%.,^^

J

I

\

)

/

\

« \

i '

/ y

Vol

ume

94

/ /

Number

/

Volume 94, Number 1 April 2007

Annals of the

Missouri Botanical Garden

The Annals

s

b

buted from the Missouri Botanical Garden, St. Louis. Papers originating outsid

the Card

(

be acceDted. All

J

Bed

dependent reviewers. Authors should write the Managing Editor for information

for Dublishinii in the Annals. Instructions to A

b

K

of th

:)f

01 eac

h

umc

y

labl

www.m])gpress.org.

Editorial Conmiittee

Victoria C. HoHovvell

Scientific Editor, Missouri Botanical Garden

Beth Parada

Managing Editor, Missouri Botanical Garden

Diana Gunter Associate Editor, Missouri Botanical Garden

Barbara Mack Editorial Assistant,

Missouri Botanical Garden

Monica Anderson MBG Press Assistant, Missouri Botanical Garden

Thsan A. Al-Shehbaz Missouri Botanical Garden

Gerrit Davidse

Missouri Botanical Garden

Roy E. Gereau

Missouri Botanical Garden

Peter GohJblalt

Missouri Botanical Garden

Gord

on

McPl

lerson

Missouri Botanical Garden

P. Mick Richardson Missouri Botanical Garden

Charlotte Taylor Missouri Botanical Garden

ricnk

van uer

d

Werff

Missouri Botanical Garden

For subs<'ription information conlart Ann MS

OFTHK MlSSOl HI tJOTANICAL Gardfa, % Allen Mar-

ketiiifj; & Managenuml, I^.O. Box 18*17. Lawrence, KS 66044-8897. Snl>scription pHee for 2007 is $165 per volume U.S., $175 Canada & Mexico, $200 all olher coUiilrievS. Four issues per voluni(\ The journal Novon is inelnded in tlie subscription price of I he Annats.

annals@niobot.org (editorial qiuMies)

The Annals of 'vwv. Missoiin BorwicAi. Cvkokn

(ISSN 0026-6493) is published quarterly I>y the Missouri Botanical Garden, 2345 Tower Grove Avenue, St. Louis, MO 63110. Periodicals post- age paid at Si. bonis, MO and additional mail- ing offices. Posi.vrxsTER: Send address ehanges to

Annals of ihk Misson^i Hotanicm Gardln, %

Allen Marketing & Management, P.O. Box 1897,

Lawrence, KS 66044-8897.

'YheAnnah are abstracted and/or indexed in AGRICOLA (tlirough 1994), \I^^ Onl in.s BIOSIS®, CAB Ab- stract/Global Heahli databases, ingenta, ISI® databases, JSTOR, Research Alert®, and Sei Search®.

Typesetting by Allen Press, Inc. using A(K'ent3B2 automatic paging system.

© Missouri Botanical Garden Press 2007

The mission of the Missouri Botanical Garden is to discover and shan^ knowledge about plants and their environment, in order to preserve and enrich tile.

@ This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).

Vol

ume

94

2007

ISSOURl BOTANICAL

MAY

! 2007

GARDEN LIBRARY

nn

of th

Missouri

Botanical Garden

A PHYLOGENETIC ANALYSIS OF ALYXIEAE (APOCYNACEAE) BASED ON RBCU MATK, TRNL IN IRON, TRNL-Y SPACER SEQUENCES, AND MORPHOLOGICAL CHARACTERS

Mary E. Eudress,~ Raymond W. J. M. ran dcr Ham/' Siircrt Nilsson,^ lAiare Clveyrel/ Mark W. Chase/' Bengl Sentd)lad/ Kurt Polgieler/ .h'Jfrey Joseph/ Marlyn Powell/ David Lorenee/ Ylra-Maria Zimmerman/ and Victor A, Alheri^^

Abstract

Wilhin RauvolfioiilfUf' (Apocynacear), gciicra lia\e long berii assigned to ttibes based riiaiiily on only one or two sn|)eirieial liuit and .'^eerl rhaiaclcrs. Taxa witli (lni]>ac(n)us Imits ^vrrr inclndrd in \l)xieae. To elucidate relationships \silhin \l\\i{\ie, we analyzed j)h\"logenetieally a data set oi se(]i[(Mie(^s from fnnr [)lastid DNA regions {r})c\ .. ukuK. trril. inlron. and lni\.-F intergenic spacer) an<l a morphologieal da I a set for 33 genera ol \poe> naceae. ineludinji; rep resent at iaos of all genera pre\iously ineliKled in Alyxieae and luo nun-Apoeynaceae species. l\<-sidls of parsimony anaJ).sis indicate lliat Al).\ieae as pre\ iously deliniilcd are [)o]) phyletie, with most gcaiera falling into tuo main elades. Tlie/l/jA/a clade includes seven genera:

Alyxia Ranks ex K. Hr., Lepinia Decno., Irpi/fiopsis Valeton, Flenilyxia K. Schiim., and Condylocarpon Desf. togellier wilh

Pli'ct(unna Thouars. (earlier included in I'lunieiieae) and Chilocarpits Hlume (earlier inelud<"d In (^Inlocarjieae). The Vuica clade inelud(\s (Mglil geneia; Cabucala Pielio!), Pclchla Livera. iiaiiroljla L., Calliaranlhus C. Don, Vaiva L., Neisospcrmd Raf., Oclirosid Juss.. and kopsia Blume. l«//c,s;(/ lun/ ^K: Paw i\m\ A nccliifcs Criseb. are not relalctl to either clade and come oul as sister to Aspidospcrma Mart. & Zuce, (Aspidosjx'rmeae) and T}}<'ie!i<i L. (Plumerieae). respf^Tively. The fruit and seer]

'We wish to thank ihe followmg persons who provided ])lant mateiial or DNA sam|)les: A. Assi. Paul Berry, F. liilliet, Birgilta Bremer. \. F<"rreira. P. (iarnnck-Jones, P. Kessler. T. Koch. A. Tceuwenberg, S. Lieih-, JJ. Neill, R. Onilor. M. Pelignat, M. I'rcvost, C. R{Mnei"o, A. Specht, S. Thicker, and S. Zona. Vov technical assistance and [jholographic help wilh llie |)ollcn contribution sincen^ thanks are due to Elisabeth (irafstrfim and Magnus llellbom, Palynological Laboraloiy, Sweth'sh Museiun of Natural Hisloiw Stockholm, and Berlie Joan \an H(an'en and \\ ini Star, Nalionaal Ilcibarinm Nederland. Leiden. This stud\ was su[)j)orted b\ a grant from the llelge Ax:son Johnson Loundation and the Sw^niisli FouncUition for International

107. 8008 ZiJrich. Switzerland. mcnfh'ess@svstbot.

Cooperation Rest-ardi (STI\T) to B. Sennblad.

^Institute of S\stemalic Botanv. Uni\crsil\" of Ziirich. Zollikerslrassc uni/h.eh.

"Nationaal Herbarium \ederland. P.O. Box 9514. 2300 RA Leiden. The Netherlands.

' Laboratoire, D\iiami(|ue dv, la Biodiversilc, UMR 5172, Univcrsitc Paiil Sabatitn", 31062 Toulouse, Cedcx 9, France.

■'^Jodrell Lal)orator}', Ko}al Botanic Gardens, Kew, Richmond Surrey, TW9 3DS, United Kingdom.

''Stockholm Bioinformalics Center, Stockholm LIniversitv, All)aNo\'a l^esearch Center, SIv 10691 Stockholm, Sweden.

Colle^o of \etcrinary Medicine, LIniversitv of Illinois, Url>ana. fllinois 60801. U.S.A.

'^National Tropical Botanical Garden, 3530 Papalina Road, Kalahco Kauai, lla\saii 96741. U.S.A.

'MnstituI fiir F\()hilionsi)iologie imd Mor[)liologie, Uni\ei"silat \\ itlcti/l lerdecke, Sockurncr Str. 10 l)-58il8 \\ itt<'n, (/crmanv.

"Natural History Museum, University of Oslo. Box 1172 Blindern, NO-0318 Oslo. Norway.

t Siwert NilsscMi passed away unc\pectedl\ belore th(^ manuscript was completed. We losi both an excellent collalnnatoi' and a dear friend. We dedicate this |>apei- lo hinn

Ann. MissuLiii Bot. Gabd. 94: 1-35. Plblishld on 26 Aphil 200

i

2

Annals of the

Missouri Botanical Garden

characters j)rc\ i(His1y used to detiiarcalc^ Aly\i(\ie arc tiotiiefilasiniis. as arc cllicr niorpliolngieal cliaraelers such as sl\ h' head structure and s\in'arp\ versus apocarjn'. Corn (^rselv. pcHcn morpli(tl()<2:v. which lias not pr"c\iouslv pla\'cd inueh of ;i role in tribal <!r]irnitatioti. was shown to he the niosl uselid niorpliologieal cliaraclcr for dehiuitin^ MwJeae (roni other li'ihes of

RaiivoHioidcae.

Kty nords: \\)\\cae, Apocynaccac, classificatioji, nialK, niorphohjgy, ph\h)^cny, polU-ii, rhc\.^ sysletnatics, !rn\., //7/L-F,

Tribal del inn tal ion in l{au\()l(i()Hleae (usually

Hcuth. Notonerium has since been shown to Ixditng to

rcfcTrcd to as I'luiiicrioi^h^ac in tlio older lil(^rature) l?orai:iriaceae (Cris[), 1983).

has previously b(^en based on (ruit and seed

hi his classification from iy94a, Leeuvvenbertr

cliaraelers (Seluutiann, 1P)95; l^ielit)n, 1949a, I949e; nunnlainetl Pichon's (1949a) cireuniseiiplion of Kau- Ly. 19t>(): Leeu\venl)erg. 1994a). One reason ihat fruit \()irieae as well as the live subtribes intdudtM! th(M"e

but gave no insight into ihe deliniilalion of the sul)lril)es, slalini^ oidy that the relatively slight

and seed cliaraelers ha\e been so frc(|uenlly used in elassKiealious and keys is ihal tlu'y are readily observed, pennilting easy nn-ognilion of inanv i^euera. The

J JD

(lidcrcnees between ils subtrd)ej

s are nol easuy

asil

other main reason for the fruit-based tribal described in a concise wav. The only dillereiices

between beeuwenberg's (1994a) and Piehorrs (1949a)

and there are few ol)\ ious distinguishing floral characters uselul loi" dillerenliating tribes in dns

elassilicalions is that die (lowers of many Ranvol(i(»i- deae tend lo be superlieially similar. Many are tiibal cireuniscriptions art' that b<'euw^enl)erg changed relati\elv small with a whitish, salverhtrm corolla, ll]c name o( the trdic to \l\Aiea<\ put Podochrosia inlo

synonym) LUider Hdnroljui and Rlupidin in (^ondyln- Citrpon (lol lowing Tal leu, 19o3b), and intduded snl)fatnil} . Hie most detailed studies o( Ran\ol[ioi- Afwchitcs in CouiKlocarpinae, al though Falh^n deae were those of Piclion (194Ha, 1948b, 1919a, (19o.'5a) had suggested earlier that a [xisitiofi (doser 195()b), who {)ublished extensively on die family. His lii (axa previously included in Cerbereae {Cfimrrdrid elassilicalion was a grcal iin[)rov(ancnl over its I.., Cerbcni 1,., Cerhcriopsis Vicilb ex l*aiudier & predecessors. Its main weakness was thai tribal S<d)ert. Thefcda L., and SkyUinlluis Vleven) was more delimitation was based mainly on a single fruit appropriate. More ii-ccnlly, I.ei-nuenberg (1997) pul eliaracder. lie split the rauv(dfioid hibes int<t two (lal)U(:ala inlo synonymy under Pelchia, main groups, depending on whether llie deepei" la\'ers In ad(hlion to the taxa mentioned abo\e, there arc

of die mesoear[i were (lesliy or dry. The group with hiur other genera (diaracterized by (hu|)accous fruils: a fleshy inner tnesoearp was divich-d inlo five tribes: (^.crhcrcL Thcvclia, Cerhcriopsis, and (Uimerarid. These, Carisseae, Ambtdanieae, and Maeoid)eeae c(jnipiised together widi Skylanllnis, witli lollicular fruits, wei'e those taxa with indehiseent berries and Chiloeaipeae split oiil tif Hauv(tl(ioideae and treated by Pi(dion ancrrabernaemonlancae incduded those taxa with fruit (19dl5b) as a separate subfamily, Cerberoideae. The

( haracters he used h)i d(diniitatIon of this subfamily are ambiguous. beeuwenl)erg (199 1aj r(M'ogni/ed the

consisting of chdiiscent hdlicdes with arillate seeds. In

tl

le group wiin a my mesocarp, he recognized lliree

ith a d

trd)es. Two of them are idiaracteri/cd by d(diiscent grouj) as defined by Picdion, Lul al l!ie tribal level as

fruils: Alsloui<Nie (riumcrieae sensu Teeuweiiberg. Cerbereae. Mor|»hological studies by Kallen (198.^)

I99da), in whi(di the (ruit consists of a pair of (ollicles, suggested a close re]ationshi|) btdween Cerbereae and

and the monotypii: Allamaudeae. in ulii(di the fruit is AlUuuandu L.. the sole gc.'uus placed in Allarnandcac

a spiny nnilocidar capsule. 1 lie last tribe, Rauvolfieae by l^itdion (1949aj and Leeuwenberg (1994a), Studies

(Alyxieae sensu l.eeu\ve!d)erg, 199da), coulaiued all based on molecular (tr combined morphological and

taxa in whi(di the fruit is an indehiseent drii[)e with molecular dala (Endress et al.. 1996; Semiblad &

a stony endocarp. ll is the relalionshi[is williiri this Bremer, 1996, 2000, 2002) indicated that th<^ genera

grouf) thai are the focus o( this paper. pfcviously iiu Inded in Cerbenvu' do form a natural

Pi(dion (1949a} recogniz(Ml five sulitribes within his group (see Potgieter & Albeii, 2001, h)r a different

I{au\oinea<': Rauvolfinae (iiKduding Ciihiicida Pi- opinion) and that they an- onl) a part of a largci' grouj)

ehon, Prlchid Livera, Rnuvolfid b., and Podochrosid that irududes not only Alldtudiidd. but also Pluiumid

Baill.), Alyxiiiiae (iucduding Alyxia Panks ex [{. Pr., L. (usual!) uududed ui the Phimerieae: HainoHIol-

LepifUd Deem-., and Lepinlop.sis \'ahdoii), Otdirosiinae deae). An aual\sis of Cerbereae is not the aim of this

(iufdudirig only r^c/z/YM/V/ Juss.), Vallesiinae (including study, allhongh some re]»resenlat ives from that tribe

Vallcsia Ruiz & Pav. and Kop.sia Blunu^). and are included in our analyses.

Condyloearpinae (including Rhipidia Markgr. and Lsing fruit charatders to (hdimil tribes in Rauvol-

Condylncdrpon Desf.). Pi(dion in(du(hHl two genera fioideae is appealing because it allows taxa to be

as Incertae sedis: Afieclnlcs (^riscb. and Noloncrluni easily t-alegorized and keys to be constructed.

Volume 94, Number 1 2007

Endress et al.

Phylogenetic Analysis of Alyxieae

3

However, other charaelers of these taxa do not (following I.e(nnvenberg, 1997). Simoes et al. (2007)

indicate the same |)aUerns of relationships. Pliyloge- treated Amsonia as a genus incertae sedis and

nelic analyses of inaiidy moleeular data have shown transferred Laxoplnmeria Markgr., Touduzia Piltier,

that these fruit- and seed-based classifications are and Kdinctlia Kostel. to Vineeae, brins^ina; the total

considerably more arlifi<'ial than [)reviously sus- number of genera in the tribe up to 10. peeled. An rbcL analysis by Sennbhul and l>remer '\\\e classification of Sennblad and Bremer (2002)

(1996) indicated that Call tar anihus G. Don (with dry pro|)osed a new system that is compatible with

dehiscent folli(des and included in Plumerieae) was traditional Linnaean nomenclature but uses a variant

mole closely related to taxa previously placed in ol the delinitions used in plivlogenetic nomenclature

Alyxieae sensu Leeuvv(^nl)erg (1994a) than to other to improve the stability of classifications. Although

1997; they do not provide lists of included genera, their

Plumerieae. In

larger studies (Sennldad,

Sennblad & Bremer, 2000, 2002), CdtharaiJihus and definitiuns of the tribes containing traditional Alyx- \inca L. formed a well-suppoiled clade tng(Mh(M- with ieae genera are completely congruent with those of Rauvolfla, Oclirosia, and Kopsia, which have fleshy Endress and Biuyns (2000). The aim of this study is to

'\ ivxieae anr

I Vine

nceae as

drupes. In the same study, Chilocarpiis Blume, which cladistically evaluate Al

has always been thought to be most closely related to circumscribed by Endress and Bruyns (2000) in

Carisseae, formed a strongly supported elade with com|)arison with previous classifications, to re-

Alyxia and U'pinia (Alyxieae), confirming results of examine the usefulness of fruit and seed characters

a strongly su[){)orted C I lilocarp us- Aiyxiaie clade ft^i trilial delimitation within Rauvolfioideae, and to

reported previously by Civeyrel (1996) and van der discover new morphological characters that have

Ham et al. (2001). Pichon (1949a) already realized hitherto received little attention in classification of

that Geissospermum AUemao (with indehiscent fruits diis sublamUy but show phyli»geneti<^ potential.

and seeds embcdd(Hl in pulp) is proljably the nearest

relative of Aspulosperjiia Mart. & 7a\cc. (with dry Maikk! Al.s AND iMethods

dehiscent follicles and wind-dispersed seeds with

a diaphanous wing), a position supported by Polgieter

and Albert (2001) and Simoes et al. (2007). In

TAXON SAMIM.ING

additujn, Potgietcr and Albert (2001) found diat Vallesia (with drupaceous fruits and seeds embedded

in juicy pulp) is closely related to HaplopJiyioji A. DC.

(with diT dehiscent follicles and wind-dispersed r f

The ingroup taxa were chosen to include represen- tatives of all genera of Alyxieae (sensu Leeuwenberg, 1994a), as well as other putalivtdy related genera. The outgroup taxa are one genus each of Loganlaceae and semiaceae, which several studies (Bremer &

cnniose seeds). Such results md.cate thai fmit ^^^^^^^^ ^,y,y^. ^^^^ ^^ ^^^ ^,^^3. Savolainen et al., characters i.i Apocynaceae are evolutmnanly plastic 2OOO; Sultis el al.. 2000) have riemonstrated to be ,n response to selective pressures for adaptations ^j^^^j^ ,.^,^j^j ^^ Apocynaceae (Appendices 1, 2).

associated with wind or animal dispersal.

The most recent classifications of Apocynaceae s.l.

Other more naiTOwly focused studies on Gentianal<\s (Struwe et al., 1994; Endress el al., 1996; Backlund et

(Endress & Bruyns, 2000; Sennblad & Brimier, 2002) ^] ^ 2OOO) have also indicated that Loganlaceae and altempted to rectify some of these anomalies. In the Gelseniiaceae are the closest families to Apocyna-

classification ol Endress and Bruyns (;

, Anechites was moved to a newly defined Plumerieae (includin Cerbereae sensu Leeuwenberg (1994a) as well as

ceae. For the morphological analyses, we omitted the outgroups altogether because, in preliminaiy analyses, one or the other of these genera was embedded in

Allamanda), a position suggested by Fallen (1983a), different portions of the ingroup due to obvi

loirs

and Vallesia was Included with Aspidosperma^ Geis-

parallelisms of certain characters; we arranged the

rest as in ihe molecular results.

sosprrmuni, and Haplophyton in a newly defined morphological tree with the same group sister to the Alstonieae. The remainder of Alyxieae (sensu Leeu- wenberg, 1994a) was split into two tribes, vVlyxieae and Vineeae, based on molecular results as well as additional morphological characters. Alyxieae sensu Endress and Bruyns (2000) included seven genera:

FLOKALSTKLCTLKE

Fixed flowers at or near anthesis (only buds were

Alyxia, Pteralyxia K. Schum., Lepinia, L^piniopsis, available for hqnniopsis) were dehydrated in an

Plectaneia Thouars, Gofidylocarpon, and Chilocarpus. alcohol-xylene series, embedded in paraplast, cut

Vineeae included Amsonia Walter, Caiharanthus, with a rotary microtome at 10 )J.ni, and stained with

Vinca, Raavolfia, Pelchia, Kopsia, Neisospcrma Raf., safranin and astra blue. For SEM studies, material

and Ochrosia. Rhazyn Decne. was considenMl to be was critical-point dried and then sputter-coated with

synonymous with Anisonia, and Cabucala with Petchia gold.

4

Annals of the

Missouri Botanical Garden

POLLEN MOHI'llOLOGY

Dovle and Doyle (1987). For ihc latter, DNA sampler

s

Pollen material was sampled from the lollowin^

r

herbaria: BISH, RR, COL, G, U P, PTBC, QCA, S,

UB, WAG, and Z. Pollen stLuHes were earrietl (lut in Leiden and Stoekholm. For light mleroseopy (LM). p()ll(Mi material was aeetolyzed (exeept for Condylo- carpon and Vinca), mounterl in glyeerine jelly, and sealed with paraiTiiL Generally, 10 pollen gi'ains were

were puriiied by ultracentrifngation in CsCl— ethidium bromide gradients (1.55 g/ml). Additional pnrifieation using lh(^ QlAquifk PCK purification kil (Qiagen, Valeneia, California) was performer! in rases with problematic polymerase chain reaction (PCR) ampli- fiealion using the manufacturer's protocol. Double- stranded DlNyV was amplified w^ilh PC II priuKM's for

7>eL from Fay et al. (1998); the tni\. intron and inih-V

^ _ -/-I ci ^ f,f^^,^ horn ray et al. (^iVVoj; the lni\. nitron ana inii.-v

measured for polar axis (P) and equatorial diameter . . . ., r. . t i?\ it i

,^, ^ .-,.,,. 1. 1-1 11 1 mlergenie spacer (t:iereailer, r77/L-rj w^ere ampliliea

using ihe c and f primers of Ta])erlet et al. (1991); matK primers were those of Endress et al. (1996) aiid Johnson and Sollis (1994). Direct sequencing ol PCR products was performed using the PCR primers plus

(E). For SEM, pollen w^as sputter-coated wdth gold and examined with a JSM 5300 or JSM 6300 scannin electron microscope (JEOL, Tokyo). Frozen sections were made using an Ames Tissue-TEK CryoslaL For transmission electron microsco[)y (Tl^M), unaceto- lyzed material (w'hole anthers) was embedded in Spun- resin or 3/7 Epon, sectioned wilh a LKB LUtrotonie 111

or V, poststained wilh uranylacetate and lead citrate, and examined with a Z('iss 10, a JEOL 100-S, or a JEM 1010. Terminology is according to Punt et al. (1994).

internal sequencing primers. For rbc\., the internal primers w^ere those of Fay et al. (1998); for //7iL-E, w^e used the d and e primers of Taberlet et al. (1991); and for nialK, we designed Iwo new internal pritners:

5'-ATGTAT(;TGACTACGAATCA-3

ant

I

734F,

829R, 5'-ACTTTC rATTTTTCCATAGA-3'. In a num

her of eases, w^e also used the internal sequenein

primers as PCR primers to amplify shorter [>i'oduets.

T r .- r ■. 1 1 1 1 f For senueneiuii, we us(m1 either the Dye Deoxy

Iniormation on I mil and seeds were taken Irom i &- . .'

OniLR MORPHOLOGICAL AM) ClILMICAL CHARACTERS

Terminator Cy(de Sequencing or Big Dye kils o

f

observations of herbarium specimens as available.

c If'. 1 11 * ^ .. . .r ... fv- .... iL.v Ap!)lied Biosystems (ABI; Warrington, Cheshire,

Several IruH and sc^ed ciiaraeters were taken Irom ttie ' i -^ ^ ' & '

literature, as were data on the presence of latieifers United Kingdom). Sequencing reactions were carried

and intraxylarv phloem (Solereder, 1892; Schumann, ""^ 'l"«^tly on the elean^Ml PCR products and run on

1895; Valeton, 1895; Degener, 1946; Pichon, 1947a, '^^ ^BI 277a automated sequencer at Kew following

1947b, 1948a, 19481., 1948c, 1948(1, 1949a, 1949b, the manufacturer's protocols.

1949e, 1950a, 1950b, 195()c, 1952; Censel, 1969;

Markgraf, 1971, 1976, 1979; Markgraf c'i- Huber, CI.ADiSTIC ANAI.Y.St;s

1975; Corner, 1976; Conn, 1980; beeuwenl)erg &

The dala matrix comprised four submatrices: each

. Tl

le

Leenhouts, 1980; Rogers, 1986; Rudjiman, 1986;

o ino-7 iv/i , If e r\ n M^vn v> ir inod ef the three DNA re^^ions plus morpholoy;y

rageu, 198/; Meteaii & Chalky 1989; RosaHi, 1989; ^ ^ ^ '^;

morpliological subniatj"i\ (Appendix 3) comprises 54

wru- innr r\ ' i nn/^ i I ^ nn^ characters from floral, fruit, vegetative, and pollen

W 1 1 hams, 1 996; i )nuno, 1996; Leeuwenljerg, 1 99 ^ ; ' ^ b i

morpliology and phytochemistry (Appendix 4). All

Wagner et al., 1990; Sevenet et al., 1994; Forster &

Sidiyasa, 1998; Lin & Bernardello, 1999). Chemical

] , .1 r nil I n u f 1 lOAo analyses were performed using PAUP^' 4.01)10

dala were taJ^en irom ttie literature (Jorins et al., i9oo; -^ . .

II in^M i(W(\ r Sr r 11 ioqq (Swofford, 2002). Heurislie searches were peiTorrned

I iegnauer, i 9 i 0, 19o9; Loppen oi LoIjd, I9oo; ^ ' ^ ^

1^- i 1 1 1 looo TT .a ...^... x^ u^ ^ lOQi. w^ith all characters given unit weight (Fitch parsirnonv;

Kisakurek et al., 198.3; tiomberger oi Hesse, lyo4; ^ t? v r- ^ ■>

Bisset, 1987; Endress et al., 1990; Wagner et ah, 1990; Zhu et ah, 1990; Attaurrahman el ah, 1989, 1991; Arambewela & Ranatunge, 1991; Jensen, 1992;

Fitch, 1971), and each submalrix was analyzed

separately before their joint combined analysis, Fach of the searches used the lollovviiig sellings: 1000

MOLECULAR MLTHOOS

Sevenet et ah, 1994; Zeches el ah, 1995; Kam et ah, replicates of random taxon entry, the subtree pinning \oi)j\ re-grafting (SPR) branch swapping algorithm, and

MULTREFS on (saving multiple equally parsimonious

trees) but hohh'ng only 10 trees per replicate. All llie

shortest trees were tlien collected and used as starling

Nine new sequences of ricF, 16 of ma/K, and 11 of trees for a search with a 25,000-tree limit. If the tree

the irn\. intron and truL-Y intergeiiic spacer were limit was reached, then swapping was allowed to

produced for this study; the other sequences were continue until all 23,000 trees weie swapped to

published [)reviously (Appendix 2). Total DNA was completion.

extracted from fresh leaves, sihea gel-<lried material Bootstrap percenlages (BP) (Felsenstein, 1985)

(Chase & Hills, 1991), or herbarium material using w^ere ealculated w^ilh 500 replicates on each of the die methods of Saghai-Maroof et al. (1984) or modified submatrices as well as on the two combined matrices

Volume 94, Number 1 2007

Endress et al.

Phylogenetic Analysis of Alyxieae

5

(molecular conilMncd and molecular/morphological (Maddison & Maddisoii, 2000) to illuslialc cliaracler

evolution and compare llie usefulness of characters branch f^wapping and MULTREES on, holding only dial have previously been used in deliniilation of

combined). Vie used the following sellings: SPR

Vlyxieae.

KF,SrF.TS

1()KP[10L()(,\

10 trees per step. All other settings were those uf the standard defaults of PA UP* 4.0. This strategy produces results statistically indistinguishable from odier, more dnorough bootstrap protocols (Salamin et al., 2003). We apjily the following arbitrary sc:ale ui our discussion: 50%— 74%, weakly supportcnl; 7S%— 84%, moderately sup[iorted; 85%— 100%, strongly

supp(ji'led. Alignment for r/>cL and matK was a s!mj>le

,, ,1 r II 1 .1 .- ] ,1 parsimonious trees, each of 246 steps with a eonsis-

matter; tiie tormer had no iength variation, and Ihe ' _ ^ \

1 .. 1 1 1 r -1 1 . 1 .■ / tencv index (CI) of 0.37 and a retejdion index (Rl) of

latter iiad only a tew easny characterized mscrtions/ ' . .

vXnalysis of the morphological matrix produced 31 trees in three islands (18, 4, and 9 trees) of equally

deletions (indels). For //7zL-F, we started with the alignment of Polgieter and Albert (2001) and added

the additional taxa needed for this anal) sis, which required adding a few more insertions; we did adjust their aligniricnt in places following Kelchner (2()()()). We analyzed the Irnh intron and trnh-F intergenic

0.62. A slricl consensus tree of all three islands (Fig. 1; numbers below the branches are BPs) shows that llie position of several genera is unclear with these morphological data, and ihe three islands place them in different relative positions to the clades that are consistently resolved in all three islands. The

spacer in a single analysis (which can be considered P«si'»>"-^ "^ ^"^1'^^^^^ antUlana Woodson, V. glabra

"non-co<n.,f,^" because there is only alK,ut a ;50 bp (^av.) Link, and the Plumeria clade {Allamanda

region of ihe tniL exon included); this region, termed <^MhxirUvAi L., Anechites nerinm (Auhl.) Uib., Phnnena

lni\A', is ecjniposed of two unrelated parts, but the ''"^^^" L., and Theveiia peruviana (Pers.) K. Sclium.)

number of variable sites is the lowest and, even when ^'"^- f^<>r'S''stent in all three islands, as is a clade

combined, lh<'se pro<]ueed a higlily unresolved strict composed of" all remaining taxa. Within the last,

consensus Iree. We do report statistics for these two Acohmihera ohlongifolia (Hochsl.) Co.ld, A. oppusiti-

regions separately (Appendix 2) but consider results M" (Lam.) Codd, Molongum laxum (Renlh.) Pichon,

only for the two combined.

Picralima nitida (Stapf) T. Duiand & II. Durand,

Incongruence of different regions of plastid DNA P^^iocarpa mutica Benth., Tahernaemontana dirar-

would be unexpected because recombination is i^cata (P.) K. or. ex noem. & cjchull., and I.

unknown in the generally unipanmtally inherited pnndacaqui Lam. occupy different positions in each

plastid genome. Several tests for combinability have ^^ ^'^^ three islands with respect to the two

been developed, but we have not used any of them in consistently resolved clades: (1) Alsfonia scholaris

this paper. Such tests have proven to be utireliable

(L.) K. Br. to Rhazya stricla Decne. and (2) Alyxia

indicatorsofincongruence(Reeveseta].,2001), so we oblongata Domin and A. rnscifolia R. Br. to Nerinm attach no particular significance to results uf these oleander L, (Fig. 1), The other conspicuously differ- tesls but instead prefer to look for cases of strongly ently placed group is that composed of /U;?.soma cv/mia

su[)ported and incongruent patterns of relationshij)s.

Walter, .4. tabernaemxyni ana Waller, Rhazya stncta.,

Differences in relationships are to be expected wilh and Catharanlhus roseus (L.) G. Don -f Vinca major L.

different matrix components simply due to sampling and V, minor L., which in two of the islands (18 and 9

effects where there are too few variable characters to tree islands) are a clade but in the oUicr island form

ol)tain clear patterns. If, however, there are only a few a grade. Clades that receive moderate to strong BPs

characters in a matrix, such as is the case here wilh are the following: Neisosperma nakaiana (Koidz.)

the morphological characters, differentiating between Fosberg & Sachet + Ochrosia coccinea (Teijsm. &

sam[)ling (MTor and incongruent patterns is extremely Binn.) Miq. (BP 83), Alyxia ohlongata and A.

difficult. We note that the morphologically based rascifolia + Lepinia marquisensis Lorence & W. L.

estimates of relationships deviate from those based on Wagner, L. solomonensis Hemsl. and /.. taitensis

DNA data, particularly for the Vinceae, but the small Decne. + Lepiniopsis ternalensis Valeton and L

number of morphological characters (only 54) does not trilocalaris Markgr. + Pteralyxia kanaiensis Caum

permit us to say whether the differences between and /■*. rnacrocarpa (Hillebr.) K. Schum. (BP 97; the

molecular and morj>hoIogieal patterns are evidence of last two genera with BP 96; the Alyxia clade),

true incongruence. Kihatalia gitingensis (Elmer) Woodson + Mascaren-

Charaider slate distributions of selected morpho- hasia arhoresceiis A. DC. + Nerinm oleander (BP 98;

logical <'haracters were individually mapped onto the the first two genera w^ith BP 94), PirraJima, nitida. +

total evidence tree (Figs. 10, 11) using MacClade 4.0 Pleiocarpa mutica (BP 84), and Allamanda cathartica

6

Annals of the

Missouri Botanical Garden

53

72

98

84

Acokanthera ob long i folia

Acokanthera oppositifolia Alstonia scholaris

Amsonia oil lata

Amsonia tabernaemontana

Cabucala polysperma

Kopsia fruticosa Neisosperma nakaiana

Ochrosia coccinea

Petchia ceylanica

Rauvolfia vomitoria Catharanthus rose us

Vinca minor Vinca major

Craspidospermum verticillatum

Rhazya strict a

Alyxia oblongata

l_ Alyxia ruscifolia Lepinia marquisensis

Lepinia solomonensis Lepinia taitensis

Lepiniopsis ternatensis Lepiniopsis trilocularis

Pteralyxia kauaiensis Pteralyxia macrocarpa

J— Condylocarpon guyanense L Condylocarpon isthmicum

Ctiilocarpus denudatus Chilocarpus suaveolens

r- Plectaneia stenophylla

"— Plectaneia thouarsii

Kibatalia gitingensis Mascarenhasia arborescens

Nerium oleander

Molongum laxum

Picralima nitida

Pleiocarpa mutica

Tabernaemontana divaricata Tabernaemontana pandacaqui

All am and a cathartica Anechites nerium

Thevetia peruviana Plumeria rubra

Vallesia antillana Vallesia glabra

Aspidosperma parvifolium

Figure 1. Slritt toti:5Ciisus tree ol the three islands found m the morphological analysis. Numbers Ijeiow ihe iirauches

indicalc bootstrap piMcenta^es greater ihan 30%.

Volume 94, Number 1 2007

Endress et al.

Phylogenetic Analysis of Alyxieae

7

+ Anechites nerium + Thevelia peruviana (BP 84). nitida + Pleiocarpa miitica as sister lo the Alyxia Calliaranlhus roseiis + Vinca major and V. niiiior clade, plus Pleclanela slcnof)hylUi (BP 70); aiul the

nearly reached the moderate level (BP 72).

ANALYSIS OF RBCL

Of the 1398 included positions, 244 (18%) were variable and 146 (10%) were potentially parslnuuiy informative. Analysis produced two islands o( equally parsimonious trees, one of 39 trees and the other of 16 trees; they had 479 steps with CI (including un-

Mnca clade (BP 65; Cahucala polyspcnna, Peuhia

ceylanica, Ochrosia coccinea, Catharaulhus roscus^ Vinca minor, Nelsosperma opj)osi(iJolia^ and Rauroljla man nil).

J

\N A LYSIS or M\TK

The aligned wafK matrix contained 1647 ]>{), oi which 561 (34%) were variable and 250 (15%) were

informative positions) of 0.59 and Bl of 0.63. The two IH>lentially parsimony inh.rmative. We were unahic to

islancls differ in Ihe relative positions of AV/:.so,v/>cr/r;a amplify the following taxa for matK: Anechites^

opposilijolia (Lam.) Fosberg & Sachet and liamoljui W^^^'^^^^ L^piniopsis, Ochrosia, and Plcclaneia. Anal-

manmi Stapf and Clnlocarpus suaveolcns Blume + ysis produced a single, most parsimonious Iree of 970

Condylocarpon ^uyanense Desf. In island one, the

steps with a CI of 0.73 and an Bl of 0.59. Patterns of

latter form a clade with the KlbataUa gUingensis + relalicnships are nearly ideiilieal lo those found wilh Mascarenhasia arborescens + Nerium oleander clade, ''^>'-^- '^»' '" gf^n'^''>'. ^Ps an- high.T Oian with rhcL or and Seisosperma opposUiJoUa and Rauvol/la mannii ""^^^ (^ig- ^)- I'atleni. Lliat were nol observe.) with

are unresolved members of the clade inclnding Calnicala polyspernia (Scolt-Elliott) Piclion + Petchia ceylanica (Wight) Livera H- Ochrosia cocrinca. In island two, Ncisosperma opposilifolia + Rauvoljia

rhcL include: Chilocarpus suaveolcns + Condylocarpon iruvanense (BP 100) as sister ([JP 99) to (he Alyxia

^

(

■lade (BP 100); Alslonia scladarls strongly supported in an isolated position as sister (BP 93) to all but

, . /^ ,7 ,1 I !■■ Asf)idos()crnia IrllrrnaUun + Vallcsia anlillana: and

mannii are sister to Latnarantnus roseus + \ inca ' '

1 /v ■/ 1 s r 1 1 Aspidosucrma Iritcrnatum + Vallesia antillana (BP 82)

minor, and Lnilocarpus suaveolcns + Lonaylocarpon ^ ^ _ ^ -'

1 1 rp. . . , r m(Klerat(Iv supported as sister lo (he rest of the

i^uyanense are unresolvea. trie strict consensus ol . i i

both islands (Fig. 2) therefore shows these laxa to be unresolved. Supported clades that are also foniul in the morpholt)gical analysis include the following: Alliunanda calhartica + Pliuncria inodora Jac{|. (BP

ingroup (BP 97).

ANALYSIS OF 77nL-F

The aligned //v/I^-F

matrix consisted of 1206 1

M>

100) and Anechites nerium + Thevelia peruviana (BP ^-^^, ,^^^ ^^,^^^^^ „^^ ^^.^^,^ ■_^,^.,,,^ .^^^^1 ^^^^ ,^^^ j-^.^^^^^ ^|,,, ^^.^^p_

97; the w^holc Pliuncria clade, BP 62); Alyxia ruscifolia + Ij'pinia taitensis + Lepiniopsis Inlocularis + Pleralyxia kauaiensis (the Alyxia (4adc; BP 99j; Picralima nitida sister to Pleiocarpa mulica (BP 100); Calharanthus roseus -L Vinca minor (BP 68j; and

F s[)acer). We w^ere unable to amplify the lollouing taxa for lrn\-V: KihalaVia C Don and Ix^pinia. Anahsis produced over 25,000 tices of 160 steps

with a CI of 0.78 and an Pil of 0.70 {irnV intron: 292 steps with a CI of 0.76 aufl an Bl of 0.69; trn\~V

Kibatalia gilingensis ^ Mascaradwsia jirhorescens + iniergenie spacer: 168 steps with a CI of 0.82 and an

Bl (»f 0.72) (Fig. 4). Rclati(tnships from these two. largely non-coding regions are similar to those

Nerium oleander (the Aerium clade; BP 93). Well- supported clades that are not strongly in conflict with the morphological results inchuh^: Aspidosperma Iritcrnatum Rojas Acosta as sister to Vallesia

estimated from rhc]. and fiuilK (figs. 2, 3). The major noteworthy result (also observed in the mnlK results.

anlillana (BP 100); Plcclaneia sleuophylla Jum. as j^^^j ^^^^-^^^^ rcL-cI\c(l BP < 50) is a strongly supported

sister to the Alyxia clade (BP 98), within which ihe ^.].^^]^ ^j^p 97^ (.ompos(ul of Nerium oleander and

topology is the same as in the morphological results; Mascarenhasia arborescens of ihe Nerium clad(%

Amsimui labernaemonlana as sister to Rhazya slricta observed with morphology, /7>cL. and nmlK, with

(BP 100); Cabucala polyspcrma as sister to Petchia Acokanlhera opposilifolia weakly su[iported as sisttu"

ceylanica (BP 100); and Molongum laxuni as sister to ^gp 50) also observed but without support with matK,

Tabernaemonlana divaricata (BP 95). Chilocarpus and Allamanda indcL ~\- Planicria alba Kimth (BP 95),

suaveolcns as sister to Condylocarpon guyanense (BP Anechites nerium, and Thevelia uhoiiai {}..) A. DC. (the

85) is contradicted in the morphological results by j^st i^^o unresolved with resixH't to the Nerium clade). a weak BP 70 for the latter lo l)e sister to Alyxia claile. \\'eaklv su[)j)orled results not contradicting relation- ships produced by morphology are the Plumeria clade

(BP 62: Allamaada calhartica, Anechites nerium. The c:ombined data set [>roduced 46 equally

Plumeria inodora, 'dud Theveda peruviana); Picralima parsimonious trees of 1925 steps with a CI of 0.*0

COMBINLt) MOLECULAR \NALVSLS

8

Annals of the

Missouri Botanical Garden

100

62

100

97

100

70

98

99

79

93

100

100

65

53

100

68

85

93

98

95

Geniostoma

Gelsemium

Acokanthera

Allamanda

Plumeria

Anechites

Thevetia

Alstonia

Aspidosperma Vallesia A lyxia Lepin ia Lepiniopsis

Pteralyxia Plectaneia Pi era lim a Pleiocarpa Amsonia

Rhazya Cabucala Pe tch ia Ocfirosia

Catfiarantlius Vin ca

Neisosperma

Rauvolfia Ciiilocarpus

Condyiocarpon Cra spidosp erm urn Kibatalia

Mascareniiasia Ne hum

Kopsia

Moiongum

Tabernaemontana

Fi^iii'i^ 2.

Striil foiisiMisus iree of llic luo islands ftjuiid willi llu- rbcL data. Niiinhcrs It. 'low ilir hratuln's iiiili(*;ilt

]KK>|straj) pf'iX'ciita^cs greater lliaii 50%.

Volume 94, Number 1 2007

Endress at al.

Phylogenetic Analysis of Alyxieae

9

Z6_ Geniostoma

^ Gelsemium

22

3

1

1

25/90

7

93

12

97

16

25

100

100

21

14

8

58

100

40

26

8

98

2

11

13

4

100

2

35

11

13

99

100

15

17

14

100

35

5

2

9

5

14

6

2

96

73

52

97

4

28

in

52

28

51

23

9

100

21

29

82

Acokanthera

12_ Kibatalia

Mascarenhasia

Nerium

^ Allamanda

Plumeria

- Thevetia

- Am son i a

15_ Rhazya

Craspidospermum

Picralima

Pleiocarpa

A lyxia

Pteralyxia

Chilocarpus

Condylocarpon

Cabucala

Petchia

Catharanthus

Rauvolfia

Vine a

Neisosperma

Kopsia

Molongum

29_ Tabernaemontana

Alston id

1^ Aspidosperma

^ Vallesia

Fij^iirc 3. The single most |)arsiinoni(>iis Iree founrl willi lh<' malK data. Nunihers above llie hraiieites if](lieat(; esliniated siiltstiliitioiiH, ACCTRAN oplinii/alion. Numbers below the braiu-Iu^s iiidii-ate boulslrap percenlaj^^es j^i'ealer tlum 50%.

aiul an RI of 0.62. One of the shorlesL individual Irees IniL-F inlero;enic spacer, 170 steps (vs. 168 on tlie

tniL-F trees). Patterns of relationsbijjs are nujcli like

is shown in Figuie 5, with estimated substitutions

(ACCTRAN optimization) indicated above the those in the previous analyses, and BPs are generally branches and BPs below; groups not present in all higher llian in any of the inthvidual analyses.

shortest Ifees are marked wilh an arro\v]iead. We show

a single tree to ilhistrate relalive levels of genetic (hvcrg(Mice. The conlribution o( (*acl] region to this tree was: rbcL, 488 steps (vs. 479 for the rhcL trees);

ANAI,YSTS OF AtJ. DATA COMRINFD

The combined data produced a single, most malK, 974 steps (vs. 970 for the malK tree); tni\. parsimonious tree of 2226 sle[)s with a CI of 0.65 intron, 293 steps (vs. 292 on the IniL-F trees); and and an RI of 0.60 (Fig. 6). llie DNA o])limized onto

10

Annals of the

Missouri Botanical Garden

94

64

72

97

60

100

95

78

89

52

98

100

56

68

99

50

99

91

99

Genlostoma Gelsemium Acokanthera Mascarenhasia

Ne rium Allamanda Plume ha Anechites The vet la A lyxia

Lepiniopsis

Pteralyxia

Plectaneia

Chilocarpus

Condylocarpon

Amsonia

Rhazya Cabucala Petchia Catharanthus

Neisosperma Ochrosia

Rauvolfia

Vinca

Craspldosperm urn Kop si a Molongum Tabernaemontana Pic rail ma

Pleiocarpa Alstonia Aspldosperma V all e si a

l''igiiiT 1. Strict consciisiis tree ofllu- 25, ()()() e(|iiall)' most paisiiiioiiioii.s tiers foiiii(! vulli the ////T.-F data. Nuiiihers lielovv

ihe liiaiiclies itulieale hdolstrap jieii.Ttltaj5<'s j^ri-aler than 50%.

^^ISCl^^ll)^

K\ ALLATION Ol TU ADmONAL t.lHCl MSCIill'TIUiNs OK AIA WVAV.

ImiIIi llic ni()j'plinlii^i<;il ;m(l llic niiileculai' analysis

ihis Iree is 1926 s|<'])s, cme step 1oii^<t than the

(:oiiiL>iiied DNA tree; this step is eausetl !)) shifting

Kopsid fnilicosd (Ker. Cavvl.) A. DC. ant] Mahnigum

](i\]im + TabcriKu'iuoniiinti (two sji])., Afipendix 2)

fiiitii an uiuesolved piisilion with res|)e(l In llu; niaior

clades into positions as a grade with iesj)ccl to the indicatr tlial Alyxieae as i>n'viously circiiriiscrihed are

Vima (lade. OlheiAvise, reIationshij)s arc exactly as polyiihyletic. Of the indi\ idiial data sets analyzed, lh(>

with llie eoiiibiiied niolccidar data. The nu>r])lioh)gieal l'''e' hased on nialK prov ided the best sup|iort.

data optimized onto the combined trees (ACCTIIAN followed hy that of lrn\A\ Bl^s in the tree based on

o})tiaiizalion) was 276 steps witli a CI of ().,'53 and an rhcl, an<l ihe morphologital data set were low, with

Rt of O.SS, versus 2(')6 steps with a CI of O.-'^J') and an riuieh of the Iree a polylomy, and the positions of

lil ol 0.61 in the morphological analysis.

several genera were' e(jnivocal. Kven llien, however.

Volume 94, Number 1 2007

Endress et al.

Phylogenetic Analysis of Alyxieae

11

84_ Geniostoma

41

13

2

9

61/100

10

5

70

15

21

97

98

W

100

95

83_ Gelsemium Acokanthera

28

49

100

100

28

32

14

98

100 48

98

54

100

22

24

100

100

lA

100

12

5

21

92

94

25

30

39

44

100

65

9

57

28

8

100

8

86

1

<

39

24

100 14

17_100

27

99

37

49

100

outgroups Carisseae

19_ Kibatalia

28_ Mascarenhasia

Nerium

45_ Alia ma nd a