Syst. Biol. 51(3):389–409, 2002 DOI: 10.1080/10635150290069869 ClassiŽcation of Apocynaceaes.l. According to a NewApproach CombiningLinnaean and Phylogenetic T axonomy

BENGT SENNBLAD1 AND BIRGITTA BREMER Department of SystematicBotany, Evolutionary Biology Center (EBC),Uppsala University, Norbyv.18D, SE-752 36 Uppsala,Sweden

Abstract.— Anew approachto anomenclatural system,including elements fromboth Linnaean and phylogenetic nomenclature, is proposed. It is compatible with the existing Linnaeansystem, includ- ing“ standardnames” corresponding to principal andsecondary ranks, and uses avariantof the deŽnitions fromthe Phylocode system.A new infrafamilial classiŽcation, usingthis nomenclatural approach,of the Apocynaceae s.l.(i.e., including the Asclepiadaceae) basedmainly on analysesof rbcL and ndhFdatais discussed. Twenty-one tribes andfour ranklesstaxa are deŽ ned. [Apocynaceae s.l.; classiŽcation; Linnaeannomenclature; ndhF;nomenclature; phylogenetic nomenclature; phylogeny; rbcL.]

Discussionabout different nomenclature The Phylocode systemaims at explicitly systemsin recent literaturehas been in- binding namesto speciŽ c clades;the name tense (e.g., de Queirozand Gauthier ,1992, ofa taxonis tightly coupled toa speciŽc 1994;Schander andThollesson, 1995; Bryant, clade(or a speciŽc ancestor)through a def- 1996,1997; de Queiroz,1996, 1997a; Liden inition.Three different kinds ofdeŽ nitions andOxelman, 1996; Reveal, 1996; Cantino (node-based, stem-based,and apomorphy- et al.,1997; Liden et al.,1997; Moore, 1998; based;e.g., de Queirozand Gauthier, 1992; Schander,1998; Sereno, 1999).A new nomen- Cantinoand de Queiroz,2000) have been claturesystem, termed the Phylocode or proposed.Because the Phylocode system phylogenetic taxonomy(de Queirozand abandonsranks, it is (with itspresent de- Gauthier,1992;Cantino and de Queiroz, scription)incompatible with the Linnaean 2000)has been proposed,challenging the system. traditionallyused “Linnaeansystem” (e.g., Bothsystems have certain advantages Greuteret al.,1994). Applications of phyloge- asa communicationtool in biology:The netic taxonomyto plantsystematics include, Linnaeansystem has a morestable set of forexample, classiŽcations of the Lamiaceae namesin use (but the exactcircumscrip- (Cantinoet al.,1997), Ericaceae (Kron, 1997), tionsof their correspondingtaxa may vary), Malvaceae(Baum et al.,1998), and Scrophu- whereasthe phylogenetic systemprovides lariaceae(Olmstead et al.,2001). moreexact deŽ nitions, thus reducing in- In the Linnaeansystem, nomenclature stabilitycaused by subjective changesin andclassiŽ cation (circumscription) are not taxoncircumscriptions (e.g., splittersand stronglycoupled. Namedtaxa are Ž xed lumpers). In addition,the principal ranksof totwo reference points:the type (type the Linnaeansystem provide universalstan- genus/species or,in the caseof species,type dardnames that are important, for example, specimen) andthe rank;the only mandatory fortextbooks, databases, and  oras. rankis genus (Greuter et al.,1994; in prac- Weproposea compromiseapproach that tice,further ranksare often treatedas manda- combinesthe advantagesof the twosystems. tory).An exactcircumscription of the taxon Itis compatible with the systempresently in isnot required; the only restrictionis that a use—the Linnaeansystem— in using asys- taxoncannot include anothertaxon with the temof hierarchicstandard names (compare sameor higher rank.Nevertheless, a listof in- with“ primaryranks” in the Linnaeansys- cluded taxais often addedin classiŽcations tem) andtypes, but italso adopts a variantof oftaxa above species level. the deŽnitions from the phylogenetic system toreduce the impactof subjective changes 1 CurrentafŽ liation andaddress for correspondence: in circumscriptions.Because the mainaim StockholmBioinformatics Center/Center forGenomics withthe present studyis to present anew andEvolution, KarolinskaInstitute, SCFAB,SE-106 91 tribalclassiŽ cation of the angiospermfam- Stockholm,Sweden; E-mail: [email protected] ily Apocynaceaes.l., we will mainlyconcern

389 390 SYSTEMATICBIOLOGY VOL. 51 ourselveswith family/ tribal-level plantclas- Sennblad andBremer, 1996; Civeyrel etal., siŽcation. Likewise, ouruse ofthe termthe 1998;Potgieter and Albert, 2001),three Linnaeansystem mainly refers tothe In- tribeswere monogeneric,and the mono- ternationalCode of Botanical Nomenclature phyly ofone tribe hasnot been evalu- (ICBN; Greuter et al.,1994). W ewill not ated.The studyof Sennblad et al.(1998) discussspecies-level nomenclature,nor that wasaimed mainly at the tribe Wrightieae ofextinct taxa (see, e.g.,de Queirozand (Apocynoideae) sensu Leeuwenberg (1994a) Gauthier,1992;Eriksson et al.,1998; Mishler , but alsoincluded representativesof the other 1999). tribesof the Apocynoideae,as well asof the The Apocynaceaes.l. belong tothe or- Asclepiadaceae.Sennblad et al.showed that der Gentianalesand have a mainlypantrop- none ofthe described tribeswithin the Apoc- ical/subtropicaldistribution, with a few ynoideae ismonophyletic and suggested temperaterepresentatives. The family com- areclassiŽcation of the Wrightieae sensu prisesmany well-known ornamentalssuch Leeuwenberg (1994a)into the Wrightieae,the as Nerium (oleander) and Hoya (wax ower). Nerieae, andthe Malouetieae. The plantsare typically laticiferous and pro- The infrafamilialclassiŽ cation of Ascle- duce variousalkaloids and cardenolides, piadaceaeproposed by Brown(1810) when someof whichhave medical properties. The describing the family hasbeen followed mostwell-known example is Catharanthus by mostsubsequent authors(Liede and (roseperiwinkle), which containsvinblas- Albers, 1994)and comprises three subfam- tine andvincristine, compounds now used ilies,of whichthe Periplocoideae andSe- worldwideto treatchildhood leukemia. camonoideaeare monotribal and the As- The family Asclepiadaceaewas segregated clepiadoideae usually compriseŽ ve tribes. fromthe Apocynaceaes.l. by Brown(1810). An exception fromthis traditional view is Although someauthors have questioned this the classiŽcation of Swarupanandanet al. separation(e.g., Demeter,1922;Safwat, 1962; (1996),which reduced the number oftribes Judd et al.,1994; Struwe et al.,1994), most ofthe Asclepiadoideaeto twoand included classiŽcations subsequent tothat of Brown the Secamonoideaeas a thirdtribe in the havefollowed his “ two-family”treatment. Asclepiadoideae.Furthermore, a revisionof Molecularstudies of the Apocynaceaes.l. the Periplocoideae by Venter andV erhoeven (Sennblad andBremer, 1996; Civeyrel etal., (1997)recognizes three tribesin thissubfam- 1998;Potgieter and Albert, 2001)have indi- ily.The monophyly ofthe Asclepiadaceae catedthat the Asclepiadaceaeform a sub- isuncertain; in the studyof Sennblad and cladeof the Apocynaceaes.str .,thusrender- Bremer (1996),however ,the subfamily ing the Apocynaceaes.str .nonmonophyletic. Periplocoideae,itself monophyletic,did not These studiesalso indicated major problems forma monophyleticgroup withthe other withthe infrafamilialclassiŽ cation in the two twosubfamilies (sensu Liede andAlbers, families.A detaileddiscussion of different 1994),whereas in the studyof Civeyrel classiŽcations of the Apocynaceaes.str .and et al.(1998) the three subfamilies didform a the Asclepiadaceaeis given in Sennblad and monophyleticgroup. The studyof Civeyrel Bremer (1996),and only ashortreview will et al.(1998) also indicated that the tradi- be given here, concentratingon the mostre- tionalSecamonoideae is monophyletic with centclassiŽ cations and results from molecu- apositionas sister to Asclepiadoideae (see larstudies. alsoSennblad andBremer ,1996;Potgieter Since the classiŽcation of Schumann andAlbert, 2001).The traditionalAsclepi- (1895),the Apocynaceaehave tradition- adoideaereceive goodsupport (Sennblad allybeen divided intotwo subfamilies, the andBremer ,1996;Civeyrel et al.,1998; Plumerioideae andthe Apocynoideae,and Potgieterand Albert, 2001).Of the tribes approximatelyinto 12 tribes and 27 sub- ofthe Asclepiadoideaein the classiŽcation tribes(Pichon, 1948a,b,c,1950; Leeuwenberg, of Liede andAlbers (1994),Asclepiadeae 1994a).The studiesby Sennblad andBremer andMarsdenieae have been shownto be (1996)and Endress et al.(1996) showed that nonmonophyletic. boththe Plumerioideae andApocynoideae The Žndings ofthe molecularstudies indi- arenonmonophyletic. Four of the tribes catedabove and also from a previousversion ofthe Plumerioideae havebeen shownto of thispaper (included in the Ph.D. thesis be nonmonophyletic(Endress et al.,1996; ofB.S.,1997; see alsoSennblad andBremer , 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 391

2000)were summarizedand developed fur- TABLE 1.ClassiŽ cation ofthe Apocynaceae s.l.and therby use ofmorphology in the mostre- samplingof taxa.ClassiŽ cation areaccording to Endress centclassiŽ cation of the Apocynaceaes.l. by andBruyns (2000). Endressand Bruyns (2000).The Žve tradi- RAUVOLFIOIDEAE tionalsubfamilies were kept, but 19newly Alstonieae Melodineae recircumscribedtribes were alsorecognized Alstonia R.Br. Craspidospermum A.DC. (see Table 1). Aspidosperma Mart. Diplorhynchus Ficalho et Zucc. et Hiern However,the resultsof the present study Vallesia Ruiz Melodinus J.R. Forst. indicatethat although the classiŽcation of et Pav. et G. Forst. Endressand Bruyns (2000) are a majorad- Vinceae Hunterieae vancein makingthe classiŽcation congru- Catharanthus G. Don Picralima Pierre ent withcurrent best estimates of phylogeny , Kopsia Blume Pleiocarpa Benth. Ochrosia Juss. Plumerieae problematiccases still need revision.Fur- RauvolŽa L. Allamanda L. thermore,for many of the recognizedtribes, Vinca L. Anechites Griseb. taxonsampling for molecular data is weak Willughbeieae Cameraria L. andthe morphologicalhomologies in these Ancylobotrys Pierre Cerbera L. Dictyophleba Pierre Plumeria L. groupsare difŽ cult to interpret (Endressand Vahadenia Stapf Thevetia Adans. Bruyns,2000). Further rearrangementsof Tabernaemontaneae Carisseae tribalcircumscription will almostcertainly Carvalhoa K. Schum. Acokanthera G. Don be needed in the future. Molongum Pichon Carissa L. The Apocynaceaes.l. may thus provide a Schizozygia Baill. Alyxieae Tabernaemontana L. Alyxia R.Br. suitabletest for the classiŽcation system pro- Tabernanthe Baill. Chilocarpus Blume posedin thisstudy .Wetherefore proposea Lepinia Decne. new classiŽcation of the Apocynaceaebased APOCYNOIDEAE mainly on an rbcLanalysisthat includes Wrightieae Apocyneae extended taxonsampling as well asaddi- Adenium Roem. Aganosma (Blume) tionaldata from length variationof rbcL in its et Schult. G. Don Nerium L. Apocynum L. 30-end anddownstream. Certain relation- Stephanostema K. Schum. Baissea A.DC. shipsof special interest have been testedwith Strophanthus DC. Beaumontia Wall. asmallerdata set for the subfamily Apoc- Wrightia R.Br. Trachelospermum Lem. ynoideae andthe traditionalAsclepiadaceae, Malouetieae Mesechiteae Funtumia Stapf Mandevilla Lindl. including additionalnucleotide sequences of Holarrhena R.Br. Mesechites Mull.-Arg.¨ the chloroplastgene ndhF.Thisgene often Kibatalia G.Don Echiteae hasa greatersubstitution rate than rbcL (see Mascarenhasia A.DC. Parsonsia R.Br. Olmsteadand Sweere, 1994;Kim andJansen, Pachypodium Lindl. Peltastes Woodson 1995)and may therefore forma complement Prestonia R.Br. Rhabdadenia Mull.-Arg.¨ to rbcLwhen the latterproves too conserved. PERIPLOCOIDEAE Wehaveused only the 3’region ofthe gene Mondia Skeels Periploca L. because thatregion appearsto show the most Parquetina Baill. Petopentia Bullock variability(see Kim andJansen, 1995). Fur- Pentopetia Decne. Tacazzea Decne. thermore,results from other recent molecu- SECAMONOIDEAE larstudies have also been takeninto account Secamone R.Br in the classiŽcation. ASCLEPIADOIDEAE Marsdenieae Asclepiadeae Fockea Endl. Araujia Brot. Hoya R.Br. Asclepias L. MATERIALSAND METHODS Micholitzia N.E.Br. Calotropis R.Br. Wesampled77 representatives of the Stephanotis Thouars Orthosia Decne. Apocynaceaes.l. (T able 1).All subfamilies Ceropegieae Fischeria DC. Ceropegia L. Matelea Aubl. andtribes of the classiŽcation of Endressand Stapelia L. Schizostephanus Benth. Bruyns (2000)are represented in the analy- et Hook. f. sis. Gelsemium Juss. and Mostuea Didr. of the Tweedia Hook.et Arn. Gelsemiaceaewere chosenas outgrouptaxa, Tylophora R.Br. because they were indicatedto be the closest Vincetoxicum Wolf 392 SYSTEMATICBIOLOGY VOL. 51 sistergroup tothe Apocynaceaes.l. in the inferred throughcomparisons with complete recent analysisof Gentianalesby Backlund Apocynaceae ndhF-sequences fromthe study et al.(2000). ofBacklundet al.(2000). Forty-onenew sequences of rbcL are pub- Multiple sequence alignmentwas made by lished in thisstudy; rbcLsequences for handto reduce the number ofgapswhile in- the outgrouptaxa and for 35 ingroup taxa creasingthe percentage similarity.Weused were published previously (Table 2).For the criteriaof Golenberg et al.(1993) with mi- ndhF, the 30 region of the gene wasse- normodiŽ cations. quenced for18 taxa, representing the tra- Twocladistic analyses were performed. ditionalApocynoideae andAsclepiadaceae. The datamatrix for the Žrstanalysis covered TotalDNA wasextracted from fresh leaves allincluded taxaand comprised characters orherbarium materialby using the meth- correspondingto the nucleotide positions odsof Saghai-Maroofet al.(1984) and Doyle 27–1,425 of the rbcLgene andcharacters andDoyle (1987).The extractionswere pu- correspondingto gaps and informative nu- riŽed by ultracentrifugationin CsClgra- cleotidepositions downstream of position dientsor ethanol precipitation. Additional 1,425.EMBL databaseaccession numbers puriŽcation with the QiaquickPCR puriŽ - for rbcLsequences aregiven in Table 2. cationkit (Qiagen Inc.) wasperformed in Only parsimonyinformative characters were caseswith problematic polymerase chain re- analyzed. action(PCR) ampliŽcation. Double-stranded The secondanalysis was of a subsetof DNAwasampliŽ ed by PCRby using Taq- the taxaincluded in the Žrstanalysis and polymerasekit (Promega Corp.). Synthetic includes 18genera representing the Apoc- PCRprimer sequences for rbcL were taken ynoideae sensu Leeuwenberg (1994a)and fromOlmstead et al.(1992); ndhF primers the traditionalAsclepiadaceae. The datama- were takenfrom Olmstead and Sweere (1994) trixconsisted of two submatrices. One sub- andOxelman et al.(1999). In caseswhere matrixcomprised positions 27– 1,425 of the PCRampliŽ cation proved difŽ cult, a PCR rbcLgene. The secondsubmatrix comprised reactionusing Taqextender PCRadditive 733aligned nucleotide positionsfrom the (Stratagene Inc.) wasperformed, following 3’ region of ndhF.EMBL databaseaccession the protocolprovided by the supplier. Asec- numbers forthe ndhFsequences aregiven in ondround of PCR, with only one ofthe Table 2.Only parsimonyinformative charac- primers,was performed toobtain single- terswere analyzed. strandedDNA (Kaltenboeck et al.,1992). The cladisticanalyses were performed us- Single-stranded DNA wassequenced man- ing PAUP4.0b2(Swofford, 1998).An initial ually (Sanger et al.,1977) by using internal heuristicsearch (P AUP settings:HSEARCH primersdesigned for rbcLby G.Zurawski [ADDSEQ RANDOMNREPS 200 (DNAX ResearchInstitute) and for ndhF by SWAP TBR];D other options with default D set- Olmsteadand Sweere (1994)and Oxelman tings) withD all characters given aunit weight et al.(1999). wasfollowedby asuccessiveweighting anal- The length ofthe sequences of rbcL, ex- ysis(Farris, 1969, 1989) also using heuris- cluding the 26Ž rstnucleotides, but includ- ticsearches (P AUPsettings:HSEARCH ing positionsjust downstream from the gene, [ADDSEQ RANDOMNREPS 10 obtainedin thisstudy varied between 1,411 SWAP TBR];D other options with default D set- and1,509 nt. For all taxa except Parsonsia, the tings) andD characters reweighted according sequence includes the stopcodon. In posi- totheir rescaledconsistency index. tions27– 1,425, no structural mutations occur Bootstrapfrequencies (Felsenstein, 1985) in rbcL.Alignment ofthis part of the gene were calculatedwith 10,000 replicates on the isthus very simple. However,after position unit-weighted ( uwboot)andsuccessive 1,425,that is, just before the typicalposition weighted (swboot)datamatrices (P AUP ofthe stopcodon, gaps of different sizesoc- settings:BOOTSTRAP [NREPS 10,000 cur,complicating alignment. Similarly ,gaps METHOD HEURISTIC CONLEVELD occurin the codingregion ofthe ndhF se- 50][/ ADDSEQD RANDOMNREPS D1 quences. Becauseof these alignmentprob- SWAP SPR NOMULPD ARS]; otheroptions D lems,the sequences were truncated23 bases withdefault D settings). before the typicalposition of the stopcodon Tosimplify the discussionof the results, in Apocynaceaes.l. The reading framewas weused arough scaleof the relativesupport 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 393

TABLE 2.V ouchers andEMBL accession numbersfor sequences published in this study.Forpreviously published sequences andfor sequences frompreviously published voucher specimens, the reference is given.

Taxon Voucher/source b EMBL number rbcL/ndhF Acokanthera oblongifolia (Hochst.)Codd Sennbladand Bremer, 1996 X91758/- Adeniumobesum (Forssk.)Roem. et Schult. Sennbladet al.,1998 AJ002880/AJ420130 Aganosma marginata (Roxb.)G. Don Ryding490, UPS AJ419730/- Allamanda cathartica L. Sennbladand Bremer, 1996 X91759/- Alstonia scholaris R.Br. Sennbladand Bremer, 1996 X91760/- Alyxiaruscifolia R.Br. Sennblad236, UPS AJ419731/- Ancylobotryspetersiana Pierre Sennblad201, UPS AJ419732/- Anechitesnerium Urb. Bremer et al.3,386, UPS AJ419733/- Apocynumcannabinum L. Sennbladand Bremer, 1996 X91761/AJ420113 Araujia hortorum E. Fourn. Bremer 3,006,UPS AJ419734/- Asclepiascurassavica L. Sennbladand Bremer, 1996 X91774/- Aspidosperma triternatum RojasAcosta Bremer 3,029,UPS AJ419735/- Baissea leonensis Benth. Sennbladet al.,1998, Y allah112, UPS a AJ002881/AJ420114 Beaumontia grandiora Wall. Sennbladet al.,1998 AJ002882/AJ420115 Calotropis procera (Aiton) W.T.Aiton Sennblad222, UPS AJ419736/- Cameraria latifolia L. Houghton1,085, FTG AJ419737/- Carissa bispinosa (L.) Merxm. Sennblad235, UPS AJ419738/- Carvalhoa campanulata K. Schum. Sennblad217, UPS AJ419739/- Catharanthusroseus G. Don Sennbladand Bremer, 1996 X91757/- Cerbera venenifera Steud. No voucher AJ419740/- Ceropegia woodii Schltr. Sennbladand Bremer, 1996 X91775/- Chilocarpussuaveolens Blume Endresset al.,1996 X92445/- Fockea multiora K. Schum. Specks 248,cult, MSTR c AJ419741/- Craspidospermum verticillatum Bojer Pettersson andNilsson 742 AJ419742/- Dictyophlebalucida Pierre Sennbladand Bremer, 1996 X91762/- Diplorhynchuscondylocarpon (Mull.-Arg.)¨ Pichon Sennblad203, UPS AJ419743/- Fischeriastellata (Vell.) E.Fourn. Clarkand W att793, UPS AJ419744/- Funtumiaelastica Stapf Sennbladet al.,1998 AJ002884/- Gelsemium sempervirens Aiton Olmstead et al.,1993 L14397/- Holarrhena pubescens G. Don Sennbladet al.,1998 AJ002884/- Hoya bella Hook. Sennbladand Bremer, 1996 X91776/- Kibatalia gitingense (Elmer) Woodson Liede 3,268,Z AJ419745/- Kopsia fruticosa A.DC. Sennbladand Bremer, 1996 X91763/- Lepiniataitensis Decne. Perlman15071, NTBG AJ419746/- Mandevillasanderi (Hemsl.)W oodson Sennbladand Bremer, 1996 X91764/AJ420116 Mascarenhasia arborescens A.DC. Sennbladet al.,1998 AJ002885/AJ420117 Matelea hirsuta (Vahl)W oodson Sennblad263, UPS AJ419747/- Melodinusmonogynus Roxb. Sennblad261, UPS AJ419748/- MesechitestriŽ da Mull.-Arg.¨ Bremer et al.3,351, UPS AJ419749/- Micholitziaobcordata N.E.Br. Bremer 3,010,UPS AJ419750/- Molongumlaxum (Benth.)Pichon Sennbladand Bremer, 1996 X91765/- Mondiaecornuta (N.E.Br.)Bullock Sennblad215, UPS AJ419751/- Mostuea brunonis Didr. Olmstead et al.,1993 L14404/- Nerium oleander L. Sennblad265, UPS AJ002886/AJ420118 Ochrosia coccinea Miq. v.d.Laan 389, W AG AJ419752/- Orthosia serpyllifolia Decne. Bremer et al.3,372, UPS AJ419753/- Pachypodiumlamerei Drake Sennbladet al.,1998 AJ002887/AJ420119 Parquetinanigrescens (Afzel.) Bullock Sennbladand Bremer, 1996 X91777/- Parsonsia heterophylla A. Cunn. Sennbladet al.,1998 AJ002888/- Peltastes peltatum (Vell.) Woodson Sennblad262, UPS AJ419754/- Pentopetia sp. No voucher AJ419755/- Periplocagraeca L. Sennbladet al.,1998 AJ002889/AJ420120 Petopentianatalensis (Schltr) Bullock Sennblad237 a ands.n. cult. MSTRAJ419756/ AJ420121 Picralima nitida T. et H. Dur. Sennbladand Bremer, 1996 X91766/- Pleiocarpamutica Benth. Bremer 3,017,UPS AJ419757/- Plumeria inodora Jacq. Sennbladand Bremer, 1996 X91767/- Prestonia quinquangularis Spreng. Sennbladand Bremer, 1996 X91768/AJ420122 RauvolŽa mannii Stapf Sennbladand Bremer, 1996 X91769/- Rhabdadeniabi ora Mull.-Arg.¨ Zona616, FTG AJ419759/AJ420123 Schizostephanusalatus K. Schum. Cult. MSTR AJ419758/- Schizozygiacoffaeoides Baill. Sennblad207, UPS AJ419760/- Secamone afzelii (Schult.) K.Schum. Sennbladand Bremer, 1996 X91779/AJ420124 394 SYSTEMATICBIOLOGY VOL. 51

TABLE 2.Continued

Taxon Voucher/source EMBL number rbcL/ndhF Secamone geayi Constatinet Gallaud Civeyrel 1,200,LC AJ419761/- Stapelialeendertziae N.E.Br. Sennbladand Bremer ,1996 X91778/- Stephanotis oribunda Brongn. Sennblad256, UPS AJ419762/AJ420125 Stephanostema stenocarpum K. Schum. Sennbladand Bremer ,1996 X91770/AJ420126 Strophanthuseminii Pax Sennbladand Bremer ,1996 X91771/AJ420127 Tabernaemontana divaricata Roem.et Schult. Sennbladand Bremer ,1996 X91772/- Tabernantheiboga Baill. Leeuwenberg 12544,W AG AJ419763/- Tacazzea apiculata Oliver Venter 9,188,cult. MSTR AJ419764/- Thevetiaperuviana (Pers.) K.Schum. Sennbladand Bremer ,1996 X91773/- Trachelospermum jasminoides (Lindl.) Lem. Sennbladet al.,1998 AJ002890/AJ420128 Tweedia coerulea Sweet Sennblad254, UPS AJ419765/- Tylophorasylvatica Decne. Sennbladand Bremer ,1996 X91789/- Vahadeniacaillei (A.Chev.)Hutch.et Dalziel Leeuwenberg 12,275,WAG AJ419766/- Vallesia antillana Woodson Meagher966, FTG AJ419767/- Vinca minor L. Sennblad230, UPS AJ419768/- Vincetoxicumhirundinaria Medik. Sennblad257, UPS AJ419769/- Wrightiaarborea (Dennst.) Mabb. Sennbladet al.,1998 AJ002891/AJ420129

a Voucher for rbcLsequenceonly . b Herbariumabbreviations are according to Index Herbariorum,except LC Privateherbarium of LaureCiveyrel. D c Thisspecimen was originally determined as Cibirhizaalbersiana inKunzeet al. (1994), but has recently been redeterminedas Fockea multiora K.Schum.(V erhoevenet al., 2002). ofthe cladesin the successiveweighting (Fig. 1).The unit weight analysisof the com- analysis:Clades with a successiveweighted bined setof the rbcL and ndhFdata(153 char- bootstrap >63%were consideredwell sup- acterstotal; 0.43% of the cellswere scoredas ported,those with a successiveweighted missingdata) resulted in four trees345 steps bootstrap >85%were consideredstrongly long (ci 0.539, ri 0.482).The successive supported.In atheoreticalcontext, a clade weightingD analysisgave D three trees97.8 steps supported by one uncontradictedcharacter long (ci 0.840, ri 0.849).The strictconsen- correspondsto a bootstrapsupport of 63%, sustree D ofthe successiveD weighting analysis anda cladesupported by twouncontradicted ispresented in Figure 2.These treesare one charactersto 85% (Harshman, 1994; Farris step longer withunit weightsthan the trees et al.,1996). Here, however,this scale merely fromthe unit weight analysis.Unless other- relatesto the supportdistribution within the wisestated, the followingdiscussion refers analysis. tothe successiveweighting analysesof the complete rbcL data set. Investigatingthe justiŽcation of a priori RESULTS weighting (Sennblad andBremer, 2000), we The unit weight analysisof the complete madesome additional analyses of our rbcL rbcLdataset (237 characters total; 2.14% of dataset, including testsfor base composition the cellswere scoredas missing data) re- biasand rate heterogeneity .Thisdid not pro- sultedin 19,003trees 828 steps long, with duce signiŽcantly different results,and all aconsistencyindex ( ci) 0.378,and a re- well-supported groupswere congruent with tentionindex ( ri) 0.678.D The consensus thoseof the present study.The general struc- tree fromthe successiveD weighting analysis ture andthe well-supported groupsof the of 252most-parsimonious trees (each 177.3 resulting trees(Figs. 1and2) are to a large steps long, ci 0.661, ri 0.880)is presented degree congruent withthe molecularstud- in Figure 1.The D lengthD ofthese treeswith ies ofSennblad andBremer (1996),which unit weight charactersis 830 steps; thus these are based on rbcLdata;of Civeyrel et al. treesare not identical to any of the most- (1998),based mainly on matKsequence data; parsimonioustrees from the unit weight andof Potgieter and Albert (2001),based analysis.Branches not present in the strict mainly on trnL–Fdata.Although there are consensustree fromthe unit weight analysis differences in the weakly supported rela- areindicated on the combinable consensus tionshipsamong the well-supported groups, tree fromthe successiveweighting analysis in particularthe exactposition of the 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 395

FIGURE 1.Combinable component consensus tree of the 252most-parsimonious trees fromthe successive weighting analysisof the complete rbcLdataset. Subfamilial andtribal classiŽcation is accordingto Endressand Bruyns(2000). For subfamilies, athree-letter code is used:RAU, APO, PER,SEC, and ASC denote RauvolŽoideae, Apocynoideae, Periplocoideae, Secamonoideae,and Asclepiadoideae, respectively; OUT denotes the outgroup. Numbersbelow branchesare unit-weighted bootstrapvalues; successive weighted bootstrapvalues areindicated abovebranches. Branches not present in the strict consensus tree fromthe unit weight analysisare indicated with a cross (†).Forbranches marked A orB, the frequencies in the combinable component consensus are95% or 50%, respectively; all other brancheshave 100% frequency . 396 SYSTEMATICBIOLOGY VOL. 51

FIGURE 2.Strict consensus tree fromthe successive weighting analysisof the combined ndhF and rbcL data set. Tribal classiŽcation sensu Endressand Bruyns (2000) is indicated. Numbersbelow branchesare unit-weighted bootstrapvalues, whereas successive weighted bootstrapvalues areindicated abovebranches. Branches not present in the strict consensus tree fromthe unit weight analysisare indicated with across ( †).

Periplocoideae andof the rootof the in- The Apocynoideae–Asclepiadaceae clade, group, allwell-supported groupsin ouranal- whichis strongly supported ( uwboot 86%, ysisare congruent with those studies. The swboot 99%),is in turnnested withinD the combined analysisusing datafrom rbcL- subfamilyD RauvolŽoideae sensu Endressand and matK-sequences and oraland pollen Bruyns (2000). morphologyin the studyof the Apocy- Comparisonof our results to the most naceaes.str. by Endresset al.(1996), and recent classiŽcations of the Apocynaceae the analysisof Sennblad et al.(1998) of Endressand Bruyns (2000;see Table 1and the tribe Wrightieae sensu Leeuwenberg Fig. 1),shows that, although their circum- (1994a)and related taxa, using rbcL and scriptionof the subfamilies Periplocoideae, morphologicaldata, are largely congruent Secamonoideae,and Asclepiadoideae are withthe resultfrom the present study.Fur- monophyleticin ouranalysis, their sub- thermore,preliminary resultsfrom collab- familiesRauvolŽ oideae and Apocynoideae orativestudies by Sennblad et al.(aimed arenonmonophyletic. Similarly ,the tribes atthe traditionalApocynoideae andAs- ofthe RauvolŽoideae except Melodineae clepiadaceae,pers. comm.) and Endress aremonophyletic, but three of the tribes et al.(aimed atthe traditionalAlyxieae, ofthe Apocynoideae—the Apocyneae, the pers.comm.) based on combined analy- Echiteae,and the Wrightieae—are nonmono- sisof severalmolecular and morphologi- phyletic. Of the tribesof the Asclepiadoideae, caldatasets are also congruent withour the Ceropegieae andAsclepiadeae aremono- results. phyletic. The positionof Fockea as sister Support isgood for the splitbetween tothe restof the Asclepiadoideaeren- the ingroup andthe twooutgroup taxa, dersthe Marsdenieaenonmonophyletic. All Gelsemium and Mostuea (uwboot 96%, the included Periplocoideae taxabelong swboot 100%).Thus, the initialassump- D tothe tribe Periploceae sensu Venter and tionof Dmonophyly ofthe Apocynaceae Verhoeven (1997),except Pentopetia, which s.l.is not violated. Conforming to ear- they placed in the tribe Cryptolepideae. lier studies(e.g., Sennblad andBremer , Given thisunresolved positionof Pentopetia, 1996;Civeyrel etal., 1998; Sennblad etal., the monophyly ofthese tribescannot be eval- 1998;Potgieter and Albert, 2001),the tra- uated;however ,the reductionof Parquetina ditionalAsclepiadaceae are nested within tosynonymy with Periploca,proposedby the traditionalsubfamily Apocynoideae Venter andV erhoeven (1997),is not sup- (e.g., sensu Endressand Bruyns, 2000). portedby the present study. 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 397

DISCUSSION tribe Apocyneae, anda new namewould One of the mostimportant tasks for a be needed forthe taxoncorresponding to nomenclaturesystem is to provide asta- the currentview ofApocynaceae(e.g., sensu ble meansto communicate classiŽ cations Endressand Bruyns, 2000). A node-based (which wehere will understandas represent- deŽnition would handle thisparticular prob- ing phylogenetic relationships).Although lem but mayhave other problems, as illus- boththe Linnaeanand the Phylocode sys- tratedhere withanother hypothetical worst- temsare able tocommunicate phylogenies, casescenario: Early classiŽ cations of the they aresensitive to changes in phylogenetic Apocynaceaeincluded Plocosperma Benth. hypotheses,which in bothsystems will de- Anode-based deŽnition representing this mandtaxonomic changes. In addition,the view couldhave included asreference taxa, Linnaeansystem is subject toa potential say, Apocynum and Plocosperma .The recently instabilityrelating to subjective changesin proposedposition of Plocosperma close to circumscription,that is, lumpers andsplit- the Boraginaceae(Backlund et al.,2000) tersmay refer todifferent groupsof organ- wouldwith this deŽ nition have made Apoc- ismsby the samename (see deQueirozand ynaceaea very largetaxon (containing at Gauthier,1992).For many taxa, a consensus leastboth Gentianales and Boraginales), pos- circumscriptionhas eventually been estab- sibly synonymouswith (a hypothetical def- lished,but forother groups this is stilla prob- initionof) the informaltaxon Euasterids lem.A currentexample ofthis from Apoc- II(Bremer et al.,1998), and again a new ynaceaeis the genus Tabernaemontana sensu namefor the Apocynaceaesensu Endress Leeuwenberg (1991,1994b), which in the andBruyns (2000;i.e., excluding Plocosperma ) classiŽcation of Allorge (1985)is split into wouldhave been needed. Becausethe eight genera. The Phylocode systemwas de- Linnaeansystem allows recircumscription of signed toavoid this problem; by use ofstrict synonymousnames, the potentialturnover deŽnitions of taxa, subjective changesin in the setof names in use isreduced with circumscriptionare avoided. thissystem (albeit atthe expense ofallowing However,this rigidity in deŽnitions leads subjectivity). Additionally,the ranksof the toa potentialinstability and high turnover Linnaeansystem provide asetof namesthat ofnames associated with a group ofnested workas relativelystable universal standard taxa.Relatively small changes in phyloge- namesin communication;note that equal netic views,e.g., when adeŽning taxon, ranksdoes not guarantee comparable evo- “speciŽer,” receives alessnested position, lutionaryunits, however .Continuityin us- maycause some (or all) of these nested names age ofnames is important, for example, in tobecome synonymous(e.g., de Queiroz journals,literature databases,  oras,teach- andGauthier ,1994;Bryant, 1996; Liden and ing, andinventory work, and the need for Oxelman,1996; Sereno, 1999).The junior syn- suchcontinuity has been recognizedby pro- onymsshould, according to the Phylocode, ponentsof both nomenclature systems (de be rejected, andnew namesmay then be Queirozand Gauthier, 1992; Greuter et al., needed forthe new systemof nested clades 1994;Reveal, 1996). (de Queirozand Gauthier, 1994). The names Weproposea compromiseapproach that associatedwith this group oftaxaare, thus, usessolutions relating to these aspectsfrom replaced by new names.This can be espe- bothsystems, concentrating mainly on ciallyproblematic when carelessdeŽ nitions botanicalclassiŽ cation of extant plant taxa havebeen made,as have been notedby atthe tribaland familial levels andonly Schander andThollesson (1995) and Cantino consideringmonophyletic taxa. W erecom- et al.(1997) among others. Consider a hy- mend asystemthat uses “ standardnames,” potheticalworst-case example: The Ascle- (i.e., universally used communicationunits piadaceaewere traditionallyconsidered as atconvenient hierarchicallevels), suchas separatefrom Apocynaceae. A stem-based thoseprovided by the principal ranksof the deŽnition of Apocynaceae re ecting this Linnaeansystem (Greuter et al.,1994), but in view couldbe the largestclade that in- whichsensitivity toward subjective changes cludes Apocynum but not Asclepias. On the in circumscriptionis reduced. One wayto do present tree,such a deŽnition would limit thisis to adopt the deŽnitions of the phy- the Apocynaceaeto be synonymouswith the logenetic systembut use constraintssimilar 398 SYSTEMATICBIOLOGY VOL. 51 tothose that apply between ranksin the the explicit reference oftype specimens in the Linnaeansystem. T oreduce sensitivityto- minimum circumscriptionprovided by the wardchanges in phylogenetic hypotheses node-based partof the deŽnitions circum- (Schander andThollesson, 1995; Bryant, ventsthe tautologyproblem raisedby 1996,1997; Cantino et al.,1997; Moore, 1998), Bryant,1996). In the currentexample, this we proposeusing combined node stem- wouldmean that regardless of which name baseddeŽ nitions that will provokeCexplicit ofApocynaceae and Asclepiadaceae is re- incompatibilitiesbetween taxain caseof jected, the otherwill be deŽned ascorre- unfortunatechanges in circumscription(de sponding tothe Apocynaceaesensu Endress Queiroz,1996, has proposed a similartype of andBruyns (2000).The choicewill there- deŽnition for designating nonmonophyletic fore simply be whichname to keep. In this taxa;somewhat similar variants are also case,because Apocynaceaehas priority un- discussedin the Phylocode). Anode-based der the Linnaeansystem, we wouldchoose partof the deŽnition provides a minimal tokeep Apocynaceaeto promoteconsistency circumscription,whereas a stem-basedpart withthe currentsystem. Nevertheless, out- providesboundaries to competing taxa side these recommendations,choices of what (maximalcircumscription). If, ona particu- namesto choose will probably toa certain larphylogeny ,the minimal(monophyletic) degree be arbitrary.Wealsopropose that circumscriptionsof competing taxaare over- the rejected nameshould be reinstatedwith lapping, then the twotaxa are incompatible. anemended deŽnition if appropriate,to po- Tomakethis meaningful, atleast two taxa tentiallyallow for more stability in the set shouldbe included in the node-based part of namesin use. Thus,in oursecond ex- ofthe deŽnition; this will alsoavoid unnec- ample above,a node stem-baseddeŽ ni- essarymonotypic taxa. This incompatibility tionre ecting the inclusionC of Plocosperma will provide anobjective pointwhen nomen- in Apocynaceae(e.g., by exchanging Dic- claturalreconsideration is needed. Returning tyophleba for Plocosperma in the deŽnition tothe Žrstof ourexamples above, node above) wouldpresumably be incongruent stem-baseddeŽ nitions of the ApocynaceaeC withthe deŽnition of Gentianales as well as andAsclepiadaceae could be asfollows: the deŽnitions of Rubiaceae, Gentianaceae, andso forth,and therefore wouldbe rejected. Thefamily Apocynaceae is the most inclusive clade in In sucha case,where the only realchange isin the orderGentianales, including the type specimens the positionof asingle taxon,a reinstatement of Apocynum and Dictyophleba ,but not the taxaAscle- piadaceae[, Gentianaceae,Rubiaceae : : : ] ofthe Apocynaceaewith an emended def- initionis motivated. However, a drawback and isthat this may also introduce elements of subjectivity in the circumscriptionsof taxa. Thefamily Asclepiadaceae is the most inclusive clade Notall named clades need tobe provided in the orderGentianales, including the type speci- withstandard names. In fact,for the purpose mens of Asclepias and Periploca,but not the taxaApoc- ynaceae[, Gentianaceae,Rubiaceae : : : ] ofauniversalset used in journals,databases, andso forth, it may sufŽ ce touse standard Onthe tree in Figure 1,these twodeŽ nitions namescorresponding to the principal ranks aremutually exclusive, anda decisionon ofthe Linnaeansystem (Greuter et al.,1994), whichname to keep mustbe made.Mini- suchas species, genus, family,andorder . mizingthe number ofclassiŽcatory changes Nevertheless,to provide compatibilitywith shouldtake precedence in thisdecision. The the present system,we suggestthat addi- reasonfor this is to avoid the large,incon- tionalnames corresponding to commonly venient “standardname” changes that could used secondaryranks (e.g., tribes) maybe resultfrom relatively small rearrangements. recognized.The hierarchicallevel ofa stan- Rejected namesare then ignored when oc- dardname needs tobe indicated.This does curring in deŽnitions of other taxa. This is notimply thattaxa of the samehierarchi- because the reference taxain the stem-based callevel arecomparable evolutionary units partof the deŽnitions are “ deŽned” taxa but simply indicatesthe hierarchicallevel ratherthan physical types (thus,reliance on of ataxonrelative to nested taxa.Because deŽned reference taxa,which has been crit- suchmisinterpretations will in anycase be icizedfor stem-based deŽ nitions by Sereno, difŽcult to prohibit, one mayas well pro- 1999,is what we aimfor; on the otherhand, vide compatibilitywith the present Linnaean 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 399 systemby using the namesof the Linnaean node stem-baseddeŽ nitions may be able ranks(e.g., species,genus, tribe,family , tosubsume C suchrearrangements without order)to indicatehierarchical level. The prin- numerousexplicit recircumscriptions.W e ciple ofexhaustive subsidiary taxa is not will here use the resultfrom the present study accepted,so redundant taxa need notbe rec- asabasisto proposesuch a classiŽcation of ognized.Other ,“nonstandard,”taxa could the Apocynaceaes.l. be deŽned by using Phylocode deŽnitions— Ouraim is to base taxa on cladesthat are thatis, without the restraintsdiscussed for well supported.A secondaim is to make standardtaxa above. ourclassiŽ cation compatible with the cur- One advantageof the described system rentsystem. W ewill therefore primarilyrec- isthat utilizing the correspondencebetween ognizenonoverlapping tribesthat are valid the standardnames and Linnaean ranks al- under the Linnaeansystem. W ewill further lowsthe systemto be largely compatiblewith adoptthe principle ofnested referencing (as the establishedLinnaean system. A change of suggested by Lee, 1999,and Sereno, 1999) systemcould therefore be gradual,with the fromthe Linnaeansystem. Thus, the Žrstref- twosystems coexisting during aconversion erence taxonin the node-based partof the time(see e.g.,de Quieroz,1997b). deŽnition is the primarytype, whichcorre- Even though the node stem-based spondsto the Linnaeantype used forthe deŽnitions reduce the sensitivityC towards nameof the taxon.Instead of the (optional) changesin circumscriptions,they maystill be enumerationof subsumed taxaof a lower sensitivetoward “ unfortunate”or “bad”def- rank(e.g., genera orsubtribes), we will in- initions(see e.g.,Cantino et al.,1997; Cantino clude node stem-baseddeŽ nitions as dis- andde Queiroz,2000). Many ofthe recom- cussedabove. C Citations to primary types of mendationsfor such things as choicesof ref- reference taxa,in node-based partof deŽ - erence taxa(speciŽ ers) in phylogenetic def- nitions,are also nested, and thus refer to initionsare applicable alsoto the node “the type specimen ofthe type species of stem-baseddeŽ nitions (e.g., Schander andC the type genus : : : of the deŽned taxon.”A Thollesson,1995; Bryant, 1996; Cantino et al., problem isthatthe relationshipsbetween the 1997;Sereno, 1999).Most likely the compro- tribesare weakly supported.This may lead to misesystem described abovewill havesev- very cumbersomedeŽ nitions, enumerating eralfurther problems,and we hope thatthis allcompeting tribesin the stem-basedpart paper will invite further discussionon the ofthe deŽnition (e.g., Moore,1998). One way subject. toreduce thisproblem isto deŽ ne interme- diate,well-supported, taxaand use themas competing taxa.W etherefore will further rec- ANew ClassiŽcation ofthe Apocynaceae s.l. ognizefour nonstandardtaxa (note thatan Ourresults indicate that some problem- assignmentof “ nonstandard”taxa does not atictaxa remain in the classiŽcation of imply lessreliability; standard names relate Endressand Bruyns (2000).Monophyly for tocommunicationpurposes only). The taxa twoof the subfamilies andŽ veofthe tribes discussedare indicated in Figure 3.Wewill iscalled into question. If otherrecent molec- discussthe tree fromthe topof Figure 3,start- ularstudies (e.g., Potgieterand Albert, 2001) ing withthe taxaof the traditionalPlumeri- aretaken into consideration, a further four oideae.T ribes andsubtribes given in paren- tribes(Alstonieae, Alyxieae, Plumerieae, and thesesin the textrefer tothe classiŽcations Vinceae) maybe nonmonophyletic.This can ofEndressand Bruyns (2000);in othercases, in somecases be simply analysisartifacts, ourclassiŽ cation is implied. Forsome tribes but asEndressand Bruyns (2000)themselves wehavechosen a namewith priority under pointout, their classiŽcation is to be consid- Linnaeansystem that implies inclusionof a ered preliminary.InsufŽcient taxonsampling genus notsampled in the present study.For in molecularstudies and difŽ cult homology allbut one (Willughbeieae) ofthese cases, decisionsrelating to morphological charac- additionalmolecular or morphologicalphy- tersmake the circumscriptionof many of logenetic analysessupport the inclusionof the tribes,for example, in the Apocynoideae the genus. Even if toouncertain deŽ nitions sensu Endressand Bruyns (2000),uncertain. shouldbe avoided,this is less critical with There is,thus, risk for future rearrangements node stem-baseddeŽ nitions. In the case in manyof the tribes.A classiŽcation using of Willughbeieae,C we havedecided, from 400 SYSTEMATICBIOLOGY VOL. 51

FIGURE 3.The tribal reclassiŽcation discussed in the text indicated onthe combinablecomponent consensus tree fromthe successive weighting analysisof the complete rbcLdataset. Fournonstandard taxa— Apocynoidina, Euapocynoidina,Asclepiadacina, andAsclepiadoidina— are also indicated. discussionswith a morphologicalexpert itis not, our system allows redeŽ nition of (M. Endress,pers. comm.), that morphologi- the taxon.W ehavealso tried to provide calsupport exists for including Willughbeia commentson potential morphological char- Roxb.in Willughbeieae. If thisis correct, acterstaken from literature for the differ- the deŽnition given below will be valid;if ent taxarecognized. However, we havenot 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 401 performed anymorphological analysis, but asimilartype of stylehead (i.e., the apical rely oninformation external to this study enlarged partof the style), witha stigmatic forthese comments.In mostcases we can hollowand upper andlower hair wreaths therefore notidentify synapomorphiesfor (except in somespecies of Ochrosia), the pres- the taxa. ence ofanectardisk, and a linearhilum on the seed (Pichon, 1948b).On the present tree, TAXONOMIC TREATMENT Vinceae alsoinclude Kopsia;however,thispo- TribusWillughbeieae A.DC. sition of Kopsia isvery weakly supported (see alsoSennblad andBremer ,1996). The Ancylobotrys,Dictyophleba , and Va- DeŽnition.— Tribe Vinceae isthe most hadenia cladeis strongly supported ( uwboot inclusive cladethat includes the primary and swboot 100%)and corresponds to the types of Vinca and RauvolŽa but notAlyxieae, WillughbeieaeD ofEndressand Bruyns (2000). Aspidospermeae,Carisseae, Hunterieae, The present samplingcontains closely re- Melodineae, Plumerieae, Tabernaemon- latedgenera thathave been traditionally taneae,Willughbeieae, orApocynoidina. kept together. Landolphia P.Beauv.s.l.in- cludes allthese taxa(e.g., Schumann, 1895). In additionto the taxasampled here, the TribusTabernaemontaneae G.Don studyof Potgieter and Albert (2001)sup- The stronglysupported associationbe- portsthe inclusionof Couma Aubl., Lacmellea tween Tabernaemontana, Tabernanthe,Carval- H. Karst., Pacouria Aubl., and Saba (Pichon) hoa,Schizozygia , and Molongum (Tabernae- Pichon in thisclade. Plants in thisclade montaneae; uwboot 84%, swboot 98%) allhave a gynoecium thatis congenitally hasbeen suggested previouslyD (BoiteauD et al., syncarpousand indehiscent andthat con- 1978;Fallen, 1986;Endress et al.,1996; tainsa eshy pulp including numerousseeds Sennblad andBremer ,1996).Inclusion of taxa withcopious horny endosperm (Fallen, 1986; traditionallyassociated with Tabernaemon- Persoonet al.,1992). This probably con- tana (e.g., Voacanga Thouars and Callichilia stitutessynapomorphies for this clade Stapf; see alsoT abernaemontaneaesensu (M. Endress,pers. comm.), although not un- Leeuwenberg, 1994a)and Macoubea Aubl. ambiguously;for example, syncarpyis par- issupported by the analysisof Potgieter alleled in Carisseae.W ewill therefore as- andAlbert (2001).The taxashare sclerenchy- sume thatthe reference genus Willughbeia is maticanthers, free fromthe stylehead, and included inthistribe. acharacteristicband of heavily cutinized DeŽnition.— Tribe Willughbeieae isthe cellsat the insertionon the staminalrib mostinclusive cladethat includes the prim- (Endresset al.,1996). The taxatradition- ary types of Willughbeia and Dictyophleba but allyassociated with Tabernaemontana are notAlyxieae, Aspidospermeae,Carisseae, characterizedby their apocarpousfruits Hunterieae, Melodineae, Plumerieae, Taber- witharillate seeds, whereas genera asso- naemontaneae,V inceae, orApocynoidina. ciated with Molongum (Ambelanieae sensu Leeuwenberg, 1994a)have syncarpous fruits TribusVinceae Bartl. andlack an arillus.However, Macoubea forms alink between the two,having a syncar- Ochrosia,RauvolŽ a, V inca , and Catharanthus pousfruit witharillate seeds (Zarucchi et al., (allVinceae) forma stronglysupported clade 1995). (uwboot 89%, swboot 99%).The associa- DeŽnition.— Tribe Tabernaemontaneaeis tionbetween D RauvolŽa andD Catharanthus was the mostinclusive cladethat includes the pri- indicatedby Sennblad andBremer (1996). marytypes of Tabernaemontana and Schizozy- The studyof Potgieterand Albert (2001)sup- gia but notAlyxieae, Aspidospermeae,Caris- portsinclusion of the genera Laxoplumeria seae,Hunterieae, Melodineae, Plumerieae, Markgr., Neisosperma Raf., and Tonduzia Pit- Vinceae, Willughbeieae, orApocynoidina. tier,previously classiŽed with Aspidosperma . The Vinceae containboth herbaceous and fruticosetaxa. The fruitsare apocarpous, TribusAspidospermeae Miers but whereasthe fruitsof Catharanthus,Lax- Aspidosperma and Vallesia (Aspidosper- oplumeria,Neisosperma, T onduzia , and Vinca meae) forma stronglysupported clade( uw- havedry fruit walls,the fruitsof Ochrosia and boot and swboot 100%)Ž rstsuggested RauvolŽa aredrupes. Plantsin Vinceae share in Sennblad (1997;D see alsoSennblad and 402 SYSTEMATICBIOLOGY VOL. 51

Bremer,2000). Subsequently , Geissospermum ofthe twogenera (Pichon, 1948a,b),the fol- Allemao,˜ Haplophyton A.DC., Microplumeria lowingsimilarities also emerge: presence of Baill., and Strempeliopsis Benth. were indi- astipularline, pollen in tetrads,and a punc- catedto belong tothis clade (Potgieter and tiformhilum. Albert, 2001).The Aspidospermeaehave DeŽnition.— Tribe Melodineae isthe most simple styleheads, lacking bothstigmatic inclusive cladethat includes the pri- hollowand upper hairwreath, and have marytypes of Melodinus and Craspidosper- apocarpousfruits with variation in eshi- mum but notAlyxieae, Aspidospermeae, nessand scleriŽ cation. Alstonia has tradi- Carisseae,Hunterieae, Plumerieae, Taber- tionallybeen placed in the Plumerieae. Be- naemontaneae,Vinceae, Willughbeieae, or causeits seeds have a hairymargin, it has Apocynoidina. been suggested toform a link tosubfam- ily Apocynoideae.This was contradicted by TribusHunterieae K.Schum. Sennblad andBremer ’sstudy(1996), how- The Picralima and Pleiocarpa clade is ever, where Alstonia wasplaced in aniso- stronglysupported ( uwboot and swboot latedposition as the sistergroup tothe 100%)and corresponds to the subtribeD restof the Apocynaceaes.l. Here itgroups Hunterieae sensu Endressand Bruyns (2000). withAspidospermeae Sensu Endressand The studyof Potgieterand Albert (2001) Bruyns (2000).However ,thisassociation is alsosupports the inclusionof Hunteria Roxb. very weakly supported ( uwboot and swboot These taxaare characterized by apocar- < 50%),and in otherrecent studiesAs- pous,sometimes pluricarpous, ovaries and pidospermeae and Alstonia do not form a eshy,Žbrousfruit walls(Omino, 1996). clade(Potgieter andAlbert, 2001).Because Diplorhynchus (Melodineae) ishere weakly ofthis,we donotuse itas areference taxon associated( uwboot and swboot < 50%) with in the deŽnition. Thus, although the Aspi- the Pleiocarpeae.It shares a few similari- dospermeaeinclude Alstonia onthe present ties,such as stipular lines anda stylehead tree,some other position of Alstoniacan withouta stigmatichollow and hair wreaths, be accommodatedwithout amending the but alsohas differences, suchas a dry de- deŽnition. hiscentfruit. With the present deŽnition, DeŽnition.— Tribe Aspidospermeaeis the Diplorhynchus will tentativelybe included in mostinclusive cladethat includes the the Hunterieae. primarytypes of Aspidosperma and Vallesia DeŽnition.— Tribe Hunterieae isthe most but notAlyxieae, Carisseae,Hunterieae, inclusive cladethat includes the primary Melodineae, Plumerieae, Tabernae- types of Hunteria and Picralima but notAlyx- montaneae,Vinceae, Willughbeieae, or ieae, Aspidospermeae,Carisseae, Melod- Apocynoidina. ineae, Plumerieae, Tabernaemontaneae, Vinceae, Willughbeieae, orApocynoidina. TribusMelodineae G. Don The well-supported associationbetween TribusPlumerieae Endl. Melodinus and Craspidospermum (uwboot The stronglysupported relationship 70%, swboot 75%)was Ž rstreported D (uwboot and swboot 100%)between Alla- by Sennblad (1997;D see alsoSennblad and manda and PlumeriaDcorrespondsto earlier Bremer,2000) and has subsequently been studies(Endress et al.,1996; Sennblad and reected in the tribe Melodineae sensu Bremer,1996;Civeyrel et al.,1998; Potgieter Endressand Bruyns (2000;their inclusionof andAlbert, 2001).In allearlier treatments, Diplorhynchus ,however,isnot supported). Allamanda hashad uncertain relationships. Craspidospermum ,althoughits fruit isdry The associationwith Plumeria Žndssupport anddehiscent asopposed to the indehis- in pollen morphology,suchas perforate centfruits of Melodinus,wasnoted by Pichon mesocolpialdepressions and similar inner (1948b)to present severalcharacters of the exine pattern.The twogenera alsocontain Carisseae,such as stamensinserted near the secoiridoidsrather than the indole alka- baseof the corollatube anda syncarpous loidsand cardenolides that are common ovary.In particular,Pichon found the very in the traditionalPlumerioideae (Endress dense suprastaminalindumentum type rem- etal., 1996). Cameraria, Cerbera, and Thevetia iniscent of Melodinus.Fromhis descriptions form astronglysupported association 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 403

(uwboot 83%, swboot 91%) with Ane- Afurther inclusionof Condylocarpon Desf., chites,whichD has been suggestedD by Fallen Lepiniopsis Valeton,and Plectaneia Thouars (1983).This association was based mainly wasindicated by Potgieterand Albert (2001). onthe latrorse/sublatrorseanthers and The constituenttaxa are the only Apocy- the broadstyle head with large apical naceaetaxa outside the Apocynoidinathat appendages andstigmatic hollow .The study haveporate pollen. by Potgieterand Albert (2001)suggested DeŽnition.— Tribe Alyxieae isthe mostin- afurther inclusionof Cerberiopsis Viell. Ex. clusive cladethat includes the primarytypes Pancher & Sebert´ and Skytanthus Meyen in of Alyxia and Lepinia but notAspidosper- thistribe. These taxaform a well-supported meae,Carisseae, Hunterieae, Melodineae, clade with Plumeria and Allamanda, which Plumerieae, Tabernaemontaneae,Vinceae, correspondsto the Plumerieae sensu Willughbeieae, orApocynoidina. Endressand Bruyns (2000).Characters supporting thisrelationship are the presence ofinfrastaminal as well assuprastaminal APOCYNOIDINA appendages (notpresent in Anechites and Onthe present tree,the traditionalsubfam- Plumeria)andwinged seeds(not present in ily Apocynoideae includes allsubfamilies Anechites)(Endresset al.,1996). (Periplocoideae, Secamonoideae,and Ascle- DeŽnition.— Tribe Plumerieae isthe piadoideae) ofthe Asclepiadaceaesensu mostinclusive cladethat includes the Liede andAlbers (1994).All ofthese taxa primarytypes of Plumeria and Allamanda areestablished names, and selecting one as but notAlyxieae, Aspidospermeae,Caris- asubfamily atthe expense of the others seae,Hunterieae, Melodineae, Tabernae- mightbe consideredunfortunate from dif- montaneae,Vinceae, Willughbeieae, or ferent pointsof view .Wehavehere chosen Apocynoidina. notto recognizesubfamilies; instead, we will deŽne the traditionalApocynoideae, Ascle- TribusCarisseae Dumort. piadaceae,and Asclepiadoideae as nonstan- dardtaxa. The correspondenceto the tradi- Acokanthera and Carissa (both Carisseae) tionaltaxa will be indicatedin the namesby aretraditionally joined andare sometimes replacing the sufŽx -eae witha neutralsuf- even treatedas one genus (e.g., Pichon, Žx-ina (Kron,1997). W ewill also,for practi- 1948a).In the present studythey forma calreasons, recognize as anonstandardtaxon stronglysupported clade( uwboot 96%, the informalgroup euapocynoids,suggested swboot 99%).Among the charactersD sup- by Sennblad et al.(1998). portingDthis tribe aresyncarpous fruits with- The taxaof the traditionalApocynoideae outaxile placentation and endocracks on the andAsclepiadaceae ( uwboot 86%, swboot inner exine ofthe pollen (Endresset al.,1996). 99%)clade will in manycases D be moredif- D DeŽnition.— Tribe Carisseaeis the mostin- Žcultto safely delimitto tribes, because clusive cladethat includes the primarytypes manyof the groupsin thisclade have weak of Carissa and Acokanthera but notAlyxieae, ornosupport (although preliminary results Aspidospermeae,Hunterieae, Melodineae, froman unpublished collaborativestudy by Plumerieae, Tabernaemontaneae,Vinceae, Sennblad et al.indicate additional support Willughbeieae, orApocynoidina. forthe tribesdiscussed below). The tribal andsubtribal classiŽ cation of the traditional TribusAlyxieae G. Don Apocynoideae hasbeen shownto be prob- Chilocarpus hasbeen agenus withun- lematic(Leeuwenberg, 1994a;Endress and certainafŽ nities. In Endresset al.(1996), it Bruyns,2000). The circumscriptionof tribes wasthe sistertaxon to the Apocynoideae. forthese taxamight therefore in somecases Here itis associated with Alyxia and Lep- be preliminary. inia in acladecorresponding to the Alyxieae In allfollowing taxa,the basalpart of the sensu Endressand Bruyns (2000).This antherconnective, called the retinacle,is ad- cladeis strongly supported withsucces- nateto the stylehead. siveweighted bootstrap( swboot 91%), but DeŽnition.— Apocynoidinaare the most notwith unit-weighted bootstrapD support inclusive cladethat includes the pri- (uwboot 58%);such clades will henceforth marytypes of Apocynum and Wrightia but be termedD moderately well supported clades. notAlyxieae, Aspidospermeae,Carisseae, 404 SYSTEMATICBIOLOGY VOL. 51

Hunterieae, Melodineae, Plumerieae, Taber- (uwboot 57%, swboot 74%),whereas naemontaneae,Vinceae, orWillughbeieae. the correspondingD clade Dis well supported in the combined rbcL and ndhF analysis TribusWrightieae G. Don (uwboot 66%, swboot 100%).On the ba- sisof identiŽ D ed synapomorphiesD (presence The Stephanostema and Wrightia (both ofcalciumoxalate crystals in the stomiumof Wrightieae) cladeis strongly supported the anthers,absence ofinterpetal vein, ses- (uwboot and swboot 100%)and corresponds D sile Žlaments,and absence ofair spaces in tothe Wrightieae sensu Sennblad et al.(1998), the anthers),Sennblad et al.(1998) predicted whoalso suggested aninclusion of Pleio- aninclusionof AlaŽa Thouars, Kibatalia, and ceras Baill.The tribe couldbe characterized Malouetia A.DC.in the tribe.The inclusionof by acombinationof synapomorphicand ple- Kibatalia isstrongly supported in thisstudy . siomorphiccharacters, for example presence Followingthese suggestions,we will here as- ofachalazaland absence ofmicropylar coma, sume the inclusionof the reference genus left contortedaestivation, and absence ofair Malouetia in thistribe. (The deŽnition below spacesin the anthers(1998). will include Strophanthus onthe present tree, DeŽnition.— Tribe Wrightieae isthe most but see commentsunder Nerieae.) inclusive cladein the Apocynoidinathat in- DeŽnition.— Tribe Malouetieae isthe most cludes the primarytypes of Wrightia and inclusive cladein the Apocynoidinathat in- Stephanostema but notMalouetieae, Nerieae, cluded the primarytypes of Malouetia and orEuapocynoidina. Funtumia but notNerieae, Wrightieae,or Euapocynoidina. TribusNerieae (Benth.) M.Pichon Adenium,Nerium , and Strophanthus (all EUAPOCYNOIDINA Wrightieae) constitutedthe Nerieae in In allthe remaining representativesof Sennblad et al.(1998); however, the clade the Apocynoideae sensu Endressand received weaksupport. Here Adenium and Bruyns (2000),that is, Mandevilla, Mesechites, Nerium forma cladethat is absent in half Aganosma, Apocynum, Trachelospermum , ofthe most-parsimonioustrees in the com- Rhabdadenia , Beaumontia, Parsonsia, and plete rbcLanalysisbut isstrongly supported Prestonia,the anthersare adnate to the style in the combined ndhF and rbcL analysis head bothby the retinacleand by the thecae. (uwboot 98%, swboot 99%). Strophan- D D These taxatogether with the taxaof the tra- thus groupswith the Mascarenhasia,Pachy- ditionalAsclepiadaceae form a moderately podium,Kibatalia, Funtumia , and Holarrhena supported clade( uwboot < 50%, swboot clade,which corresponds to the Malouetieae 86%)that corresponds to the informaltaxon D sensu Sennblad et al.(1998). This position of euapocynoidssensu Sennblad et al.(1998). Strophanthus is,however ,weakly supported DeŽnition.— Euapocynoidinaare the most (uwboot and swboot < 50%)andis not present inclusive cladethat includes the primary in the combined ndhF and rbcL tree. Fur- types of Apocynum and Echites but not thermore,certain morphological characters Malouetieae,W rightieae,or Nerieae. (unfused slitsin the corollatube andan apicalanther appendage) indicatea posi- TribusMesechiteae Miers tioncloser to the Nerieae (Sennblad etal., 1998). Mandevilla and Mesechites (both Mese- DeŽnition.— Tribe Nerieae isthe mostin- chiteae) forma moderatelysupported clade (uwboot < 50%, swboot 87%),equivalent clusive cladein the Apocynoidinathat in- D cludes the primarytypes of Nerium and Ade- tothe Mesechiteae. Thistribe couldbe char- nium but notMalouetieae, Wrightieae, or acterizedby antherswith obtuse, truncated Euapocynoidina. tailsand a retinaclelacking hairs,and also a stronglypentagonal style head with a stig- matichollow . TribusMalouetieae M ull-Arg.¨ DeŽnition.— Tribe Mesechiteae isthe most The associationof Mascarenhasia , Pachy- inclusive cladein the Euapocynoidinathat podium,Kibatalia, Funtumia , and Holar- includes the primarytypes of Mesechites rhena (all Wrightieae) ismoderately well and Mandevilla but notApocyneae, Echiteae, supported in the complete rbcL analysis Periploceae, orAsclepiadacina. 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 405

TribusApocyneae Bercht. etJ.Presl. Potgieterand Albert, 2001).The taxaof Aganosma,Apocynum, Beaumontia , and the traditionalAsclepiadaceae have “ trans- Trachelospermum (Apocyneae) forma weakly lators,”structures related to pollination supported cladewith Rhabdadenia (Echiteae) specialization.In the Periploceae, the pollen (uwboot and swboot < 50%)in the complete isdeposited as tetrads in aspoon-like rbcLanalysis,whereas the corresponding translatorwith an adhesivedisc. cladein the combined ndhF and rbcL analy- DeŽnition.— Tribe Periploceae isthe most sisgains a littlebetter support( uwboot < 50%, inclusive cladein the Euapocynoidinathat swboot 91%).Likewise, thisclade (with the includes the primarytypes of Periploca inclusionD of Chonemorpha G.Don) isweakly and Pentopetia but notApocyneae, Echiteae, supported in Potgieterand Albert (2001).Po- Mesechiteae, orAsclepiadacina. tentialcharacterizing traits could be astyle headwithout stigmatic hollow in combina- ASCLEPIADACINA tionwith thecae that are adnate to the style The taxaof the Asclepiadoideaesensu head (Sennblad etal., 1998). Liede andAlbers (1994)form a cladetogether DeŽnition.— Tribe Apocyneae isthe most with the two Secamone species and Baissea inclusive cladein the Euapocynoidinathat (Apocyneae; uwboot < 50%, swboot 73%). includes the primarytypes of Apocynum The exclusionof Periploceae (traditionallyD and but notEchiteae, Mese- Trachelospermum associatedwith this clade; see Civeyrel et al., chiteae,Periploceae, orAsclepiadacina. 1998)from this taxon is weakly supported. Therefore, Periploceae isnot included asa TribusEchiteae Bartl. reference taxonin the stem-basedpart of Parsonsia,Peltastes , and Prestonia (all the deŽnition of the Asclepiadacina,thus Echiteae) forma moderatelysupported or leaving open the possibilityfor a laterin- unsupported cladewith the Periplocoideae clusionof this taxon. In the Asclepiadacina representativesof the study( uwboot and (except Baissea)the pollen isagglutinated swboot < 50%, and uwboot < 50%, swboot intopollinia that are connected to aclasping 89%in the complete rbcLandthe combinedD translator. rbcL and ndhFanalysis,respectively). In the DeŽnition.— Asclepiadacinaare the most studyof Sennblad et al.(1998), Prestonia and inclusive monophyleticclade in the Euapoc- Parsonsia formeda weakly supported clade, ynoidinathat includes the primarytypes of andin the studyof Potgieter and Albert Asclepias and Secamoneae but notApocyneae, (2001),inclusion of both Parsonsia and Echites Echiteae,or Mesechiteae. P.Brownein the Echiteaeis supported. Parsonsiaand Echiteae also share certain TribusSecamoneae G.Don characters,such as ahorseshoe-shapedreti- The tribe Secamoneaesensu Bruyns nacleand a stylehead withstigmatic hollow (uwboot 89%, swboot 96%)is mono- (see Sennblad et al.,1998). phyletic inD the present analysis.D However ,be- DeŽnition.— Tribe Echiteaeis the mostin- causerepresentatives from only one ofthe clusive cladein the Euapocynoidinathat constituentgenera areincluded, monophyly includes the primarytypes of Echites and isnot well tested,but the studyof Civeyrel Prestonia but notApocyneae, Mesechiteae, et al.(1998), which included alargernumber Periploceae, orAsclepiadacina. oftaxa, including Pervillea Decne., showed thatmonophyly ofthe tribe wassupported. TribusPeriploceae Bartl. The traditionalcharacter for this tribe isfour polliniaper translator. The representativesof the Periplo- DeŽnition.— Tribe Secamoneaeis the most coideaesensu Endressand Bruyns (2000)— inclusive cladein the Asclepiadacinathat in- Pentopetia,Parquetina, Petopentia, Periploca, cludes the primarytypes of Secamone and Tacazzea, and Mondia—group togetherin Pervillea but notAsclepioidina or Baissea. awell-supported clade( uwboot 67%, swboot 93%).The monophyly ofthe D tra- ditionalDperiplocoid taxais also supported Baissea in earlierstudies that included awider or The positionof Baissea correspondswith different sampling(Civeyrel et al.,1998; thatin the studyof Sennblad et al.(1998) 406 SYSTEMATICBIOLOGY VOL. 51 andis also supported by the combined ndhF cludes the primarytypes of Fockea and Fockea and rbcL analysis (uwboot 55%, swboot but notCeropegieae, Marsdenieae,or Ascle- 97%).The unexpected positionD of the BaisseaD piadeae. (Žrst suggested in Sennblad, 1997,see also Sennblad andBremer, 2000) nested within TribusCeropegieae Decne. the traditionalAsclepiadaceae prompted us toverify the rbcLsequence by resequencing; The tribe Ceropegieae sensu Endressand the resultsturned outto be identical.Further- Bruyns (2000)is represented by Stapelia and more,in the studyof Potgieter and Albert Ceropegia andis very stronglysupported (uwboot and swboot 100%).This traditional (2001),based on trnL–Fdata,an indepen- D dent extraction(but sampledfrom the same group isstrongly supported by morphol- ogy,andthe inclusionof further traditional specimen) placed Baissea in the sameposi- tion.No obvious characters connect Baissea stapeliadtaxa is indicated in Potgieterand tothe traditionalAsclepiadoideae and Seca- Albers (2001). monoideae,but suggestive qualitiesmay be DeŽnition.— Tribe Ceropegieae isthe most the tendencies of the stamensto have dor- inclusive cladein the Asclepiadoidinathat salstaminal appendages, ofthe styleto have includes the primarytypes of Stapelia and but notAsclepiadeae, Fockeeae, or elongatedapices (compare certain Secamone), Ceropegia andof the bulbs orridges tobe below oratthe Marsdenieae. Žlamentinsertions of certain species (sug- gesting abasaltube). Translatorsof a very TribusMarsdenieae Benth. simple type arealso found in Baissea. Because Stephanotis,Micholitzia , and Hoya repre- ofitsuncertain position, there areno obvious sentthe tribe Marsdenieaesensu Endress sistergroups to Baissea.Also,to avoidmono- andBruyns (2000)and form a very well typic taxa,we donot assign Baissea to a tribe supported clade( uwboot 91%, swboot (the principle ofexhaustivesubsidiary taxa 95%).The positionof FockeaD assister group D isnot followed). tothe restof the Asclepiadoidinamakes the Marsdenieaesensu Endressand Bruyns (2000)nonmonophyletic. Stephanotis is most ASCLEPIADOIDINA likely congeneric with,or sister to, Marsdenia The genera ofthe traditionalAsclepi- R.Br.,whichwe therefore will assumeto be oideaeform a moderatelysupported clade included in the tribe.The studyof Potgieter (uwboot 58%, swboot 84%),which will andAlbers (2001)supports further inclusion D D here be treatedas a nonstandardtaxon. of Dischidia R.Br., Dregea E. Mey., and Telosma The traditionalcharacters for the Asclepi- Coville. adoideaeare clasping translators with two DeŽnition.— Tribe Marsdenieaeis the most polliniaper translator. inclusive cladein the Asclepiadoidinathat DeŽnition.— Asclepiadoidinaare the most includes the primarytypes of Marsdenia and inclusive cladethat includes the primary Hoya but notAsclepiadeae, Fockeeae, or types of Asclepias and Fockea but notSeca- Ceropegieae. moneae. TribusAsclepiadeae (R. BR.) Duby TribusFockeeae Kunzeet al. Schizostephanus,Asclepias, Calotropis, V ince- Fockea (Marsdenieae) hasbeen placed with toxicum,T ylophora,Orthosia, Araujia, T weedia, Cibirhiza Endl.in the tribe Fockeeae. In this Fischeria, and Matelea forma stronglysup- analysis Fockea isthe sistergroup tothe rest portedclade ( uwboot 89, swboot 96) corre- ofthe Asclepiadoideae( uwboot 58%, swboot sponding toAsclepiadeaeD sensu EndressD and 84%).This position is similar Dto the posi- Bruyns (2000).The inclusionof Tylophora and tionD of Fockea in the studiesby Civeyrel et al. the traditionalGonolobeae (here represented (1998)and Potgieter and Albert (2001).Inclu- by Fischeria and Matelea)in the Asclepiadeae sionof the tribe Fockeeae (Kunze et al.,1994) (e.g., Liede, 1996;Sennblad andBremer ,1996, in the Marsdenieae(Endress and Bruyns, 2000;Swarupanandan et al.,1996; Civeyrel 2000)is thus not supported. et al.,1998; Endress and Bruyns, 2000; DeŽnition.— Tribe Fockeeae isthe mostin- Potgieterand Albert, 2001)is thus sup- clusive cladein the Asclepiadoidinathat in- ported.The studiesof Civeyrel et al.(1998) 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 407 andPotgieter and Albert (2001)also indicate BRYANT,H.N.1997.Cladistic informationin phyloge- supportfor inclusion of othertraditional As- netic deŽnitions anddesignated phylogenetic con- texts forthe use of taxonnames. Biol. J.Linn.Soc. clepiadeae taxa. 62:495–503. DeŽnition.— Tribe Asclepiadeae isthe most CANTINO, P. D., AND K. DE QUEIROZ.2000.PhyloCode: inclusive cladein the Asclepiadoidinathat Aphylogenetic code of biologicalnomenclature. Draft includes the primarytypes of Asclepias and available onhttp:/ /www.ohiou.edu/phylocode/ Matelea but notFockeeae, Marsdenieae,or CANTINO,P.D.,R. G.O LMSTEAD, AND S. J. WAGSTAFF. 1997.A comparison ofphylogenetic nomenclature Ceropegieae. with the current system:A botanicalcase study.Syst. Biol. 46:313–331. CIVEYREL, L., A. LE THOMAS, K. FERGUSON, AND M. W. ACKNOWLEDGMENTS CHASE.1998.Critical reexaminationof palynological WethankK. Bremer,M. Fishbein,R. Olmstead, charactersused to delimit Asclepiadaceae in compari- andtwo anonymousreviewers forvaluable comments sonto the molecular phylogenyobtained from plastid anddiscussion ofthe manuscript. N.Heidariis ac- matKsequences. Mol.Phylogenet. Evol. 9:517–527. knowledged formuch help with the sequencing.M. DE QUEIROZ,K.1996.A phylogenetic approachto bi- Backlund,M. Chase,L. Civeyrel, J.Clark,M. Endress,A. ological nomenclature asan alternative to the Lin- Leeuwenberg, U.Meve, B.Oxelman,B. Pettersson, neansystem in current use. In Biological nomencla- E.Robbrecht,and S. Zonakindly shared their mate- ture in the 21stcentury (J.L. Reveal). http://www. rial. M.Endress,A. Leeuwenberg, andU. Meve also inform.umd.edu/PBIO/nomcl/dequ.html.,Univ .of controlled the determination ofmany specimens. K. Maryland. Andreasen,A. Backlund,and J. Kukkaare thanked DE QUEIROZ,K.1997a.Misunderstandings about the forvaluable help andcompany during Ž eld trips. phylogenetic approachto biological nomenclature: A TheUppsala Botanical Garden,the Uppsala Botanical reply to Liden andOxelman. Zool. Scr.26:67– 70. Museum,the Fairchild Tropical Garden,the Munster¨ DE QUEIROZ,K.1997b.The Linnaean hierarchy and the Botanical Garden,the AarhusBotanical Garden,and evolutionization of taxonomy,with emphasison the the StockholmUniversity were most helpful in provid- problem ofnomenclature. Aliso 15:115–144. ingmaterial. This study was supported bythe Swedish DE QUEIROZ, K., AND J. GAUTHIER.1992.Phylogenetic ResearchCouncil to B.B.anda travel grantfrom J. A. taxonomy.Annu.Rev .Ecol. Syst.23:449– 480. Wahlberg’s memorial fundto B.S. DE QUEIROZ, K., AND J. GAUTHIER.1994.Toward a phy- logenetic systemof biological nomenclature. Trends Ecol. Evol. 9:27–31. DEMETER,K.1922.V ergleichende Asclepiadeenstudien. REFERENCES Flora(Jena) 115:130– 176. ALLORGE,L.1985.Monographie des Apocynacees—´ DOYLE, J. J., AND J. L. DOYLE.1987.A rapidDNA isola- Tabernamontanoidees´ Americaines. M´em.Mus. Nat. tion procedure forsmall quantities offreshleaf tissue. Hist.Nat. S er´ .BBot.30, 216pp. Phytochem.bull. 19:11–15. BACKLUND, M., B. OXELMAN, AND B. BREMER. 2000.Phy- ENDRESS, M. E., AND P. V. BRUYNS.2000.A revised clas- logenetic relationships within the Gentianalesbased siŽcation of the Apocynaceae s.l.Bot. Rev .66:1–56. on ndhF and rbcLsequences, with particular reference ENDRESS, M. E., B. SENNBLAD, S. NILSSON, L. CIVEYREL, to the Loganiaceae.Am. J. Bot.87:1029– 1043. M. W. CHASE, S. HUYSMANS, E. GRAFSTRO¨ M, AND BAUM,D.A.,W.S.A LVERS ON, AND R. NYFFELER. 1998.A B. BREMER.1996.A phylogenetic analysisof Apoc- durianby any other name:T axonomyand nomencla- ynaceaes.str. and some related taxain Gentianales:A ture of the core Malvales.Harv .Pap.Bot. 3:313– 330. multidisciplinary approach.Op. Bot. Belg. 7:59– 102. BOITEAU, P., L. ALLORGE, AND C. SASTRE. 1978. Mor- ERIKS SON, T., M. J. DONOGHUE, AND M. S. HIBBS. phologie orale des Apocynaceae:II. Caract` eres 1998.Phylogenetic analysisof Potentilla using DNA distinctifs entre Ambelanieae (Plumerioideae) et sequences of nuclear ribosomalinternal transcribed Macoubeae(T abernaemontanoideae).Adansonia n.s. spacers (ITS),and implications forthe classiŽcation of 18:267–277. Rosoideae (Rosaceae).Plant Syst.Evol. 211:155–179. BREMER, K., M. W. CHASE, P. F. STEVENS , A. A. FALLEN,M.E.1983.A systematic revision of Anechites ANDERBERG, A. BACKLUND, B. BREMER, B. G. BRIGGS, (Apocynaceae).Brittonia 35:222–231. P. K. ENDRESS, M. F. FAY, P. GOLDBLATT, M. H. G. FALLEN,M.E.1986.Floral structure in the Apocy- GUSTAFSSON, S. B. HOOT, W. S. JUDD, M. KALLERSJ¨ O¨ , naceae:Morphological, functional, andevolutionary E. A. KELLOGG, K. A. KRON, D. H. LES, C. M. aspects. Bot.Jahrb. Syst. P anzengesch. P anzen- MORTON, D. L. NICKRENT, R. G. OLMSTEAD, R. A. geogr.106:245– 286. PRICE, C. J. QUINN, J. E. RODMAN, P. J. RUDALL, FARRIS,J.S.1969.A successive approximationsap- V. SAVOLAINEN, D. E. SOLTIS, P. S. SOLTIS, K. J. proachto characterweighting. Syst.Zool. 18:374–385. SYTSMA, AND M. THULIN.1998.An ordinalclassiŽ - FARRIS,J.S.1989.The retention index andthe rescaled cation forthe families of owering plants.Ann. M. consistency index.Cladistics 5:417–419. Bot.Gard. 85:531– 553. FARRIS,J.S.,V.A.A LBERT, M. KALLERSJ¨ O¨ , D. LIPS COMB, BROWN,R.1810.On the Asclepiadeae. Preprinted from AND A. G. KLUGE.1996.Parsimony jackkniŽ ng out- 1811,Mem. Wernerian Nat. Hist. Soc. 1:12– 78. performs neighbor-joining.Cladistics 12:99–124. BRYANT,H.N.1996.Explicitness, stability,anduni- FELSENSTEIN,J.1985.ConŽ dence limits onphylogenies: versality in the phylogenetic deŽnition andusage of An approachusing the bootstrap.Evolution 39:783– taxonnames: A case studyof the phylogenetic taxon- 791. omyof the Carnivora(mammalia). Syst. Biol. 45:174– GOLENBERG,E.M.,M.T.C LEGG, M. L. DURBIN, 189. J. DOEBLEY, AND D. P. MA.1993.Evolution of a 408 SYSTEMATICBIOLOGY VOL. 51

noncodingregion of the chloroplast genome.Mol. OLMSTEAD, R. G., AND J. A. SWEERE.1994.Combin- Phylogenet.Evol. 2:52–64. ingdata in phylogenetic systematics: An empiri- GREUTER, W., F. R. BARRIE, H. M. BURDET, W. G. cal approachusing three molecular datasets in the CHALONER, V. DEMOULIN, D. L. HAWKSWORTH, Solanaceae.Syst. Biol. 43:467–481. P. M. JØRGENSEN, D. H. NICHOLSON, P. C. SILVA, OMINO ,E.1996.A contribution to the leaf anatomyand P. TREHANE, AND J. MCNEILL.1994.International Code taxonomyof Apocynaceae in Africa. Theleaf anatomy of Botanical Nomenclature (TokyoCode). Regnum ofApocynaceae in EastAfrica. Amonographof Pleio- Vegetabile 131. carpinae(Series of revisions of Apocynaceae XLI). HARSHMAN,J.1994.The effect of irrelevant characters WageningenAgr. Univ .Pap.96:1– 178. onbootstrap values. Syst.Biol. 43:419–424. OXELMAN, B., M. BACKLUND, AND B. BREMER. 1999. Re- JUDD,W.S., R.W.S ANDERS, AND M. J. DONOGHUE. 1994. lationships of the Buddlejaceae s.1.investigated using Angiosperm familypairs: Preliminary phylogenetic parsimonyjackknife and branch support analysisof analyses.Harv .Pap.Bot. 5:1– 51. chloroplast ndhF and rbcLsequence data.Syst. Bot. KALTENBOECK, B., J. W. SPATAFORA, X. ZHANG, K. G. 24:164–182. KOUSOULAS, M. BLACKWELL, AND J. STORZ. 1992. PERSOON,J.G.M.,F.J.H. VAN DILST, R. P. KUIJPERS, EfŽcient production ofsingle-strandedDNA aslong A. J. M. LEEUWENBERG,andG. J.A.V ONK. 1992. The as2 kbfor sequencing of PCR-ampliŽed DNA. African species of Landolphia P.Beauv.(Series ofrevi- Biotechniques 12:164–171. sions of Apocynaceae XXXIV).W ageningenAgr .Univ. KIM, K.-J., AND R. K. JANSEN. 1995. ndhFsequence evo- Pap.92:1– 232. lution andthe majorclades in the sunower family. PICHON,M.1948a.ClassiŽ cation des Apocynac´ees I. Proc. Nat.Acad. Sci. USA92:10379–10383. Carissees´ et Ambelani´´ ees. Mem.´ Mus. Nat. Hist. Nat. KRON,K.A. 1997.Exploring alternative systemsof clas- 24:111–181. siŽcation. Aliso 15:105–112. PICHON,M.1948b.ClassiŽ cation des Apocynac´ees: KUNZE, H., U. MEVE, AND L. LIEDE. 1994. Cibirhizaalber- IX,RauvolŽees,´ Alstoni´ees, Allamandees´ et siana,anew species ofAsclepiadaceae, andestablish- Tabernaemontanoid´ ees´ . Mem.´ Mus. Nat. Hist. ment of the tribe Fockeeae.T axon43:367– 376. Nat.27:153– 252. LEE,M.S.Y.1999.Reference taxaand phylogenetic PICHON,M.1948c.ClassiŽ cation des Apocynac´ees: V, nomenclature. Taxon48:31– 34. Cerberoid´´ ees. Notul. Syst.13:212– 229. LEEUWENBERG ,A.J.M.1991. Tabernaemontana : 1. The PICHON,M.1950.ClassiŽ cation des Apocynac´ees: XXV, Old World species. RoyalBotanical Garden,Kew , Echito¨´ õ dees´ . Mem.´ Mus. Natl. Hist. Nat. S er.´ B Bot. England. 1:1–143. LEEUWENBERG ,A.J.M.1994a.Taxa of the Apocynaceae POTGIETER, K., AND V. A. ALBERT.2001.Phylogenetic re- abovethe genuslevel. Series ofrevisions of Apocy- lationships within Apocynaceae s.l.based on trnL in- naceaeXXXVIII. W ageningenAgr. Univ .Pap.94:45– tron and trnL-Fspacer sequences andpropagule char- 60. acters. Ann.Mo. Bot. Gard. 88:523– 549. LEEUWENBERG ,A.J.M.1994b.A revision of Tabernae- REVEAL,J.L.1996.Solutions forbiological nomencla- montana:2.The New World species andStemmadenia. ture. In Biological nomenclature in the 21stcentury RoyalBotanic Gardens,Kew ,England. (J.L.Reveal). http://www.inform.umd.edu/PBIO/ LIDEN, M., AND B. OXELMAN.1996.Point ofview— Do nomcl/reve.html, Univ.ofMaryland. we need “phylogenetic taxonomy”? Zool. Scr.25:183– SAFWAT,F.M.1962.The  oralmorphology of Secamone 185. andthe evolution of the pollinating apparatusin As- LIDEN, M., B. OXELMAN, A. BACKLUND, L. ANDERS SON, clepiadaceae. Ann.Mo. Bot. Gard. 49:95– 129. B. BREMER, R. ERIKSSON, R. MOBERG, I. NORDAL, SAGHAI-MAROOF,M.A.,K.M.S OLIMAN, R. A. K. PERSSON, M. THULIN, AND B. ZIMMER.1997.Char- JORGENSEN, AND R. W. ALLARD.1984.Riboso- lie is our darling.T axon46:735– 738. malDNA spacer-length polymorphisms in barley: LIEDE, S. 1996. Cynanchum–Rhodostegiella– V incetoxicum– Mendelianinheritance, chromosomallocation, and Tylophora (Asclepiadaceae): New considerations onan population dynamics.Proc. Nat.Acad. Sci. USA old problem. Taxon45:193– 211. 81:8014–8018. LIEDE, S., AND F. ALBERS.1994.Tribal disposition ofgen- SANGER, F., S. NICKLEN, AND A. R. COULSON. 1977.DNA erain the Asclepiadaceae. Taxon43:201– 231. sequencing with chain-terminatinginhibitors. Proc. MISHLER,B.D.1999.Getting rid of species? Pages307– Nat.Acad. Sci. USA 74:5463–5467. 315. In Species: New interdisciplinary essays(R. A. SCHANDER,C.1998.T ypes,emendations and names— A Wilson, ed.).MIT Press, Cambridge,Massachusetts. reply to Liden et al.T axon47:401– 406. MOORE,G.1998.A comparison of traditional andphy- SCHANDER, C., AND M. THOLLESSON.1995.Phyloge- logenetic nomenclature. Taxon47:561– 579. netic taxonomy—some comments. Zool. Scr.24:263– OLMSTEAD, R. G., B. BREMER, K. M. SCOTT, AND J. D. 268. PALMER.1993.A parsimonyanalysis of the Asteridae SCHUMANN,K.M.1895.Apocynaceae. Pages 109– 189 in sensu lato basedon rbcLsequences. Ann.Mo. Bot. Die naturlichen¨ Panzenfamilien (H. G.A.Englerand Gard.80:700– 722. K.A.E.Prantl,eds.). Wilhelm Engelmann,Leipzig, OLMSTEAD,R.G.,C. W.D EPAMPHILIS , A. D. WOLFE, Germany. N. D. YOUNG, W. J. ELISONS , AND P. A. REEVES. 2001. SENNBLAD,B.1997.Phylogeny of the Apocynaceae s.l. Disintegration of the Scrophulariaceae.Am. J. Bot. Acta Univ.Upsaliensis 295:18pp. 88:348–361. SENNBLAD, B., AND B. BREMER.1996.The familial and OLMSTEAD,R.G.,H.J.M ICHAELS, K. M. SCOTT, AND J. subfamilial relationships ofApocynaceae andAscle- D. PALMER.1992.Monophyly of the Asteridae and piadaceaeevaluated with rbcLdata.Plant Syst.Evol. identiŽcation of their majorlineages inferred from 202:153–175. DNA sequences of rbcL.Ann.Mo. Bot. Gard. 79: SENNBLAD, B., AND B. BREMER.2000.Is there ajusti- 249–265. Žcation fordifferential apriori weighting in coding 2002 SENNBLADAND BREMER—CLASSIFICATIONOFAPOCYNACEAE 409

sequences? Acase studyfrom rbcLandApocynaceae. VENTER, H. J. T., AND R. L. VERHOEVEN.1997.A tribal Syst.Biol. 49:43–55. classiŽcation ofthe Periplocoideae (Apocynaceae). SENNBLAD, B., M. E. ENDRESS, AND B. BREMER. 1998. Taxon46:705– 720. Morphologyand molecular datain phylogenetic VERHOEVEN, R. L., S. LIEDE, AND M. ENDRESS . 2002. The fraternity—The tribe Wrightieae (Apocynaceae) revis- tribal position of Fockea and Cibirhiza (Apocynaceae: ited. Am.J. Bot.85:1143– 1158. Asclepiadoideae): Evidence frompollinium structure SERENO,P.C.1999.DeŽ nitions in phylogenetic taxon- andcpDNA sequence data.Grana (in press). omy:Critique andrationale. Syst. Biol. 48:329–351. ZARUCCHI,J.L.,G.N.M ORILLO, M. E. ENDRES S, B. F. STRUWE, L., V. A. ALBERT, AND B. BREMER.1994.Cladis- HANSEN, AND A. J. M. LEEUWENBERG .1995.Apoc- tics andfamily level classiŽcation of the Gentianales. ynaceae.Pages 471– 571 in Floraof the Venezuelan Cladistics 10:175–206. Guayana(J. A. Steyermark,P .A.Berry,andB. K. SWARUPANANDAN , K., J. K. MANGALY, T. K. SONNY, Holst,eds.). Missouri Botanical Garden/TimberPress, K. KISHOREKUMAR , AND S. CHAND BASHA. 1996. The St.Louis/ Portland,Oregon. subfamilial andtribal classiŽcation ofthe familyAs- clepiadaceae. Bot.J. Linn.Soc. 120:327– 369. SWOFFORD,D.L.1998.P AUP*. Phylogenetic analysisus- Firstsubmitted 27 October 2000;revision submitted ingparsimony (*and other methods),version 4.0b3a. 17December 2001;Ž nalaccepted 31 December 2001 SinauerAssociates, Sunderland,Massachusetts. Associate Editor:R. Olmstead