Botanical Journal of the Linnean Society (1999), 129: 267–303. With 4 figures Article ID: bojl.1998.0226, available online at http://www.idealibrary.com on Support for an expanded family concept of Malvaceae within a recircumscribed order Malvales: a combined analysis of plastid atpB and rbcL DNA sequences CLEMENS BAYER,1∗ MICHAEL F. FAY,2 ANETTE Y. DE BRUIJN,2 VINCENT SAVOLAINEN,3 CYNTHIA M. MORTON,4 KLAUS KUBITZKI,1 WILLIAM S. ALVERSON,5 MARK W. CHASE2 1 Universität Hamburg, Institut für Allgemeine Botanik und Botanischer Garten, Ohnhorststrabe 18, 22609 Hamburg, Germany 2 Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3DS 3 Conservatoire et Jardin Botaniques, 1292 Geneva & IBSG, University of Lausanne, 1015 Lausanne, Switzerland 4 Department of Botany, University of Reading, Reading RG6 2AS 5 Harvard University Herbaria, 22 Divinity Avenue, Cambridge, MA 02138, U.S.A. Received April 1998; accepted for publication September 1998 Sequence analyses of the plastid genes atpB and rbcL support an expanded order Malvales. Within this alliance, core Malvales are clearly supported and comprise most genera that have previously been included in Sterculiaceae, Tiliaceae, Bombacaceae, and Malvaceae. Additional well supported malvalean alliances include the bixalean clade (Bixaceae, Diegodendraceae, and Cochlospermaceae), the cistalean clade (Cistaceae, Dipterocarpaceae, and Sarcolaenaceae) and Thymelaeaceae (including Gonystyloideae and Aquilarioideae). Our results indicate sister-group relationships between (1) Neuradaceae and the cistalean clade; (2) Sphaerosepalaceae and Thymelaeaceae; (3) these two clades (1 and 2); and (4) all these and an alliance comprising the bixalean clade and core Malvales, but this pattern is weakly supported by the bootstrap. The affinities of Muntingiaceae and Petenaea are especially ambiguous, although almost certainly they are Malvales s.l. The traditional delimitation of families within core Malvales is untenable. Instead, we propose to merge Sterculiaceae, Tiliaceae and Bombacaceae with Malvaceae and subdivide this enlarged family Malvaceae into nine subfamilies based on molecular, morphological, and biogeographical data: (1) Byttnerioideae, including tribes Byttnerieae, Lasiopetaleae and Theobromeae (all of which have cucullate petals) and Hermannieae; (2) Grewioideae, including most genera of former Tiliaceae; (3) Tilioideae, monogeneric in our analysis; (4) Helicteroideae, comprising most of the taxa previously included in Helictereae, plus Mansonia, Triplochiton (indicating that apocarpy evolved at least twice within Malvaceae) and possibly Durioneae; (5) Sterculioideae, defined by apetalous, apocarpous, usually unisexual flowers with androgynophores; (6) Brownlowioideae, circumscribed as in previous classifications; (7) Dombeyoideae, expanded to include Burretiodendron, Eriolaena, Pterospermum, and Schoutenia; (8) Bombacoideae, corresponding to former Bombacaceae (without Durioneae) but including Fremontodendreae ∗ Corresponding author. Email: c.bayer@botanik.uni-hamburg.de 0024–4074/99/040267+37 $30.00/0 267  1999 The Linnean Society of London C. BAYER ET AL. 268 and Pentaplaris; (9) Malvoideae, monophyletic but difficult to delimit from Bombacoideae, which with more data and taxon sampling than here might prove to be paraphyletic without Malvoideae.  1999 The Linnean Society of London ADDITIONAL KEY WORDS:—Sterculiaceae – Tiliaceae – Bombacaceae – Bixales – Cistales – apocarpy – pollen – molecular systematics. CONTENTS Introduction . . . . . . . . . . Material and methods . . . . . . DNA extraction . . . . . . . Amplification and sequencing of atpB Data analysis . . . . . . . . Results . . . . . . . . . . . Analysis of rbcL . . . . . . . Analysis of atpB . . . . . . . Combined analysis . . . . . . Discussion . . . . . . . . . . Malvales sensu lato . . . . . . Circumscription of core Malvales . Subdivision of core Malvales . . . Conclusion . . . . . . . . . . Acknowledgements . . . . . . . References . . . . . . . . . . Appendix 1 . . . . . . . . . . Appendix 2 . . . . . . . . . . . . . . . . . . . and rbcL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268 269 270 270 271 272 272 274 275 277 277 279 280 290 290 290 296 299 INTRODUCTION Order Malvales has been variously defined: narrow circumscriptions have restricted it to Tiliaceae, Sterculiaceae, Bombacaceae, Malvaceae and Elaeocarpaceae (Cronquist, 1988), but many authors have included additional families such as Bixaceae, Cistaceae, Cochlospermaceae, Diegodendraceae, Dipterocarpaceae, Dirachmaceae, Huaceae, Peridiscaceae, Plagiopteraceae, Sarcolaenaceae, Scytopetalaceae, Sphaerosepalaceae and Thymelaeaceae (e.g. Dahlgren, 1983; Huber, 1991; Thorne, 1992; Takhtajan, 1997). It has become apparent that the closely related Tiliaceae, Sterculiaceae, Bombacaceae, and Malvaceae constitute the monophyletic core Malvales, if putative relatives such as Elaeocarpaceae, Flacourtiaceae, Muntingiaceae and Neuradaceae are removed (Chase et al., 1993, and unpubl.; Judd & Manchester, 1997; Fay et al., 1998a; Bayer, 1999 and unpubl.; Bayer, Chase & Fay, 1998; Alverson et al., 1998). The expanded order Malvales includes some but not all of the families mentioned above. Previous morphological and molecular studies support the exclusion of Elaeocarpaceae, Flacourtiaceae, Huaceae, Scytopetalaceae, Dirachmaceae, Peridiscaceae, and Plagiopteraceae (Chase et al., 1993, and unpubl.; Appel, 1996; Morton et al., 1996; 1997; Fay et al., 1998a; Alverson et al., 1998; Thulin et al., 1998; C. Bayer, unpubl.). Although the component families of the expanded Malvales have been identified in previous studies, the interrelationships between malvalean families, especially within core Malvales, are largely unknown. This is largely a result of the problematic delimitation of the core families, which appear to be based on tradition rather than MOLECULAR SYSTEMATICS OF MALVALES 269 characters. The limits of Malvaceae/Bombacaceae, Sterculiaceae/Tiliaceae and Sterculiaceae/Bombacaceae remain nebulous. As a consequence, taxa such as Fremontodendreae, Gossypieae, Hibisceae, Corchoropsis, and Nesogordonia have been moved between families. Especially Tiliaceae in their traditional circumscription (e.g. de Candolle, 1824; Bentham & Hooker, 1862; Bocquillon, 1866; Baillon, 1873; von Szyszylowicz, 1885; Hutchinson, 1967) have been used for filing genera that could not be otherwise placed. Even authors who had a narrower and more critical concept of Tiliaceae (e.g. Burret, 1926) included aberrant genera such as Neotessmannia (Bayer et al., 1998). Ongoing morphological studies (C. Bayer, unpubl.) indicate that many Tiliaceae s.l. such as Burretiodendron, Schoutenia, and even Tilia are more closely related to Sterculiaceae than to the grewioid alliance, which includes the majority of Tiliaceae. However, all such considerations remain vague because none of these families is unambiguously defined by morphological characters. The distribution of distinctive characters derived from inflorescence, flower, and pollen morphology is only partly consistent with the traditional classifications. For instance, spinose or spinulose pollen occurs in all four families of core Malvales (Erdtman, 1952), and it is not known to what extent this scattered distribution is due to common ancestry or to parallel evolution; the same applies to the occurrence of an epicalyx (Bayer, 1999). Sterculiaceae, to mention another example, have been subdivided into subfamilies Sterculioideae and Byttnerioideae (Thorne, 1992; Takhtajan, 1997; or even elevated as families: Edlin, 1935) on the basis of the apetalous, unisexual, and apocarpous flowers of Sterculioideae. If secondary apocarpy has evolved only once within Sterculiaceae, genera such as Helicteres, Mansonia, and Triplochiton would have to be referred to Sterculioideae on the basis of their apocarpous gynoecia, even though they exhibit hermaphrodite flowers with petals, which are not found in the other apocarpous genera. In view of the difficulties with the distribution of morphological characters, the present study based on DNA sequences was undertaken as an additional approach to clarify the interrelationships among malvalean families. We decided to analyse two plastid genes, rbcL and atpB, since in some cases the latter provides greater resolution than rbcL (Hoot, Culham & Crane, 1995), and a combination of data sets may increase support of clades (as estimated by the bootstrap, Felsenstein, 1985; Soltis et al., 1997). MATERIAL AND METHODS As is evident from previous molecular studies (Chase et al., 1993; Gadek et al., 1996; Fay et al., 1998a; Alverson et al., 1998), Malvales are related to expanded Capparales and Sapindales, which have been used as outgroups in the present study. To cover the major lineages within Malvales we included representatives of most known or suspected alliances as well as some isolated taxa of unknown affinities; this was restricted by the availability of suitable plant material. For most taxa, the same samples were used to sequences both atpB and rbcL (Appendix 1). On certain occasions, however, we combined sequences of the respective genes obtained from different DNA samples or closely related taxa in the combined data set. Full names, authorities, sources, vouchers and database accessions are listed in Appendix 1. 270 C. BAYER ET AL. DNA extraction Leaf tissue, flowers or seeds from fresh or silica gel dried material, or leaf fragments from herbarium specimens were used. Total genomic DNA was extracted using a modified 2×CTAB method (based on Doyle & Doyle, 1987). Since we had difficulties obtaining DNA even from fresh material, DNA precipitation with ethanol or isopropanol was generally extended to about one month at −20°C (Fay et al., 1998a). Due to the mucilage content of many samples, it was sometimes difficult to remove the mucilaginous supernatant after centrifugation without losing most of the DNA. DNA was purified by ultracentrifugation on a CsCl2-ethidium bromide density gradient (1.55 g/ml) followed by dialysis. Subsequently, total DNA samples were purified using QIAquick silica columns (Qiagen, Ltd., Crawley, U.K.) according to the manufacturer’s protocols. We took this additional step because amplification of atpB was generally more difficult than amplification of rbcL, which succeeded without problems from the same total DNA samples. Following purification on QIAquick columns, success with atpB was much more consistent; we suspect that phenolic compounds, which are known to be inhibitors of DNA polymerases, are present, but why such differential effects should occur is not clear. For a few samples, especially some of those extracted from herbarium material, the usual protocol including precipitation, gradient centrifugation and dialysis was replaced by using QIAquick columns to purify the raw extract directly, after treatment with chloroform/isoamyl alcohol (24:1) to remove proteins ( J. Ronnholm, University of Uppsala, pers. comm.). Again, due to the high mucilage content of some samples, prolonged and repeated centrifugation of the columns was required. Amplification and sequencing of atpB and rbcL PCR amplification of rbcL was generally performed in two overlapping segments, using synthetic primers that anneal at base positions 1 and 636 (forward), and 724 (reverse) and a downstream ribosome control site (Olmstead et al., 1992; Fay et al., 1998a). In contrast, atpB was usually amplified in a single piece from position 2 (forward) to 1494 (reverse). The internal primers start at position 611 (forward) and 766 (reverse), respectively (Hoot et al., 1995). Bovine serum albumin (1–4 % of 0.4 % aqueous solution; Savolainen et al., 1995) was added to all PCR reactions to bind phenolic compounds. PCR products were purified using Wizard minicolumns (Promega U.K., Ltd., Southampton, U.K.) according to the manufacturer’s protocol. Modified dideoxy cycle sequencing with dye terminators was used to produce the new sequences presented here (Perkin-Elmer Applied Biosystems, Inc., Warrington). In most cases, the total reaction volume suggested by the manufacturer (20 ll) was reduced to 5 ll. Diluting the reaction products with 15 ll water prior to precipitation and cleaning considerably improved the quality of the first part of each sequence by eliminating most or all of the unincorporated dye terminators. Products for both strands were sequenced directly on an automated sequencer (ABI 377, Perkin-Elmer Applied Biosystems, Inc.) following the manufacturer’s instructions. Individual strands were edited and assembled using Sequence Navigator and Auto Assembler (PerkinElmer Applied Biosystems, Inc.). MOLECULAR SYSTEMATICS OF MALVALES 271 Data analysis For each of the three matrices (rbcL, atpB and combined) under equal weights (Fitch parsimony; Fitch, 1971), we performed 1000 replicates of random taxonaddition with TBR (tree bisection-reconnection) branch swapping, which is the most thorough swapping algorithm of PAUP 3.1.1 (Swofford, 1993). In these replicates, we limited the number of trees being held at each step to ten so that excessive amounts of time were not spent swapping on suboptimal island, but this method of search could not find all most parsimonious trees or detect islands of equally parsimonious trees (Maddison, 1991). We therefore used all trees collected in the 1000 replicates as starting trees for a more complete search with a tree limit of 5000. We limited this search for rbcL and atpB due to computer memory limitations, and thus we could not determine how many trees in total existed at the shortest tree-lengths found. Using five single trees found in separate replicates as starting trees, we were able to find all 5000 trees, thus indicating that islands were not present. Following these searches under the Fitch criterion, we performed successive approximations weighting (SW; Farris, 1969) to reduce or eliminate the effects of positions that change frequently. In most cases, the effect of SW is to reduce tree number, but not always (Fay et al., 1998b). Using all trees found in the Fitch search (above), we reweighted characters using the menu command in PAUP 3.1.1 with the following settings: a base weight of 1000 and best fit for each character based on the rescaled consistency index (RC). Using the consistency (CI) or the retention index (RI) has no effect on the topologies produced with these matrices. Rounds of search followed by re-weighting were performed until branch lengths were the same in two consecutive searches. Each round consisted of ten replicates of random taxonaddition with TBR swapping and a tree limit of 50 trees. All trees were then swapped on to completion (up to 5000 trees) before preceeding to the next round of SW. Trees favoured by SW are never radically different from those found by Fitch analysis (equal weights), and this is the case here (Table 1). Internal support was evaluated with the bootstrap (Felsenstein, 1985). Use of extensive swapping makes this process very slow, so we made use of a modified procedure that is much faster but not much less accurate. We performed 5000 replicates of bootstrapping with nearest-neighbour interchange (NNI) swapping, but permitting only ten trees to be held per replicate. The greatest effect here results from group presence in the starting trees (which are the result of quick distance calculations), but we have found that some minimal amount of swapping improves the estimates for larger clades (which are under-estimated when only ‘no swapping’ is employed). We have compared, on smaller matrices, the effects of such minimal swapping on bootstrap percentages with those found with extensive TBR swapping, and they are highly correlated. Thorough bootstrapping with TBR and no tree limit is not practical for such large matrices because it would require several months of analysis. If there is any effect from this faster method, it would only be an underestimate rather than an exaggeration of support. Patterns of sequence evolution were estimated using MacClade (Maddison & Maddison, 1992) from matrices stripped to only the base positions included in the analyses. Because we believe that the SW combined tree is the most accurate (due to higher overall levels of bootstrap support), we assessed the evolution of each gene on this tree rather than on the trees produced in the separate rbcL and atpB analyses. C. BAYER ET AL. 272 T 1. Statistical values from our analysis of rbcL and atpB gene sequences (CI: consistency index, RI: retention index) Invariant sites Uniquely variable sites Informative sites Transitions CI RI Transversions CI RI Transition/transversion ratio Number of equally parsimonious Fitch trees Fitch tree length CI RI Number of SW trees SW tree length CI RI Fitch length of SW tree CI RI Steps, first position (% assessed on the combined tree) CI RI Steps, second position (% assessed on the combined tree) CI RI Steps, third position (% assessed on the combined tree) CI RI rbcL atpB 946 (67.62%) 144 (10.29%) 309 (22.09%) 930 0.42 0.71 684 0.38 0.57 1.36 >5000 1619 0.40 0.66 1509 434569 0.72 0.85 1626 0.40 0.66 375 0.31 0.54 161 (9.98%) 0.40 0.57 1078 (66.79%) 0.43 0.68 914 (63.69%) 209 (14.56%) 312 (21.74%) 1002 0.41 0.73 454 0.58 0.64 2.21 >5000 1435 0.50 0.69 >5000 492134 0.77 0.84 1438 0.50 0.69 (23.23%) 0.51 0.60 139 (9.55%) 0.59 0.52 1068 (73.35%) 0.47 0.71 combined >5000 3066 0.44 0.66 81 912505 0.75 0.84 3070 0.44 0.66 249 (17.10%) To calculate the number of transitions and transversions observed on one of the shortest combined SW trees (as well as their Cls and RIs), we used the following step matrix to calculate the number of transversions at each base position: A C G T A C G T – 1 0 1 1 – 1 0 0 1 – 1 1 0 1 – From this number of transversions and their collective CI and RI, we could calculate those of transitions. RESULTS Analysis of rbcL Due to the position of the forward PCR primer, we deleted the first 29 bases of the total 1428 bases from our rbcL matrix and used 1399 characters, of which 453 MOLECULAR SYSTEMATICS OF MALVALES 273 A 20 18 16 rbcL 14 Steps 12 10 8 6 4 2 0 B 16 200 400 600 800 1000 1200 Site 14 12 atpB Steps 10 8 6 4 2 0 200 400 600 800 1000 1200 Site Figure 1. Number and distribution of base changes in (A) rbcL and (B) atpB (produced by MacClade; Maddison & Maddison, 1992). These are optimized substitutions on one of the 81 combined SW trees. were variable and 309 were potentially informative. Heuristic search under the Fitch criterion yielded more than 5000 equally parsimonious trees of 1619 steps with a CI of 0.40 and a RI of 0.66. Successive weighting produced 1509 trees of 434569 steps (CI 0.72, RI 0.85), which corresponds to a Fitch length of 1626 steps (CI 0.40, RI 0.66; trees not shown). Changes in rbcL are not particularly clustered; the abundance and distribution of substitutions are shown in Figure 1A. The transition/ transversion ratio was 930/684 (1.36). Although more numerous, transitions had both a higher consistency index (CI) and retention index (RI) than transversions (0.42 versus 0.38 and 0.71 versus 0.57, respectively; Table 1). Third positions contributed the most steps (66.79 % as assessed on the combined tree) and had the highest CI and RI, whereas the CI and RI of the first positions were lowest (Table 1). Malvales s.l. form a clade in all most parsimonious Fitch trees found, but their monophyly is not supported by the bootstrap. There is strong support for core Malvales (bootstrap values: 99 % with Fitch weights, 100 % with SW), Muntingiaceae (97/96), Thymelaeaceae (including Aquilaria and Gonystylus; 97/99) and a clade 274 C. BAYER ET AL. including Cistaceae, Dipterocarpaceae and Sarcolaenaceae (99/100). Bixa is sister to Diegodendron (96/97), but their sister-group relationship with Cochlospermaceae is only weakly supported (73/85). Within core Malvales, only a few clades are well supported, such as the Dombeya alliance (Dombeya, Helmiopsiella, Eriolaena, Ruizia, Trochetiopsis and Paramelhania; 100/100), Brownlowia and Pentace (100/100), Gossypium and Thespesia (90/93), Hibiscus and Pavonia (100/100), Helicteres and Triplochiton (81/ 92) and Thomasia and Lasiopetalum (96/100). Tilia is sister to all other core Malvales, but this position is not supported by the bootstrap. There is some support for a clade including Grewia, Colona, Microcos and Goethalsia, if SW is applied (-/88). The byttnerioid alliance (see discussion) is neither resolved nor supported. Analysis of atpB The beginning and the end of each sequence were not reliable due to the annealing positions of the PCR primers, so we cut the first 34 and the last 28 bases of the total 1497. We thus used 1435 base pairs of atpB in the analysis, of which 521 were variable and 312 were potentially informative. A single six-base deletion was detected in the atpB gene of Sparrmannia, corresponding to AGATAG starting at position 156 in the Nicotiana reference sequence (GenBank accession X61319). Insertions in the atpB sequence as compared to the Nicotiana reference sequence were found in the following taxa: CTTAG starting at position 1472 in Heliocarpus, TAGAA starting at position 1474 in Lavatera, GGA starting at position 1098 in Aquilaria. All these were deleted from the PAUP matrix because they were all unique to single taxa. Heuristic search under the Fitch criterion yielded more than 5000 equally parsimonious trees of 1435 steps (CI 0.50, RI 0.69). Successive weighting produced more than 5000 trees of 492134 steps (CI 0.77, RI 0.84), which corresponds to a Fitch length of 1438 steps (CI 0.50, RI 0.69; trees not shown). The abundance and distribution of base changes are shown in Figure 1B. The transition/transversion ratio was 1002/454 (2.21). As with rbcL, transitions were more numerous; their CI was lower than those for transversions, but the RI was higher (Table 1). Third positions contributed the most steps (73.35 % as assessed on the combined tree) and had the lowest CI and the highest RI, whereas the second positions had the highest CI and the lowest RI (Table 1). Malvales s.l. are supported by atpB data (bootstrap values: 84 % with Fitch weights, 95 % with SW). Well supported clades within Malvales s.l. include Muntingiaceae (96/96), Thymelaeaceae (including Aquilaria and Gonystylus; 98/100), Cistaceae, Dipterocarpaceae and Sarcolaenaceae (99/100) and Bixa with Diegodendron (90/97). The sister-group relationship between Neurada and the remaining Malvales is not supported (-/52). There is no support for the monophyly of core Malvales (-/53). Within core Malvales (68/53), the Dombeya alliance (as mentioned for the rbcL analyses but without Helmiopsiella, for which atpB was not available) is well supported (100/100). Its sister-group relationship with a clade comprising Pterospermum, Schoutenia and Burretiodendron (78/83) is also supported by the bootstrap (62/93). Other clades include Abroma and Byttneria (81/88), Lasiopetalum and Thomasia (80/77) and Pavonia and Hibiscus (60/87). Malvaceae (Gossypium, Hibiscus, Pavonia and Lavatera) are only very weakly supported (-/69). There is some support for an enlarged Grewia alliance (as above for rbcL plus Apeiba, Sparrmannia and Heliocarpus; 63/81). The sterculioid genera (Sterculia, Hildegardia, Cola) form a weakly supported clade (53/69). The MOLECULAR SYSTEMATICS OF MALVALES 275 brownlowioid alliance (Christiana, Berrya, Brownlowia and Pentace) is even more weakly supported (-/58). The clade formed by Christiana and Berrya has slightly stronger support (56/78). Combined analysis The combined matrix under the Fitch criterion yielded more than 5000 equally parsimonious trees of 3066 steps (CI 0.44, RI 0.66), the strict consensus tree of which (for core Malvales) is shown in Figure 2. Successive weighting of the combined data set produced 81 trees of 912505 steps (CI 0.77, RI 0.84), which corresponds to 3070 Fitch steps (CI 0.44, RI 0.66). A randomly selected tree (of the 81 SW trees) is shown in Figure 3 (Malvales s.l., Sapindales, Capparales) and Figure 4 (core Malvales). Branch lengths are shown above the branches (ACCTRAN optimization), bootstrap values (equal/SW) are indicated below the branches (to save space in Figures 3 and 4, we indicated only 99 if support was 99 or 100%; if values for both Fitch and SW were 99 or 100%, we indicated a single 99). Branches not present in the SW strict consensus tree are marked with arrows. Expanded Malvales (68/58) comprise core Malvales (Malvaceae s.l., see below; 99/100), which are the sister group to the bixalean clade (Bixaceae, Diegodendraceae, Cochlospermaceae; 85/ 81), and these in turn are sister to a larger clade in which Thymelaeaceae (including Gonystylus and Aquilaria; 100/100) and Sphaerosepalaceae together are the sister group to the cistalean clade (Cistaceae, Dipterocarpaceae, Sarcolaenaceae and Neuradaceae). The outlier to this whole assemblage is a clade comprised of Muntingiaceae and Petenaea. However, the interrelationships between the clades mentioned, as well as the sister-group relationships between Sphaerosepalaceae/ Thymelaeaceae (-/70) and Neuradaceae and the cistalean clade (-/67), are either unsupported or only weakly supported. The positions of Muntingiaceae and Petenaea are especially unstable. Core Malvales (Malvaceae s.l. in Fig. 3) are strongly supported by our data. Hibiscus, Pavonia, Gossypium, and Lavatera, which represent Malvaceae s.s. in the combined analysis, form the only monophyletic clade (56/83) that corresponds to a traditionally circumscribed family (Figs 2, 4). None of other core Malvales families is indicated to be monophyletic: Bombacaceae are unresolved but are paraphyletic with respect to Malvaceae, and those alliances or single genera usually classified as Tiliaceae or Sterculiaceae are interwoven (Fig. 2). This is also evident from the separate analyses of both atpB and rbcL sequences (not shown), which yield similar topologies but are generally more weakly supported. Since the commonly applied family names are not suitable to describe most of the clades shown in Figure 4, we refer here to the alliances and their names as outlined in the discussion: Bombacoideae (without Pentaplaris, not supported by the bootstrap); Dombeyoideae (72/96); Sterculioideae (-/69); Brownlowioideae (93/94), sister to Mortoniodendron (50/74); Helicteroideae (74/91), sister to Durio (55/56); Grewioideae (70/78); and Byttnerioideae (not supported) including Hermannia with Theobroma (not supported), Lasiopetaleae (55/64) and Abroma with Byttneria (86/96). Malvoideae, Bombacoideae, Dombeyoideae, Sterculioideae, Brownlowioideae, Helicteroideae and Tilia form a clade (74/94), in which Tilia is sister to the remaining taxa (-/57). Most other sister-group relationships among the clades mentioned as well as those within Byttnerioideae have no bootstrap support. 276 C. BAYER ET AL. Gossypium Thespesia Lavatera Hibiscus Pavonia Pentaplaris Adansonia Ochroma Fremontodendron Bombax Matisia Chorisia Pachira Dombeya Helmiopsiella Eriolaena Paramelhania Trochetiopsis Ruizia Burretiodendron Pterospermum Schoutenia Sterculia Hildegardia Cola Mortoniodendron Christiana Berrya Pentace Brownlowia Helicteres Triplochiton Reevesia Durio Tilia Keraudrenia Thomasia Lasiopetalum Rulingia Colona Microcos Goethalsia Grewia Heliocarpus Apeiba Sparrmannia Hermannia Theobroma Abroma Byttneria Leptonychia Malvaceae Tiliaceae Bombacaceae Sterculiaceae Bombacaceae Sterculiaceae Tiliaceae Sterculiaceae Tiliaceae Sterculiaceae Tiliaceae Sterculiaceae Bombacaceae Tiliaceae Sterculiaceae Tiliaceae Sterculiaceae outgroups Figure 2. Core Malvales in the strict consensus tree of more than 5000 equally parsimonious Fitch trees (equal weight) obtained by heuristic search using the combined rbcL and atpB data set (no atpB data for Helmiopsiella and Thespesia; length 3066 steps, CI = 0.44, RI = 0.66). Family affiliation of genera according to Brummitt (1992). MOLECULAR SYSTEMATICS OF MALVALES 33 99 10 –/56 10 –/54 59 99 18 –/97 24 Bixa Bixaceae 21 Diegodendron Diegodendraceae Cochlospermum Cochlospermaceae 99 43 99 83 99 18 Helianthemum 39 Tuberaria 55 39 Anisoptera 47 23 7 –/70 15 84/ 98 17 99 Petenaea 45 Muntingia 39 Aesculus Hippocastanaceae 33 Acer Aceraceae Koelreuteria Sapindaceae 15 Rhus 72 52 55 26 Pistacia Anacardiaceae Schinus Reseda Resedaceae Capparis Capparaceae Stanleya Brassica Brassicaceae Floerkea Limnanthaceae Carica Caricaceae Tropaeolum Tropaeolaceae Capparales 27 98/99 19 86/ 92 33 44 99 Muntingiaceae 56 39 35 99 incertae sedis Dicraspidia 17 6 73/65 17 21 93/ 99 Thymelaeaceae 23 56 52 97/ 99 Phaleria Sapindales 13 80/ 97 42 99 44 99 Sphaerosepalaceae Thymelaea 34 38 99 Rhopalocarpus Dais 53 37* –/– Neuradaceae Aquilaria 17 6 19 99 15 86/ 99 Sarcolaenaceae Neurada Gonystylus 40 44 99 Dipterocarpaceae Sarcolaena Malvales 85 5 –/53 Cistaceae Cistus 42 22 –/67 29 68/ 58 Malvaceae s.l. 12 99 31 85/ 81 277 Figure 3. Malvales, Sapindales and Capparales in one of the 81 equally parsimonious trees selected at random from heuristic search of the combined data set (no atpB data for Petenaea) using successive weighting (SW tree length 912505, CI = 0.44, RI = 0.66; Fitch length 3070, CI = 0.44, RI = 0.66). Branch lengths are indicated above the branches (ACCTRAN optimization; note that the branch length marked with an asterisk is not comparable to the others, since atpB data is lacking for Petenaea), numbers below the branches represent bootstrap values without/with SW. Bootstrap percentage of 99 or more is marked as 99; if both values are 99 or more, only a single 99 is given. All branches in this portion of the strict consensus tree are resolved. Malvaceae s.l. (core Malvales) are shown in Figure 4. DISCUSSION Malvales sensu lato The content and relationships of Malvales s.l. have been discussed elsewhere (Fay et al., 1998a, Bayer et al., 1998, Alverson et al., 1998), but some differences between C. BAYER ET AL. 278 5 –/– 3 –/67 38 17 19 99 7 –/– 1 –/– 3 –/64 1 –/– 29 7 –/– 9 56/83 35 9 12 4 9 1 –/– 3 –/– 11 6 3 –/– 1 –/– 13 13 5 65/86 23 99 5 77/80 3 1 52/– 12 –/53 4 14 72/96 4 2 9 12 6 75/87 3 –/– 12 16 3 –/69 5 –/57 11 50/74 11 74/94 7 4 –/– 10 2 –/– 4 93/94 14 8 80/96 5 4 18 99 4 8 11 84/98 6 74/91 3 55/56 6 23 25 27 31 3 –/– 7 70/78 5 57/81 3 –/– –/– 18 13 26 54 6 –/– 1 14 47 5 57/75 3 –/– 33 99 5 9 69/68 7 70/78 6 24 62/98 19 17 6 21 55/64 2 –/– 18 99 8 12 86/96 8 22 14 28 Gossypium Lavatera MALVOIDEAE Hibiscus Pavonia Pentaplaris Adansonia Chorisia Pachira BOMBACOIDEAE Fremontodendron Ochroma Bombax Matisia Dombeya Eriolaena Paramelhania Trochetiopsis DOMBEYOIDEAE Ruizia Burretiodendron Pterospermum Schoutenia Sterculia Cola STERCULIOIDEAE Hildegardia Mortoniodendron (incertae sedis) Christiana Berrya BROWNLOWIOIDEAE Pentace Brownlowia Triplochiton Helicteres HELICTEROIDEAE Reevesia Durio (incertae sedis) Tilia TILIOIDEAE Colona Microcos Goethalsia GREWIOIDEAE Grewia Heliocarpus Apeiba Sparrmannia Hermannia Theobroma Keraudrenia Rulingia BYTTNERIOIDEAE Thomasia Lasiopetalum Abroma Byttneria Leptonychia (incertae sedis) Figure 4. Malvaceae s.l. (core Malvales) portion of the same single tree as in Fig. 3. Branch lengths are indicated above the branches (ACCTRAN optimization), numbers below the branches represent bootstrap values without/with successive weighting. Bootstrap support of 99 or more is marked as 99; if both values are 99 or more, only a single 99 is given. Branches not present in the strict consensus of the SW trees are indicated by arrows; for the strict Fitch consensus tree, see Figure 2. The suprageneric names given on the right margin correspond to the preliminary classification proposed here (see text). MOLECULAR SYSTEMATICS OF MALVALES 279 these and this study need to be addressed. Based on rbcL sequence data, Rhopalocarpus (Sphaerosepalaceae) falls either in the bixalean clade (Alverson et al., 1998) or is sister to Thymelaeaceae (Fay et al., 1998a), although neither placement is well supported. The latter placement is supported here in the combined data set, although not strongly. Neurada, which agrees with many Malvales in the lysigenous mucilage canals, exotegmic seed coat and cyclopropene acids in the seed oil (Huber, 1993), is either sister to the cistalean clade (our Fig. 3; Fay et al., 1998a; Alverson et al., 1998: fig. 2) or represents the sister group of all other Malvales (Alverson et al., 1998: fig. 3). Muntingiaceae belong to the cistalean clade according to rbcL analyses (Fay et al., 1998a; Alverson et al., 1998), but in our combined analysis they are (together with Petenaea, see below) sister to the remaining Malvales. However, this topology is not supported by the bootstrap (Fig. 3). These taxa require further research to elucidate their relationships with confidence. Petenaea, which has not been included in previous phylogenetic studies, merits special attention. The monotypic genus from southern Mexico, Guatemala and Belize was originally placed in Elaeocarpaceae (Lundell, 1962). The suggestion that Petenaea should be placed in Elaeocarpaceae was claimed to be supported by anatomical studies, which also emphasized the differences between Petenaea and Muntingia (Kukachka, 1962; Gasson, 1996). Petenaea is characterized by multicellular simple or branched trichomes, palminerved, cordate leaves with minute stipules, tetra- or pentamerous flowers without petals but with moniliform trichomes, receptacular glands that alternate with the staminal filaments, dorsifixed anthers that open by apical slits, prolate and tricolporate pollen with microperforate tectum, massive axile placentae with numerous ovules and baccate fruits (C. Bayer, pers. obs.). The structure of the ovary is especially reminiscent of Muntingiaceae, which are close to Petenaea according to our results (Fig. 3). Even though this position is not supported by the bootstrap it seems to be the least inconvenient option to treat Petenaea as tentatively related to Muntingiaceae. Circumscription of core Malvales Core Malvales are well supported by our sequence data. The clade is usually characterized by features such as palminerved leaves, stellate hairs, mucilage, layered phloem with dilated rays, valvate calyces, and more or less numerous stamens. However, these characters are not rare outside core Malvales. The few known morphological apomorphies for Malvales s.s. include occurrence of a unique repeating unit within the inflorescences (bicolor unit: Bayer, 1999), trichomatous floral nectaries localized mainly on the adaxial side of the perianth (Knuth, 1898, 1904; Brown, 1938; Frei, 1955; Vogel, 1977) and perhaps a valvate calyx, even if this feature is not rare outside Malvales s.l. The occurrence of tile cells (Chattaway, 1933; Manchester & Miller, 1978) is restricted to core Malvales (and Karwinskia, Rhamnaceae; Schirarend, pers. comm.) but is known only from relatively few genera. Contrary to widespread opinion (e.g. Cronquist, 1981; Judd & Manchester, 1997), the occurrence of cyclopropenyl fatty acids is not restricted to core Malvales: positive Halphen reaction and/or the detection of (dihydro-) malvalic or (dihydro-) sterculic acids have been reported for Sarcolaenaceae and Thymelaeaceae and also for some remote families such as Boraginaceae, Elaeocarpaceae, Leguminosae, Rhamnaceae, 280 C. BAYER ET AL. and Sapotaceae (and even for Gnetum; Vickery, 1980, 1981; Gaydou & Ramanoelina, 1983; Hosamani, 1994, 1995). Subdivision of core Malvales As evident from our results (Figs 2, 4) there is strong support for the polyphyly of Tiliaceae and Sterculiaceae and no evidence for the subdivision of core Malvales into the four traditional families Sterculiaceae, Tiliaceae, Bombacaceae and Malvaceae, only the last of which is monophyletic. Other clades of the consensus tree also correspond to previously recognized alliances, which however are usually ranked below the family level. Even if some of them are not supported by the bootstrap or found in the consensus tree (Figs 2, 4), most of them can be circumscribed by morphological characters and roughly correspond to some of the traditionally accepted suprageneric taxa. These clades are here considered as subfamilies of a single family Malvaceae that is expanded to comprise all core Malvales. The name Malvaceae Juss. (1789: 271) is preferred over Tiliaceae Juss. (1789: 289) because the former has already been used in a similar broader sense by earlier botanists (e.g. Jussieu, 1789; Baillon, 1873; van Tieghem, 1884; for comments see Masters, 1869; Schumann, 1895; Judd, Sanders & Donoghue, 1994; Judd & Manchester, 1997). Certainly, there will be some objections to the fusion of the established families Tiliaceae, Sterculiaceae, Bombacaceae, and Malvaceae. However, in view of our data and others ( Judd & Manchester, 1997), there seems to be no advantage to the maintenance of the traditionally accepted families by simply changing their circumscriptions. For instance, representatives of former Tiliaceae are scattered among Sterculiaceae. To achieve a more natural delimitation of Tiliaceae, all ‘tiliaceous’ genera except Grewioideae (the largest clade of Tiliaceae, see below) would have to be transferred to Sterculiaceae. Since Tilia would also have to be removed, the remaining genera would have to be named Grewiaceae, not Tiliaceae. Based on molecular as well as on morphological data it is, therefore, not possible to maintain both Sterculiaceae and Tiliaceae in their broader delimitation. If, in turn, Tiliaceae were expanded to include Sterculiaceae, and Bombacaceae were sunk into Malvaceae, then the two remaining core families of Malvales would be difficult to delimit by morphological characters, and the dombeyoid group especially would occupy an intermediate position. In addition to the fact that Tiliaceae (incl. Sterculiaceae) would be paraphyletic with respect to Malvaceae (incl. Bombacaceae), there is no molecular evidence in favour of a deep split between these two families. If it is admitted that neither a subdivision of core Malvales into four families nor an expansion of both Tiliaceae and Malvaceae are supported by available data, then two possibilities remain: the clades found in our trees can be treated either as separate families or as infrafamilial taxa of a single family, Malvaceae. To us it would seem inappropriate to rank these clades as families, some of which would have to be formally described as new. It is true that there are no objective criteria or formal obstacles against treating the clades within core Malvales as families, but for practical reason we feel that any increase in the number of small families should be generally avoided if possible. Furthermore, the morphological differences as well as the plastid gene sequence divergence within Malvaceae s.l. are not larger than in other families. The monophyly of Malvaceae s.l. is well established, whereas the MOLECULAR SYSTEMATICS OF MALVALES 281 suprageneric taxonomy within this group obviously requires far more research. For now, ambiguously placed genera can simply be treated as taxa incertae sedis within Malvaceae, whereas splitting core Malvales into numerous smaller families would leave these genera without clear family affiliation, which is also impractical and undesirable. The emerging suprageneric alliances are treated here as subfamilies, which has the advantage that some of the former tribes (especially those within former Malvaceae) may be maintained in their commonly used sense, even if certain changes in their circumscription may be necessary. In view of the tentative character of the classification proposed here it seems favourable to use only names that are already available, which is true for all subfamilies listed below. Previous classifications made little use of the subfamily rank, so this category allows us to circumscribe the major lineages within core Malvales without changing the commonly used names any more than is required. This is desirable since the circumscriptions of these entities have to be drastically narrowed (e.g. Tilioideae) or broadened (e.g. Dombeyoideae) to achieve presumably natural groups with roughly comparable ranges of variation. Except for Malvoideae, Bombacoideae, and Durioneae, for which only a few representatives are listed, the taxonomic position of every genus according to some previous classifications ( Jussieu, 1789; Baillon, 1873; Schumann, 1895; Hutchinson, 1967) as well as our circumscriptions here are given in Appendix 2. The placement of those genera, which are not included in our molecular analyses, is based on morphological characters (C. Bayer, unpubl.). Distinctive characters of the subfamilies, some of which are likely to be synapomorphies, are mentioned in the following paragraphs and summarized in Table 2. Byttnerioideae Burnett Byttnerioideae include genera that represent tribes Byttnerieae, Lasiopetaleae, Theobromeae and Hermannieae (see below). Byttnerioideae can be circumscribed by their peculiar cucullate (‘hooded’) petals (Schumann, 1886; Leinfellner, 1960). However, this character is absent from Hermannieae, which might be interpreted as the result of a secondary transformation within Byttnerioideae. As far as our molecular data are concerned, this clade represents the most problematic alliance within core Malvales. The assumption that Byttnerioideae are monophyletic is neither supported nor strongly rejected by our molecular data: Byttnerioideae appear to be paraphyletic with respect to Grewioideae in all most-parsimonious trees (both Fitch and SW), but this topology is not supported by the bootstrap. Although Figure 4 indicates that Grewioideae are monophyletic but embedded within Byttnerioideae, we suspect a sister-group relationship between them. This relationship is only two steps less parsimonious with these data (as determined with MacClade and a constraint experiment in PAUP). In view of the morphological differences between these entities, we treat them as separate subfamilies. Of the tribes mentioned above, only Lasiopetaleae are supported. Based on morphological characters, Byttnerioideae could be subdivided into the four tribes discussed below. Although the combined rbcL/atpB matrix does not identify the first two, it does not strongly refute their existence. The topology that we obtained lacks a clear pattern because of the only modest level of divergence. Hence, we suspect that with more information these four clades will emerge. Although we presently lack evidence for these tribes’ monophyly, use of the names is not precluded and serves a useful exploratory purpose. Inflorescences Flowers Pollen other features Byttnerioideae flowers in many- to 3(-2)flowered bicolor units, more rarely solitary with epicalyx (many Lasiopetaleae), often in sympodia petals cucullate (except Hermannieae) to reduced; stamens epipetalous, fused to clusters or solitary; episepalous staminodia present or reduced; ovary 5–1-locular usually 3-colporate to pororate, sometimes operculate; reticulate to perforate or occasionally spinulous pantropical; mostly small trees or shrubs Grewioideae bicolor units usually many-to 3-flowered (epicalyx e.g. in Luehea) anthocladia, condensed sympodia, or panicles, sometimes terminal nectaries, if present, mostly on petals or androgynophore; stamens arising from alternipetalous primordia or from ringwall primordia, distinct, numerous, some occasionally sterile, but not resembling byttnerioid staminodia ± prolate; 3-colporate; usually micro-perforate, often with suprategillary reticulum pantropical; trees, shrubs, rarely herbs Tilioideae bicolor units usually many-flowered, stamens numerous, distinct their axillary, axis with primordia and (if present) staminodia wing-like bract (Tilia) epipetalous ± oblate; (brevi-) colporate, finely reticulate trees; northern hemisphere, mainly from temperate regions Helicteroideae bicolor units sometimes reduced to flower pairs with 4 bracts, sometimes arranged in anthocladia; rarely panicles of flowers with or without epicalyx sometimes zygomorphic; often gamosepalous; petals often with lateral constrictions; androgynophore usually present; staminodia present; sometimes apocarpous ± oblate; 3 (–5)– angulaperturate; brevicolpate, sometimes with verrucae or tiny spinules sometimes with extrafloral nectary on inflorescence ramifications Brownlowioideae probably at least sometimes in many-flowered bicolor units gamosepalous; sometimes apetalous and/or unisexual; stamens numerous, free, thecae divergent at base and touching each other on top of connective; often apocarpous ± oblate; breviaperturate; often finely reticulate predominantly palaeotropical; trees, rarely large shrubs; occasionally lepidote Sterculioideae mostly paniculate, axillary, rarely condensed and cauliflorous; epicalyx absent gamosepalous; apetalous; usually unisexual; androgynophore present; staminodia absent; apocarpous spheroidal to prolate; tricolporate; suprategillary reticulate, often micro-perforate pantropical; trees; leaves sometimes digitate or uniforliolate; fruits follicles or nuts continued C. BAYER ET AL. Subfamily 282 T 2. Summary characteristics of the subfamilies of Malvaceae s.l. as recognized in our treatment; most distinctive characters or potential apomorphies underlined (for sources see text) T 2. continued Inflorescences Flowers Pollen other features Dombeyoideae flowers axillary on open shoots, often solitary or in few-flowered cymes; epicalyx usually present stamens in bundles usually separated by staminodia, sometimes numerous, ± free or forming a tube suboblate to spheroidal; often 3-porate, occasionally polyaperturate; spinose mainly Madagascar and Pacific islands; mostly shrubs or herbs; seeds sometimes winged; cotyledons usually bifid Bombacoideae flowers solitary axillary or variously arranged, rarely paniculate; sometimes in anthocladia; epicalyx present gamosepalous, stamens (sometimes very) numerous, filaments more or less fused, sometimes forming phalanges or tubes, anthers di-, tetra-, or polysporangiate spheroidal to ± oblate; usually 3-(col)porate; often reticulate, rarely spinulose mainly tropical America and Africa; tall to small trees; leaves often digitate; often chiropterophilous Malvoideae flowers usually with epicalyx, in axillary condensed sympodia or solitary, rarely in anthocladia gamosepalous, stamens ± numerous, forming a tube, anthers always disporangiate; number of carpels often increased often large and more or less spheroidal, 3- to polyaperturate, mostly spinose cosmopolitan, temperate to tropical; herbs or shrubs, rarely trees, fruits rarely capsular or baccate MOLECULAR SYSTEMATICS OF MALVALES Subfamily 283 284 C. BAYER ET AL. (1) Byttnerieae (Byttneria, Ayenia, Rayleya and Megatritheca) obviously represent a natural entity. The flowers of Byttnerieae are characterized by a whorl of single antepetalous stamens and, more specifically, their peculiar clawed petals, which Leinfellner (1960) regarded as the most complicated in the angiosperms (Cristóbal, 1960, 1976). Abroma, which is sister to Byttneria according to our analysis, has cucullate petals with a much broader base, and the stamens are not solitary but in bundles. Judging from these characters, which are typical of Theobromeae (see below), it is not clear why Abroma should be more closely related to Byttneria than to Theobroma, as our cladograms indicate. (2) Theobromeae include Theobroma, Herrania, Guazuma, Abroma, Scaphopetalum, Kleinhovia and Leptonychia. These genera share petals with a broadly based cucullus. The distal petal appendix is laminar in Abroma, Guazuma, Herrania and most Theobroma species but is lacking in Scaphopetalum, Leptonychia and Theobroma sect. Telmatocarpus (Cuatrecasas, 1964). The androecium includes staminodia, which are usually conspicuous, and alternisepalous bundles of two or more stamens that are produced through secondary increase of primordia (e.g. Payer, 1857; Baillon, 1861/1862, 1870; van Heel, 1966; Bayer & Hoppe, 1990). It is quite obvious that the neotropical genera Herrania and Guazuma are closely related to Theobroma. The palaeotropical genera Abroma and Scaphopetalum, and possibly also Leptonychia, appear to fit well into this alliance, even if Corner (1976) regarded the last as a misfit within Sterculiaceae. Nevertheless, there is no molecular support for a placement of Abroma and Leptonychia within Theobromeae and especially the position of Leptonychia (here treated as incertae sedis) remains unclear. (3) Lasiopetaleae include a well delimited core group that is confined to Australia and comprises Thomasia, Hannafordia, Lysiosepalum, Lasiopetalum and Guichenotia. Their flowers are mostly arranged in bracteose monochasia and have an epicalyx (Classen, 1988; Bayer & Kubitzki, 1996). Further characters are the reduced or absent petals and staminodia, the occurrence of anthers opening with short slits or pores, often less than five carpels, tubular stigmas and adaptions to myrmecochory (Gay, 1821; Schumann, 1886; Jenny, 1985). In addition to this core group, Lasiopetaleae are generally considered to include genera such as Keraudrenia and Seringia. However, the separation of such a broadly circumscribed Lasiopetaleae from other Byttnerioideae has always been somewhat arbitrary. There is a morphological continuum between the flowers of Theobromeae and Lasiopetaleae, and some of the ‘intermediate’ genera here placed in Lasiopetaleae have previously been included in Theobromeae. We cannot see any obvious reason to include the traditionally accepted Keraudrenia and Seringia in Lasiopetaleae if the similar genera Rulingia and Commersonia are excluded ( Jenny, 1985; Bayer & Kubitzki, 1996). It is more likely that these four genera are closely related to each other and belong to a slightly expanded tribe Lasiopetaleae, which is represented by Lasiopetalum, Thomasia, Rulingia and Keraudrenia in our analysis (Fig. 4). This clade is present but only weakly supported in all most parsimonious trees produced by the combined matrix. Lasiopetaleae extend from Australia to Madagascar and tropical Asia. Their inflorescences are arranged in anthocladia, and the flowers have an epicalyx or are united in many-flowered units, in which the first, sterile bract is displaced on the main axis (Bayer & Kubitzki, 1996; C. Bayer, pers. obs.). Their relatively small albeit cucullate petals link them with Theobromeae. As in Byttnerieae and in contrast to Theobromeae, the androecium of Lasiopetaleae includes only five fertile stamens. MOLECULAR SYSTEMATICS OF MALVALES 285 (4) Hermannieae (Hermannia, Melochia, Dicarpidium and Waltheria) are united by the presence of five stamens and reduced or absent staminodia. Since we sampled only a single species of this tribe, we can only speculate about its monophyly. Our data indicate that Hermannieae are embedded within Byttnerioideae. Accordingly, one could postulate the loss of the cucullate condition of the petals within this lineage. In Hermannieae, flower pairs surrounded by four bracts prevail, and these are often arranged in sympodia (Bayer, 1994). Pollen is usually spheroidal to prolate; spinulose grains are restricted to the short-styled flowers of heterostylous Waltheria and Melochia species (Köhler, 1973; M. Jenny, pers. comm.). Anatomically, Hermannieae appear to be homogeneous and similar to other Byttnerioideae (Dumont, 1887). Grewioideae Hochr. Grewioideae, which comprise Burret’s (1926) Sparmanniinae and Grewiinae plus Tetraliceae, are strongly supported by our DNA data. If there is any consistency in former Tiliaceae apart from Brownlowioideae, it is found in this subfamily, which includes the vast majority of ‘tiliaceous’ genera, but not Tilia itself. Their floral nectaries, if present, are located at the ventral base of the petals and rarely on adjacent tissue such as the androgynophore. Staminodia equivalent to those of Byttnerioideae are lacking. The fact that the stamens are free and indeterminate in number, but usually more numerous than in most former Sterculiaceae, constitutes the traditional character used to discriminate between Tiliaceae and Sterculiaceae. These traits have also been cited as support for an alleged primitive position of Tiliaceae (Edlin, 1935), even if it was known that the increased number of stamens is due to a ‘dédoublement’ and is therefore secondary (Ronse Decraene & Smets, 1993). Unlike other Malvaceae, stamens usually arise from alternipetalous primordia or from a ringwall-shaped primordium (Payer, 1857; Čelakovský, 1875; Hirmer, 1917; van Heel, 1966; Kortum, unpubl.; C. Bayer, pers. obs.). Two whorls of androecial primordia are only rarely found (Mollia: W. Kortum, unpubl.). Pollen of Grewioideae is more or less prolate, its exine usually micro-perforate and often bearing a suprategillary reticulum. It appears to occur throughout the tribe; somewhat similar, albeit finer sculptured, prolate grains are found in Lasiopetaleae (Erdtman, 1952; Sharma, 1969; Presting, Straka & Friedrich, 1983; M. Jenny, pers. comm.). Tilioideae Arn. Tilia occupies an isolated position in our analaysis. Judged from morphological and molecular data, Tilia appears to stand outside the clade comprising Malvoideae, Bombacoideae, Dombeyoideae, Brownlowioideae, Sterculioideae, and Helicteroideae. The remaining genera included in Hutchinson’s (1967) Tilieae (Duboscia, Muntingia, Brachypodandra, Schoutenia) are only remotely related: Schoutenia is much better placed in expanded Dombeyoideae rather than close to Tilia, as indicated by molecular data as well as the presence of an epicalyx and spinose pollen. Duboscia and Brachypodandra were not included in this study; the former should be palced in Grewioideae, and the latter is a synonym of Vatica (Dipterocarpaceae; Ashton, 1982). Muntingia is not a member of core Malvales and, with Dicraspidia and Neotessmannia, forms a distinct family, Muntingiaceae (Fig. 3; Bayer et al., 1998). Tilia is by no means typical of what was formerly called Tiliaceae. In contrast to most genera of former Tiliaceae, the genus exhibits generalized malvalean characters, such as the presence of sepal nectaries and alternisepalous androecial primordia. 286 C. BAYER ET AL. The peculiar sympodial shoot and conspicuous wing-like bract of the inflorescence (Hall & Swain, 1971; Manchester, 1994; Bayer, 1994) apparently represent autapomorphies. The high chromosome numbers of Tilia (x=41; Fedorov, 1969) are certainly an indication of polyploid origin, and it is likely that the genus represents an early temperate offshoot of Malvaceae s.l. (for distribution of Tilia, see Tang & Zhuge, 1996). A genus not included in our study that might prove to be sister to Tilia is Craigia ( Judd & Manchester, 1997). On the other hand, Zhuge (1989) suggested close affinities between Maxwellia and Craigia, both of which he considered to be sterculiaceous. From a palynological comparison between Craigia and few other genera, Long, He & Hsue (1985) concluded that Craigia agrees better with Sterculiaceae than with Tiliaceae. In fact, the pollen of Craigia is tilioid, thus resembling the pollen of Mortoniodendron, Brownlowioideae, certain Helicteroideae, Bombacoideae, and of course Tilia (Erdtman, 1952; Sharma, 1969; Nilsson & Robyns, 1986; Ying, Zhang & Boufford, 1993; M. Jenny, unpubl.). This character seems to be the decisive one in Judd & Manchester’s (1997) assumption of a sister-group relationship between Tilia and Craigia. Flower structure, however, might provide more specific characters in favour of a relationship between Craigia and Tilia. The flowers of Craigia have been misinterpreted as apetalous (Smith & Evans, 1921; Ying et al., 1993), which might have led Hutchinson (1967) to place the genus in Lasiopetaleae. Alternating with the sepals, five clustered structures are found, in each of which four fused stamens are enclosed by an outer and an inner organ. The outer organ is obviously homologous to the petals of other Malvaceae. The inner one corresponds to a staminode that most probably developed from the epipetalous androecial primordium, whereas the episepalous sector remains empty. Therefore the inner staminodia do not correspond to the staminodia of other Malvaceae; those of both these taxa are from the inner androecial whorl and arise from the episepalous sectors (C. Bayer & M. Jenny, unpubl.). However, a similar arrangement with staminodia that represent the central members of epipetalous androecial clusters are found in several Tilia species (Payer 1857; Schumann, 1890; C. Bayer, pers. obs.). In view of this rare character Craigia certainly merits further attention to clarify the sister-group relationships of Tilia. Helicteroideae (Schott & Endl.) Meisn. Helicteroideae are well supported by our data. They include the genera of Helictereae sensu Hutchinson (1967) except for Pterospermum, which is here transferred to Dombeyoideae, and Kleinhovia, which is referrable to Byttnerioideae according to ndhF data (Alverson et al., in press). Within Helicteroideae, Neoregnellia is obviously related to Helicteres. As to the other genera included in Helictereae by Hutchinson (1967), the position of Reevesia in this clade is supported by our molecular data; our sampling did not include Ungeria. In addition to the generally accepted Helictereae, we also include Triplochiton and Mansonia. Triplochiton was described by Schumann (1900) as representing a new family. The genus exhibits a peculiar combination of characters of both Sterculioideae (secondary apocarpy, androgynophore) and many cucullate Byttnerioideae (presence of petals and staminodia). These characters, however, are met within Helicteroideae, into which Triplochiton falls in our analyses. Mansonia, which has not been included in the molecular study, is morphologically similar and probably closely related to Triplochiton (see Prain, 1905; Mildbraed, 1921; MOLECULAR SYSTEMATICS OF MALVALES 287 Schulze-Motel, 1964; but also Emberger, 1960). Mansonia and Helicteres share the occurrence of unique glands on the inflorescence ramifications (Fahn, 1979; C. Bayer, pers. obs.). Pollen of Triplochiton, Mansonia, and the remaining Helicteroideae is quite similar. The exine is often microperforate and occasionally spinulose (Neoregnellia, Helicteres, Triplochiton: Erdtman, 1952; M. Jenny, pers. comm.). The petals of some Helicteroideae have been described as being cucullate (Schumann, 1886; Leinfellner, 1960); they often exhibit lateral constrictions that are reminiscent of the petals of the cucullate Byttnerioideae. Durioneae (Durio, Neesia, Coelostegia, Kostermansia, Cullenia, Boschia), which are generally considered to represent a tribe of Bombacaceae, appear to be related to Helictereae according to our data. They differ from Bombacoideae (see below) in characters such as their pinnately nerved leaves, lepidote indumentum, more or less fused epicalyx, distinctive muricate to spinose fruit, usually arillate seeds with thick and flat cotyledons, special pollen type with mostly smooth microperforate exine, considerably lower chromosome numbers (n=14, di- or tetraploid), vegetative anatomy, and exclusively Asian distribution (Dumont, 1887; Masters, 1875; Bakhuizen van den Brink, 1924; Nilsson & Robyns, 1986; Krutzsch, 1989; Baum & Oginuma, 1994). These differences indicate that this homogeneous alliance has been generally misplaced in Bombacaceae. However, an inclusion in Helicteroideae is neither unambiguously evident from these sequence data nor corroborated by morphological characters. Therefore the systematic position of Durioneae remains puzzling, and we consider them incertae sedis here. Brownlowioideae Burret Our molecular data strongly support Brownlowioideae sensu Burret (1926), who remarked that it is so different from other Tiliaceae that it might even be raised to family rank. Its members are characterized by fused sepals and a special arrangement of staminal thecae, which are divergent at the base and touching each other on the top of the connective. Unlike Sterculioideae, which also possess fused sepals, the pollen of Brownlowioideae corresponds to the Tilia-type (Erdtman, 1952; Sharma, 1969). Some Brownlowioideae are described as apocarpous or have at least free carpels in fruit, which is also reminiscent of Sterculioideae. However, due to a lack of suitable material for ontogenetic studies, it is not known whether true secondary apocarpy exists in Brownlowioideae (Kubitzki, 1995). The position of Mortoniodendron within Brownlowioideae is only weakly supported by the SW bootstrap. Based on morphology, there is no obvious justification for including Mortoniodendron in this clade. Neither the typical anthers nor the fused calyx of Brownlowioideae are present, and the arillate seed is as uncommon in this group as in the remaining former Tiliaceae. Agreement with Brownlowioideae can be found in the tilioid pollen (Erdtman, 1952; Graham, 1979), which is, however, more coarsely reticulate (G. El-Ghazaly and K. Kubitzki, pers. obs.). Sterculioideae Burnett The clade representing Sterculioideae is weakly supported by the bootstrap (SW only) but is readily characterized by the usually unisexual and always apetalous, apocarpous flowers with androgynophores. In addition to Hutchinson’s (1967) Sterculieae, this subfamily also includes Hildegardia and Heritiera (Tarrietieae sensu Hutchinson, 1967). The remaining genera of Hutchinson’s (1967) Tarrietieae, 288 C. BAYER ET AL. Mansonia and Triplochiton, differ in their hermaphrodite flowers with petals and staminodia and the more or less oblate pollen with short apertures (Erdtman, 1952; M. Jenny, pers. comm.) and are transferred here to Helicteroideae (see above). The wood anatomy of Sterculioideae has been described as unique within Malvales and treated as an apomorphy (Chattaway, 1932, 1937; Taylor, 1989). This alliance has been considered as a primitive tribe or subfamily of Sterculiaceae (e.g. Brizicky, 1966). Nonetheless, apetaly, unisexuality of flowers and, less widely known, apocarpy of Sterculioideae are secondary and by no means primitive (Endress, Jenny & Fallen, 1983; Jenny, 1985, 1988). These features, the presence of axillary paniculate inflorescences, androgynophores, absence of staminodia and an epicalyx can be regarded as advanced character states of this clade, although these traits also occur scattered in other taxa. This also indicates that secondary apocarpy evolved at least twice within core Malvales. Dombeyoideae Beilschm. Dombeyoideae are strongly supported by our molecular data (both with Fitch and SW). Most genera of the subfamily belong to a monophyletic core group that corresponds to Hutchinson’s (1967) Dombeyeae with addition of Eriolaena, Helmiopsiella, and obviously also Helmiopsis and Corchoropsis (no molecular data). This alliance is morphologically homogeneous and unambiguously supported by the bootstrap. They also resemble Malvoideae in certain respects (de Candolle, 1823; Erdtman, 1952; Heslop-Harrison & Shivanna, 1977; Jenny, 1985, 1988; Barnett, 1987, 1988; Bayer, 1994). The core group of Dombeyoideae as outlined above is centred in Madagascar and Pacific islands, extending to Africa (some Dombeya and Melhania species and Harmsia incl. Aethiocarpa), St. Helena (Trochetiopsis), SE Asia (Cronk, 1990). As indicated by our data, there are at least there more Asian genera that share palynological and inflorescence characters with Dombeyoideae and should be included in this subfamily. Pterospermum, which was misplaced in Helictereae (Dumont, 1887; Schumann, 1886; Zebe, 1915; Schulze-Motel, 1964; Jenny, 1985; Tang, 1992; Bayer, 1994) or separated as Pterospermeae Wu & Tang (Tang, 1992), has a seed wing of the same type as Helmiopsis and Helmiopsiella (Barnett, 1988). However, Pterospermum lacks the bifid cotyledons (Mohana Rao, 1976), the apomorphy of Dombeyeae s.s. (Barnett, 1988), and appropriately branches off at a lower node of the cladogram. The same is true for Schoutenia, which was formerly included in Tiliaceae-Tilieae, and Burretiodendron (Tiliaceae-Enteleeae according to Hutchinson, 1967). However, it is not known if this is true for all species recognized by Zhuge (1990), who included Excentrodendron H.T. Chang & R.H. Miao, since Burretiodendron s.l. appears to be palynologically heterogeneous (Tang & Gao, 1993). The species investigated in the present study, B. esquirolii (Lév.) Rehder, has the spinose pollen of Dombeyoideae. As a rough rule, this character seems to prevail in the ‘advanced’ subfamilies outlined here and, for instance, helped to assign genera of uncertain position to Dombeyoideae. However, spin(ul)ose pollen must not be taken as the only criterion for an inclusion in this subfamily, since spines, spinules or similar structures apparently evolved independently in other taxa (e.g. Byttnerieae: Ayenia; heterostylous Hermannieae: see above; Helicteroideae: Helicteres p.p.; Erdtman, 1952; M. Jenny, pers. comm.). Bombacoideae Burnett The clade representing Bombacoideae (and Malvoideae) in our study is largely resolved as monophyletic, but weakly or unsupported. Bombacoideae comprise most MOLECULAR SYSTEMATICS OF MALVALES 289 genera that formerly have been referred to Bombacaceae. In contrast to previous classifications, we exclude Durioneae from this alliance but include Fremontodendreae (Fremontodendron and Chiranthodendron). The latter have been been placed in Sterculiaceae by most authors, although bombacaceous affinities have been generally admitted (Erdtman, 1952; Metcalfe & Chalk, 1950). Our data favour the latter placement, even if Fremontodendron has a relatively low chromosome number (n=20) as compared with the general n=43–46 typical of other Bombacoideae. However, Baum & Oginuma’s (1994) karyological review did not include representatives of the Matisia alliance; counts for Chiranthodendron have not been reported. Another genus that, contrary to the placement in Tiliaceae–Brownlowieae proposed by Williams & Standley (1952), could belong to Bombacoideae, is Pentaplaris. This is supported by features such as the basally fused, slightly imbricate calyx that is penetrated by the contorted petals before anthesis, and the presence of a staminal column with five phalanges of monothecal anthers (C. Bayer, pers. obs.). Pollen of Pentaplaris resembles Nilsson & Robyns’ (1986) Rhodognaphalopsis type (C. Bayer, pers. obs.). According to our cladograms (Figs 2, 4), Pentaplaris is sister to Malvoideae. However, this position is not supported by the bootstrap, and only three additional steps (as assessed with MacClade) are required to shift the genus into Bombacoideae. Except for the uncertain position of Pentaplaris and apart from Durioneae (see Helicteroideae), the monophyly of Bombacoideae is not refuted by our data (Figs 2, 4). Therefore, we tentatively maintain former Bombacaceae as a distinct subfamily, even if we are aware of the fact that the lack of unambiguously discriminating characters and future evidence for paraphyly may lead to sinking it into Malvoideae ( Judd et al., 1994). Malvoideae Burnett Our molecular data provide only weak to moderate support for a monophyletic clade Malvoideae that corresponds to former Malvaceae in the circumscription accepted by most authors, thus including Hibisceae and Gossypieae. The dehiscent fruits of the latter tribes are probably plesiomorphic, and Edlin (1935) transferred them to Bombacaceae, many of which have capsules. In contrast, our analyses confirm Hutchinson’s (1967: 538) comment that “Malvaceae without the great genus Hibiscus would be like a horse without a tail” and support the original placement (see La Duke & Doebley, 1995). There appears to be no single morphological character discriminating Bombacoideae from Malvoideae. Malvoideae are rarely arborescent and usually have lower chromosome numbers. The free portions of their staminal filaments are relatively short, monothecal anthers are found throughout, and pollen is almost always spinose and often pantoporate (Bakhuizen van den Brink, 1924; Erdtman, 1952; Robyns, 1963; Fryxell, 1968; Christensen, 1986; Nilsson & Robyns, 1986; Baum & Oginuma, 1994). Nevertheless, these and other characters fall within the range found in Bombacoideae. A character that has been claimed clearly to discriminate between these taxa, the persistence of the nucleolus during mitosis (Baker & Baker, 1968), is neither known for a sufficient number of species nor can it be regarded as a convenient character to distinguish the two former families. However, even if there are exceptions (e.g. Camptostemon), the vast majority of genera can be easily referred to one of the respective groups by applying the traditional combination of characters. 290 C. BAYER ET AL. The topology within the clade comprising Malvoideae, Bombacoideae, Dombeyoideae, Brownlowioideae, and Sterculioideae, all in the extended sense outlined here, is not supported by the bootstrap. Morphologically, Malvoideae resemble both Bombacoideae and Dombeyoideae more than the remaining subfamilies. Malvoideae and Dombeyoideae share inflorescence and pollen features that are rare in Bombacoideae (Erdtman, 1952; Christensen, 1986; Nilsson & Robyns, 1986; Bayer, 1994). On the other hand some similarities between Bombacoideae and Malvoideae (e.g. staminal tube with monothecal anthers) are so striking that it cannot be excluded that the latter represent an offshoot within the bombacoid clade. CONCLUSION Previous workers rarely questioned the traditional family borders within core Malvales but instead erected highly structured infrafamiliar classifications. Some earlier authors such as Schumann, who were familiar with the broad range of Malvales groups, recognized inconsistencies in character distribution (e.g. the occurrence of spinose pollen in Pterospermum; Schumann, 1886). Nevertheless, the available information may have been insufficient for attempting a revised family circumscription, and authors working only on a single family did not perceive the problems of higher order classification and simply followed tradition. Our new molecular data provide a basis for an revised subdivision of core Malvales that is more consistent with the distribution of morphological characters. Therefore we are convinced that the classification proposed here will provide an improved basis for future work in Malvales, but we by no means consider this as a final solution to these longstanding problems. ACKNOWLEDGEMENTS We would like to thank our colleagues at Kew for help in the Jodrell Laboratory, F. R. Blattner for some DNA samples, D. A. Baum and T. Terrazas for the permission to use their sequence data and O. Appel and L. J. Dorr for helpful comments on the manuscript. Travel support for the first author was kindly provided by the Deutsche Forschungsgemeinschaft (Ku 174/14–1). REFERENCES Albert VA, Williams SE, Chase MW. 1992. Carnivorous plants: phylogeny and structural evolution. Science 257: 1491–1495. Alverson WS, Karol KG, Baum DA, Chase MW, Swense SM, McCourt R, Sytsma KJ. 1998. Circumscription of the Malvales and relationships to other Rosidae: evidence from rbcL sequence data. American Journal of Botany: 85: 876–887. Alverson WS, Whitlock BA, Nyffeler R, Bayer C, Baum DA. In press. Phylogeny of the core Malvales: evidence from ndhF sequence data. American Journal of Botany. Appel O. 1996. Morphology and systematics of the Scytopetalaceae. Botanical Journal of the Linnean Society 121: 207–227. MOLECULAR SYSTEMATICS OF MALVALES 291 Ashton PS. 1982. Dipterocarpaceae. Flora Malesiana, series 1, vol: 9(2). The Hague, Boston, London: Nijhoff, Junk Publishers, 237–552. Baillon H. 1861/1862. Etudes organogéniques sur quelques genres de Byttnériacées. Adansonia 2: 166–181. Baillon H. 1870. Traité du développement de la fleur et du fruit (Suite), Buettneriées. Adansonia 9: 336–351. Baillon H. 1873. Histoire des plantes, Vol. 4: Paris. Baker HG, Baker I. 1968. Chromosome numbers in the Bombacaceae. Botanical Gazette (London) 129: 294–296. Bakhuizen van den Brink RC. 1924. Revisio Bombacacearum. Bulletin du Jardin Botanique de Buitenzorg, série 3 6: 161–240. Bakker FT, Vassiliades DD, Morton C, Savolainen V. 1998. Phylogenetic relationships of Biebersteinia Stephen (Geraniaceae) inferred from rbcL and atpB sequence comparisons. Botanical Journal of the Linnean Society 127: 149–158. Barnett LC. 1987. Tribal realignment of certain palaeotropical Sterculiaceae. American Journal of Botany 74: 724. Barnett LC. 1988. Systematics of Nesogordonia Baillon (Sterculiaceae). Ph.D. thesis, University of Texas at Austin. Baum DA, Oginuma K. 1994. A review of chromosome numbers in Bombacaceae with new counts for Adansonia. Taxon 43: 11–19. Bayer C. 1994. Zur Infloreszenzmorphologie der Malvales. Dissertationes Botanicae 212. Bayer C. 1999. The bicolor unit – homology and transformation of an inflorescence structure unique to core Malvales. Plant Systematics and Evolution 214: 187–198. Bayer C, Hoppe JR. 1990. Die Blütenentwicklung von Theobroma cacao L. Beiträge zur Biologie der Pflanzen 65: 301–312. Bayer C, Kubitzki K. 1996. Inflorescence morphology of some Australian Lasiopetaleae. Telopea 6: 721–728. Bayer C, Chase MW, Fay MF. 1998. Muntingiaceae, a new family of dicotyledons with malvalean affinities. Taxon 47: 37–42. Bentham G, Hooker JD. 1862. Genera plantarum 1(1), London, 195–241. Bocquillon H. 1866. Mémoire sur le groupe des Tiliacées. Adansonia 7: 17–64. Brizicky GK. 1966. The genera of Sterculiaceae in the southeastern United States. Journal of the Arnold Arboretum 47: 60–74. Brown WH. 1938. The bearing of nectaries on the phylogeny of flowering plants. Proceedings of the American Philosophical Society 79: 549–595. Brummitt RK. 1992. Vascular plant families and genera. Kew: Royal Botanic Gardens. Burret M. 1926. Beiträge zur Kenntnis der Tiliaceen. Notizblatt des Botanischen Gartens und Museums zu Berlin-Dahlem 9: 592–880. Candolle AP de. 1823. Mémoire sur quelques genres nouveaux de la famille des Buttneriacées. Mémoires du Muséum d’Histoire Naturelle, Paris, Vol. 10: 97–115. Candolle AP de. 1824. Prodromus systematis naturalis regni vegetalis, Vol. 1. Paris. Čelakovský L. 1875. Ueber den “eingeschalteten” epipetalen Staubgefässkreis. Flora 58: 481–489, 497–504, 513–525. Chase MW, Soltis DE, Olmstead RG, Morgan D, Les DH, Mishler BD, Duvall MR, Price RA, Hills HG, Qiu Y-L, Kron KA, Rettig JH, Conti E, Palmer JD, Manhart JR, Sytsma KJ, Michaels HJ, Kress WJ, Karol KG, Clark WD, Hedrén M, Gaut BS, Jansen RK, Kim K-J, Wimpee CF, Smith JF, Furnier GR, Strauss SH, Xiang Q-Y, Plunkett GM, Soltis PS, Swensen SM, Williams SE, Gadek PA, Quinn CJ, Equiarte LE, Golenberg E, Learn GH, Graham SW, Barrett SCH, Dayanandan S, Albert VA. 1993. Phylogenetics of seed plants – an analysis of nucleotide sequences from the plastid gene rbcL. Annals of the Missouri Botanical Garden 80: 528–580. Chattaway MM. 1932. The wood of the Sterculiaceae. I. Specialisation of the vertical wood parenchyma within the subfamily Sterculieae. New Phytologist 31: 119–132. Chattaway MM. 1993. Tile-cells in the rays of the Malvales. New Phytologist 32: 261–273. Chattaway MM. 1937. The wood anatomy of the Sterculiaceae. Philosophical Transactions of the Royal Society of London 228: 313–365. Christensen PB. 1986. Pollen morphological studies in the Malvaceae. Grana 25: 95–117. Classen R. 1988. Beiträge zur Kenntnis der Gattung Lasiopetalum (Sterculiaceae). Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 109: 501–527. 292 C. BAYER ET AL. Conti E, Litt A, Sytsma KJ. 1996. Circumscription of Myrtales and their relationship to other rosids: evidence from rbcL sequence data. American Journal of Botany 83: 221–233. Corner EJH 1976. The seeds of dicotyledons. Cambridge: Cambridge University Press. Cristóbal CL. 1960. Revision del género ‘Ayenia’ (Sterculiacae). Opera Lilloana (Tucuman) 4: 1–230. Cristóbal CL. 1976. Estudio taxonómico del género Byttneria Loefling (Sterculiaceae). Bonplandia 4: 1–428. Cronk QCB. 1990. The history of the endemic flora of St Helena: late Miocene ‘Trochetiopsis-like’ pollen from St Helena and the origin of Trochetiopsis. New Phytologist 114: 159–165. Cronquist A. 1981. An integrated system of classification of flowering plants. New York: Columbia University Press. Cronquist A. 1988. The evolution and classification of flowering plants. 2. ed. New York: The New York Botanical Garden. Cuatrecasas J. 1964. Cacao and its allies – a taxonomic revision of the genus Theobroma. Contributions from the United States National Herbarium 35: 379–614. Dahlgren R. 1983. General aspects of angiosperm evolution and macrosystematics. Nordic Journal of Botany 3: 119–149. Doyle JJ, Doyle JL. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11–15. Dumont A. 1887. Recherches sur l’anatomie comparée des Malvacées, Bombacacées, Tiliacées, Sterculiacées. Annales des Sciences Naturelles, Botanique, série 7, 6: 129–242, pl. 4–7. Edlin HL. 1935. A critical revision of certain taxonomic groups of the Malvales. New Phytologist 34: 1–20. Emberger L. 1960. Les végétaux vasculaires. In: Chadefaud M, Emberger L, Traité de botanique (systematique), vol 2 (2) Paris: Masson & Cie. Endress PK, Jenny M, Fallen ME. 1983. Convergent elaboration of apocarpous gynoecia in higher advanced dicotyledons (Sapindales, Malvales, Gentianales). Nordic Journal of Botany 3: 293–300. Erdtman G. 1952. Pollen morphology and plant taxonomy. Angiosperms Stockholm: Almquist & Wiksell. Fahn A. 1979. Secretory tissues in plants. London, New York, San Francisco: Academic Press. Farris JS. 1969. A successive approximation approach to character weighting. Systematic Zoology 18: 374–385. Fay MF, Bayer C, Alverson WS, Bruijn AY de, Chase MW. 1998a. Plastid rbcL sequence data indicate a close affinity between Diegodendron and Bixa. Taxon 47: 43–50. Fay MF, Olmstead RG, Richardson JE, Santiago E. Prance GT, Chase MW, 1998b. Molecular data support the inclusion of Duckeodendron cestroides in Solanaceae. Kew Bulletin 53: 203–212. Fedorov AA. 1969. Chromosomnye cisla zvetkovych rastenij. Leningrad: Nauka. Felsenstein J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. Fitch WM. 1971. Toward defining the course of evolution: minimal change for a specific tree topology. Systematic Zoology 20: 406–416. Frei E. 1955. Die Innervierung der floralen Nektarien dikotyler Pflanzenfamilien. Berichte der Schweizerischen Botanischen Gesellschaft 65: 60–114. Fryxell PA. 1968. A redefinition of the tribe Gossypieae. Botanical Gazette (London) 129: 296–308. Gadek PA, Fernando ES, Quinn CJ, Hoot SB, Terrazas T, Sheahan MC, Chase MW. 1996. Sapindales: molecular delimitation and infraordinal groups. American Journal of Botany 83: 802–811. Gasson P. 1996. Wood anatomy of the Elaeocarpaceae. In: Donaldson LA, Singh AP, Butterfield BG, Whitehouse LJ, eds. Recent advances in wood anatomy. Rotorua: New Zealand Forest Research Institute. Gay J. 1821. Monographie des cinq genres de plantes que comprend la tribu des Lasiopétalées dans la famille des Büttnériacées. Paris. Gaydou EM, Ramanoelina ARP. 1983. A survey of the Sarcolaenaceae for cyclopropene fatty acids. Phytochemistry 22: 1725–1728. Graham A. 1979. Mortoniodendron (Tiliaceae) and Sphaeropteris/Trichipteris (Cyatheaceae) in cenozoic deposits of the Gulf-Caribbean region. Annals of the Missouri Botanical Garden 66: 572–576. Hall JW, Swain AM. 1971. Pedunculate bracts of Tilia from the Tertiary of western Unites States. Bulletin of the Torrey Botanical Club 98: 95–100. Heel WA, van. 1966. Morphology of the androeceum in Malvales. Blumea 13: 177–394. Heslop-Harrison Y, Shivanna KR. 1977. The receptive surface of the angiosperm stigma. Annals of Botany 41: 1233–1258. MOLECULAR SYSTEMATICS OF MALVALES 293 Hirmer M.1917. Beiträge zur Morphologie der polyandrischen Blüten. Flora 10: 140–192, 11 p1. Hoot SB, Culham A, Crane PR. 1995. The utility of atpB gene sequences in resolving phylogenetic relationships: comparison with rbcL and 185 ribosomal DNA sequences in the Lardizabalaceae. Annals of the Missouri Botanical Garden 82: 194–207. Hosamani KN, 1994. Ricinoleic and cyclopropen acids in Trichodesma zeylanicum seed oils. Phytochemistry 37: 1621–1624. Hosamani KN. 1995. A minor source of vernolic, malvalic, and sterculic acids in Pithecollobium dulce (syn. Inga dulcis) seed oil. Journal of the American Oil Chemists’ Society 72: 489–492. Huber H. 1991. Angiospermen. Leitfaden durch die Ordnungen und Familien der Bedecktsamer. Stuttgart, New York: Fischer. Huber H. 1993. Neurada, eine Gattung der Malvales. Sendtnera 1: 7–10. Hutchinson J. 1967. The genera of flowering plants (Angiospermae) Vol. 2. Oxford: Clarendon Press. Jenny M. 1985. Struktur, Funktion und systematische Bedeutung des Gynoeciums bei Sterculiaceen. Doctoral Dissertation, Universität Zürich. Jenny M. 1988. Different gynoecium types in Sterculiaceae: Ontogeny and functional aspects. In: Leins P, Tucker SC, Endress PK, eds. Aspects of floral development. Berlin, Stuttgart: Cramer, 225–236. Judd WS, Sanders RW, Donoghue MJ. 1994. Angiosperm family pairs: preliminary phylogenetic analyses. Harvard Papers in Botany 5: 1–51. Judd WS, Manchester SR. 1997. Circumscription of Malvaceae (Malvales) as determined by a preliminary cladistic analysis of morphological, anatomical, palynological, and chemical characters. Brittonia 49: 384–405. Jussieu A-L de. 1789. Genera Plantarum, Paris. Knuth P. 1898. Handbuch der Blütenbiologie, vol. 2. Leipzig. Knuth P. 1904. Handbuch der Blütenbiologie, vol. 3. Leipzig. Köhler E. 1973. Über einen bemerkenswerten Pollendimorphismus in der Gattung Waltheria L. Grana 13: 57–64. Krutzsch W. 1989. Paleogeography and historical phytogeography (paleochorology) in the Neophyticum. Plant Systematics and Evolution 162: 5–61. Kubitzki K. 1995. Asterophorum and Tahitia congeneric with Christiana (Tiliaceae). Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 116: 537–542. Kukachka BF. 1962. Wood anatomy of Petenaea cordata Lundell (Elaeocarpaceae). Wrightia 3: 36–40. La Duke JC, Doebley J. 1995. A chloroplast DNA based phylogeny of the Malvaceae. Systematic Botany 20: 259–271. Leinfellner W. 1960. Zur Entwicklungsgeschichte der Kronblätter der Sterculiaceae-Buettnerieae. Österreichische Botanische Zeitschrift 107: 153–176. Long H, He L-K, Hsue H-H. 1985. Pollen morphology of Craigia with reference to its systematic position. Acta Phytotaxonomica Sinica 23: 179–184. Lundell CL. 1962. Plantae mayanae – V. Petenaea cordata, a new genus and species in the Elaeocarpaceae, and other taxonomic notes. Wrightia 3: 21–25. Maddison DR. 1991. The discovery and importance of multiple islands of most-parsimonious trees. Systematic Zoology 40: 315–328. Maddison WP, Maddison DR. 1992. MacClade: analysis of phylogeny and character evolution, version 3. Sunderland, Massachusetts: Sinauer Associates. Manchester SR. 1994. Inflorescence bracts of fossil and extant Tilia in North America, Europe, and Asia: patterns of morphologic divergence and biogeographic history. American Journal of Botany 81: 1176–1185. Manchester SR, Miller RB. 1978. Tile cells and their occurrence in malvalean fossil woods. IAWA Bulletin 1978/2–3: 23–28. Masters MT. 1869. On some points in the morphology of the Malvales, together with a description of a new genus of Buettnerieae. Journal of the Linnean Society Botany 10: 18–30. Masters MT. 1875. Monographic sketch of the Durioneae. Journal of the Linnean Society, Botany 14: 495–508. Matcalfe CR, Chalk L 1950. Anatomy of the dicotyledons. 1. Oxford: Clarendon. Mildbraed J. 1921. Zur Kenntnis der afrikanischen Sterculiaceae-Mansonieae. Notizblatt des Botanischen Gartens und Museums zu Berlin-Dahlem 7: 486–490. Mohana Rao PR. 1976. Seed and fruit anatomy of Pterospermum acerifolium (Sterculiaceae). Phytomorphology 26: 363–369. Morgan DR, Soltis DE, Robertson KR. 1994. Systematic and evolutionary implications of rbcL sequence variation in Rosaceae. American Journal of Botany 81: 890–903. 294 C. BAYER ET AL. Morton CM, Chase MW, Kron KA, Swensen SM. 1996. A molecular evaluation of the monophyly of the order Ebenales based upon rbcL sequence data. Systematic Botany 21: 567–587. Morton CM, Mori SA, Prance KG, Karol KG, Chase MW. 1997. Phylogenetic relationships of the Lecythidaceae: a cladistic analysis using rbcL sequence and morphological data. American Journal of Botany 84: 530–541. Nilsson S, Robyns A. 1986. Bombacaceae Kunth. World Pollen and Spore Flora 14. Olmstead RG, Michaels HJ, Scott KM, Palmer JD. 1992. Monophyly of the Asteridae and identification of their major lineages inferred from DNA sequences of rbcL. Annals of the Missouri Botanical Garden 79: 249–265. Payer J-B. 1857. Traité d’organogénie comparée de la fleur. Paris. Prain D. 1905. Mansonieae, a new tribe of the natural order Sterculiaceae. Journal of the Linnean Society of London, Botany 37: 250–263. Presting D, Straka H, Friedrich B. 1983. Palynologia Madagassica et Mascarenica, Familien 128 bis 146. Tropische und Subtropische Pflanzenwelt 44. Price RA, Palmer JD. 1993. Phylogenetic relationships of the Geraniaceae and Geraniales from rbcL sequence comparisons. Annals of the Missouri Botanical Garden 80: 661–671. Robyns A. 1963. Essai de monographie du genre Bombax s.l. (Bombacaceae). Bulletin du Jardin Botanique de l’Etat Bruxelles 33: 1–311. Rodman JE, Price RA, Karol KG, Conti E, Sytsma KJ, Palmer JD. 1993. Nucleotide sequences of the rbcL gene indicate monophyly of mustard oil plants. Annals of the Missouri Botanical Garden 80: 686–699. Ronse Decraene LP, Smets EF. 1993. Dédoublement revisited: towards a renewed interpretation of the androeceum of the Magnoliophytina. Botanical Journal of the Linnean Society 113: 103–124. Savolainen V, Cuénoud P, Spichiger R, Martinez MDP, Crévecour M, Manen J-F. 1995. The use of herbarium specimens in DNA phylogenetics: evaluation and improvement. Plant Systematics and Evolution 197: 87–98. Schulze-Motel W. 1964. 32. Reihe Malvales (Columniferae). In: Melchior H, ed. A. Engler’s Syllabus der Pflanzenfamilien vol. 2. 12. ed. Berlin: Borntraeger, 304–316. Schumann K. 1886. Vergleichende Blütenmorphologie der cucullaten Sterculiaceae. Jahrbuch des Königlichen Botanischen Gartens und des Museums zu Berlin 4: 286–332. Schumann K. 1890. Neue Untersuchungen über den Blüthenanschluss. Leipzig. Schumann K. 1895. Elaeocarpaceae, Tiliaceae, Malvaceae, Bombacaceae, Sterculiaceae. In: Engler A, Prantl K, eds. Die natürlichen Pflanzenfamilien, Vol. 3(6), Leipzig, 1–99. Schumann K. 1900. Eine neue Familie der Malvales. Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 28: 330–331. Sharma BD. 1969. Pollen morphology of Tiliaceae in relation to plant taxonomy. Journal of Palynology 5: 7–29. Smith WW, Evans WE, 1921. Craigia, a new genus of Sterculiaceae. Transactions of the Botanical Society of Edinburgh 28: 69–71. Soltis DE, Soltis PS, Nickrent DL, Johnson LA, Hahn WJ, Hoot SB, Sweere JA, Kuzoff RK, Kron KA, Chase MW, Swensen SM, Zimmer EA, Chaw S-M, Gillespie LJ, Kress WJ, Sytsma KJ. 1997. Angiosperm phylogeny inferred from 18S ribosomal DNA sequences. Annals of the Missouri Botanical Garden 84: 1–49. Swofford DL. 1993. PAUP: Phylogenetic analysis using parsimony, Version 3.1.1. Champaign, Illinois: Illinois Natural History Society. Szyszylowicz I von. 1885. Zur Systematik der Tiliaceen I. Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 6: 427–457. Takhtajan A. 1997. Diversity and classification of flowering plants. New York: Columbia University Press. Tang Y. 1992. On the affinities of Pterospermum Schreb. (Sterculiaceae). Guihaia 12: 8–14. Tang Y, Gao XF. 1993. Pollen morphology of Burretiodendron sensu lato (Tiliaceae) and its systematic significance. Cathaya 5: 81–88. Tang Y, Zhuge R. 1996. Geographical distribution of Tilia Linn. Acta Phytotaxonomica Sinica 34: 244–264. Taylor EL. 1989. Systematic studies in the tribe Sterculieae: a taxonomic revision of the neotropical species of Sterculia L. (Sterculiaceae). Ph.D. Thesis, Cambridge, Massachusetts: Harvard University. Tieghem P van. 1884. Traité de Botanique. Paris. Thorne RF. 1992. Classification and geography of the flowering plants. Botanical Review 58: 225–348. Thulin M, Bremer B, Richardson J, Niklasson J, Fay MF, Chase MW. 1998. Family relationships MOLECULAR SYSTEMATICS OF MALVALES 295 of the enigmatic rosid genera Barbeya and Dirachma from the Horn of Africa region. Plant Systematics and Evolution 213: 103–119. Vickery JR. 1980. The fatty acid composition of seed oils from ten families with particular reference to cyclopropene and dihydrosterculic acids. Journal of the American Oil Chemists’ Society 57: 87–91. Vickery JR. 1981. The occurrence of dihydromalvalic in some seed oils. Journal of the American Oil Chemists’ Society 58: 731–732. Vogel S. 1977. Nektarien und ihre ökologische Bedeutung. Apidologie 8: 321–335. Williams LO, Standley PC. 1952. Pentaplaris, a new genus of Tiliaceae from Costa Rica. Ceiba 3: 139–142. Ying T-S, Zhang Y-L, Boufford DE. 1993. The endemic genera of seed plants of China. Beijing: Science Press. Zebe V. 1915. Monographie der Sterculiaceen-Gattungen Kleinhovia, Helicteres, Reevesia, Ungeria und Pterospermum (Allgemeiner Teil). Doctoral dissertation, Universität Breslau. Zhuge R. 1989. The cladistic analysis of the systematic position of Craigia. Acta Botanica Yunnanica 11: 17–23. Zhuge R. 1990. On the genus Burretiodendron sensu lato (Tiliaceae). Journal of the Arnold Arboretum 71: 371–380. 296 APPENDIX 1 Sources of plant material (family affiliation following Brummitt, 1992, except for Muntingiaceae) Family Species Accession/Voucher Reference for atpB Reference for rbcL Aceraceae Anacardiaceae Acer saccharum L. Pistacia vera L. Rhus vernix L. Schinus molle L. Bixa orellana L. Adansonia rubrostipa Jum. & H. Perrier Bombax buonopozense P. Beauv. Bombax ceiba L. Chorisia speciosa A. St.-Hil. Durio zibethinus Murray Durio zibethinus Murray Matisia cordata Humb. & Bonpl. Ochroma pyramidale (Cav. ex Lam.) Urb. Pachira aquatica Aubl. Brassica campestris L. Megacarpaea polyandra Benth. Stanleya pinnata (Pursh) Britton Stanleya pinnata (Pursh) Britton Capparis hastata Jacq. Capparis spinosa L. Carica papaya L. Carica papaya L. Cistus revolii Coste & Soulié Helianthemum grandiflorum DC. Tuberaria guttata Gross Cochlospermum intermedium Mildbr. Diegodendron humbertii Capuron Anisoptera marginata Korth. Aesculus pavia Castigl. Floerkea proserpinicoides Willd. Gossypium robinsonii F. Muell. Gossypium hirsutum L. Chase 106, NCU Terrazas s.n., CHAPA Terrazas s.n., CHAPA Anderson 13601, MICH Chase 243, NCU Chase 3043, K Alverson s.n., WIS Chase 3049, K Chase 3188, K Alverson 2180, WIS Chase 3039, K Kubitzki, Bayer & Appel 21, HBG Chase 244, NCU Chase 3189, K unknown Chase 565, K Price s.n., IND Chase 2748, K Iltis 30-315, WIS Chase 2751, K WIS Botanical Garden Chase 2508, K Chase 524, K Chase 525, K Chase 1075, K Fay s.n., K Capuron 23034, K Chase 2486, K Chase 503, K Reznicek 8609, MICH Wendel s.n., ISC Chase 3014, K Bakker et al., 1998; AF035893 unpublished Bakker et al., 1998; AF035912 Bakker et al., 1998; AF035914 Bakker et al., 1998; AF035897 this paper, AJ233050 Albert, Williams & Chase, 1992; L01881 Terrazas, unpublished Terrazas, unpublished Chase et al., 1993; U39270 Fay et al., 1998a; Y15139 this paper, AJ233115 Chase et al., 1993; AF022118 Bixaceae Bombacaceae Capparaceae Caricaceae Cistaceae Cochlospermaceae Diegodendraceae Dipterocarpaceae Hippocastanaceae Limnanthaceae Malvaceae this paper, AJ233053 this paper, AJ233054 Bakker et al., 1998; AF035910 this paper, AJ233056 this paper, AJ233116 Alverson et al., 1998; AF022119 C. BAYER ET AL. Brassicaceae this paper, AJ233051 this paper, AJ233052 this paper, AJ233117 this paper, AJ233118 this paper, AJ233119 Olmstead et al., 1992 unpublished Chase et al., 1993; M95753 unpublished Rodman et al., 1993; M95754 Bakker et al., 1998; AF035900 Rodman et al., 1993; M95671 Bakker et al., 1998; AF035901 Bakker et al., 1998; AF035902 Bakker et al., 1998; AF035907 this paper, AJ233059 this paper, AJ233060 this paper, AJ233061 Bakker et al., 1998; AF035918 Bakker et al., 1998; AF035894 Bakker et al., 1998; AF035904 Fay et al., 1998a; Y15140 Fay et al., 1998a; Y15141 this paper, AJ233120 Fay et al., 1998a; Y15143 Fay et al., 1998a; Y15138 Fay et al., 1998a; Y15144 Gadek et al., 1996; U39277 Chase et al., 1993; L12679 Chase et al., 1993; L13186 this paper, AJ233063 continued APPENDIX 1 continued Species Accession/Voucher Reference for atpB Reference for rbcL Malvaceae Hibiscus punaluuensis (Skottsb.) O. Deg. & I. Deg. Lavatera acerifolia Cav. Pavonia multiflora A.St.-Hil. Thespesia populnea (L.) Sol. ex Corrêa Dicraspidia donnell-smithii Standl. Chase 3045, K this paper, AJ233064 this paper, AJ233121 Chase 3035, K Chase 323, NCU Wendel s.n., ISC Pennington, Owen & Zamora 13583, K Chase 346, NCU Collenette 8-93, K Chase 3017, K Price s.n., IND Chase 115, NCU Chase 903, K Chase 906, K Chase 3081, K Alverson s.n., WIS Chase 3228, K Chase 3190, K Chase 273, NCU Kubitzki & Appel 102, HBG Thorne 54717, RSA Chase 3037, K this paper, AJ233065 Bakker et al., 1998; AF035916 this paper, AJ233122 this paper, AJ233123 Albert et al., 1992; L01961 Fay et al., 1998a; Y15145 Muntingiaceae Neuradaceae Resedaceae Sapindaceae Sarcolaenaceae Sphaerosepalaceae Sterculiaceae Muntingia calabura L. Neurada procumbens L. Reseda alba L. Reseda alba L. Koelreuteria paniculata Laxm. Sarcolaena sp. Rhopalocarpus sp. Abroma augusta (L.) l.f. Byttneria aculeata ( Jacq.) Jacq. Byttneria filipes Mart. ex K. Schum. Cola nitida (Vent.) Schott & Endl. Dombeya sp. Eriolaena spectabilis Planch. ex Hook.f. Fremontodendron mexicanum Fremontodendron californicum (Torr.) Cov. X mexicanum Davidson Helicteres baruensis Jacq. Helmiopsiella madagascariensis Arènes Hermannia erodioides Kuntze Hildegardia barteri (Mast.) Kosterm. Keraudrenia hermaniifolia J. Gay Lasiopetalum sp. Leptonychia pallida K. Schum. Paramelhania decaryana Arènes Pterospermum celebicum Miq. Reevesia thyrsoidea Lindl. Chase 3048, K Capuron 18625, K Chase 3046, K Chase 3187, K Chase 2194, K Chase 2195, K Cable 4571, K Chase 3038, K Chase 2142, K Chase 3185, K this paper, AJ233067 Bakker et al., 1998; AF035908 this paper, AJ233069 unpublished unpublished this paper, AJ233070 this paper, AJ233071 this paper, AJ233072 this this this this paper, paper, paper, paper, AJ233073 AJ233074 AJ233075 AJ233076 Fay et al., 1998a; Y15146 Morgan, Soltis & Robertson, 1994; U06814 Rodman et al., 1993; M95756 Gadek et al., 1996, U39283 Fay et al., 1998a; Y15147 Fay et al., 1998a; Y15148 this paper, AJ012208 Alverson et al., 1998; AF022123 this paper, AJ233124 this paper, AJ233123 this paper, AJ233126 Alverson et al., 1998; AF22124 Davidson this paper, AJ233077 this paper, AJ233078 this this this this this this this this paper, paper, paper, paper, paper, paper, paper, paper, AJ233080 AJ233081 AJ233082 AJ233083 AJ233084 AJ233085 AJ233114 AJ233086 this this this this this this this this this this paper, paper, paper, paper, paper, paper, paper, paper, paper, paper, AJ233127 AJ233129 AJ233130 AJ233131 AJ233132 AJ233133 AJ233134 AJ233135 AJ233136 AJ233137 MOLECULAR SYSTEMATICS OF MALVALES Family continued 297 298 APPENDIX 1 continued Species Sterculiaceae Ruizia cordata Cav. Chase 3227, K Rulingia sp. Chase 2196, K Sterculia apetala ( Jacq.) G. Karst. Chase 352, NCU Theobroma cacao L. Solheim BUF296, WIS Theobroma cacao L. Chase 3016, K Thomasia solanacea J. Gay Chase 3186, K Triplochiton zambesiacus Milne-Redhead Cubr 36100, B Trochetiopsis erythroxylon (G. Forst.) Marais Chase 3040, K Aquilaria beccariana Tiegh. Chase 1380, K Dais cotinifolia L. Chase 1381, K Gonystylus macrophyllus (Miq.) Airy Shaw Chase 1382, K Phaleria chermsideana (Bailey) C.T. White Conti 106, WIS Phaleria capitata Jack Chase 1383, K Thymelaea hirsuta Endl. Chase 1882, K Apeiba tibourbou Aubl. Kubitzki, Bayer & Appel 1, HBG Berrya javanica (Turcz.) Burret Chase 2143, K Brownlowia elata Roxb. Chase 2144, K Burretiodendron esquirolii (Lév.) Rehder Beusekom et al., 3852, K Christiana africana DC. Luke 2916, K Colona floribunda Craib Kubitzki & Appel 104, HBG Goethalsia meiantha Burret Richards 5873, K Grewia occidentalis L. Chase 3042, K Heliocarpus americanus L. Kubitzki, Bayer & Appel 11, HBG Microcos latistipulata (Ridl.) Burret Coode 7923, K Mortoniodendron anisophyllum (Standl.) Thomsen 74, K Standl. & Steyerm. Pentace polyantha Hassk. Chase 2147, K Pentaplaris doroteae L.O. Williams & Standl.Hammel 17697, K Petenaea cordata Lundell Pennington & MacQueen 13427, K Schoutenia glomerata King Kubitzki & Appel 108, HBG Sparrmannia ricinocarpa (Eckl. & Zeyh.)Chase 3229, K Kuntze Tilia americana L. Alverson s.n., WIS Tilia platyphyllos Scop. Chase 3018, K Tropaeolum majus L. Chase 113, NCU Tropaeolum tricolor Sweet Chase 2518, K Thymelaeaceae Tiliaceae Tropaeolaceae Accession/Voucher Reference for atpB Reference for rbcL this paper, AJ233087 this paper, AJ233088 this paper, AJ233089 this paper, AJ233138 this paper, AJ233139 this paper, AJ233140 Chase et al., 1993; AF022125 this this this this this this this paper, paper, paper, paper, paper, paper, paper, AJ233090 AJ233091 AJ233092 AJ233093 AJ233079 AJ233094 AJ233095 this paper, AJ233096 this paper, AJ233097 this paper, AJ233098 Bakker et al., 1998; AF035896 Bakker et al., 1998; AF035898 this paper, AJ233101 this paper, AJ233102 this paper, AJ233103 this paper, AJ233104 this paper, AJ233105 this paper, AJ233106 this paper, AJ233107 this paper, AJ233108 this paper, AJ233109 this paper, AJ233110 this paper, AJ233111 this paper, AJ233112 this paper, AJ233141 this paper, AJ233142 this paper, AJ233143 Fay et al., 1998a; Y15149 this paper, AJ233144 Fay et al., 1998a; Y15150 Conti, Litt & Sytsma, 1996; U26332 Fay et al., 1998a; Y15151 this paper, AJ233145 this paper, AJ233146 this paper, AJ233147 this paper, AJ233148 this paper, AJ233149 this paper, AJ233150 this paper, AJ233151 this paper, AJ233152 this paper, AJ233153 this paper, AJ233154 this paper, AJ233155 this this this this this paper, paper, paper, paper, paper, AJ233156 AJ233157 AJ233158 AJ233159 AJ233128 Chase et al., 1993; AF022127 this paper, AJ233113 Price & Palmer, 1993; L14706 Bakker et al., 1998; AF035917 C. BAYER ET AL. Family APPENDIX 2 Placement of core Malvales genera in previous classifications and in the present treatment Genus Jussieu (1789) Baillon (1873) Schumann (1895) Hutchinson (1967) this paper (Malvaceae–) Hermannia Waltheria Melochia Dicarpidium Rayleya Megatritheca Byttneria Ayenia Rulingia Commersonia Keraudrenia Seringia Lasiopetalum Thomasia Hannafordia Lysiosepalum Guichenotia Maxwellia Herrania Theobroma Guazuma Abroma Kleinhovia Scaphopetalum Leptonychia Glossostemon Tiliaceae dubiae Malvaceae — Malvaceae–Hermannieae Malvaceae–Hermannieae Sterculiaceae–Hermannieae Byttnerioideae — — Malvaceae–Buettnerieae Sterculiaceae– Büttnerieae– Büttnerinae Sterculiaceae–Byttnerieae Malvaceae–Lasiopetaleae Sterculiaceae–Lasiopetaleae Sterculiaceae–Lasiopetaleae Malvaceae–Buettnerieae — (sub Theobroma) Sterculiaceae– Büttnerieae– Theobrominae Sterculiaceae–Helictereae Sterculiaceae– Büttnerieae– Theobrominae — Malvaceae — Malvaceae–Helictereae Malvaceae– Buettnerieae MOLECULAR SYSTEMATICS OF MALVALES Malvaceae — Sterculiaceae–Theobromeae Sterculiaceae–Helictereteae Sterculiaceae– Theobromeae continued 299 incertae sedis (Byttnerioideae?) APPENDIX 2 continued 300 Genus Jussieu (1789) Goethalsia Pseudocorchorus Schumann (1895) Hutchinson (1967) this paper (Malvaceae–) — — Grewioideae Tiliaceae–Tilieae Tiliaceae–Tilieae Flacourtiaceae Tiliaceae– Pseudocorchoreae Tiliaceae– Enteleeae Tiliaceae– Sparrmanieae Tiliaceae–Tilieae Tiliaceae– Desplatzieae Tiliaceae verae — Tiliaceae–Grewieae — Tiliaceae–Tilieae Tiliaceae verae (sub Grewia) — — Tiliaceae–Tilieae Tiliaceae–Tilieae (sub Grewia) Tiliaceae–Tilieae/Grewieae Tiliaceae–Grewieae (sub Grewia) — — (sub Columbia) Tiliaceae–Tilieae Tiliaceae–Grewieae C. BAYER ET AL. Entelea Corchorus Sparrmannia Clappertonia Duboscia Desplatsia Hydrogaster Vasivaea Mollia Luehea Trichospermum Grewia Microcos Colona Eleutherostylis Lueheopsis Tetralix Diplophractum Erinocarpus Triumfetta Heliocarpus Apeiba Glyphaea Ancistrocarpus Tilia Craigia Baillon (1873) Tiliaceae– Lueheeae Tiliaceae–Lueheeae/Grewieae Tiliaceae–Grewieae (sub Grewia) Tiliaceae–Triumfetteae Tiliaceae verae (sub Triumfetta) Tiliaceae–Tilieae Tiliaceae–Apeibeae Tiliaceae–Apeibeae — Tiliaceae–Tilieae — Tiliaceae–Tilieae Sterculiaceae– Lasiopetaleae — Tiliaceae verae — Tilioideae incertae sedis (Tilioideae?) continued APPENDIX 2 continued Genus Jussieu (1789) Baillon (1873) Schumann (1895) Triplochiton Sterculiaceae–Tarrietieae Helicteroideae Sterculiaceae–Helictereteae Malvaceae — Malvaceae–Helictereae Sterculiaceae–Helictereae Malvaceae–Bombaceae Bombacaceae–Durioneae Bombacaceae–Durioneae incertae sedis (Helicteroideae?) — — Tiliaceae–Diplodisceae Brownlowioideae Tiliaceae–Brownlowieae Tiliaceae–Brownlowieae Tiliaceae–Brownlowieae Tiliaceae–Berryeae — — Malvaceae — Malvaceae–Sterculieae — Tiliaceae–Enteleeae Sterculiaceae–Sterculieae incertae sedis Sterculioideae Sterculiaceae–Sterculieae — (sub Firmiana) Sterculiaceae–Tarrietieae Malvaceae–Sterculieae — 301 Jarandersonia Hainania Diplodiscus Pityranthe Brownlowia Christiana Berrya Carpodiptera Pentace Mortoniodendron Acropogon Brachychiton Sterculia Cola Octolobus Scaphium Pterocymbium Pterygota Firmiana Hildegardia Heritiera Franciscodendron this paper (Malvaceae–) MOLECULAR SYSTEMATICS OF MALVALES Mansonia Neoregnellia Helicteres Reevesia Ungeria Durio alliance Hutchinson (1967) — — continued 302 APPENDIX 2 continued Genus Jussieu (1789) Baillon (1873) Schumann (1895) Hutchinson (1967) this paper (Malvaceae–) Helmiopsiella — — — Sterculiaceae– Dombeyoideae Helmiopsis Helmiopsideae Eriolaena Dombeya Malvaceae Malvaceae–Helictereae Sterculiaceae–Eriolaeneae Sterculiaceae–Eriolaeneae Malvaceae–Dombeyeae Sterculiaceae–Dombeyeae Sterculiaceae–Dombeyeae — — Melhania Ruizia C. BAYER ET AL. Pentapetes Astiria — Trochetia Cheirolaena Harmsia Paradombeya Paramelhania Trochetiopsis Corchoropsis Tiliaceae–Tilieae — Tiliaceae–Corchoropsideae Pterospermum Malvaceae–Helictereae Sterculiaceae–Helictereae Sterculiaceae–Helictereteae Schoutenia Tiliaceae–Tilieae Tiliaceae–Tilieae/Brownlowieae Tiliaceae–Tilieae Burretiodendron — — Tiliaceae–Enteleeae Sicrea Nesogordonia Sterculiaceae– Helmiopsideae incertae sedis (Dombeyoideae?) continued Genus Jussieu (1789) Baillon (1873) Schumann (1895) Hutchinson (1967) this paper (Malvaceae–) Bombax, Matisia, Adansonia, Ochroma, etc. Malvaceae (as far as known) Malvaceae–Bombaceae (as far as known) Bombacaceae: various tribes (as far as known) Bombacaceae: various tribes (as far as known) Bombacoideae Fremontodendron — (sub Chiranthodendron) Sterculiaceae–Fremontieae Sterculiaceae–Fremontieae Chiranthodendron Malvaceae–Chiranthodendreae Pentaplaris — — Tiliaceae–Brownlowieae Gossypium, Malvaceae (as far Hibiscus, Pavonia, as known) Lavatera, etc. Malvaceae: various tribes (as far as known) Malvaceae: various tribes (as far as known) Malvaceae: various tribes (as far as known) Malvoideae MOLECULAR SYSTEMATICS OF MALVALES APPENDIX 2 continued 303