PHYLOGENY OF TRICALYSIA (RUBIACEAE) AND ITS RELATIONSHIPS WITH ALLIED GENERA BASED ON PLASTID DNA DATA: RESURRECTION OF THE GENUS EMPOGONA1 James Tosh,2 Aaron P. Davis,3 Steven Dessein,4 Petra De Block,4 Suzy Huysmans,2 Mike F. Fay,3 Erik Smets,2,5 and Elmar Robbrecht4 ABSTRACT Recent studies on the circumscription of the tribe Coffeeae (Rubiaceae) revealed a weakly supported clade containing Tricalysia A. Rich. and the allied genera Argocoffeopsis Lebrun, Calycosiphonia Pierre ex Robbr., Belonophora Hook. f., Diplospora DC., Discospermum Dalzell, Nostolachma T. Durand, and Xantonnea Pierre ex Pit. The phylogenetic relationships of Tricalysia and these allied taxa are investigated further using sequence data from four plastid regions (trnL-F intron and intergenic spacer, rpL16 intron, accD-psa1 intergenic spacer, and PetD). Our results demonstrate that Tricalysia sensu Robbrecht is not monophyletic. The genus name Tricalysia should be restricted to taxa from subgenus Tricalysia; subgenus Empogona (Hook. f.) Robbr. is sister to the genus Diplospora and is recognized at the generic level. The 34 necessary new combinations for Empogona Hook. f. are provided: E. acidophylla (Robbr.) J. Tosh & Robbr., E. aequatoria (Robbr.) J. Tosh & Robbr., E. africana (Sim) J. Tosh & Robbr., E. aulacosperma (Robbr.) J. Tosh & Robbr., E. bequaertii (De Wild.) J. Tosh & Robbr., E. bracteata (Hiern) J. Tosh & Robbr., E. breteleri (Robbr.) J. Tosh & Robbr., E. buxifolia (Hiern) J. Tosh & Robbr. subsp. buxifolia, E. buxifolia subsp. australis (Robbr.) J. Tosh & Robbr., E. cacondensis (Hiern) J. Tosh & Robbr., E. concolor (N. Hallé) J. Tosh & Robbr., E. coriacea (Sond.) J. Tosh & Robbr., E. crepiniana (De Wild. & T. Durand) J. Tosh & Robbr., E. deightonii (Brenan) J. Tosh & Robbr., E. discolor (Brenan) J. Tosh & Robbr., E. filiformistipulata (De Wild.) Bremek. subsp. filiformistipulata, E. filiformistipulata subsp. epipsila (Robbr.) J. Tosh & Robbr., E. glabra (K. Schum.) J. Tosh & Robbr., E. gossweileri (S. Moore) J. Tosh & Robbr., E. kirkii Hook. f. subsp. junodii (Schinz) J. Tosh & Robbr., E. lanceolata (Sond.) J. Tosh & Robbr., E. macrophylla (K. Schum.) J. Tosh & Robbr., E. maputenis (Bridson & A. E. van Wyk) J. Tosh & Robbr., E. ngalaensis (Robbr.) J. Tosh & Robbr., E. nogueirae (Robbr.) J. Tosh & Robbr., E. ovalifolia (Hiern) J. Tosh & Robbr. var. ovalifolia, E. ovalifolia var. glabrata (Oliv.) J. Tosh & Robbr., E. ovalifolia var. taylorii (S. Moore) J. Tosh & Robbr., E. reflexa (Hutch.) J. Tosh & Robbr. var. reflexa, E. reflexa var. ivorensis (Robbr.) J. Tosh & Robbr., E. ruandensis (Bremek.) J. Tosh & Robbr., E. somaliensis (Robbr.) J. Tosh & Robbr., E. talbotii (Wernham) J. Tosh & Robbr., and E. welwitschii (K. Schum.) J. Tosh & Robbr. Key words: accD-psal, Coffea, coffee, Coffeeae, Empogona, molecular systematics, petD, rpl16, Rubiaceae, Tricalysia, trnL-F. The genus Tricalysia A. Rich. is one of the largest genera of Rubiaceae in Africa and occurs in continental Africa (ca. 95 species), Madagascar (12 species), and the Comoros (one species). The genus typically possesses the distinguishing characteristics of the tribe Coffeeae (Bridson & Verdcourt, 2003; Davis et al., 2007). These include axillary inflorescences paired at the nodes with obvious calyculi, flowers with left contorted corolla aestivation and a distinctly 2-lobed style, and relatively small and fewseeded fleshy fruits. Most Tricalysia species can be separated readily from other Coffeeae by the presence of stipules with needlelike awns, truncate to distinctly lobed calyces, and seeds with a shallow hilum. Identification of Tricalysia at the species level is notoriously difficult, as the genus contains a large number of species across a broad geographic and ecologic range, often separated by minor and continuous characters. In a series of papers, Robbrecht (1978, 1979, 1982, 1983, 1987) conducted a taxonomic revision of Tricalysia, with later contributions by Ali and 1 James Tosh would like to acknowledge all members of the conservation genetics, molecular systematic, and Rubiaceae research groups at the Royal Botanic Gardens, Kew, who provided help and support during my research visit in 2006. The authors would also like to thank the reviewers of the paper for their helpful comments and suggestions. This research was supported financially by grants from the Fund for Scientific Research–Flanders (FWO, G.0250.05 and G.0268.04). 2 Laboratory of Plant Systematics, Katholieke Universiteit Leuven, Kasteelpark Arenberg 31, P.O. Box 2437, BE-3001 Leuven, Belgium. Corresponding author: james.tosh@bio.kuleuven.be. 3 Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3AE, United Kingdom. 4 National Botanic Garden of Belgium, Domein van Bouchout, BE-1860 Meise, Belgium. 5 National Herbarium of The Netherlands, Leiden University Branch, P.O. Box 9514, NL-2300 RA Leiden, The Netherlands. doi: 10.3417/2006202 ANN. MISSOURI BOT. GARD. 96: 194–213. PUBLISHED ON 23 APRIL 2009. Volume 96, Number 1 2009 Tosh et al. Phylogeny of Tricalysia 195 Robbrecht (1991) and Ranarivelo-Randriamboavonjy et al. (2007). Robbrecht (1979, 1982, 1983, 1987) recognized and revised two subgenera: subgenera Tricalysia A. Rich with five sections (Probletostemon (K. Schum.) Robbr., Tricalysia, Rosea (Klotzsch) Robbr., Ephedranthera Robbr., and an unnamed Madagascan section) and subgenus Empogona (Hook. f.) Robbr. with two sections (Empogona Hook. f. and Kraussiopsis Robbr.). Separation of the two subgenera in Tricalysia was supported by differences in calyx lobe morphology, corolla throat pubescence, fruit color, and the presence/absence of a sterile appendage on the anther connective. Empogona Hook. f. was originally recognized at the generic level by Hooker (1873) based on a single Zambezian species, E. kirkii Hook. f. Brenan (1947) reduced the genus Empogona to a section of Tricalysia, containing six mainly eastern and southern African species. During his revision of Tricalysia and, in particular, his treatment of subgenus Empogona, Robbrecht (1979) showed that ca. 20 other tropical African species, many of them with Guineo-Congolian distribution, also belonged to this subgenus. Robbrecht (1978) also investigated the closely related genus Neorosea N. Hallé, consisting of 17 species, many of which were formerly included in Tricalysia. Two of these 17 species, including the type species N. jasminiflora (Klotzsch) N. Hallé, proved to be genuine Tricalysia species; a new genus, Sericanthe Robbr., was described to accommodate the remaining species (Robbrecht, 1978). The close association between Diplospora DC. and Discospermum Dalzell with Tricalysia has long been recognized, with some authors (e.g., Schumann, 1891) considering Diplospora and Tricalysia to be synonymous. Ali and Robbrecht (1991) broadly surveyed Diplospora and Discospermum and enumerated a whole suite of characters that could be used to distinguish these Asian taxa from the closely related African Tricalysia species. They also justified Diplospora and Discospermum as separate genera on the basis of fruit morphology. The most recent taxonomic work on Tricalysia, by Ranarivelo-Randriamboavonjy et al. (2007), focused on the unnamed Madagascan section that was alluded to, but not treated by, Robbrecht (1987). Of the 12 species of Tricalysia occurring in Madagascar, only one species belongs to subgenus Empogona (T. ovalifolia Hiern). The other 11 species, characterized by the presence of unisexual flowers, belong to subgenus Tricalysia. Ranarivelo-Randriamboavonjy et al. (2007) observed that the Madagascan taxa could be accommodated within section Tricalysia were it not for the presence of unisexual flowers. As a result, they formally placed these 11 taxa in Androgyne Robbr., a new section within subgenus Tricalysia. Recent phylogenetic investigations incorporating morphological and molecular data sets have enabled us to improve our understanding of the systematic position of Tricalysia and its relationships with associated genera (Andreasen & Bremer, 2000; Persson, 2000; Bridson & Verdcourt, 2003; Robbrecht & Manen, 2006; Davis et al., 2007). Andreasen and Bremer (2000) assessed tribal and generic delimitation in subfamily Ixoroideae using morphology, plastid and nuclear ribosomal DNA sequences, and restriction site (restriction fragment length polymorphism) data. Their results highlighted the close affinity between Coffea L. and Psilanthus Hook. f. (Coffeeae s. str.) and several members of the Gardenieae subtribe Diplosporinae (Diplospora and Tricalysia), resulting in an expanded circumscription of the tribe Coffeeae to include Tricalysia, Diplospora, Discospermum, Sericanthe, Coffea, Psilanthus, and Bertiera Aubl. Bridson and Verdcourt (2003) further enlarged and modified the concept of Coffeeae on the basis of morphology and provisional plastid data (provided by A. P. Davis, unpublished). In contrast to the studies of Andreasen and Bremer (2000) and a broader study of the Rubiaceae (Robbrecht & Manen, 2006), the genus Bertiera was excluded from Coffeeae and placed in its own tribe, Bertiereae. Davis et al. (2007) reexamined the circumscription and phylogeny of Coffeeae and Bertiera using sequence data from three plastid regions (trnL-F intron and intergenic spacer, accD-psa1, and rpl16) in combination with morphological data. Their study confirmed the placement of Tricalysia and related taxa (Sericanthe, Diplospora, and Discospermum) with Coffea and Psilanthus, and expanded Coffeeae to include six other genera (Argocoffeopsis Lebrun, Belonophora Hook. f., Nostolachma T. Durand, Calycosiphonia Pierre ex Robbr., and Xantonnea Pierre ex Pit.). However, this study only surveyed a limited number of Tricalysia species, all of which belong to subgenus Tricalysia. Bertiera was excluded from Coffeeae and retained in Bertiereae, in agreement with Bridson and Verdcourt (2003), and Gardenieae subtribe Diplosporinae was placed in synonymy with Coffeeae. The current investigation uses DNA sequence data to test the monophyly of Tricalysia as currently circumscribed and to assess the accuracy of the subgeneric classification for the genus (Robbrecht, 1979, 1982, 1983, 1987). This is the first molecular study to include widespread and representative sampling of Tricalysia. In addition, we reassess the phylogenetic relationships within the broadly circumscribed Coffeeae, with an expanded sampling from both subgenera of Tricalysia. Given the wealth of trnL-F, rpl16, and accD-psa1 sequence data already 196 Annals of the Missouri Botanical Garden Table 1. Summary of species from Tricalysia subgen. Empogona and subgen. Tricalysia sampled in this study (following classification of Robbrecht, 1979, 1982, 1983, 1987). A) Tricalysia subgen. Empogona (ca. 27 spp., Robbrecht, 1979) Section Tricalysia sect. Empogona Hook. f. 12 spp. sensu Robbrecht, 1979 Species group Species T. discolor group T. acidophylla Robbr. T. junodii group T. junodii (Schinz) Brenan T. ngalaensis Robbr. T. ovalifolia Hiern No known group affiliation within sect. Empogona Tricalysia sect. Kraussiopsis Robbr. 15 spp. sensu Robbrecht, 1979 T. crepiniana group T. concolor N. Hallé T. gossweileri S. Moore T. bequaertii De Wild. T. talbotii (Wernham) Keay T. ruandensis group T. cacondensis Hiern T. lanceolata (Sond.) Burtt Davy T. ruandensis Bremek. B) Tricalysia subgen. Tricalysia (ca. 75 spp., Robbrecht, 1982, 1983, 1987) Section Species group Tricalysia sect. Probletostemon (K. Schum.) Robbr. 4 spp. sensu Robbrecht, 1983 Species T. anomala E. A. Bruce T. elliottii (K. Schum.) Hutch. & Dalziel Tricalysia sect. Ephedranthera Robbr. 9 spp. sensu Robbrecht, 1982 T. aciculiflora Robbr. T. acocantheroides K. Schum. T. bridsoniana Robbr. Tricalysia sect. Tricalysia 40 spp. sensu Robbrecht, 1987 T. angolensis group T. griseiflora K. Schum. Core group for sect. Tricalysia T. bagshawei S. Moore T. coriacea (Benth.) Hiern T. microphylla Hiern T. okelensis Hiern T. pallens Hiern Tricalysia sect. Rosea (Klotzsch) Robbr. 9 spp. sensu Robbrecht, 1987 T. jasminiflora (Klotzsch) Benth. & Hook. f. ex Hiern T. schliebenii Robbr. Tricalysia sect. Androgyne Robbr. 11 spp. sensu Ranarivelo-Randriamboavonjy et al., 2007 T. ambrensis Randriamb. & De Block T. analamazaotrensis Homolle ex Randriamb. & De Block T. cryptocalyx Baker T. dauphinensis Randriamb. & De Block T. leucocarpa (Baill.) Randriamb. & De Block T. perrieri Homolle ex Randriamb. & De Block Volume 96, Number 1 2009 Tosh et al. Phylogeny of Tricalysia 197 Table 2. Amplification primers used in this study. Region trnL-F rpl16 accD-psa1 petD Primer Primer sequence (59-39) Forward (c) Reverse (f) Forward (71f) Reverse (1661r) Reverse (1516r) Internal forward Internal reverse Forward (769 F) Reverse (75 R) Forward (1365) Reverse (738) CGA AAT CGG TAG ACG CTA CG AAT TGA ACT GGT GAC ACG AG GCT ATG CTT AGT GTG TGA CTC GTT G CGT ACC CAT ATT TTT CCA CCA CGA C CCC TTC ATT CTT CCT CTA TGT TG GTA AGA AGT GAT GGG AAC GA TCG TTC CCA TCA CTT CTT AC GGA AGT TTG AGC TTT ATG CAA ATG AGA AGC CAT TGC AAT TGC CGG AAA TTG ACY CGT TTT TAT AGT TTA C AAT TTA GCY CTT AAT ACA GG available for Coffeeae (Davis et al., 2007), we have focused on these three plastid regions in the current investigation and included further sequence data from the plastid region petD. MATERIALS AND METHODS TAXON SAMPLING An expanded sampling of Tricalysia, Diplospora, Discospermum, Sericanthe, and Bertiera was combined with sequence data generated by Davis et al. (2007). Tricalysia samples representing both subgenera and all of the seven sections of the genus (Robbrecht, 1979, 1982, 1983, 1987) were included in the analyses (Table 1). Representative taxa from Ixoreae, Octotropideae, and Gardenieae were selected as the outgroup. A list of the 80 accessions used in the study is given in Appendix 1. DNA EXTRACTION, POLYMERASE CHAIN REACTION AMPLIFICATION, AND SEQUENCING Most DNA samples were obtained from silica gel collections or, alternatively, from seed, flower, or leaf samples taken from herbarium specimens (BR, K, MO, WAG). A small number of DNA samples were obtained from fresh leaf material collected from the living collections of the National Botanic Garden of Belgium. For silica gel samples, DNA was isolated using a modified DNA Mini Extraction Protocol (Royal Botanic Gardens, Kew [K] protocol). DNA samples were obtained from herbarium material using the 23 CTAB protocol of Doyle and Doyle (1987), with the DNA subsequently purified using cesium chloride/ ethidium bromide gradients and concentrated by dialysis before inclusion in the DNA Bank at K. All DNA samples were purified using a NucleoSpin purification column (Macherey-Nagel, Bethlehem, Reference Taberlet et al., 1991 Jordan et al., 1996 Shaw et al., 2005 Davis et al., 2007 Mendenhall, 1994 Löhne & Borsch, 2004 Pennsylvania, U.S.A.) according to the manufacturer’s instructions in order to remove any potential polymerase chain reaction (PCR) inhibitors. Amplification of the trnL-F, rpl16, petD, and accDpsa1 plastid regions was carried out using the primers listed in Table 2. Amplification of the rpl16 region was primarily carried out using the forward primer 71f and the reverse primers 1661r (Jordan et al., 1996) and 1516R (Shaw et al., 2005), although Coffeeae specific internal primers designed by K were also required for certain taxa (Davis et al., 2007). All PCR and sequencing reactions were performed using a Perkin Elmer (Waltham, Massachusetts, U.S.A) GeneAmp 9700 Thermal Cycler machine. Amplification of trnL-F was carried out using the following profile: 94uC for 3 min.; 32 cycles of 94uC for 1 min., 51uC for 1 min., 72uC for 2 min.; and a final extension of 72uC for 7 min. accD-psa1 and rpl16 were amplified as follows: 94uC for 3 min.; 32 cycles of 94uC for 1 min., 52uC for 1 min., 72uC for 1 min. 30 sec.; and a final extension of 72uC for 7 min. Amplification of petD was carried out as follows: 96uC for 2 min.; 34 cycles of 94uC for 1 min., 50uC for 1 min., 72uC for 1 min. 30 sec.; and a final extension of 72uC for 10 min. For the trnL-F, petD, and rpl16 regions, 25 ml PCR reactions were made using a commercial PCR master mix (2.5 mM MgCl2 ReddyMix; ABgene; Epsom, Surrey, U.K.). accD-psa1 did not amplify successfully with the commercial master mix, so 25 ml PCR master mixes were prepared using Biotaq DNA polymerase (Bioline, London, U.K.), 2.5 ml of 103 NH4 reaction buffer (Bioline), 1.5 ml of 50 mM MgCl2, and 2.5 ml of dNTPs (Promega, Madison, Wisconsin, U.S.A.). All amplified PCR products were purified using NucleoSpin purification columns following the manufacturer’s protocol. Cycle sequencing reactions were carried out using BigDye Terminator Mix version 3.1 (Applied Biosystems, Inc., Warrington, Cheshire, U.K.). The cycle 198 Annals of the Missouri Botanical Garden sequence reaction consisted of 26 cycles of 10 sec. at 96uC, 5 sec. at 50uC, and 4 min. at 60uC. Cycle sequencing products were cleaned with the MagneSil Clean-Up System (Promega) using an automated robot (Biomek NX S8; Beckman Coulter, High Wycombe, Buckinghamshire, U.K.). Analysis of cycle sequencing products was performed using an AB 3730 DNA Analyzer (Applied Biosystems). In addition, a number of the trnL-F and petD samples were sent to Macrogen (Seoul, South Korea) for sequencing. was selected for the trnL-F sequence matrix. The combined data set was partitioned into five discrete units. In addition to the four plastid regions, there was a fifth partition for the phylogenetically informative indels. The restriction site (binary) model of evolution was implemented for the indel data, following the recommendation of Ronquist et al. (2005). Four independent Bayesian searches, each consisting of two simultaneous parallel analyses, were carried out using MrBayes 3.1 (Huelsenbeck & Ronquist, 2001). In each Bayesian analysis, four Markov chains (three heated, one cold) were run simultaneously for 2,000,000 generations, sampling trees every 100 generations. The initial 25% of trees were discarded as a conservative burn-in. After confirming by eye that trees generated from separate analyses had consistent topologies, the ‘‘post-burn-in’’ trees from each analysis were pooled together, imported into PAUP* version 4.0b10 (Swofford, 2003), and summarized by majority rule consensus, with values on the tree equating to posterior probabilities (PP). ALIGNMENT AND GAP CODING Sequences were assembled and edited using the Staden software package (Staden et al., 1998). All sequences were aligned manually in MacClade (version 4.04, Maddison & Maddison, 2002). Low levels of sequence variation enabled sequences to be aligned without difficulty. Regions of ambiguous alignment, such as the beginning and end of sequences, were removed. The edited sequences were analyzed with gaps treated as missing data and phylogenetically informative indels (insertions and/or deletions) coded according to the ‘‘simple indel coding’’ method of Simmons and Ochoterena (2000). PHYLOGENETIC ANALYSES Phylogenetic analyses were performed on the four separate plastid data sets in addition to the combined four-region plastid matrix. Maximum parsimony. Heuristic tree searches were carried out in PAUP* version 4.0b10 (Swofford, 2003) using 10,000 replicates of random taxon sequence addition, holding 10 trees at each step, with tree bisection-reconnection (TBR) branch swapping, delayed transformation (DELTRAN) optimization, and MULTREES in effect, and saving no more than 10 trees per replicate. Support values for clades recovered in the analyses were estimated using bootstrap analysis (Felsenstein, 1985). One thousand replicates of simple sequence addition, TBR swapping, and saving 10 trees per replicate were performed in PAUP*. We interpreted bootstrap values greater than 85% as being well supported, 75%–84% as being moderately supported, and 50%–74% as having low support. Bayesian inference. Evolutionary models for each plastid region were selected using Modeltest v3.06 (Posada & Crandall, 1998) under the Akaike information criterion. In the case of accD-psa1, petD, and rpl16, the nucleotide substitution model that best fits the data was HKY + I + G. The HKY + I model RESULTS This study generated 229 sequences, which were combined with the 75 sequences obtained by Davis et al. (2007). In total, this study included 79 accD-psa1 sequences (53 newly generated), 80 trnL-F sequences (54 newly generated), 78 rpl16 sequences (55 newly generated), and 67 petD sequences (all newly generated). The rpl16 region proved to be the most problematic region to amplify, due in part to two polyA stretches (one 373 bp from the 59 end, the other 466 bp from the 39 end). As a result, it was often difficult to obtain sufficient overlap during sequence assembly. Internal primers, designed specifically for Coffeeae taxa (Davis et al., 2007), were used to obtain a complete sequence for rpl16 in problematic taxa. In general, the amount of genetic variability in all plastid regions was low (Table 3). A large proportion of the total genetic variation occurred between the ingroup (Coffeeae) taxa and outgroup (other Ixoroideae). We observed considerable length variability in the accD-psa1 region. As with all the plastid regions investigated, accD-psa1 is particularly AT-rich and subject to several repeat units, giving rise to a number of potentially phylogenetically informative indels. In the case of Tricalysia subgen. Empogona, all taxa included in the study share a 250 bp deletion in the accD-psa1 region. Less length variation was observed in petD, rpl16, and trnL-F. The gross tree topologies of all four individual analyses were examined by eye and found to be topologically consistent, and the four data sets were subsequently combined in all further analyses. 80 4415 — 3796 50 352 (7.9) 929 0.816 0.908 8853 80 889 772–822 765 9 74 (8.3) 174 0.822 0.916 9990 No. of taxa Total length (base pairs) Sequence length variation No. of constant characters No. of phylogenetically informative indels No. of variable PI characters (% of total characters) Tree length Consistency index Retention index No. of trees saved 79 1255 737–1061 1075 22 117 (9.3) 283 0.827 0.923 9920 78 1207 995–1068 982 11 116 (9.6) 339 0.814 0.890 1056 67 1064 937–966 974 8 45 (4.2) 123 0.854 0.937 1392 Combined plastid accD-psa1 Table 3. Characteristics of accD-psa1, rpl16, petD, trnL-F, and combined data sets and tree statistics. rpl16 petD trnL-F Volume 96, Number 1 2009 Tosh et al. Phylogeny of Tricalysia 199 The aligned combined matrix had a total length of 4465 bp. There were 669 variable characters and, of these, 352 characters were parsimony informative (7.9% of total number of characters). In total, the matrix contained 50 parsimony informative indels, consisting of repeat sequences in addition to insertion/ deletion events. Exclusion of outgroup taxa (Ixoreae, Gardenieae, Octotropideae, and Bertierieae) revealed 211 parsimony informative characters within Coffeeae. PHYLOGENETIC RESULTS The heuristic maximum parsimony (MP) analysis of the combined plastid data matrix generated 8853 most parsimonious trees with a length of 929 steps, a consistency index (CI) of 0.816, and a retention index (RI) of 0.908. Table 3 summarizes the tree statistics for the individual and combined analyses. The topologies of the MP strict consensus tree and the Bayesian majority rule tree (Fig. 1) were consistent with each other. Figure 2 displays one of the most parsimonious trees and indicates both bootstrap support (BS) and branch length. Both MP and Bayesian analyses reconfirm the monophyly of the ingroup (BS 99%, PP 1.00). Bertiera, here represented by its two subgenera, is monophyletic (BS 100%, PP 1.00) and is sister to the ingroup (BS 79%, PP 1.00). The clade of Coffea and Psilanthus is well supported (BS 100%, PP 1.00) and is sister to the remaining ingroup taxa (BS 93%, PP 1.00). There is also strong support for the clade of Argocoffeopsis and Calycosiphonia (BS 99%, PP 1.00). The sister relationship of Calycosiphonia and Argocoffeopsis to the rest of the ingroup receives weak bootstrap support (BS 50%), but is supported by a PP of 0.98. Both MP and Bayesian analyses recovered a clade including Sericanthe, Diplospora, Discopermum, and Tricalysia subgen. Empogona. Although there is no bootstrap support for this clade (BS , 50%), the clade does receive support in the Bayesian analyses (PP 0.98). Within this clade, there is strong support for the monophyly of Sericanthe (BS 99%, PP 1.00), Discospermum (BS 100%, PP 1.00), and the group of Diplospora and Tricalysia subgen. Empogona (BS 99%, PP 1.00). The monophyly of both Diplospora (BS 90%, PP 1.00) and Tricalysia subgen. Empogona (BS 98%, PP 1.00) is confirmed. Within Tricalysia subgen. Empogona, two groups receive high levels of support: the group of T. cacondensis Hiern, T. lanceolata (Sond.) Burtt Davy, and T. ruandensis Bremek. (BS 85%, PP 1.00); and the group of T. junodii (Schinz) Brenan, T. ovalifolia, and T. acidophylla Robbr. (BS 98%, PP 1.00). The clade of Belonophora and Tricalysia subgen. Tricalysia is present in both the MP strict consensus 200 Annals of the Missouri Botanical Garden Figure 1. Maximum parsimony strict consensus/Bayesian majority rule consensus tree. Bayesian posterior probabilities are indicated above branches. Sectional groupings are annotated after species names: AND, Tricalysia sect. Androgyne; EMP, Tricalysia sect. Empogona; EPH, Tricalysia sect. Ephedranthera; KRA, Tricalysia sect. Kraussiopsis; PRO, Tricalysia sect. Probletostemon; ROS, Tricalysia sect. Rosea; TRI, Tricalysia sect. Tricalysia. See Table 1 for species authorities and provenance. Volume 96, Number 1 2009 Tosh et al. Phylogeny of Tricalysia 201 Figure 2. One of the 8853 most parsimonious trees generated in the maximum parsimony analysis. Bootstrap values . 50% are indicated above branches, and selected branch lengths are indicated below branches. Sectional groupings are annotated after species names: AND, Tricalysia sect. Androgyne; EMP, Tricalysia sect. Empogona; EPH, Tricalysia sect. Ephedranthera; KRA, Tricalysia sect. Kraussiopsis; ROS, Tricalysia sect. Rosea; PRO, Tricalysia sect. Probletostemon; TRI, Tricalysia sect. Tricalysia. See Table 1 for species authorities and provenance. tree and the Bayesian majority rule tree, although there is negligible support for this clade (BS , 50%, PP 0.82). However, the monophyly of Belonophora (BS 100%, PP 1.00) and Tricalysia subgen. Tricalysia (BS 99%, PP 1.00) is strongly supported. Within Tricalysia subgen. Tricalysia, several groups receive strong support: the group of T. elliottii (K. Schum.) Hutch. & Dalziel and T. anomala E. A. Bruce (BS 95%, PP 1.00), and a group of predominantly Madagascan taxa with the inclusion of T. jasminiflora (Klotzsch) Benth. & Hook. f. ex Hiern (BS 97%, PP 1.00). There is also moderate bootstrap (BS 75%) and high PP (PP 1.00) for the clade of T. acocantheroides K. Schum., T. griseiflora K. Schum., T. bridsoniana Robbr., T. 202 Annals of the Missouri Botanical Garden microphylla Hiern, T. schliebenii Robbr., and the aforementioned Madagascan group together with T. jasminiflora. of Davis et al. (2007). First, we did not recover an Asian clade. Instead, Diplospora formed a wellsupported monophyletic group with Tricalysia subgen. Empogona (BS 99%, PP 1.00). Second, both the MP strict consensus tree and the Bayesian majority rule consensus tree indicated sister relationships between Tricalysia subgen. Tricalysia and Belonophora, and recovered a clade containing Sericanthe, Discospermum, Diplospora, and Tricalysia subgen. Empogona. The clade of Belonophora and subgenus Tricalysia received poor internal support (BS , 50%, PP 0.82), but there was support for the second clade in the Bayesian analysis (BS , 50%, PP 0.98). DISCUSSION Previous taxonomic work on Tricalysia has focused on the use of traditional morphological and anatomical characters to infer relationships within the genus. In the most recent classification of the genus, Robbrecht (1979, 1982, 1983, 1987) subdivided it into two subgenera and seven sections. Here, for the first time, we have addressed relationships in this group using molecular data. In the current investigation, we obtained sequence data from four plastid regions for both subgenera and all seven sections of Tricalysia and generated estimates of phylogeny using both MP and Bayesian inference methods. The consensus tree topologies of both analyses (strict consensus for MP, majority rule consensus for Bayesian) were consistent. As is often observed, Bayesian PP were higher than bootstrap support values for any given node (Huelsenbeck et al., 2002; Erixon et al., 2003; Randle et al., 2005). TESTING THE MONOPHYLY OF THE GENUS TRICALYSIA Our phylogenetic analyses indicate that Tricalysia, as currently circumscribed, is not monophyletic. The monophyly of subgenera Tricalysia and Empogona is confirmed, but they are not sister to each other. This represents a new, though perhaps unsurprising, observation, which has implications for the taxonomy of the group (see below). Davis et al. (2007) included five species of Tricalysia in their molecular and morphological reassessment of the circumscription and phylogeny of Coffeeae. All five species were representatives of subgenus Tricalysia. In both their combined molecular and combined morphological-molecular phylogenies, Tricalysia (subgen. Tricalysia) was placed in a poorly supported and unresolved clade containing Sericanthe, Belonophora, and an Asian clade (including Diplospora and Discospermum). The study of Davis et al. (2007) incorporated molecular data from three plastid regions (trnL-F, accD-psa1, and rpl16). In our investigation, we included an additional plastid region, the group II intron petD. The extra characters provided by this fourth plastid marker were still not sufficient to fully elucidate systematic relationships within the clade containing Tricalysia subgen. Tricalysia, Sericanthe, Belonophora, Diplospora, and Discospermum. The inclusion of taxa from Tricalysia subgen. Empogona led to results conflicting with the study TAXONOMIC IMPLICATIONS FOR GENERIC CONCEPTS The revelation that Tricalysia sensu Robbrecht is not monophyletic calls for a reconsideration of the taxonomic delimitation of Tricalysia and closely related taxa. One taxonomic option would be to widen the genus Tricalysia to include Belonophora, Diplospora, Discospermum, and Sericanthe. However, these genera are easily identified (e.g., by the use of a key) and are so diverse morphologically and anatomically that consolidating them into one genus does not seem justified (Table 4). A more logical option would be to separate these taxa into groups at the generic level, based on morphological and molecular synapomorphies. Robbrecht (1979) enumerated four potential field characters that distinguish the subgenera Empogona and Tricalysia. Taxa of subgenus Empogona are identified by the presence of distinctly lobed calyces (vs. short and truncate in subgenus Tricalysia), densely pubescent corolla throats (vs. glabrous to sparsely hairy), the presence of a large flattened sterile appendage protruding from the anther connective (vs. blunt anthers, occasionally forming a short triangular appendage), and fruits that turn black at maturity (vs. red fruits). Robbrecht (1979) considered recognizing Empogona at the generic rank, but opted to incorporate it as a subgenus of Tricalysia, given the similarity in a number of other key characters (placentation, pollen morphology, fruit and seed morphology, and seed coat anatomy). This decision was also pragmatic in terms of taxonomic stability, as it required the fewest nomenclatural changes (Robbrecht, 1979). The revision of Sericanthe (Robbrecht, 1978) and the survey of the Asian relatives of Tricalysia (Ali & Robbrecht, 1991) provided ample morphological and anatomical evidence to justify the exclusion of these genera from Tricalysia. The genus Sericanthe is distinguished from Tricalysia by the presence of bacterial leaf galls (rare in Rubiaceae), wing-shaped Volume 96, Number 1 2009 Tosh et al. Phylogeny of Tricalysia colleters, and pollen with a verrucate sexine (in contrast to the reticulate sexine occurring in all other members of Coffeeae). Davis et al. (2007) also presented the following synapomorphic characters for the genus: 7- to 9-merous flowers, distinctly basifixed anthers, and horizontal micropyle orientation. Diplospora and Discospermum consistently have tetramerous flowers, which occur only rarely in African Tricalysia, and the flowers of Asian taxa are smaller than their African counterparts (Ali & Robbrecht, 1991). In addition, there is a strong tendency toward unisexual flowers in Asian taxa, a trait that is absent in all but a few representatives of Tricalysia confined to Madagascar (Ranarivelo-Randriamboavonjy et al., 2007). Ali and Robbrecht (1991) also justified maintaining Diplospora and Discospermum as separate genera on the basis of rather divergent fruit types (small, fleshy, and red fruits in Diplospora and large, leathery, and purplish black fruits in Discospermum). The decision to maintain Diplospora and Discospermum as separate genera is also supported by our molecular analyses. The tribal position of Belonophora has been fairly unstable since its initial description by Hooker (1873), partly due to the erroneous observation by Hooker that Belonophora possesses a solitary, pendulous ovule in each of the two locules. Keay (1958) observed that Belonophora species actually possess two collateral ovules per locule, on the inner surface of a pendulous placenta, but he felt it premature to assign the genus to a new tribe until a more satisfactory tribal classification within Rubiaceae had been proposed. Robbrecht and Puff (1986) tentatively placed Belonophora in the tribe Aulacocalyceae, although the axillary inflorescences of Belonophora contrasted with the terminal or subterminal inflorescences possessed by other members of the tribe. The placement of Belonophora in the tribe Coffeeae was first proposed by Bridson and Verdcourt (2003) and later supported by the study of Davis et al. (2007). The imbricate calyx lobes of Belonophora were synapomorphic for the genus in the study of Davis et al. (2007), and the genus is also distinguished from other members of Coffeeae by the presence of a superior embryo radicle (Cheek & Dawson, 2000). In light of evidence from our own molecular investigation, and in combination with morphological and anatomical observations reported elsewhere, we believe it is appropriate and fully justified to recognize Empogona (sensu Robbrecht, 1979) at generic rank. The necessary taxonomic changes for the inclusion of many former Tricalysia species in the genus Empogona are provided at the end of the Discussion. RECOGNITION OF INTRAGENERIC GROUPS IN TRICALYSIA 203 In addition to testing the monophyly of Tricalysia sensu Robbrecht, we were able to assess the levels of support for his sectional groups within the genus. All seven sections were sampled in our analysis, although some were better represented. Low levels of genetic variation between species limited the amount of resolution between taxa, but there are some provisional findings from this study. Tricalysia subgen. Tricalysia was subdivided into five sections by Robbrecht (1982, 1983, 1987). Tricalysia sect. Probletostemon, here represented in our molecular study by T. elliottii and T. anomala (Table 1), was thought to possess many morphological and anatomical features regarded as primitive for the group. These included free bracteoles, standard colleters (Robbrecht, 1988), large pleiomerous flowers with many ovules per placenta, and large fruits (Robbrecht, 1983). Our study confirms the monophyly of section Probletostemon (BS 95%, PP 1.00), but it remains unresolved in a basal polytomy. Tricalysia sect. Ephedranthera, here represented by three species, is characterized by the presence of anthers that are sessile in the corolla throat and partly included within the corolla tube (Robbrecht, 1982). The monophyly of this section is not supported in our investigation. Tricalysia aciculiflora Robbr. falls within the basal polytomy, whereas T. acocantheroides and T. bridsoniana are situated within the moderately to well-supported clade (BS 75%, PP 1.00) containing all the remaining taxa of subgenus Tricalysia. The other three sections (Tricalysia, Rosea, and Androgyne) are very similar morphologically. Most species in subgenus Tricalysia belong to section Tricalysia, which Robbrecht (1987) further subdivided into four informal groups. Only two of these informal groups are included in this investigation. The core group of taxa within section Tricalysia, here represented by T. coriacea (Benth.) Hiern and the weakly supported clade of T. pallens Hiern, T. okelensis Hiern, and T. bagshawei S. Moore, is unresolved in the basal polytomy. The group of T. angolensis A. Rich. ex DC., represented by T. griseiflora K. Schum., falls within the clade containing T. bridsoniana, T. microphylla, and representatives from sections Rosea and Androgyne. In section Rosea, species differ conspicuously from those in section Tricalysia due to the presence of a spathaceous calyx (Robbrecht, 1987). In section Androgyne, which comprises the Madagascan representatives of subgenus Tricalysia, species are characterized by the presence of unisexual flowers. There is weak bootstrap and significant Bayesian support Discospermum (ca. 7 spp.) Bracts and bracteoles Corolla length (mm) Flower organization Merosity Calyx free or fused into calyculi 8–15 free or fused into calyculi 5–10 Corolla throat Anthers glabrous to bearded medifixed; on short filaments in throat, exserted glabrous or hairy medifixed; on short filaments in throat, exserted Anther connective sometimes protruding in very short triangle sometimes protruding in very short triangle Placentation 1–3(–6) ovules on a (3–)5–15 ovules on a hemihemi-circular to circular to 6 hemi-ellipsoid 6 hemi-ellipsoid placenta; attached to the placenta; attached middle of the septum to the middle of the septum 20–30 5–7 Empogona (29 spp.) Tricalysia, excluding sect. Androgyne (ca. 80 spp.) free alternate, sect. fused into calyculi; Empogona; fused into free alternate in calyculi, sect. sect. Kraussiopsis Probletostemon (6–)8–17 8–50 hermaphroditic hermaphroditica Tricalysia sect. Androgyne (11 spp.) fused into calyculi 5–10 unisexual ((4–))5(–6) (4–)5–6(–12) 4–7 tube short, lobes well- tube well-developed; tube welldeveloped and lobes none, developed, often overlapping triangular or with minute linear teeth glabrous to hairy hairy densely beardedb medifixed; on long medifixed; on long medifixed; on filaments in throat, filaments in short exserted throat, exserted filaments in throat or sessile, exserted protruding in mostly protruding in short apical conspicuous short triangle appendage ribbon-like appendagesc 1–ca. 25 ovules on a 1–12 ovules on a 2–8 ovules on a hemi-circular to 6 hemi-circular to hemi-circular hemi-ellipsoid 6 hemi-ellipsoid to 6 hemiplacenta; attached to placenta; attached ellipsoid the middle of the to the middle of placenta; septum the septum attached to the middle of the septum 8–10 5–20 5–9 Belonophora (5 spp.) Sericanthe (ca. 20 spp.) lower bracts fused into fused into calyculi calyculi; upper bracts free, opposite (10–)20–30(–40) (8–)12–25 hermaphroditic; heterostyly ((4–))5 tube short, lobes welldeveloped hermaphroditic glabrous medifixed; sessile in tube, included glabrous or hairy basifixed; on short filaments in throat, exserted protruding in short triangle strongly flattened; no appendage 2 collateral ovules on inner face of a pendulous hemicircular 6 hemiellipsoid placenta; attached to the apex of the septum (1–)2(–5) ovules on a hemi-circular 6 hemiellipsoid placenta; attached to the apex of the septum 10–30 10–15 (5–)7–8(–9) tube welldeveloped, with minute teeth Annals of the Missouri Botanical Garden 4 tube short, rounded lobes present or absent hermaphroditic or unisexual 4–((–5)) tube short, lobes mostly triangular Fruit (mm) hermaphroditic or unisexual Diplospora (ca. 10 spp.) 204 Table 4. Salient morphological characters of Tricalysia and close relatives. Characters in boldface represent unique features for Empogona. Figures in single parentheses 5 rarely; figures in double parentheses 5 very rarely. lateral superior inferior inferior away from septum Embryo radicle OTHER RELATIONSHIPS WITHIN COFFEEAE AND THE RELATIONSHIP TO BERTIERA Heterostyly in section Ephedranthera. b Glabrous in a few species, e.g., Empogona concolor. c Some species with an inconspicuous appendix, e.g., Empogona welwitschii. entire, ruminate in some spp. Endosperm entire or ruminate massive, mostly surrounding seeds Placental outgrowth 205 (BS 61%, PP 0.98) for a clade containing these two sections. Tricalysia schliebenii (section Rosea) is sister to a strongly supported clade (BS 97%, PP 1.00) containing members of section Androgyne and T. jasminiflora of section Rosea. Robbrecht (1979) recognized two sections within subgenus Empogona: section Empogona is characterized by free bracteoles and distinct non-overlapping calyx lobes; in contrast, the bracteoles in section Kraussiopsis are fused to form calyculi, and the calyx lobes either touch or overlap each other (with the exception of Tricalysia bequaertii De Wild., where the calyx lobes are not touching). Tricalysia ngalaensis Robbr., previously thought to be closely related to T. junodii (Schinz) Brenan (Robbrecht, 1979), is in an unresolved position (Figs. 1, 2). There is weak bootstrap but significant Bayesian support for the monophyly of section Kraussiopsis (BS 56%, PP 0.99), and the informal group of T. ruandensis is also well supported (BS 85%, PP 100). The remaining taxa of section Empogona are weakly supported (BS 60%, PP 0.96), although the clade of T. junodii, T. ovalifolia, and T. acidophylla is well supported (BS 98%, PP 1.00). a entire entire mostly none; massive in some spp. entire massive, surrounding seeds sclerified or leathery Pericarp mostly none none, with weak outgrowths in some spp. entire, ruminate in some entire spp. inferior inferior none fleshy fleshy fleshy fleshy; rarely sclerotic, sect. Probletostemon mostly none orange yellow red red, rarely orange first white, turning purple, then black fleshy turning from yellow and orange to red fleshy purplish black Fruit color Belonophora (5 spp.) Tricalysia sect. Androgyne (11 spp.) Empogona (29 spp.) Diplospora (ca. 10 spp.) Discospermum (ca. 7 spp.) Table 4. Continued. Tosh et al. Phylogeny of Tricalysia Tricalysia, excluding sect. Androgyne (ca. 80 spp.) Sericanthe (ca. 20 spp.) Volume 96, Number 1 2009 The sister relationship of Bertiera and Coffeeae is recovered with moderate bootstrap and significant Bayesian support (BS 79%, PP 100), although our outgroup sampling is not complete. In order to confirm this result, more extensive sampling of representative groups within subfamily Ixoroideae is needed. Robbrecht and Manen (2006) opted to place Bertiera in subtribe Bertierinae, sister to Coffeinae, as the characteristic features of Bertiera differ from those of Coffeeae. Davis et al. (2007) found only weak bootstrap support for the sister relationship between Bertiera and Coffeeae (BS , 40%) based on molecular data alone, and the sister relationship was not recovered following the addition of morphological characters in their combined molecular-morphological analysis. Based on the decision of Bridson and Verdcourt (2003), they opted to place Bertiera in the monogeneric tribe Bertiereae. Whether Bertiera is recognized at the tribal or subtribal level is still open to debate, but we agree with Davis et al. (2007: 321) that ‘‘Coffeeae, with the addition of new genera and the removal of Bertiera, is both scientifically coherent and practical.’’ In the three-region plastid analysis of Davis et al. (2007), Coffea and Psilanthus form a well-supported monophyletic clade supported by a bootstrap of 87%, and are placed sister to the rest of Coffeeae. This relationship is recovered in our four-region analysis, 206 Annals of the Missouri Botanical Garden with increased support values (BS 93%, PP 1.00). There was also strong support for the sister relationship between the well-supported Argocoffeopsis and Calycosiphonia clade and the remaining ingroup taxa in our Bayesian analysis (PP 0.98), but weak support for this relationship in the MP analysis (BS 50%). This relationship was also recovered in the strict consensus tree of Davis et al. (2007). analysis. They have a basal position in the clade corresponding to section Empogona. TAXONOMIC NOVELTIES RESULTING FROM THE GENERIC RESURRECTION OF EMPOGONA An outline of an emended infrageneric classification for Empogona is provided below. It contains a formal new combination for one of the two sections recognized. The outline is followed by a checklist of species, providing all necessary new combinations at the species level and below. OUTLINE OF AN EMENDED CLASSIFICATION FOR EMPOGONA Empogona Hook. f., Hooker’s Icon. Pl. 11: 72, t. 1091. 1871. TYPE: Empogona kirkii Hook. f. Tricalysia subgen. Empogona (Hook. f.) Robbr., Bull. Jard. Bot. Natl. Belg. 49: 259. 1979. The further synonymy of subgenus Empogona (Robbrecht, 1979: 259) remains applicable to the genus Empogona. Empogona Hook. f. sect. Empogona. Tricalysia subgen. Empogona (Hook. f.) Robbr. sect. Empogona (Hook. f.) Brenan. Empogona sect. Kraussiopsis (Robbr.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia subgen. Empogona sect. Kraussiopsis Robbr., Bull. Jard. Bot. Natl. Belg. 49: 309. 1979. TYPE: Empogona crepiniana (De Wild. & T. Durand) J. Tosh & Robbr. EMPOGONA RUANDENSIS SPECIES GROUP This corresponds to the group of Tricalysia ruandensis (Robbrecht, 1979: 310). The group comprises the species numbered 8, 9, 19, 26, and 27 in the checklist below. EMPOGONA GLABRA SPECIES GROUP This corresponds to the group of Tricalysia glabra (Robbrecht, 1979: 292). This small group comprises only two species, numbers 16 and 23 in the checklist below. EMPOGONA CREPINIANA SPECIES GROUP This corresponds to the group of Tricalysia crepiniana (Robbrecht, 1979: 329). It is the most speciose group comprising 11 species, numbered 2, 3, 5, 7, 12, 13, 15, 20, 21, 28, and 29 of the checklist below. CHECKLIST OF SPECIES AND INFRASPECIFIC TAXA, INCLUDING TAXONOMIC NOVELTIES EMPOGONA KIRKII SPECIES GROUP This corresponds to the group of Tricalysia junodii (Robbrecht, 1979: 269). The group comprises the species numbered 11, 18, 22, and 24 in the checklist below. The position of Empogona ngalaensis (species 22 below) was not confirmed by our molecular analysis. EMPOGONA DISCOLOR SPECIES GROUP This corresponds to the group of Tricalysia discolor (Robbrecht, 1979: 292). The group comprises the species numbered 1, 4, 6, and 14 in the checklist below. The group is only represented by Empogona acidophylla in the analysis, which falls in a clade corresponding to the previous species group. Section Empogona further comprises the three species numbered 10, 17, and 25 in the checklist below. They were considered to be of isolated position (Robbrecht, 1979: 300). Two of these species (10. E. concolor and 17. E. gossweileri) are included in the The list below, ordered alphabetically, enumerates all known taxa of Empogona, including the four species (species numbered 3, 7, 21, and 27 below) treated or described after Robbrecht’s (1979) revision. The infrageneric assignment of the species is given in the preceding section of the present paper. Taxa preceeded by an asterisk (*) were included in the molecular analysis (see Table 1). The checklist includes taxonomic novelties for all species, i.e., 34 new combinations and three modifications of infraspecific status. In his revision, Robbrecht (1979) used varietal status for all infraspecific taxa recognized. Here we reconsider the appropriateness of that treatment in applying du Rietz’s criteria (as cited in Stace, 1991) for distinguishing subspecies and varieties. Therefore, when infraspecific taxa are allopatric and differing in several features, we propose subspecific rather than varietal status. (*) 1. Empogona acidophylla (Robbr.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia acid- Volume 96, Number 1 2009 ophylla Robbr., Bull. Jard. Bot. Natl. Belg. 49: 292. 1979. TYPE: Tanzania. Eastern Usambaras, 2 mi. E of Sigi railway station, 27 July 1953, R. B. Drummond & J. H. Hemsley 3490 (holotype, K!; isotypes, B!, BR!, LISU!, S!). Tosh et al. Phylogeny of Tricalysia 207 8b. Empogona buxifolia subsp. australis (Robbr.) J. Tosh & Robbr., comb. et stat. nov. Basionym: Tricalysia buxifolia var. australis Robbr., Bull. Jard. Bot. Natl. Belg. 48: 465. 1978. TYPE: Angola. Tchivinguiro, 13 Dec. 1961, G. Barbosa 9650 (holotype, LISC!; isotypes, COI!, K!, LUAI!). 2. Empogona aequatoria (Robbr.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia aequatoria Robbr., Bull. Jard. Bot. Natl. Belg. 48: 465. 1978. TYPE: [Democratic Republic of the Congo.] Congo belge. Yangambi, 4 Dec. 1937, J. Louis 6887 (holotype, BR!; isotypes, B!, BR!, C!, COI!, EA!, HBG!, K!, MO!, P!, PRE!, UPS!, WAG!). (*) 9. Empogona cacondensis (Hiern) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia cacondensis Hiern, Cat. Afr. Pl. (Hiern) 1(2): 467. 1898. TYPE: Angola. Rd. from Quipaca to fortress near Ferão, Oct. 1859, F. Welwitsch 3112 (lectotype, designated by Robbrecht [1979: 320], LISU!; duplicates, BM!, COI!, K!). 3. Empogona africana (Sim) J. Tosh & Robbr., comb. nov. Basionym: Diplospora africana Sim, Forest Fl. Cape, 238. 1907. Tricalysia africana (Sim) Robbr., S. African J. Bot. 51: 331. 1985. TYPE: South Africa. E Pondoland, Egossa Forest, Aug. 1899, T. R. Sim 2386 (holotype, NU!). (*) 10. Empogona concolor (N. Hallé) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia concolor N. Hallé, Fl. Gabon 17: 283. 1970. TYPE: Gabon. Bélinga, mine de fer, 21 July 1966, N. Hallé & A. Le Thomas 119 (holotype, P!; isotypes, K!, P!). 4. Empogona aulacosperma (Robbr.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia aulacosperma Robbr., Bull. Jard. Bot. Natl. Belg. 49: 296. 1979. TYPE: [Democratic Republic of the Congo.] Congo belge. Musenge, 20 Dec. 1958, A. Léonard 2088 (holotype, BR!; isotypes, EA!, K!, MO!, WAG!). 11. Empogona coriacea (Sond.) J. Tosh & Robbr., comb. nov. Basionym: Kraussia coriacea Sond., Linnaea 23: 54. 1850. Tricalysia sonderiana Hiern, Fl. Trop. Afr. [Oliver et al.] 3: 119. 1877, replacement for Kraussia coriacea Sond., non Randia coriacea Benth., Niger Fl. [W. J. Hooker] 387. 1849 [5 Tricalysia coriacea (Benth.) Hiern]. TYPE: [South Africa. KwaZulu-Natal:] Natal: Durban, s.d., W. Gueinzius 100 (holotype, W!; isotypes, BM!, C!, K!, PRE!, S!). (*) 5. Empogona bequaertii (De Wild.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia bequaertii De Wild., Pl. Bequaert. 3: 157. 1925. TYPE: [Democratic Republic of the Congo.] Congo belge. [Kisangani] Stanleyville, Tshopo River, 25 Feb. 1915, J. Bequaert 6969 (holotype, BR!). 6. Empogona bracteata (Hiern) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia bracteata Hiern, Fl. Trop. Afr. [Oliver et al.] 3: 120. 1877. TYPE: [Guinea.] Senegambia. Karkandy, s.d., Heudelot 855 (holotype, K!). 7. Empogona breteleri (Robbr.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia breteleri Robbr., Bull. Jard. Bot. Natl. Belg. 51: 166. 1981. TYPE: Gabon. Moanda–Franceville Km 23, 12 Sep. 1970, F. J. Breteler 6431 (holotype, WAG!; isotypes, BR!, P!). 8. Empogona buxifolia (Hiern) J. Tosh & Robbr. 8a. Empogona buxifolia (Hiern) J. Tosh & Robbr. subsp. buxifolia, comb. nov. Basionym: Tricalysia buxifolia Hiern, Fl. Trop. Afr. [Oliver et al.] 3: 119. 1877. TYPE: Angola. Ambriz, Nov. 1872, J. Monteiro s.n. (holotype, K!; isotype, W!). 12. Empogona crepiniana (De Wild. & T. Durand) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia crepiniana De Wild. & T. Durand, Ann. Mus. Congo Belg., Bot. ser. 3, 1: 120. 1901. TYPE: [Democratic Republic of the Congo.] Wangata, 17 Feb. 1896, A. Dewèvre 740 (holotype, BR!; isotype, COI!). 13. Empogona deightonii (Brenan) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia deightonii Brenan, Kew Bull. 8: 112. 1953. TYPE: Sierra Leone. Jama, 10 Mar. 1948, F. C. Deighton 4723 (holotype, K!; isotype, P!). 14. Empogona discolor (Brenan) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia discolor Brenan, Kew Bull. 2: 72. 1947. TYPE: [Ghana.] Gold Coast. Mampong Scarp, Feb. 1933, C. Vigne 2748 (holotype, K!; isotype, MO!). 15. Empogona filiformistipulata (De Wild.) Bremek. 15a. Empogona filiformistipulata (De Wild.) Bremek. subsp. filiformistipulata, Bot. Jahrb. 71: 208 Annals of the Missouri Botanical Garden 201, 222. 1940. Basionym: Urophyllum filiformistipulatum De Wild., Pl. Bequaert. 3: 211. 1925. Tricalysia filiformi-stipulata (De Wild.) Brenan, Kew Bull. 8: 112. 1953. TYPE: [Democratic Republic of the Congo.] Congo belge. Kisangani, Tshopo River, 12 Jan. 1915, J. Bequaert 6580 (holotype, BR!; isotype, K not seen). 15b. Empogona filiformistipulata subsp. epipsila (Robbr.) J. Tosh & Robbr., comb. et stat. nov. Basionym: Tricalysia filiformistipulata (De Wild.) Brenan var. epipsila Robbr., Bull. Jard. Bot. Natl. Belg. 48: 465. 1978. TYPE: [Democratic Republic of the Congo.] Congo belge. Yangambi, Feb. 1933, J. Louis 14233 (holotype, BR!; isotypes, COI!, K!, MO!, P!, WAG!). 16. Empogona glabra (K. Schum.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia glabra K. Schum., Bot. Jahrb. Syst. 23: 445. 1896. TYPE: Angola. Catete, Nov. 1856, F. Welwitsch 3117 (holotype, LISU!; isotypes, BM!, C!, COI!, K!, P!). (*) 17. Empogona gossweileri (S. Moore) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia gossweileri S. Moore, J. Linn. Soc. Bot 37: 305. 1906. TYPE: Angola. Cuanza Norte, Cazengo, 1903, J. Gossweiler 688 (holotype, BM!; isotypes, K!, P!). 18. Empogona kirkii Hook. f. 18a. Empogona kirkii Hook. f. subsp. kirkii, Hooker’s Icon. Pl. 11: 72, t. 1091. 1871. Tricalysia junodii (Schinz) Brenan var. kirkii (Hook. f.) Robbr., Bull. Jard. Bot. Natl. Belg. 49: 271. 1979. TYPE: Malawi. Cape Maclear, Oct. 1861, J. Kirk s.n. (holotype, K!). Mus. 3: 122. 1912. TYPE: [South Africa. KwaZulu-Natal:] Natal: Durban, W. Gueinzius 68 (lectotype, designated by Robbrecht [1979: 313], W!; duplicates, P!, S!). 20. Empogona macrophylla (K. Schum.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia macrophylla K. Schum., Bot. Jahrb. Syst. 28: 66. 1899. TYPE: Cameroon. Bipinde, Zenker 1569 (lectotype, designated by Robbrecht [1979: 339], COI!; duplicates, BM!, BR!, COI!, E!, G!, GOET!, HBG!, L!, M!, MO!, P!, S!, W!, WAG!). 21. Empogona maputenis (Bridson & A. E. van Wyk) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia maputensis Bridson & A. E. van Wyk, Fl. Zambes. 5(3): 475. 2003. TYPE: Mozambique. Matutuı́ne, 8 Aug. 1957, L. A. G. Barbosa & F. L. de Lemos 7807 (holotype, LISC not seen). (*) 22. Empogona ngalaensis (Robbr.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia ngalaensis Robbr., Bull. Jard. Bot. Natl. Belg. 49: 277. 1979. TYPE: Malawi. North Ngala, 20 mi. N of Chilumba, 17 Dec. 1969, J. Pawek 3095 (holotype, K!). 23. Empogona nogueirae (Robbr.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia nogueirae Robbr., Bull. Jard. Bot. Natl. Belg. 48: 466. 1978. TYPE: Angola. Musenge, 14 Oct. 1966, J. B. Teixeira 10701 (holotype, LISC!; isotype, COI!). 24. Empogona ovalifolia (Hiern) J. Tosh & Robbr. Empogona allenii Stapf is the only species validly published in the genus Empogona not taken up as a result of the present study. It is a synonym of the present taxon (Robbrecht, 1979: 272). (*) 24a. Empogona ovalifolia (Hiern) J. Tosh & Robbr. var. ovalifolia, comb. nov. Basionym: Tricalysia ovalifolia Hiern, Fl. Trop. Afr. [Oliver et al.] 3: 119. 1877. TYPE: [Tanzania.] Zanzibar: s. loc., s.d. [acc. K Sep. 1868], J. Kirk s.n. (lectotype, designated by Robbrecht [1979: 339], K!). (*) 18b. Empogona kirkii subsp. junodii (Schinz) J. Tosh & Robbr., comb. et stat. nov. Basionym: Empogona junodii Schinz, Mém. Herb. Boiss. 10: 67. 1900. Tricalysia junodii (Schinz) Brenan, Kew Bull. 2: 60. 1947. TYPE: Mozambique. Baia de Laurenço Marques (Delagoa Bay), s.d., H. Junod 311 (holotype, Z!). 24b. Empogona ovalifolia var. glabrata (Oliv.) J. Tosh & Robbr., comb. nov. Basionym: Empogona kirkii Hook. f. var. glabrata Oliv., Trans. Linn. Soc., Bot., 2: 336. 1887. Tricalysia ovalifolia Hiern var. glabrata (Oliv.) Brenan, Kew Bull. 2: 58. 1947. TYPE: Kenya or Tanzania. 40–60 mi. from coast, [1884], H. H. Johnston s.n. [Kilimanjaro Exp.] (holotype, K!). (*) 19. Empogona lanceolata (Sond.) J. Tosh & Robbr., comb. nov. Basionym: Kraussia lanceolata Sond., Linnaea 23: 53. 1850. Tricalysia lanceolata (Sond.) Burtt Davy, Ann. Transvaal 24c. Empogona ovalifolia var. taylorii (S. Moore) J. Tosh & Robbr., comb. nov. Basionym: Empogona taylorii S. Moore, J. Bot. 63: 145. 1925. Tricalysia ovalifolia Hiern var. taylorii (S. Volume 96, Number 1 2009 Moore) Brenan, Kew Bull. 2: 59. 1947. TYPE: Kenya. Giriama, Oct. 1887, W. E. Taylor s.n. (holotype, BM!). 25. Empogona reflexa (Hutch.) J. Tosh & Robbr. 25a. Empogona reflexa (Hutch.) J. Tosh & Robbr. var. reflexa, comb. nov. Basionym: Tricalysia reflexa Hutch., Kew Bull. 1915: 44. 1915. TYPE: Sierra Leone. Kessewe, 17 Apr. 1913, C. E. Lane-Poole 131 (holotype, K!). 25b. Empogona reflexa var. ivorensis (Robbr.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia reflexa var. ivorensis Robbr., Bull. Jard. Bot. Natl. Belg. 48: 466. 1978. TYPE: Ivory Coast. W of Niapidou, 20 Jan. 1959, A. J. M. Leeuwenberg 2500 (holotype, WAG!; isotypes, BR!, K!). (*) 26. Empogona ruandensis (Bremek.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia ruandensis Bremek., Bull. Jard. Bot. État Bruxelles 26: 253. 1956. TYPE: [Rwanda.] Mayaga, Mutema, 19 May 1954, L. Liben 1416 (holotype, U!; isotypes, BR!, WAG!). 27. Empogona somaliensis (Robbr.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia somaliensis Robbr., Bull. Jard. Bot. Natl. Belg. 56: 149. 1986. TYPE: Somalia. 17 km W of Badade, 30 June 1983, J. B. Gillett, C. F. Hemming, R. M. Watson & H. Julin 25153 (holotype, K!). (*) 28. Empogona talbotii (Wernham) J. Tosh & Robbr., comb. nov. Basionym: Cremaspora talbotii Wernham, Cat. Pl. Oban 49. 1913. Tricalysia talbotii (Wernham) Keay, Bull. Jard. Bot. État Bruxelles 28: 291. 1958. TYPE: Nigeria. Southern Nigeria, Oban, 1911, P. A. Talbot 287 (holotype, BM!; isotype, K!). 29. Empogona welwitschii (K. Schum.) J. Tosh & Robbr., comb. nov. Basionym: Tricalysia welwitschii K. Schum., Bot. Jahrb. Syst. 23: 449. 1897. TYPE: Angola. Near Ponte do Felix Simões, Apr. 1855, F. Welwitsch 3106 (holotype, LISU!; duplicates, BM not seen, COI!, K!, P!). CONCLUSIONS AND FUTURE DIRECTIONS We have been able to demonstrate that the two subgenera comprising the large Afro-Malagasy genus Tricalysia do not form a monophyletic group and should be treated as separate genera. Empogona has been previously recognized at generic rank, and Tosh et al. Phylogeny of Tricalysia 209 subsequent authors have considered reviving its generic status. On the basis of our molecular evidence, it is now fully justified to revive Empogona at the generic rank. The Asian genus Diplospora is sister to Empogona, with both genera forming a strongly supported monophyletic group. As a consequence, the weakly supported Asian clade reported by Davis et al. (2007) is not recovered in this investigation. Further data are still required to fully elucidate the phylogenetic relationships between Belonophora, Diplospora, Discospermum, Empogona, Sericanthe, and Tricalysia. There is increased support for the placement of a Coffea and Psilanthus clade as sister to the rest of Coffeeae. Future work requires the inclusion of nuclear ribosomal and low copy nuclear DNA sequence data, as well as expanded taxon sampling, in an effort to improve resolution between terminal taxa within the genera Tricalysia and Empogona. It seems prudent to defer detailed discussion on the biogeography of Tricalysia and Empogona until we have a broader sampling and a more resolved phylogenetic hypothesis of both genera. Literature Cited Ali, S. J. & E. Robbrecht. 1991. Remarks on the tropical Asian and Australian taxa included in Diplospora or Tricalysia (Rubiaceae-Ixoroideae-Gardenieae). Blumea 35: 279–305. Andreasen, K. & B. Bremer. 2000. Combined phylogenetic analysis in the Rubiaceae-Ixorideae: Morphology, nuclear and chloroplast DNA data. Amer. J. Bot. 87: 1731–1748. Brenan, J. P. M. 1947. Empogona Hook. f. and its relation to Tricalysia DC. Kew Bull. 1947: 53–63. Bridson, D. M. & B. Verdcourt. 2003. Rubiaceae. Pp. 379– 720 in G. V. Pope (editor), Flora Zambesiaca, Vol. 5(3). Royal Botanic Gardens, Kew. Cheek, M. & S. Dawson. 2000. A synoptic revision of Belonophora (Rubiaceae). Kew Bull. 55: 63–80. Davis, A. P., M. Chester, O. Maurin & M. Fay. 2007. Searching for the relatives of Coffea (Rubiaceae, Ixoraoideae): The circumscription and phylogeny of Coffeeae based on plastid sequence data and morphology. Amer. J. Bot. 94: 313–329. Doyle, J. J. & J. L. Doyle. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem. Bull. 19: 11–15. Erixon, P., B. Svennblad, T. Britton & B. Oxelman. 2003. Reliability of Bayesian posterior probabilities and bootstrap frequencies in phylogenetics. Syst. Biol. 52: 665–673. Felsenstein, J. 1985. Phylogenies and the comparative method. Amer. Naturalist 125: 1–15. Hooker, J. D. 1873. Rubiaceae. Pp. 7–151 in G. Bentham & J. D. Hooker (editors), Genera Plantarum, Vol. 2(1). Lovell Reeve & Co., London. Huelsenbeck, J. P. & F. Ronquist. 2001. MRBAYES. Bayesian inference of phylogeny. Bioinformatics 17: 754–755. ———, B. Larget, R. E. Miller & F. Ronquist. 2002. Potential applications and pitfalls of Bayesian inference of phylogeny. Syst. Biol. 51: 673–688. 210 Annals of the Missouri Botanical Garden Jordan, W. C., M. W. Courtney & J. E. Neigel. 1996. Low levels of interspecific genetic variation at a rapidly evolving chloroplast DNA locus in North American duckweed (Lemnaceae). Amer. J. Bot. 83: 430–439. Keay, R. W. J. 1958. Notes on Rubiaceae for the Flora of West Tropical Africa, 2nd ed. Bull. Jard. Bot. État. 28: 297–298. Löhne, C. & T. Borsch. 2005. Molecular evolution and phylogenetic utility of the petD group II intron: A case study in basal angiosperms. Molec. Biol. Evol. 22: 317–332. Maddison, D. R. & W. P. Maddison. 2002. MacClade 4: Analysis of phylogeny and character evolution, version 4.01. Sinauer Associates, Sunderland, Massachusetts. Mendenhall, M. 1994. Phylogeny of Baptista and Thermopsis (Leguminosae) as Inferred from Chloroplast DNA and Nuclear Ribosomal DNA Sequences, Secondary Chemistry, and Morphology. Ph.D. Dissertation, University of Texas, Austin. Persson, C. 2000. Phylogeny of Gardenieae (Rubiaceae) based on chloroplast DNA sequences from the rps16 intron and trnL(UAA)-F(GAA) intergenic spacer. Nord. J. Bot. 20: 257–270. Posada, D. & K. A. Crandall. 1998. Modeltest: Testing the model of DNA substitution. Bioinformatics 14: 817–818. Ranarivelo-Randriamboavonjy, T., E. Robbrecht, E. Rabakonandrianina & P. De Block. 2007. Revision of the Malagasy species of the genus Tricalysia (Rubiaceae). Bot. J. Linn. Soc. 155: 83–126. Randle, C. P., M. E. Mort & D. J. Crawford. 2005. Bayesian inference of phylogenetics revisited: Developments and concerns. Taxon 54: 9–15. Robbrecht, E. 1978. Sericanthe, a new African genus of Rubiaceae (Coffeeae). Bull. Jard. Bot. Natl. Belg. 48: 3–78. ———. 1979. The African genus Tricalysia A. Rich. (Rubiaceae-Coffeeae). 1. A revision of the species of subgenus Empogona. Bull. Jard. Bot. Natl. Belg. 49: 239–360. ———. 1982. The African genus Tricalysia A. Rich. (Rubiaceae-Coffeeae). 2. Ephedranthera, a new section of subgenus Tricalysia. Bull. Jard. Bot. Natl. Belg. 52: 311–339. ———. 1983. The African genus Tricalysia A. Rich. (Rubiaceae). 3. Probletostemon revived as a section of subgenus Tricalysia. Bull. Jard. Bot. Natl. Belg. 53: 299–320. ———. 1987. The African genus Tricalysia A. Rich. (Rubiaceae). 4. A revision of the species of section Tricalysia and section Rosea. Bull. Jard. Bot. Natl. Belg. 57: 39–208. ———. 1988. Tropical woody Rubiaceae. Opera Bot. Belg. 1: 1–271. ——— & C. Puff. 1986. A survey of the Gardenieae and related tribes (Rubiaceae). Bot. Jahrb. Syst. 108: 63–137. ——— & J.-F. Manen. 2006. The major evolutionary lineages of the coffee family (Rubiaceae, angiosperms). Combined analysis (nDNA and cpDNA) to infer the position of Coptosapelta and Luculia, and supertree construction based on rbcL, rps16, trnL-trnF and aptBrbcL data. A new classification in two subfamilies, Cinchonoideae and Rubioideae. Syst. & Geogr. Pl. 76: 85–146. Ronquist, F., J. P. Huelsenbeck & P. van der Mark. 2005. MrBayes 3.1 manual. ,http://mrbayes.csit.fsu.edu/mb3. 1_manual.pdf., accessed 30 October 2008. Schumann, K. 1891. Rubiaceae. Pp. 1–156 in A. Engler & K. Prantl (editors), Die naturlichen Pflanzenfamilien, Vol. 4 (4). Wilhelm Engelmann Verlag, Leipzig. Shaw, J., E. B. Licky, J. T. Beck, S. B. Farmer, W. Liu, J. Miller, K. C. Siripun, C. T. Winder, E. E. Schilling & R. L. Small. 2005. The tortoise and the hare II: Relative utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis. Amer. J. Bot. 92: 142–166. Simmons, M. P. & H. Ochoterena. 2000. Gaps as characters in sequence-based phylogenetic analysis. Syst. Biol. 49: 369–381. Stace, C. A. 1991. Plant Taxonomy and Biosystematics, 2nd ed. Edward Arnold, London. Staden, R., K. Beal & J. Bonfield. 1998. The Staden Package. Pp. 115–130 in S. Miseners & S. Krawetz (editors), Computer Methods in Molecular Biology. Humana Press, New York. Swofford, D. L. 2003. PAUP* 4.0b10: Phylogenetic Analysis Using Parsimony (* and other methods). Sinnauer Associates, Sunderland, Massachusetts. Taberlet, P., L. Gielly, G. Pautou & J. Bovet. 1991. Universal primers for amplification of three non-coding regions of chloroplast DNA. Pl. Molec. Biol. 17: 1005–1109. accD-psa1 petD rpl16 trnL-F Argocoffeopsis eketensis (Wernham) Robbr. Argocoffeopsis rupestris (Hiern) Robbr. subsp. thonneri (Lebrun) Robbr. Argocoffeopsis scandens (K. Schum.) Lebrun Belonophora coriacea Hoyle Belonophora coriacea Hoyle Belonophora sp. indet. Bertiera bicarpellata (K. Schum.) N. Hallé Bertiera breviflora Hiern Bertiera iturensis K. Krause Bertiera sp. indet. Calycosiphonia macrochlamys (K. Schum.) Robbr. Calycosiphonia macrochlamys (K. Schum.) Robbr. Calycosiphonia spathicalyx (K. Schum.) Robbr. Canephora sp. indet. Coffea homollei J.-F. Leroy Coffea mangoroensis Portères Coffea moratii J.-F. Leroy ex A. P. Davis & Rakotonas. Didymosalpinx norae (Swynn.) Keay Diplospora dubia (Lindl.) Masam. Diplospora sp. indet. Diplospora sp. indet. Discospermum abnorme (Korth.) S. J. Ali & Robbr. Discospermum sp. indet. Doricera trilocularis (Balf. f.) Verdc. Gardenia thunbergia L. f. Hyperacanthus microphyllus (K. Schum.) Bridson Hyperacanthus perrieri (Drake) Rakotonas. & A. P. Davis Hyperacanthus sp. indet. Ixora guillotii Hochr. Psilanthus ebracteolatus Hiern Psilanthus mannii Hook. f. Psilanthus semsei Bridson Polysphaeria sp. indet. Sericanthe andogensis (Hiern) Robbr. Sericanthe andogensis (Hiern) Robbr. Davis 3031 (K), Cameroon Harris 8168 (K), Central African Republic Davis 3016 (K), Cameroon Maurin 5 (K), Cameroon Maurin 19 (K), Cameroon Tadjouteu 480 (K), Cameroon Davis 3051 (K), Cameroon Van Caekenberghe 41 (BR), Gabon* Van Caekenberghe 40 (BR), Gabon* Davis 3017 (K), Cameroon Davis 3044 (K), Cameroon Davis 3036 (K), Cameroon Davis 2925 (K), Tanzania Davis 2727 (K), Madagascar Davis 2305 (K), Madagascar Rakotonasolo 41 (K), Madagascar Davis 2326 (K), Madagascar Van Caekenberghe 62 (BR), Zimbabwe* Van Caekenberghe 49 (BR)a Bremer 15238 (K), Borneo (Brunei) Nangkat 15238 (K), Borneo (Brunei) Sidiyasa 2148 (K), Borneo (Kalimantan) Ismail 16846 (K), Borneo (Brunei) Friedmann 2939 (K), Mascarenes (Rodrigues) Davis et al. 1961-29703 (K), SE Africa Goyder 5024 (K), Madagascar Davis 2584 (K), Madagascar Davis 2586 (K), Madagascar Tosh et al. 408B (BR), Madagascar Billiet 53054 (BR), Ivory Coast* Van Caekenberghe 78 (BR), Ghana* Kisera 1473 (K), Tanzania Mvungi 15 (K), Tanzania Bidgood 3490 (K), Tanzania Dessein 1097 (BR), Zambia DQ180497 DQ180496 DQ180498 DQ180499 DQ180500 DQ180501 DQ180502 NA FM160622 DQ180504 DQ180507 DQ180506 DQ180509 DQ180510 DQ153402 DQ153503 DQ153502 FM160621 AM999388 DQ180511 AM999389 AM999380 AM999390 DQ180513 DQ180514 AM999387 FM160619 FM160620 FM160624 AM999392 FM160623 DQ153395 DQ180517 DQ180522 FM177157 AM999399 NA AM999400 AM999401 AM999402 AM999403 AM999396 AM999397 AM999398 NA NA AM999404 AM999405 NA NA AM999406 AM999407 AM999395 AM999408 NA AM999409 AM999410 AM999411 NA NA NA NA NA AM999394 AM999412 AM999413 AM999414 NA AM999416 AM999415 DQ180531 DQ180532 DQ180533 DQ180534 DQ180535 DQ180536 DQ180537 AM999524 AM999525 DQ180539 DQ180542 DQ180541 DQ180544 AM999523 DQ153651 DQ153752 DQ153751 AM999522 AM999526 DQ180546 AM999527 AM999528 AM999529 DQ180548 DQ180549 AM999520 AM999519 AM999521 AM999518 AM999530 AM999531 DQ153644 DQ180552 DQ180557 AM999532 DQ180566 DQ180567 DQ180568 DQ180569 DQ180570 DQ180571 DQ180572 AM999466 AM999467 DQ180574 DQ180576 DQ180575 DQ180578 DQ180579 DQ153769 DQ153870 DQ153869 AM999465 AM999468 DQ180580 AM999510 AM999469 AM999470 DQ180582 DQ180583 AM999464 AM999462 AM999463 AM999461 AM999471 AM999472 DQ153762 DQ180586 DQ180591 AM999473 211 Voucher Tosh et al. Phylogeny of Tricalysia Taxon Volume 96, Number 1 2009 Appendix 1. Taxon voucher and accession data. 212 Appendix 1. Continued. Taxon Carvalho 4169 (K), Gulf of Guinea Islands (Bioko) Valkenberg 3160 (WAG), Gabon Manktelow 91215 (K), Tanzania Luke 7071 (K), Tanzania Kindekat 122 (BR), Tanzania Dessein 1212 (BR), Zambia Brummit 320 (K), Malawi De Block 1313 (BR), Madagascar Tosh et al. 11 (BR), Madagascar De Block et al. 1874 (BR), Madagascar Davis 3045 (K), Cameroon Malaisse 2052 (K), Democratic Republic of the Congo Walters 942 (MO), Gabon De Block 389 (BR), Kenya Dessein 1031 (BR), Zambia Degreef 95 (BR), Gabon Dessein 1283 (BR), Zambia Dessein 1359 (BR), Zambia De Block 527 (BR), Madagascar Tosh et al. 322 (BR), Madagascar De Block 694 (BR), Madagascar Tosh et al. 349 (BR), Madagascar Rabevohitra 2115 (K), Madagascar Jongkind 1806 (K), Ghana Senterre 4041, Equatorial Guinea Dessein 1044 (BR), Zambia Dessein 305 (BR), Zambia Ayami 42 (K), Malawi Van Caekenberghe 79 (BR), Zimbabwe* Bagliss 1519 (K), South Africa Gautier 2442 (K), Madagascar Tosh et al. 398 (BR), Madagascar De Block 405 (BR), Kenya Bidgood 2966 (K), Tanzania Schmidt 2139 (K), Ghana accD-psa1 petD rpl16 trnL-F DQ180523 AM999391 AM999345 AM999344 AM999346 AM999347 AM999348 AM999349 AM999350 AM999351 DQ180526 AM999352 AM999353 AM999354 AM999355 AM999356 AM999358 AM999357 AM999359 AM999360 AM999361 AM999362 AM999363 AM999364 AM999365 AM999367 AM999366 AM999368 AM999369 AM999370 AM999371 AM999372 AM999373 AM999374 AM999375 NA AM999417 AM999419 AM999418 AM999420 AM999421 AM999422 AM999423 AM999424 AM999425 AM999426 AM999427 AM999428 AM999429 AM999430 AM999431 AM999433 AM999432 AM999434 AM999435 AM999436 AM999436 AM999438 AM999439 AM999440 AM999442 AM999441 AM999443 AM999444 AM999445 AM999446 AM999447 AM999448 AM999449 AM999450 NA AM999533 AM999535 AM999534 AM999536 AM999537 FM160581 FM160582 FM160583 FM160584 DQ180560 FM160585 FM160586 FM160587 FM160588 FM160589 FM160591 FM160590 FM160592 FM160593 FM160594 FM160595 FM160596 FM160597 FM160598 FM160600 FM160599 FM160601 FM160602 FM160603 FM160604 FM160605 FM160606 FM160607 FM160608 DQ180592 AM999511 AM999475 AM999474 AM999512 AM999476 AM999513 AM999477 AM999478 AM999514 DQ180595 AM999479 AM999480 AM999481 AM999482 AM999483 AM999485 AM999484 AM999486 AM999487 AM999488 AM999489 AM999490 AM999491 AM999492 AM999494 AM999493 AM999495 AM999496 AM999497 AM999498 AM999499 AM999500 AM999501 AM999505 Annals of the Missouri Botanical Garden Sericanthe jacfelicis (N. Hallé) Robbr. Sericanthe sp. indet. Tricalysia aciculiflora Robbr. Tricalysia aciculiflora Robbr. Tricalysia acidophylla Robbr. Tricalysia acocantheroides K. Schum. Tricalysia acocantheroides K. Schum. Tricalysia ambrensis Randriamb. & De Block Tricalysia analamazaotrensis Homolle ex Randriamb. & De Block Tricalysia analamazaotrensis Homolle ex Randriamb. & De Block Tricalysia anomala E. A. Bruce var. guineensis Robbr. Tricalysia bagshawei S. Moore Tricalysia bequaertii De Wild. Tricalysia bridsoniana Robbr. Tricalysia cacondensis Hiern Tricalysia concolor N. Hallé Tricalysia coriacea (Benth.) Hiern Tricalysia coriacea (Benth.) Hiern Tricalysia cryptocalyx Baker Tricalysia cryptocalyx Baker Tricalysia dauphinensis Randriamb. & De Block Tricalysia dauphinensis Randriamb. & De Block Tricalysia dauphinensis Randriamb. & De Block Tricalysia elliottii (K. Schum.) Hutch. & Dalziel Tricalysia gossweileri S. Moore Tricalysia griseiflora K. Schum. Tricalysia griseiflora K. Schum. Tricalysia jasminiflora (Klotzsch) Benth. & Hook. f. ex Hiern Tricalysia junodii (Schinz) Brenan Tricalysia lanceolata (Sond.) Burtt Davy Tricalysia leucocarpa (Baill.) Randriamb. & De Block Tricalysia leucocarpa (Baill.) Randriamb. & De Block Tricalysia microphylla Hiern Tricalysia ngalaensis Robbr. Tricalysia okelensis Hiern Voucher Volume 96, Number 1 2009 Appendix 1. Continued. Taxon ovalifolia Hiern ovalifolia Hiern ovalifolia Hiern pallens Hiern pallens Hiern pallens Hiern perrieri Homolle ex Randriamb. & De Block ruandensis Bremek. schliebenii Robbr. talbotii (Wernham) Keay De Block et al. 1072 (BR), Madagascar De Block et al. 1090 (BR), Madagascar Butly 309 (K), Tanzania Dessein 1266 (BR), Zambia Dessein 953 (BR), Zambia Adams 831 (K), Liberia De Block 766 (BR), Madagascar Kuchar 22323 (BR), Tanzania Bidgood 1913 (K), Tanzania Latilo 67674 (K), Nigeria accD-psa1 petD rpl16 trnL-F AM999378 AM999376 AM999377 AM999381 AM999382 AM999379 AM999383 AM999384 AM999385 AM999386 AM999453 AM999452 AM999451 AM999455 AM999456 AM999454 AM999457 AM999458 AM999459 AM999460 FM160611 FM160609 FM160610 FM160613 FM160614 FM160612 FM160615 FM160616 FM160617 FM160618 AM999504 AM999503 AM999502 AM999515 AM999516 AM999506 AM999507 AM999517 AM999508 AM999509 * Leaf material and vouchers collected from the living collections of National Botanic Garden of Belgium. Country of origin given in the table. Origin unknown. Living material given to National Botanic Garden of Belgium by Hong Kong Herbarium. a Tosh et al. Phylogeny of Tricalysia Tricalysia Tricalysia Tricalysia Tricalysia Tricalysia Tricalysia Tricalysia Tricalysia Tricalysia Tricalysia Voucher 213