A new subfamilial and tribal classification of the pantropical flowering plant family Annonaceae informed by molecular phylogenetics


E-mail: lars.chatrou@wur.nl


The pantropical flowering plant family Annonaceae is the most species-rich family of Magnoliales. Despite long-standing interest in the systematics of Annonaceae, no authoritative classification has yet been published in the light of recent molecular phylogenetic analyses. Here, using the largest, most representative, molecular dataset compiled on Annonaceae to date, we present, for the first time, a robust family-wide phylogenetic tree and subsequent classification. We used a supermatrix of up to eight plastid markers sequenced from 193 ingroup and seven outgroup species. Some of the relationships at lower taxonomic levels are poorly resolved, but deeper nodes generally receive high support. Annonaceae comprises four major clades, which are here given the taxonomic rank of subfamily. The description of Annonoideae is amended, and three new subfamilies are described: Anaxagoreoideae, Ambavioideae and Malmeoideae. In Annonoideae, seven tribes are recognized, one of which, Duguetieae, is described as new. In Malmeoideae, seven tribes are recognized, six of which are newly described: Dendrokingstonieae, Fenerivieae, Maasieae, Malmeeae, Monocarpieae and Piptostigmateae. This new subfamilial and tribal classification is discussed against the background of previous classifications and characters to recognize subfamilies are reviewed. © 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2012, 169, 5–40.


Inventories of tropical forests on all continents invariably list Annonaceae as one of the most diverse plant families (Phillips & Miller, 2002). In terms of species richness and abundance of individuals, Annonaceae contributes significantly to the diversity of trees in Neotropical forests (e.g. Gentry, 1988; Valencia, Balslev & Paz Y Miño, 1994) and lianas and trees in rain forests of the Old World (e.g. van Gemerden et al., 2003; Slik et al., 2003; Tchouto et al., 2006). Around 2400 species in 108 genera are currently recognized in the family (Rainer & Chatrou, 2006), > 300 of which have been described in taxonomic papers, monographs and regional or continental floras since the start of the international Annonaceae project almost 30 years ago (Maas, 1983; Chatrou, 1999). In parallel with renewed taxonomic efforts, recent years have seen increasingly detailed studies of the phylogenetics of Annonaceae (e.g. Doyle & Le Thomas, 1994, 1996; Mols et al., 2004; Pirie et al., 2006; Couvreur et al., 2008; Erkens, Maas & Couvreur, 2009). The polyphyly of notorious ‘dustbin’ genera, such as Polyalthia Blume, has been demonstrated (Mols et al., 2004; Saunders, Su & Xue, 2011), and even easily recognizable genera in morphological terms have been shown to be nonmonophyletic (Chatrou, Koek-Noorman & Maas, 2000; Erkens et al., 2007; Chatrou et al., 2009; Couvreur et al., 2009). On the basis of these results, some generic circumscriptions have been realigned following the primary principle of monophyly (Chatrou et al., 2000; Su et al., 2005; Rainer, 2007; Erkens & Maas, 2008; Mols et al., 2008; Nakkuntod et al., 2009; Su, Chaowasku & Saunders, 2010; Surveswaran et al., 2010; Xue et al., 2011).

Given the large numbers of species in Annonaceae, a useful and stable infrafamilial classification is necessary to aid communication and information retrieval. Although a number of formal or informal classifications have been proposed (e.g. Baillon, 1868; Hutchinson, 1923; Fries, 1959; Walker, 1971; van Heusden, 1992; van Setten & Koek-Noorman, 1992), none has yet proved to be stable in the face of increasing knowledge of the diversity of the family. These classifications were based on different sources of data, such as floral morphology (van Heusden, 1992), fruit and seed morphology (van Setten & Koek-Noorman, 1992) and palynology (Walker, 1971). In each case, the data were interpreted intuitively, resulting in often contradictory conclusions/classifications. The classification of Fries (1959), primarily based on floral characters, remains perhaps the most widely used. However, with few exceptions, his tribes and informal groups of genera are neither characterized by unequivocal (combinations of) characters nor demonstrably monophyletic. To quote from a recent monographic work: ‘Systems of informal classifications [in Annonaceae] have proliferated to the point that classification of the family into smaller units is in disarray’ (Johnson & Murray, 1995: 249).

The phylogenetic reconstruction in Annonaceae was initiated with cladistic analyses of macromorphological and palynological characters (Doyle & Le Thomas, 1994, 1996, 1997). Although indicating the earliest diverging position of Anaxagorea A.St.-Hil., such characters showed high levels of homoplasy and limited phylogenetic utility compared with subsequent studies employing DNA sequence data (Doyle, Bygrave & Le Thomas, 2000; Mols et al., 2004; Richardson et al., 2004; Pirie et al., 2006; Couvreur et al., 2008, 2011). To date, a new formal classification based on molecular phylogenetic work has been postponed because of limitations in taxon representation and phylogenetic resolution. Clades are currently referred to by informal names relating to aspects of their molecular evolution [e.g. long branch clade (LBC) and short branch clade (SBC) sensuRichardson et al., 2004, in reference to the differing levels of genetic divergence between the two major clades identified in the early molecular phylogenetic studies]. These names are neither comprehensive in scope nor usefully memorable. Improvements in generic classification have thus yet to be matched by an improved higher level classification.

A robust and maximally representative hypothesis of relationships between clades in this important angiosperm family is clearly warranted. In this article, we present a phylogenetic analysis of Annonaceae inferred from multiple plastid DNA loci, representing 94 of the 108 currently recognized genera and marking an important improvement in both the representation of taxa (at the generic level) and phylogenetic resolution in Annonaceae compared with previous efforts (e.g. Richardson et al., 2004; Couvreur et al., 2011). We place this analysis in the context of previous infrafamilial classifications, evaluate the monophyly of the groupings identified and discuss the relative utility of various morphological characters for the diagnosis of groups in Annonaceae. The rank at which monophyletic groups might be classified remains a more or less subjective decision; we discuss the potential classifications that might be adopted given a number of secondary criteria, such as diagnosability and size of the groups. Finally, based on the results, we formally describe four subfamilies and 12 tribes. The subfamilies are Anaxagoreoideae (corresponding to the genus Anaxagorea), Ambavioideae (corresponding to the ambavioid clade of Doyle & Le Thomas, 1994, 1996), Annonoideae (corresponding to the LBC of Richardson et al., 2004, and the inaperturate clade of Doyle & Le Thomas, 1994, 1996) and Malmeoideae (corresponding to the SBC of Richardson et al., 2004, and the malmeoid/piptostigmoid/miliusoid clade of Doyle & Le Thomas, 1994, 1996). Seven tribes are recognized in Annonoideae, of which Duguetieae is described as new, and seven tribes are recognized in Malmeoideae, of which six are newly described, namely Dendrokingstonieae, Fenerivieae, Maasieae, Malmeeae, Monocarpieae and Piptostigmateae.


Taxon sampling

Most genera sampled were represented by two species. Exceptions were monotypic genera (with just single samples) and genera that have previously been demonstrated to be para- or polyphyletic, for which each segregate clade was represented by two samples. We selected taxa to bracket the crown node of each clade, if known. Of the 108 currently recognized genera (Rainer & Chatrou, 2006), 94 were represented, five of which were sampled for the first time, i.e. they were not represented in Richardson et al. (2004), who sampled 79 genera, Pirie et al. (2006), who increased sampling for the SBC, Couvreur et al. (2008), who increased sampling for the LBC, and Couvreur et al. (2011), who added another few unsampled genera (Table 1). Subsequent to analyses performed for this article, the genera Anomianthus Zoll., Balonga Le Thomas, Cyathostemma Griff., Dasoclema J.Sinclair, Ellipeia Hook.f. & Thomson, Ellipeiopsis R.E.Fr. and Rauwenhoffia Scheff. have been brought into synonymy with Uvaria L. (Zhou, Su & Saunders, 2009; Zhou et al., 2010). With the exception of Balonga and Dasoclema (which are not sampled), these former genera are thus still represented individually. A list of currently recognized genera of Annonaceae is presented in Table 2, with the numbers of recognized species, representation of species in previous phylogenetic studies (i.e. evidence for monophyly) and representation in this study.

Table 1.  Collections and GenBank numbers
Species CountryrbcLmatKndhFtrnTLtrnLFpsbA-trnHatpB-rbcLtrnSG
Persea americana Mill.UUBG 87GR00058Cultivated in UUBG, of Neotropical originAY841592JQ437545JQ742021AY841669JQ513882JQ513883
Coelocaryon preussii Warb.Wieringa, J.J. 3640 (WAG)GabonAY743437AY743475JQ437546AY743456AY841424
Degeneria vitiensis L.W.Bailey & A.C.Sm./D. roseiflora J.M.Mill.Mixed originOrigin unknownL12643AY220414 (intron)
AY220361 (spacer)
Eupomatia bennettii F.Muell.Chatrou, L.W. s.n. (U)Cultivated in UUBG, origin AustraliaDQ861790JQ437547DQ861842JQ513885
Galbulimima belgraveana (F.Muell.) SpragueMixed originOrigin unknownL12646AY220415 (intron)
AY220362 (spacer)
Liriodendron chinense SargentChatrou, L.W. 279 (U)Cultivated in UUBG, origin ChinaAY841593AY841670
Magnolia kobus DC.Chatrou, L.W. 278 (U)Cultivated in UUBG, origin JapanAY743438AY743457
Alphonsea boniana Finet & Gagnep.Keßler, P.J.A. 3116 (L)VietnamAY318965AY319077
Alphonsea elliptica Hook.f. & ThomsonVan Balgooy, M. 5141 (L)IndonesiaAY318966AY319078
Ambavia gerrardii (Baill.) Le ThomasRabevohitra, R. 2035 (MO)MadagascarJQ513886JQ513889
Anaxagorea phaeocarpa Mart.Maas, P.J.M. 8592 (U)EcuadorAY238952AY238960EF179279DQ861643AY231284 (intron)AY841426EF179244EF179321
AY238944 (spacer)
Anaxagorea silvatica R.E.Fr.Maas, P.J.M. 8836 (U)BrazilAY743439AY743477EF179280DQ861644AY743458AY841427AY578140EF179322
Annickia chlorantha (Oliv.) Setten & MaasSosef, M.S.M. 1877 (WAG)GabonAY841594AY841393AY841401AY841571AY841671AY841442AY841370AY841550
Annickia pilosa (Exell) Setten & MaasSosef, M.S.M. 1803 (WAG)GabonAY743450AY743488AY841402AY841572AY743469AY841444AY841371AY841551
Annona glabra L.Chatrou, L.W. 467 (U)Cultivated in UUBG, origin FloridaAY841596DQ125050EF179281JQ742022AY841673DQ125116EF179246EF179323
Annona herzogii (R.E.Fr.) H.RainerChatrou, L.W. 162 (U)PeruAY841656DQ125062EF179308JQ742023AY841734DQ125132EF179273EF179350
Annona muricata L.Chatrou, L.W. 468 (U)Cultivated in UUBG, of Neotropical originAY743440AY743478EF179282DQ861648AY743459AY841428EF179247EF179324
Anonidium sp.Cheek, M. 7896 (K)CameroonAY841598DQ125051EF179283JQ742024AY841675DQ125117EF179248EF179325
Artabotrys hexapetalus (L.f.) BhandariUUBG 94GR01614 (U)Cultivated in UUBG, origin IndiaAY238953AY238962EF179284DQ861649AY231286 (intron)AY841429EF179249EF179326
AY238946 (spacer)
Artabotrys sp.Wieringa, J.J. 4018 (WAG)GabonAY841599DQ125052EF179285JQ742025AY841676DQ125118EF179250EF179327
Asimina angustifolia A.GrayWeerasooriya, A. s.n. (U)USADQ124939DQ125053EF179286JQ742026AY841677DQ125119EF179251EF179328
Asimina rugelii B.L.Rob.J. R. Abbott 22361 (FLAS)USAJQ513887GQ139881
Asimina triloba (L.) DunalChatrou, L.W. 276 (U)Cultivated in UUBG, origin USAAY743441AY743479EF179287JQ742027AY743460AY841430EF179252EF179329
Asteranthe asterias (S. Moore) Engl. & DielsRobertson, A. 7548 (WAG)KenyaEU169757EU169757
Bocageopsis multiflora (Mart.) R.E.Fr.Jansen-Jacobs, M.J. 5789 (U)GuyanaAY841600AY841678
Bocageopsis pleiosperma MaasMiralha, J.M.S. 300 (U)BrazilAY841601AY841679
Cananga odorata (Lam.) Hook.f & ThomsonChatrou, L.W. 93 (U)Costa RicaAY841602AY841394AY841403AY841680AY841431AY841372AY841548
Cleistopholis glauca Pierre ex Engl. & DielsWieringa, J.J. 3278 (WAG)GabonAY841603AY841395AY841404AY841681AY841432AY841373AY841549
Cremastosperma brevipes (DC.) R.E.Fr.Scharf, U. 76 (U)French GuianaAY743527AY743550AY841405AY841573AY743573AY841447AY841374AY841552
Cremastosperma cauliflorum R.E.Fr.Chatrou, L.W. 224 (U)PeruAY743519AY743542AY841406AY841574AY743565AY841448AY841375AY841553
Cyathocalyx martabanicus Hook.f. & ThomsonMols, J.B. 11 (L)Cultivated in Kebun Raya Bogor, IndonesiaAY841605DQ125054EF179288JQ742028AY841683DQ125120EF179253EF179330
Cymbopetalum brasiliense (Vell.) Benth. ex Baill.UUBG 84GR00275Cultivated in UUBG, originating from BrazilAY841608DQ125055EF179289DQ861646AY841686DQ125121EF179254EF179331
Cymbopetalum torulosum G.E.SchatzChatrou, L.W. 54 (U)Costa RicaAY743442AY743461
Dasymaschalon macrocalyx Finet & Gagnep.Keßler, P.J.A. 3199 (L)ThailandAY841610EF179277EF179290JQ742029AY841688EF179313EF179255EF179332
Dasymaschalon sootepense CraibKeßler, P.J.A. 3201 (L)ThailandAY743443AY743462
Desmopsis microcarpa R.E.Fr.Chatrou, L.W. 85 (U)Costa RicaAY319059AY319173
Desmopsis schippii Standl.Chatrou, L.W. 94 (U)Costa RicaAY319060AY319174
Desmos chinensis Lour.C.-C. Pang N2 (HKU)Hong KongJQ762414JQ762415
Desmos elegans (Thwaites) Saff.Kostermans 24761 (L)Sri LankaHQ214067HQ214069
Dielsiothamnus divaricatus (Diels) R.E.Fr.Johnson, D.M. 1903 (OWU)TanzaniaEU169781EU169759
Disepalum pulchrum (King) J.SinclairChan, R. 192 (FLAS)MalaysiaJQ513888GQ139909
Disepalum platipetalum· Merr.Takeuchi 18201 (L)IndonesiaAY841612AY841690
Drepananthus biovulatus (Boerl.) Survesw. & R.M.K.SaundersWong 46009 (L)IndonesiaHM173779HM173751
Duguetia hadrantha (Diels) R.E.Fr.Chatrou, L.W. 181 (U)PeruAY738161AY740541EF179293DQ861650AY740573DQ125123EF179258EF179335
Duguetia staudtii (Engl. & Diels) Chatrouvan Andel, T.R. 3290 (U)CameroonAY738178AY740558EF179294JQ742030AY740590DQ125124EF179259EF179336
Enicosanthum membranifolium J.SinclairKeßler, P.J.A. 3198 (L)ThailandAY318974AY319086
Enicosanthum paradoxum Becc.Keßler, P.J.A. 2746 (L)IndonesiaAY318975AY319087
Ephedranthus boliviensis Chatrou & PirieChatrou, L.W. 301 (U)BoliviaAY841614AY841692
Ephedranthus sp.Maas, P.J.M. 8826 (U)BrazilAY841616AY841396AY841407AY841575AY841694AY841463AY841376AY841554
Fissistigma glaucescens (Hance) Merr.Law, C.L 00/07b (L)Hong KongAY743444AY743463AY743444
Fissistigma uonicum (Dunn) Merr.Law, C.L 00/05 (L)Hong KongAY841617AY841695
Fitzalania heteropetala (F.Muell.) F.Muell.Forster, P.I.F. 8326 (K)AustraliaAY318977AY319089AY318977
Friesodielsia desmoides (Craib) SteenisKeßler, P.J.A. 3189 (L)ThailandAY841618AY841696
Friesodielsia sp.Wieringa, J.J. 3605 (WAG)GabonAY841619AY841697
Fusaea longifolia (Aubl.) Saff.Chatrou, L.W. 175 (U)PeruAY841620AY841698
Fusaea peruviana R.E.Fr.Chatrou, L.W. 179 (U)PeruAY743445AY743483EF179295DQ861652AY743464AY841436EF179260EF179337
Goniothalamus griffithii Hook.f. & ThomsonKeßler, P.J.A. 3188 (L)ThailandAY743446AY743484EF179296JQ742031AY743465DQ125125EF179261EF179338
Goniothalamus tapis Miq.Keßler, P.J.A. 3193 (L)ThailandAY841622DQ125058EF179297JQ742032AY841700DQ125126EF179262EF179339
Greenwayodendron oliveri (Engl.) Verdc.Jongkind, C.C.H. 1795 (WAG)GhanaAY743451AY743489AY841408AY841576AY743470AY841465AY841377AY841555
Greenwayodendron suaveolens (Engl. & Diels) Verdc.Semsei 2376 (K)KenyaAY841524AY841538
Guatteria anomala R.E.Fr.Ishiki, M. 2233 (U)MexicoAY740962AY740913EF179298DQ861657AY741011AY841437EF179263EF179340
Guatteria pudica N.Zamora & MaasChatrou, L.W. 107 (U)Costa RicaAY740994AY740945JQ769093DQ861663AY741043DQ125197JQ513884FJ842397
Haplostichanthus longirostris (Scheff.) HeusdenTakeuchi 15656 (L)Papua New GuineaAY318979AY319091
Hexalobus crispiflorus A. Rich.Sosef, M.S.M. 2287 (WAG)GabonEU169782EU169760
Hexalobus salicifolius Engl.Sosef, M.S.M. 2376 (WAG)GabonEU169783EU169761
Hornschuchia citriodora D.M.JohnsonMaas, P.J.M. 8828 (U)BrazilAY841625AY841703
Isolona campanulata Engl. & DielsUUBG 86GR00240UUBG, of tropical African originAY238954AY238963EU169715JQ742033AY231287 (intron)DQ125127EF179266EU169806
AY238947 (spacer)
Isolona cooperi Hutch. & Dalziel ex G.P.Cooper & RecordUUBG 84GR00382UUBG, originating from Ivory CoastAY841626AY841704
Klarobelia inundata ChatrouChatrou, L.W. 205 (U)PeruAY743452AY743490AY841409AY841577AY743471AY841469AY841378AY841556
Klarobelia stipitata ChatrouChatrou, L.W. 113 (U)Costa RicaAY841628AY841706
Letestudoxa bella Pellegr.Wieringa, J.J. 2797 (WAG)GabonAY841629DQ125059EF179302DQ861653.AY841707DQ125128EF179267EF179344
Letestudoxa glabrifolia Chatrou & RepeturBreteler, F.J. 12858 (WAG)GabonAY841630AY841708
Lettowianthus stellatus DielsRobertson, A. 7505 (WAG)KenyaEU169775EU169753
Maasia discolor (Diels) Mols, Keßler & RogstadTakeuchi & Ama 16394 (L)Papua New GuineaAY319021AY518872AY841416AY841584AY319135AY841500AY841385AY841563
Maasia glauca (Hassk.) Mols, Keßler & RogstadMols, J.B. 20 (L)IndonesiaAY319023AY319137
Maasia sumatrana (Miq.) Mols, Keßler & RogstadSAN 143918 (SAN)MalaysiaAY319039AY518873AY841418AY841586AY319153AY841503AY841387AY841565
Malmea dielsiana R.E.Fr.Chatrou, L.W. 122 (U)PeruAY238955AY238964AY841410AY841578AY231288 (intron)AY841473AY841379AY841557
AY238948 (spacer)
Malmea sp.Chatrou, L.W. 8 (U)PeruAY841527AY841397AY841411AY841579AY841541AY841475AY841380AY841558
Marsypopetalum littorale (Blume) B.Xue & R.M.K.SaundersRastini 153 (L)IndonesiaAY319026AY319140   
Marsypopetalum pallidum (Blume) KurzKeßler, P.J.A. 3192 (L)ThailandAY318980AY319092
Meiocarpidium lepidotum (Oliv.) Engl. & DielsBreteler, F. 13947 (WAG)GabonEU169776EU169687EU169754EU169731EU169798
Meiogyne cylindrocarpa (Burck) HeusdenRidsdale, C.E. DV-M1-1930 (L)MalaysiaAY318981AY319093
Meiogyne sp.Rainer, H. 1593 (WU)MexicoAY841623AY841701
Meiogyne stenopetala (F.Muell.) HeusdenJessup, L.W. 706 (K)AustraliaAY318971AY319083
Meiogyne virgata (Blume) Miq.Keßler, P.J.A. 2751 (L)IndonesiaAY318982AY319094
Mezzettia parviflora Becc.Okada 3388 (L)IndonesiaAY318983AY319095
Miliusa horsfieldii (Benn.) PierreMols, J.B. 1 (L)IndonesiaAY318986AY319098
Miliusa mollis PierrePholsena 1756 (L)ThailandAY318989AY319101
Mischogyne michelioides ExellBamps, P. 4459 (WAG)AngolaEU169786EU169764
Mitrella kentii (Blume) Miq.Gardette, E. 2239 (K)MalaysiaAY841633AY841711
Mitrephora polypyrena (Blume) Miq.Mols, J.B. 7 (L)IndonesiaAY318997AY319110
Mitrephora teysmannii Scheff.Keßler, P.J.A. 3226 (L)ThailandAY318996AY319109
Mkilua fragrans Verdc.Chatrou, L.W. 474 (U)Cultivated in UUBG, origin KenyaAY841634DQ125060EF179303DQ861647AY841712DQ861696EF179268EF179345
Monanthotaxis whytei (Stapf) Verdc.UUBG 84GR00388Cultivated in UUBG, origin NigeriaAY841635EF179278EF179304JQ742034AY841713EF179315EF179269EF179346
Monanthotaxis sp.Wieringa, J.J. 3833 (WAG)GabonAY841636AY841713
Monocarpia euneura Miq.Slik, J.W.F. 2002–2931 (L)IndonesiaAY318998AY518865AY841412AY841580AY319111AY841477AY841381AY841559
Monocyclanthus vegnei KeayJongkind, C.C.H. 6992 (WAG)LiberiaEU169765EU169787
Monodora crispata Engl.UUBG E64GR00066Cultivated in UUBG, origin Ivory CoastAY841637AY841715
Monodora myristica (Gaertn.) DunalUUBG E84GR00389Cultivated in UUBG, origin Ivory CoastAY743447EU169700EU169721JQ742035AY743466DQ125129EF179270EU169812
Mosannona costaricensis (R.E.Fr.) ChatrouChatrou, L.W. 90 (U)Costa RicaAY743510AY743503AY841413AY841581AY743496AY841479AY841382AY841560
Mosannona papillosa ChatrouPitman, N. s.n. (U)EcuadorAY743514AY743500
Mwasumbia alba Couvreur & D.M.JohnsonCouvreur, T.L.P. 85 (WAG)TanzaniaEU747680EU747674
Neostenanthera myristicifolia (Oliv.) ExellWieringa, J.J. 3566 (WAG)GabonAY743448AY743486EF179306JQ742036AY743467DQ125130EF179271EF179348
Neo-uvaria acuminatissima (Miq.) Airy ShawRidsdale, C.E. DV-SR-4671 (L)MalaysiaAY318999AY319112
Neo-uvaria parallelivenia (Boerl.) H.Okada & K.UedaKeßler, P.J.A. sub IV-H-73 (L)IndonesiaAY319000AY319113
Onychopetalum periquino (Rusby) D.M.Johnson & N.A.MurrayChatrou, L.W. 425 (U)BoliviaAY319065AY518876AY841414AY841582AY319179AY841485AY841383AY841561
Ophrypetalum odoratum DielsRobertson, A. 7547 (WAG)KenyaEU169789EU169767
Orophea celebica (Blume) Miq.Keßler, P.J.A. 2953 (L)IndonesiaAY319004AY319117
Orophea creaghii (Ridl.) Leonardía & KeßlerKeßler, P.J.A. 1605 (L)IndonesiaAY841632AY841710
Orophea enterocarpa Maingay ex Hook.f. & ThomsonChalermglin 440403 (TISTR)ThailandAY319006AY319119
Orophea kerrii KeßlerChalermglin 440416-1 (TISTR)ThailandAY319008AY319121
Orophea polycarpa A.DC.Keßler, P.J.A. 3234 (L)ThailandAY319010AY319123
Oxandra asbeckii (Pulle) R.E.Fr.University of Guyana, course Neotrop. Botany UG-NB-55 (U)GuyanaAY841639AY841717
Oxandra longipetala R.E.Fr.Chatrou, L.W. 114 (U)Costa RicaAY841641AY841719
Oxandra macrophylla R.E.Fr.Chatrou, L.W. 204 (U)PeruAY841642AY841720
Oxandra polyantha R.E.Fr.Chatrou, L.W. 215 (U)PeruAY841643AY841721
Oxandra venezuelana R.E.Fr.Chatrou, L.W. 120 (U)Costa RicaAY841645AY841723
Oxandra xylopioides DielsChatrou, L.W. 165 (U)PeruAY841646AY841724
Phaeanthus ebracteolatus (C.Presl.) Merr.Utteridge, T. 17 (KL)Papua New GuineaAY319012AY319125  
Piptostigma mortehani De Wild.Wieringa, J.J. 2779 (WAG)GabonAY743454AY743492AY841415AY841583AY743473AY841498AY841384AY841562
Piptostigma pilosum Oliv.Wieringa, J.J. 2030 (WAG)CameroonAY841648AY841726
Platymitra macrocarpa Boerl.Okada 3457 (L)IndonesiaAY319013AY319127
Polyalthia borneensis Merr.Ridsdale, C.E. DV-SR-7921 (L)MalaysiaAY319014AY319128
Polyalthia cauliflora Hook.f. & ThomsonKeßler, P.J.A. 3114 (L)SingaporeAY319015AY319129
Polyalthia celebica Miq.Mols, J.B. 9 (L)IndonesiaAY319016AY319130
Polyalthia cerasoides (Roxb.) Benth. & Hook.f. ex BeddomeChalermglin 440214-4 (L)ThailandAY319017AY319131
Polyalthia cinnamomea Hook.f. & ThomsonRidsdale, C.E. DV-M1-347 (L)MalaysiaAY319018AY319132
Polyalthia congesta (Ridl.) J.SinclairRidsdale, C.E. DV-S-5105 (L)MalaysiaAY319019AY319133
Polyalthia debilis (Pierre) Finet & Gagnep.Keßler, P.J.A. 3228 (L)ThailandAY319020AY319134
Polyalthia flagellaris (Becc.) Airy ShawDuling 38 (K)BruneiAY319022AY319136
Polyalthia cf. glabra (Hook.f. & Thomson) J.SinclairRastini 224 (L)IndonesiaAY319032AY319146
Polyalthia lateriflora (Blume) KingHort. Bot. Bog. XII-B-VII-37 (L)IndonesiaAY319024AY319138
Polyalthia longifolia (Sonn.) ThwaitesJohnson, D.M. 1965 (OWU)TanzaniaAY319027AY319141
Polyalthia cf. longifolia (Sonn.) ThwaitesMols, J.B. 14 (L)IndonesiaAY319025AY319139
Polyalthia longipes (Miq.) Koord. & Valet.Ridsdale, C.E. DV-M2-11443 (L)MalaysiaAY319028AY319142
Polyalthia obliqua Hook.f. & ThomsonAmbriansyah 1694 (L)IndonesiaAY319029AY319143
Polyalthia pendula Capuron ex G.E.Schatz & Le ThomasRabevohitra 2386 (K)MadagascarAY319030AY319144
Polyalthia rumphii (Blume ex Hensch.) Merr.Van Balgooy, M. 5654 (L)IndonesiaAY319031AY319145
Polyalthia sclerophylla Hook.f. & ThomsonHort. Bot. Bog. XX-D-82 (L)IndonesiaAY319033AY319147
Polyalthia stenopetala (Hook.f. & Thomson) Finet & Gagnep.Chalermglin 440302 (TISTR)ThailandAY319034AY319148
Polyalthia stuhlmannii (Engl.) Verdc.Luke 1424 (K)KenyaAY319035AY319149
Polyalthia subcordata Blume (Blume)Gravendeel, B. 678 (L)IndonesiaAY319037AY319151
Polyalthia suberosa (Roxb.) ThwaitesUUBG 83GR00317Cultivated in UUBG, origin IndiaAY238956AY238965AY841417AY841585AY231289 (intron)AY841502AY841386AY841564
AY238949 (spacer)
Polyalthia viridis CraibChalermglin 440214-3 (L)ThailandAY319040AY319154
Polyalthia xanthopetala Merr.Ridsdale, C.E. DV-S-5107 (L)MalaysiaAY319041AY319155
Polyceratocarpus microtrichus (Engl. & Diels) Ghesq. ex Pellegr.Bos, J.J. 6684 (WAG)CameroonEU747683EU747677
Polyceratocarpus pellegrinii Le Thomasde Wilde J.J.E. 8718 (WAG)CameroonEU747684EU747678
Popowia odoardi DielsRidsdale, C.E. DV-SR-7422 (L)MalaysiaAY319043AY319157
Popowia pisocarpa (Blume) Endl.Van Balgooy, M. 5683 (L)IndonesiaAY319044AY319158
Porcelia steinbachii (Diels) R.E.Fr.UUBG 99GR00210Cultivated in UUBG, origin BoliviaAY841649AY841727
Pseudartabotrys letestui Pellegr.Wieringa, J.J. 3273 (WAG)GabonAY841650DQ125061EF179307AY841728DQ125131EF179272EF179349
Pseudephedranthus fragrans (R.E.Fr.) Aristeg.Maas, P.J.M. 6878 (U)VenezuelaAY841651AY841729
Pseudomalmea diclina (R.E.Fr.) ChatrouChatrou, L.W. 211 (U)PeruAY319068AY841398AY841419AY841587AY319128AY841506AY841388AY841566
Pseudomalmea sp.Idarraga, A. 13 (U)ColombiaAY841652AY841730
Pseudoxandra polyphleba (Diels) R.E.Fr.Maas, P.J.M. 8227 (U)PeruAY841654JQ769091JQ769092JQ742037AY841732AY841512
Pseudoxandra spiritus-sancti MaasMaas, P.J.M. 8833 (U)BrazilAY841533AY841399AY841421AY841589AY841547AY841513AY841390AY841568
Pseuduvaria megalopus (K.Schum.) Y.C.F.Su & MolsTakeuchi 15599 (L)Papua New GuineaAY319011AY319124AY319011
Pseuduvaria pamattonis (Miq.) Y.C.F.Su & R.M.K.SaundersSlik, J.W.F. 2002–2911 (L)IndonesiaAY319049AY319163AY319049
Pseuduvaria phuyensis (R.M.K.Saunders, Y.C.F.Su & Chalermglin) Y.C.F.Su & R.M.K.SaundersKeßler, P.J.A. 3221 (L)ThailandAY319001AY319114AY319001
Pseuduvaria rugosa (Blume) Merr.Keßler, P.J.A. .3209 (L)ThailandAY319048AY319162AY319048
Ruizodendron ovale (Ruiz & Pav.) R.E.Fr.Maas, P.J.M. 8600 (U)EcuadorAY841657HQ214070AY841735AY841514
Sageraea lanceolata Miq.Ridsdale, C.E. DV-M2-1692 (L)MalaysiaAY319050AY319164   
Sanrafaelia ruffonammari Verdc.Kayombo 3027 (MO)TanzaniaEU169790EU169768
Sapranthus microcarpus (Donn.Sm.) R.E.Fr.Maas, P.J.M. 8457 (U)HondurasAY319052AY319166
Sapranthus viridiflorus G.E.SchatzChatrou, L.W. 55 (U)Costa RicaAY319051AY743493AY841422AY841590AY319165AY841515AY841391AY841569
Sphaerocoryne gracilis (Oliv. ex Engl. & Diels) Verdc.Robertson, A. 7554 (WAG)KenyaEU169755EU169777
Sphaerocoryne sp.Chalermglin 440214-2 (L)ThailandAY319071AY319185
Stelechocarpus burahol (Blume) Hook.f. & ThomsonMols, J.B. 13 (L)IndonesiaAY319053AY319167
Stelechocarpus cauliflorus (Scheff.) J.SinclairHort. Bot. Bog. XV-A-196 (L)IndonesiaAY319054AY319168
Stenanona costaricensis R.E.Fr.Chatrou, L.W. 67 (U)Costa RicaAY319069AY319183
Stenanona panamensis Standl.Chatrou, L.W. 100 (U)Costa RicaAY319070AY319184
Tetrameranthus duckei R.E.Fr.Stevenson, D.W. 1002 (U)BrazilAY841658AY841736
Tetrameranthus laomae D.R.SimpsonPipoly, J. 13407 (U)PeruAY841659AY841737
Toussaintia orientalis Verdc.Johnson, D.M. (OWU)TanzaniaEU169778EU169756
Tridimeris sp.Schatz, G.E. 1198 (K)MexicoAY319055AY319169
Trigynaea duckei (R.E.Fr.) R.E.Fr.Chatrou, L.W. 129 (U)PeruAY841660AY841738
Trigynaea lanceipetala D.M.Johnson & N.A.MurrayChatrou, L.W. 234 (U)PeruAY743449AY743487EF179309JQ742038AY743468EF179274EF179351
Trivalvaria macrophylla (Blume) Miq.Chase, M.W. 1207 (K)IndonesiaAY319056AY319170
Unonopsis pittieri Saff.Chatrou, L.W. 68 (U)Costa RicaAY841661AY841739
Unonopsis stipitata DielsChatrou, L.W. 253 (U)PeruAY841662AY841400AY841423AY841591AY841740AY841519AY841392AY841570
Uvaria chamae P.Beauv.Chatrou, L.W. 482 (U)Cultivated in UUBG, origin TogoAY841663AY841741
Uvaria cherrevensis (Pierre ex Finet & Gagnep.) L.L.Zhou, Y.C.F. Su & R.M.K. SaundersMaxwell 90–625 (L)ThailandFJ743823FJ743858
Uvaria clementis (Merr.) Attanayake, I.M.Turner & R.M.K.SaundersKeßler, P.J.A. 3211 (L)ThailandAY841606FJ743853
Uvaria cuneifolia (Hook.f. & Thomson) L.L.Zhou, Y.C.F.Su & R.M.K. SaundersMohtar S48169 (L)IndonesiaFJ743822FJ743857
Uvaria dulcis DunalMaxwell, J.F. 88–509 (L)ThailandFJ743815FJ743849
Uvaria grandiflora Roxb. ex Hornem.Saunders 05/1 (HKU)ThailandFJ743836FJ743870
Uvaria griffithii L.L.Zhou, Y.C.F.Su & R.M.K.SaundersChalermglin 440402-2 (TISTR)ThailandFJ743820FJ743855
Uvaria lucida Benth. subsp. virens (N.E.Br.) Verdc.UUBG 84GR00334Cultivated in UUBG, origin West AfricanAY238957AY238966EF179310JQ742039AY231290 (intron)AY841440EF179275EF179352
AY238950 (spacer)
Uvaria siamensis (Scheff.) L.L.Zhou, Y.C.F.Su & R.M.K.SaundersSaunders 07/3 (HKU)Cultivated in Hong Kong Botanic GardensFJ743824FJ743859
Uvariastrum insculptum (Engl. & Diels) Sprague & Hutch.Jongkind, C.C.H. 4707 (WAG)Ivory CoastEU169791EU169769
Uvariastrum pynaertii De Wild.Wieringa, J.J. 2620 (WAG)GabonEU169792EU169770
Uvariodendron kirkii Verdc.Robertson, A. 7550 (WAG)KenyaEU169793EU169771
Uvariodendron molundense (Diels) R.E.Fr.Sosef, M.S.M. 2219 (WAG)GabonEU169794EU169772
Uvariopsis korupensis Gereau & KenfackRichardson, J.E. 212 (WAG)GabonEU169796EU169774
Uvariopsis vanderystii Robyns & Ghesq.Sosef, M.S.M. 2241 (WAG)GabonEU169773EU169795
Uvariopsis tripetala (Baker.f.) G.E.SchatzJongkind, C.C.H. 4356 (WAG)Ivory CoastEU169780EU169758
Woodiellantha sp.Lugas 311 (K)MalaysiaAY841665AY841743
Xylopia ferruginea (Hook.f. & Thomson) Hook.f. & ThomsonSlik, J.W.F. 2002-S 558 (L)IndonesiaAY841666DQ125063EF179311JQ742040AY841744DQ125133
Xylopia hypolampra Mildbr. & DielsWieringa, J.J. 3748 (WAG)GabonAY841668AY841746
Xylopia peruviana R.E.Fr.Chatrou, L.W. 483 (U)Cultivated in UUBG, origin PeruAY238958AY238967EF179312DQ861654AY231291 (intron)DQ125134EF179276EF179353
AY238951 (spacer)
Table 2.  Currently recognized genera of Annonaceae, number of species and number of species sampled in phylogenetic analyses, demonstrating (lack of) monophyly. Unless indicated otherwise, studies that demonstrate monophyly are given, including the sampling size. A dash indicates either a lack of presence altogether in any phylogenetic study or the presence of a single species only. Genera followed by an asterisk were not included in the phylogenetic analyses presented in this paper
GenusNo. of recognized speciesEvidence for monophyly: no. of species sampled and reference
Desmopsis142g (unresolved)
Enicosanthum184g (unresolved)
Klarobelia126j, 2a
Meiogyne154a (paraphyletic)
Neo-uvaria52a, g
Oxandra2811d (polyphyletic)
Phaeanthus92a, g
Polyalthia13526l (polyphyletic)
Trivalvaria42a, t

Character sampling

We used previously published plus unpublished sequence data from up to eight plastid loci: protein coding rbcL, matK and ndhF genes plus an intron, trnL, and spacer regions trnT-L, trnL-F, trnS-G, atpB-rbcL and psbA-trnH. Total genomic DNA was extracted following a protocol adapted from the cetyltrimethylammonium bromide (CTAB) method (Doyle & Doyle, 1987), as described in Erkens et al. (2008). Conditions for the polymerase chain reactions (PCRs) and primers for the plastid markers were standard, and are identical to Pirie et al. (2006) and Erkens et al. (2008). PCR products were purified using QIAquick PCR purification kits (Qiagen) and sequenced with the PCR primers.

The relative importance for the phylogenetic accuracy of sampling either characters or taxa has been discussed extensively (Graybeal, 1998; Mitchell, Mitter & Regier, 2000; Cummings & Meyer, 2005; Rokas & Carroll, 2005). We adopted a sampling strategy that addressed both issues at once, specifically by following a supermatrix approach in which missing data are tolerated (Philippe et al., 2004; Wiens, 2005, 2006; Pirie et al., 2008). In this way, we focused sequencing effort on the resolution of relationships between the major clades of Annonaceae, which is of particular relevance to classification in the family. For all 200 taxa, rbcL, the trnL intron and trnL-F spacer were sampled. After phylogenetic analyses of these three markers (results not shown), 56 species were selected, paying particular attention to the inclusion of early diverging species in clades at all levels. These 56 species were selected as placeholders to be sampled for additional characters using the remaining six markers. All data (both taxa and characters) were subsequently combined in a single supermatrix, i.e. a data matrix including incompletely sampled taxa.

DNA sequences were aligned manually using PAUP* version 4.10b (Swofford, 2000) and MacClade (Maddison & Maddison, 2000) following the guidelines in Kelchner (2000). Characters in regions for which alignment was ambiguous were excluded from the analyses. Microsatellites were also excluded, as these regions are variable within species (Kelchner & Clark, 1997; Provan, Powell & Hollingsworth, 2001; personal observations on species for which the same spacer region from different accessions was sequenced). Gaps in the alignment shared by two or more taxa were coded as a single binary character (presence/absence) according to the simple indel coding method of Simmons & Ochoterena (2000). Single-nucleotide indels where verified once more against the tracer files to ensure that they were not sequence editing artefacts. Nucleotide characters included in these indels were excluded from the analyses, with a few exceptions when insertions in clades contained parsimony informative variation at the nucleotide level. Two short sequences, of 15 positions in psbA-trnH and of 12 positions in the trnT-L spacer, appeared to represent inversions. Around half of the species exhibited the reverse-complement sequence of the other half and transitions between the motifs appeared to be frequent, with different motifs apparent in closely related species (as reported in Pirie et al., 2006). We aligned one motif with the reverse complement of the other and, as the informative base changes that were revealed displayed little or no homoplasy, we assumed them to be effectively homologous and included them in the analyses (following Pirie et al., 2006).

Phylogenetic analyses

As plastid DNA is inherited as a unit, individual markers were not analysed separately to look for incongruence; we excluded a paralogous second copy of the trnL-F region from these analyses (Pirie et al., 2007). For the combined analyses, a supermatrix approach was adopted, i.e. including all taxa, even where data were not available for particular markers, which were coded as missing.

Parsimony analysis

Analyses were performed using PAUP* version 4.10b (Swofford, 2000) with the heuristic search option, tree bisection–reconnection (TBR) branch swapping, the accelerated transformation (ACCTRAN) criterion and the multiple parsimonious trees (MULPARS) option invoked. Character states were specified as unordered and equally weighted (Fitch parsimony; Fitch, 1971). Alignment gaps were treated as missing data, but larger indels were coded as above. The search strategy consisted of 10 000 replicates of random addition sequence, saving 25 trees per replicate. To ensure that the tree island with the globally shortest tree had been visited, we performed a parsimony ratchet (Nixon, 1999) search as implemented in PAUPRat (Sikes & Lewis, 2001), with 1000 ratchet iterations, perturbing 25% of the characters in each round. The robustness of the phylogenetic relationships was assessed by nonparametric bootstrapping of the data. Following Müller (2005), the number of bootstrap replicates was set at a high level (50 000), whereas the thoroughness of searches and computing time per bootstrap replicate were minimized by limiting the number of random addition sequence replicates to one, saving a single tree. When evaluating the results, we used the following descriptions of support by bootstrap values: 50–74% represents weak support, 75–84% moderate support and 85–100% strong support.

Maximum likelihood (ML) analysis

In recent years, ML algorithms have become more efficient, allowing for fast and accurate estimation of ML trees and even bootstrapping, which is especially useful for large datasets (Guindon & Gascuel, 2003; Stamatakis, 2006; Zwickl, 2006; Morrison, 2007). For this study, we used the RAxML web-server program available at the CIPRES portal in San Diego, CA, USA (http://www.phylo.org/portal2), which implements an efficient and rapid heuristic bootstrap in RAxML (Stamatakis, Hoover & Rougemont, 2008). For each analysis, the ‘maximum likelihood search’ and ‘estimate proportion of invariable sites’ boxes were selected, with a total of 1000 bootstrap replicates performed. The dataset was not partitioned, as the number of missing data per marker (excluding trnL-F and rbcL) resulted in aberrant results. Indel characters were necessarily excluded from the analyses, resulting in a total of 7657 included characters.

Bayesian inference

Bayesian analysis was performed on the combined dataset using MrBayes 3.2 (Huelsenbeck et al., 2001; Ronquist & Huelsenbeck, 2003). We used the 56 completely sampled species to identify the best partitioning of the data, employing the Bayes factor criterion (2ln Bayes factor > 10; Kass & Raftery, 1995; Sinsheimer, Lake & Little, 1996) following Brandley, Schmitz & Reeder (2005). Bayes factors were calculated as the ratio of the harmonic means of each partitioning strategy, which are produced by MrBayes in the output from the sump command. We tested the following partitioning strategies: 1, combined protein coding (‘coding’) regions/combined intron and spacer (‘noncoding’) regions/combined binary coded indel characters (‘indels’) (three partitions); 2, codon positions for the three coding regions (rbcL, matK, ndhF) separately/noncoding/indels (11 partitions); 3, nucleotide and indel characters for each marker separately (14 partitions). Values of the 2ln Bayes factor were all between 0 and 1, showing no preference for any of the partitioning strategies. This being the case, the data were partitioned according to strategy 1. This relatively simple strategy represents an attempt to best reflect differences between markers whilst maximizing the proportion of topology relative to substitution parameter change proposals in the Markov Chain Monte Carlo (MCMC) chains. Both rates and substitution models were allowed to vary across partitions. Priors for the number of parameters in the DNA substitution models were applied to each partition [as determined using ModelTest 3.06 (Posada & Crandall, 1998), with the topology in each case derived from a randomly selected most parsimonious tree]. In each case, this corresponded to models with NST = 6, gamma distributed rates and proportion of invariable sites. Runs were set to continue indefinitely, and the outputs were tested periodically for convergence through both visual inspection of cumulative clade posterior probabilities (PPs) (using AWTY; Nylander et al., 2008) and according to effective sample sizes (ESS) calculated using Tracer 1.4 (Rambaut & Drummond, 2007).


A small number of our sequencing attempts were unsuccessful, e.g. in the case of trnT-L for Cananga odorata (Lam.) Hook.f. & Thomson, Meiocarpidium lepidotum (Oliv.) Engl. & Diels and Cleistopholis glauca Pierre ex Engl. & Diels, psbA-trnH and trnS-G for Pseudoxandra polyphleba (Diels) R.E.Fr. and Ruizodendron ovale (Ruiz & Pav.) R.E.Fr. and ndhF for Meiocarpidium lepidotum. The database contained a total of 7787 characters after alignment and exclusion of ambiguous regions of the alignment. Table 3 shows the number of positions in the aligned data matrix and the number of indel characters per marker.

Table 3.  Characteristics of individual markers, which have been assessed using all data available for each locus, i.e. 193 sequences for rbcL and trnL-F and 59–61 sequences for the remaining loci. Outgroup taxa were excluded from the calculations. Consistency index (CI) includes all (i.e. variable and invariant) nucleotide characters
MarkerNumber of characters in aligned matrixNumber of potentially parsimony informative characters (%)Number of indel charactersCIRIModel selected by MrModeltest
  1. RI, retention index.

rbcL1376284 (20.6)00.400.80GTR + I + G
matK831247 (29.7)10.610.70GTR + G
ndhF1956715 (36.6)40.500.70GTR + I + G
atpB-rbcL747206 (27.6)270.700.82GTR + G
trnT-L673225 (33.4)00.680.73GTR + G
trnL intron520197 (37.9)230.600.83GTR + G
trnL-F377214 (56.8)320.560.84GTR + G
psbA-trnH433209 (48.3)240.540.69GTR + G
trnS-G744241 (32.4)190.700.81GTR + G

The heuristic search resulted in 20 960 most parsimonious trees with a tree length of 9806 steps, an overall consistency index (Kluge & Farris, 1969) of 0.55 and an overall retention index (RI; Farris, 1989) of 0.77. The ratchet search did not find shorter trees. The total number of potentially parsimony informative characters was 2729 (35.0%). The greatest number of parsimony informative characters was for ndhF (715) and then atpb-rbcL (206); psbA-trnH (209) had the least (Table 3). The final ML optimization likelihood was −67 480.27. The two MrBayes runs were terminated after ten million generations, having reached the same likelihood plateau after c. 350 000 generations, which were discarded as burn-in. Analysis of the tree output using AWTY (Nylander et al., 2008) showed that the clade PPs of the two runs were consistent with each other, and clade PPs of each run had reached values that no longer changed with additional generations. The effective sampling size of all parameters of the combined output, minus burn-in, as estimated using Tracer (Rambaut & Drummond, 2007), exceeded 200.

Phylogenetic relationships

The monophyly of Annonaceae (clade A; Fig. 1A) and the sister group relationship between Anaxagorea and the remaining Annonaceae (clade B; Fig. 1A) are maximally supported in all three analyses. Clade C (Fig. 1) is weakly supported in the parsimony analyses [parsimony bootstrap percentage (PBP), 66], mainly because of a degree of uncertainty in the position of Meiocarpidium Engl. & Diels. Both parametric analyses, however, assign high support [ML bootstrap percentage (MBP), 97; Bayesian PP, 0.95] to this node. The rest of the strongly supported relationships in clade C are the same among all three analyses.

Figure 1.

Figure 1.

Phylogram showing one of many most parsimonious tree topologies, with support [maximum parsimony bootstrap (BS) percentages, maximum likelihood BS percentages and Bayesian posterior probabilities (PP) indicated in the key]. A, Phylogenetic relationships among the outgroup taxa and species of Anaxagoreoideae and Ambavioideae. B, Phylogenetic relationships in Annonoideae. C, Phylogenetic relationships in Malmeoideae.

Figure 1.

Figure 1.

Phylogram showing one of many most parsimonious tree topologies, with support [maximum parsimony bootstrap (BS) percentages, maximum likelihood BS percentages and Bayesian posterior probabilities (PP) indicated in the key]. A, Phylogenetic relationships among the outgroup taxa and species of Anaxagoreoideae and Ambavioideae. B, Phylogenetic relationships in Annonoideae. C, Phylogenetic relationships in Malmeoideae.

Figure 1.

Figure 1.

Phylogram showing one of many most parsimonious tree topologies, with support [maximum parsimony bootstrap (BS) percentages, maximum likelihood BS percentages and Bayesian posterior probabilities (PP) indicated in the key]. A, Phylogenetic relationships among the outgroup taxa and species of Anaxagoreoideae and Ambavioideae. B, Phylogenetic relationships in Annonoideae. C, Phylogenetic relationships in Malmeoideae.

The monophyly of the remainder of Annonaceae, representing > 97% of the species diversity of the family, is well supported in all analyses (clade D; PBP, 99; MBP, 100; PP, 1.00; Fig. 1A). The sister clades E and F receive maximum support in all three analyses.

Most of the relationships in Annonoideae, clade E, receive strong support (maximum parsimony, ML and Bayesian; Fig. 1B). Bocageeae (clade G) receives maximum support (maximum parsimony, ML and Bayesian), as does the sister relationship between Bocageeae and the remaining Annonoideae (clade H). The latter splits into five strongly supported, species-rich clades, the relationships of which, however, are mostly unresolved: Xylopia L./Artabotrys R.Br. (clade I: c. 250 species; PBP, 97; MBP, 100; PP, 1.00); Duguetia A.St.-Hil., Fusaea (Baill.) Saff., Letestudoxa Pellegr. and Pseudartabotrys Pellegr. (clade J: c. 100 species; PBP, 100; MBP, 100; PP, 1.00); Guatteria Ruiz & Pav. (clade K: c. 250 species; PBP, 100; MBP, 100; PP, 1.00); Annona L., Anonidium Engl. & Diels, Asimina Adans., Disepalum Hook.f., Goniothalamus (Blume) Hook.f. & Thomson and Neostenanthera Exell (clade M; PBP, 100; MBP, 100; PP, 1.00); and a clade containing Palaeotropical species only (clade N: c. 600 species; PBP, 100; MBP, 100; PP, 1.00). Clade M is sister to clade N (PBP, 100; MBP, 100; PP, 1.00). Clade N consists of three maximally supported clades: clade O, including two monotypic African genera Ophrypetalum Diels and Sanrafaelia Verdc.; clade P, including c. 80 African tree species; and clade Q, including c. 475 Palaeotropical climbing species. A small number of more shallow nodes in the tree are much more strongly supported by Bayesian PPs compared with the results of the maximum parsimony and ML analyses. For example, the sister group relationship between Toussaintia Boutique and a clade containing Friesodielsia Steenis and Monanthotaxis Baill. (clade R; PBP, 62; MBP, 78; PP, 1.00) and the monophyly of a clade of five African genera (clade S; PBP, 62; MBP, 85; PP, 1.00).

Resolution and support in Malmeoideae (clade F) are lower than in Annonoideae and Ambavioideae (Fig. 1C) across all methods. The first dichotomy in Malmeoideae divides five African genera (clade T; PBP, 74; MBP, 88; PP, 0.97) from the remaining Malmeoideae (clade U; maximally supported). The latter comprises three major clades: V, including c. 175 Neotropical species (PBP, 85; MBP, 95; PP, 1.00); W, comprising the genus Maasia Mols, Keßler & Rogstad; and X (PBP, 99; MBP, 100; PP, 1.00), in which the Asian genus Monocarpia Miq. is sister to the rest of the c. 525 species (PBP, 71; MBP, 78; PP, 1.00), mainly distributed in Asia, with four genera endemic to Central America [Desmopsis Saff., Sapranthus Seem., Stenanona Standl. and Tridimeris Baill. (35 species in total) plus a small number of species of Polyalthia from Madagascar (e.g. Ppendula Capuron ex G.E.Schatz & Le Thomas) and eastern Africa (e.g. Pstuhlmannii (Engl.) Verdc.].

Overall resolution in this clade is poor in all three analyses, but, nevertheless, the nonmonophyly of Polyalthia, with c. 135 currently recognized species, is indicated.


Molecular phylogenetic studies, such as this, benefit in part from the availability of many more characters (Chase & Cox, 1998). A further benefit of DNA sequence data over morphological characters that have been used previously to infer phylogenetic relationships is the ability to isolate conflicting phylogenetic signals that can be confounded in patterns of inheritance of morphological variation. We used sequence data from eight plastid loci, representing a large number of characters that are always inherited as a single unit (the plastid genome) without recombination (Birky, 2001). In one instance, differing phylogenetic signals caused by paralogy have been identified for putative plastid sequences in Annonaceae (Pirie et al., 2007), but, in general, congruence of plastid loci has been demonstrated. Congruence between low-copy and plastid loci analyses has been documented in other families (e.g. Górniak, Paun & Chase, 2010). In the absence of contradictory evidence (in the form of independent DNA loci), we assume that this plastid tree (Fig. 1) broadly reflects the potentially more complex phylogenetic history underlying the complete genomes of the taxa involved. Support for this topology is largely robust, at least for the purposes of defining major clades, although resolution in these clades is, in some cases, poor. Poor resolution does not affect the creation of an infrafamilial classification for Annonaceae because weakly supported/resolved clades are not given taxonomic recognition. The supermatrix approach employed here works well, particularly for the resolution of deeper nodes, and the generally higher Bayesian PPs (compared with BPs) reported here are a further attribute known to be associated particularly with matrices comprising a proportion of missing data (Wiens, 2006; Pirie et al., 2008). We consider PP ≥ 0.95 to be robust, even if BPs are considerably lower.

Morphological characters and the history of Annonaceae classification

‘The family of Annonaceae is a very natural one’. With this statement, King (1893) was hardly exaggerating the perceived monophyly of the family, although he clearly would not have used that term. Apart from frequent inclusion of the monotypic Eupomatiaceae in the past (e.g. Baillon, 1868; Diels, 1912), delimitation of the Annonaceae has never been ambiguous, because of the presence of synapomorphies, such as simple vessel perforations, a ‘cobweb-like’ wood structure in cross-section (caused by broad and high multiseriate xylem rays with many narrow, tangential parenchyma bands perpendicular to the xylem rays; Koek-Noorman & Westra, 2012), alternate, distichous leaves, trimerous calyx and corolla, and perichalazal ovules (Sauquet et al., 2003). On the other hand, ambiguity has governed the delimitation of groups within the family from the stance of their recognition. A number of attempts have been made to formally classify Annonaceae in tribes or subfamilies (e.g. Endlicher, 1839; Hutchinson, 1923) or to produce informal groupings of genera (e.g. Diels, 1932; Fries, 1959; Walker, 1971). Each of these classifications differs in the kinds of characters that were emphasized (a summary is provided in Table 4). Following the first treatment of the family by Dunal (1817), several authors used fruit characters for the main divisions of the family (Endlicher, 1839; Diels, 1932). Baillon (1868) and Prantl (1891) produced similar treatments, as the groups they proposed were based solely on floral characters. Fries (1959) produced a comprehensive treatment of the family, the size of which had greatly expanded because of his numerous revisionary works (e.g. Fries, 1930, 1931, 1934, 1937, 1939). Floral characters plus a single fruit character (free vs. fused carpels) formed the basis for the delimitation of informal groups of genera (‘natürliche Gruppen’) in tribes. The work of Fries (1959) is probably the most influential classification of Annonaceae to date, in particular having served as a guide for the planning of taxonomic studies in the family. It has, however, not gone without criticism, even before the application of molecular phylogenetic data. Le Thomas (1969), for example, expressed doubts as to the systematic value of sepal and petal aestivation, one of the key floral characters in the classification of Annonaceae, notably in the system by Fries (1959), but also by Hooker & Thomson (1855). In Le Thomas' (1969) treatment of Annonaceae for the Flore du Gabon, the limited systematic value of petal aestivation was illustrated with reference to Uvarieae. This tribe featured genera with both apocarpous (e.g. Uvaria, Cleistopholis Pierre ex Engl.) and syncarpous [Letestudoxa, Pachypodanthium Engl. & Diels (now Duguetia)] fruits. In addition, substantial variation in the number of ovules and placentation type occurs within these seven genera. In effect, Uvarieae was considered to be related only because of imbricate petals and occasional stellate hairs. In this article, we will limit ourselves to a few examples, such as that of Uvarieae given here. Almost any past classification can be justified to some extent, in that the responsible author hypothesized a close relationship between particular genera, which, with hindsight, appears to have been correct. However, in each case, there are also as many (if not more) problems to be pointed out. We consider that there is little point in evaluating in detail these various classification schemes, as the common methodological ground and explicit justification that are necessary to clarify or test the reasons for differences between them are missing. They are all highly intuitive and therefore irrefutable.

Table 4.  Overview of previous infrafamilial classifications of Annonaceae. Informal names given in inverted commas. Lists of genera included in infrafamilial taxa are not comprehensive, and are restricted to major genera and generally to names currently recognized
PublicationTaxaDiagnostic characters (and key genera included)
  • *

    Partial study, geographically focused on taxa from ‘British India’.

  • Listed as tribe ‘Saccopetaleae’ (Hooker & Thomson, 1855: pp. 91, 92).

  • Similar infrafamilial classification adopted by Hooker & Thomson (1872), although without recognition of subtribes within tribe Mitrephoreae.

  • §

    Descriptions not provided for tribes and subtribes, although with extensive descriptions of genera included.

  • Now excluded from Annonaceae as Eupomatiaceae.

Rafinesque (1815)Subfamily Anonoideae Raf.Monocarps fused (Annona)
 Subfamily Uvarioideae Raf.Monocarps free, indehiscent (Guatteria, Melodorum, Porcelia and Uvaria)
 Subfamily Xylopioideae Raf.Monocarps free, dehiscent (Xylopia)
Kosteletzky (1836)Subfamily Anonoideae Raf.Monocarps many, fused (Annona)
 Subfamily Monodorioideae Kostel.Monocarps single, possibly fused (Monodora)
 Subfamily Uvarioideae Raf.Monocarps many, free (Artabotrys, Polyalthia, Uvaria and Xylopia)
Reichenbach (1837)‘Anonariae’Petals free; monocarps free or fused
  ‘Uvarieae’ (subgroup) Stamens many; monocarps free (Asimina, Porcelia and Xylopia)
  ‘Bocageeae’ (subgroup) Stamens 6, opposite the petals (Bocagea)
  ‘Anoneae genuinae’ (subgroup) Stamens many; monocarps fused (Annona and Monodora)
 ‘Cardiopetaleae’Petals basally fused (Hexalobus, Miliusa and Orophea)
 ‘Guatterieae’Petals connivent or patent; monocarps single-seeded (Anaxagorea, Artabotrys, Duguetia, Guatteria and Polyalthia)
Endlicher (1839)Tribe Anoneae Endl.Stamens many; ovules basal (Anaxagorea, Annona, Artabotrys, Duguetia and Guatteria)
 Tribe Bocageeae Endl.Stamens few; ovules ventral (Bocagea, Miliusa, Orophea and Popowia)
 Tribe Xylopieae Endl.Stamens many; ovules ventral (Hexalobus, Polyalthia, Uvaria and Xylopia)
Hooker & Thomson (1855)*Tribe Anoneae Endl.Monocarps fused (Annona)
 Tribe Guatterieae Hook.f. & ThomsonPetals flat or basally slightly curved (Anaxagorea, Artabotrys, Cananga, Cyathocalyx, Guatteria, Phaeanthus and Polyalthia)
 Tribe Miliuseae Hook.f. & ThomsonStamens loosely imbricate (Miliusa and Orophea)
 Tribe Mitrephoreae Hook.f. & ThomsonInner petals clawed (Goniothalamus, Mitrephora, Orophea and Popowia)
 Tribe Uvarieae Hook.f. & ThomsonPetal aestivation imbricate (Sageraea, Stelechocarpus and Uvaria)
 Tribe Xylopieae Endl.Inner petals thick, 3-angled (Melodorum and Xylopia)
Bentham (1863)Tribe Miliuseae Hook.f. & ThomsonPetals diverse; stamens loosely imbricate, connective dorsally conspicuously or not enlarged (Alphonsea, Bocagea, Miliusa and Orophea)
 Tribe Mitrephoreae Hook.f. & ThomsonPetals valvate, outer apert, inner connivent or connate
  Subtribe Eumitrephoreae Benth. & Hook.f.Inner petals shorter than outer or subequal, often basally clawed (Goniothalamus, Mitrephora and Monodora)
  Subtribe Phaeantheae Benth. & Hook.f.Inner petals much larger than outer, erect (Cymbopetalum and Phaeanthus)
 Tribe Unoneae Benth. & Hook.f.Petals flat, slightly unequal or inner smaller than outer, generally apert (Anaxagorea, Artabotrys, Asimina, Cananga, Cyathocalyx, Disepalum, Polyalthia, Popowia, Hexalobus and ‘Unona’)
 Tribe Uvarieae Hook.f. & ThomsonPetals flat, spreading, at least inner imbricate; stamens densely crowded, connective apex enlarged (Duguetia, Guatteria, Porcelia, Sageraea, Stelechocarpus and Uvaria)
 Tribe Xylopieae Endl.Outer petals thick, connivent or slightly apert; inner petals enclosed, small or absent (Annona, Melodorum and Xylopia)
Baillon (1868)§Tribe Anoneae Endl. 
  Subtribe Oxymitreae Baill.(Cymbopetalum, Mitrephora, Orophea and ‘Oxymitra’)
  Subtribe Rollinieae Baill.(Artabotrys, Cyathocalyx, Hexalobus and ‘Rollinia’)
  Subtribe Unoneae Baill.(Anaxagorea, Bocagea, Disepalum, Popowia and ‘Unona’)
  Subtribe Uvarieae Baill.(Cananga, Oxandra, Sageraea and Uvaria)
  Subtribe Xylopieae Baill.(Annona and Xylopia)
 Tribe Eupomatieae Baill.(Eupomatia)
 Tribe Miliuseae Hook.f. & Thomson(Miliusa and Phaeanthus)
 Tribe Monodoreae Baill.(Monodora)
Prantl (1891)Tribe Eupomatieae Baill.Perianth absent; perigynous (Eupomatia)
 Tribe Melodoreae PrantlAll or only inner petals erect, petals touching; stamens numerous; apocarpous; hairs simple (Melodorum, Phaeanthus and Piptostigma)
 Tribe Miliuseae Hook.f. & ThomsonStamens connective apex not expanded over anthers; stamens few; apocarpous; hairs simple (Alphonsea, Bocagea, Mezzettia, Miliusa, Oxandra and Sageraea)
 Tribe Mitrephoreae Hook.f. & ThomsonInner petals usually clawed, apically connivent; stamens sometimes few; apocarpous; flowers small (Mitrephora, Orophea and Popowia)
 Tribe Monodoreae Baill.Syncarpous, unilocular with parietal placentation; petals more or less fused at base; valvate (Monodora)
 Tribe Unoneae Benth. & Hook.f.Petals subequal, spreading or slightly converging; stamens numerous; hairs simple (Anaxagorea, Cananga, Cymbopetalum, Disepalum, Polyalthia and ‘Unona’)
 Tribe Uvarieae Hook.f. & ThomsonAt least outer petals imbricate, spreading or slightly erect; stamens numerous (Asimina, Duguetia, Guatteria, Porcelia, Stelechocarpus and Uvaria)
 Tribe Xylopieae Endl.Petals spoon-shaped at base, more or less constricted above and then spreading or laterally compressed (Annona, Artabotrys, Cyathocalyx, Hexalobus and Xylopia)
Hutchinson (1923, 1964)Subfamily Annonoideae Raf.Carpels free or fused into multilocular syncarp; stigmas erect (rarely radiating)
  Tribe Miliuseae Hook.f. & ThomsonPetals in one or two whorls, valvate; outer petals smaller than inner (Cymbopetalum, Marsypopetalum, Miliusa, Orophea, Phaeanthus, Piptostigma and Trivalvaria)
  Tribe Unoneae Benth. & Hook.f.Petals in one or two whorls, valvate; outer petals subequal or larger than inner
   Subtribe Annonineae Hutch.Carpels fused (Annona)
   Subtribe Xylopiineae Hutch.Carpels free or only slightly fused (Alphonsea, Anaxagorea, Artabotrys, Asimina, Cyathocalyx, Dasymaschalon, Desmopsis, Desmos, Disepalum, Drepananthus, Goniothalamus, Haplostichanthus, Hexalobus, Hornschuchia, Meiogyne, Melodorum, Mezzettia, Mitrephora, Monanthotaxis, Monocarpia, Orophea, Platymitra, Polyalthia, Popowia, Unonopsis, Uvariastrum, Uvariopsis and Xylopia)
  Tribe Uvarieae Hook.f. & ThomsonPetals in two whorls, both or inner imbricate; leaf indument stellate or lepidote (Duguetia, Enicosanthum, Ephedranthus, Fusaea, Guatteria, Malmea, Oxandra, Porcelia, Sageraea, Sapranthus, Stelechocarpus and Uvaria)
 Subfamily Monodoroideae Kostel.Carpels fused into unilocular ovary, placentation parietal; stigmas radiating (Isolona and Monodora)
Fries (1959)Subfamily Annonoideae Raf.Carpels spirally arranged (sometimes few in a whorl); apocarpous, rarely syncarpous (forming multilocular syncarp)
  Tribe Tetramerantheae R.E.Fr.Petals in whorls of four, imbricate; stigma 3-lobed, appressed to ovary; leaves spirally arranged (Tetrameranthus)
  Tribe Unoneae Benth. & Hook.f.Petals (at least outer) valvate (imbricate in Porcelia); leaves distichous (Alphonsea, Anaxagorea, Annona, Anonidium, Artabotrys, Cananga, Cyathocalyx, Cymbopetalum, Dasymaschalon, Desmos, Disepalum, Drepananthus, Fissistigma, Goniothalamus, Marsypopetalum, Meiocarpidium, Meiogyne, Mezzettia, Miliusa, Mitrephora, Monanthotaxis, Monocarpia, Neostenanthera, Onychopetalum, Orophea, Phaeanthus, Piptostigma, Polyalthia, Polyceratocarpus, Popowia, Porcelia, Pseuduvaria, Sphaerocoryne, Trigynaea, Trivalvaria, Unonopsis, Uvariastrum, Uvariodendron, Uvariopsis and Xylopia)
  Tribe Uvarieae Hook.f. & ThomsonPetals imbricate; leaves distichous (Asimina, Cleistopholis, Cremastosperma, Desmopsis, Duguetia, Enicosanthum, Ephedranthus, Fusaea, Guatteria, Hexalobus, Malmea, Oxandra, Pseudoxandra, Sageraea, Sapranthus, Stelechocarpus, Stenanona and Uvaria)
 Subfamily Monodoroideae Kostel.Carpels whorled, fused in unilocular ovary, placentation parietal (Isolona and Monodora)

Students of Annonaceae in the 19th and most of the 20th century assembled impressive numbers of collections, observations and taxonomic papers, but simply lacked the rigour of using any formal methodology to establish a classification. In reviewing those leading up to that of Fries (1959), it does, however, become apparent that a limited number of key characters have consistently been used for infrafamilial classification. These can be summarized as those concerning phyllotaxy, indument, inflorescence position, sepal and petal aestivation, petal fusion, shape and form, anther connective form, carpel fusion and placentation.

Subsequent to Fries (1959), Annonaceae systematics has focused on a wider range of differing sources of data and has generally involved a variety of more or less formal methodologies. Walker (1971) used pollen and, to a lesser extent, floral morphology and phytogeography. Following this pioneering work, Le Thomas (1980, 1981), Le Thomas & Lugardon (1976) and Walker (1971, 1972) assembled a comprehensive overview of palynological characters of Annonaceae. Although these data were largely novel, the analytical approach was similar to that of several preceding studies: observed similarities between Annonaceae, on the one hand, and presumed primitive angiosperms, extant and fossil, on the other, together with assumed transformation series, were taken as a primary guide to hypothesize relationships. In addition, Le Thomas (1980, 1981) based several decisions on ancestral and derived pollen characters on presumed evolutionary trends (‘series’), for instance from simple to complex characters or from free to fused parts, and on an analogy between morphological series in pollen characters and macromorphological characters. Although Le Thomas (1980, 1981) mentioned the concepts of homology and convergent evolution, she lacked the analytical tools to demonstrate the significant levels of homoplasy that have since been demonstrated (Doyle & Le Thomas, 2012). Noteworthy contributions to the classification of Annonaceae (van Heusden, 1992; van Setten & Koek-Noorman, 1992) appeared immediately before the first cladistic papers (Doyle & Le Thomas, 1994, 1995), allowing the last two publications to benefit from the comprehensive overviews of floral morphology (van Heusden, 1992) and fruit and seed morphology (van Setten & Koek-Noorman, 1992). Observations that led to intuitive groupings of genera in each of these publications separately were combined and used for a phenetic analysis (Koek-Noorman, van Setten & van Zuilen, 1997). Koek-Noorman et al. (1997) used a neighbor-joining tree derived from these data to produce an informal classification (‘grouping’) that was similar to that of van Setten & Koek-Noorman (1992). In a few cases, particular results from the neighbor-joining tree were disregarded in the classification, for example, in the case of Porcelia Ruiz & Pav.: established opinion based on inflorescence type, anther septation and pollen size was that the genus was allied to the other genera in Bocageeae (Fig. 1B), but, in the analysis of Koek-Noorman et al. (1997), it was clustered distant from its putative close relatives, possibly as a result of the absence of a seed appendage, either an aril or a caruncle (Johnson & Murray, 1995), that is otherwise present in all species of Bocageeae.

The first formal cladistic analyses based on morphological, anatomical and palynological characters revealed rampant homoplasy. The consistency index of 79 characters scored for 42 genera of Annonaceae was as low as 0.27 in Doyle & Le Thomas (1996). The retention index was not reported, but, as autapomorphies were absent and the number of symplesiomorphies was low, the proportion of similarity in their tree to be interpreted as synapomorphies is also likely to have been low. There are several reasons for this high level of homoplasy, one of which is the difficulty of homology assessment, as acknowledged by Doyle & Le Thomas (1996), in cases in which observations on living material were absent and characters were scored on the basis of observations of herbarium material (e.g. petal connivence, fruit wall thickness). Moreover, some characters were included that have been shown to be derived via different developmental pathways. Because of the lack of similarity in anatomy and development, it comes as no surprise that these characters were demonstrated by Doyle & Le Thomas (1996) to be homoplasious. An example of such a character is syncarpous fruits, as opposed to those consisting of free monocarps. Several papers have demonstrated that syncarpy can better be interpreted as different, nonhomologous, suites of characters related to fusion or coherence between carpels, fusion or coherence between carpels and receptacle, and syncarpy in flowering stage (Briechle-Mäck, 1994; Chatrou & He, 1999; Chatrou et al., 2000; Couvreur et al., 2008). Finally, part of the homoplasy as found by Doyle & Le Thomas (1996) can be explained by patterns apparent in their family-wide analysis. In many cases, homoplasy was not distributed evenly across the tree. This can be illustrated by the fact that similar characters, when used in phylogenetic analyses of morphological characters of clades within Annonaceae (Johnson & Murray, 1995; Chatrou et al., 2000), showed little to no homoplasy. Examples are the occurrence of different trichome types, lianescent habit and presence vs. absence of styles in the clade comprising Duguetia and related genera (Chatrou et al., 2000), and the occurrence of bilobed arils and different shapes of the floral receptacle in Bocageeae (Johnson & Murray, 1995). These characters do not exhibit homoplasy in these relatively small, more exclusive clades, but have all evolved in parallel multiple times in Annonaceae as a whole.


Because classifications are important for communication and information retrieval, stability and universal applicability should be their main features. Previous classifications of Annonaceae, to a large extent, fail to meet both criteria.

We assign the rank of subfamily to four clades: Anaxagorea (Anaxagoreoideae); clade C (Ambavioideae), corresponding to the clade previously referred to informally as the ambavioids (e.g. Doyle & Le Thomas, 1994, 1995, 1996; Doyle et al., 2000; Sauquet et al., 2003); clade E (Annonoideae), the LBC or inaperturate clade (e.g. Richardson et al., 2004; Erkens et al., 2007; Couvreur et al., 2008; Zhou et al., 2009); and clade F (Malmeoideae), the SBC or malmeoid/piptostigmoid/miliusoid clade (e.g. Richardson et al., 2004; Pirie et al., 2006). These four clades include all the genera sampled for these analyses (i.e. 90% of all accepted genera).

Clade E, which includes the genus Annona, must be assigned the name Annonoideae Raf. (McNeill et al., 2006; art. 19.4). All other previously described subfamilies of Annonaceae are based on names of genera that belong to Annonoideae, namely Bocageoideae Pfeiff., Monodoroideae Kostel., Uvarioideae Raf. and Xylopioideae Raf., and are therefore all superfluous. The somewhat curious bias towards subfamilies based exclusively on annonoid genera can be explained by their greater morphological variability (particularly in fruit characters). In the absence of further available subfamilial names, Anaxagoreoideae, Ambavioideae and Malmeoideae are newly described here.

As with all recent classificatory endeavours in angiosperms (e.g. APG III, 2009), strongly supported monophyly is the foremost principle, at any taxonomic level. For our classification of Annonaceae, this is also true. Monophyly is the only characteristic that these subfamilies have in common, however: they are not otherwise equivalent in terms of age, geographical distribution or any other biological attributes (and should not be expected to be so). Following monophyly, morphological diagnosability is an important subsidiary criterion for the classification of groups, which, in this case, is challenging. As stated above, clear morphological synapomorphies have yet to be identified for most clades of Annonaceae. However, we do not believe that this should prevent us from recognizing infrafamilial taxa, especially because the use of different terms for the same clades, as mentioned above, is likely to cause misunderstanding (APG I, 1998). The plethora of informally named groups resulting from the classifications proposed to date demonstrates the need for named units within the family. Furthermore, there are many characters awaiting evaluation for their diagnostic value; two recently published examples of characters that have a good, although not perfect, fit onto the phylogenetic tree for Annonaceae are phyllotaxis (Johnson, 2003) and orbicules (Huysmans et al., 2010).

The need to identify clades and to name them extends beyond the level of subfamilies. A comprehensive classification of Annonaceae recognizes further taxa at lower, namely tribal, ranks. Characters that are diagnostic for these less inclusive clades are more straightforward to identify than for the four subfamilies. Nevertheless, extensive homoplasy still makes the diagnosis of tribes only possible by the enumeration of a suite of characters that individually are found in several clades (i.e. that are individually homoplasious). Some useful characters include phyllotaxis (distichous phyllotaxis is shared by members of clade L), habit (a climbing habit is common to species of clade Q, although with exceptions, e.g. in most species of Dasymaschalon (Hook.f. & Thomson) Dalla Torre & Harms: Wang, Chalermglin & Saunders, 2009) and pollen characters (clade R is characterized by cryptoaperturate/disulculate pollen grains; T. Chaowasku et al., unpubl. data). The phylogenetic tree presented here includes some poorly resolved clades, but the lack of resolution does not prevent formal classification at the tribal level. The composition of taxa at lower ranks, such as tribes, has frequently changed, and they have rarely been used in formal classifications. Fries (1959) named only three tribes in his classification of the family: Uvarieae, Unoneae and Tetramerantheae. Of these, Tetramerantheae included the ambavioid genus Tetrameranthus R.E.Fr. only. Uvarieae and Unoneae, by contrast, included genera dispersed across Annonoideae, Malmeoideae and Ambavioideae as circumscribed here. A more recent revision and recircumscription of Saccopetaleae (Keßler, 1988) also resulted in a nonmonophyletic group of genera (this study; T. Chaowasku et al., unpubl. data). As many clades are well supported, it makes sense to recognize them formally, and thus, in addition to the naming of subfamilies, we provide a set of tribes in order to avoid adding further disorder to the classification of Annonaceae.

Eupomatia R.Br., containing three species, has previously been included in Annonaceae (Bentham, 1863; Baillon, 1868; Prantl, 1891; Diels, 1912), usually in its own tribe: Eupomatieae Baill. Subsequent precladistic classifications (e.g. Hutchinson, 1973; Cronquist, 1988; Takhtajan, 1997) always considered Eupomatiaceae to be a separate family from Annonaceae. Given this classificatory history and the fact we are presenting a new classification of Annonaceae, the classification of Eupomatia needs to be addressed here. Eupomatia has been shown to be sister to Annonaceae in several phylogenetic analyses of angiosperms (e.g. Qiu et al., 2005) and Magnoliales (Sauquet et al., 2003). Therefore, both recognition and rejection of familial status for Eupomatia would be in agreement with the principle of monophyly. Despite the ‘annonaceous appearance’ of Eupomatia, there are hardly any characters that are synapomorphic for Annonaceae and Eupomatia combined. Sauquet et al. (2003) only listed fibrous mesotesta as a possible synapomorphy. The similarities between Annonaceae and Eupomatia are often symplesiomorphies as they are shared with the sister clade of Annonaceae/Eupomatia, consisting of Himantandraceae and Degeneriaceae. These characters include adaxial prophylls, a flat-concave floral receptacle, apical extension of the connective and testal ruminations (Doyle & Le Thomas, 1997; Endress & Doyle, 2009; Endress & Armstrong, 2011). Furthermore, the clearest synapomorphies of Annonaceae (that is, characters showing no or hardly any homoplasy) are not shared with Eupomatia. Despite considerable variation in floral morphology (Xu & Ronse De Craene, 2010), the floral bauplan of Annonaceae is uniform (Saunders, 2010) and provides several synapomorphies (Endress & Armstrong, 2011), such as a whorled floral phyllotaxis (vs. spiral in Eupomatiaceae), plicate carpels (vs. ascidiate carpels) and trimerous perianth (vs. many tepals). The broad and high multiseriate xylem rays with many narrow, tangential parenchyma bands perpendicular to the xylem rays is a typical wood anatomical feature characterizing every species of Annonaceae investigated so far (Koek-Noorman & Westra, 2012). This wood structure is absent in Eupomatia (L. Y. T. Westra & L. W. Chatrou, pers. observ.; http://insidewood.lib.ncsu.edu). Given the character distribution described here, we do not favour the inclusion of Eupomatia in Annonaceae and leave it out of the classification presented here.

Because subfamily Anaxagoreoideae consists of Anaxagorea only, we feel that it would be taxonomically redundant to also recognize this clade at the tribal level. Subfamily Ambavioideae comprises nine genera and just over 50 species. To split this subfamily further into tribes would involve the recognition of three tribes, consisting of Meiocarpidium, the clade containing Cananga (DC.) Hook.f. & Thomson, and the clade containing Ambavia Le Thomas, respectively. Apart from creating undesirably species-poor tribes, this alternative would focus on differences between the three groups, such as the basic chromosome number x = 7 for the Canaga clade and x = 8 for the Ambavia clade. In order to avoid insoluble debates about the level of morphological difference required for a clade to be recognized as a classificatory unit, not just in Ambavioideae but in other subfamilies as well, we focus on synapomorphies recognizing clades that are as inclusive as possible. In the case of Ambavioideae, these are the presence of a middle integument (Christmann, 1989) and some palynological synapomorphies (Doyle & Le Thomas, 2012).

Parts of the tree for Annonaceae that require additional sampling of taxa and character include the relationships among species-rich clades in Annonoideae (clades I, J, K and L). We argue that these species-rich clades can be given tribal status without their interrelationships being fully resolved. The clades are morphologically distinct and, to a large extent, they have been recognized in the past as groups of related genera. An exception is the clade comprising Artabotrys and Xylopia, for which clear synapomorphies have yet to be identified.

The clade that would require much better resolution before any classificatory conclusion can be drawn is clade X, containing most Asian and Central American endemic genera in Malmeoideae. Sequence divergence is low among species in this clade and, as a result, support is low in general. So far, our knowledge of phylogenetic relationships is based on only one-eighth of the species in clade X. An advantageous strategy would seem to be to increase taxon sampling, as this would provide a more accurate estimation of phylogenetic model parameters (Heath, Hedtke & Hillis, 2008). In the light of this, we recircumscribe tribe Miliuseae (clade X excluding Monocarpia) and envisage that its classification be revisited once relationships are better resolved. The inclusion of the genus Monocarpia in Miliuseae would have been possible from the perspective of monophyly; we prefer, however, to erect a new tribe for Monocarpia alone as this genus has never previously been included in Miliuseae (Mols & Keßler, 2003; Mols et al., 2004). Monocarpia furthermore lacks cryptoaperturate/disulculate pollen, which is synapomorphic for Miliuseae as circumscribed here (Chaowasku, Keßler & van der Ham, 2012). Relationships among clades V, W and X are unresolved. We apply a similar reasoning here as for clades I, J, K and L in Annonoideae: the lack of interclade resolution does not hinder the recognition of tribes, which are morphologically distinct. Saunders et al. (2011) demonstrated that the Malagasy genus Fenerivia Diels, unsampled in this study, is part of this polytomy. Hence, clades V, W, X and Fenerivia are given tribal status. Furthermore, we classify the genus Dendrokingstonia Rauschert in tribe Dendrokingstonieae, in accordance with results of phylogenetic analyses (Chaowasku et al., 2012). In doing so, we assign all genera of Annonaceae without exception to a tribe.

Considering the nomenclature of tribes, the principle has been adopted that published names of tribes, based on a generic name, are assigned to clades containing that genus. To a large extent, we could use tribal names that have been published before, using the oldest names that apply to clades in our tree. Names that have become superfluous are Tetramerantheae R.E.Fr., Melodoreae Prantl, Mitrephoreae Hook.f. & Thomson and Unoneae Benth. & Hook.f. Six new tribes need to be described as none of the existing names could be applied.

The macromorphological characters most frequently emphasized in classifications of Annonaceae concern the form of the fruit, inflorescence position, and shape and aestivation of the sepals and petals. Some are demonstrably nonhomologous, such as syncarpous fruits (as discussed above). The primary homology of further historically important characters, such as numbers of carpels, numbers of ovules per carpel, fusion of petals, and shape and insertion of petals, can best be tested in the light of our tree. We have not attempted a formal character analysis here. It is, however, clear from the phylogenetic relationships presented here that historically important characters do not consistently characterize major clades. Indeed, it is not obvious which, if any, individual traits might be interpreted as unequivocal diagnostic characters for more inclusive clades. In our description of subfamilies and tribes, we document what we see as suites of characters most useful for broadly defining clades. These characters have been taken from our own observations and those of Maas & Westra (1984), Westra (1985), Morawetz & Le Thomas (1988), van Heusden (1992), van Setten & Koek-Noorman (1992), Johnson & Murray (1995), Doyle & Le Thomas (1996), Svoma (1998), Johnson (2003), Maas, Westra & Chatrou (2003), Tsou & Johnson (2003), Scharaschkin & Doyle (2005, 2006), Su & Saunders (2006), Maas, Westra &Vermeer (2007), Couvreur (2009), Huysmans et al. (2010), Surveswaran et al. (2010) and Weerasooriya & Saunders (2010).

Anaxagoreoideae Chatrou, Pirie, Erkens & Couvreur, subfam. nov.—TYPE: Anaxagorea A.St.-Hil.

Trees with distichous phyllotaxis; hermaphroditic; carpels free in flower and fruit; stamen apex pointed or rounded; connective extension anthers nonseptate; inner staminodes present; orbicules present; ovules two (basal); monocarp abscission at base of stipe; monocarps ventrally dehiscent; seeds not arillate, asymmetrical; middle seed integument absent; basic chromosome number x = 8.

Included genus: Anaxagorea.

Ambavioideae Chatrou, Pirie, Erkens & Couvreur, subfam. nov.—TYPE: Ambavia Le Thomas

Trees with spiral or distichous phyllotaxis; hermaphroditic; carpels free in flower and fruit; apical connective prolongation tongue-shaped, peltate-apiculate or peltate-truncate; anthers nonseptate; staminodes absent; ovules two—numerous (lateral); orbicules present; monocarp abscission at apex or base of stipe; monocarps indehiscent; seeds sometimes arillate, symmetrical; middle seed integument usually present; endosperm ruminations irregular; basic chromosome number x = 7 or 8.

Included genera: Ambavia, Cananga, Cleistopholis, Cyathocalyx Champ. ex Hook.f. & Thomson, Drepananthus Maingay ex Hook.f., Lettowianthus Diels, Meiocarpidium, Mezzettia Becc., Tetrameranthus.

A synapomorphy for this clade is a middle integument (Christmann, 1989; Svoma, 1998; Lucas et al., 2012).

Annonoideae Raf., Anal. Nat. 175., Apr–Jul 1815 (‘Annonidia’), descr. emend.—TYPE: Annona L.

Trees or lianas with spiral or distichous phyllotaxis; hermaphroditic, sometimes (andro)dioecious, rarely (andro)monoecious; carpels free or fused in flower and fruit; ovules one (basal, rarely apical) to numerous (lateral); apical connective prolongation peltate-truncate, peltate-apiculate, rarely tongue-shaped or absent; anthers septate or nonseptate; outer staminodes rarely present; orbicules usually absent; monocarps indehiscent or dehiscent (adaxially or abaxially); seeds sometimes arillate, symmetrical; middle seed integument absent (only present in Artabotrys); endosperm ruminations usually lamelliform, sometimes irregular; basic chromosome number x = 7, 8 or 9.

Although Bocagea A.St.-Hil., Cardiopetalum Schltdl. and Froesiodendron R.E.Fr. have not been sampled for our phylogenetic analyses, we believe their inclusion in Annonoideae is warranted. Together with Cymbopetalum Benth., Mkilua Verdc., Porcelia and Trigynaea Schltdl., these genera belong to Bocageeae, which are clearly set apart from other Annonaceae by a combination of inflorescence, pollen and seed characters (Johnson & Murray, 1995). The inflorescence is an internodal pedicel that is articulated at the base (Murray, 1993; Johnson & Murray, 1995). The lack of bracts, below and above the articulation, precludes sympodial growth of the inflorescence and, as a consequence, the flowers are solitary. Like solitary flowers, columellar polyads (Johnson & Murray, 1995; Tsou & Fu, 2007; Doyle & Le Thomas, 2012) only occur in this clade of Annonaceae, and therefore are clear synapomorphies. Bilobed arils (Murray, 1993; Johnson & Murray, 1995) are an additional feature characteristic of the genera in clade G, even though these are also present in Asimina, Xylopia and Cananga. Given that these three characters occur in the three unsampled genera, they are included here.

The unsampled genus Afroguatteria Boutique is also included in Annonoideae. This genus of two species has not been included in any molecular phylogenetic analysis. However, in a morphological cladistic analysis by Doyle & Le Thomas (1996), it was found to be in a clade of climbers that corresponds to clade O in our analyses. Species of Afroguatteria are climbers, which, on its own, merits inclusion in Annonoideae. The many-seeded apocarpous fruits, valvate sepals and imbricate petals could indicate a close relationship between Afroguatteria and Uvaria. A climbing habit also supports the addition of two monotypic African genera, Exellia Boutique and Gilbertiella Boutique, to Annonoideae. These two genera have habitually been treated as Monanthotaxis s.l., based on pollen and flower morphology (Le Thomas, 1969; Le Thomas & Doyle, 1996). Bygrave (2000) included Exellia scamnopetala (Exell) Boutique in his phylogenetic analysis of rbcL sequences of Annonaceae, which placed it in an unresolved position in a clade of climbing taxa, compatible with our clade O. We were, however, unable to obtain that rbcL sequence.

Cleistochlamys Oliv. is a third monotypic genus that has been associated with Monanthotaxis (in this case, Popowia Endl.; Verdcourt, 1971). Its facultative climbing habit and association with Hexalobus A.DC. and other annonoid genera based on pollen characters (Walker, 1971) underpin its incorporation in Annonoideae.

Despite its absence in our phylogenetic analyses, the monotypic African genus Boutiquea Le Thomas is placed in Annonoideae, based on palynological characters. Boutiquea has pollen tetrads, with a granular infratectum similar to Neostenanthera (Le Thomas, 1980; Doyle & Le Thomas, 2012). Pollen characters are also among the characters to support inclusion of Duckeanthus R.E.Fr. in Annonoideae. Pollen of this relatively poorly known monotypic genus bears strong resemblance to that of Fusaea in having large tetrads with a minutely granular exine structure (Walker, 1971; Le Thomas, Lugardon & Doyle, 1994; Doyle & Le Thomas, 2012). Moreover, Fusaea and Duckeanthus share characters of inflorescences, stamen anatomy and aril structure, which made them appear as sister genera in a morphological cladistic analysis (Chatrou et al., 2000). The only genus we include in our classification without reference to published data is Schefferomitra Diels. Just before submitting this paper, sequence data became available (PhD project of S. Meinke), showing affiliation of this monotypic genus with clade R in our analyses.

The genus Diclinanona Diels was shown to belong to Malmeoideae (Erkens et al., 2009). Because of suspicion of a mixed origin of the available sequence data of Diclinanona, T. Chaowasku et al. (unpubl. data) extracted and sequenced Diclinanona again, and showed it to belong to Annonoideae, confirming the phylogenetic position found by Richardson et al. (2004), and congruent with the absence of orbicules (Huysmans et al., 2010).

Circumscription of tribes in Annonoideae

Bocageeae Endl., Gen. Pl.: 830, Jun 1839—TYPE: Bocagea A.St.-Hil.

Trees with spiral phyllotaxis; indument of simple hairs; bracts absent; flowers solitary, terminal, bisexual; carpels free in flower and fruit; placentation lateral, uni- or biseriate; monocarp abscission at base of stipe; aril bilobed, rarely absent; ruminations lamelliform; pollen inaperturate.

Included genera: Bocagea, Cardiopetalum, Cymbopetalum, Froesiodendron, Hornschuchia, Mkilua, Porcelia, Trigynaea.

This tribe is equivalent to clade G (Fig. 1B).

Xylopieae Endl., Gen. Pl.: 831, Jun 1839—TYPE: Xylopia L.

Trees or lianas with spiral phyllotaxis; indument of simple hairs; bracts present; inflorescences rhipidiate, terminal or axillary; flowers bisexual; carpels free in flower and fruit; placentation basal or lateral and uniseriate; monocarp abscission at base of stipe; aril absent, rudimentary or bilobed; ruminations spiniform to lamelliform; pollen inaperturate or sulcate.

Included genera: Artabotrys, Xylopia.

This tribe is equivalent to clade I (Fig. 1B).

Duguetieae Chatrou & R.M.K.Saunders, tribus nov.—TYPE: Duguetia A.St.-Hil.

Trees, sometimes lianas, with spiral phyllotaxis; indument of lepidote, stellate and/or simple hairs; bracts present; inflorescences rhipidiate, terminal; flowers bisexual; carpels free in flower, in fruit usually fused; placentation basal; stipe absent; aril rudimentary, rarely absent; ruminations lamelliform; pollen inaperturate, sometimes sulculate.

Included genera: Duckeanthus, Duguetia, Fusaea, Letestudoxa, Pseudartabotrys.

This tribe is equivalent to clade J (Fig. 1B).

Guatterieae Hook.f. & Thomson, Fl. Ind. 1: 92, 126, Jul 1855—TYPE: Guatteria Ruiz & Pav.

Trees, phyllotaxis unknown; indument of simple hairs; bracts present; inflorescences rhipidiate, axillary, rarely terminal; flowers bisexual; carpels free in flower and fruit; placentation basal; monocarp abscission at base of stipe; aril absent or rudimentary; ruminations spiniform to lamelliform; pollen sulculate.

Included genus: Guatteria.

This tribe is equivalent to clade K (Fig. 1B).

Annoneae Endl., Gen. Pl.: 833, Jun 1839—TYPE: Annona R.E.Fr.

Trees with distichous phyllotaxis; indument of simple hairs, rarely stellate hairs; bracts present; inflorescences rhipidiate, terminal or axillary; flowers bisexual, sometimes (andro)dioecious; carpels free in flower, free or fused in fruit; placentation basal or lateral and uni- or biseriate; monocarp abscission at base of stipe, sometimes at apex, or stipe absent; aril bilobed, rudimentary or absent; ruminations spiniform, irregular pegs or lamelliform; pollen inaperturate.

Included genera: Annona, Anonidium, Asimina, Boutiquea, Diclinanona, Disepalum, Goniothalamus, Neostenanthera.

This tribe is equivalent to clade M (Fig. 1B).

Monodoreae Baill., Hist. Pl. 1: 263, 288. Aug–Dec 1868—TYPE: Monodora Dunal.

Trees with distichous phyllotaxis; indument of simple hairs; bracts present; inflorescences rhipidiate, terminal or axillary; flowers bisexual, sometimes dioecious; carpels free or fused in flower, free or fused in fruit; placentation parietal, or lateral and uni- or biseriate; monocarp abscission at base of stipe, or stipe absent; aril absent; ruminations lamelliform; pollen inaperturate.

Included genera: Asteranthe Engl. & Diels, Hexalobus, Isolona Engl., Mischogyne Exell, Monocyclanthus Keay, Monodora Dunal, Ophrypetalum, Sanrafaelia, Uvariastrum Engl., Uvariodendron (Engl. & Diels) R.E.Fr., Uvariopsis Engl.

This tribe is equivalent to clade O plus clade P (Fig. 1B).

Uvarieae Hook.f. & Thomson, Fl. Ind. 1: 91, 92. 1–19 Jul 1855—TYPE: Uvaria L.

Lianas, rarely trees, with distichous phyllotaxis; indument of stellate, sometimes simple hairs; bracts present; inflorescences rhipidiate, terminal or axillary; flowers bisexual; carpels free in flower and fruit; placentation basal, or lateral and uni- or biseriate; monocarp abscission at base of stipe; aril rudimentary or absent; ruminations lamelliform; pollen inaperturate; sometimes sulculate.

Included genera: Afroguatteria, Cleistochlamys, Dasymaschalon, Desmos Lour., Dielsiothamnus R.E.Fr., Exellia, Fissistigma Griff., Friesodielsia, Gilbertiella, Melodorum Lour., Monanthotaxis, Pyramidanthe Miq., Schefferomitra, Sphaerocoryne (Boerl.) Scheff. ex Ridl., Toussaintia, Uvaria.

This tribe is equivalent to clade Q (Fig. 1B).

Malmeoideae Chatrou, Pirie, Erkens & Couvreur, subfam. nov.—TYPE: Malmea R.E.Fr.

Trees with spiral phyllotaxis; hermaphroditic, sometimes (andro)dioecious, rarely (andro)monoecious; carpels free in flower and fruit; apical connective prolongation peltate-truncate, peltate-apiculate, tongue-shaped or absent; anthers nonseptate; outer staminodes rarely present; ovules one to numerous (basal or lateral, rarely apical); orbicules usually present; monocarps indehiscent; seeds not arillate, symmetrical; middle seed integument absent; endosperm ruminations usually spiniform, sometimes lamelliform or irregular; basic chromosome number x = 8 or 9.

Dendrokingstonia Rauschert is placed in Malmeoideae even though we did not include it in our analyses. Analyses by Chaowasku et al. (2012) put it in this subfamily, in a moderately supported position sister to clade X (Fig. 1C). The genus Fenerivia is absent from our analyses. Ten species that were until recently included in Polyalthia have been transferred to Fenerivia by Saunders et al. (2011). Oncodostigma Diels is included in Malmeoideae, but it may not be worthy of recognition: van Heusden (1994) brought the then recognized Oncodostigma spp. into synonymy under Meiogyne Miq., but treated the type species Oleptoneura Diels as a ‘dubious name’ as the holotype appeared to be of mixed origin; the generic name was therefore never formally reduced to synonymy with Meiogyne. Okada (1996) subsequently described Oncodostigma microflorum H.Okada, in effect revitalizing the generic name. The taxonomy of Oncodostigma needs further clarification and, for now, we list the name in our classification. Fitzalania F.Muell. has been accommodated here even though a proposal has been published to give priority to the name Meiogyne (Chaowasku, Zijlstra & Chatrou, 2011).

Circumscription of tribes in Malmeoideae

Piptostigmateae Chatrou & R.M.K.Saunders, tribus nov.—TYPE: Piptostigma Oliv.

Trees; indument of simple hairs; bracts present; inflorescences rhipidiate, terminal or axillary; flowers bisexual or androdioecious; carpels free in flower and fruit; placentation basal or lateral and uni- or biseriate; monocarp abscission at base or apex of stipe; aril absent; ruminations spiniform, sometimes irregular pegs; endosperm glass-like; pollen monosulcate.

Included genera: Annickia Setten & Maas, Greenwayodendron Verdc., Mwasumbia Couvreur & D.M.Johnson, Piptostigma, Polyceratocarpus Engl. & Diels.

This tribe is equivalent to clade T (Fig. 1C).

Malmeeae Chatrou & R.M.K.Saunders, tribus nov.—TYPE: Malmea R.E.Fr.

Trees; indument of simple hairs; bracts present; inflorescences rhipidiate, terminal or axillary; flowers bisexual or androdioecious; carpels free in flower and fruit; placentation basal, rarely lateral; monocarp abscission at base of stipe; aril absent; ruminations spiniform, irregular pegs or lamelliform; endosperm glass-like or soft; pollen monosulcate.

Included genera: Bocageopsis R.E.Fr., Cremastosperma R.E.Fr., Ephedranthus S.Moore, Klarobelia Chatrou, Malmea, Mosannona Chatrou, Onychopetalum R.E.Fr., Oxandra A.Rich., Pseudephedranthus Aristeg., Pseudomalmea Chatrou, Pseudoxandra R.E.Fr., Ruizodendron R.E.Fr., Unonopsis R.E.Fr..

This tribe is equivalent to clade V (Fig. 1C).

Maasieae Chatrou & R.M.K.Saunders, tribus nov.—TYPE: Maasia Mols, Keßler & Rogstad

Trees; indument of simple hairs; bracts present; inflorescences rhipidiate, axillary; flowers bisexual; carpels free in flower and fruit; placentation basal, rarely ventral; monocarp abscission at base of stipe; aril absent; ruminations spiniform; endosperm glass-like; pollen monosulcate.

Included genus: Maasia.

This tribe is equivalent to clade W (Fig. 1C).

Fenerivieae Chatrou & R.M.K.Saunders, tribus nov.—TYPE: Fenerivia Diels.

Trees; indument of simple hairs; bracts present; flowers solitary, axillary, bisexual; carpels free in flower and fruit; placentation basal; monocarp abscission at base of stipe; aril absent; ruminations spiniform; endosperm glass-like; pollen monosulcate.

Included genus: Fenerivia.

This tribe is not represented in our analyses. Monophyly of Fenerivia has been demonstrated by Saunders et al. (2011), who also found a polytomy comprising the tribes Malmeeae, Maasieae, Monocarpieae together with Miliuseae, and Fenerivieae. This result is confirmed by T. Chaowasku et al. (unpubl. data).

Dendrokingstonieae Chatrou & R.M.K.Saunders, tribus nov.—TYPE: Dendrokingstonia Rauschert.

Trees; indument of simple, rarely stellate, hairs; bracts present; inflorescences rhipidiate, axillary, or flowers solitary; flowers bisexual; carpels free in flower and fruit; placentation lateral, uni- or biseriate; monocarp abscission at base of stipe; aril absent; ruminations lamelliform; endosperm soft; pollen monosulcate.

Included genus: Dendrokingstonia.

Monocarpieae Chatrou & R.M.K.Saunders, tribus nov.—TYPE: Monocarpia Miq.

Trees; indument of simple hairs; bracts present; inflorescences rhipidiate, terminal; flowers bisexual; carpels free in flower and fruit; placentation lateral; monocarp abscission at base of stipe; aril absent; ruminations spiniform; endosperm glass-like; pollen monosulcate.

Included genus: Monocarpia.

Miliuseae Hook.f. & Thomson, Fl. Ind. 1: 147. 1–19 Jul 1855—TYPE: Miliusa Lesch. ex A.DC.

Trees; indument of simple hairs, rarely T-shaped hairs; bracts present; inflorescences rhipidiate, terminal or axillary, flowers rarely solitary; flowers bisexual or (andro)dioecious, rarely monoecious; carpels free in flower and fruit; placentation basal or lateral; monocarp abscission at base of stipe; aril absent; ruminations spiniform, irregular pegs or lamelliform; endosperm soft or glass-like; pollen cryptoaperturate/disulculate.

Included genera: Alphonsea Hook.f. & Thomson, Desmopsis, Enicosanthum Becc., Fitzalania, Haplostichanthus F.Muell., Marsypopetalum Scheff., Meiogyne, Miliusa, Mitrephora (Blume) Hook.f. & Thomson, Neo-uvaria Airy Shaw, Oncodostigma, Orophea Blume, Phaeanthus Hook.f. & Thomson, Phoenicanthus Alston, Platymitra Boerl., Polyalthia, Popowia, Pseuduvaria Miq., Sageraea Dalzell, Sapranthus, Stelechocarpus (Blume) Hook.f. & Thomson, Stenanona, Tridimeris, Trivalvaria (Miq.) Miq., Woodiellantha Rauschert.

This tribe is equivalent to clade X excluding Monocarpia (Fig. 1C).


The phylogenetic tree presented here represents a significant improvement in both the generic representation and resolution when compared with previous work on Annonaceae. Previous informal classifications failed to find general acceptance and were unstable with the addition of new data or the application of different methods of analysis. The knowledge of the phylogenetics of Annonaceae has now reached the point at which it is possible to define a formal classification, with the four subfamilies and 14 tribes treated here, which is likely to be stable in the face of new data. Such a classification is warranted as an aid to communication in this important and widely distributed tropical plant family.


We gratefully acknowledge critical comments by Gea Zijlstra and Tanawat Chaowasku on an earlier draft, translations of Latin and German literature by Lubbert Westra and Daniel Thomas, respectively, and editorial support from Mike Fay and Hassan Rankou. This work has been supported in part by the Netherlands Organization for Scientific Research (NWO, grant S85-324) to LWC and the Innovational Research Incentives Scheme (VENI, nr. 863.09.017; NWO-ALW) to RHJE.