The polHnator
of F. tsiangii is not known, but it could be a Blastophaga without corbiculae (as
in Fig. 1) as those of Blastophaga group A.

Section Sycidium, — In the new classification this group has two subsections,
Scahrae and Varinga. These groups are related by their pollinators of the
Blastophaga group B, which are characterized by their open sternal corbiculae

(Fig. 7).

aeop

Phaeopilo

b-

section Paleomorphe (both sensu Corner, 1965) are elevated to sectional rank.
Phaeopilosae is pollinated by Blastophaga group C with closed sternal corbic-
ulae (Fig, 9). Paleomorphe has two subsections, Paleomorphe being pollinated
by Liporrhopalum with closed sternal corbiculae (Fig. 8) and Copiosae (series
Copiosae, sensu Corner, 1965) by Blastophaga group D having closed sternal
corbiculae (as in Fig. 9).

306 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

SUBGENUS SYCOMORUS

In the new classification the subgenus Sycomonis is expanded and comprises
eiglit sections; Adenosperma^ Neomorphe^ Prostratae^ Pungentes, Pseudopal-
meae, Rivulares, Sycocarpus and Sycomorus (Table 3). Of these sections,
Adenosperma, Neomorphe and Sycocarpus were considered by Corner (1965)
as sections of the subgenus Ficus; Prostratae and Pun^entes as series of subsec-
tion Sycidium; Pseudopalnieae and Rivulares as series of subsection Ficus.

All the sections included here in Sycomorus, excepting Rivulares, are known
to be pollinated by Ceratosolen wasps. The pollinator of Ficus rivularis (the
only species of section Rivulares) is not known, but I suspect this species to be
pollinated by a Ceratosolen with a short ovipositor and closed sternal corbicuhxe.
All the dioecious sections (Adenosperma, Neomorphe, Prostratae, Pungentes,
Pseudopalmcae and Sycocarpus) are inhabited by Ceratosolen wasps with short
ovipositors. Nevertheless, Corner (1965: 85) inchided in section Sycocarpus
(subsection Papuasyce) the species F. microdictya (of New Guinea) which has
the perianth similar to that of Sycocarpus, but is monoecious Hke Sycomorus-,
which does not occur in New Guinea (Corner, 1958: 31, personal communication).
Section Sycomorus is a monoecious group pollinated by Ceratosolen with long ovi-
positors.

RELATioNsmps Among Groups of Figs Included in

Subgenus Sycomorus

SECTION ADEISOSPERMA

This section aligns with the unistaminate sections Sycidium and Sycocarpus,
which differ in the form of the seed if not in that of the flower (Corner, 1969b:
320). The section is related to section Sycocarpus, subsection Auriculispertim,
of the Solomon Islands, and connects with the origin of section Ficus through the
Philippine species F. pseudopalma and F. rivularis (Corner, 1969b: 319).

SECTION NEOMORPHE

Corner (1967: 51) stated that this section has much in common with tlie sub-
genus Sycomorus, Neomorphe may have come from the stock of Adenosperma
on the Melanesian Foreland, and this stock may have been connected with that
of Sycomorus, so that Neomorphe is an eastern parallel of it (Corner, 1967: 51).
Neomorphe must be divided into two series (Table 1), Variegatae and Auricula-
tae, which show alliance with the subgenus Sycomorus in the first case and sec-
tion Sycocarpus in the second. Series Variegatae can be divided, like-
wise, into two subseries. The first Corner ( 1965: 32-33 ) called subseries
Laciniatae, It has tepals characteristic of subgenus Sycomorus, but it is fur-
ther removed geographically from the African subgenus Sycomorus (Cor-
ner, 1967). The second, subseries Variegatae, has only two spe-
cies, F. variegata and F. viridicarpa. Ceratosolen striatus (=C. appendiculatus) ,
an agaonid collected from F. variegata in Java, was illustrated by Grandi (1917:

'^ Ficus priichardii, a inonot'cious fig, also belongs to Sycocaqms (Corner, 1970).

1977] RAMIREZ— CLASSIFICATION OF FICUS 3Qy

Fig. XII, 6) as a wasp with a long ovipositor like the wasps found in section
Sycomorus (as in Table 3) .

Neomorphe as well as subgenus Sycomorus of Corner (1965) are pollinated
by Ceratosolen wasps which are apparently related. Wiebes (1963a: 104) re-
ported that the species of the Ceratosolen appendiculatus group live in the re-
ceptables of section Neomorphe and subgenus Sycomorus (sensu Corner, 1965),
and one species is known from series Prostratae. The occurrence of a group of
such closely related species of Ceratosolen in the figs of both dioecious Neomorphe
and monoecious Sycomorus would suggest that the floral characters in w^hich
Neomorphe is close to Sycomorus are more important than the distribution of the
flowers in the receptacles. A parallel is found in F. microdictya^ which is a mon-
oecious species in the dioecious Sycocarpus^ (Wiebes, 1963a: 104).

SECTIONS PROSTRATAE AND PUNCENTES

These sections are also pollinated by Ceratosolen wasps. Corner (1965) con-
siders them to be two series of section Sycidium, According to Wiebes (1963a:
102), the greater part of the Indomalayan and Papuan species of Ceratosolen live
in the sections Neomorphe and Sycocarpus, but some are known from Prostratae
and Pungentes, two series of Sycidium (sensu Corner, 1965). These series have
usually been placed in section Sycocarpus and only recently have been assigned
to Sycidium (Wiebes, 1963a). Botanically these two series point to a common
ancestor which would combine Sycidium with Sycocarpus and Sycomorus, in-
cluding Neomorphe (Corner, 1958: 31). In the opinion of Wiebes (1963a: 102)
the wasps from the series Prostratae connect those from the section Neomorphe
with those of the subgenus Sycomorus, and the wasps from the series Pungentes
appear to be related to the wasps from the section Sycocarpus. According to
Corner (1959: 444), series Prostratae relates with section Ficus but habit and
convenience place it in Sycidium.

SECTIONS PSEVDOPALMEAE AND RIVULARES

Each of these taxa has a single species. Corner (1965) included them as two
series of the subgenus Ficus, Both species are found in the Philippines. Ficus
rivularis is an advanced leptocaul shrub with lanceolate leaves, distinguished
in section Ficus by the gamophyllous perianth with distinct tepal lobes, com-
pressed auriculiform seed, and the more or less gynobasic style in the female

flower. The perianth is intermediate between that of section Ficus and Syco-
carpus. In perianth, style and seed, F. rivularis agrees with Adenosperma; it ap-
pears as a relic, fitting no section of the ancestral line of section Ficus from which
those of Auriculisperma and Adenosperma diverged (Corner 1969b: 328).

Icrpalma

dammaropsis ( section

Sycocarpus, subsection Auriculisperma) of New Guinea, and thus, with section
Adenosperma. It connects also with the ancestry of the F. deltoidea complex
(section Ficus series Erythrogyneae) and has the three tepals of section Ficus

'^ Ficus pritchardii (a monoecious fig) also belongs to Sycocarfms.

308

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

10

Fu;uuEs 9-10. — 9. Mesosternuni of Blastopha^a jacohsi {Blastophaga group C) with closed
sternal c'orbiculiu\ — 10. Mesosternuni of Ccratosolen pilipis with closed corbiciilae.

(Corner, 1969b: 326). Ceratosolen hakeri is the pollinator of F. pseudopalma.
This wasp appears to be related to the C. ahnormis and the C. armipes groups.
Ficus pseudopahna was classified in section Ficus because of its bistaminate
male flowers, but it does show some relationship with F. dammaropsis (section
Sycocarpus), the host of C. ahnormis (Wiel)es, 1963a: 101),

SECTION SYCOCARPUS

This group of Ficus is mostly dioecious; however, F. microdictya and F.
pritchardii are monoecious. It is pollinated by Ceratosolen wasps with short
ovipositors, but the ovipositors of the pollinators of F. microdictya and F. pritchar-
dii are probably much longer than the abdomens. The chief character of the
section is the entirely gamophyllous perianth. In the male flower the perianth is
saccate and covers one, or less often, two stamens (Corner 1960: 38). For the
relationship of the pollinators of Sycocarpus with the pollinator of F. pseudo-
pahna and those of section Nemorphe, see under sections Pseudopahneae and
Neomorphe, See also under section Adenosperma.

SECTION SYCOMORUS

In the new classification, this group contains all the monoecious figs in-
cluded in the subgenus Sycomorus of Corner (1965). It is polHnated by Cerato-
solen wasps with long ovipositors.

Galil (1973) noted that in spite of numerous structural differences between
the syconia of the dioecious F. fistulosa (section Sycocarpus) and the monoe-
cious F. sycomorus (section Sycomorus sensu Ramirez, 1974) which belong to
different subgenera of Ficus, namely Ficus and Sycomorus (sensu Corner, 1965)
respectively, the two have several biological features in common. In both, the
pollinating wasps are species of Ceratosolen which behave very similarly in re-
lation to the figs, and such likeness in behavior indicates that physiological con-
ditions within the figs are probably also similar in both cases.

1977] RAMIREZ— CLASSIFICATION OF FICUS 309

Chahactehs of the Subc;enus Sycomorus

Corner (1967: 51) stated that Sycomorus, Sycocarpus, Adenosperma, Neo-
morphe, and two series of SycicUum {Prostratae and Fung^entcs) are distinguished
by having Ceratosolen as pollinating insects. Despite their differences, he sug-
gests it may be necessary to combine them in the subgenus Sycomorus in con-
trast with the remainder of the subgenus Ficus pollinated by Blastophaga.

The newly defined subgenus Sycomorus is characterized by the following
characters: Male flowers: (a) in 1 or 2 (in some cases 3) ostiolar rings; (b)
few per fig; (c) usually without pistillode; (d) perianth with free petals, gamo-
phyllous or utriculate; (e) mostly sessile; (f) usually with only one or two
stamens (few species with three). Anthers: (a) enfolded by the perianth; (b)
usually small; (c) pollen not exposed at male phase. Female flowers: (a)
stigma simple; (b) styles usually short excepting those of section Sycomorus
and of F. microdictya and pritchardii^. Syconia: (a) with internal bristles; (b)
helicoidal ostiolar entrance with several (more than three) interleafing super-
ficial bracts; (c) dioecious (excepting section Sycomorus and F. microdictya and
pritchardii; (d) ostiolum usually does not open at male phase. Leaf: (a)
stomata usually superficial; (b) leaf not coriaceous; (c) plicate in bud. Trees:
independent, not epiphytic. Pollinators: Ceratosolen wasps which are character-
ized by closed sternal corbiculae (as in Fig. 10), and coxal combs, and which col-
lect the pollen from detached anthers cut by the males (Gahl, 1973); short
ovipositors (except the Ceratosolen wasps of section Sycomorus and F. micro-
dictya and pritchardii) and by the ability of the male to perforate the fig in
order to gnaw an exit that allows the females to escape. The males in all species
probably cut the stamens before the females emerge from the galls (Galil,

1973 ) .

The figs of sections Adenosperma, Sijcocarpus, and Sycomorus are para-
sitized by Eukoehelea wasps (tribe Sycophagini, Hill, 1967: 92), while the spe-
cies of section Sycomorus are inhabited by Sycophaga wasps (tribe Sycophagini,
Hill, 1967 : 92 ) .

Literature Cited

Corner, E. J. H. 1958, An introduction to the distribution of Ficus. Reinwardtia 4: 15-45.
. 1959. Taxonomic notes on Ficus Linn., Asia and Australasia. Sections 1-4. Card.

Bull. Singapore 17: 368-485.

1960. Taxonomic notes on Ficus Linn., Asia and Australasia. Sections 5-6. Card.

Bull. Singapore 18: 1-69.

1965. Cheek-list of Ficus in Asia and Australasia with keys to identification. Card.

Bull. Singapore 21: 1-186.

1967. Ficus in the Solomon Islands and its bearing on the Post-Jurassic History of

Melanesia. Philos. Trans., Ser. B, 253: 23-159.

1969a. The complex of Ficus deltoidea; a recent invasion of the Sunda Shelf.

Philos. Trans., Ser. B, 256: 281-317.

— . 1969b. Ficus section Adenosperma. Philos. Trans., Ser. B, 256: 319-355.

1970. Ficus subgen. Pharmacosycea with reference to the species of New Cale-

donia.

383-433

* Corner (1970: 383) suggested that subsection Papuasyce (of section Sycocarpus), to
which F. itoana, microdictya and pritchardii belong, should become a fifth subgenus as a
monoecious group distinct from subgenus Ficm but with F. itoana us the dioecious product.

310 ANNALS OF THE MISSOURI BOTAXICAL GARDEN [Vol. 64

Galil, J, 1973. Pollination in dioecious figs; pollination of Ficus fistulosa by Ceratnsolcn

hewitti. Card. Bull. Singapore 26: 303-311.
Crandi, G. 1917. Contributo alia a^noscenza degli Agaonini ( Hymenoptcra, Chalcididae)

di Ciava. Boll. Lab. Zool. Gen. Agrar. R. Scuola Agric. Portici 12: 1-60.
Hill, D. S. 1967. Figs (Ficus spp.) of Hong Kong. Hong Kong University Press, Hong

Kong. 128 pp.
Ramihkz B., W. 1970. Host specificity of fig wasps (Agaonidae). Evolution 24: 680-691.
. 1974, Coevolution of Ficus and Agaonidae. Ann. Missouri Bot. Garden 61: 770-

780.

WiEBES, J. T. 1963a. Taxonomy and host preferences of Indo-Australian fig wasps of the
genus Ceratosolcn (Agaonidae). Tijdschr. Entomol. 106: 1-112.

1963b. Indo-Malayan and Papuan fig wasps ( Hymenoptera, Chaleidoidea) 2.

Tl»e genus Pleistochmtes Saunders (Agaonidae). Zool. Meded., Leiden 38; 303-321.

STUDIES IN BIGNONIACEAE 25: NEW SPECIES

AND COMBINATIONS IN SOUTH
AMERICAN BIGNONIACEAE'

Alwyx II. Gentry-

Absthact

Five new species of South American Bignoniaceae are described — Anemopacgma gran-
villei A. Gentry, Arrahidaea ornithophila A. Gentry, Cuspidaria octoptera A. Gentry, Memom
cristicalijx A. Gentry, and Tahehuia catarinensis A. Gentry — and two new combinations —

Lundia virginalis \dr. nitidula (DC.) A. Gentry and Mcmora imperatoris-maximilianii (Wawra)
A. Gentry — arc made.

Anemopaegma granvillei A. Gentry, sp nov.

Frutex scandens; sine pseudostipulis vel consocicbus glandularnm in nodis inter petiolos;
folia 2-foliolata, foliolis oblongo-ellipticis, infra omnino puberulis; inflorcscentia axillaris,
racemosa, piiberula; florcs calyce cniinlato, corolla lutea, tnbo extus glabro, ovario ovoideo,
minute lepidoto, ad basim non contractor fructiis ignotus.

Vine; branchlets finely but prominently striate, elenticellate, piiberulous,
without interpetiolar glandular fields; pseudostipules (only 1 seen) subulate,
4 mm by 1 mm. Leaves 2-foliolate, sometimes tendrillate (tendril tip not seen);
leaflets ( ovate- ) oblong-elliptic, shortly and obtusely acuminate, rounded or
truncate at the base, 10-15 cm long, 4.5-6.5 cm wide, chartaceous, puberulous
throughout beneath with rather scattered erect trichomes, above puberulous
only along the main veins, the secondary veins looped and connecting several
mm from the margins, not very prominent nor strongly ascending, drying olive,
glossy above, dull below. Inflorescence a contracted axillary raceme, densely
tannish puberulous; pedicels subtended by linear 2-3 mm long bracts. Flowers
with the calyx cupular, asymmetrically truncate, 7-10 mm long, 7-8 mm wide,
puberulous at the base and around the margin, with fields of plate-shaped glands
in the upper half; corolla tube pale to lemon yellow, the lobes pale yellow,
tubular-campanulate, ca. 5 cm long, the tube 3.5-4 cm long, the lobes ca. 1 cm
long, the tube glabrous outside, the lobes glandular-lepidote with ciliate mar-
gins, large glands absent below the lobes; stamens didynamous; ovary (in bud)
ovoid, longitudinally ridged, minutely lepidote, not contracted at the base; disc
large, patelliform, 2 mm long, 3 mm wide. Fruit unknown.

Type: French Guiana. Riviere Petite Ouaqui, vegetation ripicole verse
lembouchure de la crique Carbet Brule, 27 July 1973, de Granville 1935 (CAY,
holotype and isotype; MO, fragments and photocopy).

The combination of glabrous corolla tube and puberulous leaves indicates
affinity with A. puherulum (Seib.) Miranda which belongs to the A. grandi-
folium (Jacq. ) Merrill & Sandw. complex. However, all species of this complex

^ Supported by grants from the National Science Foundation.

^Missouri Botanical Garden, 2345 Tower Gnwe Avemie, St. Louis, Missouri 63110.

Ann. Missouri Bot. Card. 64: 311-319. 1977.

312 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

differ in having more prominently lenticellate branchlets and much more promi-
nent and strongly ascending secondary venation. Inflorescence bracts of A.
puherulum and its relatives are also smaller and less prominent, and leaves are
relatively ovate, never as oblong as in the new species. Of the other pubescent-
leaved Anemopaegma species, A. villosum A. Gentry, A. rugosum (Schlecht.)
Sprague, A. goyazense K, Schum., and A. velutinum Mart, ex DC. have promi-
nently lepidote corolla tubes. The former two also differ in possessing foliaceous
pseudostipules, the latter two have more densely pubescent 3-foliolate leaves.
Only two conspicuously pubescent-leaved species have glabrous corolla tubes —
A. hilarianum Bureau & K. Schum. has more pubescent, smaller, ovate leaves
and a paniculate inflorescence; A. brevipes S. Moore, which may be the closest
relative of A. granvillei, has foliaceous pseudostipules, a shorter calyx and
smaller leaflets which are much more densely puberulous beneath.

Arrabidaea ornithophila A. Gentry, sp. nov. — Fig. 1.

Frutex scandens; interdum consociebus glandularum in nodis inter petioles; folia tri-
foliolata, foliolis elliptico-oblongis; inflorescentia floribus in panicula terminali dispositis;
calyx tubulosiis, bilabiatus, puberulus; corolla rosea, tubulosa; stamina subexserta, thecis pen-
dulis; ovarium oblongum, lepidotum; discus annulatim pulvinatus; capsula linearis.

Liana; branchlets terete, puberulous, with or without inconspicuous inter-
petiolar glandular fields; pseudostipules lacking. Leaves 3-foliolate (sometimes
simple in part); leaflets elliptic-oblong, rounded to acuminate at the apex,
rounded at the base, 7-21 cm long, 4-10 cm wide, chartaceous, above minutely
lepidote, otherwise glabrous or with a few inconspicuous trichomes along the
main vein, below densely puberulous, drying dark above, light gray below;
petiolules 1-4 cm long; petioles 2.5-6 cm long, lepidote and puberulous. In-
florescence a terminal panicle, its branchlets densely puberulous with short
glandular trichomes and longer nonglandular trichomes. Flowers with the calyx
tubular, bilabiate, 10-13 mm long, 5-7 mm wide, the lobes almost 2 mm long,
densely pubescent with conspicuously reddish glandular and eglandular tri-
chomes; corolla cherry red, tubular, 2.3-2.5 cm long, 0.4-0.5 cm wide at the
mouth of the tube, the tube 1.9-2.1 cm long, the lobes 2-3 mm long, densely
puberulous outside, inside very densely pilose with exceedingly long (several
mm) tangled trichomes in the upper part of the tube, these completely filling
the throat, less densely pilose with shorter trichomes below, villous at the level
of stamen insertion; stamens didynamous, inserted 3-4 mm from the base of the
corolla tube, subexserted, the anther thecae subparallel, 3 mm long (including
connective), joined for the upper 1.4-2 mm, slightly divergent basally, the con-
nective extended 1.4 mm beyond the point of attachment, the filaments 1.6-1.8
cm long, the staminode 4 mm long; pistil ca. 2 cm long, the ovary oblong, taper-
ing slightly towards the top, 2 mm long, 0.8 mm wide, densely glandular lepidote;
disc pulvinate, 1 mm long, 2-2.5 mm wide. Fruit linear, compressed, the apex
and base subacute, ca. 20 cm long, drying dark.

Type: Brazil, para: Distrito Acara, Thome Assii, Pao Vermelho, 50 m,

■

border woods in sun, vine climbing small trees, cherry red flower, occasional, 3
Aug. 1931, Y. Mexia 6041 (MO, holotype; NY, US, isotypes).

1977]

GENTRY— SOUTH AMERICAN BIGNONIACEAE

313

CJix)K^+fc

Figure 1. Arrahidaea ornithophila A. Gentry. — A. Inflorescence; X%. — B. Leaves;

X%. [AitGx Mexia 6041 (MO).]

3]^4 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Additional collections examined: Buazil. mahanhao; Km 374 Belem-Brasilia, cipo
sobre arvores, flores vermolhas, 26 Aug. 1960, Olivcira 1046 (IAN), amapa; Santa Patricia,
mar^'em esqucrda do Rio Jari, 160 m, 13 Mar. 1970, Silva 2971 (IAN), paba: Km 32, Belem-
Brasilia highway, liana, corolla red, 27 Aug. 1964, Prance ir Silva 58902 (NY). Km. 289-293^
Belem-Brasilia, cipo sobre arvores, flores vermelhas em fonna de tubo, frutos em vagem com-
pridos e chatos, 31 July I960, Oliveim 945 (IAN). Estrada Belem-Brasilia, cipo, flores
vermelhas, vistosas ornamental, May 1960, Froes 34938 (IAN). Rodovia Belem-Brasilia, km
92, flores vermelhas, 21 Aug. 1959, Kuhlmann i:r Jimbo 78 (IAN). Km 167-173 da Estrada
Belem-Brasilia, flores vermelhas em cachos, 25 Apr. 1960, Olivcira 549 (IAN). Km 243-
239 da Rodovia Belem-Brasilia, 8 July 1960, Olivcira 878 (IAN). Region do Rio Jari, estrada
de Monte Dourado ao Mungaba, 27 June 1968, Olivcira 4681 (IAN, NY); 2 July 1968,
Olivcira 4739 (IAN, NY); 10 June 1968, Silva 2149 (IAN). Rio Jari, Planalto de Monte
Dourado, 16 June 1968, Olivcira 4548 (IAN, NY); 2 Oct. 1968, Silva 1084 (NY). Rio Jari,
serra de Monte Dourado, 140 m, 10 Nov. 1967, Olivcira 3604 (IAN, NY); 13 Nov. 1967,

Olivcira 3524 ( IAN ) ; 18 June 1970, Silva 3227 ( IAN ).

The closest relatives of this very distinct, presumably hummingbird-pol-
hnated, species are Arrabidaea trailii Sprague and Fridericia speciosa Mart.
Arrahidaea trailii has a similar (but smaller) tubular red corolla with sub-
exserted (but much smaller and with a minute connective) anthers and gray-
drying (but never simple) leaves with densely puberulous undersurfaces. The
calyx of A. trailii is smaller, subtruncate, and evenly 5-denticulate. The flower
and calyx of A. ornithophila are even more similar to those of Fridericia speciosa
except that the calyx of Fridericia is somewhat inflated and conspicuously
5-ridged. The anthers of Fridericia are included and the thecae divergent. The
new species is so clearly intermediate between Arrahidaea trailii and F. speciosa
that its existence seriously weakens the case for retention of monotypic Fridericia
as a distinct genus.

Cuspidaria octoptera A. Gentry, sp. nov.

Frutex scandens; ramidi teretes, sine pseudostipulis, consociebus glandularum in nodis
inter petiolos; folia 3-foliolata vel 2-foHolata, interdum cum cirrho, foliolis ovatis vel ellipticis,
infra sparsim puberulis saltem nervisequentibus; inflorescentia paniculata, bracteata; flores
calyce cupulato, 5-denticidato, corolla tubulo-infundibuliformi, extus puberula, intus fere
glabra, staminibus didynamis, antherarum thecis reflexis, ovario oblongo, dense lepidoto;
capsula anguste oblonga, subtetragona, alis octo longitudinalibus.

Vine; branchlets terete, glabrous or subpuberulous, when older with scat-
tered, pale, round lenticels, the interpetiolar glandular fields divided, the two
halves separated by a nonglandular medial strip; pseudostipules lacking. Leaves
3-foliolate or 2-foliolate with a (presumably simple) tendril or tendril scar, never
simple even at the base of branchlets; leaflets ovate to elliptic, acuminate, the
base rounded, 3-8 cm long, 1.2-4 cm wide, chartaceous, the main veins plane
or prominulous above, slightly raised below, somewhat puberulous along the
midvein above, sparsely puberulous or pilose along the main veins below and
sometimes scattered subpuberulous on the lower surface, the margins noticeably
ciliate, drying brownish olive; petiolules 0.3-1.6 cm long; petioles 1.3-3.5 cm
long, varyingly puberulous. Inflorescence a terminal panicle, lepidote and
puberulous, bracts and bracteoles linear, to 2 mm long. Flowers with the calyx
cupular, puberulous, ca. 2 mm long (with teeth), 2-3 mm wide, 5-denticulate,
the teeth 0.5 mm long, extended as ribs on the outside of the calyx; corolla
magenta, tubular-infundibuliform, 2.6-3 cm long, 0.9-1.3 cm wide at the mouth

1977] GENTRY— SOUTH AMERICAN BIGNONIACEAE

315

berulo

side and on the lobes inside, the tube mostly glabrous inside, slightly glandular
pubescent at the level of stamen insertion; stamens didynamous, inserted 5-6
mm above the base of the corolla, the anther thecae divaricate, pilose, reflexed
forward from near the base, ca. 1.5 mm long, the blunt pilose connective ex-
tended 0.3-0.4 mm; pistil 1.6-1.7 cm long, the ovary oblong, densely lepidote,
1.5 mm long, 1 mm wide; disc small, pulvinate, 0.3 mm long, 1 mm wide. Cap-
sule linear-oblong, basically subtetragonal, 4-30 cm long, 1.3-2.3 cm wide in-
cluding the 8 thin longitudinal wings, each wing 3-8 mm wide, glabrous, drying
dark brown; seeds thin, bialate, ca. 1 cm long, ca. 3 cm wide, the wings brownish-
hyaline, not sharply demarcated from the seed body.

Type: Brazil. Without locality, Nacleaud s.n. (P, holotype; MO, P, iso-

types ) .

Additional collections examined: Brazil, hio de Janeiro : Rio Paraliyha, 29 Nov. 1880,
Netto et al. s.n. ( R-23675, MO). Baixada Flumincnse, Pilar, 30 Dec. 1939, Lutz 1565

With

Sao Paulo: Without locality. We

516 (BM, mixed with flowering material of Arrahidaca florida DC).

This overlooked species is closely related to C. convohita (Veil.) A. Gentry
[C. pterocarpa (Cham.) DC] and Sandwith (in herb.) has identified flowering
material (Lutz 1565) with that species, which differs most conspicuously in a
merely 4-winged fruit and a larger much more deeply divided calyx with teeth
2.5-4 mm long. Vegetatively the two are extremely similar but C. octoptera
differs in having uniformly 2-parted interpetiolar glandular fields (these may
be 2-parted, undivided, or absent in C. convoluta), typically darker-drying
leaves with noticeably ciliate leaflet margins ( the leaflet margins of C. convoluta
are pubescent only in var. pubescem Mello which has the whole leaf undersur-
face pilose), and in the complete absence of simple leaves at the base of vege-
tative shoots. CuspicJaria convoluta ranges from northern Argentina and ad-
jacent Paraguay north to the states of Minas Gcrais and Rio de Janeiro in Brazil
where it overlaps with C. octoptera. However, the two species are probably
ecologically separated since all altitudinal records for C. convoluta are from

above 500 m while C. OCtontem is nnrmrpntlv rpsh-iftfrl fn tliP r-nncfnl ln\x;hinr1c

Lundia virginalis DC. var. nitidula (DC.) A. Gentry, comb. nov.

L. nitidula DC, Prodr. 9: 181. 1845. syxtypes: Brazil, Minas Gerais, Martius s.n. (M, frag-
ment G-DC). Brazil, Sebastianapolitana, Martius s.n. ( M ).
Bignonia nitidula Mart, ex DC, Prodr. 9: 181. 1845, nom. nud., pro syn.

De Candolle (1845) was the first to systematically treat Lundia. His spe-
cific concepts have proven overly narrow and four of his nine species are now
generally regarded as conspecific. Bureau (1868) united L. hehantha DC. wnth
L. virginalis DC. but maintained L. nitidula as specifically distinct. Baillon
(1888) was the first to unite these two concurrently published species of de
Candolle which differ only in calyx length. Since Baillon adopted L. virginalis,
that name takes precedence for the species. Later Bureau & Schumann ( 1896-
1897) likewise concluded that L. virginalis and L. nitidula were conspecific but

316

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

d

d form as L.

nitudula var. virginalis (DC.) Bureau & K. Schum. Since the name for the com-
bined taxon must be L. virginalis under Article 57 of the Intermtional Code of
Botanical Nomenclature, the new combination proposed above is needed if the
long-calyxed plant is to be recognized at varietal rank. An additional problem in
this complex is posed by the existence of a second short-calyxed form which dif-
fers in more greenish-drying calyx and leaves, shorter, broader corolla, and
white (not magenta) flower color. This has been separated as L. glazioviana

d

and Sandwith ( in herb. ) .

Memora imperatoris-maximilianii (Wawra) A. Gentry, comb. nov.

Bignonia imperatom-maxhniliam Wawra, Bot. Ergeb. Reise Maximilian Bras. 73, tab. 10.

1866. type: Brazil, Bahia, Wawra 6 Maly 156 ( W).
Pleonotoma imperutoria-inaximiliani (Wawra) Bureau & K. Schum. in Mart., Fl. Bras. 8(2):

279. 1897.

reviewing

Wawra

ma

tion. Salient characters of the plant include the 5-denticulate calyx, open
paniculate inflorescence, red corolla with both tube and lobes glabrous out-
side, plate-shaped glands at base of the lobes, triternate leaves with rather large
leaflets, and especially the terete branchlets. The latter two characters alone
almost mandate placement in Memora; all other features are also consistent with

this placement.

Memora imperatoris-maximilianii is unusual in Memora because of its red

flowers but M. magnifica (Mart, ex DC.) Bureau has bright orange or red orange
flowers and several Memora species have yellow orange flowers. The combina-
tion of conspicuously 5-denticulatc but otherwise truncate calyces and very
minutely bracteate inflorescences are matched in Memora only by M. campicola
Pilger which has a very different inflorescence, yellow flowers, and multiple
compound leaves with small pubescent leaflets and M. hiternata A. Samp, which
has sessile leaflets and corolla lobes pubescent outside. Both of these species have

Memora

nuixtmt

now available. This is A. Lima 57-2799 (IAN) from Nazare da Mata, Pernam-
buco, and permits amplification of Wawra's description. The most noteworthy
additional characteristic is the presence of conspicuous subulate pseudostipules
3-5 mm long (cf. M. cristicalyx below). The leaflets of this collection are entire
to serrulate and it has the distinctly paniculate inflorescence figured by Wawra.
Field notes on the Lima collection describe the flowers as "roseo nos lobos (vari-
ando de r6seo a lilaz bem claro) e amarelo no tubo; calice esverd."

Memora cristicalyx A. Gentry, sp. nov.

Memora amtiloha Bureau, Bull. Soc. Bot. France, Mem. 58 (3f): 523. 1911, nom. nud.

Habitus ignotus; ranuili teretcs, glabri, sine consoeiebus glandulanun in nodis inter peti-
oles; folia biternata, foliolis anguste ovatis, plerumque scrratis; inflorcscentia anguste paniculata,

1977] GENTRY—SOUTH AMERICAN BIGNONlACEAE 3]^^

axillaris, plus minusve glabrata; f lores calyce campanulato, 5-denticulato dcntibus extensis in
cristis, glabro, corolla tubiilo-infundibuliformi, tubo extus glabro, lobis puberulis, ovario cy-
lindrico dense lepidoto; fructus ignotus.

Habit unknown; branchlets terete, finely striate, glabrous, drying dark with
numerous round whitish lenticels, the nodes without interpetiolar glandular
fields, with a raised ridge connecting opposite petioles; pseudostipules prominent,
linear-subulate, 4-10 mm long. Leaves biternate, the tendril tip not seen; leaf-
lets narrowly ovate, acute to acuminate, basally rounded, usually conspicuously
serrate, chartaceous, 2.5-9 cm long, 1.3-4.2 cm wide, mostly glabrate, incon-
spicuously scattered-lepidote, minutely subpuberulous at the base of the midvein
above and below, drying blackish above, dark oHve with reddish black midvein
below; petiole and petiolules adaxially grooved, subpuberulous. Inflorescence
a racemose axillary panicle, the lateral branches subsessile to 1.5 cm long, glabrate
to subpuberulous, terminated by a pair of several mm long thinly subulate bracts,
these subtending a cluster of 1 to 8 fk)wers on ebracteolate pedicels up to 2.5
cm long. Flowers with the calyx campanulate, subtruncate, 5-denticulate, the
0.5 mm long teeth extended as raised lateral ridges to the base of the calyx, 6-7
mm long, 5-6 mm wide, glabrous; corolla probably yellow (drying blackish
yellow in type), tubular-infundibuhform, ca. 3 cm long, ca. 1 cm wide at the
mouth of the tube, the tube ca. 2.5 cm long, the lobes ca. 1 cm long, the tube
glabrous and the lobes puberulous outside, the lobes puberulous inside, tlie tube
glabrous inside except at the level of stamen insertion; stamens didynamous, the
filaments ca. 1.5 cm long, the anther thecae 3 mm long, somewhat divergent;
ovary cylindric, 2 mm long, 1 mm wide, densely minutely lepidote; disc annular-
pulvinate, 0.6 mm long, 1.5 mm wide. Fruit unknown.

Type: Brazil, ceara: Without data, Fr. Allerndo 6- A/, de Cystieiros 1045
f R-127332, holotvDe: MO. isotvoe).

Additional collection examined: Brazil. Without data, Glaziou 11232 (P, 2 sheets).

This species is related to A/, imperatoris-jnaximilianii (see above) but is other-
wise remarkably isolated. Biternate leaves and terete stem mandate placement
in Memora where it is the only species with serrate leaflets. Ridged calyces,
rather reminiscent of Clytostoma pterocahjx Sprague, are unique in Memora
and similar long subulate pseudostipules are found only in M. imperatoris-
maximilianii. Memora imperatoris-muximiUanii differs in an unridged glandular
calyx, tlie corolla lobes glabrous outside and with plate-shaped glands at their
bases, entire to subentire leaflets and a more open inflorescence. It is possible
that additional collections from poorly known and apparently Bignoniaceae-rich
northeastern Brazil will show that this and M. imperatoris-maximilianii repre-
sent opposite extremes of a variable population but the available evidence sug-
gests specific separation.

The Paris specimens have been annotated by Bureau as ''Memora acutiloba
Bur., n. sp." and that nomen nudum was used in Glaziou's "Liste des plantes du
Bresil central recueillies en 1861-1895." I prefer the more descriptive epithet
cristicalyx.

318

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Tabebuia catarinensis A. Gentry, sp. nov.

Frutex ad 3 m altiis; folia palmatini (6-)7-f{)liolata, foliolis oblonRo-ellipticis vel
obovatis, serratis, jrlabroscentibus; inflorescentia paniculata, aliquantiim congcsta, ramis dense
stellato-riifescentibus; flores calyce tubulo-campanulato, piloso, corolla Iiitea, extus glabra,
intus fauce pubcnila, ovario ovoideo, dense lepidoto; capsula angiiste oblonga, tonientosa
trichomatibiis barbatis, irreguliter rugnlosa.

Shmb 0.5-3 m tall; twigs terete, striate, minutely and glabrescently stellate-
tomentose. Leaves palmately (6-)7-foliolate; leaflets oblong-elliptic to obovate,
acute or very briefly acuminate, the base rounded, conspicuously and evenly
serrate, the terminal leaflet to 11 cm long and 5 cm wide, the lateral leaflets pro-
gressively smaller, chartaceous, when young sparsely stellate-pubescent along
the main veins above and below, almost completely glabrescent at maturity,
drying blackish or dark olive above and below; petiolules to 4 cm long; petioles
5-13 cm long, glabrescently stellate-tomcntose. Inflorescence a several- to many-
flowered, short terminal panicle, its branches densely rufescent with stellate,
barbate and simple trichomes to 1 mm long, the bracts minute, subulate, to 3
mm long. Flowers with the calyx tubular-campanulate, irregularly 3-5-lobed,

12-20 mm long, 8-1 _ \ '

to 1 mm long; corolla yellow, tubular-infundibuliform, 5-7 cm long, 1.4-2.2
cm wide at the mouth of the tube, the tube 3.5-5 cm long, the lobes 1-2 cm long,
drying dark brown with blackish venation, glabrous, the tube glabrous outside,
inside pubescent with rather short (0.5-0.8 mm long) stiff erect trichomes de-
scending from sinuses of the corolla lobes to above the level of stamen insertion,
more or less glabrous at and below stamen insertion; stamens didynamous, in-
serted ca. 10 mm abo\'e the base of the corolla tube, the filaments 1.7-2.2 cm
long, the staminode ca. 6 mm long, the anther thecae widely divergent, 3 mm
long; pistil 3.2-3.4 cm long, the ovary

densely lepidote, drying blackish, the ovules ca. 4-seriate in each locule; disc
shortly cyUndric, 1 mm long, 2 mm wide.
1.5-1.8 cm wide, densely reddish-brown tomentose with mostly barbate tri-
chomes ca. 0.5 mm long, the surface finely and irregularly rugulose, not regu-
larly striate; seeds (very immature) bialate, the wings hyaline membranaceous.

5-9

Type: Brazil, santa catarina: Monte Crista, Garuva, campo, 750 m, ar-
busto 2 m, flores amarela, 21 Oct. 1966, Klein 6 Ravenm 6834 (K).

Additional collections examined: Brazil, santa catarina: Monte Crista, Garnva, 750
m, campo, arbusto 2 m, fnito imatnro marron, 21 Oct. 1966, Klein 6 Ravenna 6828 (K);
matinha, flor amarela, arbusto 3 m, 21 Oct. 1966, Klein i:r Ravenna 6843 (K); 900 m, campo,
arbusto 0-5 m, flores amarela, 2 Sep. 1960, Reitz 6 Klein 9790 (K). Morro do Campo Alegre,
Sao Francisco do Sul, 1,200 m, campo, arbusto 1 m, flores amarela, 5 Sep. 1960, Reitz 6 Klein
9766 (K); sterile, 24 Mar. 1961, Reitz ir Klein s.n. (K). parana: Mun. Campina Grande
do Sul, Pico Caratiiva, 1,950 m, arbusto do topo do morro, flor amarela, 5 Oct. 1967, Hutsch-
bach 1 7325 ( MO ) .

This species is superficially most similar to T. bureauvii Sandw. endemic to

J

very

shaped glands, a fewer-flowered, more finely tomentose inflorescence, a sparsely
papillose-puberulous corolla throat, a longer (ca. 4 mm long) ovary, narrower

1977] GENTRY— SOUTH AMERICAN BIGNONIACEAE

319

leaflets, and longer, smooth-surfaced glabrous fruit. The new species was sup-
posed by Sandwith & Hunt (1974) to be a form of T. chrysotricha (Mart, ex
DC.) Standley. A hybrid origin from that species and T. alba (Cham.) Sandw.
was also suggested as the new species is intermediate in most respects between

two

catarinemis

is ecologically distinct from lowland T. chrysotricha (below 800 m) but not from
T. alba. Its flowers and inflorescence are identical to those of T. alba of which
it could be a glabrescent-leaved derivative. However, the shrubby habit, shorter
rough-surfaced (not striate) fruit, and uniformity of the strikingly different
glabrate, rather than densely canesccnt, leaves support specific recognition.
Hatschbach (personal communication) reports that the new species can be
rather common locally at high altitudes in the Serra do Mar.

Literature Cited

Baillon, H. 1888. (1891). Histoire des Plantos. Vol. 10. L. Ilachette et Cie, Paris.
Bureau, E. 1868. Revision des j^enres Tynanthus et Lundia. Adansonia 8: 270-294.

& K. Schumann. 1896-1897. Bignoniaceae. In K. F. P. von Martins, Flora Bra-

siliensis. Vol. 8(2) : 1-434. Leipzig.

Candolle, a. p. de. 1845. Prodromus Systcniatis Naturalis Regni Vegetabilis. Vol. 9.
Treutel et Wiirtz, Paris.

Sandwith, N. Y. 1959. Contribntions to the Flora of Tropieal America LXV Studies in
Bignoniaceae XXIV. Kew Bnll. 1958: 427-443.

& D. Hunt. 1974. Bignoniaceae. In P. R. Reitz, Flora Ilustrada Catarinense.

Fasc. BIGN. Itajai, Sta. Catarina, Brazil.

NEW RECORDS OF APOCYNACEAE
FOR PANAMA AND THE CHOCO'

Alwyn H. Gentry-

Abstract

Tahcrnaenwtitana peiuhila and T. hmgipes, synonymized under T. chrysocarpa
in the Flora of Panama, are recognized as distinct from it. Stemmadenia allenii was origi-
nally described from fruiting and flowering material of different species, one of which —
the most conunon wet-forest species of the genus in Panama— is now described as S. minima
A. Gentry. The first South American record of the North American T abernaemontana ar-
borea, the first North American record of the South American Odontadenia cognata, and the
reconfirmation of the occurrence of Fosteronia myriantha in Panama are reported.

Panamanian plants referred to Tahernaemontana chrysocarpa in the Flora
of Panama treatment ( Nowicke, 1970 ) prove to represent three distinct species.
These three species, somewhat similar on the basis of floral characteristics, are
easily separated by vegetative and fruiting characters.

Tabernaemontana pendula Woodson

This species was described from a single specimen from El Valle {Allen
1734). It was compared by Woodson (1940) with T. amy gdalifolia Jacq. be-
cause of its cxserted anthers but lumped with T. chrysocarpa, a species charac-
terized by included or subexserted anthers, by Nowicke (1970) in the Flora
of Panama. Tahernaemontana pendula has a much longer peduncle than either
T. amy gdalifolia or T. chrysocarpa. It also has wider, more elliptical leaves and
a characteristically wrinkled-reticulate fruit surface. The long peduncle is
also obvious in fruit. The fruit, previously undescribed, is similar in shape to
that of T. chrysocarpa. Two additional collections of this species, both in fruit,
are now at hand. These are Mori et al. 1912 from La

Mesa

Mori

Province.

Tabernaemontana longipes Donnell Smith

This species was described from Costa Rica and has been thought endemic
to that country. It is closely related to T. chrysocarpa and the Panamanian spe-
cimens of r. longipes were included with that species in the Flora of Panama.
Vegetatively T. longipes differs from T. chrysocarpa in its elliptic leaves, always
broadest near the middle; the latter has narrowly obovate to oblanceolate-el-
liptic leaves, broadest above the middle. The fruit of T. longipes, previously
undescribed, is very distinctive with a verrucose muricate-ridged surface quite
unlike the smooth, papillose or finely reticulate-ridged fruits of other Panamanian
soecies of Tahernaemontana and Stemmadenia. I have seen no fruits of this

^ Supported by NSF grant OIP75-18202.

=' Missouri Botanical Garden, 2345 Tower Grove Avenue, St. Louis, Missouri 63110.

. Missouri Box. Card. 64: 320-323. 1977.

1977]

GEXTRV— APOCYXACEAE 32]

Species from Costa Rica, but the Costa Rican and Panamanian collections seem
indistinguishable on the basis of vegetative and floral characters.

Tabernaemontanci hmgipes has been collected in Panama only above El Valle
de Anton, Code Province, where it is locally very common. It is represented b)^
thirteen collections from El Valle in the Missouri Botanical Garden Herbarium
including all the Code Province collections cited as T. chrysocarpa in the Flora

1

(see above). The additional

collections of this species — Kennedy et al. 3035, Liesner 747, Croat 14383, Gentry
6 Dayer 3612, and Gentry 6873 — were all identified and distributed as T. chryso-
car pa.

The four Panamanian Tahernaeniontana species with anthers tinged blue
green can be separated by the following key:

a. AiitluTS exsertL'd or half-c\sertcd; follicles narrow (more than twice as long as wide)
or reniform and finely reticulate-wrinkled.

b. Peduncle very long, exceeding the lea\es; fruit reniform and finel>' reticulate
r. pcmlula

1)1). Peduncle not elongate, inflorescence not exceeding the leaves; fruit narrowly

elliptic, smooth ____ ._. T. amygdalifolia

aa. Anthers included or l>arely exserted; follicles n^niform, smooth or verrucose and

muricate-ridged.

c. Leaves narrowly obovate to ohlanceolate-clliptic, broadest above the middle;

fruit smooth - — - — T. clinjsocarpa

cc. Leaves elliptic, broadest at the middle; fruit verrucose and muricate-ridged

.___ ...._ T. longipes

Tabernaemontana aiborea Rose

This species has previously been recorded from Belize to Panama. It is easily
recognized by its yellow anthers inserted near the base of the corolla tube. Like
many other "Central American" species, it also occurs in the northern Choco.
Two Choco collections have been seen— Dn/cc 12233 (MO, NY, OSU) and
Duke 11169 (OSU), both from the Rio Truaiido.

Stemmadenia allenii Woodson

This species, described from El Valle, Panama, was separated by Woodson
from closely related Costa Rican S. alfari (Donnell Smith) Woodson because
of its longer calyx lobes and larger corolla with broader throat and longer lobes.
The type collection of S. allenii is in fruit and the fruit is of the same narrow,
long-acuminate form as that of S. alfari. Many fruiting collections of S. allenii
are now at hand, all with fruits of the same characteristic slender form. How-
ever, a vegetatively similar species with a very different obovoid to almost orbic-
ular fruit also occurs in the same wet-forest areas and the flowering material
attributed to S. allenii by Woodson (1941) and Nowicke (1970) actually be-
longs to the thick-fruited species (see below). The first flowering collection of
the real S. allenii is Liestier 765 (MO) from El Valle, the type locality, which
has the corolla throat narrower (rather than broader!) than S. alfari and calyx
lobes only 4-5 mm long; in fact S. allenii proves separable from S. alfari not by
the characteristics cited by Woodson but by their opposites! Tlie real S. allenii

322 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

keys out to S. aJfari in the Flora of Panama^ but that species does not occur in
Panama unless S. allenii itself should be considered as merely a geographic vari-
ant.

Stemmadenia minima A. Gentry, sp. nov.

Fnitrx lac'tiftT. Folia parva, angnste clliptica, acuminata, glal)ra. Flor calycis lobis mem-
branacris, 4-10 nun lougis, corolla iufuudihulifonni, albida, tube torto, stanu'nibus tubo
corollac propc mcdiain inscrtis. Folliculi crassi, obtusi, fcrc suborl)iculati.

Shrub or small tree 1.5-5 m tall; branchlets somewhat angulatc, minutely
papillose, lactiferous. Leaves small, narrowly elliptic, acuminate, cuneate at the
base, to 11 cm long and 3.9 cm wide (largest leaf 1.9-3.9 cm wide, x — 2.97 cm),
glabrous above and below, membranous; petioles 2-10 mm long, not clearly
differentiated from the leaf base. Inflorescence a single flower, terminal from
between two dichotomous lateral branches, glabrous, with a minute triangular
bracteole; calyx lobes ine(iual, membranous, narrowly oblong, 4-10 mm long,
glabrous; corolla infundibulifonu, white to cream, the tube proper 12-17 mm
long, twisted 180^ at the top, the throat 15-20 mm long, the lobes obovate, ca.
1 cm long: stamens attached at the middle of the corolla tube, the anthers 4-5
miu long. Follicles thick, blunt, almost orbicular ("obovoid subreniform"), 2-3
cni long, 1,5-2.5 cm l:)road.

Typk: Panama, panama: Cerro Jefe, 800-1,000 m, 21 Dec. 1972, Gentry
6763 (MO, holotype; duplicates were distributed as S. cf. alfari).

Addilional collections examined: Panama, canal zone: Madden Lake, Dtctjcr 6-
LaUathin HH27A (MO), cocle: El Valle, AUen 2239, 2364; Croat 25347; Dwtjcr ct al
4502(1 (all MO), colon: Santa Rita Rid^e, GnUry 6090, 6562 (both MO). Panama:
Cerro Canipana, Busctj S61; Croat 12144; Dressier 3523; Dwyer & Kirkhrkle 7829A; Gentry
4934; Mori b Kalhniki 1930; Porter et al 5249 (all MO). Cerro Jete, Croat 13028, 14444;
Dressier 3333; Duke 9449 (all MO). El Llano-Carti Road, Gentry 5071; Gentry et al 14201,
14214; Mori 6 Kalhtnki 2915 (all MO). vkhac;uas; Month of Rio Concepeion, Lewis ct al
2853 (MO).

This species lias been geuerally coufouuded with S. allenii, and, in fact, the
original description of the flowers of that species are based on S. mininui (see
above). Discovery of the short thick fruits of S. minima prove that it is quite
unrelated to S. allenii which has narrow, long-acuminate fruits. Calyx lobe
length of these plants also exceeds that of S. allenii and, in fact, approaches that

of S. hii^nnae Woodson, otherwise reported only from Bocas del Toro Province.
I have previously identified collections of this entity as S. lagunae, which has
a similar thick, rounded fruit. Numerous additional collections of S. mininui are
now at hand and it proves to differ constantly from S. la^iunae in smaller lea^Ts
[largest U*af 1.9-3.9 cm wide (x = 2.97 cm) versus largest leaf 4.3-6.2 cm wide
(x = 5.32 cm)] as well as shorter (x for longest lobe = 8.05 mm versus 17.2 mm),
narrower, not at all imbricate calyx lobes. It usually has white flowers (some-
times pale yellow or white with a yellowish center) while S. la<^unae has yellow
flowers (one collection reported as light yellow) and has a distinct geographic
distribution. No collections are available from the critical area between Santa
Fe de \^eraguas where S. lagnnae occurs and El Valle, the westernmost locality
tor S. minima, but the available evidence suggests spt^cific recognition.

1977] (^.EXTRY— APOCYXACEAE

323

Stemmadenia minima is fairly common in all the accessible middle elevation
wet-forest areas of eastern and central Panama and is the only wet-forest species
of Stemmadenia occurring east of the Canal Zone. It overlaps with S. allenii
at El Valle and Cerro Campana but can be easily distinguished from that species
by its very different fruit, wider corolla throat, smaller, less membranous leaves,
longer calyx lobes and paler flower color.

Forsteronia myriantha Donnell Smith

This species was not treated in the Flora of Panama although Woodson
(1935) had reported it from the Republic on the basis of a single Hayes col-
lection. It has recently been recollected in Panama (Foster 4107^ Barro Colo-
rado Island, Canal Zone). The two Panamanian collections key to this species
but have the petals sparsely pilosulose both inside and outside. Although this
disagrees with Woodson's description of the petals as glabrous or very minutely
papillate without, the type (Ileyde i^ Lux 4533 from Guatemala) also has a few
long trichomes on the outside of some petals and otherwise matches the Pana-
manian material. Forsteronia myriantha differs vegetatively from the other Pana-
manian species in having long trichomes scattered along the leaf midvein and
sometimes over the surface beneath as well as in the nerve axils.

Odontadenia cognata (Stadelm.) Woodson

This widespread South American species was cited from extreme eastern Pana-
ma by Woodson (1935) based on a single immature collection from Puerto Obaldia,
but was subsequently rejected from the Flora of Panama, Ten Panamanian col-
lections are now at hand, all collected in wet-forest areas during the last few
years. It is the most common species of Odontadenia on the El Llano-Carti road
and has also been collected on Santa Rita Ridge, Colon Province, and above
Santa Fe, Veraguas Province. The species also reaches Costa Rica, based on Opier
1724 (MO) from La Selva, Ileredia Province. Odontadenia cognata is easily
told from O. puncticulosa, which it somewhat resembles, by the corolla tube ta-
pering inore evenly to a narrower base with the anthers inserted near the base of
the tube proper rather than at the base of the throat. Several collections are
noted as having pink or orange red corollas, but I have seen corollas of the more
frequently reported yellow or pale j^ellow color only. Similar color variations
occur in South America but do not appear taxonomically significant.

LrrKRATUHK CriED

Nowk:kk, J. W. 1970. Apocynaccae. In R. E. W^xxlson, Ji\ & R. W. Schcry, Flora of

Panama. Ann. Missouri Bot. Card. 57: 59-130.
Woodson, R. E., Jr. 1935. Studies in tlie Apocynaceac l\\ The Ameriean genera of

Ecliitoicleae. Ann. Missouri Bot. Garel. 22: 153-306.
. 1940. Contri1)utions toward a Flora of Panama l\\ Ann. Missouri Bot, Card. 27:

265-364.

— . 1941. Apocynaeeae. 7/j Contributions toward a Flora of Panama V. Ann. Mis-
souri Bot. Card. 28: 461-462.

NEW TAXA AND COMBINATIONS
IN ERAGROSTIS (POACEAE)^

John T. Witherspoon

AliSIRACT

Erof^rostis f:,iuitc}iialensis Witherspoon and E. intermedia Ilitclic. var. appressa Wither-
spoon are descril)ed as new. Tlie former is distinguished from other North American Eragros-
tis by having long pilose hairs on the lemmas and paleas. The latter is distinguished from
other members of the E. intermedia group by its short lemmas and appressed primary
branchlets and pedicels. In addition, fom* new combinations are made — E, intermedia var.
oreophila (L. II. Harvey) Witherspoon, £. intermedia var. praetermissa ( L. II. IIar\'oy)
Witherspoon, E. hirta Fourn. var. longiramea (Swallen) Witherspoon, E. tricJioeoIea Hack.
& Arech. var. floridana (Ilitchc. ) Witherspoon — and keys are provided to the varietirs of
£. intermedia, E, hirta, and E. trieJioeolea,

Eragrostis guatemalensis Witberspoon, sp. nov.

Species E. intermediae varietatum intermediae et praetermissae affinis, sed differt ab
utrcxjue paucis pilosis propre margines lemmatum et palearum et a prima innovationibus
extravaginalis, a secunda paniculis axillis dense pilosis, nervis lateralibus inconspicuis, et
caryoiisibus valde sulcatns.

Perennial, 65-115 cm tall. Cnhm erect to ascending from a somewhat knotty
base, sometimes geniculate at tbe middle nodes; innovations extravaginal, in-
frequently intravaginal, few to many, variously papillose-pilose. Leaven 4-8
per culm; sheaths slightly overlapping below, shorter than the internodes alx)ve,
sparsely to densely papillose-pilose over the rounded back, along the margins
and on the sides of the collar, particularly dense in the region between the
margins and the "keel," occasionally glabrate; ligules 0.3-0.4 mm long; blades
linear-lanceolate, attenuate, flat to involute, 10-22 cm long, 2-5 mm wide, with
dense supraligular hairs and scattered to dense papillose-pilose pubescence on
the adaxial surface, glabrous to densely papillose-pilose abaxially. Panicles el-
liptic to ovate, open, 24-30 cm long, 12-21 cm wide, very long exserted when
mature; branches 24-30 per panicle, 10-13 cm long, ascending to spreading,
curved or flexuous, floriferous 0.8-2.5 cm above the base, the lower ones soli-
tary, paired or in verticels, the upper likewise and equidistant, the primary axils
pilose laterally and adaxially, occasionally bearded all around the branch bases,
the secondary axils often pilose laterally; primary branchlets 4-10 per branch,
30-46 mm long, ascending to spreading, capillary, mostly flexuous; secondary
branchlets 0-2 per primary branchlet, mostly ascending, capillary, flexuous;
pedicels 5-10 mm long ascending to spreading, capillary, flexuous. Spikelets
2-S per primary branchlet, oblong to ovate, acute, slightly compressed, dark
green to plumbeous or brownish, 3-8-flowered, 3.5-7 mm long, 1.2-2.1 mm

^ This paper is taken in part from u Ph.D. dissertation submitted to the University of
Montana, Missoula, Montana. I would h'ke to thank Dr. L. H. Harvey for his support through-
out the study and Dr. John Hay for his assistanee with the Latin diagnoses. A grant-in-ai<l
from Tlie Society of the Sigma Xi is gratefully acknowledged.

"2311 East Livingston, Springfield, Missouri 65803.

Ann. Missouai Bor. Gahd. 64: 324-329. 1977.

19771 WiTUKHSVOOK—ERAGROSTlS 325

wide, the florets slightly imbricate; ghimes with hyaline margins and mem-
branous bodies or the first hyaline throughout, scabrous on the keels^ the first
ovate-lanceolate, acuminate, 1-1.5 mm long, 0.5-0.7 mm wide, the second ovate,
acute, 1.5-2 mm long, 1-1.2 mm wide; lemmas ovate, acute, rounded on the
back below, scabrous on the keel above, the margins and tips hyaline, the bodies
membranous, often tinged with reddish purple, 2-2.4 mm long, 1.2-1.5 mm wide,
the lateral nerves inconspicuous, 2-6 pik)se hairs ca. 0.5 mm long in a longitudinal
row on the broadest part of the lemmas between the lateral nerves and the mar-
gins, the hairs fragile and caducous; paleas with hyaline margins and mem-
branous bodies, usually also hyaline between the keels on the evenly convex
back, ciliolate on the keels, 1.4-1.8 mm long, often with a row of hairs similar
to those on the lemmas, the hairs in a longitudinal row between the keels and
the margins. Caryopses oblong, strongly grooved adaxially, 0.6-0.8 mm long,
0.4-0.5 mm wide, 0.4-0.5 mm thick.

Type: Guatemala: Alamedo, 15 July 1937, /. R. Johnston 930 (F, holo-
typc ) .

Paratvpes: Guatemala: Solola, San Pedro, Stcycrnuirk 47100 (F, US). Guatcmaki,

AguilarSis (F).

Mexico. Mexico, between Nieolas Roinero and Progreso Industrial, Sohn.s 540 (US).
4 mi E of Ixtlhuaea, Sodersirom 515 (US), puebla: El Chaniizal, mun. de Mazapillepec,
Ventura A. 1729 (ENCB).

This new species is similar in many respects to Eragrostis intermedia Hitchc.
var. intermedia and E. intermedia var. praetermissa (proposed below). It dif-
fers from the first by the hairs on the lennnas and paleas and by the pubescence
on the fohage. It differs from the second by the hairs on the lemmas and paleas
and by the inconspicuous lateral nerves.

The only other known species of the E. intermedia group in the Western
Hemisphere with hairs on the body of the lemma is the South American £. semi-
nuda Trin. Plowever, the latter is a strikingly different taxon with extremely
long, villous, involute blades and a very large, diffuse panicle. The spikelets of
E, seminuda are smaller and have only three or four florets. The relationships
of £. seminuda seem to lie more witli E, polytrieha Nces and E. triehoeolea

Hack. & Arech. s. lat.

Era^rostis guatenudensis occurs between 1,900 and 2,650 m on rocky moun-
tain slopes and in oak-pine forests of central Mexico and central Guatemala.
The specific epithet was derived from the taxon's occurrence in Guatemala. When
first discovered among specimens from the Field Museum (F), the only known
collections were those from Guatemala. Since then, however, it has been found
from the states of Mexico and Puebla in Mexico.

Eragrostis intermedia Hitchc. var. appiessa Witherspoon, var. nov.

Varietas oreophilae affinis, sed differt leniinatibus l)re\ ioribus et pedieellis appressis.

Perennial, 31-63 cm tall. Culms erect, weakly tufted, strongly anthocyanic;
innovations intravaginal, rarely extravaginal, few, glabrous. Leaves 2-4 per

326 AWALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

culm; slieutlis overlapping below, sliglitly exceeding to sliglitly shorter than the
internodes above, pilose along the margins and on the collar sides and throat;
ligules 0.2-0.3 mm long; blades linear-lanceolate, acuminate-attenuate, flat or
involute below, tightly involute above, appressed or ascending, 2-9 cni long,
1-2 mm wide, densely pilose above the ligule and a short way up the adaxial
surface. Panicles elliptic when young, mostly deltoid when mature, open to
somewhat condensed, stiff, straight or curved at the tip, 12-24 cm long, 5-20
cm wide, short to long exserted; branches 10-24 per panicle, 5-13 cm long, ap-
pressed-asccnding to ascending-spreading, infrequently spreading, straight, florif-
erous 0.9--3.1 cm above the base, the primary and secondary axils pilose laterally
and abaxially, occasionally also adaxially; primary branchlets 0-9 per branch,
13-45 mm long, appressed to ascending, capillary; secondary branchlets absent;
pedicels 1-6 mm long, appressed to appressed-ascending, capillary. Spikelcts
1-6 per primary branchlet, ovate, acute, barely compressed, reddish purple, 3-5-
flowered, 2.4-3.7 mm long, 1.2-1.6 mm wide, the florets slightly imbricate, tightly
so when young; glumes hyaline or the second with hyaline margins and a mem-
branous body, strongly scabrous on the keels, the first lanceolate to ovate, acu-
minate to acute, 1-1.5 mm long, 0.5-0.6 mm wide, the second ovate, acute, 1.3-1.6
mm long, 0.7-0.8 nmi wide; lemmas ovate to broadly so, acute, rounded or
slightly keeled below, scabrous on the keel above, the tips hyaline, the bodies
membranous, often tinged with reddish puiple, 1.5-L7 mm long, 1-1.4 mm wide,
the lateral nerves very weak; paleas hyaline, ciliolate on the keels, 1.2-1.5 mm
long. Carijopscs oblong to quadrate, weakly grooved adaxially, reticulate, 0.5-0.7
mm long, 0.3-0.5 mm wide, 0.3-0.5 mm thick.

Type: Mexico, jalisco: 7 mi S of Zacatecas-Jalisco border, 11 Oct. 1972,
llarveij 6 Witherspoon 9344 (US, holotype; ENCB, MO, MONTU, NY, RM,
TAES. W, isotypes).

This taxon is phenetically close to Eragrosth intermedia var. orcophila (pro-
posed below) but is distinguished from it by its shorter lemmas and appressed
pedicels.

This variety is known only from the type locality. The area is a roadside
swale containing limestone, at approximately 1,830 m. The vegetation is sparse
and the plants of var. appressa grow in small tufts scattered throughout the
area. Oaks and junipers are common on the margins of the swale.

The plants are a striking deep-red color and upon emergence from the sheath,
the panicles appear spikelike, resembling a Sporobolus or Muhlenhcrgia with
narrow, compact panicles. All orders of branching remain appressed for some
time following emergence, but even after the main branches spread, the pri-
mary branchlets and pedicels remain appressed, hence the derivation of the name.

Eragrostis intermedia Ilitchc. var. oreophila (L. II. Harvey) Witherspoon,
comb, et stat. no v.

Eragrostis oreophila L. II. Ilarvcy, Bull. Torrcy Bot. Club 81: 407. 1954. type: Mexico,
Hidalgo, Jacala, stony mountain sicle, 4,500 ft, 29 June 1939, V. //. Chase 7223^/-^ (US,
holotype; ARIZ, ClI, MICH, MO, TEX, isotypes).

1977J WITHERSrOON— LHAGROSriS

327

This variety occurs primarily on mountain slopes in central Mexico. It grows
in fairly deep, sandy-clay soils at elevations al)Ove 1,380 m.

It is not a common taxon in most of its range, but seems to be quite abundant
in the mountains around Jacala, Hidalgo. It is known from single collections in
both liaja California and Nuevo Leon. The wide separation of these collections
may reflect inadequate sampling but it may be an indication that the taxon has
arisen more than once. Variety intermedia is common in all these areas. How-
ever, the two taxa remain relatively distinct in these areas of sympatry. One
collection from Chihuahua (Gentry 8157, Sierra Charuco, MICII, UC, US) has
pubescent foliage not characteristic of other known specimens.

Eragrostis intermedia Hitchc. var. praetermissa (L. H. Harvey) Witherspoon,
comb, et stat. nov.

Eraf^w.stis praetermissa L. H. Harvey, Bull. Torrcy Bol. Club 81: 408. 1954. typi-:: Guatc-
iiiala, Dopt. Baja Vcrapaz, Santa Rosa, July 1887, //. von Tiircklieim 1292 (US, holo-

typc).

Era<i^rostis intermedia var. praetermissa occurs at high elevations in Mexico
and Central America. It grows primarily in the pine-oak zone in deep, loamy
soils.

It is not common in any part of its range but may be more prevalent than
herbarium material indicates. Since it grows in seemingly native vegetation,
most collectors would probably not encounter it if they botanize mostly along

roads and established trails.

This variety, although more common than \ar. oreophila^ is perhaps not as

distinct. Intermediates between var. intermedia and var. praetermissa have been
noted rarely, but var. praetermissa can usually be distinguished by its extra-
vaginal buds and papillose-pilose sheaths.

The following key will effectively separate the varieties of Erag^rostis inter-
media.

a. Buds c\tra\a^inal; at least the lower slieatlis deusely papillose-pilose; blades usually
flat, often exeeeding 5 mm wide; spikelets dark green, the lemmas with strong lateral

nerxes

i, rarely weak var. praetermissa

aa. Buds mostly intravaginal; sheaths glabrous except for a tuft of hairs at the apex or

with scattered hairs; blades mostly involute, usually less than 5 mm wide; spikelets

light green to i:)lTmibe()us, often reddish, the lenimas with weak to inconspicuous

lateral nerves, rarely strong.

b. Panicles decompound, witli secondary branchlets at least on the lower primary

branchlets, averaging jnore than 6 priuiary branchlets per branch; leaves few to

many, not crowded at the base; pedicels generally equal to or shorter than the

spikelets .__.. ....- var. intermedia

1)1). Panicles lacking secondary branchlets, usualK' with less than 4 primary branelv
lets per branch; leaves few, crowded at the base; pedicels generally longer than

the spikelets.
c. Branchlets and pedicels appressed to the main branches; lemmas less than

1.8 mm long - - - \ar. appressa

cc. Branchlets and pedicels ascending to spreading; lemmas greater than or

equal to 1.8 nun long var. oreophila

328 AWALS OF THE MISSOUHI BOTANICAL GARDEN [Vol. 64

Eragrostis hirta Fourn, var. longiramea (Swallen) Witherspoon, comb, et
stat. iiov.

Era^rodL- hn^iraniva Swallen, J. Wash. Acad. Sci. 21; 437. 1931, type: Mexico, Tainaiilipas
Sierra dc San Carlos, Pico del Diablo, vie, Mannolejo, 12 Aug. 1930, //. 7/. Bariktt

10910 (US, holotypci GH, UUl;' MICH, isotypes).

bold

900

generally more northern than typical var.
hirta. It is known only from the mountains of Tamanlipas, Nnevo Leon, and San
Luis Potosi in Mexico.

Variety hnfi^iramea is much larger than var. hirta, often approaching 2 m in
height. The two varieties are quite similar vegetatively, var. longiramea usually
having more flattened, longer and wider blades. However, they are easily dis-
tinguished by their panicles, as shown in the following key.

a. Panicle less than or etiual to 45 cm long and 10 cm wide, the branches less than or
equal to 10 cm long __ var. hirfn

aa. Panicle greater than 50 cm long and 18 cm wide, the branches great(T than 15 cm

long _____ _ „_. var. longiramea

Eragrostis trichocolea Hack. & Arech. var. floridana (Hitchc.) Witherspoon,
comb, et stat, nov.

Eragrostis floridana Hitchc, Amer. J. Bot. 2: 308. 1915. type: United States, Florida, dry
pine woods near Tampa, Oct. 1885, A. H. Ciuiis 3494 [US, holotype; BR, F, ISC, LE,
M, MO, NY (2), PH, TAES, TENN, US (2), isotypes].

Eragrostis purpusii Jedw., Bot. Arch. 5: 201. 1924. type: Mexico. Puebla, Cerro de Gavilan,
Aug. 1909, C. A. Ptirpus 4084 ( UC, lectotype, here designated; US, fragment of liolo-
type ex B). The holotype was destroyed in Berlin; the US fragment consists only of
a small branch with a few spikelcts.

Erafirostis tricJiocolea var. floridatia is found in sandy pinelands in coastal
and central Florida and in sandy ground near Huntsville, Texas. It also occurs
in sandy ground and sandy pine forests m central Mexico and Central America
at higher elevations.

This variety is distinguished by its smaller lemmas, more pubescent foliage
and its distribution.

Hitchcock described £. floridana in 1915 based on collections from Florida
and Orizaba, Mexico. However, after seeing material of £. trichocolea in Euro-
pean herbaria, he decided that E. floridana was cx)nspecific with that taxon. From
notes on the type specimen of E. floridana at US, he apparently made this
change in 1918 and the taxon was treated as £. trichocolea in both editions of
Manual of the Grasse.s of the United States (1935 and 1950). However, Hitch-
cock had not seen the holotype of E. trichocolea that is housed in Montevideo
(MVM). The "types" seen by Hitchcock and most other North American work-
ers were fragments of two Areehavaleta collections procured by Agnes Chase
in 1922 and placed in the U.S. National Herbarium. She obtained these frag-
ments from the Hackcl Herbarium in Vienna and the labels on those specimens
do not match the label on the holotype. Therefore, the fragments at US do not
represent types, although they are, in fact, E. trichocolea var. trichocolea.

''LIIH denotes the personal herbarium of Dr. L. H. Harvey, University of Montana.

1077] WlTHEnsrOON—i: KAGROSTIS

329

After examining the holotype and other South American material, I feel tliat
tlie North and Central American specimens are phenetically distinct at the vari-
etal level.

Eragrostis purpusii falls within the moiphological limits of E. tricJiocoIca var.
floridancL

The varieties of Eragrostis trichocolea may be distinguished by the following

key.

a. Lemmas greater tlian or e(iiial to 1.8 mm long and 1.4 mm wide, averaging o\er 2
mm in length; blades variously papillosc^-xiilosi- less than Ms their length; South Amer-
ican - var. trichocolea

aa. Lemmas less than or equal to L8 mm long and L3 nun wide, averaging loss than
L7 nnn in lengtli; blades densely papillose-pilose at least % their length; North and
Central American \dv. jhridana

REALIGNMENT OF THE SPECIES PLACED

IN EXOGONIUM (CONVOLVULACEAE)

I

Damkl F. AUSTIN-

Exo^i^omuDi has never been widely accepted as a genus although tliere have
been proponents of this rank since Choisy (1834, 1838, 1845) first described
the taxon. House (1908), Matuda (1963) and Standley & Williams (1970) have
been among the recent authors keeping the species as a separate genus. Others
have suggested that the species could better be ranked at some infrageneric level.
Grisebach (1864) reduced it to a section of Ipomoea^ while Meisner (1869)
considered the plants a subgenus.

Since the origin of the name Exogonium by Choisy (1834) 31 species have
been placed in the taxon, many authors varying the definition of the group
slightly. Although usually unstated, the major criteria for inclusion in the taxon
were red flowers, saberform corollas, and exserted stamens and stigmas. Floral
moiphology suggests that the species included in Exogonium are mostly adapted
for bird pollination; the species exhibit the characters classically associated
with this syndrome (van der Pijl, 1960, 1961; Meeuse, 1961; Percival, 1965;
Faegri & van der Fiji, 1971). However, a polyphyletic taxon has been created
because species from several lines have been lumped solely on the basis of a
common pollination system.

The following treatment is a revision of the binomials placed in Exogonium.
Several other related species are also included. Some of the nomenclatural and
biological problems within Ipomoea have been discussed elsewhere (Verd-
court, 1957, 1963; Austin, 1975a, 1975b).

Consideration of all morphological criteria indicates that the species pro-
posed for inclusion in Exogonium should be placed in the following taxonomic
groups.

Group 1. Exogonium vclutifolium House [Bull. Torrey Dot. Club 35: 100.
1908. TYPE: Mexico, Oaxaca, Nelson 1877 (GH, holotype; US, isotype)] is not
a member of the Convolvulaceae but the Acanthaceae. The correct name is
RucUia vclutifolia ( House) Wasshausen & Austin (Phytologia 25: 433-437.
1973 ) .

^ Tliis study was supported by grants from tli(^ Di\ision of Sponsored Research, Florida
Atlantic University, and a grant from the National Science Foundation througli tlie Smithson-
ian Institution ( Summer Institute in Systematics — V. Species Diversity.)

Spc^cimens in tlie herbaria at A, FAU, F, G, GH, IJ, K, L, LE, M, MEXU, MO, NY, US,
and W were examined chiring \'isits and on loan. Except where types are cited from other
lierbaria, material from tliese institutions formed tlie basis of the study. My tlianks are ex-
tended to curators and staff of the institutions cited. K. R. Robertson (Arnold Arboretum)
lias revised the genus Jacqucmontia and pn)vided especially useful comments. W. G. D'Arey
( Missouri Botanical Garden) criticized the original manuscript.

^ Department of Biological Sciences, -Florida Atlantic University, Boca Raton, Florida
33431.

Ann. Missouri Bot. Gahd. 64: 330-339, 1977.

1^>77] AUSTIN— REALIGNMENT OF EXOGONIUM

331

Group 2. Exogonium filiforme (Desr.) Choisy is the only member of the

/

Tlie small seeds and eapsule

which breaks into several sections at dehiscence, among other characteristics,
support the conclusion that this is a Jacquemontia (see Robertson, 1971).

Jacquemontia whnifolia (L.) Hall, f., Bot. Jahrb. Syst. 16: 542. 1893.

Basionym: Ipomoea solanifolia L., Sp. PI. 161. 1753. lectotype: Ipomoea
foliis cordatis Plumier, Cat. Pi. Amer., p. 3 in

Synonyms; Quamoclit soJanifolia ( L. ) Clioisy in DC, Prodr. 9: 335. 1845. Exogoninm
sohnifolium (L.) Britton, Mem. Brooklyn Bot. Card. 1: 82. 1918.

Ipomoea filifonim Jacq., Enum. Pi. Carib. 13. 1700; Sel. Stirp. Amcr. 27, pL 19. 1763.
lectotype: illustration by Jacquin, pL 19, 1763. Convolvulus filiformis (Jacq.) Dcsr. in
Lam., Encycl. Mcth. Bot. 3: 555. 1789. Exogonhnn filiforme (Jacq.) Choisy, Mem. Soc.
Phys. Geneve 8: 129. 1838.

Distribntion: Puerto Rieo, Guadalupe, Antigua, St. Barthelemy, St. Croix,
St. Thomas, Martinique, Tortola.

Group 3. House ineluded Exogonhtm racemosiim^ E, wrightii and E. ru-

(loIpJiii in his eoncept of tlie genus. These tlnee names have been reduced to
two species and placed in Turhina by O'Donell (1960) and Liogier (1968).
The accrescent sepals indicate tliat this is the proper disposition. A new com-
bination does need to be made.

1. Tiirbina racemosa (Poir. ) D. Austin, comb. nov.

Basionym: Ipomoea racemosa Poir. in Lam., Encycl. Meth. lk)t. Suppl. 4:
633. 1816, non RoUi, 1821, nee Griseb., 1866. tyim:: St. Domingo. Possibly in
LAM-P, not in the microfiche. Interpretation based on the protologue.

Synonyms: Convolvulus racemosus (Poir. ) Sprengcl, Syst, Veg. 1 : 600. 1825. Exo-
^oniutn racemosum (Poir.) Choisy, Mem. Soc. Phys. Geneve 8: 128. 1838.

Cahjstegia herterii Sprengel ex Hall, f., Bot. Jahrb. Syst. 16: 558. 1893, nom. nnJ., pro
syn.

Ipomoea hracteata Ruclol. e\ Ledeb. in Schrad., Neues J. Bot. 2: 292. 1807, non Cav.,
1799. TYPE: St. Domingo, Rudolphi (not at LE). Ipomoea rudolphii Roem. & Schnlt.,
Syst. Veg. 4: 222. 1819. Pharhitis hraeteata (Rudol. e\ Ledeb.) Clioisy in DC, Prodr. 9:
M4. 1845. mvea hracteata (Rudol. ex Ledeb.) Hall, f., Bot. Jahrb. Syst. 18: 158. 1894.
Turbina rudolphii (Roem. & Sehult.) O'Donell, Lilloa 30: 64. 1960.

Convolvulus altissimtis Sprengel, Syst. Veg. 1: 613. 1825. TvrK: Ilispaniola, Bertero
(MO, isotype). Ipomoea altissima (Sprengel) Bert, ex G. Don, Gen. Syst. 4: 273. 1838.

Distribution: Cuba, Haiti, Dominican Republic.

o

\

2. Turbina wrightii (House) Alain, Brittonia 20: 152. 1968.

Basionym: Exogonium wrightii House, Bull. Torrey Bot. Club 35: 99. 1908.
TYPE: Cuba, Wright 1650 (GH, holotype; MO, isotype).

Synonyms: Ipomoea wrightii (llonse) Alain, Mem. Soe. Cub. Ilist. Nat. "Felipe Poey"
22: 123. 1955.

Ipomoea racemosa sensu Griseb., Cat. Pi. Cub. 205. 1866, non Poir., 1816.

Distribution: Endemic to Cuba.

332 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Group 4, Throe binomials illustrated by Sesse & Moyifio have been placed
in Ipomoca sect. QuamocUt by Choisy (1845) and O'Donell (1959). Their
caudate sepals clearly indicate affinity with that species group.

1. Ipomoca hastv^era II.B.K., Nov. Gen. Sp. PL 3: 87, 1819; O'Donell, Lilloa
29: 42. 1959.

E.Vf^f^onhi/?* ciiniflorum Scssc & Mo<,'ino ex Choisy in DC, Prodr. 9: 336. 1845, noni.

pro syn.

2. Ipomoca twci (Sprengel)O'Donell, Lilloa 29: 69. 1959.

Exogouium unihcJIatton Moyino & Scsst' ex Choisy in DC, Prodr. 9: 336. 1845, nom.
pro syn.

3. Ipomoca ncci (Sprengel) O'Donell, Lilloa 29: 69. 1959.

Exog()uiu?7i corinil)()stn7i Sesse & Moyino ex 0*Donell, Lilloa 29: 71. 1959, noni. pro syn.

A fourth species was not inchided in O'Donell's discussion of the QuamocUt
group. The species is an Ipomoca; the flowers appear to be bird pollinated; and
the morphology indicates affinity with the QuamocUt group.

4. Ipomoca iihdeana (Fenzl, ex Ilall f.) D. Austin, comb. nov.

Basionym: Exogouium uhdcanum Fenzl. ex Hall, f., Bot. Jahrb. Syst. 16:
559. 1893. A ncw^ name for QuamocUt tuhulosa Mart. & Gal.

Synon>ins: QuomocJit tuhulosa Mart. & Cal., Bnll. Aead. Roy. Sei. Brnxelles 12: 270.
hS45. ivrt:; CMdlcotti 1393 (W, isotype). Ipomoca tuhulosa (Mart. & Gah) Ilemsl., Biol,
Centr. Anier., Bot. 2: 395. 1882, non Roeni. & Sehnlt., 1819.

Group 5. Ipomoca bractcata, the type species of Exogouium, shares many
characteristics with 7. purga and its allies. Only three of tlic nine species in the
alliance have been placed in Exogouium.

Ipomoca: Exogonium alliance.

Exogouium Choisy, Mcni. Soc. Phys. Geneve 6: 443, 1834.

Ipomoca suhyen. Exogouiut)i (Choisy) Meisn. in Mart., Fl. Bras. 7: 221. 1869.

Ipomoca seet. Batatas snliseet. Emcticac House, Ann. New York Aead. Sei. 18: 239. J908.

1. Ipomoca hractcata Cav., Icon. Descr. PL 5: 51, pJ. 477. 1799. xYrE: Cited
as Ipomoca ?hractcata by Cav.; based on two collections: Mexico, "Mazatlan
duabus Icucis" Scssc ir Mogiiio and "quattuor a Chipalcingo" Scsse ir Mo-
yino (probably MA, not seen). Interpretation based on plate 477 by Cavanilles.

S>non>ins: Ipomoca cincfa Roeni. & Sehnlt., Syst. Veg. 4: 254. 1819, nom. illeg. Exo-
gouitoti hractcatutu (Cav.) Choisy, Mom. Soe. Pliys. Geneve 6: 443. 1834.

Ipomoca spicata II.B.K., Now Gen. Sp. Pi. 3: 112. 1819. type; Mexico, Ilutuholdt i:r
Bonplauil (P, not seen, mierofiehe seen). Exogoniu))i spicatum (H.B.K.) Choisy, Mem. Soc.
Phys. Geneve 8: 128. 1838, nom. illeg.

Exogouium oJirac Bareena, Viaje Cav. Caeahnam. 29. 1874. type: Mexico, Cuernavaca.
Bdrceua (pres!imal)ly MKXU, not found). Interpretation based on the Biircena plate.

Convohuhis hractiflonis Sesse & Movino, Ph Nov. Hisp, 38. 1887; PI. Nov. Hisp. 22.
1893. rvPK: Mexico, Scssc 6- Mo^'itlo (probably MA), Their leones 207 is cited in the 1893
pnblication, but I have not been able to obtain a copy of this. The interpretation used here
is based on the protologue.

1977] AUSTIX— REALIGXMEXT OF EXOCOXIUM 333

Ipomoca hracteata var. puhcscens Rol)inson & Greoninan, Amer. J. Sci. 50: 160. 1895.
type; Mexico, Jalisco, Pringle 4734 (MO, holotypc). Exogouiurn hwcteaftim var. puhcsccns
(Robinson & Greennian) Honsc, Bnll. Torrcy Bot. Chih 35: 101. 1908.

Distribution: Baja California, Jalisco, Sinaloa, Oaxaca, Tepic, Morclos, So-
nora, Chihuahua, Michoacan, and Cuerrero (Mexico).

2. Ipomoea dumosa (Benth.) L, (). Williams, Fieldiana, Bot 32: 190. 1970.
Basionyni: Exogonium dumosum Benth., PL llartw. 46. 1840. lYrK: Uart-

iceg s.n, (K, holotype, not seen; F, photo).
Distribution: Mexico.

This species is very close to I, purga and has been considered synonymous
with that population ])y some.

3. Tpomoea elongala Choisy in DC, Prodr. 9: 355. 1845. type: Mexico, Oaxaca,
Andrieux212 (G-DC, holotype; US, photo).

Distribution: Mexico. See Matuda, Anales Inst. Biol. Univ. Nac. Mexico 35:

75, 1964.

4. Ipomoea emetica Choisy in DC, Prodr. 9: 376. 1845. type: Mexico. Based
on an unpublished plate by Sesse & Moyino (not found).

Synonyms: Ipomoca sugiltata Sessc & Mo^ifio e\ Choisy in DC, Prodr, 9: 376. 1S45,
nom. pro syn., non Poir., 1791.

Ipomoca caudata Fcrnald, Proc. Anicr. Acad. Arts 36: 498. 1901. type: Mexico, Morclos,
rringlc 8448 (Gil, liolotypc).

Distribution: Mexico. See Matuda, Anales Inst. Biol. Univ. Nac. Mexico
36: 85. 1965.

5. Ipomoea hintonii L. O. Williams, Ecou. l^ot. 24: 400. 1970. type: Mexico,
Ilinton et al 8474 ( F, holotype ) .
Distribution : Mexico.

6. Ipomoea purga (Wender. ) Ilayne, Arzneigewachse 12: tal). 33, 34. 1833.
Basionyni: Convolvulus purga Wender., Pharm. Central-Blatt 1: 457. 1830.
type; Based on plants grown from seed collected in Mexico by Schiede, prob-
ably not preserved. Interpretation based on Hayne plates.

Synonyms; Exogoniu7n jJitr^a (WVndcr. ) Bcntli., Pi. Hartw. 46. 1840.

Ipomoca scliiedcami Zncc, Flora 14: 801. 1831. TvrK; Based on plants collected in
Mexico by Schiede and cnltivated by Zuccarini. Proba1)ly based on the same Schiede col-
lections as /. ]mrga.

Distribution: Mexico, Cuatemala, El Sahador, Honduras, Costa Rica, Pan-
ama.

7. Ipomoea seducta House, Ann. New York Acad. Sci. 18: 241. 1908. type:
Guatemala, Alta Verapaz, Tuerckheim 7926 (Gil, US, isotypes).

Distribution: Mexico, Guatemala, Sec Standley & Williams, Fieldiana, Bot.
24 (9): 51. 1970.

8. Ipomoea suffulla (II.B.K.) G. Don, Gen. Syst. 4: 276. 1838.

Basionym: Convolvulus stiffultus II.B.K., Nov. Gen. Sp. Pi. 3: 102, ;>/. 211.

234 ANXALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

1819. typk: Mexico, Volcan de Jonillo, Ilumbohlt 6- Bonphnd (P, not seen,
microfiche seen).

9. Ipomoea urlnnei House, Miihlenbergia 3: 41, pi 2. fig. /;. 1907. type: Mex-
ico, Volcan de Coliina, Bdrcena 214 (presumably MEXU, not found).

Distribution: Known only from the type collection. See Matuda, Analcs

Inst. Biol. Univ. Nac. Mexico 35: 67. 1964.

Group 6. The species in this group form a relatively homogeneous assem-
blage within Eriospermum which is normally recognized as a section or sub-
genus oi Ipomoea (Verdcourt, 1963; Austin 1975a). Comose seeds separate these
species from the others that have been placed in Exogonium. Some justifica-
tion for the separation of this group from Eriospermum could be made. Most
of the species listed here are adapted for bird pollination while others in EMo-
spermum are bee pollinated. If the two groups were placed in different taxa,
several sp(x^ies-pairs (e.g., Z. eggersii/L steudelii and I. viridiflora/L Carolina)
would be separated. While closely related, one species of these pairs conforms
to the bee-pollination syndrome (/. eggersii), and the other to the bird-pollina-
tion syndrome (/. steudelii). If this aUiance were separated from Eriospennum,
the closely related species of the pairs would be placed in separate taxa.

Ipomoea miramlina/L microdactijla alliance,

Exof^onium Clioisy, Mem. Soc. Phys. Gcncxc 6: 443. 1834, in part, excl. type species.
Ipomoea sect. Exogonium (Choisy) Griseb., Fl. Brit. W. I. 472. 1862, excl. type species.

1. Ipomoea argentifolia A. Rich, in Sagra, Hist. Cuba 11: 131. 1850. type:
Isle of Pines, Richard (P, not seen),

Synonvius: Exogonium argcnlifolium ( A. Rich. ) House, Bull. Torrey Bot, Club 35:

102. 1908,'

Ipomoea praeeox Wriglit in Sauv., Fl. Cubana 107. 1873; Anales Acad. Ci. Med. Habana

7:46. 1870. wvi:: VVV/g/i? 3644 ( USJsotype).

Distribution: Cuba, Isle of Pines, Mexico (Oaxaca, Puebla).

Although I have not seen the type of Richard's species, it is the only popula-
tion on the Isle of Pines matching the protologue. The distribution of this spe-
cies is worthy of note in that the plants are disjunct from western Cuba to the
western slopes of the Sierra Madre Occidental in Mexico, The species is prob-
ably native to Cuba and not introduced since the allied species Ipomoea lachnea
Sprengel {Bertero, MO, isotype) occurs in the Dominican Republic.

2. Ipomoea Carolina L., Sp. PL 160. 1753. type: Based on illustration in Catesby,
Nat. Hist Carolina 2: 9, tab. 91 (lectotype).

Synonyms: Exogouiu7n pedatum Cboisy, Mem. Soc. Pliys. Geneve 8: 130. 1838. type:
Santo Domingo, Poiteau (G-DEL, not seen).

?Ipi)moca clausa Rudolphi ex. Ledeb. & Adlerstam, Pi. Doming. 14. 1805; Ledelj.,
NeuesJ. Hot. 2: 292. 1807. type: fliir/o//^/i/ coll.? (not in LE ).

Distribution: Haiti, Dominican Republic, Bahama Islands.

The synonymy listed here is based entirely on the descriptions since the types
have not been seen. The protologue of 7. clausa is general enough that one can-
not be sure of the species. Some have thought that this name applies to 7. triloba

l'J'^"J AUSTIN— RKALIGXMENT OF EXOCOMUM

335

(Matilda, 1965: 100). Perhaps 7. clausa does apply to that species, but the de-
scription appears to me to bettcT fit the population treated here as 7. Carolina.
While I was in Geneva I looked for the type of E. pedattim but did not find
it. Additional searching will perhaps solve that problem of synonymy.

3. Ipomoea conzattii Greenman, Publ. Field Golumbian Mus., Bot. Ser. 2: 258.
1907. tyi'e: Mexico, Conzatti 1666 (F, holotype).

Synonym: Exo^oniiim conzattii (Crccnnian) Honsc, ]?ull. Torrey Bot. Club 35: 102. 1908.

Distribution: Mexico (Guerrero).

4. Ipomoea concolora (Matuda) D. Austin, comb. nov.

liasionym: Exogonium concolonim Matuda, Anales Inst. Biol. Univ. Nac.
Mexico 36: 116. 1965(1966). tyi-e: Kriise 844 (MEXU, holotype).

5. Ipomoea claremis Alain, Mem. Soc. Cub. J list. Nat. "Felipe Poey" 22: 121.

1955. tyi'e: Cuba, Leon I- Roca 7.95,9 (NY, holotype, not found). Although the

type was not found, there is a sp(x-imen [Howard 6565 (US)] annotated by
Alain.

Distribution: Endemic to Cuba. This species is very similar to 7. micro-
dactyla. The main difference is the flower color: red in 7. microdactyla and
white in 7. clarensis.

6. Ipomoea cuhensh (House) Urban, Symb. Antil. 9: 427. 1925.
Basionym: Exogonmm cu])L'n.se House, Bull. Torrey Bot. Club 35: 105. 1908.

TYI'E: Britton ir Shaffer 495 (NY, holotype).

7. Ipomoea desromseaiixii Steud., Nom. Bot., ed. 2. 816. 1841. type: Based
on Convolvulus eriospermus Desr.

liasionym: Convolvulus eriospermus Desr. in Lam., Encycl. Meth. Bot. 3:
567. 1789. TYI'E: probably in P-LAM, not seen.

Synonyms: Exogonitim eriospcnuum (Desr.) Choisy, Mem. Soc. Pliys. Ccnt-ve 8: 180.
18;38. Ipomoea eriospcrma (Desr.) Ruf., Fl. Tell. 4: 74. 1838, non Beaiiv., 1807.

The specimen in the DeCandolle herbarium [Santo Domingo, Bertero s.n.
(G-DC)] matches the description of Desrousseaux and the synonymy is based
on that specimen.

Distribution: Santo Domingo.

8. Ipomoea eggersii (House) D. Austin, comb. nov.

Basionym: Exogonium eggersii House, Bull. Torrey Bot. Club 35: 104. 1908.
type: St. Thomas, Feb. 1887, Eggers (NY, holotype, not found; G, L, isotypes).
Distribution: St. Thomas, Tortola.

This species has been confused with 7. steudelii. Differences between them
are few; the major distinction is that 7. eggersii has white bee-pollinated flowers
and 7. steudelii has red bird-pollinated flowers.

9. Ipomoea fuchsioides Griseb., Cat. PI. Cub. 205. 1886. type: Cuba, Wriglit
3095(MO, isotype).

ty

336

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Synonyms: Exoiioiuum ftichsioidcs (Griscb.) House, Bull. Torrey Bot. Club 35: 101.

1908.

Distribution: Endemic to Cuba.

10. Ipomoea inccrta (Britton) Urban, Symb. Antil. 9: 247. 1924.
Basionym: Exogonium incertum Britton, Mem. Torrey Bot. Club 16: 94.

1920. tyi-k: Culia, SJwfer 1235 (NY, holotype).
Distribution: Endemic to Cuba.

11. Ipomoea juhpohJes Griscb., Cat. PI. Cub. 202. 1886. tyve: Cuba, Wright
3097 (MO, isotype).

Synonym: Exogonium jdlapoUh's ( Crisrb. ) House, Bull. Torrey Bot. Club 35: 101. 1908.

Distril)ution: Endemic to Cuba.

12. Ipomoea longhtaminea O'Donell, Lilloa 23: 488. 1950. type: Brasil,
Bahia, Rose 6 Rm.seU 19784 ( US, holotype ) .

Distribution: Eudeuiic to Brasil and apparently to the state of Bahia.

13. Ipomoea leucomura Urban, Syuib. Antil. 3: 350. 1902. syntypes: Haiti,
Ehrenherg 134 (US, fragment), Picarda 16 (US, fragment), Biich 5 (not seen).

Synonym: Exogonitint h'ticoiwiirum (Urban) House, Bull. Torrey Dot. Club 35: 106.
1 908 .

Distribution: Endemic to Haiti.

14. Ipomoea miewdaetyla Criseb., Cat. PI. Cub. 204. 1886. type: Cuba, Wriglit
3094 (MO, isotype).

Synonyms: Exugonhim micwdcictylum (Criseb.) House, Bull. Torrey Bot. Club ;35: 102.

1908.

Exogonium micnnJacfyJtim var. intcgrifolium Ilonse, Bull. Torrey Bot. Club 35: 103.

1908. TYiM-:: Cuba, Wn^'/i/ 3/02 (MO, i.sotype).

llHxnora repmuUi sensu Criseb., Cat. Pi. Cub. 204. 1886, non Jaeq., 1760.

Distribution: Florida, Bahamas, Cuba. These vines are found only in the
rocky pinelands of Dade and Monroe counties in Florida. In the Bahamas
(Inagua) the Bahama Woodstar hummingbird visits and pollinates the flowers.

15. Ipomoea mirandimi (Pitticr) O'Donell, Lilloa 26: 370. 1953.
Basionym: Exogoniwn miramUmim Pittier, J. Wa.sh. Acad. Sei. 21: 143. 1931.

type: Venezuela, Fittier 12217 (YEN, holotype; US, isotype).

Distribution: Known from Venezuela and Panama; undoubtedly in Co-
lombia also but no spcciuiens .seen.

16. Ipomoea repanda Jacq., Enum. PI. Carib. 13. 1760; Sel. Stirp. Amer. 28,
pi 20. 1763, non Criseb., 1886. lectotype: illustration by Jacquin, pi. 20. 1763.

Synonyms: Convolvulus rcpomhis (Jaeq.) Desr. in Lam., Encycl. Meth. Bot. 3: 555. 1789.
ExogouiuDi iviHiiiduni (Jaeq.) Cboisy, Mem. Soe. Pbys. Geneve 8: 128. 1838.

Distribution: Puerto Rico, Tortola, Cuba, Barbuda, Antigua, Martiniciue,
Dominica, Cuadeloupe, St. Vincent, Montscrrat, St. Lucia, St. Jan.

1977] AUSTIN— REALIGNMENT OF EXOGONIUM 30

i

17. iDomoea

D. Austin, comb. nov.

Basionym: Exogonium retropUosum Pittier, J. Wash. Acad. Sci. 21: 143.
1931. type: Venezuela, Merida, Pittier 12698 (VEN, holotype; MO, US, isotypes).
Distribution: Endemic to the coastal mountains of Venezuela.

18. Ipomoea shinnersii D. Austin, nom. nov.

Basionym: Exogonium hiteum House, Bull. Torrey Bot. Club 35: 103. 1908.
type: Mexico, Conzatti I- Gonzalez 668 (GH, holotype; NY, isotype).

Because of Ipomoea hitea Ilemsley (Diagn. PI. Nov. 34, tab. 60, 1878) a
new name is required for House's species. The new name commemorates the
late Lloyd Shinners, a student of Convolvulaceae.

Distribution : Mexico. ( Guerrero ) .

19. Ipomoea shinnersii var. woronovii (Standley) D. Austin, comb, et stat. nov.

Basionym: Exogonium woronovii Standley, Field Mus. Nat. Hist., Bot. Ser.

11:171. 1932. type: Mexico, Woranoir 2906* (F, holotype).
Distribution : Mexico ( Michoacan ) .

20, Ipomoea signata House, Muhlenbergia 3: 46. 1907. type: Guatemala,
Nelson 3595 ( US, holotype ) .

Distribution: Mexico, Guatemala, Venezuela. See Matuda, Anales Inst.
Biol. Univ. Nac. Mexico 35: 72. 1964; Standley & Williams, Fieldiana, Bot. 24
(9): 53. 1970. The distribution of this species is unusual in that it appears to
be absent from a large part of Central America and reappears in the coastal
mountains of Venezuela.

21. Ipomoea sieudelii Millsp., Publ. Field Columbian Mus. Nat. Hist., Bot. Ser.
2: 86. 1901. type: Based on Ipomoea arenaria (Choisy) Steud.

Synonyms: Exogonium arcnarium Clioisy, Mem. Soc. Pliys. Geneve 8: 129, /;/. Z. 1838.
type: Puerto Rico, Bertero (C-DC, lectotype). Ijwtnoca arenaria (Choisv) Steucl, Nom.
Bot., ed. 2. 815. 1841, non Rocm. & Sclmlt., 1819.

Ipomoea eggersiana Peter in Enj^rler & Prantl, Nat. Pflanzcnfam. IV (^a): 30. 1891, nom.
niul.

Distribution: Puerto Rico, St. Croix, Virgin Corda, St. Thomas, St. John.

Choisy based Exogonium arenarium on four collections. These collections
came from Puerto Rico, St. Thomas, Santo Domingo, and the Bahamas. Accord-
ing to the interpretation that has been used for the past 60 to 70 years, the species
does not occur on either the island of Hispaniola or in the Bahamas. Therefore,
neither of the collections cited by Choisy from these islands should be chosen
as the type. The specimen in Geneva matches the concept of historic use and has
been chosen to be the lectotype.

22. Ipomoea avicola D. Austin, nom. nov.

Basionym: Exogonium verrucuhsum Pittier, J. Wash. Acad. Sci. 21: 142.
1931. TYPE: VcnezAiela, Aragua, Pittier 12118 (YEN, holotype; G, US, NY,
isotypes).

Synomin: Jpomoea verruculosa (IMttier) O'Donell, Lilloa 26: 379. 1953, non 7. verm-
culosa Mart, ex Choisy in DC, Prodr. 9: 378. 1845, nom. i^ru s>n.

338

AXNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

Distribution: Endemic to the coastal mountains of Venezuela.

This and the Brasilian /. Jon<i,istaminea O'Donell are similar. These two
populations apparently represent local endemic bird-pollinated flowers derived
independently.

23. Tpomoca viridiflora Urban, Symb. Antil. 3: 348. 1902. type: Elirenherg
345(USJsotype).

Synonym: Exo^otwtm viridiflonnn (Urban) House, Bull. Torrey Bot. Club 35: 106. 1908.

Distribntion: Cuba, Hispaniola.

Group 7. These two species belong to different alliances within the Erio-
spermum group of Ipomoea. Flowers on the plants are apparently bee pollinated
and the inclusion of tliese species in Exogonium appears anomalous.

1. Ipomoea argentea Meisn. in Mart., Fl. Bras. 7: 247. 1869. syxtyi'es: Brasil,
Gardner 33S6 {wot ^cQw) . Venezuela, Spn/cc? 3605 (K).

Synonyms: Ipomoea comosa House, Ann. New York Acad. Sci. 18: 201. 1908. type:
Basod on I. vilhmi (Clu)isy) Meisn. Batatas villosa Choisy in DC, Prodr. 9: 337. 1845.
Tvri:: Brasil. Martins 609 (M, syntype). Ipomoea villosa (Clioisy) Meisn. in Mart., Fl. Bras.
7: 244. 1869, non Ruiz & Pavon, 1799. Exo^ouium villosum (Choisy) Peter in Engler &
Prantl, Nat. Pflanzenfani. IV (:3a): 28. 1891.

Distribution: Known from savannas in Venezuela, south to Brasil. The plants
show considerable variation throughout the range, prompting the division of the

1960

same.

2. Ipomoea steerei (Standley) L. O. Williams, Fieldiana, Bot. 32: 195. 1970.

Basionym: Exogonium steerei Standley, Carnegie Inst. Wash. 461: 83. 1935.
type; Steere 1545 (F, lectotype), Stecre 1599 (F, syntype).

Distril)ution: Reported from Mexico and Guatemala.

LiTKnATUHE Cited

Austin, D. F. 1975a. Typification of the New World subdivisions of Ipomoea. Taxon 24:

107-110.

. 1975b. Convolvulaceae. In R. E. Woodson & R. W. Scliery, Flora of Panama. Ann.

Missouri Bot. Card. 62: 157-224.

Choisy, J. D. 1834. Convolvulaceae Orientales. Mem. Soe. Pbys. Geneve 6: 385-502.
. 1838. De Ccmvolvulaeeis Dissertatio Seeunda. Mem. Soc. Phys. Geneve 8: 121-

164, tab, Z-7V.

. 1845. Convolvulaceae. In A. de Candolle, Prodromus Systematis Naturalis Regni

Veffctabilis. Vol. 9: 323-465.
Fakciu, K. & L. VAN OEH PijL. 1971. The Principles of Pollination Ecology. Ed. 2. Per-

gamon Press, New York.
GiusKUAcii, A. H. R. 1864. Flora of the Britisli West Indian Islands. Lovell Reeve & Co.,

London.
HousK, H. I). 1908. Studies in the North American Convolvulaceae — IV. The Genus

Exogonium. Bull. Torrey Bot. Club 35: 97-107, /;/. 1-2.
LioGiEu, A. H. 1968. Novitates Antillanae. 111. Brit(mnia 20: 152.
Matuda, E. 1963 (1964). El genero Ipomoea en Mexico — I. Anales Inst. Biol. Univ.

NaJ. Mexico 34: 85-145.
. 1964 (1965). El genero Ipomoea en Mexico (II). Anales Inst. Biol. Univ. Nac,

Mexico 35: 45-76.

1-^77] AUSTIN— REALIGNMENT OF EXOGONIUM

339

. 1965 (1966). El gcnero Ipomoea en Mexico (III). Anales Inst. Biol Univ Nac

Mexico 36: 83-106.

Mekuse, B. J. D. 1961. The Story of Pollination. Ronald Press, New York.
Meisneh, C. F. 1869. Convolvulaceae. In K. F. P. von Martins, Flora Brasilicnsis Vol
7: 200-424, Leipzig.

O'DoNELL, C. A. 1959. Las especies aniericanas de Ipomoea L. sect Quamocht (Moench)
Griseb. Lilloa29: 19-86.

. 1960. Notas sobre Convolvulaceas aniericanas. Lilloa 30: 39-69.

Pkhcival, M. S. 1965. Floral Biology. Perganion Press, London.

Pjjl, L. van deh. 1960. Ecological aspects of flower evolution I. Phyletic evolution
Evolution 14: 416-463.

1961. Ecological aspects of flower evolution II. Zoopbilous flower classes. Evo-

lution 15: 44-59.

RoHEinsox, K. R. 1971. A revision of tlie genus Jacqiwmontia (Convolvulaceae) in North
and Central America and the West Indies. Ph.D. dissertation, \\'ashington University,
St. Louis. 285 pp.

Staxdley, p. C. & L. O. Williams, 1970. Convolvulaceae. In Flora of Guatemala
Fieldiana, Bot. 24(9): 4-85.

Vkhdcouht, B. 1957. Typification of the subdivisions of Ipomoea L. (Convolvulaceae)
with particular regard to the East African species. Taxon 6: 150-152.

1963. Convolvulaceae. In C. E. Hubbard & E. Milne-Redhead (editors), Flora

of Tropical East Africa. Crown Agents for 0\'ersea Governments and Adnu'uist'rations,
London.

THE LUPINUS MONTANUS COMPLEX OF MEXICO

AND CENTRAL AMERICA'

David B. Dunn- and William E. Harmon''

Abstract

The recognition of the Lupiuus monfanus complex by morpliological traits is discussed.
Ecological modification of traits is discussed and the island nature of distribution from uioim-
tain peak to mountain peak produces semi-isolated gene pools. Long range dispersal and uitro-
gression from other lupines has occurred at the northern end of the distribution in San Luis
Potosi, Mexico, de\eloping L. caciitnhius. A similar situation occurred in Costa Rica, with
L. vuicnoi the product of introgression from, as yet, an unknown taxon. In Guatemala \ar.
austrovolcanicus represents local introgression from L. kellennunianus, into L. montamis. Both
of the Peru\ian (L. pracstdhilis and L. procuhustrinus) taxa are, likewise the result of long
range dispersal and introgression. The geographic range of each of the taxa of the complex
is plotted and the interrelationship is discussed. The alkaloids have been plotted from random
samples of each of the taxa and the data supports the taxonomic treatment and interpretation
of their interrelationship.

The lupines of Mexico liave never been studied monogruphically. Previous
studies have been t'loristic for states or regions or miscellaneous descriptions, as
contrilnitions to the flora of Mexico. To avoid further nonienclatural complica-
tions, the earhest named taxa should be identified first. In this sense, the first
taxon named for Mexico was Lupmiis mexicanus Cerv. ex. Lag. (1816), which
has been identified (Dinm, 1972). Lupinus monttmus H.B.K. (1823) was tlie
second epithet published for Mexico, concurrently with L. elegans H.B.K.
(Humboldt et ah, 1823: 478). Both of the types of these taxa are available at
Paris, France, with microfiche illustrations now widely distributed. The topo-
type material was studied and dissections of 50 collections, representing the
geographic range of L. nwntanm were made, and the mean measurements were
used to prepare the illustrations of L. montanus and allies presented in this
paper. The illustration of L. montanus was sent to Paris and the curator of the
herbarium kindly varified that the illustration accurately represents the species
by matching it with the type specimen. Since L. elegans H.B.K. is the first
epithet in a different complex of lupines, it will be treated, as soon as the rest
of the complex is understood. With this approach it is believed, after ten years
of study of the Mexican lupines and dissection of over 100 types for Mexico, that
the taxonomic treatment of the L. montanus complex for Mexico and Central
America can be presented. C. P. Smith (1948: 608) reported two South Aineri-

1 The authors wish to express appreciation to the multiple curators of herbaria who loaned
material for the study, as cited in the distributions b) the code letters from Index Ucrhanorum
(Holmgren & Keuken, 1974). Special appreciation is expressed to the curators at Paris for
comparing the illustration prepared for L. numtanus witli the type specimen. Two additional
herbaria are cited which were not in the code. GUN = University of Northern Colorado,
Greeley, Colorado, U.S.A., and WUP = Wiscxmsin University-Platteville, Platteville, Wiscon-
sin, U.S.A.

= Division of Biological Sciences, University of Missouri-Columbia, Columbia, Missouri

" ]>partment of Biology, University of Northern Colorado, Greeley, Colorado 80G3L
Ann. Missouiu Bot. Gauo. 64: 340-365. 1977.

1977] DUNN & HAWMOS—LVPINVS

341

can species to belong to what lie interpreted as the ''Lupinus montanus com-
plex." Only one of these was available for study. It is a member of tlie complex
and is very distinctive in multiple characteristics. Both are only treated in
the key.

Morphological Rlcogxition

Lupinus montanus, per se, can be readily identified by the large sheathing
stipules, the largest per specimen being 3-33 cm long. The infraspecific taxa
show various modifications of the shape, size, texture, and vestiture of the stip-
ules. (The allies are so different that they cannot be recognized by the stipules.)
The stems, 0.5-2.3 m tall (to 4 m shrubs in Peru), are clustered from a w^oody
caudex, and in age become short woody trunks, to 5 cm diameter. The current
year's growth is normally hollow and fistulose or subfistulose, varying from 5-
30 mm in diameter. (Allies and products of introgression have w^oody, solid
stems, as small as 3 mm in diameter, and are only 3-7 dm tall.) The leaves
show no general relationship other than a tendency to have many (9-17) leaf-
lets, palmately compound, linear-elliptic to oblanceolate and generally glabrous
above, but the last trait is modified by increasing hairiness above, with higher
elevations and by introgression from the allies. The flower structure shows the
greatest degree of uniformity, with only very subtle changes in size, shape and
vestiture of the calyx, and size and position of the bracteoles. Keels are gen-
erally glabrous but both infraspecific taxa and the allies may have ciliate keels.
The base of the deep sulcus of the banner appears to have a nectary, which is
uncommon in Lupinus. The number of ovules varies from taxon to taxon, with
L. montanus and the infraspecific taxa varying from 7-13 per pod (the allies
less). The seeds of L. montanus and the infraspecific taxa are very similar in
shape to those of L. pohjphyllus, a species distributed from southern California
to British Columbia, with which it has been confused. The seeds are generally
4.5 mm long and 3 mm wide, with a deep funicular pit at the side of one end.

While the large stipules represent tlie most distinctive trait of the L. mon-
tanus complex, they are very small in some of the allies and reduced in taxa
considered to be products of introgression. The large flowers with the banners
reflexing near the midpoint are perhaps tlie most consistent character of the
complex. They may indicate the utilization of specific pollinators.

Habitat axu Ecolocical Moduications

The taxa of the complex are associated with the upper forest openings extend-
ing upward to timberline. Thus the population on each mountain represents a
breeding population with a chance for some genetic drift or fixation. This is
reflected by variations in the hair type from mountain to mountain. However,
to view the population on each mountain as having achieved some taxonomic
status would be an exaggeration. It appears probable that migratory birds con-
tribute to the separated populations of each mountain peak intermittently. There

is also an altitudinal modificatioti of the density of pubescence of both the stems

and leaflets. On Popocatepetl the smalU^st specimens were at the lowest altitude

342 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 01

and the area above tiinberline. Tlie largest specimens were witliin the upper
margins of timber. Tlie densest pubescenee was at the highest elevation and the
sparsest pubescence was at the lower elevation, with the leaflets glabrous above
at the lower elevations. On Nevada de Toluca, the type locality, the situation
was similar, Sharma (1967, in an unpublished part of his thesis) demonstrated
experimentally that hair frequency increased with aridity and heat. The vege-

3-2

while those within L. montanus vary from 0.5-2.5 m tall.

Evolution and Geochaphical Relationships (Fig. 1)

Lupinws muelleri and L. keUermanianus, treated here as allies, can be uti-
lized to illustrate both the islandlike mountain isolation and long range disjunc-
tion of taxa with morphological similarities. The two are very closely related,
but L, kellertminianus is known only from two volcanic peaks in Guatemala,
while L. mucIIeri is known only from the opposite end of the distributional
range of the complex on Cerro Potosi, Nuevo Leon, Mexico. Vegetatively they
are very different from the rest of the L. montanus complex, having woody
stems, low (3-5 dm) stature, only 2-2.5 cm long petioles, only 2-2.5 cm long
leaflets with strigose-sericeous pubescence on both sides, and only 8-20 mm
long stipules. The flowers, however, are very similar to L. montanus except
that both are pubescent on the back of the banner. The A^egetative traits, except
pubescence, resemble L. ar<ienteus of the Rocky Mountains of the United States
but the closest geographic approach of this species is in northern New Mexico,
while L. muelleri occurs in southern Nuevo Leon and disjunctly in Guatemala,
where L. hellermanianus also occurs. The pubescence of the leaflets and the
banner resembles that of the L. sericeus complex, but the closest approach of
this complex is in northern Arizona, north of the Grand Canyon. If the source
of the characteristics is from the above two complexes of the United States, then
long range dispersal seems the only plausible explanation, via migratory seed-
eating birds. Lupinus muelleri is on a mountain ridge surrounded by Chihua-
huan desert, but inhabits the lower pine zone. Lupinus kellermanianus was
collected high on Volcan Agua near timberline, hence both are in somewhat

xeric situations where abundant pubescence is adaptive.

Introgressive Hyhridization

Lupinus cacuminus is vegetatively intermediate between L, muelleri, with
which it has geographic proximity, and L. montanus, from which it is geograph-
ically completely isolated. Lupinus cacuminus has intermediate stipules, leaf-
lets, and petioles, the fistulose stems of L, montanus, but the pubescent banner
of L. vwelleri, and extends above timberline, an ecological trait of L. nwntanus.
In some populations of L. cacuminus an intermediate amount of ciliation occurs
on the keel, a trait derived from L. muelleri. Tlie multiple. collections are very
similar and appear to be a stabilized entity, derived from the hybridization of
the L. montanus genome with the L. muelleri genome. The alkaloid spectrum
of L. cacuminus is identical with that of L. montanus as shown below.

1977]

DUNN & HARMON— LUP/AUS

343

Fk;ure 1. Distribution of the taxa of the Lupimis montamis complex and their allies
in Mexico and Central America.

Lupinus montanus subsp. montanus var. auslrovolcanicus may l)c the i^rod-
uct of introgression between ancestral L. montamis and L. kellernuinianus. In
this case L, keUermaniamis could liave provided the woody stem, reduced stat-
ure, and intermediate vegetative traits. Both L. keUernuinianus and L. mon-
tanus var. auslrovolcanicus are limited to one or two mountain peaks and liave

only been collected a few times. It is thus questionable whether the process of

introgression has progressed long enough to have established a stabilized taxon
of intermediate appearance.

Another possible example of a long range introduction of some portion of
the L. montanus genome into a lupine population is provided by L. valerioi on
Cerro Vueltas in Costa Rica. There is a long distance between this population
and the nearest known material of L. montanus in Guatemala. In this case the
stipules are intermediate in size, somewhat similar to those of L. cacumirms.
The bracts, however, are quite large and broad, typical of L. montanus. The
stems are very slender, similar to those of L. keUernuinianus, but the petioles
are long, 6-12 cm, and have spreading pilose hairs to 4 mm long. These traits
all suggest a uu'xing of genetic traits of L. montanus with some, as yet undeter-
mined taxon of Lupinus. None presently known from Costa Rica provides

these characteristics.

A fourth example of long range introduction of genetic material from the

344

ANNALS OF THE MISSOURI BOTANICAL GARDLN [Vol. 64

L. montatms genome is provided by L. proculaustrinus C. P. Smith of Peru. In
this case the plants have retained many of the traits of L. montanus, inchiding
tlic large sheathing stipules and large broad bracts. Tlie species has multiple
unique traits, however, which are not present in L. montanus. A conspicuous
one is the glabrous, glaucous surface and also a shrubby stature, reported on
some specimens to be up to 4 meters in height. The other Peruvian species,
L. praestahilis C. P. Smith has not been available for study.

Alkaloid Ciiemlstry

The material utilized was dried leaflets in all cases, since we have observed
some cases where the alkaloids stored in the seeds were different from those
present in the leaves. Seed material is not always present. Four to five leaflets
were fragmented in a new coin envelope for each sample and transferred to a
test tube. In species with large leaflets tjuly one is necessary. Enough 30%
KOH was added to wet the leaf fragments. Enough chloroform was added to
cover the fragments. The rack with the test tubes was then stored in a refrig-
erator for one day (24 h). A micropipette was utilized to spot 50 fA of the
clear chloroform solution from the bottom of the test tube, for each sample,
onto a thin-layer chromatographic plate (TLC). If no clear bottom layer was
present, a few drops of chloroform was added to the test tube. The soKent
utilized to separate the alkaloids was 95 parts chloroform, 4 parts anhydrous
methyl alcohol, and 1 part ammonium hydroxide, by capillary flow, against
gravity, in a Brinkman tank. The flow was stopped at 15 cm and the plate
dried and developed first with Dragcndorf's reagent. All visible spots were
marked with a pencil. The plate was then sprayed with iodaplatinate to bring
out any trace substances not observed with the first stain. The Rt values plotted
in Table 1 have been correlated with standards supplied by Cho & Martin
(1971) for sparteine, lupanine, hydroxylupanine and cytisene, on each plate that

was prepared.

While the number of samples plotted is not large, it is quite clear that while
some variation occms in the trace alkaloids, the presence of the principal alka-
loids is fairly consistent. Within L. montanus both varieties retain the same
principal alkaloids, while the two subspecies show distinct alterations in the
principal alkaloids. The suggestion that L. keJler mania nus, L. valerioi, and L.
cacumhuis are allied is supported by the fact that the samples analyzed show
the same principal alkaloids as those in L. montanus proper.

Lupinus mudleri has a distinctly different spectrum of alkaloids from L.
cacuminus suggesting that the two taxa are maintaining their isolation at the
present time, even though they are in close proximity and are separated only

altitudinally by a few hundred feet.

The lone specimen available from Volcan Colima, probably in Jalisco, sug-
gests that this population may be sufficiently isolated to require recognition.
However, a single sample is not sufficient to permit an analysis of the situation,
particularly since it appears to be morphologically very little different from the
main population of L. montatms, even though it is geographically isolated.

1977]

DUNX & HARMON— LL/P/NL/S

345

O

o

o

I-

I-

o

^ +

^w

00

* ^H

o

CC

uu

,5 "S

<U

rt o

on ^

o +

3 +

a

• rv

o 3

*u

a;

+

^

■ #

OJ

"c

0^

c

^■^

^

CA

c

fl

rt

(U

^^ ^

s

a;

5

a!

a>

o.

en

ir

;rt

'W

cc

c
H

X

■

o

oo

I-

CI

05

o

cc

•

CJ

o"

CO
CI

o

CI
CI

cc

10

o

CO

o

CI

o

X

lO

CI

o

>

rt

o

hJ

+
+

+

++

o

CI CO

^

c
o o

nj n

^ (U OJ

+

+

+

+

+
+

+ +

F

+

+ + +
+ + +

+

+
+

+
+

4- +

H H

H H

+

+ + + +

++++-f + + +
+-f + + + +++

+

+

+ + ++ +

h h b< +++

H r-

H H

h H H

0^

H H H H H H

H h

2

H ==+ + + $

U++ +

H H

+ +

fie

;5

in

oc

■^■^ t^

<7>

05

"ocoi^ric :::

CI

c^

^ ;^ — ■ ^ ai ^

to

CJ >-i-. -I- ^1— ^
^^ (« (« Wl O >

O C C V Q.*

i;>^oi

^■^ , c ^

^ "^ ^ i- ^

c == a c c

> CO

" — lo

CO

ON

^V

ci

^ c S §

^ *^ -.,

CO

:= "^ ^^ r- v=

CI

<o

4^ C C C

^^ - • • a> (C

'y-^ -£

^ ^ fc^ ^

t

-— ^

#« rt

^ 5u

3 C ^ ^

~ o c

^ ^ a c g
K =^ ^ c c

^ '^ ^

V

-* ►--. »^

L^

c o

> >

^

2 -J C C C C -

3'— '-O^C^C -O ^

o , Oj oj 0; a; -J

r*. ^ '•^ ■'^ ^ w !i;

=2j3 > > > > g

c rt Si i: Si 2i fi

Q-C Vv^ / y !/ tn

►J

<.

> E

^ c o o

^ -^^ ^ Ti —

to

, ococ

1 I

h H :- H H

H H

H

++ +++++

H+

+H

*—

h+H

I-++

I-+

+H

—

h+H

I-++

H H

H H h H h

+ + ++H+ +

H H

H H + H

H + H + +

00

5^ ^

k;o

COCO

g I. -2 -2

^ C) "— '—

^ ^
^t^

3 l^^-^*^

^ •

.- S •: ^

(J ^ c 2

^ v:^:

'i)

». ■»-.

--J

-U ■ <«f

^ r^ C3 Cm Cw

t w^ c c c

c« b c c

Cr CI iJ 1-J C^

s « rt - t;

.-ji; 3 3 3

2-UOOO

^5

• o . ^

=: T >:

*u "!:: •>

j;;-u =:

■w^

WWW

i 1^

»* ^ r -v

c? c3 o O ^'^

o o 5 o ^

i< ^ ir= ^ —

— ^ i> Oj

rt :^ :^

O o O

n <r. cz

X X X

^^

CO

■^ ■■ ^TJ -,*^ *^ --^ '^

-- Cj Ci rt

- £ S S

^ 1j 0^ 0;

£ 3 "■ '"■

^^

-OOO -OOO

'^ii' V^ VS A ^

E O C O O

*" ti: tc u tjc

;. = c c g

:-' cs fS c3 ra

':: 1- u J- i-

v3

2

S C 3

2 r^-^

^ £ £

O; Oj qj
— S C5

-S d 3

S-00

h3

^* _d jv4 .^ "i

-.* rf S ':?

c ^ l:^ S^ a; c

■j; ,1^ 3 2 C

^^ ^^ ^^ • • ■

r^ ^

c j:^ p:; o:: Di pd

34(5 ANNALS OK THE MISSOURI BOTANICAL GARDEN [Vol. 64

Taxonomy

Key to Lupititis numtanus and allies

a. Banner sparsely strigoso dorsally, near the distal half, along the eiest; stipules 4.5
cm long or less (sec also var. atistralovolcauiats) ; leaflets linear to linear-lanceolate,
2.5-5 cm long, strigose to sericeons above.

b. Plants abont 14 dm tall or more, glabrous or minutely pubernlent; bracts per-
sistent; kuown only from Pern _ (not treated) L. pracstabilis

bb. Plants 4-9 dm tall, sericeous to strigose; known from Guatemala or Mexico.
c. Petioles 5-13 cm long; stems hollow, fistulose below; pubescence apprcssed-

sericcous; Nue\o Leon, Mexico .— 2. L. cacuminus

cc. Petioles 2-4.5 cm Icmg; stems solid and ligneous.

d. Keels glabrous; pubescence of leaflets strigose-sericeous; known from

Guatemala - 4. L. kellennaniautts

dd. Keels ciliate; pubescence of leaflets strigose-sericeous; known from

Coahnila and Nuevo Leon __ 3. L. mucUcri

aa. Banner glabrous; largest stipules more than 4.5 cm long or the stems with long
spreading pilose hairs, 3-4 nun long; leaflets mostly glabrous above, sometime stri-
gose above, variable in size up to 15 cm long.
e. Pubescence of the stems with abundant spreading pilose hairs to 4 mm long;

sliMUS becoun'ng ligneous; petioles fi-12 cm long; largest leaflets 4-7 nun long;

kuown froui Costa Rica - - - — - 5. L. valcrioi

ee. Pubescence of the stems strigose, glabrous, or canescent; petioles variable in size;

stems ligneous or herbaceous; largest leaflets 5-15 cm long; widely distributed.

f. Stems and stipules glabrous to minutely pubernlent; leaflets glabrous above;

keels glabrous or sparsely ciliate above distally; known from Chihuahua,

Durango, or Pern.

g. Stems glabrous and glaucous; flowers 18-20 nun long; shrubs to 3.5 m

tall; known onl> from Peru (not treated) L. ))roculaustrimis

gg. Stems glabrous or glabrate; flowers 15-18 mm long; herbaceous stems
5-15 dm tall; known from northwestern Mexico.

h. Bracts lance shaped, strigose dorsally; upper lip of the cab'x trun-
cate with an irregiilar notch; known from Chihuahua and north-
ern Durango __ — Id. L. montanufi subsp. ^Jahrior

hh. Bracts lanceolate, the tips attenuate and setaceous hairy, the
lower dorsal area glabrous; upper lip of the calyx triangular;

knowii from southwest Durango and Sinaloa — —

_ _ __ __„ _.. le. L. montanus subsj). montcsii

ff. Stems, stipules, and bracts abundantly pubescent; known from other areas

of Mexico and Guatemala.

i. Stipules hispidulous to canescent within, the larget 10-33 cm loug;
stems fistulose, hispidulous to retrorsely hispidulous, 12-33 mm in

diameter; known only from Ixtlan, Oaxaca, Mexico ____

_ __ lb. L. montanns subsp. montanus var. nchonii

ii. Stipules glabrous within, (u* if pubescent, not with the abfwe combina-
tion of characteristics, generally less than 10 cm long; stems pubcru-

Fk;uue 2. Illustration of the t\'pical structures of Lu})intis niontautt.s subsp. 7nontantis
var. montanus. The floral and vegetative parts are drawn to the mean value of a set of
30 dissections from the geographic range of the taxon. The lettering is the same on all
of the figures: F = lateral view^ of the left side of the flower; B = banner petal flattened,
dorsal view; Br = bract, outside lower portion, inside upper portion; Ca := cal>x, cut at
the left lateral sinus and opened so that the inside is illustrated; K = keel petals, enclosing
staminal tube and pistil, with the mean number of ovules drawn; L =^ average largest leaflet
drawn to Vi the scale used for the floral part, the lower half shows different hair types
observed on the lower surface, the upper half shows hair types observed on the upper sur-
face; S := stem structure of first year growth and hair types, half scale; St := sheathing pair
of stipules, half scale, showing hair types; W = wing petal.

1977]

DUW & HAHMOS—LUPINUS

347

348

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol, 64

lent to stri^ose or hispidulous, hollow, 4-12 mm in diameter on tlie

first year's growth, becoming ligneous.

j. Stems to 4 mm in diameter, but hollow on the first yearns growth,
finely apprcssed puberulent; stipules only 2-4.5 em long and
pubeseent within; racemes generally less than 8 cm long; rare,
known froni Volcan Santa Maria, Cnatemala

Ic. L. montanus subsp. Diontantis var. austrovolcanicus
jj. Stems of the first year's growth over 5 mm in diameter, hollow
and flattening on pressing, variously pubescent; largest stipules
4-9 cm long, counnonly glabrous o\er most of the area within;
racemes o\er 15 cm long at maturity; known from Guatemala to
central Mexico la. L. montanus subsp. montanus var. montanus

la. Lupinus montanus H.B.K. subsp. montanus var. montanus, Nov. Gen. Sp.
Pi. 6:477. 1823. type: Mexico, Montosis Novae Hispaniae (Nevada de
Toluca, 9,000-10,000 ft) (P, holotype, not seen; microfiche, MO).— Fig. 2.

L. vaniuains Cham. & Sclilccht., Linnaca 5:590. 1830. tyi'e: Mexico, Monte Orizaba, Sep.,

^chifdc & Dcppe (IIAI., holotype; photos, NY, TEX, UMO).
L. flahcUam Bertol, Fl. Cuat. 30. 1840. type: Gnatemahx, Volcan d'Agua (not seen). Toro-
typk: Harmon 3646 (MO, NY, UC, UMO, US).

Plants perennial, 0.(S-2 m tall, rarely to 2.5 ni; stems hollow above the ground
level, with the current year's growth .5-12 mm in diameter, pubescence varying
with locahty and altitude, finely appressed puberulent, to strigose or sericeous,
more densely at higher altitudes, or hispidulous to retrorsely hispidulous or
cancsccnt; petioles of the primary leaves along the upper stems 6-15 cm long
with the stipules connate nearly lialf the length of the petioles, those of dwarf
lateral branches often reduced in size; stipules enshcathing % or more of the
diameter of the stems, 4-9 cm long, the triangular free tips 4-12 mm long, both
petioles and stipules pubescent dorsally as the stems, the stipules generally
glabrous or glabrate within, on dwarf branches the stipules of the multiple
leaves imbricated; leaflets of the larger primary leaves 10-15, linear to narrowly
oblanceolate, the largest 5-13 cm long, 5-14 mm wide, generally glabrous above
at lower elevations and puberulent to strigose above at higher elevations, the
tips acute or slightly attenuate; peduncles 10-22 cm long, shorter on late-season
branches; racemes 15-30 cm long at maturity, verticillate to subverticillate;
bracts large-sheathing or covering 3-5 cm of the tip of the elongating raceme,
hiding the buds, the tips attenuate-caudate, pubescence as the stipules for
each population, generally caducous; pedicels 5.5-8.4 mm long, hispid or with
appressed hairs; calyces sericeous, strigose or canescent on the outside, puberu-
lent within on the distal portion of both lips, the lower lip 8-11 mm long, gen-
erally entire, the upper lip 5.5-8.4 mm long, the notch at the tip 1-5 mm deep,
the lips connate laterally 1.4-1.8 mm, the bracteoles 1.0-3.4 mm long, attached
below the lips of the lateral sinuses; corollas glabrous, blue to lavender or

purple, occasionally white or pink; banner obovate-rotund, longer than wide,
the tip emarginate, 12.6-14.8 mm long, 11-14.5 mm wide, reflexed near tlic
midpoint, reflexed 5.7-8 mm, appressed 6.7-7.8 mm, reflexed/appressed ratio
0.81-1.06, the angle 130°-146°; wings 13.8-16 mm long, 6-10 mm wide, the
claw 2-3.7 mm long; keel 4-5 mm wide in the middle, the angle 84°-98° (aver-
age 90.9 ); ovules 7-10; pods 4-5 cm long, 9-10 mm wide when dried, arching

1977] DUXN & HARMON— L(7P/.VL/S

349

up and outward, abundantly to sparsely tangled-pilose, the hairs 1-2.5 mm
long; seeds black or brown with dark mottling, 4-4.5 mm long, 3-3.6 mm wide,
with a deep funicular pit; chromosome number n = 24 (Beaman et al., 1962).

This is one of the wide-ranging species of the volcano zone of Mexico and
Central America. It occupies a zone on most of the high mountains from tim-
berline, or above in sheltered areas, down through the upper pine forest into
the mixed pine-oak forests. The dominant alkaloid produced is sparteine, with
only traces of minor alkaloids. Nowacki's (1963) contention that sparteine is
the primitive alkaloid would suggest that this may be one of the older, more
primitive species of Ltipinus in North America. The trait of main^ leaflets and
the fistulose stems have caused many botanists to mistake this taxon for L.
polyphyllus which ranges from California and Oregon north into British Colum-
bia. The huge sheathing stipules, however, make L. montaniis easily recog-
nizable. While L. polyphyllus is considered as one of the older taxa of the West
Coast, it has the necessary genes to convert or utilize sparteine, changing it into
several other alkaloids. The two northern subspecies of L. montanus have also
modified genomes so that they concentrate other alkaloids. The presence of
what appears to be a distinct nectary at the base of the ventral sulcus of the
banner and the thickened glandlike upper surface of the base of the staminal
tube seem to be unique in Lupimis.

While the original description of L. montanus failed to mention the sheath-
ing stipules, they are clearly recognizable in the microfiche of the type speci-
men, and the material from Mt. Orizaba differs from that of Nevada de Toluca
only in the hair type being hispidulous to canescent, hence the contention that
L. vaginatus is a synonym. The type description of L. flabcllaris clearly men-
tions the sheathing stipules and topotype material is indistinguishable from the
Mexican portion of the taxon, botli morphologically ax^d chromatographically.

Guatemala, dep. chimaltexanco: Chicliaxac, Skufcli J07(US). Cliiclioy Pros, Ilttn-
newell 17145{Gli), San Marcos, 7. R. JoJinston /229(F). Sierra San Elna, Siler 2303{CU).
Volcan Acatenango, Beaman 3258{CH, MSC, TEX, US); StamUcy 61902{F). Volcan de
Afiua, J. R. Johnston 578(F). Volcan Fuc^o, N side, Beaman 4()25{CU, MSC, US), dep.
HUEiiuETENANGO: Cumlirc Papal, Stcycniiark 5()945{F). Between Tocquia and San Juan
Ixcoy, Moncnrc in 1950 (F). Between Tnnima and Quisil, Stcycrmark 48426(F). dep. que-

XAXGO: Canton La Eseranza Forest, 6 kin from San Juan Ostuncalo, Molina et a], 16648

(F, US). Cerro Lieteoreya, Koninck 302(US). 10 uii W of Qnezaltenanj^o, King 3199

(MICH, TEX, US). 11 mi SE of San Marcos, Kinfi 3173(MICH, TEX, US). Sierra Madre
Mts., Williams et al 22780{F, US, WIS), 228()(){F). Volcan Santa Maria, Skuteh S55(F, GH,
US); Steyermark 34168(F). \'olcan Santo Touias, Steyermark 34857(F). dep. sacaie

c^uEz: Volcan Agna, Harmon 359S( UNTO, US, WIS), 3646{CAS, CUN, F, MICH, MO, UC,
UMO, US); Beaman 2942(GU, MSC, TEX, US); KeUennan 4708(IJS); Maxon & Hay 3687
(US); Molina 21029(F); Salas in Jan. 1926(US); Shannon 3679(US); /. D. Smitli 2152
(US); StancUey 65093(F). dep. san maucus: Cerro El Bonito, Plotvman 5044(Gll). Be-
tween San Sebastian and summit, Steyermark 3<5553(F); sunnnit, Steyermark 35530(F). 2 mi
S of San Sebastian, Williams et al 25915 (F, WIS). Volcan Tacand, Beaman 3230 (GH,
MSC, TEX, US); Steyermark 36140(F). Volcan Tajunmlco, Beaman 3140(CR, MSC, TEX,
US); Reeder in Apr. 1952(MICH); Sliannon 56-.S(US); Williams et al 26996(F, GH, US).
DEP. solola: Volcan Atitlan, Beaman 4094 ( GH, MSC); Kellerman 5769 ( US); Steyermark
47 508 (F, US). Volcan Santa Clara, Steyermark 46988(F). Volcan Tolinuin, Steyermark
47535(F), DEP. totonicapan: Cerro Maria Tecum, Williams et al 23145(F). Boundary of
Depts. Hnehuctenango and Quezaltenango, Williaf)is et al 22707 (F, Gil, US, WIS). Los

350

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

WIS). 8 km S of TotonicaDan, Willia

WIS

Mexico, chiai-a.s: Mt. Pasitar, Matuda 70{MICH, NY, US); Mattula S-20y(MICH).
Voltan Tacana, Matuda 2333{GH, MICH, WIS), colima: Cuchilla, Nevada de Colima,
Gold.stuitJi 57(F, GH, MO, US), distiuto federal: Cerro Ajusco, Beaman 2796(CH, MSC,
US); Garcia in May 195t(IPN). Pedregal de San Angel, Barclay & Paxton 53f>(F, TEX).
Pcna de los Charros, Husscll & Souviron J39(US). cuehueiu): Mina, Cerro Teotepec, Iliuton
14266{GU, MICH, US); Rzcdowski 18493{ENCh, MICU). hidalgo: Plain.s of Actopan,
Graham i6'<S(CH). Cerro de las Venturas, N of Paelnica, Galvan in Aug. 1963 (ENCB);
Nui'wz 56(UMO). Kl Cliico, Dunn cl ul. 2029S( UMO, US). Sierra de Pachuca, Rose & Hay
5628{\JS); Prin^lc 953U{F, CH, MO, US, VT). jalisco: Nevada de Colima, Beaman 2366
(MICH, MSC, US); Matuda 3S369( MEX, UMO); McVaugh J0076(MICH, small flowers).

MEXICO: 10 km E of Amecameca, Quijanu 5i (ENCB). 11 mi E of Amemcca, Dodd.s 11

(NHCH); Montgomertj & Root 8ll4a{MSC). Cerro Joeotitlan, Matuda 38490(MEX, UMO).
CrneeroPAgua-Blanea, llinton S244(CH), 83i7(CH, MO, RSA, UMO, US). Crueero-Raiees,
llinton 903/ (F, CH, ENCB, MO, RSA, UMO, US). Estaeea, Matuda 3S503( MEX, UMO).
I.vtaehnatl, Falda, Matuda 26148{NY, US); Beaman 3477 (GH, MSC, US, n = 24); Nelson &
Goldman in Jan. 1894(US); Purpus 32(US), 203(MO); Rudd J029(UMO, US); Rzedowski
J9806(ENCB, UMO). Me.s6n Viejo, Matuda 3cS'395(MEX, UMO. Nevada de Toluca, Balls
4088{\JS); Dunn JS6'37(UMO), 1884()(ENCB, MSC, MO, NY, TENN, UMO, US); Dzicka-
nowski cl al J9J5(MEMO, MO, NY, UC, UMO, WUP), ]932(ASU, CH, MO, MSC, OSC,
SLP, TENN, UMO), ?936(MEMO, MO, RSA, UMO, US, WUP); Gallian & Leake 897
(UMO); Galliotti 3360(P); llunnewell 1314(HGU); Islas 28(MEMO); Mick & Roe 187
(ENCB); Morales-Diaz in Aug. 1962(ENCB); Rose & Painter 79()G{VS); Rzi'dowski 15782
(ENCB); Scheni 90(MICH, MO, US). Pa.so de Cortez, litis et al. i025( MICH, MSC, TEX,
WIS). Pesco Inst, de Nacional, Dunn et al. 2()373(F, GA, K, ENCB, MEMO, MO, NEL,
NY, ORE, RSA, UC, UMO, US, WUP). Rio Frio, Coutreras in July I962(ENCB). Tlamacas,
vie. of Popocatepetl, Fonseca i'.5(ENCB); Galicia in July 1962(ENCB); Garcia 3A( ENCB);
Lundell /235S(MICH, TEX); Madrigal in Dec. 1959(ENCB); Matuda 3<SJ97(MEX, UMO);
Moore 36( GH); Quijano 5J(ENCB). VoUan Popocatepetl, Balls 4227( US); Barkley et al
2353(TEX); Beaman 2fJ22(MSC); Dunn iS566(UMO); JS579( CUN, ENCB, MO, NY,
UMO, US); Galliotti 3368{P); llatheway /793(GH, MO, US); Hucrta iOi(ENCB); Leake
& Gallian 133, iJi(UMO); Lundell i235«(US); Rose & Hay 6()12{\JS); Ross 8(US); Strati)
& Gregory J603(CH, MICH, RSA). micho.\can: Cerro San Andres, 12 km N of Hidalgo,
Beaman ■/295(GH, MSC, TEX, US). Mt. Tancitaro, llinton i.5,593(US); Leavemcorth &
Iloogstraal H28{F, MO). Zitaeuaro-Caci.iue Peak, Ilinkson iJ932(US). oaxaca: Atepec,
Llano de las Flores, MacDougal 37«.S(NY). Cerro de San Felipe, Camp 2869{K\). Cerro
Zemoaltepetl, Uallherg 790(ENCB, UlCll, US). Cumbre de Sierra de Juarez, Matuda
38415{\\E\, UMO). Gueletago, Vilas 3i( WIS). 27 mi N of Ixtlan, Sierra Juarez, Roc &
Roc I9Ji(ENCB, MICH, WIS). Macuiltiangui.s, MacDougal in 1960(US). Mt. Zemoaltcpec,
summit, E. Nelson 6J9(US). Reye.s, E. Nelson i736(NHCH, US). Sierra de Ixtlan, Gentry
et al. 20272( UMO). Sierra Madre del Sur, 60 mi NE of Oaxaca, Webster i7543(MO). Siena
de San Felipe, Camp 2S69(NY); E. Nehon ii35(US); Pringle 4779(F, CH, ISC, MICH,
M.SC, ND-G, P, US, VT); C. L. Smith 333 (MO), puebla: Iztaccihuatl, Beaman 2007 (GH,
MSC, US); litis et al J025(TEX); Weber 372( ENCB). Pass between Mexico City and
Puebla, Mexia 26J7( MICH), 2647A(CAS). Alberque Piedra Grande, Beaman 3G4(){Cn,
MSC, Tl'.X. Ciudad Serdan, Beaman 2498{GU, MSC). Pico de Orizaba, GclUotti 3343iP);
Greenman 28{¥); Liebman 4892(V\ GH), 4893(F); Pringle 9,528(F, GH, MO, US, VT);
Schiede 666( HAL; photos, GH, TEX, UMO, US); Seaton 5i0( GH). Popocatepetl, Barkley
17Mo87{F); Barkley et al 2353; Beaman 1747, 2i09(GH, MSC); Dwm i8558(UMO),
JS56^(CUN, ENCB, MO, NY, UMO, US); Miranda & Barkley i7MoS7( TEX, MSC). tla.x-
cala: Mt. Malinche, Balls 4890{VS). vehacruz: Cerro de Perotc, Balls 46()4iVS). Cueva

Figure 3. Structures of Lupinus montanus subsp. montanus var. nclsonii drawn to
mean \alues, for those traits which differ from \ar. montaims. The calyx, bract and stem
are drawn to the scale shown, while the stipule and leaflet are half scale. The lettering is
Br — bract; Ca := calyx, inside \ iew; L — leaflet; S z^ stem; St ^ stipules. ( See h'gend
for Fig. 2 for full explanation.)

1977]

DUNN & HAnMON- LUPINUS

351

352

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

do MiKMto, Bcaman 1783{MSC, US). Pico de Orizaba, Bcaman 229i(MSC); E. Nelson 266
(US); Rose & Hay 5727(US); /. G. Sm(7/i 3,9i(MO). Unknown locality, £. Nelson 3S(US).

lb. Lupinus montanus siibsp. montanus var. nelsonii (Rose) C. P. Smith,
Sp. Liip. 79. 1938.— Fig. 3.

L. uehonii Hose, Contr. U.S. Natl. Herb. 8:308. 1905. type: Mexico, Oaxaca, near Cerro de
San Felipe, E. W. Nelson lli5{\JS, holotypc, photo, UMO).

Differs from L. montanus in the large fistulosc stems, up to 3 cm or more
in diameter, the fistulose nature extending throughout the above ground por-
tion to the top of the raceme; plants to 2.5 m tall; stems densely hispidulous to
retrorsely hispidulous; petioles to 50 cm long; leaflets 15, to 15 cm long and
3 cm wide, pilose to canescent below and glabrous above; stipules 10-33 cm
l(mg, connate to the petioles for all but 1-2 cm at the tip, the free tips slender
attenuate-caudate, both stipules and bracts hispidulous-canescent within as well
as densely so without; bracts numerous, plumed, the tips elongate-caudate, 3-
4.5 cm long, pilose-cancscent, as also the under side of the leaflets, the bracts
hiding 5-8 cm of the buds at the tips of the racemes, often subpersistent; flow-
ers the same as the species; pods and seeds the same as the species except the
pods densely hispidulous.

The taxon appears to be a gigas form and may represent an ecological mod-
ification since there are many typical specimens of L. montanus in the region.
Tliere are distinctive traits, however, which appear to have a genetic basis, and

tlu

1964. It is also

chromatographically similar in its alkaloids to other samples of L. montanus.

Mexico, oaxaca: Ixtlan, Atepee, Llanos de las Flores, MacDoufial 37Ss( NY). Oaxaca-
Tuxtepec Ilwy., Llanos de las Flores, MaeDoufid in 1960 (US, 4 sheets). Cerro de San
Felipe, E. Nelson 1145{\JS); rringle 5S3.9(C1I, VT). Cerro Zemoaltcpctl, Schultes S02{C.U);
Uallher^ 7,90 (MICH, in part). Lxtlan de Juarez, Sierra Ixtlan, Gentnj 20272( UMO, US).
13 mi N of Ixtlan, Anderson 48 42 {WICW) .

montanus subsp. montanus

M

Sp. Lup. 90. 1938. type: Guatemala, Volcan Santa

E. Nehon 3709 (US-250873, holotype; F, GH, isotypes).— Fig. 8.

Plants perennial, over 3 dm tall; stems hollow, ligneous, 4 mm in diameter,
finely appressed puberulent; petioles 6-10 cm long; larger stipules 3-4.5 cm
long, wide, membranous, sparsely pilose to canescent inside and outside, the
free tips 7-11 mm long; leaflets 10-11, linear, acute, mucronate, the longest
5.5-6.5 cm long, 6-S mm wide, sparsely strigose above, thinly kinky canescent
beneath; peduncles 2-8 cm long; racemes ca. 8 cm long, verticillate, the lower
whorls to 2 cm distant; bracts ca. 16 mm long below; pedicels 5-6 mm long,
slender sericeous-puberulent; calyces canescent without, finely sericeous within
near the tips of the lips, the lower lip slender, arcuate, 8-9 mm long, entire, the
upper lip ovate, bidentate, 7 mm long, the lips connate laterally 2 mm, the
bracteoles 1.5-2 mm long, attached near the lip of the lateral sinuses; banner
suborbicular, glabrous or occasionally sparsely hairy on the distal portion of
the dorsal crest, 11.5-12 mm long, 11-11.5 nun wide, widest above the midpoint,
rnfh'Mvl fi7 mill nnnrrss(>d 7 mm: wiutTs 12.8-13 mm lonti. 7 mm wide; keel

1977] DUNN & HARMON— Lt/P/.VUS

353

with minute papillae above the claws, 3.5 mm wide in the middle, the angle
95°-100° at anthesis; ovules 6-7; pods 3-4.5 cm long, 9 mm wide, pilose with
hairs 1-2 mm long.

The specimens seen appear to represent hybridization and introgression from
L. kellerrnannii. The slender woody stems, short stature of the plants, narrow
smaller leaflets, intermediate petioles, and the presence of pubescence dorsally
on the banners of about half of the specimens all suggest introgression. The
flower size and stipules are distinctly like those of L. montamis.

Guatemala, dep. quezaltenaxgo: Volcan Santa Maria, Beainan 4124{ENCB, MSC,
TEX, US); E. Nelson 3709(F, GH, US); Stetjermark 34205{F). Above Palojunoj, StancUcij
67703, 67707, 67738{F), 67683{F, intermediate to var. monianus).

Id. Lupinus montanus subsp. glabrior (Wats.) Dunn & Harmon, comb. nov.
— Fig. 4.

L. montanus var. pjahrior Wats., Proc. Anicr. Acad. Arts 23:270. 1888. type: Mexico, Chi-
huahua, summit of Sierra Madre, Prinpjc i206(GH, liolotype; F, G, K, ND-G, P, RSA,
US, VT, isotypes; photo, UMO).

L. gJahrior (Wats.) Rose, Contr. U.S. Natl. Herb. 8:308. 1905.

Plants perennial, to 1 m tall; stems glabrous to glabrate, ligneous ridged, at
least on drying, fistulose, 6-8 mm in diameter; petioles 12-20 cm long, the free
portion finely strigose; stipules membranous, sheathing and often imbricate on
dwarfed branches, 4.5-6 cm long, the free tips only 3-6 mm long; leaflets 14-
15, the large.st 9-11 cm long, 12-13 mm wide with acute-mucronate tips, gla-
brous above, sparsely and finely puberulent below; peduncles to 17 cm long;
racemes 20-30 cm long at maturity, verticillate, the lower whorls, 2.5-3 cm
distant, the rachis finely but densely puljerulent; bracts broadly lanceolate,
membranous, gradated, the lower to 2 cm long, the upper reduced, minutely
puberulent without, glabrous within; pedicels 6-8 mm long; calyces with broad
boat-shaped lips, finely puberulent without, glabrous within, the lower lip 7-8
mm long, entire, the upper lip 5-7 mm long, the apex blunt with an irregular
notch 0.5-0.8 mm deep, the lips connate 1.4 mm, the bracteoles straplike, 2-3
mm long, glabrous, except for a few setaceous hairs near the tips; corollas gla-
brous except for a few papillae above near the claws of the keel or occasionally
the keel ciliate above toward the acumen; banner orbicular, 13-14 mm lon<i,
13-15 mm wide, reflexed 7 mm, appressed 6.5-6.8 mm, the sulcus 2.4 mm deep
midway betweeu the umbo and the base, the banner angle 133''-150''; wings
15-17 mm long, 8-10 mm wide; keel 4-5.3 mm wide in the middle, the angle
80^-85^^, occasionally papillae near the claws or occasionally ciliate above near
the acumen; ovules 7-9; pods 8-8.5 mm wide, 3.5-4,5 cm long, thinly strigose;
seeds nearly black with mottling, 4.5 mm long, 3 mm wide, a pit at the funic-
ular attachment.

The subspecies glabrior is known only from northern Durango and Chihua-
hua, from the summit of the Sierra Madre Occidental, in rather inaccessible
areas. The area is north of that of subsp. montesH and subsp. montanus, as well
as the fact that there arc distinctive morphological traits in addition to distinct

354

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

CO

O

1977] DUNN & HARMON— LUPINUS

355

chromatographic differences. The best distinguishing traits are the short, blunt

upper hp of the calyx and the lanceolate bracts. While other traits are distinct,

they are not as easily recognized. Since the geography, ecology, morphology,

and chromatography suggest a distinct gene pool, subspecific rank is suggested.

The large stipules and the floral morphology clearly indicate the affinity to
L. montanus.

Mexico, chihuahua: Vic. of Chucliuichupa, summit of Sierra Madre, PringJe i206'(GH,
F, G, K, ND-G, P, RSA, US), J579(P). Cerro Mohinora, 10 mi S of Guadelupc v Calvo,
Straw & Foreman i943(ENCB, MICH), duranco: Sierra Madre, 30 mi N of Guanacevi,
E. Nelson 4785 (VS).

le. Lupinus montanus subsp. montesii (C. P. Smith) Dunn & Harmon, comb
nov.

L. montesii C. P. Smith, Sp. Liip. 41. 1938. type: Mexico, Sinaloa, Cerro de San Rafael,
San Ignacio, Monies & Salazar 112{\]S, holotype). — Fig. 5.

Plants perennial, from a woody caudex, 4-7 dm tall; stems hollow, glabrous
below, sparsely strigose on the peduncle, rachis of raceme, and petioles, ligne-
ous ridged, at least on drying, 6-8 mm in diameter; longest petioles 10-19 cm
long, strigose on the portion not fused to the stipules; stipules sheathing and
encircling % of the stem, 4-10 cm long, the free, caudate tips 1-2 cm long,
glabrous except for a few scattered setae near the tips; leaflets 9-14, linear-
elliptic to narrowly elliptic, glabrous above, sparsely strigose below, the largest
6.5-9 cm long, 6-12 mm wide, acute-mucronate at the tips; peduncles hollow,
7-14 cm long, sparsely strigose; racemes verticillate to subverticillate, 20-35
cm long, rarely only 7 cm long in depauperate specimens, the rachis more
densely strigose; bracts membranous, the long caudate tips with scattered seta-
ceous hairs, caducous, 1.0-3.5 cm long, broad and completely covering the
flower buds at the tip of the raceme, pedicels filamentous, 5-8.8 mm long,
densely strigose; calyces appressed puberulent outside, glabrous within, the

I, generally entire, occasionally with a notch 0.1 mm

8-12

6-9

base gibbous above, the lips connate 1.2-2 mm, a glabrous spatulate bracteole
1-4 mm long attached near the lip of the lateral sinuses; corollas blue and white,
glabrous but the keel sometimes ciliate; banner 11.6-15.9 mm long, 12.5-17.4
mm wide, reflexed 6.4-7.7 mm, appressed 5.5-7.4 mm, reflexed/appressed ratio
(average 1.17), banner angle 130^-149° (average 141); wings 13-18.4 mm
long, 8.4-10.4 mm wide, the claw on the average 3.2 mm long; keel 3.8-5.2 mm
wide in the middle, the angle 89°-98'' (average 91°), ciliate above near the
acumen in half of the specimens, the others glabrous; ovules 10-13; only imma-
ture legumes seen, these strigose.

Figure 4. Structures of Lupinus vwntatius subsp. glahrior drawn to mean values, for
the differences from subsp. montanus. All parts are drawn to the scale shown except the
stipules and the leaflet which are drawn to half scale. The lettering is: B = banner, dorsal
view; Br = bract; Ca = calyx, inside \iew; F — flower, left side view; K = keel; L —
leaflet; S =: stem; St = stipule; W = wing, (See legend for Fig. 2 for full explanation.)

356

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

1977] DUNN & HARMON— LUP/Nf/S

357

The subspecies is known only from soutliwestern Durango, Mexico, in the
general area of El Salto, and from Cerro de San Rafael, Sinaloa. Most of the
collections have been west of El Salto but some have been cited as 45 miles to
the south. The elevations have been near 7,000 ft, which is well below the
normal altitude for L. montanus. The distribution is also geographically dis-
tinct from that of the species. Since the ecology, geography, morphology, and
the composition of the alkaloids, are distinctive, it is suggested that there is a
sufficiently distinct gene pool to recognize the taxon at the subspecific level.
It is also distinct from subspecies glabrior in geography, morphology, and in
chromatography. The large sheathing stipules and the floral morphology leave
no doubt as to its affinity with L. monUmus.

Mexico. duran(;o: Cerro Auehiieto, S of Hiiachichelcs, Maysvillcs 725S( MICH). Be-
tween Durango and Mazatlan, Pennington 242, 243{TEX). El Salto, Guzman in Sep. 1961
(ENCB). 20 km S of El Salto, Gordon 41 (MICH, X L. madrensis). 19 mi SW of El Salto
Waterfall J5490(OKLA). 33 mi SW of El Salto, Waterfall 1271 4{OKl.A, UMO). 3 mi W
of La Ciudad, LeDoux et al. 2024(C, CAS, G, ENCB, K, MEX, MO, NY, U, UC, UMO, US).
6 mi W of La Ciudad, Fltjr 274(TEX). 13 mi E of La Ciudad, LeDoux et al. J996(ENCB,
MEX, MO, NY, UMO, US). 8.4 mi W of La Ciudad, Retcal & Aticood 3507(MARY, UMO);
Breedlove JS(S77(CAS, UMO). 14.3 mi NE of La Ciudad, Pinkava et al. 9495{ASU, UMO).
54 mi N of Estacion Coyotes, Breedlove 18789{CAS, ENCB, UMO). Metatcs, N of Cueva,
Penned 18408{Gll, US). Road to Pueblo Nuevo, Maijsvillea 7760(MICH, X L. madrensis).
W of Pueblo Nuevo, Maijsvilles 8071{M1CII, TEX), sixauja: Cerro de San Rafael, San
Ignacio, Monies & Salazar 112{\JS).

2. Lupinus cacuminus Standley, Publ. Field Mus. Nat. Hist., Bot. Ser. 22: 79.
1940. type: Mexico, Nuevo Leon, peak of Cerro Potosi, Municipio de Gale-
ana, Mueller 2269 (F, holotype; GH, MO, isotypes) [Mueller 1257 (F)
labeled type in Standley's handwriting]. — Fic. 6.

Plants perennial, caespitose, 3.5-6 dm tall; stems from a caudex, fistulose,
the internodes between fully developed leaves only 1-3.5 cm long, pubescence
all appressed-sericeous but of multiple hair types and lengths, the upper 2 or
3 nodes with branches initiated by anthesis of the primaiy racemes; largest peti-
oles 5.5-13 cm long, reduced progressively upward, persistent long after the
leaflets drop; stipules gradated from 4.5 cm long below to 1.5 cm above, imbri-
cated below, connate 3 cm below to only 7-8 mm above, the free tips subulate-
caudate; leaflets 10-14, linear-elliptic, appressed silky villous on both sides,
sparsely above and the central area often glabrous, the largest 3-4 cm long,
3^ mm wide (to 6 mm wide in a population on Pefia Nevada), the tips acute
and mucronate; peduncles 4-5 cm long, exceeded by the foliage; racemes 10-13
cm long but numerous bracts in a terminal tuft suggest that they may get much
longer, the flowers tightly and spirally arranged; bracts lance-subulate, tardily
deciduous or semipersistent; pedicels 6.5-12 mm long, hi.spidulous; calyces silky

FiGUHE 5. Structure.s of Lupinus monianus subsp. montcsii drawn to mean \alues, for
the differences from subsp. montanus. All parts are drawn to the scale shown except those
for the stipules and leaflet, which are drawn at half scale. The lettering is: B = banner,
dorsal view; Br = bract; Ca = calyx, inside view; F = flower, left side view; K = keel;
L = leaflet; S = stem; St = stipule; W = wing. (See legend for Fig. 2 for full explanation.)

358

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

I ■!

/

//

\

l\

\

L

kJ

•o

o

C/)

iA

E

O

a

(A

o.

\\

i. 1

fi

' I ,'.

//.

>'

\

L\'

/

S'

'J

1.\^

71

i^^

/

n

< ^ ^

'f /

lu

I

>S

f 'I

,//'"

\f

\ . h

^\

'i

4'^

Mt

v, u

1977] DUNN & HARMON— LUPINUS

359

white-villous with appressecl hairs, tlic lower hp oblong-hmceolate, 8-10.2 mm
long, tridentate or entire, the teetli 0.2-0.4 mm deep, the upper hp 7.7-10.4 mm
long, bifid, the notch 1-4 mm deep, the lip oblong, 3..5-4 mm wide, flattened,
the lips connate 1.4-2 mm, the bracteoles lanceolate, 1.5-6 mm long, attached
at the lip of the lateral sinuses; banner sparsely pubescent dorsally on the distal
portion, the tip emarginate, 14-17.5 mm long, 13-17,6 mm wide, reflexed 7-10
mm, appressed 6-8.5 mm, the angle 140''-155'', the sulcus 1.5-1.8 mm deep
midway; wings 15-17.4 mm long, 7-9.5 mm wide, the claw 2.5-3 mm long;
keel 4-5 mm wide in the middle, glabrous or ciliate above near the acumen,
the angle 80°-90'' at anthcsis; ovules 5-7; pods 4-5 cm long, 9-11 mm wide,
densely lanate with hairs 2 mm long.

While collectors have reported the plants as abundant in the upper pine
woods and above timber line on Cerro Potosi, they have only been collected
on three mountain peaks: Los Alpes, Cerro Potosi, and Peiia Nevada. Flower-

June through J

The taxon appear

intermediate between L. montanus and L. mucUeri in several characteristics
but has essentially the same spectrum of alkaloids as L. montanus. The speci-
mens from Pena Nevada have sparse ciliatiou near the acumen of the keel.

Mexico, coaiiutla: Los Alpes, ^0 mi E of Saltillo, Gentry et ah 2005.9(UMO, US).
xuEvo LEON: Cerro Potosi, Municipio de Galeana, Mueller 2269(F, GH, MICH, MO). Biol.
Exp., U. of 111., Schneider 95S(F, CII, MICH, MO). Cerro Potosi, top of nit., Beaman 2654
(CH, MSC, US); Gilbert 26, 29(TEX); Iliuton 17038iMlCR). Cerro Potosi, near micro-
wave tower, Dunn et ah 20203iF, K, MO, NY, RSA, UC, UMO, US); Dziekanowski et al.
1761 (GU, ENCB, K, MEMO, MEX, MICH, MO, MSC, UMO, WUP); MacGregor et al
3J4(KANU, UMO). 20 mi E of Galeana, Mueller J257(F, GH, MICH, TEX). Pefia Nevada,
26 mi NE of Dr. Arroyo, top of Picachio Onofre, Beaman 2690(01, MSC, US), tamaulipas:
Pena Nevada, E and S slopes, Stanford et al 2597 (DAO, U, US). Summit of Pena Nevada,

G. Gillett J237(MSC).

3. Lupinus muelleri Standley, Publ. Field Mns. Nat. Hist., Bot. Ser. 22: 80.
1940, type: Mexico, Nnevo Leon, Las Canoas, on Cerro Potosi, Mnnicipio
de Galeana, Mueller 2205 (F, holotype; CAS, GH, MICPI, MO, TEX, iso-
types). — Fig. 7.

Plants perennial; stems few to many from a woody eaiidex, woody with a
solid pith, erect, 5-7 dm tall, 3 mm in diameter, branching from the upper nodes,
thinly appressed strigose, with a cinereous undercoat of kinky hairs 0.2-0.4 mm
long; petioles 1-2 cm long, filiform; stipules subulate to fihform, 6-8 mm long,
connate to the petioles 2-4 mm; leaflets 6-8, the largest 2-2.5 cm long, elliptic-
oblanceolate, the tip acute, mucronate, both surfaces densely strigose; peduncles
2.5-3.5 cm long; racemes 6-10 cm long, the flowers scattered to subverticillate;
bracts subulate, 8-8.5 mm long, strigose outside; pedicels 7-11 mm long, with

Figure 6. Structures of Lupinus cacuminus drawn to the means values on the scale
shown. The stipules and leaflet are drawn to half scale. The lettering is: B = banner,
dorsal view; Br = bract; Ca ^ calyx, inside view; F — flower, left side view^; K = keel;
L = leaflet; S — stem; St = stipule; W — wing. (See legend for Fig. 2 for full explanation.)

360

ANNALS OF THK MISSOURI BOTANICAL GARDEN [Vol. 64

hispidulous hairs 0.4 mm long; calyces sericeous outside, 2-lipped, the lower lip

3-4

8-4

,8-2

mm long, linear, attached below the lateral sinus lips; banner sparsely pubescent
dorsally near the distal portion, suborbicular, widest below the midpoint, reflexed

5-8.2 mm, annressed 6-7

5^1

midway; wings glabrous, obovate, 13-15 mm long, 7-7.9 mm wide; keel ciliate
above from the middle toward the acumen, 4-5 mm wide in the middle, the

angle 83^-88^ ovules 6-8

bap

pressed kinky pilose, the hairs to 1 mm long; seeds 6.4 mm long, 5.7 mm wide,
tan with faint mottling.

Thus far known only from Coahuila and Cerro Potosi in Nuevo I.eon, where
it was reported as abundant in pine woods, at elevations near 3,000 m. Flow-
ering occurs from June to August.

Mexico, coahuila: Puerto de la Siborio, Arteaga, Marroquin l'^(MEMO). nuevo leon:
Cerro Potosi, Las Canoas, Municipio de Caleana, MucUer 2205( CAS, F, GH, MICH, MO,
TEX). Above Ejido, Beaman 33JJ(F, CH, ENCB, MSC, TEX); Dunn et al 2()247{U\lO);
Dzichnwwsh et al 7770(CAS, C, ENCB, MO, NY, ORE, P, RSA, UC, UMO, US), 1771
(OH, ENCB, MEMO, MEX, MO, MSC, NY, RSA, SEP, UC, UMO, US, WUP).

4. Liipinus kellermanianus C. P. Smith, Sp. Lup. 90. 1938. type: Guatemala,
Volean Agua, 9,000 ft, Kellerman 4746 (US, holotype).— Fig. 8.

Plants perennial, shrubby; woody stems sohd, the branehes sometimes hollow,
3-6 mm in diameter, first year stems only 3 mm in diameter, strigose; petioles

he upper branehes; stipules 8-18 mm long, the small-
est at tlu^ base of the branehes, the longest above, subulate-attenuate, eonnate
5-10 mm; leaflets 7-9, lanceolate, the tips acute, mucronate, the largest leaflets
2.5-3.5 cm long, 5 mm wide, sparsely kinky-villous above, canescent to kinky-
villous below; peduncles 3 cm long at anthesis, 3-8 em at fmiting; racemes 3-6
em louLT. vertieillate: bracts subulate-attenuate. 8-14 mm lonir, canescent: pedi-

FiGUiiE 7. Structures of Lupinus mucllcri drawn to the mean values on the scale shown,
except the stipules and leaflet which are half scale. The lettering is: B = banner, dorsal
view; Br ~ bract; Ca = calyx, inside view; F = flower, left side view; K = keel; L
leaflet; S = stem; St = stipule; W = wing. (See legend for Fig. 2 for full explanation.)

Figure 8. The vegetative structures for Lupinus nwufanus subsp. montanus var. austro-
volcanicus are shown in the upper portion for comparison with L. kellermanianus. The stip-
ules and leaflet are shown at half the scale. The floral traits are the same as shown for
L. 7nontanus subsp. 7no}da)nis \ar. montanus in Fig. 2. The lower portions shows the struc-
tures of L. hcllcnnanianus drawn to the mean values on the scale shown. The stipules and
leaflet are drawn to half scale. The lettering is: B i^ banner, dorsal view; Br = bract; C
calyx, inside view; F — flower, left side view; K = keel; L = leaflet; S — stem; St = stipule;
W =: wing. (See legend for Fig. 2 for full explanation.)

Fi(;uHE 9. Struetiues of Lupinus valcrioi drawn to the mean values at the scale shown,
except the stipules and leaflet which are half scale. The lettering is: B = banner, dorsal
view; Br — bract; Ca = calyx, inside view; F = flower, left side view; K z=. keel; L
leaflet; S = stem; St = stipule; W — wing. (See legend for Fig. 2 for full explanation.)

1977]

DUNN & HARMON— LL/r/NL/S

361

362

ANNALS OF THE MISSOURI BOTANICAL GARDEN

[Vol. 64

CO

[^

A

y.

I,

\

jrv^v

X

V

tVv\

'I

I

I

;

)

'^'y>'

J\^\

>

h

:^

(

~ft\

j\

7i

V

L\

't^

\

'V

*

X

CO

E

CO

a:
(J

(0

o

o

o

M

o

F I

o

CD

^n^j, •

(A

in

n\

\

J*

t'

O

'hi/

O

E

V

/.

'/

E

CO

a:

*

o

o

«>

M

1977]

DUNN & HARMON— LUP//Vt7S

363

354 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

eels 5 mm long at antliesis, 10-12 mm in fruit, hispidulous; calyces canescciit, the

5 mm long, strigose within on the distal half, the upper Hp 6.6-

.5-8

6.8 mm long, the notch 1.2-1.4 mm deep, the lips connate 1.5 mm laterally, the

3-4

the side of the calyx cup or at the base; banner with a sparse patch of kinky hairs
distally on the dorsal side, densest on the crest, obcordate to suborbicular, 12 nun

5^12

4-4

6-7

Very few collections have been made of this species known only from Volcan
Agua and Zunil The traits of the woody steins and pubescence on the baimt^r
have shown up on several neighboring peaks in plants which are otherwise typi-
cal L. montanus. This suggests introgression and the material named L. mon-
tanus var. ausirovolcanicus is probably of hybrid derivation.

Guatemala, quezaltenango: Suniniit of Volcan Zunil, Steijerviark 34848(F). saca-
TEPEQUKZ: Volcan Agua, 9,000 ft, Kellennan 4746{\]S); Kellerman 15, 1905(US).

5. Lupinus valerioi Standley, Publ. Field Mus. Nat, Hist., Bot. Ser. 18: 545.
1937. TYPE: Costa Rica, San Jose, Cerro de las Vueltas, Standley ir Valeria
43668 (holotype F).— Fig. 9.

Plants perennial, 6-9 dm tall, woody below; stems of current season hollow,
subfistulose, to 5 mm in diameter, with abundant spreading pilose hairs 3-4
mm long, and with an undercoat of appressed strigose hairs; petioles of mature
leaves, 6-12 cm long, pubescence as on the stems; stipules 2.5-4.5 cm long,
connate to the petioles 1-2.5 cm, pilose, the free portion slender subulate-atten-
uate; leaflets 7-10, slenderly ol)lanceolate, the largest 4-7 cm long, 7-10 mm
wide, glabrous above, subappresscd strigose below; foliage dense from short
intcrnodes causing the lower stipules to be imbricated on the branches, 2-3 cm,

8-10

6-17

subverticillate, the whorls 10-25 nun distant in age; bracts caducous, lance-
attenuate, 15-23 mm long, 2-3 mm wide in the lower portion of the raceme,
with numerous pilose hairs 2-3 mm long dorsally; pedicels 3-4 mm long at
anthesis, spreading pilose, the hairs 1-2 mm long; calyces densely subappresscd
pilose, the hairs 1-2.5 mm long, the lower lip 8.5-10.6 mm long, the tip bi- or
trifid, tlie teeth 0.1-0.3 mm long, the upper lip 6-8.5 mm long, bifid, the notch
3.5-5.5 mm deep, the lips connate laterally 1.8-2.4 mm, the bracteoles 1.5-2.5
mm long, attached on the calyx cup below the lateral sinuses, with the lower
portion fused to the calyx cup; banner glabrous, suborbicular, somewhat con-
stricted below into a broad claw, 12-14 mm long, 11-12.5 mm wide, reflexcd
5.5-6.5 mm, appressed 6..5-7 mm, reflexcd/ appressed ratio 0.76-0.83; wings
13.5-16.5 mm long, 6.5-8 mm wide, the claws 2.8-3.4 mm long; keels generally
minutely ciliate above on the distal part, 3.5-4.5 mm wide in the middle, the
angle 90°-95°; ovules 5; pods 3-3.5 cm long, 8.5-9.5 nun wide, densely villous,
the hairs 2-3 nun long; seeds 4.5 mm long, 3 mm wide, dark brown to black.

1977] DUNN ik UAnMON—LUPINUS

365

The stipules and bracts show derivation from the L. montanus genome,
which apparently was introduced and introgressed with a local taxon which at
the present time appears to have dominated most of the characteristics, with
only the vestige of traits from L. montanus. Tlie flowering and fruiting mate-
rials have been collected from September through January at elevations from
2,700-3,100 m.

Costa Rica, sax jose: Ccrro Chirripo, Evans et al. iiS(MICH). Ceno Frio, Jimenez
2677(CR, F). Cerro Vueltas, Stamlletj & Valcrio 43668{F), 43974{F, US). Cord. Tulamanca,
Wchcr 6257(MICII).

Literature Cited

Beaman, J., D, C. D DE Jong & W. P. Stoutamire. 1962. Chromosome studies in the al-
pine and subali^ine floras of Mexico and Guatemala. Amcr. J. Bot. 49: 41-50.

Cho, Y. O. & R. O. Martin. 1971. Resohition and unambiguous identifieation of micro-
gram amounts of 22 lupine alkaloids by sequential use of thin-layer and gas-liquid
chromatography and mass spectrophotometry. Anal. Biochem. 44: 49-57.

Dunn, D. B. 1972. Lupinus mexicanus Cerv. ex Lag. Rhodora 74: 489-494.

IIoLMc;i\EN, P, K. & W. Keuken. 1974. Index Herbariorium. Ed. 6. Regnum Veg. 92:
1-397.

Hu.\nK>LDT, F. H. A. von, A. J. A. Bonpland & C. S. Kunth. 1823. Nova Cenera et

Species Plantarum. Vol. 6. Paris.
Lagasca, M. 1816. Genera Species Plantarum. Madrid.
Nowac:ki, E. 1963. Proc. XI Int. Congr. Genet., The Hague. Vol. 1: 246.

& D. B. Dunn. 1964. Shrubby California lupines and relationships suggested by

alkaloidal content. Genet. Polon. 5: 47-56.
Siiarma, G. 1967. Cuticular variation in Kalanchoe and Datura. Ph.D. thesis, Univ. Mis-
souri—Columbia, Columbia, Missouri.

Smith, C. P. 1948. Correlation-key to group 9, montanus complex. Species lupinorum.
p. 608.

GUAYANIA DAVWSEI AND HEBECLINIUM GENTRYI,
NEW SPECIES FROM NORTHERN SOUTH AMERICA

( EUPATORIEAE-ASTERACEAE ) '

Robert M. Ki\g- and IIahold Robinson-

AUSTUACT

Descriptions and discussions of relationships are provided for Guaxjania davidsei R. M.
King & 11. Robinson from the Aniazonas region of Veneznela and UchccUnium gcntrtji R,
M. King ik H. Robinson from tlie Clioco region of Colombia.

Collecting efforts of two members of the staff of the Missouri Botanical
Garden liave provided two new species belonging to the HebecUnium complex
of the tribe Eupatorieae (King & Robinson, 1971a, 1971b). The complex is
notable for the partially deciduous subimbricate invohicral bracts, the simple
structure of the style base, the smooth corolla lobes, the long nonannulated
anther collars, and for a tendency to bear hairs on the receptacle. The latter
character which was tlie traditional distinction of HebecUnium is, however, not
consistent throughout the group, being absent in some species of HebecUnium
sens. str. and lacking in all species of Guayania. The genera HebecUnium and
Guayania are most easily distinguished by the extremely filiform style append-
ages of the former and by the strongly asymmetric carpopodium of the latter.
With the present additions HebecUnium has 19 species concentrated in the
northern Andes with one species, //. macrophyUum (L.) DC, widely distrib-
uted. Guayania now contains 6 species all restricted to the Guayana Highlands
region and to the surrounding lowlands of the Orinoco and Amazon.

Guayania davidsei R. M. King & H. Robinson, sp. nov. — Fig. 1.

Plantae herbaccae perennes crectac ca. 4 dm altae pauce ramosae. Caules
succulenti anguste jatrophiformes superne abbreviati et in inflorescentia abrupte
terminati glabri et in sicco irregulariter striati. Folia opposita superne congesta,
petiolis 4-6 cm longis; laminae ovatae 9-12 cm longae et 6.0-7.5 cm latae pen-
ninervatae base late obtusae margine serratae apice vix breviter acuminatae
supra et subtus glabrae vel ad marginem sparse puberulae. Inflorescentiae laxe
cymosae, ramis dense puberulis, pedicellis 0.3-1.5 mm longis. Capitula 5.0-5.5
mm longa et 3-4 mm lata; squamae involucri ca. 28 et 4-5-seriatae 1.0-4.5 nun
longae ad 1 mm latae extc^riores late ovatae interiores sensim oblongae vel an-
guste lanceolatae margine et apice distincte scariosae et minute puberulae apice
late vel anguste rotimdatae extus glabrae et plerumque 3-striatae; receptacula
leniter convexa glabra. Flores ca. 22 in capitulo; corollae albae tubulosae ca.
3 mm longae, faucis tubvilosis base indistinctis, lobis ca. 0.3 mm longis et 0.25

* This study was supported in i^art by the Naticinal Science Foundation Grant DEB77-
13457 to the senior author.

" Department of Botany, United States National Museum, Smithsonian Tnstitutioi», Wash-
ington, D.C. 20560.

Ann. Missouri Dot. Gaud. 64: 366-370. 1977.

1977]

KING & ROBINSON- Gl'AVAN/A AND IIEBECLINJUM

367

UNITCP STATES

2779111

nATioNAL mtmAmvH

FicuRE 1. Gtuiijania
tioiuil Ilerbaiiuiu. I'hoto
Natural History.

Pimi% Of VEMEZUeiA
Miras i.XA£ .

AKA?roNAS: 20 KK S of Puerto AyicVichoi
• Uv. 100 m.

Low £ort»c«d hilU &««t of bighM»r itth

At «dt« of Urt* bewitld^r outcrop in forest
m »h«de— at l«a«c iwrtUUy— «« mmil hill**

•GnrrU fiN^n«M 294S a ]iG»vwib«r 1971

(hnidsvi R. M. King & H. Robinson, holotype, United States Xa-
by Victor E. Krantz, Staff Photographer, National Museum of

mm latis, faiicis superioribiis ct loliis extus brc\1ter puberulis, pilis moniliformi-
bus; filanuMita in parte superiorc ca. 0.25 mm longa; thecae ca. 0.8 mm longae;
appendices antherarum oblongae ca. 0.15 mm longae et latae; grana pollinis ca.
18 m^u in diametro; achaenia 1.5 mm longa plcrnmqne in costis breviter setifera;

363 ANXALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

setae pappi ca. 25 teiiues ca. 3 mm longae supernc non latiorcs, cellulis api-
calibus argute aciitis.

Type : Venezuela, aaiazonas

100

forested hills E of highway with savanna at base, at edge of large boulder
outcrops in forest in shade — at least partially — on small hills, heads white, 2
Nov. 1971, Gerrit Davidse 2845 (US, holotype; MO, isotype).

Gxunjania davichci is the second member of the genus with modified stem
structure and abrupt bases on the inflorescences. In this species the stems are
somewhat thickened with more congested leaves in the upper part, providing
the superficial resemblance to a JatrophcL The related G. hulhosa (Arist.) R. M.
King & 11. Robinson has stems completely underground and tuberous. The
latter species also differs by the glanduliferous branches of the inflorescence.
The new species is apparently from low elevations, while G. hulhosa is reported
from talus slopes at 1,500 m elevation on Serrania Paru.

Hebeclinium gentry! R. M. King & H. Robinson, sp. nov. — Fig. 2.

Plantae suffrutescentes erectae ca. 1 m altae? Caules obscure tetragoni
sordidc lanosi. Folia opposita, pctiolis 0.5-1.0 cm longis; laminae ovatae 3.5-7.0
cm longae et 1,8-3.7 cm latae base breviter obtusae vel subrotundatae margine
minute serratae vel duplo-serratae apice distincte breviter acuminatae supra
virides glaljrae vel glabrescentes fere ad marginem et nervis primariis minute
pubcrulis subtus fulviores obscure glandulo-punctatae in nervis et nervulis dense
sordidc puberulae, nervis secondariis pinnatis valde ascendentibus paucis. In-
florescentiae corymboso-paniculatae, ramis dense puberulis, ramulis ultimis 1-5
mm longis. Capitula plerumque 5 mm alta et 3.5-4.0 mm lata; squamae invo-
lucri ca. 40 subimbricatae 4-5-seriatae valde inaequales 1-3 mm longae 0.3-0.6
mm latae anguste oblongae apice rotundatae margine minute dense puberulae
extus plerumque trisulcatae superne in bracteis interioribus sensim dense puber-
ulae; receptacula leviter convexa sparse puberula interne non scleroidea. Flores
ca. 25; corollae albae tubulosae ca. 3 mm longae; faucis tubulosis base indistinctis
glabris, lobis triangularibus ca. 0.4 mm longis et latis extus dense minute puberu-
lis, piHs brevibus in apicem subclavatis; filamenta in parte superiore ca. 0.15
mm longa; thecae ca. 0.8 mm longae; appendices antherarum late oblongae ca.
0.15 mm longae et latae; grana pollinis ca. 20 m/x in diametro; achaenia 1.5 mm
longa sparse minute glandulifera superne pauce sed non breviter sctifera; setae
pappi ca. 45 plerumque ca. 2.5 mm longae ad apicem vix vel non latiores, cel-
lulis apicalibus obtusis.

TvrE: Colombia, ciioco: Alto de Buey, 1,200-1,800 m, tropical wet forest,
weak-stemmed shrub, flowers white, 8 Jan, 1973. A/ Gentry 6- Enrique Forero
7290 (US, holotype; MO, isotype).

IlehccJmium gentriji is most closely related to //. reedii R. M. King & II.
Robinson (King & Robinson, 1972) of adjacent Darien Province of Panama.
The two species share the lanate stems, the pubescent outer surfaces of the
involucral bracts, less hemisplierical receptacles than usual in the genus, and

1977]

KING & ROBINSON— GL7AYAA7A AND IIEBECLIMUM

369

j^- J J ^

u. -

-fj.

UNITED STATCS

PLANTS or COLOMBIA
Depahfmknt or Choco

'■'«»?r-.it«^!^w ' •Mn;fc» fl"?^"'-^ ^h^-t**

2779193

.^--.-l.

NATIONAL HEf»SARIl>M

.-Iv-^v^I -rl ■' , , ^^J ■ ■ -yJ-^r-^M I---. ^ 'M '^ '. h--I -A^ r^JWt

Alto d* EfuAy, Alt. l^OO-iaOO B«t«rs,
tropic li »'«l for»«t

MlS«l3UR> WatAHtKtt-^ aAMIJlM Mi.««ANluM ( MQ )

Figure 2. HchccUniuni gentriji R. M. King & H. Robinson, liolotype, United States
National Herbarium.

leaves with at least partially doubly serrate margins, iiie new species

The

does

have a different appearance from //. reedii by the smaller leaves and the more
neatly oblong snbimbricate invoKicral bracts, but the more significant differ-
ences are the glabrous to glabrescent upper leaf surfaces, the less pubescent

370 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

surface and larger area of pith on the receptacles, the scarcely enlarged tips of
the pappus setae, and the glanduliferous achenes. In H. reedii the upper surfaces
of the leaves are coarsely pilose, the receptacles are shortly but densely puberu-
lous on tlie ridges, the pith of tlie receptacles is very reduced, the pappus setae
have prominently enlarged tips, and the achenes are glabrous.

LlTKHATUUE ClTEU

King, R. M. & II. Roiunson. 1971a. Studies in tlic Eupatoricac ( Astcraceac). XXW'II.

The genus IlcbccUnium. Thytologia 21: 298-301.
& . 1971b. Studies in the Eupatorieac ( Astcraceae ) . XXXIX. .A new genus,

Guaijania, Ph>tol()gia 21: 302-303.

& . 1972, Studies in the Eupatorieae ( Asteraceae). LXXIV. New species

of Critonia, Flcischmanuia and IlehecJiniuni. Phytologia 23: 405-408.

NOTES

A NEW SPECIES OF BAUHINIA {LEGUMINOSAE

FROM PERU

Coiitiiuied revisionary studies of the neotropical species of Bauhinia have
resulted in the discovery of a new species of the genus endemic to Peru.

Bauhinia hirsutissima Wunderlin, sp. nov.

Frutex scandens cirrhosus; rami juvenales fuscop()rpli>ro-hirsuti. Folia anj^uste ovata ad
oblonga, ca. % vel rare % longitudine bilohata, 5-14 cm longa, 4-10 cm lata, apice aciuiu-
nata ad obtusa, basi profunde cordata, margine revoluta, chartacea ad subcoriacea, supra
glabra, infra fuscoporphyro-hirsuta, 9- ad ll-ncr\'ata; petioli 3-5(-7) cm longi; stipulae reni-
formes. Inflorescentiae racemosae, tcrminales vel siil)tenninales et axillarcs, graciles, laxae,
fiiscoporphyro-hirsutae; rhachis 12-40 cm longa; gemmae ovoideae, excrescentibus apicalibus
gemmarum ovato-lanceolatis, 3-5 mm longis, inciirvatis; bracteae et l)racteolae angustc lan-
ceolatac, 5-7 mm longae; pedicelli graciles, 5-10 mm longi; hypantliium cyathiforme ca. 1
mm longum; calyx campanulatus vel levitcr biblabiatus, 15-nervatiis; petala 5, subaequalia,
alba vel snbrosea, 10-12 mm longa, lamina late elliptica, intra glabra, extra dense adprcsso-
pilosa, ungne lamina longiore vel subaequalia; stamina fertilia 10, tubo calycis longitudine
subaetpialia, libra, 5 glabra, 5 versus apicem pilosa, antheris oblongis, ca. 1 mm longis; gynoe-
cium stanu'nibus longitudine plus minusve aequale, stylo brevi, crasso, arcuato, glabro, ovario
dense hirsuto, gynophoro minuto, stigmate oblicjuo. Legumen dehiscens, oblongum ad anguste
obo\atum, apiculatum, ca. 7.5 cm longum, ca, 2.5 cm latum, brunneum glabratum; semina
suborbiculata, 11-12 diam., pagina bebctata puncticiilata, obscure striata, brunnea, cicatricibus
funiculi ramorum longitudine subaequalibus, ca. 1 mm longa.

Tendriled woody vine; young branches reddish brown hirsute, glabrescent
in age, older stems not seen; intrastipular tendrils single or paired, woody, cir-
cinate. Leaves narrowly ovate to oblong, bilobate ca. % or rarely % their length,
5-14 cm long, 4-10 cm wide, the apex of lobes acuminate to obtuse, the base
deeply cordate, the margin revolute, chartaceous to subcoriaceous, glabrous
above, reddish brown hirsute below, the lower surface frequently purplish-
tinged, 9-11-nerved; petioles .3-5(-7) cm long, reddish brown hirsute; stipules
reniform, 5-10 mm long, 2-5 mm wide; intrastipular excrescences other than
tendrils minute. Inflorescences racemose, terminal or subterminal and axillary,
elongate, slender, lax, reddish brown hirsute throughout; rachis 12-40 cm long,
the lower flowers soon deciduous, the inflorescence then frequently with 10-30
flowers on a long-pedunculoid rachis; buds ovoid, 6-8 mm long, the free tips
ovate-lanceolate, 3-5 mm long, incurved; bracts narrowly lanceolate, 5-7 mm
long; bracteoles similar to the bracts, but smaller, attached near or above the
middle of the pedicel; pedicels slender, 5-10 mm long; hypanthium cyathiform,
ca. 1 mm long; calyx campanulate or slightly bilabioid at anthesis, 15-nerved,
each trio of nerves ending at one of 5 ovate-lanceolate appendages at the rim
of the calyx tube, the median nerve extending the length of appendage, the
lateral 2 ending at the base of the appendage or inconspicuously extending up
to V2 its length; petals 5, subequal, white or faintly tinged with pink, 10-12 mm
long, the blade broadly elliptic, 5-7 mm long, 3-4 mm wide, glabrous internally,
densely appressed brown-pilose externally, the claw longer than to nearly equal-

372 ANNALS OF THE MISSOURI B0TANIC:AL GARDEN [Vol. 64

ling the length of the blade, brown pilose; fertile stamens 10, ± equalling the
calyx tube, free to the base, alternate ones slightly shorter, the longer 5 gla-
brous, shorter 5 brown pilose towards the tip, the filaments arcuate, tlie anthers
oblong, ea. 1 mm long, white-pilose; gynoecium ± equalling the stamens, the
style short, thick, arcuate, glabrous, the ovary densely brown-hirsute, the gyno-
phore not evident, the stigma oblique, slightly differentiated from the style.
Fruit a dehiscent legume, oblong to narrowly obovate, apiculate with a per-
sistent style, ca. 7.5 cm long, ca. 2.5 cm wide, dark brown, glabrate, gynophore
not seen; seeds suborbicular, ca. 12 mm long, ca. 11 mm wide, the surface dull,
puncticulate, obscurely striate, dark brown, funicular-branch scars subequal,
ca. 1 mm long. Chromosome number unknown.

Type: Peru, loreto: Fortaleza, near Yurimaguas, ca. 140 m, forest, Dec.
1932, G. Klug 2800 (US, holotypc- F, MO, NY, isotypes).

Specimens examined: Peru, loreto: Qiiebmda Shanucc above Yurimaguas, Croat 1806^
(MO, duplicates to be distributed). Lower Rio Huallaga, Mhp isr Smith 27601 (F, NY,
US), 28302 (F, NY, US). Fortaleza, Yurimaguas, Ll Williams 4216 (F, US), 4485 (F, US).

Distribution: Known only from near Yurimaguas on the Rio Huallaga, Lo-
reto, Peni. It occurs in forests at elevations of about 140 meters. Flowering
material lias been collected from July through December with nearly mature
fruiting material collected in December.

The newly described species is most closely related to the widespread and
highly variable B. glabra Jacq., but is distinguished by the following combina-
tion of characters: young branches, inflorescences, and lower leaf surfaces con-
spicuously deep reddish-brown hirsute; inflorescences elongate, slender, lax,
with flowers distantly arranged; flower buds ovate, with incurved lanceolate

apical

5-7 (-8) mm long, without purple spots. In

contrast, the vestiture of B. glabra, when approaching that of B. hirsutissima, is
pilose and has a coppery sheen. One local race of B. glabra whose pubescence
is very similar to that of B. hirsutissima is restricted to Panama and differs in
all other respects. The flowers of B. glabra have petal blades 10-20 mm long,
one of which is usually conspicuously marked with purple spots, although some-
times obscure in local races. In B. glabra the flower buds are lanceolate with
setiform or rarely lanceolate apical excrescences and the inflorescences, if elon-
gate, are strict and with the flowers more closely arranged.

Bauhinia reflexa Schery, a Panamanian and Colombian species, has the vesti-
ture of its leaves, young branches, and inflorescence rachis in addition to the
purple-tinged undersides of its leaves like B. hirsutissima^ but differs in all
other respects.

Finally, B. hirsutissinui also superficially resembles B. hllipiana Standley
and to a lesser degree B. vulpina Rusby and B. porphyrotricha Harms, but dif-
fers slightly in nearly all characters. Specimens determined to be B. hirsutis-
sima have frequently been identified by other workers as B. porphyrotricha.
Examination of type material of B. hllipiana, B. vulpina, and B. porphyrotricha
reveal that these species are best placed in synonymy with B. glabra sensu lato.

1977]

NOTES

373

I gratefully acknowledge John D. Dwyer, St. Louis University and Missouri Botanical
Garden for critically reading the manuscript, especially the Latin description, and the cura-
tors of the herbaria \\'ho loaned specimens for this study.

Richard P, WnmlerUn, Department of Biolo<s,i/, Universitij of South Florida,
Tampa, Florida 33620.

ALNUS MARITIMA MUHL. EX NUTT, NOT

ALNUS METOPORINA FURLOW

The new name, Ahms metoporinu Fiirlow, proposed in a recent issue of this
journal (Furlow, 1976) to replace the long-recognized Almis maritima Mulil. ex
Nutt., is unnecessary according to Article 55 of the International Code of Botani-
cal Nomenclature (Stafleu et al., 1972) which states:

When a species is transferred to another ^enus or plaeed under another generic name
for the same j^enus without change of rank, the specific epithet, if legitimate, nmst he retained
or, if it has not heen retained, must be reinstated unless one of the following obstacles exists:

(1) The resulting binary name is a later honumym (Art. 61) or a tautonym (Art. 23).

(2) An earlier legitimate specific epithet is available (but see Arts. 13f, 58, 59, 72).

The genus Bettda-Almts was validly pubHshed by Humphry Marshall in
his Arhustrum Americamim (1785), and tliree species, including B, maritima,
placed in it. Marshall's description of B. maritinui is sketchy, mentioning only
the height of the plant, the "long and narrow" leaves, and the very distinctive
August anthesis. Marshall did not cite a type nor did he keep an herbarium,
and so we do not know on what material the species was based. However, we
have no doubt that the plant was the same species that was later described by
Henry Muhlenberg in an unpublished manuscript, and subsequently validly
published by Thomas Nuttall (1842) in the first volume of his Sijha. Alnm
maritima was described as a new species in the genus Ahms, based not upon
Marshall's name [as assumed by some authors, see Little (1953)] but upon
Muhlenberg's manuscript name.

We know that Mulilenberg was well acquainted with Marshall's work for
in a letter to William Bartram dated 10 December 1792 (Darhngton, 1849),
he wrote: "Marshall has given me some satisfaction, but his Arbustrum wants
some emendations. Any observations that way where you think he is wrong,
or where another name miglit have been given, would be so pleasing to me.

It is likely that Muhlenberg may have recognized that his species was Bctida-
Alnus maritima of Marsliall, and that he intended to make the transfer to Ahms,
and to typify the sj)ecies on the Bartram collection he had. However, there is
nothing in the mantiscript presei-ved at the Philadelphia Academy of Sciences
Library to suggest this, and we may only speculate on Muhlenberg's intentions.

Nuttall apparently put even less faith in Marshall's work, for he ne\^er men-
tioned it directly in the first volume of his SyJva, Marshall is cited there only
once, and then in synonymy under Nuttall's Cartja microcarpa. Since a Muhlen-

>>

374 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

berg synonym is also cited, it is possible that the Marshall reference was copied

by Nuttall from Muhlenberg.

Furlow (1976) is correct in his contention that to transfer Marshall's earlier
epithet from Betuhi-Alnus to Almis would create a later homonym (prohibited
by Articles 55[1] and 64). His creation of a new name for B. riuiritiina to avoid
the creation of a homonym, however, is not correct. Article 55(2) requires that
if an earlier legitimate specific epithet is available for the species, it must be
adopted. Alnus maritima as proposed by Nuttall (1842) is legitimate, is avail-
able, and antedates Furlow*s A. metoporina. Ahms maritima Muhl. ex Nutt.
must be retained for this plant.

We propose to lectotypify Alnus maritima on the Pickering specimen (PH)
on which Nuttall's (1842) description and Plate X (bis) were based, and not
on the Bartram specimen (#477 in Flerb. Muhlenberg, PH) which was also
seen and indirectly mentioned by Nuttall. Furlow has supplied a neotype for
Marshairs Betula-Alnus maritima,

Wc would like to thank tlic staff at the Philadelphia Academy of Sciences for their
many courtesies, and Dr. Dan H. Nicolson of the National Museum of Natural History,
Smithsonian Institution, for his advice on the provisions of the Code discussed herein.

LiTEiiATuiiE Cited

DAHLiN(n().\, W. 1849. Memorials of John Bartram and Humphry Marshall. Lindsay &

Blakiston, Philadelphia.
Furlow, J. J. 1976. Nomenclatural changes in Alnus (Betulaceae). Ann. Missouri Bot.

Card. 63: 380-381.

Little, E. L, 1953. Check list of native and naturalized trees of the United States (inchid-
ing Alaska), Agric. Handl). 41: 1-472.

Mahshall, H, 1785. Arhustrum Americanum: The American Grove. J. Cruksliank, Phila-
delphia.

Nuttall, T. 1842. The North American Sylva. Vol. 1. Smith & Wistar, Pliiladelphia.

Stafleu, F. a. et al. (editors). 1972. International code of botanical nomenclature. Reg-
num Veg. 82: 1-426.

Virginia M. Stibolt, C. Rose Broome and James L. Reveal, Department of
Botamjy University of Maryland, College Park, Maryland 20742,

NEW TAXA AND COMBINATIONS IN THE GENUS LASIACIS

(GRAMINEAE)

The following new taxa and new combinations are published in advance of
my systematic treatment of the genus Lasiacis in order to make them available
for use in other publications. A new species, Lasiacis nigra Davidse, has also
been published (Davidse, 1974). Full explanations for my treatment will be
given in the forthcoming publication.

Lasiacis divaricata (L.) Hitchc. var. austroamericana Davidse, var. nov.

Ab L. divaricata var. divaricata spiculis globosis et ramis paniculae ascen-
dentibus differt.

1977]

NOTES

375

Type: Brazil, mixas gehais: Near Santa Barbara do Caparaco, streamsidc,
suffriitesccnt, 3 m high, climbing and rooting at lower joints, "canaviera/' 21
Nov. 1929, Mexia 4007 (NY, holotypc; F, Gil, UC, US, isotypcs).

Lasiacis divaricata (L.) Hitchc. var. leptostachya (Hitchc.) Davidse, comb,
et stat. nov.

La.siacis leptostachya Hitchc, Contr. U.S. Natl. Herb, 22: 19. 1920. Type: Nicaragua.

Jinotepe, jungle, 500 m, stout central canes, branches more or less wliorled and slender,
floral branches conspicuously flexuous, panicles all small, 7 Nov. 1911, IlitchcocJc 8718
(US, holotype).

Lasiacis grisebachii (Nash) Hitchc. var. lindelieana Davidse, var. nov.
Ab L. grisebachii var. <^mehachii laminis latioribus (1.5-2.0 cm) differt.

Type: Cuba, iiabana: Lomas de Camoa, in siloa locis umbrasis satis frc-
qiicns, 27 Nov. 1921, Ekman 13530 (US-1295003, holotype; F, NY, US-1502317,
isotypcs ) .

Lasiacis oaxacensis (Griseb.) Hitchc. var. maxonii (Swallen) Davidse, comb.
et stat. nov.

Lasiacis maxcmii Swallen, Ann. Missouri Bot. Card. 30: 231. 1943. Type: Panama. c:nauguj;
Vicinity of El Boquete, in thickets along wet trail, 1,000-1,300 m, 2-8 Mar. 1911,
Maxon 4999 (US, holotype; US, isotype).

Lasiacis rugelii (Griseb.) Hitclic. var. pohlii Davidse, var. nov.

Ab L. nigelii var. rugelii spiciilis globosis minoribns, inflorcsccntiis minori-
bns, ramis paniculae ascendentibns vel patentibns, pseudopetiolis amplis differt.

Tyt'e: Costa Rica, cartago: 1 km NE of Pejibaye, along Rio Pejil)aye,
growing at base of tree, ca. 700 m, 2 Nov. 1968, Pohl ir Davidse 11478 (ISC,
holotype; CR, EAP, K, MO, US, isotypcs).

Lasiacis ruscifolia (H.B.K.) Hitchc. var. velutina (Swallen) Davidse, comb,
et stat. nov.

Lasiacis velutina Swallen, Ceiba 4: 288. 1955. Type: IIonuuilvs. morazan: Vicinity of El

Zamorano, road to San Antonio, 17 Oct. 1951, Swallen 10834 (US, holotype).

Lasiacis sorghoidea (Desv.) Hitchc. & Chase var. patentiflora (Hitchc. &
Chase) Davidse, comb, et stat. nov.

Lasiacis patentiflora Hitchc. ik Chase, Contr. U.S. Natl Herb. 18: 338. 1917. Type: ToiiACO:
Center of island, ed^e of woods on mountainside, 20 Dec. 1912, Hitchcock 10268 (US-
865566, holotype; US-975660, isotype).

Literature Cited
Davidse, G. 1974. A new species of Lasiacis (Gramineae). Phytologia 29: 152-153.

— Gerrit Davidse, Missouri Botanical Garden, 2345 Tower Grove Avenue, St.
Louis, Missouri 63110,

376

ANNALS OF TllK MISSOUIU DOTAMCAL GARDEN [Vol. 64

SOLANUM ARMENTALIS: A NEW SPECIES

FROM COSTA RICA

Solanum armentalis J. L. Gentry & D'Aicy, sp. nov. — Fig. 1.

Frutfx 1 in altiis imMinis oninino conft'itiin stclUito-pubesccns, trichomatibus fulvis, scssili-
Inis, radio medio qiiani radiis lateralibus iiiulto longiove; foliis singularibus vel in gcminis
inaequalil)iis, pilis siibtiis cvebeniniis, supra sparsis practcM- secus costam; foliis majoribiis ovatis
vel ovato-clliplicis, 13-14 cm longis, 6-7 cm latis, apice longo-acuminatis, basim rotundis,
pctiolo 5-7 mm longo, foliis minoribus o\atis vcl late cllipticis; infloresccntia lateralibus foliis
opposita, pedunculo supra 5 cm bifurcato; floribus parvis, calyce 2-2.5 mm longo prope basim
fisso, lobis o\atis, 1.5-2.2 mm longis, apice acutis, staminibus aequalibus, filamcnlis glabris,
basim connatis, antbcris 2.5 mm longis, poris largis terminalibus aperientibus, ovario glabro,
stylo glabro; acino ignotci.

SJiruh 1 m tall, unarmed, densely stellate-pubescent throughout, the hairs
sessile, yellowish brown with the central ray greatly exceeding the lateral rays.
Leaves solitary or in pairs, unequal in size, similar or different in shape, densely
pubescent beneath, sparsely pubescent above except along the midvein; larger
leaves ovate to ovate-elliptic, the blade 13-14 cm long, 6-7 cm wide, the apex
long acuminate, the base rounded, the petiole 5-7 mm long; smaller leaves ovate
to broadly elliptic, the blade 1.8-4 cm long, 1.2-2.8 cm wide, the petiole 2-3 mm
long. Inflorescence both lateral and opposite the leaves, several llowered; pedun-
cle unbranched for 3 cm, bifurcate; pedicels slender, 10-12 mm long. Flowers
small; calyx 2-2.5 mm long, parted to near the base, the lobes ovate, 1.5-2 mm
long, the apex acute; corolla white, the limb 10-11 mm wide, parted to near the
base, the lobes 4-4.5 nun long with sessile stellate pubescence externally; stamens
equal, the filaments glabrous, ca. 0.7 mm long, basally connate, the anthers 2.5
nun long, the pores large; ovary glabrous, the style glabrous, exceeding the sta-
mens. Fruit unknown.

'e: Costa Rica. prov. puntahknas: Open forest 1 mi due south of San

Ti

J

1967, Peter IL Raven 218S7 (MO-23()419S,

holotype; F, isotype).

Additional collection cxann'ned: Costa Rica. puov. san jose: Vicinity of El General,
edge of forest, 700 m, Jan. 1939, Alcxamler F. Skufch 3919 (MO).

Solanum armentalis is recognized from other species of the genus in the

Central American flora by the softly tomentose leaf undersides, long, slender

peduncles and slender pedicels, and small white flowers which have calyces
divided nearly to the base. The indumentum is of dense, sessile stellate hairs,
with the midpoints many times longer than the several short lateral arms.

This species is a member of subgenus BrevantJwrum, Plowever, it is not
easily placed in a section within the subgenus. At first glance the plant is sug-
gestive of Solanum exfensum Bitt. and S. cordovense Sesse & Mo?., but the calyx
is different in shape and not accrescent to judge from the one immature fruit on
the Skutch specimen. Also, the indumentum and inflorescence are different.
Solanum gemellum Sendt, and S. megalochiton Mart, of eastern Brasil have caly-
ces similar in shape, but they arc conspicuously accrescent in fruit. While the
above similarities arc suggested, they do not imply a close relationship to the
species described here.

1977]

NOTES

377

Figure 1. Solantnn anunitaUs J. L. Gentry & D'Arcy. — A. Flowering branch (XV-).
B. Stem showing stellate hairs (x3).— C. Flower (X^Vj). [After Raven 21887 (MO).]

378

ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

700

Pacific side of the isthmus. Perhaps further collections will provide an insight
into its ecological requirements. The locality of San Vito suggests that this spe-
cies may be found in the nearby Chiriqui Mountains of Panama.

Because no single character delimits this well-marked species, the specific epi-
thet does not relate to the collector, ct)llection locality, or morphological features.

■Johnnie L. Gentry, Jr., Behh Ilerbarium, Department of Botanij-Uicrohiologij,
University of Oklahoma, Norman, Okhihoma 73019 and William G. D^Arcy, Mis-
souri Botanical Garden, 2345 Tower Grove Avenue, St. Louis, Missouri 63110.

HERBERTIA (IRIDACEAE) REINSTATED AS A

VALID GENERIC NAME

The genus Ilerhertia was described in 1827 by Robert Sweet for a New
World genus of Iridaccae. As currently circumscribed (Goldblatt, 1975), the
genus is a small one of approximately six species centered in temperate South
America from Uruguay to Chile with a subspecies of a South American taxon
occurring in the southern United States.

The existence of the similar name Ilerhertus Gray (also used in the form
Ilerherta) publi.shed in 1821, prompted several authors including myself to
regard Ucrhertia as a later homonym and therefore to reject it. Following Kunt7.e
(1898) who fir.st suggested that IIer])ertia be considered a homonym, both
Foster (1945) and Raveima (1968) among others, accepted Alophia Herb, (dat-
ing from 1840) as the valid name for the genus. Subsequently I discovered
(Goldblatt, 1975) that the type species of Alophia had been misinterpreted and
was in fact a species of what was then known as Eustylis. I therefore proposed

f

1

cics and for the South American representatives of Ilerhertia.

Recently it has been suggested both in print (Florschutz & Grolle 1975) and
to me personally that Ilerhertia should not have been rejected and that I sliould
carefully consider Article 75 of the Botanical Code of Nomeiielature, This article
deals with names of similar but not identical spelling, and recommends rejection

only in cases of likely confusion. The article reconunends the rejection of exam-
ples such as CohnneUa and Coin me I Ha and Eschweilera and Eschweileria as
being too similar and thus likely to cause confusion. Other examples are Pel-
tophorum and Peltophorus, Iria and Iris, neither of which are to be considered
homonyms and therefore both forms of these words are available for usage for
different genera.

Ilerhertia seems to fall into the latter category, being sufficiently different

in orthography to avoid any possibility of confusion. Florshutz & Grolle (1975)
support this view, as do several colleagues with whom I have discussed the

1977]

NOTES

379

question. Therefore, I propose reinstatement of Herhertia and tlie synonymizing
of Triftirda. This treatment involves some new combinations as follows:

1. Herbertia lahue (Molina) Goldbl., comb. nov. Basionym: Ferraha lahuc,
Molina, Sagg. Stor. Nat. Chile, ed. 2: 110. 1810.

The subspecies of this taxon are to be cited as follows:

la. H. lahue subsp. amoena (Griseb.) Goldbl., comb. nov. Basionym: Iler-

W

lb. H

1840: tab. 3779. 1840

1879.
: Trif

2. Herbertia tigridioides (Hick.) Goldbl., comb. nov. Basionym: Alophia
tigridioicks Hick., Darwiniana 1: 116. 1924.

The four remaining species were cither originally placed in Herbertia, or

have in the past been transferred to the genus. These arc: //. pulchella Sweet,

//. amatorum C. H. Wright, H. hauthaUii (O. Kuntze) K. Schum, //. bra.siUensis
Baker.

LiTEHATURE ClTEI)

Flohscuutz, p. a. & R. Grolle. 1975. Hcrhertus Cray 1821, Herhertia Sweet 1827 and

Ucrhcrta Gray nnit. Lindb. 1875. J. Bryol. 8: 479-481.
Foster, R. C. 1945. Studies in the Iridac'eae III. Contr. Gray Herb. 155: 3-55.

Goi.DRLATT, p. 1975. Revision of the Inilboiis Iridaeeae of North America Brittonia 27-
373-385.

KuxrzE, C. E. O. 1898. Revisio Genenim Plantanini. Vol. 3. Leipzig.
Ravenna, P. 1968. Notas sobre Iridaeeae. III. Bonphmdia 2: 273-291.

Veter Goldbhitt, B. A. Krukoff Curator of African Botany, Missouri Botanical
Garden, 2345 Tower Grove Avenue, St. Louis, Mi.<>souri 63110.

Garden

64

J

, Publication
IS Available in

..from

Xerox

University

l\/llcrofilms

300 North Zeeb Rd.,
Ann Arbor, Mich. 48106
(313) 761-4700

PLEASE
WRITE FOR
COMPLETE
INFORMATION

Volumes 1-48 of the Annals of the Missouri Botanical Garden are now available in com-
! sets in microform from Xerox University Microforms. However, original printed copies
dividual issues of Vol. 1-48 listed below are also available. In addition complete volumes

are for sale. These back issues may be ordered from

49-62

1041

Vol.

1
2
2
9

19

19

20
21

No.

2

1-2

3

1

2-3
4

3

3

Price

8.00
12.00
8.00
8.00
12.00
8.00
8.00
8.00

Vol

22
23
23
24
26
27
30
30

No. Price

Vol. No, Price

3

3
4
1
1
1

2
3

8.00

8.00
8.00
8.00
9.00
9.00
9.00
9.00

31

32

33
35
36

36
36
36

1

1

2
4

1

2
3
4

9.00
9.00
9.00
9.00
9.00
9.00
9.00
9.00

Vol.

45
45
46
47
47
48
48
48

No. Price

3

4

4
1

3

2

3
4

9.00
9.00
9.00
9.00
9.00
9.00

9.00
9.00

Systematics of Moraea (Iridaceae) in Tropical Africa Peter Goldhlatt ..__ 243
A New Classification of Ficus William Ramirez B. - 296

Studies in Bignoniaceae 25: New Species and Combinations in South

American Bignoniaceae Alivijn H. Gentry — 311

New Records of Apocynaceae for Panama and the Choco Altcyn H. Gentry 320

New Taxa and Combinations in Eragrostis (Poaceae) John T. Wither^oon 324

Realignment of the Species Placed in Exogonium (Convolvulaceae) Daniel

F. Austin -— 330

The Lupinus montanus Complex of Mexico and Central America David

B, Dunn ir William E, Harmon - — 340

Guyania duvidsei and Hebeclinium gentryi. New Species from Northern

South America (Eupatorieae — Asteraceae) Robert M. King ir Harold
Robinson „„ _ . 366

NOTES

A New Species of Baiihinia (Legvmiinosae) from Peru Richard P.

Wunderlin - 371

Alnus nmritima Muhl ex Nutt., not Alnus metoporina Furlow Virginia

M. Stibolt, C. Rose Broome ir James L, Reveal 373

New Taxa and Combinations in the Genus Lasiacis (Gramineae) Gerrit

Davidse 374

Solanum armentalis: A New Species from Costa Rica Johnnie L.

Gentry ir William G. D'Arcy 376

Herbertia (Iridaceae) Reinstated as a Valid Generic Name Peter

Goldbhtt ._._ -- 378

JFTHE

VOLUME 64

l^i-

\

\ '

i ^

.'*-l

^

l.ti

^ $

.t ^

■ 2^

CONTENTS

Ml

1977

NUMBER 3

wv

gsoow

BOT^«>C»

»

J\l^

191B

\

f

tww^

ubmx

■^^H,.

^•*^

SEIWA-EN JAPANESE GAHDEN, MISSOURI BOTANICAL GARDEN

Systematics of Oenothera Sect. Kneiffia ( Onagraceae ) Gerald B. Straley .. 381

The South American Species of Oenothera Sect. Oenothera (Rainuinnia,
Renneria; Onagraceae) Werner Dietrich — ^ - —

425

Wood Anatomy of Onagraceae: Additional Species and Concepts Sherwin
Carjquist - — —

62

i

A New Species of Lopczia (Onagraceae) from Sinaloa, Mexico Peter
H. Raven --

638

Reinterpretation of the Type of Godetia hottae Spach (Onagraceae) Peter

H. Raven ir Dennis R. Parnell

642

Reproductive Structures and Evohition in Ludicigia (Onagraceae). I. An-

droecium, Placentation, xMerism Richard H. Eijde - ^^

VOLUME 64

1977
NUMBER 3

OF THE

The Annals contains papers, primarily in systematic botany,
contributed from the Missouri Botanical Garden. Papers originating
outside the Garden will also be accepted. Authors should write the
editor for information concerning arrangements for publishing in
the AxxALs.

Editorial Comaiittee

Geruit Davidse, Editor-in-Chief
Missouri Botanical Garden

Barbara B. Pesch, Editor
Missouri Botanical Garden

W. G. D'Arcy, Editor— Floja of Panama

Missouri Botanical Garden

John D. Dwtter

Missouri Botanical Garden ir St. Louis University

Peter Goldblatt
Missouri Botanical Garden

Published four times a year by the Missouri Botanical Garden Press,

St. Louis, Missouri 63110.

Armals

P.O. Box 368, 1041 New Hampshire, Lawrence, Kansas 66044.

Subscription price is $40 per volume U.S., Canada, and Mexico,

?45 all other countries. Four issues per volume.

Second class postage paid at Lawrence, Kansas 66044

© xMissouri Botanical Garden 1978

OF THE

mm

VOLUME 64

1977

NUMBER 3

SYSTEMATICS OF OENOTHERA SECT. KNEIFFIA

(ONAGRACEAE)^

Gerald B. Straley-

Absthact

Based on cytology, morphology and field studies, a reevaluation of Oenothera sect.
Kneiffia, the eastern North American day-flowering Oenotheras or sundrops, is presented, wiUi
five species recognized. Subsect. Peniophylluiu contains a single annual species, O. linifoUa,
of the southeastern United States, which is a self-compatible diploid (n = 7). Subsect!
Etikneiffia contains one annual self-compatible diploid {n = 7), O. spachiana, largely of the
southern Mississippi Valley, and three perennials, O. percnnis, O. fruticosa, and O. pilosella.
Oenothera perennis is a self-compatible diploid (n — 7), complex heterozygote, forming a ring
of 14 chromosomes at meiotic metaphase I and having about 50% pollen sterility. It is dis-
tributed from Newfoundland to Manitoba and south to North Carolina and Missouri. The t\\ o
other perennials, each \\'ith two subspecies, are self-incompatible polyploids that seem to lack
chromosomal translocations but fonn numerous rings of chromosomes at meiotic metaphase I
owing to their autopolyploidy. Oenothera pilosella subsp. pilosella, a taxon largely of the
midwestern United States, is known only as an octoploid (n = 28). Oenothera pilosella subsp.
sessilis, also octoploid, is treated as a second subspecies. It is very rare at present and is
found in remnant prairies of the lower Mississippi River Valley. It differs from subsp. pilosella

' I am especially grateful to Peter II. Raven for originally suggesting this project and for
his untiring encouragement throughout the course of this study. This study has been sup-
ported in part by grants from the U.S. National Science Foundation to him. I am also grateful
to the staff of the Missouri Botanical Garden for their assistance during the summer of 1976,
where this project was completed. I would like to thank Robert M. Lloyd for his encourage-
ment and assistance, especially during the initial stages of this study, as a part of my M.S.
degree research at Ohio University. I would also like to thank Charles W. Hagen, Adolph
Hecht, Brij M. Kapoor, and Lytton J. Musselman for helpful suggestions, contributions of
unpublished information, and other assistance; Peter Hoch for taking the SEM photographs;
Julie Wilson for assistance with the maps and plates; and Lesley Bohm for the drawings (Figs.
81, 82). I was assisted in the field by John Ayers and Edward Minnick and I would like to
thank them for their assistance.

Lastly, I would like to thank the staffs of the following herbaria for the use of their
herbaria during the course of this studv and for the loan of material for study: BH BHO
BM, CAN, CM, CU, DAO, DUKE, F, FLAS, FSU, GA, Gil, lA, ILL, ILLS, ISC, KE, LINN,
MASS, MICH, MO, MSC, NA, NCU, NEBC, NFLD, NO, NY, OKLA, OKL, OS, P, PAC,
PH, POM, RSA, SIU, SMU, TENN, TEX, TRT, UARK, UBC, US, USE, VDB, VT, VPI,
WIS, WVA, YU, and Louisiana Tech University, Lynchburg College, Old Dominion Univer-
sity, St. Mary's University, and Western Kentucky University.

^ Department of Botany, LIniversity of British Columbia, Vancouxer, British Columbia
V6T IW5, Canada.

Anx. Missouri Bot. Card. 64: 381-424. 1977.

og2 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vol. 64

in being nonrliizonuitous; it has narrower, ascending leaves, more ellipsoid, nearly sessile cap-
sules, and strigose pubescence throughout. The widespread polymorphic O. fruticosa consists
of a more southern and lower elevation subspecies, subsp. fruticosa, which is tetraploid (n
14) and hexaploid (n = 21), and a more northern and higher elevation subspecies, subsp.
gJauca (O. tetnigona Roth), which as far as known is entirely tetraploid.

The Ona^^raceoiis wniis Oenothera has been the subject of classical studies

of reciprocal translocations and their effects on the evolution of a group of
plants. Much of the early cytogenetic and cytotaxonomic work was done on the
North American species of sect. Oenothera (Cleland, 1972), Other portions of
the genus have received less attention, but studies of Raimannia (Hecht, 1950),
of several subgenera by Ilagen (1950) and Gregory & Klein (I960), and of the
South American species of sect. Oenothera by Dietrich (1977) are among the
most important works. Subgenera are treated in tins paper as sections consis-
tent with the classification employed elsewhere in the family (Lewis & Lewis,
1955; Raven, 1964; Raven & Gregory, 1972b).

Oenothera sect. Kneiffia has received httle biosystcmatic attention. Ta\o-
nomic treatments of Kneiffia based largely on morphological characters have
resulted in a confusing array of specific, subspecific, varietal, and form names.
Even the most recent, rather thorough revision based upon morphological fea-
tures by Munz (1965) reflects the difficulties in circumscribing species and
particularly in delimiting infraspecific taxa, A recent biometrical and biochem-
ical study (DeTurck, 1969) has also failed to clarify the taxonomic problems in
this section. Records of chromosome numbers and meiotic pairing in species of
sect. Kneiffia are few and scattered in the literature, and many of these, partic-
ularly the older records, are from plants grown in botanical gardens for which
the ori<;inal source of nniterial is not known, and for which no voucher specimens

were made. Tliere are no previous reports of complex heterozygosity in sect,
Kneiffia, although polyploidy has been reported a number of times, especially
by Gregory & Klein (1960) and by Laws (1963).

Plants in this section are characterized by bright yellow flowers that open
after sunrise, relatively short fk)ral tubes, and 4-angled to 4-winged fruits borne
on a short stipe. They are distributed from Newfoundland to northern Florida
and westward to southeastern Manitoba and eastern Texas. Natural populations
usually have easily discernable limits, particularly in the perennial species, and
are commonly composed of a few to rarely more than several dozen plants. The
two annual species tend to be found in scattered populations with less distinct
boundaries.

Materials and Methods

This study was begun in 1973, initially in an effort to evaluate the cytology
of the section as a basis for understanding the morphological discontinuities,
particularly among the perennial species. Extensive field studies were made from
1973 to 1976 throughout much of the distribution of the section. Vouchers are

deposited at MO.

Flower buds were fixed in the field in 1:3 glacial acetic acid : ethanol and

1977] STRAhEY— OENOTHERA SECT. KNEIFFIA

383

later transferred to 95% ethanol and refrigerated. The buds were hydrolyzed in
1 : 1 concentrated hydrochloric acid : 95% ethanol at room temperature until they
were translucent (10-20 minutes depending on bud size) to remove starch grains
from the pollen mother cells (Lewis & Lewis, 1955). After hydrolysis, the anthers
were removed, stained with aceto-carminc and macerated with iron needles. A
drop of Hoyer's medium (Beeks, 1955) was added to make the slide peraianent.
The preparation was squashed and later the coverslip was ringed with Diaphane.
Observations of meiotic divisions of microsporocytes were made during diakinesis
or mctaphase I using a positive phase-contrast microscope.

Pollination systems and floral biology were studied in several natural popu-
lations, and seeds or plants were collected and grown in the greenhouse or
experimental garden for compatibility studies and further observation of charac-
ters. Field collections have been supplemented by examination of over 11,000
herbarium specimens, including large enough samples of all taxa to be of great
value in comparative morphology. Where available, herbarium voucher speci-
mens and permanent slides of cytological material from previously reported
chromosome studies of sect. Kneiffia were borrowed for further study.

Based on a reevaluation of morphological characters and the results of cyto-
logical and breeding studies, a revised classification of the taxa is presented here.

Historical and Taxoxomic Considerations

Due to their widespread distribution and abundance in eastern North Amer-
ica, the species of sect. Kneiffia have been the subject of study for more than
200 years. Linnaeus described the first species, Oenothera perennis, in 1753.
The classification of this group subsequently has undergone many reorganiza-
tions owing to the difficulties in assigning populations to clear-cut morphological
species. The greatest taxonomic problems are in the two perennial, polyploid
species treated by Munz ( 1965 ) and in most recent floras ( e.g., Gleason, 1963;
Fernald, 1950; Radford et al., 1968) as O. fruticosa L. and O. tetraQom Roth'

1

Michx. [= O. fruticosa L. subsp. fruticosa and O. frutic

L. subsp. glauca