Systematic Botany (2006), 31(1): pp. 160–170 q Copyright 2006 by the American Society of Plant Taxonomists Phylogenetic Relationships within the Subfamily Sterculioideae (Malvaceae/Sterculiaceae-Sterculieae) Using the Chloroplast Gene ndhF PETER WILKIE,1,5 ALEXANDRA CLARK,1 R. TOBY PENNINGTON,1 MARTIN CHEEK,2 CLEMENS BAYER,3 and CHRIS C. WILCOCK4 Royal Botanic Garden Edinburgh, Inverleith Row, Edinburgh EH3 5LR, U.K.; 2 Royal Botanic Garden Kew, Richmond, Surrey TW9 3AE, U.K.; 3 Palmengarten, Siesmayerstraße 61, 60323 Frankfurt, Germany; 4 Cruickshank Building, St Machar Drive, University of Aberdeen, Aberdeen AB24 3UU, U.K. 5 Author for correspondence (p.wilkie@rbge.org.uk) 1 Communicating Editor: Gregory M. Plunkett ABSTRACT. A parsimony analysis of ndhF nucleotide sequences representing 24 species and 13 genera of Sterculioideae strongly supports (100% bootstrap) the monophyly of the group. Within the Sterculioideae clade four major clades are recognized with good bootstrap support but relationships among them are not resolved. This analysis suggests the recognition of Argyrodendron as separate from Heritiera, supports Acropogon as separate from Sterculia, and Tarrietia as part of Heritiera. The current circumscriptions of Hildegardia and Firmiana are not supported. The use of fruit characters in the delimitation of genera within Sterculioideae may not be appropriate in some cases and other morphological characters need to be found. The family name Sterculiaceae is taken from the type genus Sterculia L., named after the Roman god of privies, Sterculius, and refers to the foul smelling flowers of some species (Heywood 1993). The family, as traditionally recognized, is composed chiefly of tropical trees and shrubs with a few herbaceous and climbing species. It includes the economically important genera Cola Schott & Endl. and Theobroma L. and commonly cultivated genera such as Dombeya Cav., Fremontodendron Coville, Kleinhovia L., and Brachychiton Schott & Endl. It is closely related to Tiliaceae, Malvaceae, and Bombacaceae (Alverson et al. 1998; Fay et al. 1998; Bayer et al. 1999) and contains a considerable range of morphological variation (Cronquist 1981). In trying to bring some order to this morphologically heterogeneous family some authors have proposed recognition of tribes (Schumann 1890, 1893; Edlin 1935a, 1935b; Hutchinson 1967; Takhtajan 1997) or separate families (Edlin 1935a, 1935b). Recent phylogenetic analysis of the plastid genes atpB and rbcL in the order Malvales has led to the suggestion that only the family Malvaceae is monophyletic, that the core Malvales (Sterculiaceae, Tiliaceae, Bombacaceae, and Malvaceae) should be considered a single large family, Malvaceae s.l., and that nine subfamilies be recognized based on molecular, morphological, and biogeographical data (Bayer et al. 1999). These findings have been supported by Alverson et al. (1999) using a third plastid gene, ndhF. The newly proposed division of core Malvales recognizes genera traditionally placed in tribe Sterculieae (Table 1) in subfamily Sterculioideae. The remaining genera, those of Edlin’s proposed Byttneriaceae, are shared between subfamilies Byttnerioideae, Dombeyoideae, Helicteroideae, and Bombacoideae (Bayer and Kubitzki 2003). The genera of Sterculioideae have been widely recognised as a homogenous group (Schott and Endlicher 1832; Endlicher 1850; Brown 1844; Bentham 1862; Baillon 1872; Edlin 1935a, 1935b; Takhtajan 1997). They share several characters such as apetalous flowers, 6 petaloid sepals, secondary apocarpy and mostly unisexual flowers (Edlin 1935a, 1935b; Judd and Manchester 1997; Bayer et al. 1999; Alverson et al. 1999; Bayer and Kubitzki 2003). The presence of sheath cells (upright ray cells that form a sheath about the smaller cells of a multiseriate ray) in Sterculioideae are unique within Malvaceae s.l. (Chattaway 1932, 1938). Within the subfamily the genera have a complex history of synonymy and many species have been placed in the type genus Sterculia (Table 1). As such the composition of the subfamily has varied largely depending on the subdivision of Sterculia. Taylor (1989) suggested that the first clearly monophyletic group including this genus was the tribe Sterculieae itself. Kostermans revised most of the southeast Asian genera of the subfamily, published important observations (Kostermans 1953a, 1956, 1957, 1959a, 1959b, 1988), and brought order to the genera mainly by the reduction of species and genera into synonymy. His circumscription of genera relied heavily on fruit characters as did his ideas of the evolutionary relationships of genera, many of which reflect the earlier views of Ridley (1930). However, some consider that he placed too much importance on fruit characters and undervalued the importance of floral characters in delimiting genera (Dorr and Barnett 1990). The large-scale studies of Malvaceae s.l. by Bayer et al. (1999) and Alverson et al. (1999) sampled only five of the 13 genera in the proposed Sterculioideae. This 160 161 m m m m m Tarrietieae Tarrietieae m m m m m m m m m m m m m m m m Tarrietieae Mansonieae m m m Heritiera*m m m m m m Sterculia/m Sterculia*m Sterculia*m Sterculia*m Tarrietia* m m m m m m Sterculia# m Sterculia# Sterculia# m Sterculia*m Sterculia*m Sterculia*m m m m m m m m m Sterculia# m Sterculia/m Sterculia/m Sterculia/m m Sterculia/m Sterculia/m m Sterculia/m Sterculia/m Sterculia/m m m Firmiana*m m m m Firmiana*m m Tarrietia* m Tarrietieae Sterculia1m Sterculia1m Sterculia1m m Sterculia1m Sterculia1m Sterculia1m Tarrietieae m m m m Sterculia1m m m m m m m m m m m m m m m m Sterculia (1753) Cola (1832) Acropogon (1906) Franciscodendron (1987) Heritiera (1789) Firmiana (1786) Scaphium (1832) Pterocymbrum (1844) Octolobus (1869) Pterygota (1832) Hildegardia (1832) Brachychiton (1834) Argyrodendron (1858) De Candolle Schott & 1824 Endlicher 1832 Brown 1844 Bentham 1862 Baillon 1872 Masters 1868/1875 King 1891 Schumann 1890/1893/1900 Ridley 1922 Edlin 1935a/1935b Hutchinson 1967 Takhtajan 1997 Bayer & Kubitzki 2003 WILKIE ET AL.: PHYLOGENY OF STERCULIOIDEAE Genus (desc. date) TABLE 1. The genera of Sterculioideae as recognized in 11 influential classifications. A box with m indicates that the genus was included in the tribe Sterculieae or subfamily Sterculioideae in the classification; a box without m indicates that the genus was not mentioned in the classification. If the genus was placed in another tribe or subfamily this tribe or subfamily is named in the box. A slash (/) indicates the genus was regarded as a section of the genus named in the box, ‘‘1’’ indicates the genus was regarded as a subgenus of the genus named in the box, ‘‘*’’ indicates the genus is recognised as a synonym under the genus name in the box, and ‘‘#’’ indicates the genus is recognised as a section under the genus name in the box. The genera of de Candolle (1824) are under the family Byttneriaceae, Baillon (1872) the family Malvaceae (including Bombacaceae and Sterculiaceae but not Tiliaceae), Takhtajan (1997) the subfamily Sterculioideae in Sterculiaceae and Bayer & Kubitzki (2003) the subfamily Sterculioideae in Malvaceae s.l. Edlin (1935a, 1935b) split the genera of Sterculioideae into two tribes Sterculineae and Tarrietieae. Genera misplaced under the tribe/ subfamily or have widely been accepted as synonymous with other genera have been omitted from the table. 2006] paper describes a phylogenetic analysis of ndhF nucleotide sequences for 24 species in all 13 currently recognized genera of Sterculioideae. The resulting phylogenetic hypotheses will be used to investigate the circumscription and phylogenetic position of Sterculioideae, infer relationships among the genera of Sterculioideae, investigate the monophyly of individual genera, and examine hypotheses of fruit evolution. MATERIALS AND METHODS Taxon Sampling. For this study 22 new ndhF sequences of Sterculioideae were produced (Table 2). These new sequences include eight genera for which ndhF sequence data were not previously available and constitute at least one exemplar of each genus placed in the tribe Sterculieae or subfamily Sterculioideae using previous classifications (Candolle 1824; Schott and Endlicher 1832; Brown 1844; Bentham 1862; Masters 1868, 1875; King 1891; Schumann 1890, 1893, 1900; Ridley 1922; Edlin 1935a, 1935b; Kostermans 1953a, 1953b, 1957, 1959a, 1959b; Hutchinson 1967; Takhtajan 1997; Bayer and Kubitzki 2003). A matrix of 160 taxa of Malvaceae s.l. was compiled using these new sequences, plus complete ndhF sequence data available for Malvaceae s.l. from GenBank (Alverson et al. 1999; Whitlock et al. 2001; Whitlock et al. 2003; Nyffeler and Baum 2000; Pfeil et al. 2002; Seelanan et al. 1997; Cronn et al. 2003; Small 2004), and the outgroups Muntingia L., Rhopalocarpus Bojer, and Cochlospermum Kunth. The outgroups were chosen because they were determined to belong to the expanded Malvales, but are clearly placed outside Malvaceae s.l. (Alverson et al. 1999). A second matrix of 31 taxa of Sterculioideae was compiled using our 22 new sequences, five from GenBank (Alverson et al. 1999) plus the outgroups Ochroma Sw., Bombax L., Ceiba Mill. and Pachira Aubl. The outgroups were selected as they belong to Malvatheca (Bombacoideae and Malvoideae), the putative sister group to Sterculioideae (Nyffeler et al., in press). DNA Isolation, Amplification and Sequencing. Total genomic DNA was extracted from silica gel-dried leaf material collected in the field, and in a few cases from recently collected heat-dried herbarium specimens (Table 2). DNA extraction followed the CTAB protocol modified from Doyle and Doyle (1990) except for the problematic Pterygota alata for which a CTAB extraction followed by a caesium chloride purification was undertaken. Specimens with degraded DNA were left in the freezer at 2208C for 2– 3 weeks to improve the DNA yield. For taxa that provided DNA but did not produce a PCR product, a QIAGEN DNA extraction kit (QIAGEN Ltd., Crawley, West Sussex, U.K.) was used to provide cleaner DNA by removing proteins and other carbohydrates that may have inhibited the PCR. Universal ndhF primers from Olmstead et al. (1993) were used to amplify two overlapping fragments of roughly 1000bp: ndhF1 and ndhF2. Two main primer combinations used were: 1F to 972R (ndhF1) with 803F to 2112R (ndhF2) or 252F to 1318R (ndhF1) with 972F to 2112R (ndhF2). For taxa that were difficult to amplify, different combinations were used: Scaphium linearicarpum and Scaphium macropodum (sample A), primers252F to 1318R with 803F to 2112R; Pterygota alata primers252F, 1318F, 1603F, 803R, 1318R, 1829R, 1955R and 2112R. PCR was performed on a MJ Research PTC-200 Thermal Cycler in 4 3 25mL reactions using reagents from Bioline (U.K.). Reactions contained: 2.5mL of 10X NH4 buffer, 2.5mL of 2mM dNTPs, 1.25mL of 50mM MgCl2, 0.75mL of each 10mM primer, 0.625 units of Biotaq polymerase and 1 or 2mL of DNA. Amplification for ndhF regions 1 and 2 were carried out using 1 cycle at 958C for 3 min, 30 cycles of 1 min at 958C, 1 min at 508C, 1 min at 728C and 1 cycle at 728C for 7 min. Prior to sequencing, the 4 3 25mL PCR reactions were pooled together and purified using QIAquickt PCR Purification Kit (QIAGEN Ltd.). Purified PCR products were sequenced using the dideoxy-chain termination method in 20mL reactions containing 4mL Thermo Sequenase II (Amersham Pharmacia, U.K.), 0.5mL primer (10mM), 13.5L dH2O and 2.0 mL of purified PCR product. Sequencing re- 162 [Volume 31 SYSTEMATIC BOTANY TABLE 2. Accessions of ndhF sequences used in the Sterculioideae phylogenetic analysis (new sequences are highlighted in bold), including species authorities, collector, collector number, and GenBank reference. All reference voucher material is held in the herbarium of the Royal Botanic Garden Edinburgh (E) unless otherwise indicated in parentheses. Acropogon bullatus (Pancher & Sebert) P. Morat, Gardner et al. 2069: AY795596. Acropogon dzumacensis (Guillaumin) P. Morat, Fambart et al. 14 (NOU): AY795595. Argyrodendron peralatum (Bailey) Edlin ex J. H. Boas, P. Wilkie PW305: AY795597. Bombax buonopozense Beauv., Alverson s.n. (WIS): AF111726. Brachychiton acerifolius F. Muell. (Sample A) H. Lim and S. Gale H1S1: AY795594 (Sample B) Alverson 4009 (WIS): AF111748. Ceiba rosea (Seem.) K. Schum., Alverson 2185 (WIS): AF111731. Cola acuminata Schott & Endl., Kubitzki 93-12 (HBG): AF111759. Firmiana malayana Kosterm., P. Wilkie PW164: AY795587. Firmiana simplex (L.) W. F. Wight, P. Wilkie PW469: AY795589. Franciscodendron laurifolium (F. Muell.) B. Hyland & Steenis, P. Wilkie PW312: AY795598. Heritiera aurea Kosterm. (Sample A) P. Wilkie PW37: AY795583 (Sample B) P. Wilkie PW43: AY795584. Heritiera elata Ridl., P. Wilkie PW131: AY795585. Heritiera littoralis Dryand., Alverson s.n. (WIS): AF111744. Heritiera simplicifolia (Mast.) Kosterm., P. Wilkie PW 145: AY795582. Hildegardia barteri (Mast.) Kosterm., Hill 2519 (FTG): AF111754. Hildegardia populifolia Schott & Endl., M. Cheek 11480 (K): AY795588. Ochroma pyramidale (Cav. ex Lam.) Urb., Alverson and Rubio 2246 (WIS): AF111740. Octolobus spectabilis Welw., Frimodt-Møller et al. TZ540 (K): AY795586. Pachira aquatica Aubl., Alverson et al. 2154 (WIS): AF111732. Pterocymbium tinctorium Merrill, P. Wilkie PW125: AY795590. Pterygota alata (Roxb.) R. Br., F. Effendi et al. FRI 48106: AY795599. Scaphium linearicarpum (Mast.) Pierre, P. Wilkie PW2: AY795592. Scaphium macropodum (Miq.) Beumée ex K. Heyne (Sample A) P. Wilkie PW16: AY795593 (Sample B) P. Wilkie PW111: AY795591. Sterculia balanghas L., D. Middleton 392 (BKF): AY795580. Sterculia coccinea Jack, P. Wilkie PW169: AY795579. Sterculia parviflora Roxb. ex G. Don., P. Wilkie PW120: AY795581. Sterculia stipulata Korth., M. Mendum MM34: AY795578. Sterculia tragacantha Lindl., Alverson 4011 (WIS): AF111747. action conditions were: 25 cycles of 968C for 10 sec, 508C for 5 sec and 608C for 4 min. For Pterygota alata, Sterculia balanghas and Acropogon bullatus a 10mL reaction containing 4 mL DTCS Quickstart mix (Beckman Coulter), 1 mL primer (10mM) and 5 mL of purified PCR product was used. Sequencing reaction conditions were: 35 cycles of 968C for 20 sec, 508C for 20 sec and 608C for 4 min. All sequenced products were purified by ethanol precipitation to remove unincorporated labeled ddNTPs and excess salts following the protocol of the manufacturer. DNA Sequence Analysis. All sequences were read on an ABI Prismy 377 DNA Sequencer, then edited manually using Sequence Navigator 1.0 (Applied Biosystems Inc.) except Pterygota alata, Sterculia balanghas and Acropogon bullatus, which were read on a Beckman Coulter CEQy 8000 Sequencer, then edited manually using SEQUENCHER V.4.1.4 (Gene Codes Corporation 2002). In the Sterculioideae matrix 1.28% of cells were scored as missing data. Phylogenetic Analysis. Two sets of phylogenetic analyses were undertaken. The first used the Malvaceae s.l. matrix to evaluate the placement and monophyly of Sterculioideae within the expanded Malvaceae. The second used the smaller Sterculioideae matrix to examine phylogenetic relationships within this subfamily. Nucleotide sequences were analysed using parsimony. Analyses were performed using PAUP* 4.0b10 (Swofford 2001). For the Malvaceae s.l. matrix, heuristic search options included an initial heuristic search using 1,000 random addition replicates, tree-bisectionreconnection (TBR) branch swapping with steepest decent and MULTREES not activated. The final 36 bases of the matrix were excluded from the analyses as sequencing reactions were too weak to provide sufficient signal for unambiguous base calling. Trees produced from this analysis were used as the basis for a further heuristic search, activating steepest descent and MULTREES and saving 10,000 trees which is considered to be sufficient to capture all topological variation (e.g., Sanderson and Doyle 1993). To limit the number of trees found, the ‘‘collapse branches if minimum length is 0’’ was invoked. For the Sterculioideae matrix, Branch and Bound analyses were undertaken using furthest addition sequence with stepwise addition and MULTREES selected and keeping minimal trees only. As an approximate guide to clade support, bootstrap values (Felsenstein 1985) were calculated for both sets of analyses using 10,000 bootstrap replicates with one random addition sequence replicate, and TBR branch swapping. Steepest decent and MULTREES were not activated. Bias introduced by manual alignment was evaluated with sensitivity analyses (cf. Wheeler 1995; Whiting et al. 1997; Beyra and Lavin 1999). Potentially phylogenetically informative areas of gaps were coded as separate characters, following the simple gap cod- ing procedure described by Simmons and Ochoterena (2000). Sensitivity analyses were subject to the same heuristic or Branch and Bound search options. Data matrices are available on TreeBASE (study accession S1289). Fruit Morphology Analysis. Patterns of fruit evolution were assessed in MacClade 3.05 (Maddison and Maddison 1992) by mapping characters onto the most parsimonious trees produced from the Sterculioideae matrix analyses. Six characters were conceptualized: fruit structure, fruit type, fruit texture, ovules per locule, seed position and dispersal mechanism (Appendix 1). Morphological data (Appendix 2) were obtained from examination of herbarium specimens held in Edinburgh (E) and other herbaria (BO, K, KEP, L, SAN, SAR, and SING). The ‘‘trace characters’’ function in MacClade was selected together with the ‘‘show all most parsimonious states at each node’’ option. RESULTS The aligned sequences used for the Malvaceae s.l. analyses are 2,237 base pairs (bp) in length. In this data set, there were 33 insertion or deletion (indel) events, 21 of which were potentially phylogenetically informative. Despite improved taxon sampling, parsimony analyses of the Malvaceae s.l. matrix (tree not shown) did not provide greater resolution of subfamily relationships than those previously published (Alverson et al. 1999; Bayer et al. 1999; Nyffeler and Baum 2000; Whitlock et al. 2001; Nyffeler et al., in press). All genera currently assigned to Sterculioideae fall within the Sterculioideae clade, which is strongly supported with 100% bootstrap. The sister lineage is not resolved in our analyses and Sterculioideae form a polytomy with Tilioideae, Dombeyoideae, Brownlowioideae, Helicteroideae, and a Bombacoideae/Malvoideae clade. The aligned sequences in the more restricted Sterculioideae data set were 2,156 bp in length. In these analyses there were six indels, two of which were potentially phylogenetically informative. Parsimony analysis of the Sterculioideae matrix provided 110 informative sites and produced 477 equally most-parsimonious trees with minimal length of 318 steps, consis- 2006] WILKIE ET AL.: PHYLOGENY OF STERCULIOIDEAE 163 FIG. 1. Strict consensus of 477 equally most parsimonious trees (318 steps, CI5 0.759, and RI 5 0.891) produced from the maximum parsimony analysis of the Sterculioideae matrix. Bootstrap values are shown below the branches. tency index (CI) 0.877 (0.759 excluding uninformative characters) and a retention index (RI) 0.891. The strict consensus tree is shown in Fig. 1 and one of the mostparsimonious trees showing branch lengths in Fig. 2. Sensitivity analyses for both the Malvaceae s.l. and Sterculioideae analyses produce identical topologies within Sterculioideae. Within the Sterculioideae clade, four major clades are resolved, most of them with good bootstrap support (. 62%). These are the Cola clade (99% bootstrap) comprising Cola, Octolobus Welw., Pterygota Schott & Endl., Hildegardia Schott & Endl., Firmiana Marsili, Pterocymbium R. Br., and Scaphium Schott & Endl.; the Brachychiton clade (63% bootstrap), comprising Brachychiton, Argyrodendron F. Muell., Franciscodendron B. Hyland & Steenis and Acropogon Schltr.; the Sterculia clade 164 [Volume 31 SYSTEMATIC BOTANY FIG. 2. Phylogram of one of the 477 most-parsimonious trees found in the 31-taxon, equally weighted parsimony analyses of the Sterculioideae matrix. Branch lengths are shown above the branches. (100% bootstrap), comprising only Sterculia; and the Heritiera clade (100% bootstrap), comprising only Heritiera Aiton (Fig. 1). Apocarpous flowers provide a synapomorphy for the Sterculioideae clade in all 477 trees. Leathery carpels are traced as plesiomorphic in all trees. Fibrous fruits have evolved at least once (and possibly twice) in the Heritiera clade and papery carpels once in the Cola clade (Fig. 3c). Fruit with locules containing two or more ovules are traced as plesiomorphic in all trees. Single-seeded locules have evolved once in the Cola clade, once in the Heritiera clade, and possibly inde- pendently a third time in the Brachychiton clade (Fig. 3d). Wind dispersal is traced as plesiomorphic in all trees with bird dispersal evolving up to four times within the Sterculioideae clade (Fig. 3b). Our analyses did not provide a clear plesiomorphic state for fruit type or seed position in Sterculioideae. DISCUSSION Monophyly of the Subfamily Sterculioideae. This study strongly supports (100% bootstrap) the monophyly of Sterculioideae as proposed by Bayer et al. 2006] WILKIE ET AL.: PHYLOGENY OF STERCULIOIDEAE 165 FIG. 3. Optimization of fruit characters on one of 477 trees produced from the maximum parsimony analysis of the Sterculioideae matrix. A. Fruit type. B. Dispersal mechanisms. C. Fruit texture. D. Ovules per locule. (1999). These findings are based on a complete sampling of genera traditionally placed in the tribe Sterculieae sensu Schott and Endlicher (1832) and are congruent with earlier studies using ndhF (Alverson et al. 1999) and atpB plus rbcL (Bayer et al. 1999) to investigate the Malvales, although those studies sampled only five genera of Sterculioideae. The genera of other tribes in the traditional Sterculiaceae are shared between the subfamilies Byttnerioideae, Dombeyoideae, Helicteroideae, and Bombacoideae (Bayer and Kubitzki 2003). Mansonia and Triplochiton, which have been most closely associated with Sterculieae because of their secondary apocarpy (Schumann 1900; Prain 1905), are both placed in the subfamily Helicteroideae (Bayer et al. 1999; Nyffeler and Baum 2000; Wilkie, unpubl. data). Morphologically Sterculioideae are supported by apetalous flowers, petaloid sepals, secondary apocarpy, androgynophores, and mostly unisexual flowers. The occurrence of apetaly, androgynophores, and unisexual flowers within Malvaceae s.l. is not unusual but their combination is unique to Sterculioideae. In this study, using all available ndhF data for Malvales the sister group to Sterculioideae could not be identified. Candidates are Tilioideae, Dombeyoideae, Brownlowioideae, Helicteroideae, and Malvatheca (Bombacoideae/Malvoideae), all of which form a polytomy with Sterculioideae. A recent combined analysis of the Malvadendrina clade (Malvaceae s.l. exclud- 166 SYSTEMATIC BOTANY ing Byttnerioideae and Grewioideae) using atpB, matK, and ndhF sequences (Nyffeler et al., in press) is also inconclusive in this regard. It places Sterculioideae as sister to Malvatheca, but with weak branch support (55% bootstrap, 86% posterior probability). Relationships of Genera Within the Subfamily Sterculioideae. Within Sterculioideae, four main clades are identified, although the relationships between them remain unresolved. COLA CLADE. A close relationship between Cola and Octolobus was suggested by Cheek and FrimodtMøller (1998). Edlin (1935a, 1935b) hypothesised that Pterygota was related to these genera. These hypotheses are supported in our analyses with the three genera forming a subclade that is sister to the remaining genera of the clade. However, the precise relationship between these genera requires better sampling especially of the estimated 125 species of Cola. Within Cola, priority sampling should include species in the infrageneric groupings originally proposed by Schumann (1900) and since modified, most recently by Bodard (1962). Highest priority should probably be given to the three such groupings considered by Bodard to merit generic distinction, although not generally accepted as such by the taxonomic community, these being Chlamydocola K. Schum., Ingonia Pierre and Parvenosemenocola M. Bodard nom. nud. Such extended sampling would greatly improve our understanding of inter and infra generic relationships within Cola. Both Cola and Octolobus are restricted to Africa while Pterygota, although occurring in Africa, also has species in both southeast Asia and South America. Morphologically, the leathery follicle and ring-like arrangement of the anthers in all three genera serves as a synapomorphy for the subclade. Both characters exhibit homoplasy within the Sterculioideae, with leathery follicles found in the Brachychiton and Sterculia clades and ringed anthers in Acropogon (Brachychiton clade) and Pterocymbium (Cola clade). Our phylogenetic estimate questions the generic delimitations of Hildegardia and Firmiana and does not support Kostermans’ (1954, 1956) use of follicle dehiscence versus indehiscence to separate genera. The genera are no doubt closely related, Hildegardia barteri (Mast.) Kosterm. having been moved from Firmiana to Hildegardia by Kostermans (1954). Dorr and Barnett (1990) suggested the African and Malagasy species of Hildegardia form a natural group, all having long, tubular or infundibuliform calyces with relatively short calyx lobes, and lacking leaves in flower. Our cladistic analysis supports this view. Hildegardia populifolia, which does not have these characters, is placed in a separate clade. In Firmiana, floral morphology supports the clades produced by our molecular cladistic analyses. Firmiana malayana and Hildegardia barteri both have tubular calyces and form a separate clade from [Volume 31 Firmiana simplex and Hildegardia populifolia, which have deeply lobed calyces. In Asia, some authors (Ridley 1934; Hsue 1984) have placed species with tubular calyces in the genus Erythropsis Lindl. ex Schott & Endl. but made no reference to the genus Hildegardia. Greater sampling from the eight species of Hildegardia and 12 species of Firmiana is needed before any firm conclusion can be drawn on generic boundaries/delimitation but our analyses suggest that these genera may have to be re-circumscribed. The close relationship of Pterocymbium and Scaphium has been well documented (Masters 1875; Merrill 1929; Kostermans 1953a, 1973) and is supported in our analyses (87% bootstrap). Both genera have papery, early-dehiscent follicles with a single basal seed. The monophyly of Scaphium is not supported by our analyses because of the inclusion of Pterocymbium. Greater sampling of Scaphium (c. 9 species) and Pterocymbium (c. 15 species) is needed to confirm this relationship. Morphologically, Scaphium is distinguished from Pterocymbium by the lack of a spur to the follicle, smaller flower size, and an irregularly clustered anther arrangement. BRACHYCHITON CLADE. The clade comprising Brachychiton, Argyrodendron, Franciscodendron, and Acropogon has moderate bootstrap support (63%) and consists of genera from Australia and New Caledonia. Within this clade, three subclades are recognised, although the relationships among them remain unresolved. The Brachychiton subclade has good bootstrap support (99%) and is distinctive by its hairy exotesta and dark brown, woody follicle (Guymer 1988). The two exemplars of the New Caledonian endemic genus Acropogon form the second subclade with 100% bootstrap support. Our analyses support the work of Schlechter (1906), Morat (1986, 1988) and Morat and Chalopin (2003), who recognised Acropogon as separate from Sterculia because of its (2)-3-(4) carpels, exalbuminous seeds, fleshy cotyledons, and ring arrangement of the stamens. The association of New Caledonian endemics such as Acropogon with predominantly Australian genera is also found from cladistic analyses of DNA sequence data of Maxwellia in the Lasiopetalae (Whitlock et al. 2001). The third subclade unites the Australian genus Argyrodendron to the monotypic Australian genus Franciscodendron. This subclade (100% bootstrap) is supported by their fibrous-papery follicles that are indehiscent to tardily dehiscent, in contrast to the rest of the clade, which has leathery dehiscent follicles. Our analyses refute the suggestion that Franciscodendron is sister to Hildegardia (Hyland and Steenis 1987), and that Franciscodendron is sister to Pterygota (Kostermans 1988). The placement of Argyrodendron in the Brachychiton clade rather than the Heritiera clade supports the classifications of Edlin (1935a, 1935b), Burtt Davy (1937), and Smith (1969). They con- 2006] WILKIE ET AL.: PHYLOGENY OF STERCULIOIDEAE sidered the distinct constriction of the wing near the seed, the 15–20 stamens, and the more or less transverse venation in the fruiting carpel of Argyrodendron as sufficient to recognise the genus as distinct from Heritiera. Wood anatomy supports this distinction, with Argyrodendron having broad bands of parenchyma compared to the narrow bands of Heritiera (Chattaway 1938). The classification of Kostermans (1959a, 1959b), Hyland and van Steenis (1987), and Bayer and Kubitzki (2003), who consider the differences not to merit generic separation, is rejected by our analyses. STERCULIA CLADE. Our analyses of five Asian and African species of Sterculia form a well supported (100% bootstrap) clade. To test the monophyly of this genus a far greater level of sampling is required from this pantropical genus that contains 100–150 species, especially from the poorly sampled Neotropics, but the lack of a subgeneric classification hinders the development of a future sampling strategy. Schumman (1890, 1893) proposed three series within the genus based on leaf characters but this has been found to be an artificial classification (Dorr 2004). Taylor (1989), who studied the South American members of Sterculia, and Tantra (1976), who undertook a regional monograph of the Malesian species, were not convinced the genus formed a natural group but did not propose any subgeneric classification. Potential characters for use in an infrageneric classification include floral morphology, particularly fenestrate flowers that characterise some paleotropical species (Taylor 1989), floral indumentum and foliar indumentum. Chattaway (1938) separated Sterculia into two distinct groups based on wood anatomy and this should also be investigated. HERITIERA CLADE. The four species sampled from Heritiera (c. 30 species) form a clade with 100% bootstrap support. Kostermans (1959a, 1959b) considered Tarrietia Blume a synonym of Heritiera, stating that the presence or absence of endosperm is not consistent or sufficient enough to justify separating them. This view was not shared by van Steenis (1960), who cited differences in leaf venation and indumentum as sufficiently diagnostic. This view was subsequently dismissed by Kostermans (1962). In our analyses the placement of Heritiera simplicifolia (formally in Tarrietia) within this clade tentatively supports, albeit with limited sampling, the view of Kostermans (1959a, 1959b, 1962). The wood anatomy of Heritiera and Tarrietia is similar but clearly differs from all other related genera. This led Chattaway (1932) to suggest that they be excluded from the tribe Sterculieae. This is not supported by our analyses, nor is the proposal by Edlin (1935a, 1935b) to recognise both Heritiera and Argyrodendron in a subtribe of their own. Fruit Evolution in Sterculioideae. Ridley (1930) postulated an evolutionary sequence in Sterculioideae from the many-seeded, leathery dehiscent follicles of 167 Sterculia (bird dispersed), through the single, basalseeded papery dehiscent follicle of Scaphium and Pterocymbium (wind-dispersed), and to a reduction of the wing and enclosure of the seed in the indehiscent carpel of Heritiera. He also postulated the origin of Pterygota, with its many winged seeds and leathery dehiscent follicles, from Sterculia. Kostermans (1959a, 1959b) largely followed these hypotheses, speculating that Sterculia is primitive, and that evolution followed a sequence from Sterculia to Firmiana (papery follicle with 2–4 seeds along the margin) to Scaphium and Pterocymbium to Hildegardia (indehiscent papery follicle with one basal seed) and finally to Heritiera (indehiscent fibrous follicle with a single basal seed). In some cases elements of Ridley’s (1930) and Kostermans’ (1959a, 1959b) predictions are confirmed by our analyses. For example, papery fruits are derived (Fig. 3C), as are single seeds (Fig. 3D). However, some hypotheses are refuted. The presence of wind dispersal in our outgroup taxa implies that it is likely that bird dispersal is in fact derived in Sterculioideae (Fig. 3B), and the proposed evolution from a papery follicle to a fibrous one is not supported (Fig. 3C). Unfortunately the lack of resolution amongst the major clades of Sterculioideae (Fig. 1) means that the plesiomorphic character state in Sterculiodeae is ambiguous for many characters. Despite this, character optimization on more apical branches permits a clear evaluation of some of the hypotheses of Kostermans and Ridley. Although they were in many cases correct in their assessment of plesiomorphy and apomorphy, in most cases the sequence of character state acquisition is far more complex than their scenarios envisage, as evidenced by the high homplasy of the characters. For example, indehiscence is derived, but evolved as many as five times (Fig. 3A). Future Directions. Although our analyses recognize well-supported clades within Sterculioideae, the relationship of many clades remains unresolved. Future studies should focus on obtaining additional exemplars from these clades to provide better resolution within the subfamily. Additional exemplars from the large genera Sterculia and Cola across their range, and a more complete sampling of Hildegardia, Firmiana, Scaphium, and Pterocymbium are especially critical. Additional DNA markers may also increase resolution. In addition to increased molecular data, future studies must also focus on the morphological diagnosability of monophyletic groups within Sterculioideae. Fruit characters have long played an important part in generic delimitation in Sterculioideae. Our analyses indicate that in the case of Hildegardia and Firmiana fruit characters may not help diagnose monophyletic groups and that other morphological characters will have to be found if these genera are to be maintained as currently circumscribed. 168 SYSTEMATIC BOTANY ACKNOWLEDGEMENTS. For help in the field, institutional support and stimulating discussion we would like to thank the following: in Peninsular Malaysia, staff at the Forestry Research Institute Malaysia, especially L. G. Saw, E. Soepadmo, R. Chung, and S. Kamarudin, and from the Universiti Malaya, K. M. Wong. From the Sandakan herbarium, Sabah R. Ong, J. Sugau, J. Pereira, and S. P. Lim. From the Forest Research Centre, Sarawak H. S. Lee, A. Mohtar, L. A. Julaihi, J. Sang, D. Bibian, and T. Endela. The staffs of IRD, Nouméa, Nouvelle-Calédonie, CSIRO, Australia (in particular B. Gray, B. Hyland, and R. Elick), Harvard Herbaria (in particular D. Middleton and E. Wood), and the Bangkok Forest Herbarium (especially K. Chayamarit) are thanked for their invaluable help in obtaining material. In Edinburgh M. Kranitz, P. Hollingsworth, M. Hollingsworth, M. Gardner, G. Bramley, S. Neale, and B. Mackinder are thanked for providing material and Vanessa Plana and Elspbeth Haston are thanked for help with the analyses. The Royal Botanic Garden Edinburgh is supported by the Scottish Executive Environment and Rural Affairs Department, for which acknowledgement is gratefully given. LITERATURE CITED ALVERSON, W. S., K. G. KAROL, D. A. BAUM, M. W. CHASE, S. M. SWENSEN, R. MC COURT, and K. J. SYTSMA. 1998. Circumscription of the Malvales and relationships to other Rosidae: evidence from rbcL sequence DNA. American Journal of Botany 85: 876–887. ———, B. A. WHITLOCK, R. NYFFELER, C. BAYER, and D. BAUM. 1999. Phylogeny of the core Malvales: evidence from ndhF sequence data. American Journal of Botany 86: 1474–1486. BAILLON, H. 1872. Histoire des Plantes 4: 57–160. Paris: Librairie Hachette & Cie. BAYER, C. and K. KUBITZKI. 2003. Malvaceae. Pp. 225–311 in The families and genera of vascular plants, flowering plants, dicotyledons 5, eds. K. Kubitzki and C. Bayer. Berlin: Springer Press. ———, M. F. FAY, A. Y. DE BRUIJN, V. SAVOLAINEN, C. M. MORTON, K. KUBITZKI, W. S. ALVERSON, and M. W. CHASE. 1999. Support for an expanded family concept of Malvaceae within a recircumscribed order Malvales: a combined analysis of plastid atpB and rbcL DNA sequences. Botanical Journal of the Linnean Society 129: 267–303. BENTHAM, G. 1862. Sterculiaceae. Pp. 214–228 in Genera Plantarum, eds. G. Bentham, and J. D. Hooker. London: Reeve and Co. BEYRA, M. and M. LAVIN. 1999. Monograph of Pictetia (Leguminosae-Papilionoideae) and review of the Aeschynomeneae. Systematic Botany Monographs 56: 1–93. BODARD, M. 1962. Contribution à l’étude systématique du genre Cola en Afrique Occidentale. Annales de la Faculté des Sciences Université de Dakar 7: 1–187. BROWN, R. 1844. Pterocymbium, with observations on Sterculieae, the tribe to which it belongs. London: Privately published. BURTT DAVY, J. 1937. Notes on Argyrodendron F. Muell. Tropical Woods 51: 19–20. CANDOLLE, A. P. DE. 1824. Sterculia. Pp. 481–483 in A. P. de Candolle (ed.), Prodromus Systematis Naturalis Regni Vegetablis vol. 1. Paris: Victor Masson. CHATTAWAY, M. M. 1932. The wood of the Sterculiaceae. I. Specialisation of the vertical wood parenchyma within the subfamily Sterculieae. New Phytologist 31: 119–132. ———. 1938. The wood anatomy of the Sterculiaceae VII. Philosophical Transactions of the Royal Society of London. Series B 228: 313–365. CHEEK, M. 2002. A new species of Cola (Sterculiaceae) from Usambara mountains, Tanzania. Kew Bulletin 57: 417–422. ——— and C. FRIMODT-MøLLER. 1998. The genus Octolobus (Sterculiaceae) new to East Africa. Kew Bulletin 53: 682. CRONN, R., R. L. SMALL, T. HASELKORN, and J. F. WENDEL. 2003. [Volume 31 Cryptic repeated genomic recombination during speciation in Gossypium gossypioides. Evolution 57: 2475–2489. CRONQUIST, A. 1981. An integrated system of classification of flowering plants. New York: Columbia University Press. DORR, L. J. 2004. A remarkable new species of Sterculia (Sterculioideae, Malvaceae) from Madagascar. Adansonia, sér. 3. 26: 161–165. ——— and L. C. BARNETT. 1990. A new species of Hildegardia (Sterculiaceae) from Somalia. Kew Bulletin 45: 577–580. DOYLE, J. J. and J. L. DOYLE. 1990. Isolation of plant DNA from fresh tissue. Focus 12: 13–15. EDLIN, H. L. 1935a. A critical revision of certain taxonomic groups of the Malvales Part I. New Phytologist 34: 1–20. ———. 1935b. A critical revision of certain taxonomic groups of the Malvales Part II. New Phytologist 34: 123–143. ENDLICHER, S. L. 1850. Generum Plantarum: Supplementum Quartum. Part III. Vienna: Fredrich Beck. FAY, M. F., C. BAYER, W. S. ALVERSON, A. Y. DE BRUIJN, and M. W. CHASE. 1998. Plastid rbcL sequence data indicate a close affinity between Diegodendron and Bixa. Taxon 47: 43–50. FELSENSTEIN, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. GENE CODES CORPORATION. 2002. Sequenchery user manual. Ann Arbor: Gene Codes Corporation. GUYMER, G. P. 1988. A taxonomic revision of Brachychiton. Australian Systematic Botany 1: 199–323. HEYWOOD, V. H. 1993. Flowering plants of the world. Pp. 91–93. London: B.T. Batsford Ltd. HUTCHINSON, J. 1967. The genera of flowering plants (Angiospermae) Vol. II: 497–522. Oxford: Clarendon Press. HYLAND, B. P. M. and C. G. G. J. VAN STEENIS. 1987. Franciscodendron (Sterculiaceae), a new tree genus from Queensland. Brunonia 10: 211–214. HSUE, H. H. 1984. Sterculiaceae. Flora Reipublicae Popularis Sinicae. 49: 114–188. Beijing: Science Press. JUDD, W. S. and S. R. MANCHESTER. 1997. Circumscription of Malvaceae (Malvales) as determined by a preliminary cladistic analysis of morphological, anatomical, palynological, and chemical characters. Brittonia 49: 384–405. KING, G. 1891. Materials for a flora of the Malayan Peninsula. Journal of the Asiatic Society of Bengal 60: 57–95. KOSTERMANS, A. J. G. H. 1953a. The genera Scaphium Schott and Endl. and Hildegardia Schott and Endl. (Sterculiaceae). Journal of Science Research Indonesia 2: 13–23. ———. 1953b. New and critical Malaysian plants-1. Reinwardtia 2: 357–366. ———. 1954. A note on some African Sterculiaceae. Bulletin Jardin Botanique de l’Etat Bruxelles 24: 335–338. ———. 1956. The genus Firmiana Marsili (Sterculiaceae). Communication of the Forest Research Institute Indonesia 54: 3–33. ———. 1957. The genus Firmiana Marsili (Sterculiaceae). Reinwardtia 4: 281–310. ———. 1959a. A monograph of the genus Heritiera Dry. (Sterculiaceae) (including Argyrodendron F. v. M. and Tarrietia Bl.). Penerbitan, Madjelis Ilmu Pengetahutan Indonesia 1: 1–121. ———. 1959b. A monograph of the genus Heritiera Aiton (Sterculiaceae) (including Argyrodendron F. v. M. and Tarrietia Bl.). Reinwardtia 4: 465–583. ———. 1962. Miscellaneous botanical notes 3. Reinwardtia 6: 155– 187. ———. 1973. Some new taxa. Botanisk Tidsskrift 68: 317–323. ———. 1988. A note on Franciscodendron Hyland and Steenis (Sterculiaceae). Reinwardtia 10: 475. MADDISON, W. P. and D. R. MADDISON. 1992. MacClade: analysis of phylogeny and character evolution. Version 3. Sunderland: Sinauer Associates. MASTERS, M. T. 1868. Sterculiaceae. Pp. 214–239 in Flora of Tropical Africa 1, ed. D. Oliver. London: L. Reeve and Co. 2006] 169 WILKIE ET AL.: PHYLOGENY OF STERCULIOIDEAE ———. 1875. Sterculiaceae. Pp. 353–379 in Flora of British India 1, ed. J. D. Hooker. London: L. Reeve and Co. MERRILL, E. D. 1929. Plantae elmerianae borneenses. University of California Publications in Botany 15: 1–316. MORAT, P. 1986. Rappel historique du genre Sterculia L. en Nouvelle-Calédonie et réhabilitation du genre Acropogon Schltr. (Sterculiaceae). Bulletin du Muséum National d’ Histoire Naturelle, Paris, 4e Séries, 8, Section B., Adansonia 4: 357–364. ———. 1988. Contribution à l’étude des Sterculiaceae de la Nouvelle-Calédonie. Bulletin du Muséum National d’ Histoire Naturelle, Paris, 4e Séries, 10, Section B., Adansonia 2: 93–103. ——— and M. CHALOPIN. 2003 Quatre nouvelles espèces d’ Acropogon (Malvaceae: Sterculieae) endémiques de la NouvelleCalédonie. Adansonia, Séries 3, 25: 191–203. NYFFELER, R. and D. A. BAUM. 2000. Phylogenetic relationships of the durians (Bombacaceae-Durioneae or /Malvaceae/ Helicteroideae/ Durioneae) based on chloroplast and nuclear ribosomal DNA sequences. Plant Systematics and Evolution 224: 55–82. ———, C. BAYER, W. S. ALVERSON, A. YEN, B. A. WHITLOCK, M. W. CHASE, and D. A. BAUM. In press. Phylogenetic analysis of the Malvadendrina clade (Malvaceae s.l.) based on plastid DNA sequences. Organisms, Diversity and Evolution. OLMSTEAD, R. G., J. A. SWEERE, and K. H. WOLFE. 1993. Ninety extra nucleotides in the ndhF gene of tobacco chloroplast DNA: a summary of revisions to the 1986 genome sequence. Plant Molecular Biology 22: 1191–1193. PFEIL, B. E., C. L. BRUBAKER, L. A. CRAVEN, and M. D. CRISP. 2002. Phylogeny of Hibiscus and the tribe Hibisceae (Malvaceae) using chloroplast DNA sequences of ndhF and the rpl16 intron. Systematic Botany 27: 333–350. PRAIN, D. 1905. Mansonieae, a new tribe of the natural order Sterculiaceae. Journal of the Linnean Society 37: 250–263, plate 10. RIDLEY, H. N. 1922. Sterculiaceae. Pp. 266–290 in Flora of the Malay Peninsula vol. 1. London: L. Reeve and Co. ———. 1930. The dispersal of plants throughout the world. Ashford: L. Reeve and Co. ———. 1934. Firmiana and Erythropsis. Kew Bulletin 1934: 214–217. SANDERSON, M. J. and J. J. DOYLE. 1993. Phylogenetic relationships in North American Astragalus (Fabaceae) based on chloroplast DNA restriction site variation. Systematic Botany 18: 394–408. SEELANAN, T., A. SCHNABEL, and J. F. WENDEL. 1997. Congruence and consensus in the cotton tribe (Malvaceae). Systematic Botany 22: 259–290. SCHLECHTER, R. 1906. Beiträge zur Kenntnis der Flora von NeuKaledonien. Botanische Jahrbücher Systematik, Pflanzengeschichte und Pflanzengeographie 39: 179–188. SCHOTT, H. W. and S. F. L. ENDLICHER. 1832. Meletemata Botanica. Vienna: Karl Gerold. SCHUMANN, K. 1890. Sterculiaceae. Pp. 69–96 in Die natürlichen Pflanzenfamilien 3 (6), ed. A. Engler and K. Prantl. Leipzig: Engelmann. ———. 1893. Sterculiaceae. Pp. 97–99 in Die natürlichen Pflanzenfamilien 3 (6), ed. A. Engler and K. Prantl. Leipzig: Engelmann. ———. 1900. Sterculiaceae africanae. Pp. 1–140 in Monographien afrikanischer Pflanzen-familien und-Gattungen. ed. A. Engler. Leipzig: Engelmann. SIMMONS, M. P. and H. OCHOTERENA. 2000. Gaps as characters in sequence-based phylogenetic analysis. Systematic Biology 49: 369–381. SMALL, R. L. 2004. Phylogeny of Hibiscus sect. Muenchhusia (Mal- vaceae) based on chloroplast rpL16 and ndhF, and nuclear ITS and GBSSI sequences. Systematic Botany 29: 385–392. SMITH, L. S. 1969. New species of and notes on Queensland Plants V. Contributions from the Queensland Herbarium 6: 17–19. VAN STEENIS, C. G. G. J. 1960. Bibliography. Flora Malesiana Bulletin 3: 772. SWOFFORD, D. L. 2001. PAUP*: Phylogenetic analysis using parsimony (*and other methods), 4.0b7 beta version. Sunderland: Sinauer Associates. TAKHTAJAN, A. L. 1997. Diversity and classification of flowering plants. New York: Columbia University Press. TANTRA, I. G. M. 1976. A revision of the genus Sterculia L. in Malesia (Repisi marga Sterculia L. di Malesia) (Sterculiaceae). Communication (Lembaga Penelitian Hutan) 102: 1–194. TAYLOR, E. L. 1989. Systematic studies in the tribe Sterculieae: A taxonomic revision of the neotropical species of Sterculia L. (Sterculiaceae). PhD Thesis, Harvard University. WHEELER, W. C. 1995. Sequence alignment, parameter sensitivity, and the phylogenetic analysis of molecular data. Systematic Biology 44: 321–332. WHITING, M. F. J. C., Q. D. CARPENTER, and W. C. WHEELER. 1997. The Strepsiptera problem: phylogeny of the holometabolous insect orders inferred from 18S and 28S ribosomal DNA sequences and morphology. Systematic Biology 46: 1–68. WHITLOCK, B. A., C. BAYER, and D. A. BAUM. 2001. Phylogenetic relationships and floral evolution of the Byttnerioideae (‘‘Sterculiaceae’’ or Malvaceae s.l.) based on sequences of the chloroplast gene ndhF. Systematic Botany 26: 420–437. ———, K. G. KAROL, and W. S. ALVERSON. 2003. Chloroplast DNA Sequences confirm the placement of the enigmatic Oceanopapaver within Corchorus (Grewioideae: Malvaceae s.l., formerly Tiliaceae). International Journal of Plant Sciences 164: 35– 41. APPENDIX 1 Morphological characters and character states. A. Fruit structure: (0), Syncarpous, (1) Apocarpous. B. Fruit type: (0) Dehiscent Follicle, (1) Indehiscent follicle (2) Samara (3) Capsule. In the absence of a more suitable term we have used the term follicle to include those species which have indehiscent fruit. Some Hildegardia and Franciscodendron are tardily dehiscent these are coded as dehiscent. Heritiera littoralis has been coded an indehiscent follicle rather than samara as it displays a ridge/much reduced wing. C. Dispersal mechanism: (0) Bird, (1) Wind, (2) Water (3) Primate. Bird dispersal has been coded for those species with dehiscent follicles, all of which (apart from Brachychiton) have brightly coloured red-orange follicles. Wind dispersal takes two forms, those with very small seeds which get entangled in the dense hairs produced by the endocarp (the outgroup species), and those in which the carpel wall has expanded forming a samara. The mechanisms in Cola acuminata is not recorded but the heavy, indehiscent, fleshy fruit suggest primate dispersal (Cheek 2002). D. Fruit texture: (0) Leathery, (1) Fibrous, (2) Papery. Fruit are termed leathery if they are inflexible and the pericarp is generally thicker than 2–3mm with obvious layers, fibrous if they can be flexed but are generally resistant to it and the pericarp is 1–2mm thick with no obvious layers, and papery if they are easily flexed and the pericarp is less than 1mm thick. E. Ovules per locule: (0) one, (1) two or more. Very occasionally some Scaphium fruits have 2 seeds, but one is usually aborted or very much reduced in size. Scaphium here is coded as single seeded. F. Seed position: (0) Marginal, (1) Basal, (2) Central. 170 [Volume 31 SYSTEMATIC BOTANY APPENDIX 2. Matrix of morphological characters and character states. Character states in bold are polymorphic within the genus. Characters Taxa A B C D E F Ochroma pyramidale Bombax buonopozense Ceiba pentandra Pachira aquatica Sterculia tragacantha Sterculia stipulata Sterculia coccinea Sterculia balanghas Sterculia parvifloria Heritiera simplicifolia Brachychiton acerifolius Brachychiton acerifolius Acropogon dzumacensis Acropogon bullatus Argyrodendron peralatum Franciscodendron lauriflorum Pterygota alata Heritiera aurea (sample A) Heritiera aurea (sample B) Heritiera littoralis Heritiera elata Cola acuminata Octolobus spectabilis Hildegardia barteri Firmiana malayana Hildegardia populifolia Firmiana simplex Pterocymbium tinctorium Scaphium macropodum (sample A) Scaphium macropodum (sample B) Scaphium linearicarpum 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 3 3 3 0 0 0 0 0 2 0 0 0 0 2 0 0 2 2 1 2 1 1 1 0 1 0 0 0 0 0 1 1 1 1 0 0 0 0 0 1 0 0 0 0 1 1 1 1 1 2 1 3 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1/2 1/2 0 1 1 0 1 0 0 2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 0/1 1 0 0 0 0 1 1 0/1 1 0/1 1 0 0 0 0 2 2 2 2 0 0 0 0 0 1 0 0 0 0 1 0 0 1 1 1 1 0 0 0 0 0 0 1 1 1 1