View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector ARTICLE IN PRESS

Organisms, Diversity & Evolution 5 (2005) 109–123 www.elsevier.de/ode

Phylogenetic analysis of the Malvadendrina clade ( s.l.) based on plastid DNA sequences Reto Nyffelera,1, Clemens Bayerb, William S. Alversonc, Alan Yena,2, Barbara A. Whitlocka,3, Mark W. Chased, David A. Bauma,Ã

aHarvard University Herbaria, Cambridge, USA bPalmengarten der Stadt Frankfurt am Main, Germany cThe Field Museum, Chicago, USA dRoyal Botanic Gardens Kew, United Kingdom

Received 18 April 2004; accepted 10 August 2004

Abstract

Phylogenetic relationships within Malvaceae s.l., a clade that includes the traditional families , Malvaceae s.str., , and , have become greatly clarified thanks to recent molecular systematic research. In this paper, we use DNA sequences of four plastid regions (atpB, matK, ndhF, and rbcL) to study relationships within Malvadendrina, one of the two major clades of Malvaceae s.l. The four data sets were generally in agreement, but five terminal taxa manifested highly unexpected affinities in the rbcL partition, and the non-coding sequences of the trnK intron were found to provide limited phylogenetic information for resolving relationships at the base of Malvadendrina. The remaining data strongly support the existence of six major clades within Malvadendrina: Brownlowioideae, , , Malvatheca (comprising and ), , and . These data also resolve the placement of two problematic taxa: (in Dombeyoideae) and Mortoniodendron (in Tilioideae). The relationships among the six clades are not definitively resolved, but the best-supported topology has Dombeyoideae as sister to the remainder of Malvadendrina (posterior probability PP ¼ 80%) and Sterculioideae as sister to Malvatheca (PP ¼ 86%). This early branching position of Dombeyoideae is supported by similarities in floral characters between members of that clade and outgroup taxa in . Similarly, the sister-group relationship of Sterculioideae and Malvatheca receives support from androecial characteristics, like subsessile or sessile anthers and an absence of staminodes, shared by these two clades. r 2005 Gesellschaft fu¨ r Biologische Systematik. Published by Elsevier GmbH. All rights reserved.

Keywords: Molecular systematics; Character evolution; Malvadendrina; Bombacaceae; Malvaceae; Sterculiaceae; Tiliaceae

ÃCorresponding author. Current address: Department of Botany, Introduction University of , 430 Lincoln Drive, Madison, WI 53706, USA. Morphological and molecular phylogenetic studies E-mail address: [email protected] (D.A. Baum). during the last 6 years have greatly improved our 1Current address: Institut fu¨r Systematische Botanik, Universita¨t Zu¨rich, Switzerland. understanding of systematics and resulted in 2Current address: Lifespan Biosciences Inc., Seattle, USA. the recircumscription of major taxonomic groups. 3Current address: Department of Biology, University of Miami, USA. Malvales were shown to comprise the four traditional

1439-6092/$ - see front matter r 2005 Gesellschaft fu¨ r Biologische Systematik. Published by Elsevier GmbH. All rights reserved. doi:10.1016/j.ode.2004.08.001 ARTICLE IN PRESS 110 R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 core families (Bombacaceae, Malvaceae, Sterculiaceae, Edlin 1935; Venkata Rao 1952). For example, knowl- and Tiliaceae) plus several additional families, most edge about the sister group of Malvatheca could help notably , Cochlospermaceae, Cistaceae, Dip- understand the origin of the unusual androecial terocarpaceae, Sarcolaenaceae, and (Al- structures in Bombacoideae and Malvoideae (von verson et al. 1998; Fay et al. 1998; Bayer et al. 1999; Balthazar et al. 2004) and the biogeographic history of Kubitzki and Chase 2003). Although the core families the group (Baum et al. 2004). formed a well-supported monophyletic group within In this paper, we combine previously published Malvales, the only one of the four traditional core plastid DNA data-rbcL and atpB (Bayer et al. 1999), families that represents a monophyletic group is ndhF (Alverson et al. 1999; Nyffeler and Baum 2000; Malvaceae (Judd and Manchester 1997; Alverson et al. Whitlock et al. 2001)-with new data from the trnK 1999; Bayer et al. 1999). Given these results, two basic intron in order to improve resolution among the major options were available for revising the familial ranked clades of Malvadendrina. The matK open reading frame classification while only recognizing well-supported is located in a long intron of the transfer RNA for monophyletic groups: either erecting five more families, Lysine (trnKUUU) and flanked on both sides by non- in addition to Bombacaceae, Malvaceae, Sterculiaceae, coding sequences (about 900 base pairs 50 of matK and and Tiliaceae, or merging all members into a single, about 250 base pairs 30 of matK). matK is among the expanded Malvaceae. Bayer et al. (1999) opted for the most rapidly evolving plastid DNA coding regions latter on the grounds that the level of morphological (Soltis and Soltis 1998) and has, therefore, most often differentiation among the major clades would not been used for phylogenetic studies at the intergeneric or warrant family rank. Baum et al. (1998) developed a infrageneric level (e.g. Johnson and Soltis 1994, 1995; parallel, rankless classification, providing phylogenetic Plunkett et al. 1996). Although most phylogenetic definitions for Malvaceae s.l. and various well-sup- studies in the past have focused on the matK coding ported subclades. The expanded concept of Malvaceae, region, recent studies have often also included the non- as recognized in this study, comprises about 4300 coding trnK intron, yielding in total about 2800 bp of and 245 genera (Bayer and Kubitzki 2003). Malvaceae comparatively variable plastid DNA sequences (e.g. Hu s.l. are characterized by a combination of stellate hairs, et al. 2000; Lavin et al. 2000). Including the entire trnK palmately veined , an inflorescence structure intron generally results in the addition of at least 60% consisting of characteristic repeating ‘bicolor units’, more informative characters and can, therefore, yield valvate , mucilage cavities in the cortex and pith, better-resolved phylogenetic estimates (Nyffeler 2002). and cyclopropanoid oils (Bayer 1999; Bayer and Here we use this expanded plastid DNA data set to Kubitzki 2003; Kubitzki and Chase 2003). address the major relationships in Malvadendrina and Molecular systematic studies have indicated that also to investigate the phylogenetic positions of two Malvaceae s.l. consists of nine distinct subclades problematic genera Mortoniodendron and Nesogordonia. (Alverson et al. 1999; Bayer et al. 1999; Whitlock et al. Additionally, we discuss the implications of the resultant 2001): Bombacoideae, Brownlowioideae, Byttnerioi- for the evolution of a few prominent morpholo- deae, Dombeyoideae, , Helicteroideae, gical features. Malvoideae, Sterculioideae, and Tilioideae. Byttnerioi- deae and Grewioideae form a clade (Byttneriina; Baum et al. 1998) that is sister to a clade consisting of the remainder of Malvaceae (Whitlock et al. 2001). The Material and methods latter clade, named Malvadendrina (Baum et al. 1998), is well-supported by plastid DNA data, including a Taxon sampling decisive 21 bp deletion in ndhF (Alverson et al. 1999). Within Malvadendrina, Malvoideae plus Bombacoideae Thirty-one terminals, encompassing the taxonomic constitute a well-supported clade (Alverson et al. 1999; diversity of Malvadendrina, and four outgroups from Bayer et al. 1999; Baum et al. 2004) that has been named Byttnerioideae and Grewioideae, were included in this Malvatheca (Baum et al. 1998). However, relationships study (see Appendix). Mortoniodendron and Nesogordo- between Malvatheca and the other five major subclades nia were included to clarify their phylogenetic affinities. of Malvadendrina (Brownlowioideae, Dombeyoideae, Since sequencing of the different plastid DNA data sets Helicteroideae, Sterculioideae, and Tilioideae) remained was initially conducted independently at the Harvard unresolved. Morphological diversity within Malvaden- University Herbaria and the Royal Botanic Gardens, drina is great, as seen by the fact that the different taxa Kew, some terminal taxa are now represented by have traditionally been assigned to four different congeneric individuals of different species in the families. Consequently, resolving relationships among combined plastid DNA data set. Overall, this concerns these lineages is critical for evaluating alternative 17 of the sampled 35 genera (for details on sampled scenarios for the diversification of Malvadendrina (e.g. individuals see the Appendix): , , ARTICLE IN PRESS R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 111

Ceiba, , , , , (Table 1). Sequence electropherograms were assembled , , , , Mortoniodendron, using Sequencher 3.0 (Gene Codes Corp.; Ann Arbor, , , , , and Michigan). GenBank accession numbers of all sequences . are listed in Appendix 1.

Laboratory methods Sequence alignment and phylogenetic analysis

Sequence data of the molecular markers atpB, ndhF, Sequences of the four plastid regions were aligned and rbcL were available from previous studies. Labora- manually and assembled into a combined data matrix. tory methods for these three plastid DNA data sets were Given the apparent lack (or extreme rarity) of recombi- outlined in Alverson et al. (1999) and Bayer et al. (1999). nation within the plastid genome, there is no a priori The trnK intron region was amplified in two separate reason to suppose that some of the plastid DNA PCR reactions. The 50 end was amplified with trnK- sequences could have tracked different phylogenetic 3914F (Johnson and Soltis 1994) and matK-P6R (AAG histories. However, because the exemplar species used ACT CCA GAA GAT GTT CG), modified from matK- for some genera varied between the different data sets, 1470R (Johnson and Soltis 1994). The 30 half was generic nonmonophyly could generate discordant re- amplified with matK-4F (CTT CGC TAC CGG GTG sults. Furthermore, the possibility of discordant results AAA GAT G) and matK-2R (Johnson and Soltis 1994). due to laboratory errors should also be considered. We Each 25 mL PCR reaction contained 10 mM Tris-HCl, therefore conducted partition homogeneity tests (ILD 50 mM KCl, 2.5 mM MgCl2, 0.2 mM of each dNTP, tests; Farris et al. 1994) for all possible pairs of data sets 0.4 mM of each primer, and 0.625 units of Taq as implemented in PAUP*4.0b8 (Swofford 2002). Six polymerase. The PCR temperature profile for all pairwise ILD tests were conducted using the parsimony reactions was 95 1Cfor 5 min, then 35 cycles of 94 1C optimality criterion. Heuristic searches were conducted for 30 s, 48 1Cfor 60 s, 72 1Cfor 90 s, followed by a final with simple taxon addition, TBR branch swapping, and extension of 72 1Cfor 5 min. PCRproducts were cleaned MAXTREES set to 100. using QIAquick columns (Qiagen Corp.; Valencia, We conducted parsimony analyses on the individual ) and sequenced either directly or cloned into and combined data sets using PAUP*4.0b8 (Swofford pGEM T-Easy plasmids (Promega Corp.; Madison, 2002). The four exemplars from Byttnerioideae and Wisconsin) and then sequenced using BigDye Termina- Grewioideae were defined as outgroups. In addition to tor Cycle Sequencing premix (Applied Biosystems; analyses of the four independent data sets, two Foster City, California) on an ABI 377 or ABI 3100 combined analyses were conducted, consisting either of automated sequencer (Applied Biosystems). Cloning all available sequence data, or of a reduced data set was used only in rare cases when PCR product yield excluding the non-coding trnK intron data and the rbcL was poor, and then only one clone was sequenced per data (for justification see the Results section). All sites product. Eleven internal primers, of which nine were were equally weighted in all six analyses, and gaps in the newly designed for this project, were used for sequencing alignment were treated as missing data. Each analysis

Table 1. Internal primer sequences used for direct sequencing of trnK/matK

Primer name Relative position Direction Primer sequence (first published reference) trnK-P3F 247 Forward TTCAAA GTT GGG TCGAGT GA trnK-P4R 375 Reverse AAA CCT TTA CCR CAT YAG GCA C trnK-P5F 558 Forward CCA TCT TCT TAT CCT ATA ACG AAC trnK-P7F 673 Forward CTT GTT TTG ACT GTA TCG CA matK-4F 1240 Forward CTT CGC TAC CGG GTG AAA GAT G (Manos and Steele 1997) matK-P6R 1441 Reverse AAG ACT CCA GAA GAT GTT GAT CG matK-P6F 1578 Forward TCA AGG AAA GGC AAT TCT GG matK-P8R 1900 Reverse CACGTCGGCTTACTAATG GG matK-52F 1770 Forward GGT ACG GAG TCA AAT GGT AGA A (Nyffeler 2002) matK-P9F 2107 Forward GGT TCG GAA TTT TTG GAA GA matK-P10R 2205 Reverse CAA ATA CCA AAT TCG TCC CCT A

The relative position is based on the sequence of patersonii (GenBank acc. No. AY589064) starting at the 50 end of the trnK intron. ARTICLE IN PRESS 112 R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 consisted of 1000 random taxon addition sequence reasons, some sequences remained incomplete or miss- replicates, with heuristic searches based on TBR branch ing. Four atpB sequences (for , Neesia, Nesogor- swapping with no limit on the number of trees. We donia, and Thespesia) and one rbcL sequence (for assessed support for clades with 10,000 flat-weighted Nesogordonia) were not obtained: these taxa were parsimony bootstrap replicates that were subject to excluded from analyses that only contained the missing heuristic searches with simple taxon addition, TBR gene. In addition, two sequences of atpB ( and branch swapping, and MAXTREES set to 500. ), one of trnK (Craigia), and three of rbcL We conducted Bayesian phylogenetic analyses on the (Lavatera, Nesogordonia, Pentaplaris) were missing at same six data sets as outlined above (four individual least one quarter of their total length, but these taxa genes plus two combined analyses), except that for both were retained in all analyses. combined data sets non-coding regions of the trnK Alignment required no gaps in rbcL and just one in region were excluded. The software program MrBayes atpB. In contrast, the ndhF and trnK data sets required 3.0B4 (Huelsenbeck and Ronquist 2001) was used for the insertion of several gaps. In the case of the non- these analyses. The best-available model of molecular coding regions of the trnK intron, the alignment was evolution for each partition was selected based on partially ambiguous. A parsimony analysis of one hierarchical likelihood ratio tests as implemented in the plausible alignment of the trnK intron region alone software program MrModeltest 1.1b, which only con- (1280 aligned base pairs, 93 informative sites) resulted in siders nucleotide substitution models that are currently 74,472 trees, and failed to resolve the monophyly of implemented in PAUP* and MrBayes 3.0 (Nylander Malvadendrina or any relationships among the major 2002). Since all four data sets only contained coding lineages (the focus of the present investigation). sequences, each of them was partitioned by first, second, Although these non-coding regions have utility at lower and third codon positions for all Bayesian analyses, and taxonomic scales, for example within Malvatheca model parameters were allowed to vary independently in (Baum et al. 2004), they were excluded from most of all of the either three (individual analyses), nine the parsimony [i.e. individual trnK ( ¼ matK only, (combined analysis excluding rbcL), or 12 (combined Fig. 1b) and final combined (Fig. 1f)] analyses, and from analysis of all four molecular markers) partitions. We all of the Bayesian analyses (Fig. 2b, e, f). conducted three Markov chain Monte Carlo (MCMC) Of all the pairwise ILD tests conducted (Table 2), analyses for each data set, each composed of four linked only those that included rbcL and either the informa- chains starting out from random trees. One chain was tion-rich matK or ndhF partitions failed to accept the unheated, whereas the other three were heated sequen- null hypothesis of congruent data sets at a P-value of tially with a heat parameter of 0.2. The chains where run 0.05. This discordance is unlikely to merely reflect for 1,000,000 generations, with every 100th from the different rates of evolution among partitions (e.g. unheated chain being stored. We evaluated convergence Dolphin et al. 2000; Barker and Lutzoni 2002), because by visual inspection of likelihood scores as a function of analyses of the rbcL data supported relationships in generation. A conservative burn-in period was identified strong contradiction to those found for the other and the majority-rule consensus trees were generated partitions (see below). with the SUMT command in MrBayes (Huelsenbeck The total combined plastid DNA data set, including and Ronquist 2001). The variation in clade posterior the non-coding trnK intron and rbcL sequences, probabilities between the three runs for each data set provided an aligned matrix of 7791 bp, of which 634 was used to assess mixing. Since good mixing was positions were parsimony informative. Some additional observed, the three post-burnin posterior distributions descriptive information on the four sequence partitions from each data set were pooled. is given in Table 3.

Phylogenetic analyses Results All six parsimony analyses found at most a few Sequence characteristics dozen most-parsimonious (MP) trees, except for the rbcL partition, which yielded 509 MP trees (Table 3). Thirty-five terminal taxa from Malvaceae were Fig. 1 shows the strict consensus trees of the six flat- considered for this study. For assorted technical weighted parsimony analyses. The tree topologies of the

Fig. 1. Strict consensus trees from the flat-weighted parsimony analyses of individual and combined data sets; numbers above branches indicate bootstrap support percentages. (a) atpB sequence data; (b) matK sequence data; (c) ndhF sequence data; (d) rbcL sequence data; (e) combined plastid DNA data including the non-coding trnK introns and including rbcL; (f) combined plastid DNA data excluding the non-coding trnK introns and excluding rbcL. Abbreviations: B ¼ Brownlowioideae; D ¼ Dombeyoideae; H ¼ Helicteroideae; M ¼ Malvatheca (Bombacoideae+Malvoideae); O ¼ outgroups; S ¼ Sterculioideae; and T ¼ Tilioideae. ARTICLE IN PRESS R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 113

Byttneria Byttneria 84 Theobroma 92 98 Grewia O 99 Grewia O Sparrmannia Sparrmannia Nesogordonia 100 Dombeya 66 100 Dombeya 83 Trochetiopsis Trochetiopsis D 89 Pterospermum D 55 Pterospermum Schoutenia Durio 100 Durio - 89 64 Neesia 52 Helicteres H 96 Reevesia H 65 Helicteres 96 Triplochiton 62 Mortoniodendron T 75* Mortoniodendron - 83 Craigia T 80 Pentace Tilia 73 B 95 Berrya 100 Brownlowia B Sterculia Pentace 56 63 - Cola 99 Sterculia S 88 Cola S Hildegardia Fremontodendron Fremontodendron Ochroma Adansonia Adansonia 73* 76 - - Bombax Ceiba M Pachira M Matisia 65 Matisia Pentaplaris Pentaplaris 87 - Hibiscus Hibiscus 91 ab72 70 Lavatera (atpB) 77 Gossypium (matK) 100 Gossypium Lavatera Thespesia

Byttneria Theobroma 98 Theobroma 68 Grewia 100 Grewia O 64 Sparrmannia Sparrmannia Byttneria 98 Durio Tilia 96 Neesia 99 Durio Reevesia 96 H - Neesia Helicteres 76 Reevesia Triplochiton 84 Helicteres Mortoniodendron 70 Triplochiton 96 Craigia 53 T Matisia Tilia Hildegardia Nesogordonia 100 96 Sterculia 100 Dombeya Nesogordonia - 93 Trochetiopsis D - Bombax 88 Pterospermum - 84 Pterospermum Schoutenia Schoutenia Berrya 99 Berrya 100 Brownlowia B 89 Pentace 100 Brownlowia - Pentace 100 Sterculia 100 Cola S 60 Lavatera Hildegardia 100 Dombeya Ochroma - Trochetiopsis Pentaplaris 85 Adansonia Bombax - Fremontodendron - 73 53 Ceiba - Craigia Pachira 61 Adansonia Fremontodendron Ochroma - Matisia M 66 Ceiba 52 Pentaplaris - Pachira 85 Cola 98 Hibiscus - cd91 Lavatera 74 Hibiscus (ndhF) 100 Gossypium (rbcL) 91 Gossypium Thesespesia Thespesia

Theobroma Theobroma 72 Byttneria 65 Byttneria 100 Grewia O 100 Grewia O Sparrmannia Sparrmannia Nesogordonia Nesogordonia 100 100 Dombeya 99 100 Dombeya 100 Trochetiopsis D 92 Trochetiopsis D 99 Pterospermum 98 Pterospermum Schoutenia Schoutenia 66 Mortoniodendron 100 100 Durio 67 Craigia T 99 Neesia 100 Tilia 100 Reevesia H 100 Durio 75 Helicteres 99 Neesia Triplochiton 82 100 Reevesia H 90 Mortoniodendron 99 Helicteres 67 Craigia T Triplochiton Tilia 100 Berrya 40 100 Berrya 60 100 Brownlowia B 100 Brownlowia B 50 Pentace Pentace 100 Sterculia 39 100 Sterculia 99 Cola S 100 Cola S Hildegardia Hildegardia Fremontodendron 55 Fremontodendron Ochroma Ochroma Adansonia Adansonia 99 99 Bombax 100 99 Bombax 51 Ceiba Ceiba Pachira M Pachira M 64 Matisia 73 Matisia 100 Pentaplaris 99 Pentaplaris 100 Hibiscus 100 Hibiscus ef98 Lavatera 99 Lavatera (atpB, trnK, ndhF, rbcL)(100 Gossypium atpB, matK, ndhF) 100 Gossypium Thespesia Thespesia ARTICLE IN PRESS 114 R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123

Byttneria Theobroma Theobroma Byttneria O Grewia O * Grewia * * Sparrmannia * Sparrmannia Nesogordonia Dombeya Dombeya * Trochetiopsis D * Trochetiopsis D * Pterospermum 97 Pterospermum 99 Schoutenia 83 Schoutenia Durio 85 Durio 97 * Neesia Reevesia * Helicteres H Reevesia H 53 Helicteres 76 Triplochiton * 64 Triplochiton * 62 Mortoniodendron Mortoniodendron 71 Tilia T 92 * Craigia T Pentace 0.01 Tilia 99 Berrya B Berrya 86 Brownlowia * Brownlowia B Sterculia 96 * Pentace Cola Sterculia 99 S Cola 72 68 Hildegardia * S 96 Hildegardia Fremontodendron Fremontodendron Ochroma 74 Ochroma Pachira Adansonia Adansonia Bombax Bombax * * Ceiba * 0.01 Ceiba Pachira 55 M M Matisia Matisia Pentaplaris Pentaplaris * Lavatera * ab52 Hibiscus Hibiscus Gossypium * (atpB) 98 (matK) 69 Gossypium * Lavatera * Thespesia

Byttneria Theobroma Theobroma Grewia Grewia O Sparrmannia 98 * * Sparrmannia 92 Byttneria Durio Tilia * Neesia Durio * Triplochiton H * Neesia * Helicteres 83 Reevesia 51 Reevesia * Helicteres Tilia * Triplochiton 94 * Craigia T Matisia Mortoniodendron Hildegardia Nesogordonia 51 Sterculia Dombeya Bombax 80 * Trochetiopsis D * Nesogordonia * Pterospermum Berrya 95 Schoutenia 0.01 85 Brownlowia Berrya * Brownlowia * Pentace * B Pterospermum * Pentace Sterculia 99 Schoutenia 59 Cola 77 Lavatera * S Dombeya 0.01 * Hildegardia 88 Fremontodendron * Trochetiopsis Pentaplaris Ochroma 57 Adansonia Craigia Fremontodendron * Bombax * Ceiba Adansonia 98 Pachira 96 Ochroma 70 Matisia M Ceiba Pentaplaris * Pachira 87 Hibiscus Cola * Hibiscus cd* Lavatera 52 (ndhF) 99 Gossypium (rbcL) * Gossypium * Thespesia * Thespesia

Theobroma Byttneria Byttneria Theobroma Grewia O 50 Grewia O 73 * Sparrmannia * Sparrmannia Durio Nesogordonia * Neesia Dombeya * Reevesia H * * Trochetiopsis D * Helicteres * Pterospermum * Triplochiton * Schoutenia Nesogordonia Durio Dombeya Neesia 0.01 * * * * Trochetiopsis D * Reevesia H * Pterospermum Helicteres * * 0.01 * Schoutenia 96 Triplochiton Mortoniodendron Mortoniodendron 80 * Craigia T Craigia T Tilia 85 99 Tilia Berrya 95 Berrya * Brownlowia B 61 * Brownlowia B 63 * Pentace * Pentace Sterculia Sterculia * Cola S 59 * Cola S * Hildegardia * Hildegardia 72 Fremontodendron Fremontodendron Ochroma 86 Ochroma Adansonia Adansonia * Bombax * Bombax * Ceiba * Ceiba 96 Pachira 88 Pachira 87 56 Matisia M Matisia M Pentaplaris Pentaplaris * Hibiscus * Hibiscus * * ef* Lavatera * Lavatera (combined)* Gossypium (atpB, matK, ndhF) * Gossypium * Thespesia * Thespesia ARTICLE IN PRESS R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 115

Table 2. Results from pairwise partition homogeneity tests consensus trees from the six Bayesian analyses with clade posteriors for all distinct branches. Branch lengths First partition Second partition P-value were obtained as the mean across the subset of trees atpB matK 0.156 having the branch in question (SUMT command in atpB ndhF 0.396 MrBayes; Huelsenbeck and Ronquist 2001). Examining atpB rbcL 0.010 the results of the parsimony and Bayesian analyses, matK ndhF 0.018 there is clearly a great deal of concordance among matK rbcL 0.001 partitions and between modes of analysis (except for ndhF rbcL 0.001 rbcL, as noted above). All tests with matK did not include the flanking, trnK non-coding Across all analyses, six major clades can be recognized regions. within the ingroup. These clades correspond well to those identified in previous studies (Alverson et al. 1999; Bayer et al. 1999): Brownlowioideae (‘‘B’’ in Figs. 1 and individual analyses of atpB, matK, and ndhF sequences 2), Dombeyoideae (D), Helicteroideae (H), Sterculioi- are largely congruent, whereas rbcL indicates different deae (S), Tilioideae (T), and Malvatheca (M) (the latter patterns. For example, whereas the other data sets agree including Bombacoideae and Malvoideae). The support in the monophyly of the morphologically cohesive values (i.e. parsimony bootstrap indices BS, and Sterculioideae, the rbcL data analyses imply that Cola Bayesian posterior probabilities PP) are moderate for (but not Sterculia and Hildegardia) is embedded within many of them in the individual analyses, but high in the Malvatheca, close to Malvoideae. Similarly, Lavatera,a combined analyses. Nesogordonia emerges as sister to member of the core Malvoideae (Baum et al. 2004), the remainder of Dombeyoideae. In all individual appears sister to members of Dombeyoideae. Overall, partitions except ndhF, the enigmatic Mortoniodendron five terminal taxa (Bombax, Cola, Craigia, Lavatera, and appears to be most closely related to Craigia and Tilia in Matisia) did not fall into the same major clade in Tilioideae. The combined analysis corroborates this analyses of rbcL as in analyses of other data. Despite placement and provides strong support (BS ¼ 90%; numerous checks, we found no evidence of laboratory PP ¼ 99%). Relationships within the major clades are errors. We can only hope, therefore, that more detailed often well resolved, although, in view of the limited studies on the molecular evolutionary patterns can taxon sampling, these findings are of limited signifi- explain the curious results of rbcL. In light of these cance. problems, in the remainder of this paper we will give The individual analyses of the atpB, matK, and ndhF greater prominence to combined analyses that exclude partitions remain equivocal concerning relationships rbcL (Figs. 1f, 2f). among the six major clades of Malvadendrina. How- Likelihood ratio tests indicated that the general time- ever, the analyses that were based on combined plastid reversible model (GTR model; Rodrı´ guez et al. 1990), DNA data (Figs. 1e, f and 2e, f) gave higher support for including a parameter to allow for among-site variation certain relationships among these larger clades. following a discrete approximation to a gamma rate distribution, is preferred for all data sets. MrModeltest indicated that the GTR+G model should be supple- mented in the case of the two combined data sets and the Discussion rbcL data partition with an additional parameter that designates a proportion of invariant sites (GTR+G+I). Other analyses of rbcL sequences (on Malpighiaceae, Bayesian phylogenetic analyses with MCMC showed Cameron et al. 2001; and Iridaceae, Reeves et al. 2001) rapid convergence usually by about the 12,000th have indicated misplacement of some taxa, similar to the generations. The variation in clade posteriors between patterns observed here. Although the partition homo- replicates was minimal, indicating good mixing and geneity test indicated incongruence in one case (Reeves justifying combining all three pools into a single et al. 2001), the combined analysis with other plastid posterior distribution. Fig. 2 shows the majority-rule regions was much better supported and placed, with

Fig. 2. Majority-rule consensus trees from Bayesian analyses of individual and combined data sets; numbers below branches indicate posterior probabilities ( ¼ 100 percent); branch lengths derived from means conditioned on trees having the taxon bipartition in question (SUMT command in MrBayes 3.0). (a) atpB sequence data; (b) matK sequence data; (c) ndhF sequence data; (d) rbcL sequence data; (e) combined plastid DNA data including the non-coding trnK introns and including rbcL; (f) combined plastid DNA data excluding the non-coding trnK introns and excluding rbcL. Abbreviations: B ¼ Brownlowioideae; D ¼ Dombeyoideae; H ¼ Helicteroideae; M ¼ Malvatheca (Bombacoideae+Malvoideae); O ¼ outgroups; S ¼ Sterculioideae; and T ¼ Tilioideae. ARTICLE IN PRESS 116 R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123

Table 3. Descriptive information on partitions and parsimony analyses

atpB matK ndhF rbcL Combined-all Combined-reduced

Characters 1406 1545 2162 1398 7791 5113 Informative characters 59 174 201 107 634 434 (4.2%) (11.3%) (9.3%) (7.7%) (8.1%) (8.5%) Tree length 245 629 752 423 2492 1637 Number of MP trees 64 160 12 509 14 5 CI (excl. uninf. chars.) 0.82 0.80 0.80 0.60 0.76 0.80 (0.61) (0.64) (0.64) (0.44) (0.57) (0.63) RI 0.74 0.71 0.72 0.56 0.65 0.71

The six data columns correspond to the six parsimony analyses and consensus topologies illustrated in Fig. 1. ‘‘Combined-all’’ refers to analyses of the complete concatenated data set; ‘‘Combined-reduced’’ to analysis of the complete data set after the exclusion of rbcL and the trnK non-coding regions. high bootstrap support, these ‘problem’ taxa in the The major outstanding problem in the phylogeny of clades indicated by their previous classification and Malvadendrina boils down to a ‘‘six-clade problem’’, analyses of the other DNA regions. The unusual offering potentially 945 exclusively bifurcating rooted positions of these misplaced taxa in the rbcL trees topologies (Felsenstein 1978). The addition of 1545 bp of these other studies were not well-supported and of matK sequences to the previously generated atpB and did not have a negative effect on resolution or boot- ndhF sequences and the removal of problematic rbcL strap percentages, just as in this study. The general sequences have improved resolution, but do not allow us conclusion of Cameron et al. (2001) and Reeves et al. to arrive at a single, well-supported topology. None- (2001) was that there was no conflicting signal in rbcL, theless, this combined analysis provides the best-avail- but rather that there was simply no clear signal for the able hypothesis of sister-group relationships in placement of these taxa. In the original paper in which Malvadendrina. The following discussion is guided the rbcL results were published (Bayer et al. 1999), this largely by the fully resolved topology as identified in misplacement also occurred, but only a combined tree of the final analysis of combined atpB, matK, and ndhF rbcL and atpB was shown. In Bayer et al. (1999), the sequence data (Figs. 1f and 2f). positions of these problem taxa were not supported (by bootstrap percentages greater than 50%) with either the rbcL or atpB data, and in the combined tree Cola, for Dombeyoideae example, was placed with Sterculia and Hildegardia, but also without bootstrap support greater than 50%. Such The combined analysis agrees with the individual patterns of low bootstrap support do not permit the atpB and matK partitions in indicating that Dombeyoi- separation of conflicting signal from that of sampling deae are sister to the remainder of Malvadendrina error (i.e. not enough data to obtain a clear result; (BP ¼ 82%; PP ¼ 80%). The Dombeyoideae clade Huelsenbeck et al. 1996). We cannot identify factors includes the enigmatic Nesogordonia (BP ¼ 99%; that would be responsible for this phenomenon, and it PP ¼ 100%) and thus comprises about 350 species in should be examined more closely in future studies. We 20 genera (Bayer and Kubitzki 2003). Dombeyoideae can only hope, therefore, that more detailed studies on are mainly distributed in the Paleotropics, with centers the molecular evolutionary patterns can explain the of diversity in Madagascar, the Mascarenes, and in curious results of rbcL, but this result in Malvaceae s.l. is Southeast Asia. not unprecedented. Nesogordonia consists of about 18 species, 15 of which This combined analysis of atpB, matK, and ndhF occur in Madagascar, and is aberrant within Dombeyoi- sequences confirms the presence of the six major deae due to its mostly free and antepetalous subclades identified in earlier studies (Alverson et al. staminodes, its simple (not bifid as in other Dombeyoi- 1999; Bayer et al. 1999), and provides consistently high deae, except Pterospermum; Mohana Rao 1976; Barnett support for these clades. 1988) cotyledons, and smooth . Previously, The taxa we sampled here included divergent ex- Nesogordonia had been included in Tiliaceae-Tilieae emplars from all major clades as well as taxa that had (Schumann 1897), Sterculiaceae-Dombeyeae (Burret remained problematic (i.e. Mortoniodendron and Neso- 1926), Buettneriaceae-Mansonieae (Edlin 1935), or gordonia). Thus, despite sampling only 31 ingroup taxa, Sterculiaceae-Helmiopsideae (Are` nes 1959; Hutchinson our results allow all genera of Malvadendrina to be 1967; Takhtajan 1997). This last placement in the tribe confidently assigned to one of the six major clades. Helmiopsideae was proposed based on the presence ARTICLE IN PRESS R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 117 of winged . Nonetheless, Barnett (1988) found been proposed. Durioneae are consistently resolved as that the seed wing of Nesogordonia is formed by a sister to the remaining species (i.e. Helicteroideae s.str.; tissue different from that of other representatives of Bayer and Kubitzki 2003), and there is strong evidence Helmiopsideae. that Triplochiton (and probably also Mansonia; Bayer et The general structure of flowers of Dombeyoideae al. 1999) is more closely related to Helicteres than to closely resembles that of Byttnerioideae (which are part Reevesia. Ungeria, which was not included here, is sister of Byttneriina, the sister group to Malvadendrina; to Reevesia (Nyffeler and Baum 2000). Whitlock et al. 2001): the sepals are free or almost so, Androgynophores are prominent floral structures and staminodes are present and form a tube together in Helicteroideae s.str., but these are absent in Dur- with the fertile stamens. Furthermore, like Byttnerioi- ioneae. Androgynophores are also regularly found in deae, the stamens are often arranged in antepetalous Sterculioideae, occasionally in Dombeyoideae (i.e. groups each with relatively few, tetrasporangiate Pterospermum), Brownlowioideae, Grewioideae, and anthers. All Dombeyoideae investigated so far have Byttnerioideae. A detailed comparative analysis of the polytelic synflorescences (Bayer 1994). Their ‘bicolor androgynophores found in different groups of Mal- units’ (Bayer 1999) are basically reduced to single vaceae is still lacking. The scattered distribution of this flowers with an epicalyx, as occurs also in Malvoideae characteristic, however, indicates that this feature is of and a few other lineages, including some Byttnerioideae. limited taxonomic significance in Malvaceae. In Dombeyoideae, these one-flowered units are arranged In our analysis Helicteroideae branch after Dom- in condensed, axillary cymose aggregates. Although the beyoideae (BS ¼ 82%; PP ¼ 80%) and are weakly data on Nesogordonia are limited, it appears that similar supported as sister to Tilioideae–Brownlowioideae– inflorescences with lateral cymes occur, which links the Sterculioideae–Malvatheca (BS ¼ 40%; PP ¼ 61%). In- with the rest of Dombeyoideae (Bayer and formation from the ndhF and rbcL partitions rather Kubitzki 2003). seems to favor a position of Helicteroideae as sister to Spinose pollen occurs in various lineages (e.g. most all other Malvatheca, whereas matK provides a strong Malvoideae, a few Bombacoideae) and appears to have signal for the topology as found in the present combined evolved several times in parallel, perhaps related to analysis. On the other hand, if the positions of insect . The absence of spinose pollen in Helicteroideae and Tilioideae were switched (an ar- Nesogordonia and Sicrea, both apparently relatively rangement, though, occurring in less than 10% of the early diverging branches within Dombeyoideae (Sicrea posterior distribution from the combined analysis), was not yet included in a molecular study and its then fused sepals could be a synapomorphy uniting phylogenetic relationships is here inferred from mor- Helicteroideae with Brownlowioideae, Sterculioideae, phological data), might indicate that spinose pollen is and Malvatheca. not an apomorphy of this clade but evolved later. Similarly, bifid cotyledons appear to have evolved within Dombeyoideae, with a reversal to simple in Tilioideae Pterospermum (Bayer and Kubitzki 2003). Previous molecular analyses had identified a distinct clade of Craigia (traditionally included in Sterculiaceae Helicteroideae or Tiliaceae) and Tilia (Alverson et al. 1999). The Neotropical genus Mortoniodendron was originally The subfamily Helicteroideae is a small group of thought to represent a ‘connecting link’, with respect about 12 genera and 130 species concentrated in tropical to staminal characters, between Tiliaceae and Sterculia- Australasia, but with extensions into tropical ceae (Steyermark 1938). Later, this taxon was assigned (Triplochiton and Mansonia) and two genera extending an isolated position due to the unusual feature of arillate into the Neotropics (Helicteres and Reevesia). The seeds (e.g. Miranda 1956). Bayer and Kubitzki (2003) monophyly of Helicteroideae, including Triplochiton appended Mortoniodendron as incertae sedis after and Durioneae, is well-supported (BS ¼ 99%; Tilioideae, noting that the genus might be related to PP ¼ 100%). However, this finding, which was already either Tilioideae or Brownlowioideae. The latter place- hinted at by previous analyses (Alverson et al. 1999; ment was indicated by combined analyses of atpB and Nyffeler and Baum 2000), is not well understood, due to rbcL data, although it lacked bootstrap support (Bayer the lack of any obvious common characteristics (i.e. et al. 1999). Nevertheless, inclusion of Mortoniodendron potential synapomorphies) for this clade. Whereas all in Brownlowioideae appears not satisfactory since other major clades of Malvadendrina have been Mortoniodendron differs from Brownlowioideae in the recognized in some form by previous investigators, the structure of its anthers (i.e. long parallel thecae, union of Durioneae (traditionally Bombacaceae) with connective with apical appendage) and the free sepals. disparate elements of traditional Sterculiaceae has never For this study, we generated matK and ndhF sequences ARTICLE IN PRESS 118 R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 of Mortoniodendron, and the combined analyses now Bayer et al. 1999; Bayer and Kubitzki 2003)and strongly (BP ¼ 90%; PP ¼ 99%) support inclusion of receiving further confirmation from previous molecular Mortoniodendron within Tilioideae as the sister-group to analyses (Alverson et al. 1999; Bayer et al. 1999). This the Craigia–Tilia clade (Judd and Manchester 1997; clade of 12 genera and about 400 species is cohesive, Alverson et al. 1999). Besides these three genera, no comprising primarily Old World tropical trees with others have morphological characters that would apetalous, mostly unisexual flowers, and more or less indicate that they are likely to be allied to Tilioideae. prominent androgynophores. The are composed The three genera of Tilioideae are unusual in of almost free carpels that produce one to many seeds Malvaceae in their relatively northerly distributions in per follicle. The anthers in Sterculioideae are subsessile (Mortoniodendron), China (Craigia), or sessile, a feature shared with early-diverging members and America and Eurasia (Tilia). Also, all have of Malvatheca (von Balthazar et al. 2004), but otherwise numerous, more or less free stamens, free sepals, and rare in Malvadendrina (occurring only in some ‘tilioid’ pollen (Bayer et al. 1999). According to Graham Helicteroideae). Furthermore, staminodes, which are (1979), the pollen of Mortoniodendron and Tilia differ in common in Brownlowioideae, Dombeyoideae, and size but otherwise are similar. Helicteroideae, are largely absent in Sterculioideae and The placement of Tilioideae in a clade with Brownlo- Malvatheca. These floral features lend some support to wioideae–Sterculioideae–Malvatheca is not well-sup- the hypothesis that Sterculioideae might represent the ported (PP ¼ 61%), nor is it definitive that it is the sister group to Malvatheca. Within Sterculioideae, earliest divergence within the rest of this group insufficient genera have been sampled to resolve (PP ¼ 59%). relationships among their subgroups.

Brownlowioideae Malvatheca

The monophyly of Brownlowioideae is well-sup- Monophyly of Malvatheca is well-supported in this ported (BP ¼ 100%; PP ¼ 100%), as has been undis- combined analysis (BS ¼ 100%; PP ¼ 100%). This puted since the classification by Burret (1926) and confirmation is expected given the good morphological confirmed in previous molecular studies (Alverson et al. coherence of this clade: Malvatheca are characterized by 1999; Bayer et al. 1999). This subfamily consists of eight highly modified anthers, with individual thecae being bi- genera and about 80 species, primarily from the Old or polysporangiate. Superficially comparable anther World. Brownlowioideae are well characterized by structures are found in Durioneae, but developmental stamens with thecae that are divergent at the base but investigations indicated that they derived from diffe- convergent at the top of the connective (Burret 1926; rent structural development pathways (von Balthazar Bayer et al. 1999). The sepals are fused to form a et al. 2004). In addition, Malvatheca are biogeographi- persistent, campanulate, or urceolate calyx. cally cohesive: taxa at the basal nodes on both the The internal topology of Brownlowioideae is robust malvoid and bombacoid lineage clades are Neotropical and consistent across the relevant studies (Alverson et (Baum et al. 2004) as contrasted with the Old World al. 1999; Bayer et al. 1999; Nyffeler and Baum 2000). In distribution of the other Malvadendrina clades. As particular, there appears to be a clear division between previously shown, Malvatheca are composed of two Brownlowieae (i.e. Brownlowia, Pentace, Jarandersonia, major lineages, Bombacoideae and Malvoideae. The and ) and Berryeae (i.e. Berrya, , composition of these two lineages is not easily deter- and ), although the circumscriptions of mined based on morphology, and the early molecular these tribes here differ from previous treatments (e.g. studies obtained conflicting results (Alverson et al. Hutchinson 1967). The subdivision indicated by the 1999; Bayer et al. 1999). Recent work has improved molecular data is, however, compatible with morpholo- understanding of Malvatheca (Baum et al. 2004), gical data: Brownlowieae have five antepetalous stami- indicating, for example, that Fremontodendron is nodes, whereas Berryeae have only fertile stamens. sister to the remainder of Malvatheca, but the place- Extrapolating from this single character, the single ment of a few key taxa, including Ochroma, remains unsampled genus Pityranthe, which has staminodes, is problematic. predicted to be in Brownlowieae.

Sterculioideae Conclusions

Monophyly of Sterculioideae is well-supported here Although plastid DNA has shed much light on the (BP ¼ 100%; PP ¼ 100%), agreeing with the presence relationships among the major clades of Malvaceae, of various unique morphological characteristics (see areas of uncertainty remain. The Bayesian analysis ARTICLE IN PRESS R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 119 supports one particular arrangement of lineages within relationships and morphological diversification of the Malvadendrina as being more probable a posteriori than major lineages of Malvaceae. others. However, this topology should be treated as tentative until corroborated by additional data and, ideally, until convincing morphological and/or structur- Acknowledgements al–molecular synapomorphies are discovered. In look- ing for additional data sets, it may be wise hereafter to The authors acknowledge the National Science focus on nuclear sequences. This is not so much because Foundation grant DEB-9876070 and a post-doctoral we expect there to be discrepancies between nuclear and grant by the Swiss National Science Foundation to RN plastid genealogies, although certainly we cannot rule as funding sources of this work. We also thank Stacey that out. Rather, nuclear genes, which have potentially DeWitt Smith, Steven J. Hall, Lena C. Hileman, been subjected to different evolutionary dynamics than Dianella Howarth, Tara Mehta, Ryan Oyama, and plastid genes, may resolve parts of the tree that are Karen C. Walsh for technical assistance, and Bret poorly resolved by plastid DNA (if less well resolved in Larget for providing advice on Bayesian phylogenetic other areas). Thus, it may be that once we find a nuclear analysis. Comments on the manuscript and the project region that is appropriate for this level of analysis, we in general were kindly provided by Maria von Baltha- will obtain deeper insights into the phylogenetic zar, Margaret Koopman, and Bernard Pfeil.

Appendix

Sequence data of exemplars considered for the present study. Voucher information is provided in the listed publications. All newly created trnK sequences were based on DNA material that was previously used to generate ndhF sequences (see the respective publication for voucher information), with the following exceptions: : Alverson s.n. (WIS); Hibiscus rosa-sinensis: Baum 378 (WIS); Mortoniodendron guatemalense: Lundell & Contreras 18988 (WIS); Thespesia thespesioides: Wendel 557 (ISC); Chase 3044 (K). The ndhF and trnK sequences of the following three exemplars were based on a DNA sample from Kew Gardens (voucher information in Bayer et al. 1999): Lavatera acerifolia, Pentaplaris doroteae,andSchoutenia glomerata. The rbcL sequences of the following two exemplars were based on a DNA sample from Harvard University (voucher information in Alverson et al. 1999): Craigia sp., Neesia strigosa.

Genus atpB rbcL NdhF trnK

Malvatheca L. — — AF111720 AY321168 (Alverson et al. 1999) (present study) Jum. & H. Perrier AJ233050 AJ233115 — — (Bayer et al. 1999)(Bayer et al. 1999) L. AJ233051 — — — (Bayer et al. 1999) P. Beauv. — AF233118 AF111726 AY321171 (Bayer et al. 1999)(Alverson et al. 1999) (present study) Ceiba crispiflora (H.B.K.) Ravenna [ ¼ Chorisia — — AF111730 AY321169 crispiflora H.B.K.] (Alverson et al. 1999) (present study) (A.St.-Hil., A. Juss. & Cambess.) AJF233052 AJ233116 — — Ravenna (Bayer et al. 1999,as (Bayer et al. 1999,as Chorisia speciosa) Chorisia speciosa) Fremontodendron californicum (Torr.) Cov. — — AF111721 AY321165 (Alverson et al. 1999) (present study) Fremontodendron mexicanum Davidson — AF22124 — — (Alverson et al. 1998) Fremontodendron californicum (Torr.) Cov. F. AF233077 — — — mexicanum Davidson (Bayer et al. 1999) Gossypium hirsutum L. AJ233063 — U55340 AY321158 (Bayer et al. 1999)(Seelanan et al. 1997) (present study) Gossypium robinsonii F. Muell. — L13186 — — (Chase et al. 1993) Hibiscus costatus A. Rich. — — U55323 — (Seelanan et al. 1997) ARTICLE IN PRESS 120 R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123

Appendix( continued)

Genus atpB rbcL NdhF trnK

Hibiscus punaluuenis (Skottsb.) O. Deg. & I. Deg. AJ233064 AJ233064 — — (Bayer et al. 1999)(Bayer et al. 1999) Hibiscus rosa-sinensis L. — — — AY321160 (present study) Lavatera acerifolia Cav. AJ233965 AJ233122 AY326475 AY321159 (Bayer et al. 1999)(Bayer et al. 1999) (present study) (present study) Matisia cordata Humb. & Bonpl. AJ233054 AJ233117 AF111724 — (Bayer et al. 1999)(Bayer et al. 1999)(Alverson et al. 1999) Matisia ochrocalyx K. Schum. — — — AY321167 (present study) Ochroma pyramidale (Lam.) Urb. AF035910 AJ233118 AF111727 AY321172 (Bakker et al. 1998)(Bayer et al. 1999)(Alverson et al. 1999) (present study) Aubl. AJ233056 AJ233119 AF111732 AY321170 (Bayer et al. 1999)(Bayer et al. 1999)(Alverson et al. 1999) (present study) Pentaplaris doroteae L.O. Williams & Stand. AJ233110 AJ233157 AY326476 AY321163 (Bayer et al. 1999)(Bayer et al. 1999) (present study) (present study) (L.) Correa — L01961 U55328 — (Albert et al. 1992)(Seelanan et al. 1997) Thespesia thespesiodes (Benth.) Fryxell — — — AY321161 (present study) Sterculioideae Schott & Endl. — — AF111759 AY321179 (Alverson et al. 1999) (present study) (Vent.) Schott & Endl. AJ233074 AJ233124 — — (Bayer et al. 1999)(Bayer et al. 1999) Hildegardia barteri (Mast.) Kosterm. AJ233081 AJ233131 AF111754 AY321180 (Bayer et al. 1999)(Bayer et al. 1999)(Alverson et al. 1999) (present study) (Jacq.) G. Karst AJ233089 AJ233140 — — (Bayer et al. 1999)(Bayer et al. 1999) Sterculia tragacantha Lindl. — — AF111747 AY321178 (Alverson et al. 1999) (present study) Brownlowioideae Berrya javanica (Turcz.) Burret AF035896 AJ233146 AF111755 AY321182 (Bakker et al. 1998)(Bayer et al. 1999)(Alverson et al. 1999) (present study) Brownlowia elata Roxb. AF035898 AJ233147 AF111756 AY321184 (Bakker et al. 1998)(Bayer et al. 1999)(Alverson et al. 1999) (present study) Pentace polyantha Hassk. AJ233109 AJ233156 AF111758 AY321183 (Bayer et al. 1999)(Bayer et al. 1999)(Alverson et al. 1999) (present study) Tilioideae Craigia sp. — AY864312 AF111742 AY321192 (present study) (Alverson et al. 1999) (present study) Mortoniodendron anisophyllum (Standl.) Standl. AJ233108 AJ233155 — — & Steyerm. (Bayer et al. 1999)(Bayer et al. 1999) Mortoniodendron guatemalense Standl. & AY326479 AY321190 Steyerm. (present study) (present study) L. AJ233113 AF022127 AF111760 AY321191 (Bayer et al. 1999)(Chase et al. 1993)(Alverson et al. 1999) (present study) Helicterioideae L. AJ233053 AF022119 AF111749 AY321188 (Bayer et al. 1999)(Alverson et al. 1998)(Alverson et al. 1999) (present study) Helicteres baruensis Jacq. AJ233078 AJ233127 — — (Bayer et al. 1999)(Bayer et al. 1999) Helicteres guazumaefolia H.B.K. — — AF111746 AY321186 (Alverson et al. 1999) (present study) Neesia strigosa Mast. — AY864313 AF111750 AY321189 (present study) (Alverson et al. 1999) (present study) Reevesia thyrsoidea Lindl. AJ233086 AJ233137 AF111745 AY321187 (Bayer et al. 1999)(Bayer et al. 1999)(Alverson et al. 1999) (present study) Triplochiton zambesiacus Milne-Redhead AJ233092 AJ233142 AF230256 AY321185 (Bayer et al. 1999)(Bayer et al. 1999)(Nyffeler & Baum 2000) (present study) ARTICLE IN PRESS R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 121

Appendix( continued)

Genus atpB rbcL NdhF trnK

Dombeyoideae Dombeya sp. AJ233075 AJ233123 (Bayer et al. 1999)(Bayer et al. 1999)— — Dombeya spectabilis Boj. — — AF111752 AY321173 (Alverson et al. 1999) (present study) Nesogordonia sp. — — AF111753 AY321177 (Alverson et al. 1999) (present study) Pterospermum celebium Miq. AJ233114 AJ233136 (Bayer et al. 1999)(Bayer et al. 1999) F. Muell. — — AF111748 AY321176 (Alverson et al. 1999) (present study) Schoutenia glomerata King AJ233111 AJ233159 AY326478 AY321175 (Bayer et al. 1999)(Bayer et al. 1999) (present study) (present study) Trochetiopsis erythroxylon (G. Forst.) Marais AJ233093 AJ233143 — — (Bayer et al. 1999)(Bayer et al. 1999) Trochetiopsis ebenus Cronk — — AY326477 AY321174 (present study) (present study) Outgroups: Byttnerioideae Byttneria aculeata (Jacq.) Jacq. — AJ233123 AF111775 AY321196 (Alverson et al. 1998)(Alverson et al. 1999) (present study) Byttneria filipes K. Schum. AJ233073 — — — (Bayer et al. 1999) L. AJ233090 AF022125 AF287916 AY321195 (Bayer et al. 1999)(Chase et al. 1993)(Whitlock et al. 2001) (present study) Outgroups: Grewioideae Grewia biloba G. Don — — AF111769 AY321193 (Alverson et al. 1999) (present study) L. AJ233105 AJ233152 — — (Bayer et al. 1999)(Bayer et al. 1999) L.f. — — AF11766 AY321194 (Alverson et al. 1999) (present study) Sparrmannia ricinocarpa (Eckl. & Zeyh.) Kuntze AJ233112 AJ233128 — — (Bayer et al. 1999)(Bayer et al. 1999)

References Balthazar, M. von, Alverson, W.S., Scho¨ nenberger, J., Baum, D.A., 2004. Comparative floral development and androe- Albert, V.A., Williams, S.E., Chase, M.W., 1992. Carnivorous cium structure in Malvoideae (Malvaceae s.l.). Int. J. : phylogeny and structural evolution. Science 257, Sci. 165, 445–473. 1491–1495. Barker, F.K., Lutzoni, F.M., 2002. The utility of the Alverson, W.S., Karol, K.G., Baum, D.A., Chase, M.W., incongruence length difference test. Syst. Biol. 51, 625–637. Swensen, S.M., McCourt, R., Sytsma, K.J., 1998. Circum- Barnett, L.C., 1988. Systematics of Nesogordonia Baillon scription of the Malvales and relationships to other (Sterculiaceae). Ph.D. Thesis, University of Texas, Austin. Rosidae: evidence from rbcL sequence data. Am. J. Bot. Baum, D.A., Alverson, W.S., Nyffeler, R., 1998. A by 85, 876–887. any other name: and nomenclature of the core Alverson, W.S., Whitlock, B.A., Nyffeler, R., Bayer, C., Malvales. Harvard Pap. Bot. 3, 313–330. Baum, D.A., 1999. Phylogenetic analysis of the core Baum, D.A., DeWitt Smith, S., Yen, A., Alverson, W.S., Malvales based on sequences of ndhF. Am. J. Bot. 86, Nyffeler, R., Whitlock, B.A., Oldham, R.L., 2004. Phylo- 1474–1486. genetic relationships of Malvatheca (Bombacoideae and Are` nes, J., 1959. 131e fam.: Sterculiaceae. In: Humbert, Malvoideae; Malvaceae lato) as inferred from plastid H. (Ed.), Flore de Madagascar et des Comores, DNA sequences. Am. J. Bot. 91, 1863–1871. Paris. Bayer, C., 1994. Zur Infloreszenzmorphologie der Malvales. Bakker, F.T., Vassiliades, D.D., Morton, C., Savolainen, V., Diss. Bot. 212. 1998. Phylogenetic relationships of Biebersteinia Stephan Bayer, C., 1999. The bicolor unit—homology and transforma- (Geraniaceae) inferred from rbcL and atpB sequence tion of an inflorescence structure unique to core Malvales. comparisons. Bot. J. Linn. Soc. 127, 149–158. Plant Syst. Evol. 214, 187–198. ARTICLE IN PRESS 122 R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123

Bayer, C., Kubitzki, K., 2003. Malvaceae. In: Kubitzki, K., Johnson, L.A., Soltis, D.E., 1994. matK DNA-sequences and Bayer, C. (Eds.), The Families and Genera of Vascular phylogenetic reconstruction in Saxifragaceae s.str. Syst. Plants, vol. V, Flowering Plants, Dicotyledons: Expanded Bot. 19, 143–156. Caryophyllales, Capparales and Malvales. Springer, Berlin, Johnson, L.A., Soltis, D.E., 1995. Phylogenetic inference in pp. 225–311. Saxifragaceae sensu-stricto and Gilia (Polemoniaceae) using Bayer, C., Fay, M.F., De Bruijn, A.Y., Savolainen, V., matK Sequences. Ann. Mo. Bot. Garden 82, 149–175. Morton, C.M., Kubitzki, K., Alverson, W.S., Chase, Judd, W.S., Manchester, S.R., 1997. Circumscription of M.W., 1999. Support for an expanded family concept of Malvaceae (Malvales) as determined by a preliminary Malvaceae within a recircumscribed order Malvales: a cladistic analysis of morphological, anatomical, palynolo- combined analysis of plastid atpB and rbcL DNA gical, and chemical characters. Brittonia 49, 384–405. sequences. Bot. J. Linn. Soc. 129, 267–303. Kubitzki, K., Chase, M.W., 2003. Introduction to Malvales. Burret, M., 1926. Beitra¨ ge zur Kenntnis der Tiliaceen. Notizbl. In: Kubitzki, K., Bayer, C. (Eds.), The Families and Genera Bot. Gart. Berlin-Dahlem 9, 592–880. of Vascular Plants, vol. V, Flowering Plants, Dicotyledons: Cameron, K.M., Chase, M.W., Anderson, W.R., Hills, H.G., Expanded Caryophyllales, Capparales and Malvales. 2001. Molecular systematics of Malpighiaceae: evidence Springer, Berlin, pp. 12–16. from plastid rbcL and matK sequences. Am. J. Bot. 88, Lavin, M., Thulin, M., Labat, J.-N., Pennington, R.T., 2000. 1847–1862. Africa, the odd man out: molecular biogeography of Chase, M.W., Soltis, D.E., Olmstead, R.G., Morgan, D., Les, dalbergioid legumes (Fabaceae) suggests otherwise. Syst. D.H., Mishler, B.D., Duvall, M.R., Price, R.A., Hills, Bot. 25, 449–467. H.G., Qiu, Y.-L., Kron, K.A., Rettig, J.H., Conti, E., Manos, P.S., Steele, K.P., 1997. Phylogenetic analyses of Palmer, J.D., Manhart, J.R., Sytsma, K.J., Michaels, H.J., ‘higher’ Hamamelididae based on plastid sequence data. Kress, W.J., Karol, K.G., Clark, W.D., Hedroen, M., Am. J. Bot. 84, 1407–1419. Gaut, B.S., Jansen, R.K., Kim, K.-J., Wimpee, C.F., Smith, Miranda, F., 1956. El ge´ nero Mortoniodendron y otros a´ rboles J.F., Furnier, G.R., Strauss, S.H., Xiang, Q.-Y., Plunkett, notables de las selvas del sur de Me´ xico. Ann. Inst. Biol. G.M., Soltis, P.S., Swensen, S.M., Williams, S.E., Gadek, Univ. 27, 321–336. P.A., Quinn, C.J., Eguiarte, L.E., Golenberg, E., Learn Jr., Mohana Rao, P.R., 1976. Seed and anatomy of G.H., Graham, S.W., Barrett, S.C.H., Dayanandan, S., Pterospermum acerifolium (Sterculiaceae). Phytomorphol- Albert, V.A., 1993. Phylogenetics of seed plants: an analysis ogy 26, 363–369. of nucleotide sequences from the plastid gene rbcL. Ann. Nyffeler, R., 2002. Phylogenetic relationships in the cactus Mo. Bot. Garden 80, 528–580. family (Cactaceae) based on evidence from trnK/matK and Dolphin, K., Belshaw, R., Orme, C.D.L., Quike, D.L.J., 2000. trnL-trnF sequences. Am. J. Bot. 89, 312–326. Noise and incongruence: interpreting results of the incon- Nyffeler, R., Baum, D.A., 2000. Phylogenetic relationships gruence length difference test. Mol. Phyl. Evol. 17, 401–406. of the (Bombacaceae–Durioneae or /Malvaceae/ Edlin, H.L., 1935. A critical revision of certain taxonomic Helicteroideae/Durioneae) based on chloroplast and nucle- groups of the Malvales. New Phytol. 34 (1–20), 122–143. ar ribosomal DNA sequences. Plant Syst. Evol. 224, 55–82. Farris, J.S., Ka¨ llersjo¨ , M., Kluge, A.G., Bult, C., 1994. Testing Nylander, J.A.A., 2002. MrModeltest v1.0b. Program dis- significance of incongruence. Cladistics 10, 315–319. tributed by the author. Department of Systematic Zoology, Fay, M.F., Bayer, C., Alverson, W.S., de Bruijn, A.Y., Chase, Uppsala University [download at www.ebc.uu.se/systzoo/ M.W., 1998. Plastid rbcL sequence data indicate a close staff/nylander.html] affinity between Diegodendron and Bixa. Taxon 47, 43–50. Plunkett, G.M., Soltis, D.E., Soltis, P.S., 1996. Evolutionary Felsenstein, J., 1978. The number of evolutionary trees. Syst. patterns in Apiaceae: inferences based on matK sequence Zool. 27, 27–33. data. Syst. Bot. 21, 477–495. Graham, A., 1979. Mortoniodendron (Tiliaceae) and Sphaer- Reeves, G., Chase, M.W., Goldblatt, P., Chies, T. de, Lejeune, opteris/Trichipteris (Cyatheaceae) in Cenozoic deposits of B., Fay, M.F., Cox, A.V., Rudall, P.J., 2001. Molecular the Gulf- region. Ann. Mo. Bot. Garden 66, systematics of Iridaceae: evidence from four plastid DNA 572–576. regions. Am. J. Bot. 88, 2074–2087. Hu, J.-M., Lavin, M., Wojciechowski, M.F., Sanderson, M.J., Rodrı´ guez, F., Oliver, J.L., Marı´ n, A., Medina, J.R., 1990. 2000. Phylogenetic systematics of the tribe Millettieae The general stochastic model of nucleotide substitution. (Leguminosae) based on chloroplast trnK/matK sequences J. Theor. Biol. 142, 485–501. and its implications for evolutionary patterns in Papilio- Schumann, K., 1897. Tiliaceae. In: Engler, A., Prantl, K. noideae. Am. J. Bot. 87, 418–430. (Eds.), Die Natu¨ rlichen Pflanzenfamilien, Nachtra¨ ge zu 3, Huelsenbeck, J.P., Ronquist, F., 2001. MrBayes: Bayesian 6. Leipzig, pp. 232–234. inference of phylogenetic trees. Bioinformatics 17, 754–755 Seelanan, T., Schnabel, A., Wendel, J.F., 1997. Congruence [download at http://morphbank.ebc.uu.se/mrbayes/index. and consensus in the tribe (Malvaceae). Syst. Bot. html]. 22, 259–290. Huelsenbeck, J.P., Bull, J.J., Cunningham, C.W., 1996. Soltis, D.A., Soltis, P.S., 1998. Choosing an approach and an Combining data in phylogenetic analysis. Trends Ecol. appropriate gene for phylogenetic analysis. In: Soltis, P.E., Evol. 11, 152–158. Soltis, P.S., Doyle, J.J. (Eds.), Molecular Systematics of Hutchinson, J., 1967. The Genera of Flowering Plants Plants II: DNA Sequencing. Kluwer Academic Publishers, (Angiospermae), vol. II. Clarendon Press, Oxford. Boston, pp. 1–42. ARTICLE IN PRESS R. Nyffeler et al. / Organisms, Diversity & Evolution 5 (2005) 109–123 123

Steyermark, J.A., 1938. Mortoniodendron, a new genus of Venkata Rao, C., 1952. Floral anatomy of some Malvales and Tiliaceae. Field Mus. Nat. Hist. Bot. Ser. 17, 411–415. its bearing on the affinities of families included in the order. Swofford, D.L., 2002. PAUP*: Phylogenetic Analysis Using J. Indian Bot. Soc. 31, 171–203. Parsimony (* and Other Methods), Version 4. Sinauer Whitlock, B.A., Bayer, C., Baum, D.A., 2001. Phylogenetic Associates, Sunderland, MA. relationships and floral evolution of the Byttnerioideae Takhtajan, A., 1997. Diversity and Classification of Flowering (‘‘Sterculiaceae’’ or Malvaceae s.l.) based on sequences of Plants, New York. the chloroplast gene ndhF. Syst. Bot. 23, 420–437.