Annals of Botany 112: 1723–1742, 2013 doi:10.1093/aob/mct222, available online at www.aob.oxfordjournals.org Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae): species diversification and distribution of key morphological traits inferred from dated molecular phylogenetic trees J. Tosh1,2,*, S. Dessein3, S. Buerki4, I. Groeninckx1, A. Mouly5,6, B. Bremer6, E. F. Smets1,7,8 and P. De Block3 1 Received: 9 April 2013 Returned for revision: 20 May 2013 Accepted: 6 August 2013 Published electronically: 18 October 2013 † Background and Aims Previous work on the pantropical genus Ixora has revealed an Afro-Madagascan clade, but as yet no study has focused in detail on the evolutionary history and morphological trends in this group. Here the evolutionary history of Afro-Madagascan Ixora spp. (a clade of approx. 80 taxa) is investigated and the phylogenetic trees compared with several key morphological traits in taxa occurring in Madagascar. † Methods Phylogenetic relationships of Afro-Madagascan Ixora are assessed using sequence data from four plastid regions ( petD, rps16, rpoB-trnC and trnL-trnF) and nuclear ribosomal external transcribed spacer (ETS) and internal transcribed spacer (ITS) regions. The phylogenetic distribution of key morphological characters is assessed. Bayesian inference (implemented in BEAST) is used to estimate the temporal origin of Ixora based on fossil evidence. † Key Results Two separate lineages of Madagascan taxa are recovered, one of which is nested in a group of East African taxa. Divergence in Ixora is estimated to have commenced during the mid Miocene, with extensive cladogenesis occurring in the Afro-Madagascan clade during the Pliocene onwards. † Conclusions Both lineages of Madagascan Ixora exhibit morphological innovations that are rare throughout the rest of the genus, including a trend towards pauciflorous inflorescences and a trend towards extreme corolla tube length, suggesting that the same ecological and selective pressures are acting upon taxa from both Madagascan lineages. Novel ecological opportunities resulting from climate-induced habitat fragmentation and corolla tube length diversification are likely to have facilitated species radiation on Madagascar. Key words: Rubiaceae, Ixora, Afro-Madagascan, molecular phylogenetics, molecular dating, biogeography, ETS, ITS, petD, rps16, rpoB-trnC, trnL-trnF. IN T RO DU C T IO N The pantropical genus Ixora is one of the largest genera in Rubiaceae, with approx. 530 species of shrubs and small trees that typically grow in humid rain forest (Davis et al., 2009). The centre of species diversity for the genus is in South-East Asia, in particular Borneo (Lorence et al., 2007). Although no modern monograph of Ixora exists, there have been a number of revisions focusing on specific geographical regions (e.g. De Block, 1998, revision of continental African Ixora spp.; De Block, 2013, revision of Madagascan Ixora spp.). Phylogenetic studies of Ixora have primarily focused on the tribal placement and circumscription of the genus (Andreasen and Bremer, 1996, 2000; Mouly, 2007; Mouly et al., 2009a). Most recently, Mouly et al. (2009b) identified some well-supported, geographically defined lineages, including an ‘Afro-Madagascan’ clade. There are approx. 80 Afro-Madagascan Ixora spp. distributed equally between continental Africa and Madagascar (De Block, 1998). In continental Africa, Ixora mainly occurs in the Guineo-Congolian Regional Centre of Endemism (RCE) (following White, 1983), but also in the Afromontane archipelagolike RCE, and extends into the Zambezian RCE, the Swahilian RCE and the Swahilian/Maputaland regional transition zone (RTZ) (De Block, 1998). In Madagascar, Rubiaceae are most numerous and species rich in the evergreen humid forests (Davis and Bridson, 2003). Ixora is no exception to this, occurring most frequently in the humid evergreen forest (littoral, lowland and montane) on the eastern coast of Madagascar, although Ixora spp. also occur in the semi-deciduous forest of Madagascar (De Block, 2003, 2013). Ixora is one of the most easily recognizable genera in Rubiaceae, in part due to the often striking inflorescences and tetramerous flowers (Fig. 1). Diagnostic features for the genus (adapted from De Block, 2007) include articulated petioles, narrowly tubular tetramerous flowers, bilobed stigmas, bilocular ovaries and fruits (or, rarely, with more than two locules), # The Author 2013. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For Permissions, please email: journals.permissions@oup.com Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Laboratory of Plant Systematics, KU Leuven, Kasteelpark Arenberg 31, PO Box 2437, BE-3001 Leuven, Belgium, 2Ashdown House School, Forest Row, East Sussex RH18 5JY, UK, 3National Botanic Garden of Belgium, Domein van Bouchout, BE-1860 Meise, Belgium, 4Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3AE, UK, 5Laboratoire Chrono-environnement, UMR CNRS 6249, Université de Franche-Comté, 16 Route de Gray, F-25030 Besançon cedex, France, 6Bergius Foundation, Royal Swedish Academy of Sciences and Botany Department, Stockholm University, SE-106 91, Stockholm, Sweden, 7National Herbarium of The Netherlands, Leiden University Branch, PO Box 9514, NL-2300 RA Leiden, The Netherlands and 8Netherlands Centre for Biodiversity Naturalis, PO Box 9517, NL-2300 RA Leiden, The Netherlands * For correspondence. E-mail: Tosh_J@ashdownhouse.com 1724 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) A B C D E F I L H K J M F I G . 1. Example of morphological variation in Ixora. (A) Inflorescence of Ixora regalis. (B) Inflorescence of Ixora elliotii. (C) Inflorescence of Ixora emirnensis. (D) Pendulous inflorescence of Ixora mangabensis. (E) Inflorescence of Ixora densithyrsa. (F) Inflorescence of Ixora siphonantha. (G) Front view of a flower of Ixora guillotii. (H) Flowering branch of Ixora rakotonasoloi. (I) Mature fruits of Ixora guillotii. (J) Fruits of Ixora quadrilocularis. (K) Flowering node of Ixora homolleae. (L) Articulate petioles of Ixora finlaysoniana. (M) Articulate petioles of Ixora homolleae. uniovulate locules and seeds with a large adaxial hilar cavity. In contrast, identification at the species level is difficult, with species distinguished on the basis of minor and often continuous characters, typically involving features of the inflorescence and flowers (De Block, 1998, 2003). This is particularly the case for the African representatives of the genus, which De Block (1998) described as ‘extremely homogeneous’ in their characters. On Madagascar, there are several morphological traits Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 G Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) M AT E R I A L S A N D M E T H O D S 25 mL, and contained 1 mL of each primer (100 ng mL21), 0.35 mL of Biotaq DNA polymerase, 2.5 mL of 10× NH4 reaction buffer, 1.5 mL of 50 mM MgCl2, 2.5 mL of 10 mM dNTPs, 1 mL of bovine serum albumin (BSA; 0.4 %) and 2 mL of total genomic DNA. Amplification of petD, rps16 and trnL-trnF used the following temperature profile: 94 8C for 3 min; 32 cycles of 94 8C for 1 min, 50 8C for 1 min, 72 8C for 1.5 min; final extension of 72 8C for 7 min. The amplification profile for rpoB-trnC was: 94 8C for 3 min; 32 cycles of 94 8C for 1 min, 53 8C for 1 min, 72 8C for 2 min; final extension of 72 8C for 7 min. PCR mixes for nuclear regions were the same as for plastid regions, except that 1 mL of dimethylsulfoxide (DMSO) was added per 25 mL. The ITS amplification profile was: 94 8C for 3 min; 32 cycles of 94 8C for 1 min, 52 8C for 1 min, 72 8C for 1 min; final extension of 72 8C for 7 min. The ETS amplification profile was: 97 8C for 1 min; 40 cycles of 97 8C for 10 s, 55 8C for 30 s, 72 8C for 30 s; final extension of 72 8C for 7 min. All amplification products were purified using Nucleospin purification columns and sent to Macrogen Inc. (Seoul, South Korea) for sequencing. Taxon sampling and DNA preparation Sequence alignment and phylogenetic analyses Extensive fieldwork was undertaken in eastern and northern Madagascar in order to collect herbarium, alcohol and DNA material of Madagascan Ixora spp. This material was used in the molecular and the morphological study. We included 67 Ixora accessions, representing approx. 50 species that occur throughout the global distribution of the genus (Table 1). In particular, our taxon sampling is focused on Madagascan and African species. Where possible, we included multiple accessions for each Madagascan species to test species monophyly. Thirty-eight Madagascan accessions were included, representing 24 species. We included 16 accessions (14 species) from continental Africa, and from west and east tropical Africa. The remaining 12 Ixora accessions are Asian (three species), Mascarene (two species), Neotropical (four accessions from three species) and Pacific Island (three species) taxa. Vangueria madagascariensis, representing the closely related tribe Vanguerieae (e.g. Kainulainen et al., 2013) was selected as an outgroup. Total genomic DNA was isolated from either silica gel collections, fresh leaf material from the living collections of the National Botanic Garden of Belgium (NBGB) or herbarium material (BR, MO, P) using a standard cetyltrimethyl ammonium bromide (CTAB) protocol (Doyle and Doyle, 1987). As reported elsewhere (e.g. Rajaseger et al., 1997), Ixora leaves may be highly coriaceous and can contain high levels of phenolic compounds that may affect the quality of isolated DNA. Therefore, we purified isolated DNA using Nucleospin purification columns (Macherey-Nagel), following the manufacturer’s instructions. Contiguous sequences were assembled and edited using the Staden software package (Staden et al., 1998). Sequences were manually aligned in MacClade v. 4.04 (Maddison and Maddison, 2002) without difficulty due to low levels of sequence variation. All variable nucleotide positions were verified against the original electropherograms. Gaps were treated as missing data; potentially informative indels were coded using the ‘simple indel coding’ method of Simmons and Ochoterena (2000). To minimize the time and computational effort required for phylogenetic analyses, we excluded duplicate accessions of a species if the sequences from each accession were identical across all six data sets. Congruence of the data sets was assessed using the partition homogeneity test implemented in PAUP* v. 4.0b10 (Swofford, 2003). All constant and uninformative characters were excluded. One thousand permutation cycles were run, each consisting of a heuristic maximum parsimony (MP) search of ten random sequence addition replicates with TBR (tree bisection and reconstruction) branch swapping, holding ten trees at each step and saving no more than five trees per replicate. Maximum parsimony analyses were performed with PAUP* v. 4.0b10. We conducted equal weighted parsimony heuristic tree searches on: (a) individual data sets; (b) a combined plastid data set; and (c) a combined plastid – nuclear DNA data set. Each analysis consisted of 1000 random sequence addition replicates, holding ten trees at each step, with TBR branch swapping and MulTrees in effect, DELTRAN optimization and saving no more than ten trees per replicate. Support for clades was evaluated with 1000 full-heuristic bootstrap pseudo-replicates (Felsenstein, 1985), using the same settings as outlined above. Bayesian analyses were implemented in MrBayes 3.1 (Huelsenbeck and Ronquist, 2001). The model of DNA substitution for each region was determined using Modeltest v. 3.06 (Posada and Crandall, 1998) under the Akaike information criterion (AIC; Supplementary Data Table S1). Four independent Bayesian analyses with four chains were run for each data set, Amplification and sequencing Primers for amplification of plastid and nuclear ribosomal DNA (nrDNA) regions are listed in Table 2. PCR and cycle sequencing was performed using a Perkin Elmer GeneAMPw 9700 thermocycler. Plastid PCR mixes were made up to Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 occurring in Ixora that are absent in the continental African taxa and rare in the genus as a whole. These include: (1) reduction of the number of flowers per inflorescence towards solitary flowers; (2) increase from two- to four-locular ovaries; and (3) increase towards large flowers (corolla tubes .15 cm long) and fruits (De Block, 2007, 2008, 2013). In the present study, we further investigate the phylogenetic relationships of Madagascan and continental African Ixora spp. using molecular sequence data from four plastid regions ( petD, rps16, rpoB-trnC and trnL-trnF) and nuclear ribosomal external transcribed spacer (ETS) and internal transcribed spacer (ITS) regions. The purpose of this study is to improve taxon sampling of both African and Madagascan species in order to: (1) test existing hypotheses concerning the evolutionary affinities within and between African and Madagascan species; (2) assess the distribution of key morphological innovations of the Madagascan species on our molecular phylogenetic trees; and (3) investigate the age of species diversification and dispersal using molecular dating techniques. 1725 1726 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) TA B L E 1. Taxon accession data (only first collector listed for voucher) Taxon I. ferrea (Jacq.) Benth. I. foliosa Hiern I. foliicalyx Guédès I. foliicalyx Guédès I. francii Schltr. I. guillotii Hoch. I. guillotii Hoch. I. guineensis Benth. I. hartiana De Block I. hiernii Scott-Elliot I. hippoperifera Bremek. I. hippoperifera Bremek. I. homolleae De Block & Govaerts† I. homolleae De Block & Govaerts† I. lagenifructa De Block* I. macilenta De Block I. mangabensis DC. I. mangabensis DC. I. mangabensis DC. I. mangabensis DC. I. masoalensis De Block* I. microphylla Drake I. minutiflora Hiern I. mocquerysii DC. I. moramangensis De Block* I. moramangensis De Block* I. narcissodora K.Schum. I. nematopoda K.Schum. I. nitens (Poir.) Mouly & B.Bremer I. perrieri De Block* I. perrieri De Block* I. platythyrsa Baker ETS ITS petD rps16 rpoB-trnC trnL-trnF Van Caekenberghe 82 (BR), Mozambique — — HG315109 HG315176 HG315244 HG315312 Prévost 4160 (P), French Guiana Gautier 5006 (BR), Madagascar De Block 976 (BR), Madagascar Tosh 30 (BR), Madagascar Tosh 245 (BR), Madagascar De Block 943 (BR), Madagascar De Block 857 (BR), Madagascar Tosh 256 (BR), Madagascar Dessein 1455 (BR), Cameroon Van Caekenberghe 42 (BR), Mauritius HG315378 HG315379 HG315380 HG315381 HG315382 HG315383 HG315384 HG315385 — HG315386 HG315441 HG315442 HG315443 HG315444 HG315445 HG315446 HG315447 HG315448 HG315449 HG315450 — HG315110 HG315111 HG315112 HG315113 HG315114 HG315115 HG315116 HG315117 HG315118 HG315177 HG315178 HG315179 HG315180 HG315181 HG315182 HG315183 HG315184 HG315185 HG315186 HG315245 HG315246 HG315247 HG315248 HG315249 HG315250 HG315251 HG315252 HG315253 HG315254 — HG315313 — HG315314 HG315315 HG315316 HG315317 HG315318 HG315319 HG315320 Bradley 1022 (MO), Gabon Walters 1437 (MO), Gabon Delprete s.n. (BR), Brazil Tosh 400 (BR), Madagascar Mouly 267 (P), New Caledonia Van Caekenberghe 316 (BR), China Mouly 236 (P), New Caledonia HG315387 — HG315388 HG315389 HG315390 — HG315391 HG315451 HG315452 HG315453 HG315454 HG315455 — HG315456 HG315119 HG315120 HG315121 HG315122 HG315123 HG315124 HG315125 HG315187 HG315188 HG315189 HG315190 HG315191 HG315192 HG315193 HG315255 HG315256 HG315257 HG315258 HG315259 HG315260 HG315261 HG315321 HG315322 HG315323 HG315324 HG315325 HG315326 HG315327 Groeninckx 80 (BR), Madagascar HG315392 HG315457 HG315126 HG315194 HG315262 HG315328 Mouly 659 (P), Comoro Islands De Block 987 (BR), Madagascar De Block 1773 (BR), Madagascar De Block 1977 (BR), Madagascar De Block 1786 (BR), Madagascar De Block 1788 (BR), Madagascar Merello 1716 (MO), Commonwealth of Dominica (Lesser Antilles) Taylor 11693 (MO), Puerto Rico Onana 566 (P), Cameroon Tosh 352 (BR), Madagascar De Block 696 (BR), Madagascar Mouly 241 (P), New Caledonia De Block 2091 (BR), Madagascar Tosh 408B (BR), Madagascar Gereau 5601 (MO), Ghana Bamps 4320 (BR), Angola Adam 23101 (P), Sierra Leone Dessein 1669 (BR), Cameroon Van Valkenburg 3083 (WAG), Gabon Tosh 107 (BR), Madagascar HG315393 HG315394 HG315395 HG315396 HG315397 HG315398 HG315399 HG315458 HG315459 HG315460 HG315461 HG315462 HG315463 HG315464 HG315127 HG315128 HG315129 HG315130 HG315131 HG315132 HG315133 HG315195 HG315196 HG315197 HG315198 HG315199 HG315200 HG315201 HG315263 HG315264 HG315265 HG315266 HG315267 HG315268 HG315269 HG315329 HG315330 HG315331 HG315332 HG315333 HG315334 HG315335 HG315400 HG315401 HG315402 HG315403 HG315404 HG315405 HG315406 HG315407 HG315408 HG315409 HG315410 HG315411 HG315412 HG315465 HG315466 HG315467 HG315468 HG315469 HG315470 HG315471 HG315472 HG315473 HG315474 HG315475 HG315476 HG315477 HG315134 HG315135 HG315136 HG315137 HG315138 HG315139 HG315140 HG315141 HG315142 HG315143 HG315144 HG315145 HG315146 HG315202 HG315203 HG315204 HG315205 HG315206 HG315207 HG315208 HG315209 HG315210 HG315211 HG315212 HG315213 HG315214 HG315270 HG315271 HG315272 HG315273 HG315274 HG315275 HG315276 HG315277 HG315278 HG315279 HG315280 HG315281 HG315282 HG315336 HG315337 HG315338 HG315339 HG315340 HG315341 HG315342 HG315343 HG315344 HG315345 HG315346 HG315347 HG315348 Tosh 207 (BR), Madagascar HG315413 HG315478 HG315147 HG315215 HG315283 HG315349 De Block 2036 (BR), Madagascar Dessein 1404 (BR), Cameroon Tosh 128 (BR), Madagascar Tosh 130 (BR), Madagascar De Block 2040 (BR), Madagascar De Block 2053 (BR), Madagascar Razafimandimbison 654 (BR), Madagascar De Block 985 (BR), Madagascar Dessein 1440 (BR), Cameroon Malcomber 2805 (MO), Madagascar Tosh 255 (BR), Madagascar De Block 837 (BR), Madagascar De Block 418 (BR), Kenya Dessein 1449 (BR), Cameroon Friedmann 2631 (P), Mauritius HG315414 HG315415 HG315416 HG315417 HG315418 HG315419 HG315420 HG315421 HG315422 HG315423 HG315424 HG315425 HG315426 HG315427 HG315428 HG315479 HG315480 HG315481 HG315482 HG315483 HG315484 HG315485 — HG315486 HG315487 — — HG315488 HG315489 HG315490 HG315148 HG315149 HG315150 HG315151 HG315152 HG315153 HG315154 HG315155 HG315156 HG315157 HG315158 HG315159 HG315160 HG315161 HG315162 HG315216 HG315217 HG315218 HG315219 HG315220 HG315221 HG315222 HG315223 HG315224 HG315225 HG315226 HG315227 HG315228 HG315229 HG315230 HG315284 HG315285 HG315286 HG315287 HG315288 HG315289 HG315290 HG315291 HG315292 HG315293 HG315294 HG315295 HG315296 HG315297 HG315298 HG315350 HG315351 HG315352 HG315353 HG315354 HG315355 HG315356 HG315357 HG315358 HG315359 HG315360 HG315361 HG315362 HG315363 HG315364 De Block 841 (BR), Madagascar Tosh 232 (BR), Madagascar De Block 773 (BR), Madagascar HG315429 HG315430 HG315431 HG315491 HG315492 HG315493 HG315163 HG315164 HG315165 HG315231 HG315232 HG315233 HG315299 HG315300 HG315301 HG315365 HG315366 HG315367 Continued Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Vangueria madagascariensis J.F.Gmel. Ixora aluminicola Steyerm. I. ambrensis De Block I. amplidentata De Block* I. ankazobensis De Block* I. ankazobensis De Block* I. ankazobensis De Block* I. ankazobensis De Block* I. ankazobensis De Block* I. batesii Wernham I. borboniae Mouly & B.Bremer I. brachypoda DC. I. brachypoda DC. I. brevifolia Benth. I. capuroniana De Block* I. cauliflora Montrouz. I. chinensis Lam. I. collina (Montrouz.) Beauvis. I. crassipes Boivin ex De Block I. cremixora Drake I. cremixora Drake I. densithyrsa De Block I. elliotii Drake ex De Block I. emirnensis Baker I. emirnensis Baker I. ferrea (Jacq.) Benth. Voucher/Herbarium/Country of origin Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) 1727 TA B L E 1. Continued Taxon I. praetermissa De Block I. quadrilocularis De Block* I. rakotonasoloi De Block I. regalis De Block* I. regalis De Block* I. scheffleri K.Schum. I. siphonantha Oliv. I. sp. ‘Brunei’ I. sp. ‘Malaysia’ I. tanzaniensis Bridson Voucher/Herbarium/Country of origin Dessein 1519 (BR), Cameroon Tosh 85 (BR), Madagascar Tosh 316 (BR), Madagascar De Block 835 (BR), Madagascar De Block 2083 (BR), Madagascar Luke 9162 (P), Tanzania Tosh 389 (BR), Madagascar Malcomber 2980 (MO), Brunei Billiet 7327 (BR), Malaysia Luke 9304 (P), Tanzania ETS ITS petD rps16 rpoB-trnC trnL-trnF HG315432 HG315433 HG315434 HG315435 HG315436 HG315437 HG315438 — HG315439 HG315440 HG315494 HG315495 HG315496 HG315497 — HG315498 HG315499 HG315500 HG315501 HG315502 HG315166 HG315167 HG315168 HG315169 HG315170 HG315171 HG315172 HG315173 HG315174 HG315175 HG315234 HG315235 HG315236 HG315237 HG315238 HG315239 HG315240 HG315241 HG315242 HG315243 HG315302 HG315303 HG315304 HG315305 HG315306 HG315307 HG315308 HG315309 HG315310 HG315311 HG315368 HG315369 HG315370 HG315371 HG315372 HG315373 HG315374 HG315375 HG315376 HG315377 *sp. nov. ined. nom. nov. ined. † Region petD rpoB-trnC rps16 trnL–trnF ETS ITS Primer Primer sequence (5’ –3’) PetB1365 PetD738 rpoB-F trnC-R rps16-F rps16-R trnL-c trnF-f 18S-ETS ETS-ERIT ITS 1 ITS 4 TTGACYCGTTTTTATAGTTTAC AATTTAGCYCTTAATACAGG CACCCRGATTYGAACTGGGG CKACAAAAYCCYTCRAATTG AAACGATGTGGTARAAAGCAAC AACATCWATTGCAASGATTCGATA CGAAATCGGTAGACGCTACG ATTTGAACTGGTGACACGAG GCAGGATCAACCAGGTGACA CTTGTATGGGTTGGTTGGA TCCGTAGGTGAACCTGCGG TCCTCCGCTTATTGATATGC starting from random trees, for 5 million generations, sampling trees every 1000 generations. TRACER v. 1.4 (Rambaut and Drummond, 2007) was used to assess if the search had reached stationarity and to check each parameter had an effective sample size (ESS) .100. The initial 1250 (25 %) trees were discarded as a conservative burn-in. Post-burn-in trees from the four independent analyses were pooled and summarized by a 50 % majority rule consensus tree using PAUP* v4.0b10 to obtain posterior probabilities (PPs). Morphology and optimization of morphological characters The material for the taxonomical and morphological study of Ixora in Madagascar consists of preserved samples and herbarium material of the following institutions: BM, BR, G, K, MO, P, S, TAN, TEF, UPS, W, WAG and Z (abbreviations of institutions follow Holmgren et al., 1990). In total, .1000 herbarium collections, each with several duplicates, were studied. Morphological terminology generally follows Robbrecht (1988). For the Madagascan species, morphological characters were scored on herbarium material available for this species. For the continental African species, morphological characters were taken from the revision of African Ixora (De Block, 1998). We chose morphological traits of interest (i.e. uniflorous inflorescences, four-locular ovaries, extreme corolla tube length) and characters of potential taxonomic significance (Table 3). We Reference Löhne and Borsch (2005) Shaw et al. (2005) Shaw et al. (2005) Taberlet et al. (1991) Negrón-Ortiz and Watson (2002) Baldwin and Markos (1998) White et al. (1990) assessed the distribution of these morphological characters by mapping unambiguous character state changes onto our combined Bayesian MJ consensus tree using MacClade v. 4.04 (Maddison and Maddison, 2002). Divergence time estimation A Bayesian approach was applied to infer the temporal framework of the evolution of Ixora. Due to the limited fossil record that could be unequivocally assigned to the most recent common ancestor of Ixora, an expanded family-level data set was constructed and divergence time estimation was inferred based on fossils (see the following section for more detail). This large-scale analysis allowed estimation of the temporal origin of Ixora together with a 95 % confidence interval, which was subsequently used as a prior to perform a second analysis focusing only on this genus. The family-level data set included representatives of all the major lineages in Rubiaceae (see Appendix 1) and was based on the group II plastid introns petD, rps16 and the trnL-trnF spacer (Groeninckx, 2009). In the large-scale analysis, a sub-set of taxa representative of the main clades in Ixora was included. These taxa were selected based on preliminary phylogenetic analyses. We opted for this approach rather than pooling all the data sets and taxa together mainly to avoid encountering problems related to missing data. In addition to the group II plastid introns, the Ixora data set Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 TA B L E 2. Amplification primers for plastid and nuclear regions 1728 TA B L E 3. Selected morphological characters Inflorescence Sessile/pedunculate Lax/compact African taxa I. batesii I. brachypoda I. foliosa I. guineensis I. hartiana I. hiernii I. hippoperifera I. macilenta I. minutiflora I. narcissodora I. nematopoda I. praetermissa I. scheffleri I. tanzaniensis Sessile Pedunculate Pedunculate Sessile Pedunculate Sub-sessile Pedunculate or sessile Sessile Sessile Sessile Pedunculate Sessile Pedunculate Sessile Lax Lax Compact Lax Lax Lax Compact Lax Compact Lax Lax Lax Compact Compact Madagascan taxa (Clade 2) I. amplidentata I. densithyrsa I. foliicalyx I. guillotii I. homolleae I. lagenifructa I. microphylla I. mocquerysii I. platythyrsa I. quadrilocularis I. regalis I. siphonantha Pedunculate Pedunculate Sub-sessile Pedunculate (Sub-)sessile Pedunculate (Sub-)sessile Pedunculate Pedunculate Pedunculate Pedunculate Pedunculate Lax Compact Compact Lax Uniflorous Lax Lax Compact Lax Lax Lax Lax Madagascan taxa (Clade 3) I. ambrensis I. ankazobensis I. capuroniana I. crassipes I. cremixora I. elliotii I. emirnensis I. mangabensis I. masoalensis I. moramangensis I. perrieri I. rakotonasoloi Pedunculate Sessile (+ pedunculate) Sessile Sub-sessile Sessile Sessile Pedunculate Pedunculate Sessile Pedunculate Pedunculate Sessile Lax Lax Lax or compact Lax Lax Lax Lax Lax Lax Lax Lax Uniflorous Flower Tube length (mm) Lobe length (mm) Number Corolla tube length (cm) 0.4– 0.5 0.4– 0.8 0.3– 0.5 0.3– 0.8 0.2– 0.5 0.2– 0.7 0.2– 0.4 0.2– 0.5 0.2– 0.4 0.8– 1.5 0.3– 0.6 0.2– 0.5 0.4– 0.6 0.3– 0.7 0.2– 0.4 0.5– 1.0 2.0– 4.0 0.5– 1.0 5.0– 10 5.0– 8.0 0.5– 2.0 0.5– 1.0 0.2– 0.4 4.0– 7.0 0.5– 1(– 1.5) 0.5– 1 0.4– 0.5 0.5– 0.75 0.3– 0.6 0.75–1.25 0.2– 0.75 0.2– 0.4 0.2– 0.3 0.2– 0.5 0.5– 2.0 0.25–0.5 0.3– 0.5 0.3– 0.5 0.4 –0.6 0.3 –1.0 0.4 –0.7 0.2 –0.6 0.3 –0.6 0.2 –0.5 0.2 –0.4 0.2 –0.6 0.1 –0.4 0.2 –0.6 0.8 –1.2 0.3 –0.8 0.6 –1.0 0.2 –0.4 0.5 –1.25 2.5 –4.0 1.5 –6.0 0.75– 2.5 5.0 –12 1.5 –2.0 0.5 –1.5 2.0 –6.0 0.9 –1.4 (3 –)5– 15 1–5( –9) 3.0 –11 0.2 –0.35 0.3 –0.75 0.1 –0.8 0 ,0.3 0.3 –0.5 0.3 –0.4 0.5 –1.0 0.5 –1.5 0.25– 0.8 0.2 –0.4 0.3 –0.5 30–70 50– 200 45–90 30–90 9 –30 30–90 30– 120 9 –20 9 –30 50– 100 20–50 30–70 30–90 20–50 0.5 –1.3 3.3 –11 0.9 –2.3 0.5 –2.3 1.2 –2.4 1.4 –3.3 1.6 –2.7 0.9 –2.2 1.0 –2.0 3.0 –7.5 0,4–0,9 1.1 –2.5 0.8 –2.2 2.2 –3.2 1.5 –3.3 18–23 3.5 –8.5 5.0 –8.0 2.6 –4.7 3.5 –6.5 2.2 –4.0 5.5 –11(–16) 1.4 –2.5 4.0 –8.0 1.3 –5.5 (10–)15–22 3.2 –3.9 4.0 –6.3 1.5 –3.5 17–22.5 4.5 –7.8 1.8 –3.1 1.0 –1.5 2.0 –3.2 2.2 –4.0 2.1 –3.5 4.5 –7.5 2.2 –2.8 9 –60 50– 120 45–90 50– 150 1 3( –15) (1– )3 –9 3 –45 50– 150 3(–9) 50– 120 12–80 25–50 7 –18 40–90 15–50 50– 120 30–90 9 –50 9 –30 8 –25 9 –30 30–90 1 Adapted from De Block (1998, 2013). Ovary Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Four-locular Four-locular Bilocular Bilocular Bilocular Four-locular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Bilocular Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) Taxon Calyx Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) Fossil calibration To estimate absolute ages for lineage divergences in the case of the family-level data set, we have used seven fossil calibration points (see below) in Rubiaceae to set age constraints for several nodes. As in Buerki et al. (2011), for each calibration point, the oldest fossil record was selected and the upper (younger) bound of the geological interval (Gradstein et al., 2004) in which the fossil was found was used to represent the age constraint. In the family-level Bayesian analysis, all the calibration points were modelled as follows: log-normal distribution, mean ¼ 0.5, s.d. ¼ 1, offset ¼ age fossil (see below). Before detailing the fossils used here, we would like to report the current knowledge on fossil records that have been tentatively assigned to Ixora. Palaeobotanical remains of Ixora are poorly known, and Graham (2009) reported the presence of Ixora pollen dating from the Miocene on the Marshall Islands (Micronesia). In his study, Graham (2009) stressed that this evidence could be challenged and that further studies have to be conducted to assign these pollen unequivocally to existing taxa. Therefore, we have not considered this record in our divergence time estimations. The following calibration points were selected: (a) the stem group of Rubiaceae was set with an offset of 54 million years (Ma), following the first occurrence of Rubiaceae fossils during the Eocene (Malcomber, 2002); (b) the stem of the Faramea clade was set with an offset of 40 Ma, based on distinctive diporate pollen from the late Eocene (Graham, 2009); (c) the Coprosma clade was constrained with an offset of 23.8 Ma based on fossil records of pollen from the Oligocene (Graham, 2009); (d ) the stem of the Galium and Rubia clade was assigned an offset of 5.3 Ma, following the first occurrence of fossil pollen during the late Miocene for this group (Graham, 2009); (e) the stem of the Chiococca clade was calibrated with an offset of 5.3 Ma based on leaf material from the late Miocene (Graham, 2009); ( f ) the stem of the Emmenopterys clade was set with an offset of 48 Ma, following the first record of Rubiaceae fruits in the mid Eocene (Graham, 2009); and (g) an offset of 14.5 Ma was assigned to the Gardenia clade, based on pollen data from the mid Miocene (Graham, 2009). Based on the posterior distribution of the dating uncertainty on the stem of Ixora, the most recent common ancestor of the ingroup in the second BEAST analysis was set with a normal distribution, a mean of 18 Ma and an s.d. of 2.5. R E S ULT S The 368 novel sequences generated in this study were combined with 31 sequences previously generated by Mouly (2007), resulting in a total combined data set of 399 sequences, representing approx. 50 Ixora spp. Levels of genetic variation between species were generally low across all six regions investigated (Table 4). The total number of potentially parsimony informative characters ranged from 21 in rps16 and trnL-trnF to 58 in the ITS. In terms of percentage variability, the ETS proved to have the highest proportion of potentially parsimony informative characters (13.4 %). The lowest percentage variability was observed in petD (2.3 %). Our taxon sampling included multiple accessions for 13 Ixora spp. We observed no intraspecific sequence variation in nine of these species. In contrast, varying amounts of intraspecific sequence divergence were observed in the other five species; multiple accessions from these five species were included in all subsequent phylogenetic analyses. Phylogenetic analyses Plastid data sets. Phylogenetic analyses of individual plastid data sets ( petD, rpoB-trnC, rps16, trnL-trnF) generated largely unresolved and poorly supported phylogenetic trees (data not shown). The partition homogeneity test did not reveal any significant incongruence between plastid data sets (P ¼ 0.69) and there was no supported incongruence (Bayesian PP .0.95), so these four regions were combined for all subsequent analyses. The characteristics of the individual and combined plastid data sets are listed in Table 4. In both the MP and Bayesian analyses (Fig. 2), the Asian and Pacific Ocean taxa are sister to the rest of the genus [bootstrap (BS) 70, PP 1.00]. The Mascarene Ixora clade is in turn sister to the Neotropical and Afro-Madagascan Ixora spp. (BS 77, PP 1.00). The sister relationship between the three Neotropical Central – South American taxa and the Afro-Madagascan taxa is weakly to moderately supported in the MP (BS 56) and Bayesian (PP 0.96) analyses (Fig. 2). In the Afro-Madagascan group, we identified a clade containing African and Madagascan taxa (BS 72, PP 1.00; Clade 1) and an exclusively Madagascan clade (BS 59, PP 0.99; Clade 2). There is a lack of resolution and weak node support in Clades 1 and 2 due to a paucity of potentially parsimony informative characters (Table 4). In Clade 1 there is Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 included one additional plastid (rpoB-trnC) and two nuclear (ETS and ITS) regions. These regions proved useful to resolve phylogenetic relationships in Ixora further. With the exception of the number of runs and length of the Monte Carlo Markov chain (MCMC), the settings of the Bayesian analyses were identical for the two data sets. A partitioned Bayesian MCMC analysis was conducted in BEAST v. 1.7 with a relaxed log-normal molecular clock and a Yule speciation model. The partitions were unlinked for the model of evolution, but linked for the estimation of the molecular clock and the tree topology. The other parameters were set as default (see Drummond and Rambaut, 2007). The best-fit models for each DNA region were kept identical to those in the MrBayes analyses (see above). In the case of the family-level data set, four runs of 10 million generations were performed, sampling a tree every 1000 generations. To improve the tree topology research, the MrBayes consensus tree was provided as the starting tree in the BEAST family-level analysis. In the case of the Ixora data set (which was rooted using Vangueria infausta), two runs of 5 million generations were performed, sampling a tree every 1000 generations, and the analyses were seeded with a random tree. For each parameter, convergence of runs was confirmed by the examination of their respective posterior distributions in TRACER v. 1.4 (Rambaut and Drummond, 2004). In addition, we considered the MCMC sampling sufficient when the ESS was .200 using TRACER v. 1.4 (Rambaut and Drummond, 2004). A maximum clade credibility tree with median branch lengths and 95 % confidence interval on nodes was built using TreeAnnotator v. 1.5.4 (Drummond and Rambaut, 2007) based on the set of trees after burn-in (for each run, a burn-in period of 1 million was applied). 1729 1730 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) TA B L E 4. Characteristics of individual and combined data sets and tree statistics petD Number of taxa Total length (bp) Non-parsimony informative characters Potentially parsimony informative characters (% of total) Indels Tree length Consistency index Retention index Number of trees saved 56 1023 49 24 (2.3 %) 3 81 0.914 0.955 100 trnL-trnF 57 828 41 21 (2.5 %) 2 68 0.956 0.969 68 rps16 57 722 46 rpoB-trnC Plastid ETS 55 1016 75 57 3589 211 21 (2.9 %) 33 (3.2 %) 99 (2.7 %) 55 (13.4 %) 58 (9.2 %) 1 76 0.921 0.917 511 3 140 0.850 0.811 4887 9 376 0.872 0.889 8722 0 159 0.780 0.898 1151 0 196 0.791 0.819 9580 nrDNA data sets. We were unable to obtain ETS and ITS sequences for Vangueria madagascariensis, due to amplification difficulties and ambiguous sequence reads, respectively. As our primary interests concern the Afro-Madagascan element of Ixora, phylogenetic analyses from separate and combined nuclear DNA analyses were rooted using Asian and/or Pacific Ocean Ixora spp. Relationships in the AfroMadagascan Clade were not affected by the root choice. The characteristics of the nuclear DNA data sets are listed in Table 4. The topology of the ETS phylogenetic tree (Supplementary Data Fig. S1) is similar to that of the combined plastid tree, but there are some differences. Clades 1 and 2 are not fully recovered in the ETS topology, with the deepest nodes in the Afro-Madagascan group being unresolved. Furthermore, there is weak to moderate support (BS 78, PP 0.93) for the sister relationship of the Guineo-Congolian I. nematopoda with respect to Neotropical Ixora and the remaining Afro-Madagascan Ixora (discussed by Mouly et al., 2009b). However, Clade 3 is strongly supported in the ETS topology (BS 94, PP 1.00). The ITS topology is poorly resolved and weakly supported, in particular in the MP strict consensus (Supplementary Data Fig. S2). Few clades are well supported in both MP and Bayesian analyses, and Clades 1 – 3 were not recovered. Total combined data set. Despite some topological inconsistencies between the combined plastid and nuclear DNA data sets, results from the simultaneous partition homogeneity test did not reveal significant incongruence between data sets (P ¼ 0.10). The combined plastid –nuclear DNA data set comprised 4629 characters. In total, 212 of the 545 variable characters are potentially parsimony informative (Table 4). The Mascarene clade is sister to the Neotropical and Afro-Madagascan species (BS 84, PP 1.00). The monophyly of the Afro-Madagascan group is supported (BS 65, PP 1.00), and in this group there is support for Clades 1–3 (Fig. 3). In the Bayesian analysis, I. nematopoda and I. scheffleri are sister to the rest of Clade 1 (PP 1.00). Although this relationship is recovered in the MP strict consensus, it is not supported by the bootstrap analysis. The phylogenetic position of I. nematopoda differs between the ETS and the plastid and ITS data sets (Fig. 2; Supplementary Data Figs S1 and S2). The monophyly of Clade 52 631 71 nrDNA 49 1040 122 108 (10.4 %) 0 357 0.759 0.839 3464 Total 57 4629 333 212 (4.6 %) 9 758 0.801 0.849 5362 1 is supported in both the Bayesian and MP analyses (BS 73, PP 1.00) following the exclusion of I. nematopoda (not shown). In Clade 1, the group of I. foliosa and I. hippoperifera is weakly supported (BS 57, PP 0.92), but two other groups of tropical West– Central African taxa are well supported: I. guineensis, I. minutifolia, I. hiernii and I. batesii (BS 92, PP 1.00) and I. hartiana, I. macilenta and I. praetermissa (BS 85, PP 1.00). Support for the sister relationships between these three West– Central African clades, and between the widespread I. brachypoda, is weak or lacking (Fig. 3). Branch lengths between these clades are short, which may be a contributing factor in the phylogenetic uncertainty between these West– Central African species (Supplementary Data Fig. S3). Nested in Clade 1 is the strongly supported Clade 3 (BS 100, PP 1.00), comprised of East African and Madagascan species (Fig. 3). The branch length subtending Clade 3 is relatively long, with eight synapomorphies (Supplementary Data Fig. S3). In Clade 3, the East African I. narcissodora and I. tanzaniensis are sister to the Madagascan taxa (BS 66, PP 0.94). Phylogenetic relationships between these Madagascan taxa are poorly resolved (Fig. 3; Supplementary Data Fig. S3). However, there are a few species relationships that are well supported, such as I. elliotii and I. rakotonasoloi (BS 99, PP 1.00) and I. crassipes and I. cremixora (BS 98, PP 1.00). The accessions of I. mangabensis did not group together in our analyses, due to the presence of two distinct plastid haplotypes in the four accessions of this species. The monophyly of Clade 2, comprised exclusively of Madagascan taxa, is supported in the combined plastid – nuclear DNA data set (BS 70, PP 1.00). There are two main subclades recovered in the MP strict consensus and Bayesian majority rule consensus trees, although support for these sub-clades is lacking (Fig. 3). Although the partition homogeneity test revealed no significant incongruence between individual data sets, the phylogenetic placement of certain taxa (e.g. I. densithyrsa and I. regalis 2) differed between plastid and nuclear data sets. In the plastid phylogenetic tree, I. regalis 2 is nested in the group containing I. regalis 1, I. guillotii and I. amplidentata (Fig. 2). In contrast, I. regalis 2 is unresolved in the ETS phylogenetic tree (Supplementary Data Fig. S1) and sister to I. homolleae and I. quadrilocularis in the ITS phylogenetic tree (Supplementary Data Fig. S2). Similarly, I. densithyrsa and I. mocquerysii demonstrate differing phylogenetic affinities in the plastid and nuclear data sets (Fig. 2; Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 strong support for a clade of Madagascan and tropical East African taxa (BS 85, PP 1.00; Clade 3). 52 409 51 ITS Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) 1·00/75 0·99/59 0·91/* 0·98/60 1·00/77 1·00/71 1·00/86 1·00/64 1.00/85 0·96/56 1·00/62 1·00/96 1·00/70 0·99/52 1·00/99 1·00/66 1·00/67 1·00/100 1·00/100 0·86/54 1·00/72 0·91/* 1·00/57 0·70/* 1·00/100 1·00/100 0·99/60 0·59/50 1·00/97 Madagascan lxora Madagascan and East African lxora West and Central African lxora Mascarene lxora Pacific lxora Asian lxora F I G . 2. Combined plastid Bayesian majority rule consensus tree. Bayesian posterior probabilities .0.5 and bootstrap values .50 % are indicated above branches (PP/ BS). Asterisks (*) denote nodes that have bootstrap support ,50 in the MP analysis. Clade 1, ‘Afro-Madagascan clade’; Clade 2, ‘Madagascan clade’; Clade 3, ‘East African–Madagascan clade’. Supplementary Data Figs S1 and S2). Exclusion of I. regalis 2 and I. densithyrsa (data not shown) from the Bayesian analyses results in increased support for the two sub-clades of Clade 2, with both sub-clades supported by a PP of 1.00. In the first subclade, I. foliicalyx is sister to I. homolleae and I. quadrilocularis (PP 1.00; not shown). In the second sub-clade, I. microphylla and I. platythyrsa are sister to the group of I. regalis 1, I. guillotii, I. amplidentata, I. lagenifructa, I. siphonantha and I. mocquerysii (PP 0.95; not shown). Distribution of key morphological characters The phylogenetic distribution of key morphological characters is illustrated in Fig. 4. Pedunculate and sessile inflorescences Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 1·00/85 Clade 2 0·62/* 1·00/78 Neotropical lxora Clade 3 1·00/56 V. madagascariensis I. aluminicola I. ferrea I. brevifolia I. amplidentata I. guillotti I. regalis 1 I. regalis 2 I. lagenifructa I. densithyrsa I. homolleae 1 I. homolleae 2 I. quadrilocularis I. mocquerysii I. foliicalyx I. microphylla I. platythyrsa I. siphonantha I. ankazobensis 1 I. perrieri I. ankazobensis 2 I. ankazobensis 3 I. capuroniana I. masoalensis I. emirnensis I. mangabensis 1 I. crassipes I. cremixora 1 I. cremixora 2 I. elliotii I. rakotonasoloi I. mangabensis 2 I. moramangensis I. ambrensis I. narcissodora I. tanzaniensis I. batesii I. guineensis I. hiernii I. minutiflora I. brachypoda 1 I. brachypoda 2 I. foliosa I. hippoperifera I. hartiana I. macilenta I. praetermissa I. nematopoda I. scheffleri I. nitens I. borboniae I. cauliflora I. collina I. francii I. chinensis I. sp. Malaysia I. sp. Brunei Clade 1 0·73/* 1731 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) 0·92/* 0·88/* 0·56/* 0·66/* 0·67/* 0·69/* 1·00/85 1·00/70 0·72/* 1·00/84 1·00/99 1·00/95 0·54/* 0·94/* 1·00/83 1·00/54 0·94/66 1·00/98 1·00/86 1·00/65 1·00/99 1·00/100 1·00/78 0·60/* 1·00/100 1·00/85 0·55/* 0·81/53 1·00/100 1·00/89 1·00/* 1·00/88 0·55/* 0·54/* 1·00/92 1·00/100 0·92/57 0·77/* 1·00/100 1·00/100 0·67/60 0·92/61 1·00/97 0·89/59 Neotropical lxora Madagascan lxora Madagascan and East African lxora West and Central African lxora Mascarene lxora Pacific lxora Asian lxora F I G . 3. Combined plastid– nuclear Bayesian majority rule consensus. Bayesian posterior probabilities .0.5 and bootstrap values .50 % are indicated above branches. Asterisks (*) denote nodes that have bootstrap support ,50 in the MP analysis. Clade 1, ‘Afro-Madagascan clade’; Clade 2, ‘Madagascan clade’; Clade 3, ‘East African– Madagascan clade’. occur in African taxa and both clades of Madagascan taxa (Fig. 4A). All taxa from Clade 2 and the African I. nematopoda and the Madagascan I. masoalensis (Clade3) possess calyx lobes .1 mm long (Fig. 4B). Calyx tubes .1 mm long are found in all but two species from Clade 2, in addition to the African I. narcissodora and the Madagascan I. masoalensis and I. crassipes (Fig. 4C). Species with uniflorous inflorescences or corolla tubes .15 cm long occur in both Madagascan clades, but species with four-locular ovaries are only found in Clade 2 (Fig. 4D– F). Divergence time estimation The estimated divergence times for nodes of interest are summarized in Fig. 5 and Supplementary Data Fig. S4, Tables S2 and S3. The onset of diversification of the genus started during the mid Miocene, with the emergence of most of the lineages at the Miocene – Pliocene boundary. Based on the three-gene Rubiaceae-wide data set (Supplementary Data Fig. S4, Table S2), the crown age for Ixoreae (node 152) is estimated at 16.67 million years old (9.67 –27.55, 95 % highest posterior density; hereafter HPD). The subsequent age estimates (discussed below) are based on the results of the secondary dating analysis (Fig. 5; Supplementary Data Table S3). The onset of divergence between the Asian-Pacific Ocean Ixora (node 59) and the rest of the genus is estimated at 15.37 Ma (7.39– 22.89, 95 % HPD). The estimated age of divergence between the Neotropical and Afro-Madagascan taxa (node 66) is 9.51 Ma (4.47 –14.94, 95 % HPD), with the crown age for the Afro-Madagascan group (node 67) estimated at 7.95 million years old (3.71– 12.52, 95 % HPD). The crown age of Clade 1 (node 68) is estimated at 7.22 million years old (3.36– 11.46, 95 % HPD). The crown age of Clade 2 (node 98) is estimated at 6.24 million years old (2.88– 10.03, 95 % HPD), with two separate periods of Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 1·00/98 0·99/* V. madagascariensis I. aluminicola I. brevifolia I. ferrea I. amplidentata I. lagenifructa I. guillotii I. regalis 1 I. densithyrsa I. mocquerysii I. siphonantha I. microphylla I. platythyrsa I. foliicalyx I. homolleae 1 I. homolleae 2 I. quadrilocularis I. regalis 2 I. ankazobensis 1 I. ankazobensis 2 I. ankazobensis 3 I. perrieri I. capuroniana I. masoalensis I. emirnensis I. managabensis 1 I. crassipes I. cremixora 1 I. cremixora 2 I. elliotii I. rakotonasoloi I. mangabensis 2 I. moramangensis I. ambrensis I. narcissodora I. tanzaniensis I. hartiana I. macilenta I. praetermissa I. batesii I. hiernii I. minutiflora I. guineensis I. brachypoda 1 I. brachypoda 2 I. foliosa I. hippoperifera I. nematopoda I. scheffleri I. nitens I. borboniae I. cauliflora I. collina I. francii I. chinensis I. sp. Malaysia I. sp. Brunei Clade 2 0·64/* Clade 3 1·00/84 Clade 1 1732 I. ampIidentata I. Iagenifructa I. guiIIotii I. regaIis 1 I. densithyrsa I. mocquerysii I. siphonantha I. microphyIIa I. pIatythyrsa I. foIiicaIyx I. homoIIeae 1 I. homoIIeae 2 I. quadriIocuIaris I. regaIis 2 I. ankazobensis 1 I. ankazobensis 2 I. ankazobensis 3 I. perrieri I. capuroniana I. masoaIensis I. emirnensis I. mangabensis 1 I. crassipes I. cremixora 1 I. cremixora 2 I. eIIiotii I. rakotonasoIoi I. mangabensis 2 I. moramangensis I. ambrensis I. narcissodora I. tanzaniensis I. hartiana I. maciIenta I. praetermissa I. batesii I. hiernii I. minutifIora I. guineensis I. brachypoda 1 I. brachypoda 2 I. foIiosa I. hippoperifera I. nematopoda I. scheffIeri B C I. ampIidentata I. Iagenifructa I. guiIIotii I. regaIis 1 I. densithyrsa I. mocquerysii I. siphonantha I. microphyIIa I. pIatythyrsa I. foIiicaIyx I. homoIIeae 1 I. homoIIeae 2 I. quadriIocuIaris I. regaIis 2 I. ankazobensis 1 I. ankazobensis 2 I. ankazobensis 3 I. perrieri I. capuroniana I. masoaIensis I. emirnensis I. mangabensis 1 I. crassipes I. cremixora 1 I. cremixora 2 I. eIIiotii I. rakotonasoIoi I. mangabensis 2 I. moramangensis I. ambrensis I. narcissodora I. tanzaniensis I. hartiana I. maciIenta I. praetermissa I. batesii I. hiernii I. minutifIora I. guineensis I. brachypoda 1 I. brachypoda 2 I. foIiosa I. hippoperifera I. nematopoda I. scheffIeri D E I. ampIidentata I. Iagenifructa I. guiIIotii I. regaIis 1 I. densithyrsa I. mocquerysii I. siphonantha I. microphyIIa I. pIatythyrsa I. foIiicaIyx I. homoIIeae 1 I. homoIIeae 2 I. quadriIocuIaris I. regaIis 2 I. ankazobensis 1 I. ankazobensis 2 I. ankazobensis 3 I. perrieri I. capuroniana I. masoaIensis I. emirnensis I. mangabensis 1 I. crassipes I. cremixora 1 I. cremixora 2 I. eIIiotii I. rakotonasoIoi I. mangabensis 2 I. moramangensis I. ambrensis I. narcissodora I. tanzaniensis I. hartiana I. maciIenta I. praetermissa I. batesii I. hiernii I. minutifIora I. guineensis I. brachypoda 1 I. brachypoda 2 I. foIiosa I. hippoperifera I. nematopoda I. scheffIeri F F I G . 4. Optimization of selected morphological character states on the Bayesian majority rule consensus tree. Grey boxes denote Madagascan taxa from Clade 2; white boxes denote Madagascan taxa from Clade 3; black boxes denote African taxa. (A) Inflorescence type: pedunculate inflorescence (grey); sessile inflorescence (black); pedunculate or sessile inflorescence (stippled grey); equivocal state (stippled black). (B) Calyx lobe length: ≥1 mm (black); ,1 mm (grey). (C). Calyx tube length: ≥1 mm (black); ,1 mm (grey). (D) Uniflorous taxa (black); multiflorous taxa (grey). (E) Four-locular ovaries (black); bilocular ovaries (grey). (F) Corolla tube length: ≥15 cm (black); ,15 cm (grey). Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) A 1733 Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) 0·95 0·94 61 1 62 1 0·99 59 113 1 1 Occurrence of the node (Ma) MRCA clade 2 111 1500 99 0·99 1100 101 1000 66 98 1 1 102 0·78 1 103 1 105 500 1 104 110 1 106 0·79 0 107 67 15 10 5 0 0·94 MRCA clade 1 69 0·94 1500 97 68 1000 96 1 92 70 93 1 94 0·77 95 0·59 1 0·66 500 71 0·82 0 10 5 0 Occurrence of the node (Ma) Oligocene 1 1500 88 1 77 87 1 1000 89 1 90 0·98 91 1 80 500 85 86 81 84 0 82 0·79 6 M 4 Ma 2 0 0·69 83 1 15 Asian lxora Mascarene lxora Neotropical lxora Madagascan lxora Continental African lxora Madagascan lxora L Miocene (Ma) 20 1 78 1 79 1 75 E 25 74 1 76 0·89 MRCA clade 3 8 L 73 1 72 Ma 30 0·98 0·99 15 E 0·9 108 109 1 Pacific lxora Plio 10 5 Plt 0 F I G . 5. Combined plastid –nuclear BEAST maximum clade credibility tree for Ixora. The 95 % highest posterior density intervals on time divergence estimates and posterior probabilities assigned to each node are indicated. Posterior distributions on node age estimations are also provided. Clade 1, ‘Afro-Madagascan clade’; Clade 2, ‘Madagascan clade’; Clade 3, ‘East Africa– Madagascar clade’. cladogenesis commencing 4.69 Ma (node 99) and 4.61 Ma (node104). The estimated crown age of Clade 3 (node 75) is 3.4 million years old (1.56– 5.49, 95 % HPD), and divergence in the Madagascan lineage (node 77) began 2.92 Ma (1.34– 4.69, 95 % HPD). DISCUSSION We observed topological incongruence (albeit weakly supported) in the phylogenetic placement of several taxa between the plastid and nuclear DNA data sets. Furthermore, we also observed two instances in which multiple accessions of species did not group together in the phylogenetic reconstruction (i.e. I. mangabensis and I. regalis). Phylogenetic incongruence between independent data sets can be indicative of hybridization (Linder and Rieseberg, 2004), although there are also several other processes such as incomplete lineage sorting, orthology–paralogy conflation and recombination that can produce the same pattern (e.g. Small et al., 2004; Mort et al., 2007). The conflicting placements of the Madagascan I. densithyrsa, I. mocquerysii and I. regalis 2 in the ETS, ITS and combined plastid data sets (Fig. 2; Supplementary Data Figs S1 and S2) contributed to a reduction in the overall resolution and node support in Clade 2. The ETS sequence data (Supplementary Data Fig. S1) support a close association between I. densithyrsa, I. mocquerysii and I. siphonantha. This is consistent with De Block (2013) who indicates that these three species are morphologically similar. However, the plastid haplotypes of I. densithyrsa and I. mocquerysii are similar to those of I. homolleae and I. quadrilocularis. The geographical distribution and habitat range of I. densithyrsa and I. mocquerysii overlap with those of I. homolleae and I. quadrilocularis; all four species occur either in littoral (i.e. I. homolleae) or in lowland humid forest in the eastern province of Toamasina. Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Occurrence of the node (Ma) Ma Ma 1 112 0·61 65 OUTGROUP Clade 3 60 58 V. madagascariensis I. cauIifIora I. francii I. coIIina I. sp. Brunei I. sp. MaIaysia I. chinensis I. nitens I. borboniae I. aIuminicoIa I. ferrea I. brevifoIia I. densithyrsa I. regaIis 2 I. foIiicaIyx I. quadriIocuIaris I. homoIIeae 1 I. homoIIeae 2 I. pIatythyrsa I. microphyIIa I. Iagenifructa I. ampIidentata I. siphonantha I. mocquerysii I. regaIis 1 I. guiIIotii I. nematopoda I. scheffIeri I. foIiosa I. hippoperifera I. brachypoda 1 I. brachypoda 2 I. guineensis I. batesii I. hiernii I. minutifIora I. hartiana I. praetermissa I. maciIenta I. narcissodora I. tanzaniensis I. crassipes I. cremixora 1 I. cremixora 2 I. rakotonasoIoi I. eIIiotii I. perrieri I. ankazobensis 3 I. ankazobensis 1 I. ankazobensis 2 I. mangabensis 2 I. ambrensis I. moramangensis I. mangabensis 1 I. emirnensis I. masoaIensis I. capuroniana 64 Clade 2 63 1 Clade 1 1734 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) 1735 Mascarene Ixora species Phylogenetic relationships of Afro-Madagascan Ixora spp The position of the Mascarene taxa in our phylogenetic inference, as sister to the Neotropical and Afro-Madagascan lineages, As exemplified by several studies (e.g. Malcomber, 2002; Maurin et al., 2007; Tosh et al., 2009), extremely low levels of Placement of the Afro-Madagascan Ixora spp. in the genus Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Our phylogenetic tree is consistent with that of Mouly et al. (2009b). In the genus, two main lineages are present, an Asian-Pacific lineage and an Afro-Indian Ocean-Neotropical lineage (Fig. 3). In the Asian-Pacific lineage, a clade of Asian species is sister to a clade of Pacific species. In the Afro-Indian Ocean-Neotropical lineage, a Mascarene clade is sister to the Afro-Madagascan and Neotropical species. The Neotropical clade is resolved as sister to the Afro-Madagascan species. Therefore, monophyletic groups recovered in this analysis are geographically delimited and correspond to tropical Asia, the Pacific regions, the Mascarenes, the Neotropics and continental Africa/Madagascar. In the Afro-Madagascan clade, the Madagascan taxa do not form a monophyletic group; instead they form two distinct clades (Fig. 3), one of which is nested in a group of African taxa. Mouly et al. (2009b) recovered this same general pattern, albeit with lower sampling of African and Madagascan species. may appear somewhat incongruous given its geographical proximity to Africa and Madagascar. The time of divergence between the Mascarene and the Neotropical/Afro-Madagascan taxa (Fig. 5: node 65) is estimated at 13.14 Ma (6.26– 20.23, 95 % HPD). It would seem that a single early colonization event of the Mascarenes occurred, after which the colonizing species adapted to new relatively extreme environments, resulting in markedly different morphological characters. Until recently, the Mascarene species included in this analysis were considered to belong to the endemic genus Myonima. Both this genus and the monospecific Mascarene endemic Doricera [not sampled here, but sister of Myonima in the molecular study of Mouly et al. (2009b)], differ greatly from the Neotropical and Afro-Madagascan Ixora spp., to which they are the sister group (Fig. 3). Although having articulate petioles, free stigmatic lobes and a single ovule per locule (key characters for Ixora), they also have a number of aberrant characters. They are heterophyllous ( juvenile foliage different), with coriacous leaves, short corolla tubes (,5 mm long), (2)-3–7-locular ovaries and relatively large fruits with stony pyrenes (De Block, 1997). Furthermore, the flowers of Doricera are reported to be dioecious (Verdcourt, 1983) and those of Myonima to be dioecious (Mouly et al., 2009b) or polygamous (Bentham and Hooker, 1873; Baker, 1877). These differences certainly explain why in the past the genera were considered closely related to, but distinct from, Ixora. Recently, however, Mouly et al. (2009b) showed that Ixora is paraphyletic unless several small satellite genera, including Doricera and Myonima, are included. Although the differences between the Mascarene and AfroMadagascan/Neotropical Ixora species are considerable, they can all be explained as insular adaptations to browsing pressure and selection for outcrossing. There is a high incidence of coriaceous, tough foliage and developmental heterophylly in the Mascarenes (Friedmann and Cadet, 1976), probably evolved in response to browsing pressure from giant tortoises (Griffiths et al., 2010). Several other Rubiaceae show a similar adaptation, e.g. Coptosperma borbonicum (Heine and Hallé, 1970). Many oceanic islands are particularly rich in dioecious species since selection for outcrossing in small, colonizing, hermaphroditic populations favours separation of the sexual functions (Anderson et al., 2006). Many dioecious species have small, relatively inconspicuous, whitish or greenish flowers ( pollinated by small generalist bees and flies) (Bawa, 1980; Baker, 1984), which is also the case in the species discussed here. As regards the differences in fruit morphology, these can be partly attributed as a defence against frugivory by giant tortoises (stony pyrenes) and an adaptation to dioecy. Dioecious species may have a reproductive disadvantage because not every individual in a population produces seeds. To compensate for this disadvantage, more seeds should be produced (Heilbuth et al., 2001; Queenborough et al., 2009). An increase in number of locules automatically increases the seed number since in Ixora a single seed per locule is produced. Furthermore, the increased mechanical protection of the seeds (stony pyrenes) may increase seed fitness. Although further research is needed to elucidate fully the evolutionary relationships between these taxa, one explanation consistent with the observed phylogenetic incongruence involves hybridization followed by repeated backcrossing. Mouly et al. (2009b) invoked interspecific hybridization to account for topological incongruence between nuclear and plastid data sets, particularly among Asian species of Ixora. Multiple accessions of I. regalis (Fig. 3) and I. mangabensis (Figs 2 and 3) did not form monophyletic species groups in our investigation. In both instances, collections from the Anjanaharibe-Sud Reserve (I. regalis 1, De Block 2081 and 2083; I. mangabensis 1, De Block 2040 and 2053) differed from collections made elsewhere (I. regalis 2, De Block 835, collected 12 km from Moramanga; I. mangabensis 2, Tosh 128 and 130, collected near Soanierano-Ivongo). The fact that populations of I. regalis and I. mangabensis in the Anjanaharibe-Sud Reserve in northern Madagascar are different from those further south in Moramanga and Sonierano-Ivongo is not strange since these localities represent different bioclimates. Populations of the same species may then be genetically isolated from other conspecific populations, resulting in the accumulation of genetic variation in the absence of morphological differentiation, as is the case for I. regalis. However, in the case of I. mangabensis, De Block (2013) observed some morphological differences between material collected from AnjanaharibeSud, and material collected from other localities throughout its geographical range. For example, I. mangabensis from Anjanaharibe-Sud is characterized by long and narrow bracts and bracteoles, and longer, narrowly triangular calyx lobes. The specimens also occur in different vegetation types and at different altitudes: lowland humid forest in SonieranoIvongo (altitude approx. 60 m) vs. montane humid forest (altitiude approx. 1000 m) in Anjanaharibe-Sud. The taxonomy of I. mangabensis may therefore require re-evaluation given the genetic differentiation and morphological variation observed between populations currently referred to as this species. 1736 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) length (I. narcissodora, 30– 75 mm; I. tanzaniensis, 22– 32 mm) and pubescence (greater in I. tanzaniensis than in I. narcissodora). However, there are also examples where species that are thought to be close relatives on the basis of morphological similarities are not grouped together in our phylogenetic tree(s). De Block (1998) noted the close resemblance between the Afromontane species I. scheffleri and I. foliosa and considered them to be close relatives. However, our results do not support a close relationship between these two species (Fig. 3). Although both have pedunculate, erect and compact inflorescences, there are key differences between them, notably in the morphology of bracts and bracteoles (typically absent in I. scheffleri) and in their geographical distribution (I. scheffleri from East Africa and I. foliosa from West Africa). Another example involves the Madagascan taxa I. amplidentata, I. emirnensis, I. mangabensis and I. moramangensis. De Block (2013) noted the similarity between these species as they all possess pedunculate, pendulous, moderately lax to lax inflorescences with a moderate number of flowers with relatively short corolla tubes. In our phylogenetic analyses (Figs 2 and 3), I. emirnensis, I. mangabensis and I. moramangensis are nested in Clade 3, although the exact relationship between these taxa is not fully resolved. In our individual and combined phylogenetic analyses, I. amplidentata is nested in Clade 2, with I. regalis and I. guillotii among others. Mouly et al. (2009b) also included sequence data of I. amplidentata and I. emirnensis (from different specimens) in their study of Ixoreae, and their analyses also indicated that I. amplidentata and I. emirnensis are not closely related, despite morphological similarities. There are few resolved or well-supported relationships in either clade of Madagascan taxa. In Clade 2, there is strong support for the group of I. homolleae and I. quadrilocularis (discussed in more detail in the next section) and for the group of I. platythyrsa and I. microphylla. These last two species differ markedly in inflorescence structure (long pedunculate, pendulous inflorescences with numerous flowers in I. platythyrsa vs. sessile or shortly pedunculate, erect inflorescences with few flowers in I. microphylla), but other flower/inflorescence characters are similar, notably the well-developed triangular bracteoles and calyx lobes and the corolla lobes with acuminate tip. In Clade 3, I. capuroniana and I. masoalensis are strongly supported sister taxa (Fig. 3). These taxa have sessile inflorescences; coriaceous leaves that are pale yellow when dried; and small calyces, bracts and bracteoles (De Block, 2013). Finally, I. crassipes and I. cremixora form a strongly supported clade. These species occur in humid or sub-humid (semi-)deciduous forests in Western and Northern Madagascar. Most of the other species sampled in this study occur in the humid forests (littoral, lowland or altitudinal) on the eastern and North-Eastern coasts of Madagascar. Key morphological traits in Madagascan Ixora Important morphological features for species-level identification of Madagascan Ixora include inflorescence and flower characters (De Block, 2003). However, there are some morphological features that can be used to distinguish between taxa from the two Madagascan lineages. Taxa from Clade 2 possess comparatively long calyx tubes and calyx lobes, relative to taxa from Clade 3 Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 interspecific sequence divergence in woody Rubiaceae can confound attempts to retrieve fully resolved and well-supported phylogenetic inferences. However, in the current study, there are some well-supported relationships in the Afro-Madagascan clade that are recovered in our phylogenetic analyses of the plastid and nuclear combined data set. The West– Central African I. nematopoda and the East African Afromontane I. scheffleri are sister to all other Afro-Madagascan taxa in Clade 1 (Fig. 3). There is also strong support for the West– Central African species I. foliosa (Afromontane element) and I. hippoperifera being sister to all the remaining taxa in Clade 1 (Fig. 3). These four taxa all possess pedunculate inflorescences subtended by inflorescencesupporting leaves (De Block, 1998), though the inflorescences of I. hippoperifera may also be sessile (Fig. 4A). For the continental African taxa, De Block (1998; fig. 9, p. 25) postulated that manyflowered, pedunculate, lax, corymbose inflorescences constitute the basic type. Our results corroborate De Block’s assumption (1998) that sessile inflorescences represent a derived condition in continental African Ixora. Ixora brachypoda and I. hartiana also have pedunculate inflorescences, whereas all other continental African species represented in this study have sessile inflorescences (Fig. 4A). Some African species groups, thought to be related on the basis of their morphological similarity (De Block 1998), are recovered in our phylogenetic analyses. Ixora guineensis, I. batesii and I. minutiflora (I. guineensis complex) are largely endemic to the lower Guinea regional sub-centre of endemism (RSE) occupying lowland and gallery forest. These three species are recovered in a well-supported monophyletic group (Fig. 3), with the upper Guinea endemic I. hiernii. This group is supported by the presence of sessile, lax, more or less densely pubescent inflorescences (glabrous to densely covered with minute hairs in the case of I. minutiflora), and sessile to shortly pedicellate flowers with relatively short corolla tubes (0.5 –3.0 cm). Our results also provide support for a close relationship between I. macilenta and I. praetermissa (Fig. 3). These two species are endemic to the lower Guinea sub-centre of endemism, occupying both primary and secondary forest and occurring in riverine (gallery) forest. In addition, both of these species possess sessile, lax inflorescences. Sister to these two species in our phylogenetic analyses is I. hartiana, a species with disjunct distribution in Africa, which grows in gallery forest and open woodland. This species is known from only a small number of collections, three from Congo, one from Tanzania and one collection from Angola (De Block, 1998). The morphology of I. hartiana differs somewhat from that of both I. macilenta and I. praetermissa, most notably by the presence of pedunculate, erect and lax inflorescences and long pedicellate flowers. There is strong support for the close relationship between the eastern African species I. narcissodora and I. tanzaniensis, although these two taxa differ in both morphology and habitat (De Block, 1998). Ixora narcissodora is widespread throughout the lowland evergreen and riverine forests from Kenya to Mozambique. In contrast, I. tanzaniensis is restricted to a small area of lowland forest in Tanzania. These two species differ most notably in inflorescence structure (I. narcissodora, sessile and lax; I. tanzaniensis, sessile and compact), corolla tube Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) fruit wall and stony pyrenes. The fact that I. lagenifructa is not placed with the other four-locular species (i.e. I. homolleae and I. quadrilocularis) in our separate or combined phylogenetic analyses could be considered surprising (Figs 2, 3 and 4E). Although most Ixora spp. differ from each other on the basis of minor and continuous characters, an increase in the number of locules and associated characters (e.g. thickened fruit wall, stony endocarp) represents a more profound shift in morphology. De Block (2013) regards these four-locular Ixora spp. as a natural group, and therefore additional independent material of I. lagenifructa is required to verify the results of our molecular analyses. A third morphological trait unique to the Madagascan element of Ixora is an increase in corolla tube length (Table 3). Flower size, and in particular the length of the corolla tube, is extremely variable among the Madagascan representatives of Ixora (De Block, 2007). Typically the length of corolla tubes varies between 1 and 9 cm, although there are a number of species that possess corolla tubes up to 13 cm in length (De Block, 2007). Furthermore, there are four species in which corolla tubes exceed 15 cm in length and can be as long as 23 cm. Corolla tubes of this size are rare in Rubiaceae as a whole, and the range in corolla tube lengths among Madagascan Ixora spp. (0.4 – 23.0 cm) is remarkable (De Block, 2007). In comparison, corolla tube length varies between 0.5 and 11.0 cm in continental African species (Table 3; De Block, 1998). We were able to incorporate sequence data from three of the four Ixora spp. that have extremely long corolla tubes. As postulated by De Block (2007), the increase in corolla tube length has occurred several times (Fig. 4F). Ixora densithyrsa and I. siphonantha, thought to be close relatives due to the possession of identical bracts, bracteoles and calyces (De Block, 2007), are recovered in the exclusively Madagascan Clade 2. Ixora crassipes is nested in Clade 3 in our phylogenetic analyses, and is clearly distinct from the other large flowered Ixora sampled in this investigation notably because of the reduced bracts, bracteoles and calyx lobes and the yellowish colour of the dried specimens. Therefore, with the exception of the four-locular ovaries that are exclusive to Clade 2, both lineages of Madagascan taxa contain pauciflorous and uniflorous species, and species with corolla tubes .15 cm in length. Historical biogeography Our molecular study dates the crown age of Ixora sometime during the mid Miocene (16.67 Ma). This date is similar to that in the studies of Mouly (2007) and Bremer and Eriksson (2009), who estimated the crown age of Ixora at 14– 15 Ma. Similarly, the age estimate for the crown age of the Afro-Madagascan group during the Late Miocene (7.95 Ma) is consistent with Mouly (2007). Divergence in the Afro-Madagascan group began within the last 8 Ma during the late Miocene, with extensive cladogenesis occurring throughout the Pliocene. The two separate lineages of Madagascan taxa are of different ages. The exclusively Madagascan Clade 2 started to diversify during the Miocene – Pliocene boundary, whereas Clade 3 had its origin in the late Pliocene and underwent a period of rapid speciation that continued into the Pleistocene. The results of our phylogenetic investigations for Ixora are in keeping with the general observations from the growing Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 (Fig. 4B, C). In Clade 2, calyx lobe length varies between 0.5 and 15.0 mm, compared with between 0.1 and 0.8 mm in Clade 3 (Table 3). With the exception of I. amplidentata, calyx tube length typically varies between 0.5 and 10.0 mm in Clade 2, and between 0.20 and 0.75 mm in Clade 3 (Table 3). Calyx tube and calyx lobe lengths are particularly long in the fourlocular species and I. foliicalyx, the latter of which is sister to I. homolleae and I. quadrilocularis on our plastid and nuclear combined phylogenetic inference (Fig. 3). There are three morphological traits occurring in Madagascan Ixora that are rare or absent throughout the rest of the genus. The taxonomic distribution of these traits can be interpreted in light of our phylogenetic results (Fig. 4D– F). The first of these is the extreme reduction in flower numbers towards uniflorous inflorescences (Fig. 4D). Most species in the genus have striking inflorescences containing large numbers of flowers. For example, Nilsson et al. (1990) reported up to 282 flowers in a single inflorescence of the Madagascan species I. plathythyrsa. Pauciflorous species, containing ,15 flowers per inflorescence, are rare in Ixora (De Block, 2008). The occurrence of solitary-flowered inflorescences is exceptionally rare, and only a few uniflorous Ixora spp. have been described, such as I. dzumacensis from New Caledonia (Guillaumin, 1929). On Madagascar, there are six uniflorous species currently recognized, and about 30 % of Madagascan Ixora spp. are uniflorous or contain ,15 flowers per inflorescence (De Block, 2008). Five of these uniflorous species have been accommodated in Ixora section Microthamnus (Guédès, 1986; De Block, 2008). A sixth uniflorous Madagascan species (I. homolleae) differs from members of section Microthamnus in a number of characters, most notably its large flowers, four-locular ovaries, and fruits and stigma with four stigmatic lobes. Members of section Microthamnus are poorly collected in the field, because of the inconspicuous nature of their inflorescences and their rarity (De Block, 2008). We were only able to obtain silica gel material from one representative of this section (I. rakotonasoloi) and, as a result, we were unable to test the monophyly of this seemingly natural group of uniflorous species. However, we were able to include multiple accessions of I. homolleae, and from our phylogenetic analyses it is evident that the evolution of uniflorous species has occurred at least twice in Madagascan Ixora (Fig. 4D). A second morphological trait present in Madagascan, but not continental African, Ixora is four-locular ovaries (Fig. 4E). The genus as a whole is typically characterized by uniovulate bilocular ovaries, and four-locular ovaries are rare (De Block, 2013). Mouly et al. (2009b) favoured a broad circumscription of the genus, incorporating a number of taxa previously recognized as satellite genera of Ixora. Among these are the plurilocular Mascarene endemic genus Myonima and the four-locular Ixora mooreensis, an endemic to the Society Islands (French Polynesia, Pacific Ocean) that was formerly placed in the monotypic genus Hitoa. As such, the generic circumscription of Ixora now accommodates plurilocular (two- to seven-locular) ovaries. The uniflorous four-locular I. homolleae, endemic to the littoral forests of eastern Madagascar, was formerly classified in the monotypic genus Thouarsiora. De Block (2013) recognizes several other Madagascan Ixora spp. allied to I. homolleae that have four-locular ovaries and fruits, notably I. lagenifructa, I. quadrilocularis and I. trimera. These species also share a number of other characters, including large fruits with a thick 1737 1738 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) Possible mechanisms driving the radiation of Madagascan Ixora spp Dispersal of Ixora into Africa and Madagascar, and the species diversification that followed, occurred in the late Miocene (i.e. approx. 8 Ma). Palynological and macrofossil data (reviewed in Maley, 1996; Jacobs, 2004) indicate that grass-dominated savannahs began to expand throughout sub-Saharan Africa at the expense of humid forest in the mid Miocene (16 Ma) and were widespread by the late Miocene (8 Ma). Rain forest taxa would have been restricted to small patches of humid forest in upland areas or along lowland river systems (Robbrecht, 1996; Plana, 2004) at the time when Ixora began to diverge in continental Africa. The Cenozoic climate history of Madagascar and the chronology of the development of its biomes remain largely unknown (Wells, 2003). Nevertheless, Wells (2003) surmised that humid forest conditions would probably have been present in the eastern watershed of Madagascar from the Oligocene onwards. Although several lineages of AfroMadagascan Ixora were already established at the Pliocene – Pleistocene boundary, a number of Pleistocene speciation events appear to have occurred. On continental Africa, the Pliocene and Pleistocene epochs were characterized by a mosaic of fragmented humid forest, interspersed by savannah (Plana, 2004). Expansion and subsequent contraction of each biome type was mediated by the climatic conditions of the time (Plana, 2004). In Madagascar there is considerable habitat heterogeneity resulting from a number of factors, such as a wide variety of geology, significant topographic relief and the orographic rainfall in the east (Wells, 2003; Dewar and Richard, 2007). There is also evidence of considerable displacement of vegetation zones as a result of Pleistocene and Holocene climatic fluctuations (Burney, 1996; Straka, 1996). This habitat heterogeneity, coupled with the cyclical expansion and retraction of biome ranges during periods of climate oscillation in the Pliocene and Pleistocene, may have led to formerly contiguous populations of conspecific taxa becoming geographically isolated in temporal forest refugia (Janssen et al., 2008). One of the main driving forces of rapid radiations is thought to be the development of new ecological opportunities in the absence of competition in largely unoccupied, depauperate environments (e.g. following orogenic uplift or post-glacial re-expansion) (Hughes and Eastwood, 2006; Janssens et al., 2009). In addition to novel ecological opportunities afforded by repeated habitat fragmentation during periods of climatic perturbation, pollination syndromes are thought to be a major factor in the reproductive isolation of species (Hodges and Arnold, 1994). Malcomber (2002) postulated that the rapid diversification of Gaertnera spp. (also Rubiaceae) is correlated with a change in corolla morphology, and that the elongated tubular corolla morphology (relative to the sister genus Pagamea) represented a key innovation in the group. Perhaps the most conspicuous feature of Madagascan Ixora spp. is the extreme variability in corolla tube length, which would indicate that speciation is at least partly pollinator driven (De Block, 2007). Neal et al. (1998) stated that narrow and/or long corolla tubes could preclude a wide range of potential pollinators from gaining access to the reward, thereby increasing both pollination precision and pollinator fidelity. Ixora spp. also exhibit secondary pollen presentation, whereby pollen from the protandrous anthers is deposited on the non-receptive abaxial sides of the stigmatic lobes and/or the upper part of the style prior to anthesis (Puff et al., 1996). The stigmatic lobes (and style) therefore provide the dual function of the pollen-presenting organ, and then, at maturity, the pollen recipient (Nilsson et al., 1990). In Ixora, the narrow tubular corolla provides a mechanical guide to ensure precise transfer of pollen between the style and the pollination vector (Nilsson et al., 1990). The variation in corolla tube length in Madagascan Ixora has clearly resulted in some pollinator specialization. For example, the flowers of I. densithyrsa and I. siphonantha reach lengths .20 cm long, and must be pollinated by members of the guild of long tongue Madagascan hawkmoths (Nilsson, 1998). Nilsson et al. (1990) reported that pollination in I. platythyrsa is carried out by nocturnal moths, and many other Madagascan Ixora have delicately scented, white corollas typical of moth pollination. Changes in corolla tube length over time may preclude generalist pollinators from successfully affecting their pollinator services and ultimately lead to pollinator specificity. Given the evidence from other studies (e.g. Wilmé et al., 2006; Janssen et al. 2008; Pearson and Raxworthy, 2009; Strijk et al., 2012), it is increasingly apparent that short-term climatic events and associated habitat fragmentation during the Pliocene and Pleistocene have facilitated the rapid accumulation of biodiversity of some taxonomic groups on Madagascar. However, as with other studies (e.g. Malcomber, 2002), it would appear likely that pollination biology ( pollinator specificity and phenology) has played an equally vital role in the recent diversification of Madagascan Ixora. Conclusions Madagascan Ixora do not form a monophyletic group, but are represented by two lineages of different ages. Our results indicate at least one dispersal event from East Africa into Madagascar towards the end of the Pliocene. Both Ixora lineages on Madagascar exhibit morphological innovations that are rare in the rest of the genus, including a trend towards pauciflorous inflorescences and a trend towards extreme corolla tube length. Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 literature on the historical biogeography of the flora and fauna of Madagascar (see below). Despite its prolonged geographical isolation, the current biota of Madagascar is seemingly comprised primarily of recently evolved endemics, evolving in situ following Cenozoic trans-oceanic dispersal (Yoder and Nowak, 2006). Other studies focusing both on Rubiaceae (e.g. Malcomber, 2002; Maurin et al., 2007; Groeninckx, 2009; Tosh et al. 2009; Wikström et al., 2010) and on other angiosperm families (e.g. Davis et al., 2002 on Malpighiaceae; Meve and Liede, 2002 on Apocynaceae; Plana, 2003 on Begoniaceae; Renner, 2004 on Melastomataceae; Yuan et al., 2005 on Gentianaceae; Trénel et al., 2007 on Arecaceae; Schaefer et al., 2009 on Cucurbitaceae; Buerki et al., 2013 on Sapindaceae and other families; Strijk et al., 2012 on Asteraceae) have revealed multiple independent dispersal events across the Mozambique Channel during the Cenozoic era. The most commonly observed pattern thus far is limited dispersal (one of few events) per genus from East Africa into Madagascar, followed by speciation in Madagascar. Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) This suggests that the same ecological and selective pressures are acting upon taxa from both Madagascan lineages. The recent radiation in Madagascan Ixora is likely to have been driven by increased ecological opportunities following periods of habitat contraction and expansion during the Pliocene/Pleistocene, coupled with increased pollinator specificity between Ixora spp. resulting from corolla tube length diversification in the genus. S U P P L E M E N TARY D ATA ACK N OW L E DG E M E N T S We are grateful to Madagascar National Parks ( previously known as Association Nationale pour la Gestion des Aires Protégées, ANGAP), the Ministère des Eaux et Forêts and the Parc Botanique et Zoologique de Tsimbazaza (PBZT) for permission to collect in protected areas of Madagascar. We thank the Africa & Madagascar Department of the Missouri Botanical Garden for facilitating logistics, and express special thanks to Dr Frank Rakotonasolo for all his help in the field. We thank Dr Sylvain Razafimandimbison, Dr Johan van Valkenburg and Dr Piero Delprete for providing silica gel samples. We thank Anja Vandeperre, Nathalie Geerts and Anbar Khodabandeh for technical assistance on the project, and Dr Laura J. Kelly for proofreading the manuscript. This work was supported by grants from the Fund for Scientific Research-Flanders (FWO; G.0250.05 and G.026.04). LIT E RAT URE CITED Anderson GJ, Bernardello G, Opel MR, Santos-Guerra A, Anderson M. 2006. Reproductive biology of the dioecious Canary Islands endemic Withania aristata (Solanaceae). American Journal of Botany 93: 1295–1305. Andreasen K, Bremer B. 1996. Phylogeny of the subfamily Ixoroideae (Rubiaceae). Opera Botanica Belgica 7: 119–138. Andreasen K, Bremer B. 2000. Combined phylogenetic analysis in the Rubiaceae– Ixoroideae: morphology, nuclear and chloroplast DNA data. American Journal of Botany 87: 1731– 1748. Baker HG. 1984. Some functions of dioecy in seed plants. American Naturalist 124: 149– 158. Baker JG. 1877. Flora of Mauritius and the Seychelles. London: Reeve & Co. Baldwin BG, Markos S. 1998. Phylogenetic utility of the external transcribed spacer (ETS) of 18S–26S rDNA: congruence of ETS and ITS trees of Calycadenia (Compositae). Molecular Phylogenetics and Evolution 10: 449– 463. Bawa KS. 1980. Evolution of dioecy in flowering plants. Annual Review of Ecology and Systematics 11: 15– 39. Bentham G, Hooker JD. 1873. Ordo LXXXIV. Rubiaceae. In: Bentham G, Hooker JD, eds. Genera Plantarum, Vol. 2. London: Reeve & Co., 7– 151. Bremer B, Eriksson T. 2009. Time tree of Rubiaceae: phylogeny and dating the family, subfamilies, and tribes. International Journal of Plant Science: 170: 766–793. Buerki S, Forest F, Alvarez N, Nylander JAA, Arrigo N, Sanmartı́n I. 2011. An evaluation of new parsimony-based versus parametric inference methods in biogeography: a case study using the globally distributed plant family Sapindaceae. Journal of Biogeography 38: 531– 550. Buerki S, Devey DS, Callmander MW, Phillipson PB, Forest F. 2013. Spatio-temporal history of the endemic genera of Madagascar. Botanical Journal of the Linnean Society 171: 304 –329. Burney DA. 1996. Climate change and fire ecology as factors in the Quaternary biogeography of Madagascar. In: Lourenço WR, ed. Biogéographie de Madagascar. Paris: Editions de l’ORSTOM, 49–58. Davis AP, Bridson D. 2003. Introduction to the Rubiaceae. In: Goodman SM, Benstead JP, eds. The natural history of Madagascar. Chicago: University of Chicago Press, 431– 434. Davis AP, Govaerts R, Bridson DM, Ruhsam M, Moat J, Brummitt NA. 2009. A global assessment of distribution, diversity endemism, and taxonomic effort in the Rubiaceae. Annals of the Missouri Botanical Garden 96: 68–78. Davis CC, Bell CD, Fritsch PW, Matthews S. 2002. Phylogeny of Acridocarpus-Brachylophon (Malpighiaceae): implications for tertiary tropical floras and Afroasian biogeography. Evolution 56: 2395– 2405. De Block P. 1997. Biosystematic studies in the tribe Pavetteae (Rubiaceae– Ixoroideae). PhD Thesis. University of Antwerp, UIA, Belgium. De Block P. 1998. The African species of Ixora (Rubiaceae– Pavetteae). Opera Botanica Belgica 9: 1 –218. De Block P. 2003. Ixora (Tribe Ixoreae). In: Goodman SM, Benstead JP, eds. The natural history of Madagascar. Chicago: University of Chicago Press, 442–443. De Block P. 2007. Three new Madagascan Ixora species (Rubiaceae) with flowers up to 23 cm long. Nordic Journal of Botany 25: 75– 84. De Block P. 2008. Synopsis of the uniflorous Madagascan Ixora (Rubiaceae) species belonging to section Microthamnus. Belgian Journal of Botany 141: 141–156. De Block P. 2013. The Malagasy species of Ixora (Rubiaceae– Pavetteae). Opera Botanica Belgica 16: in press. Dewar RE, Richard AF. 2007. Evolution in the hypervariable environment of Madagascar. Proceedings of the National Academy of Sciences, USA 104: 13723–13727. Doyle JJ, Doyle JL. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin, Botanical Society of America 19: 11– 15. Drummond AJ, Rambaut A. 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Biology 7: 214. Felsenstein J. 1985. Phylogenies and the comparative method. American Naturalist 125: 1– 15. Friedmann F, Cadet T. 1976. Observations sur l’hétérophyllie dans les Iles Mascareignes. Adansonia 15: 423– 440. Gradstein FM, Ogg JG, Smith AG, Bleeker W, Lourens LJ. 2004. A new Geologic Time Scale, with special reference to Precambrian and Neogene. Episodes 27: 20. Graham A. 2009. A revised fossil record of the Rubiaceae. Annals of the Missouri Botanical Garden 96: 90– 108. Griffiths CJ, Jones CG, Hansen DM, et al. 2010. The use of extant nonindigenous tortoises as a restoration tool to replace extinct ecosystem engineers. Restoration Ecology 18: 1– 7. Groeninckx I. 2009. Integrated study of the herbaceous Spermacoceae (Rubiaceae): systematics and evolution. PhD Thesis. KU Leuven, Belgium. Guédès M. 1986. Ixoras malgaches nouveaux à fleurs solitaires. Phytologia 60: 243–250. Guillaumin A. 1929. Contribution à la flore de la Nouvelle-Calédonie. Bulletin du Muséum National d’Histoire Naturelle. Paris 2: 117– 123. Heilbuth JC, Ilves KL, Otto SP. 2001. The consequences of dioecy for seed dispersal: modeling the seed-shadow handicap. Evolution 55: 880–888. Heine H, Hallé N. 1970. Une Rubiacée des Iles Mascareignes à feuilles ornamentales: Enterospermum borbonicum. Adansonia, Series 2 10: 315–327. Hodges SA, Arnold ML. 1994. Floral and ecological isolation between Aquilegia formosa and Aquilegia pubescens. Proceedings of the National Academy of Sciences, USA 91: 2493–2496. Holmgren PK, Holmgren NH, Barnett LC. 1990. Index Herbariorum. Part I. The herbaria of the world. Regnum Vegetabile 120: 1– 693. Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Supplementary data are available online at www.aob.oxfordjournals.org and consist of the following. Figure S1: ETS Bayesian majority rule consensus tree. Figure S2: ITS Bayesian majority rule consensus tree. Figure S3: one of the 5362 most-parsimonious trees generated in the maximum parsimony analysis of the combined plastid – ETS – ITS data set. Figure S4: BEAST maximum clade credibility tree for Rubiaceae based on three plastid regions. Table S1: DNA substitution models used in Bayesian analyses. Table S2: BEAST node age estimations –Rubiaceae wide analysis. Table S3: BEAST node age estimations– Ixora analysis. 1739 1740 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) Plana V. 2003. Phylogenetic relationships of the Afro-Malagasy members of the large genus Begonia inferred from trnL intron sequences. Systematic Botany 28: 693–704. Plana V. 2004. Mechanisms and tempo of evolution in the African Guineo-Congolian rainforest. Philosophical Transactions of the Royal Society B: Biological Sciences 359: 1585– 1594. Posada D, Crandall KA. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817– 818. Puff C, Robbrecht E, Buchner R, De Block P. 1996. A survey of secondary pollen presentation in the Rubiaceae. Opera Botanica Belgica 7: 369 – 402. Queenborough SA, Mazer SJ, Vamosi SM, Garwood NC, Valencia R, Freckleton RP. 2009. Seed mass, abundance and breeding system among tropical forest species: do dioecious species exhibit compensatory reproduction or abundances? Journal of Ecology 97: 555–566. Rajaseger G, Tan HTW, Turner IM, Kumar PP. 1997. Analysis of genetic diversity among Ixora cultivars (Rubiaceae) using random amplified polymorphic DNA. Annals of Botany 80: 355– 361. Rambaut A, Drummond AJ. 2007. Tracer v1.4. Available from http:// beast.bio.ed.ac.uk/Tracer. Renner SS. 2004. Multiple Miocene Melastomataceae dispersal between Madagascar, Africa, and India. Philosophical Transactions of the Royal Society B: Biological Sciences 359: 1485– 1494. Robbrecht E. 1988. Tropical woody Rubiaceae: characteristic features and progressions. Contributions to a new subfamilial classification. Opera Botanica Belgica 1: 1 –271. Robbrecht E. 1996. Geography of African Rubiaceae with reference to glacial rain forest refuges. In: van der Maesen LJG, van der Burgt XM, van Medenbach de Rooy JM, eds. The biodiversity of African plants. Proceedings XIVth AETFAT Congress. Dordrecht: Kluwer, 564 –581. Schaefer H, Heibl C, Renner S. 2009. Gourds afloat: a dated phylogeny reveals an Asian origin of the gourd family (Cucurbitaceae) and numerous oversea dispersal events. Proceedings of the Royal Society B: Biological Sciences 276: 843– 851. Shaw J, Licky EB, Beck JT, et al. 2005. The tortoise and the hare II: relative utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis. American Journal of Botany 92: 142– 166. Simmons MP, Ochoterena H. 2000. Gaps as characters in sequence based phylogenetic analysis. Systematic Biology 49: 369– 381. Small RL, Cronn RC, Wendel JF. 2004. Use of nuclear genes for phylogeny reconstruction in plants. Australian Systematic Botany 17: 145– 170. Staden R, Beal K, Bonfield J. 1998. The Staden package. In: Miseners S, Krawetz S, eds. Computer methods in molecular biology. New York: Humana Press, 115– 130. Straka H. 1996. Histoire de la vegetation de Madagascar oriental dans les derniers 100 millenaires. In: Lourenço WR, ed. Biogeographie de Madagascar. Paris: Editions de l’ORSTOM, 49– 58. Strijk JS, Noyes RD, Strasberg D, et al. 2012. In and out of Madagascar: dispersal to peripheral Islands, insular speciation and diversification of Indian Ocean Daisy Trees (Psiadia, Asteraceae). PLoS One 7: pe42932. Swofford DL. 2003. PAUP* 4.0b10: phylogenetic analysis using parsimony (* and other methods). Sunderland, MA: Sinauer Associates. Taberlet P, Gielly L, Pautou G, Bovet J. 1991. Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Molecular Biology 17: 1005– 1109. Tosh J, Davis AP, Dessein S, et al. 2009. Phylogeny of Tricalysia A.Rich. (Rubiaceae) and its relationships with allied genera based on plastid DNA data: resurrection of the genus Empogona. Annals of the Missouri Botanical Garden 96: 194–213. Trénel P, Gustafsson MHG, Baker WJ, Asmussen-Lange CB, Dransfield J, Borchsenius F. 2007. Mid-Tertiary dispersal, not Gondwanan vicariance explains distribution patterns in the wax palm subfamily (Ceroxyloideae: Arecaceae). Molecular Phylogenetics and Evolution 45: 272– 288. Verdcourt B. 1983. Notes on Mascarene Rubiaceae. Kew Bulletin 37: 521– 574. Wells NA. 2003. Some hypotheses on the Mesozoic and Cenozoic paleoenvironmental history of Madagascar. In: Goodman SM, Benstead JP, eds. The natural history of Madagascar. Chicago: University of Chicago Press, 16–34. White F. 1983. The vegetation of Africa. A descriptive memoir to accompany the Unesco/AETFAT/UNSO vegetation map of Africa. Paris: Unesco. White TJ, Bruns T, Lee S, Taylor J. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: Innis MA, Gelfand Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Huelsenbeck JP, Ronquist F. 2001. MRBAYES. Bayesian inference of phylogeny. Bioinformatics 17: 754–755. Hughes CE, Eastwood R. 2006. Island radiation on a continental scale: exceptional rates of plant diversification after uplift of the Andes. Proceedings of the National Academy of Sciences, USA 103: 10334–10339. Jacobs BF. 2004. Palaeobotanical studies from tropical Africa: relevance to the evolution of forest, woodland and savannah biomes. Philosophical Transactions of the Royal Society B: Biological Sciences 359: 1573– 1583. Janssen T, Bystriakova N, Rakatondrainibe F, Coomes D, Labat J-N, Schneider H. 2008. Neoendemism in Madagascan scaly tree ferns results from recent, coincident diversification bursts. Evolution 62: 1876– 1889. Janssens SB, Knox EB, Huysmans S, Smets EF, Merckx VSFT. 2009. Rapid radiation of Impatiens (Balsaminaceae) during Pliocene and Pleistocene: result of a global climate change. Molecular Phylogenetics and Evolution 52: 806– 824. Kainulainen K, Razafimandimbison SG, Bremer B. 2013. Phylogenetic relationships and new tribal delimitations in subfamily Ixoroideae (Rubiaceae). Botanical Journal of the Linnean Society in press. Linder CR, Rieseberg LH. 2004. Reconstructing patterns of reticulate evolution in plants. American Journal of Botany 91: 1700–1708. Löhne C, Borsch T. 2005. Molecular evolution and phylogenetic utility of the petD group II intron: a case study in basal angiosperms. Molecular Phylogenetics and Evolution 22: 317–332. Lorence DH, Wagner WL, Mouly A, Florence J. 2007. Revision of Ixora (Rubiaceae) in the Marquesas Islands (French Polynesia). Botanical Journal of the Linnean Society 155: 581– 597. Maddison DR, Maddison WP. 2002. MacClade 4: analysis of phylogeny and character evolution, version 4.01. Sunderland, MA: Sinauer Associates. Malcomber ST. 2002. Phylogeny of Gaertnera Lam. (Rubiaceae) based on multiple DNA markers: evidence of a rapid radiation in a widespread, morphologically diverse genus. Evolution 56: 42– 57. Maley J. 1996. The African rain forest – main characteristics of changes in vegetation and climate from the upper Cretaceous to the Quaternary. In: Alexander IJ, Swaine MD, Watling R, eds. Essays on the ecology of the Guinea-Congo rainforest. Proceedings of the Royal Society Edinburgh B 104: 31– 73. Maurin O, Davis AP, Chester M, Mvungi EF, Jaufeerally-Fakim Y, Fay MF. 2007. Towards a phylogeny for Coffea (Rubiaceae): identifying wellsupported lineages based on nuclear and plastid DNA sequences. Annals of Botany 100: 1565–1583. Meve U, Liede S. 2002. Floristic exchange between mainland Africa and Madagascar: case studies in Apocynaceae-Asclepiadoideae. Journal of Biogeography 29: 865– 873 Mort ME, Archibald JK, Randle CP, et al. 2007. Inferring phylogeny at low taxonomic levels: utility of rapidly evolving cpDNA and nuclear ITS loci. American Journal of Botany 94: 173– 183. Mouly A. 2007. Systématique de la tribu des Ixoreae A.Gray (Rubiaceae); phylogénie, biogeography et taxonomie. Thèse de doctorat, Muséum national d’Histoire naturelle, Paris, France. Mouly A, Razafimandimbison SG, Florence J, Jérémie J, Bremer B. 2009a. Paraphyly of Ixora and new tribal delimitation of Ixoreae (Rubiaceae): inference from combined chloroplast (rps16, rbcL and trnT-F) sequence data. Annals of the Missouri Botanical Garden 96: 146 –160 Mouly A, Razafimandimbison SG, Khodabandeh A, Bremer B. 2009b. Phylogeny and classification of the species-rich pantropical showy genus Ixora (Rubiaceae-Ixoreae) with implications of geographical monophyletic units and hybrids. American Journal of Botany 96: 686– 706. Neal PR, Dafni A, Giurfa M. 1998. Floral symmetry and its role in plant–pollinator systems: terminology, distribution, and hypotheses. Annual Review of Ecology and Systematics 29: 345– 373. Negrón-Ortiz V, Watson LE. 2002. Molecular phylogeny and biogeography of Erithalis (Rubiaceae), an endemic of the Caribbean Basin. Plant Systematics and Evolution 234: 71– 83. Nilsson LA. 1998. Deep flowers for long tongues. Trends in Ecology and Evolution 13: 259– 260. Nilsson LA, Rabakonandrinina E, Petterson B, Ranaivo J. 1990. ‘Ixoroid’ secondary pollen presentation and pollination by small moths in the Malagasy treelet Ixora platythyrsa (Rubiaceae). Plant Systematics and Evolution 170: 161–175. Pearson RG, Raxworthy CJ. 2009. The evolution of local endemism in Madagascar: watershed versus climatic gradient hypotheses evaluated by null biogeographic models. Evolution 63: 959– 967. Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) DH, Sninsky JJ, White TJ, eds. PCR protocols: a guide to methods and applications. San Diego, CA: Academic Press, 315–322. Wikström N, Savolainen V, Chase M. 2001. Evolution of the angiosperms: calibrating the family tree. Proceedings of the Royal Society B: Biological Sciences 268: 2211–2220. Wilmé L, Goodman SM, Ganzhorn JU. 2006. Biogeographic evolution of Madagascar’s microendemic biota. Science 312: 1063– 1065 1741 Yoder AD, Nowak MD. 2006. Has vicariance or dispersal been the predominant biogeographic force in Madagascar? Only time will tell. Annual Review of Ecology and Systematics 37: 405– 431. Yuan YM, Wohlhauser S, Möller M, Klackenberg J, Callmander MW, Küpfer P. 2005. Phylogeny and biogeography of Exacum (Gentianaceae): a disjunctive distribution in the Indian Ocean Basin resulted from long distance dispersal and extensive radiation. Systematic Biology 54: 21– 34. AP P E N D I X 1. List of Rubiaceae and Gentianales taxa used in molecular dating analyses Species Outgroup Voucher information (kept at BR) and accession origin Van Caekenberghe 347 (Kenya) Van Caekenberghe 346 (Zimbabwe) Van Caekenberghe 14 (China) Rubiidinae Anthospermum palustre Homolle ex Puff Coprosma repens A.Rich. Danais sp. Galium mollugo L. Mycetia malayana (G.Don) Craib Paederia foetida L. Rubia fruticosa Aiton Serissa japonica (Thunb.) Thunb. Triainolepis sp. De Block 1922 (Madagascar) Van Caekenberghe 280 (New Zealand) De Block 2011 (Madagascar) Billiet 571 (Reunion) Van Caekenberghe 5 (Thailand) Van Caekenberghe 98 (China) Van Caekenberghe 256 (Spain) Van Caekenberghe 178 (China) De Block 1958 (Madagascar) Psychotriidinae Colletoecema magna Sonké & Dessein Coussarea hydrangeifolia (Benth.) Benth. & Hook.f. ex Müll.Arg Faramea trinervia K.Schum. & Donn.Sm. Gaertnera sp. nov. Geophila repens (L.) I.M.Johnst. Ophiorrhiza mungos L. Myrmecodia tuberosa Jack Psychotria kirkii Hiern Stelechantha makakana N.Hallé Antirhea borbonica J.F.Gmel. Breonadia salicina (Vahl) Hepper & J.R.I.Wood Chiococca alba (L.) Hitchc. Cinchona pubescens Vahl Cubanola domingensis (Britton) Aiello Exostema longiflorum (Lamb.) Schult. Guettarda uruguensis Cham. & Schltdl. Hamelia patens Jacq. Hoffmannia refulgens (Hook.) Hemsl. Hymenodictyon biafranum Hiern. Rondeletia odorata Jacq. Uncaria rhynchophylla (Miq.) Miq. ex Havil. Razafimandimbisonia humblotii (Drake) Kainul. & B.Bremer Aulacocalyx caudata (Hiern) Keay Calycophyllum spruceanum (Benth.) Hook.f. ex K.Schum. Canthium sp. Coffea mangoroensis Portères Coffea moratii J.-F.Leroy ex A.P.Davis & Rakotonas. Coffea stenophylla G.Don Coptosperma nigrescens Hook.f. Cremaspora triflora (Thonn.) K.Schum. Cuviera cf. leniochlamys K.Schum. Didymosalpinx norae (Swynn.) Keay Diplospora dubia (Lindl.) Masam. Emmenopterys henryi Oliv. Empogona concolor (N. Hallé) J.Tosh & Robbr. Empogona kirkii Hook.f. Empogona ovalifolia (Hiern) J.Tosh & Robbr. Euclinia longiflora Salisb. Fadogia sp. Fernelia buxifolia Lam. Gardenia jasminoides J.Ellis Genipa americana L. Ixora ankazobensis De Block sp. nov ined. Dessein 1608 (Cameroon) Sequences obtained from GenBank Sequences obtained from GenBank De Block 1795 (Madagascar) De Block 402 (Kenya) Van Caekenberghe 15 (China) Van Caekenberghe 343 (Borneo) Van Caekenberghe 25 (DRCongo) Dessein 1483 (Cameroon) De Block 2004 (Madagascar) De Block 1151 (Madagascar) Van Caekenberghe 80 (Costa Rica) De Block 932 (Madagascar) Van Caekenberghe 168 (Dominican Republic) Van Caekenberghe 95 (Dominican Republic) Sequences obtained from GenBank Van Caekenberghe 85 (Mexico) Van Caekenberghe 67 (Mexico) Van Caekenberghe 350 (Cameroon) De Block 1407 (Cuba) Van Caekenberghe 50 (Japan) Tosh 263 (Madagascar) Dessein 1510 (Cameroon) Van Caekenberghe 318 (Ecuador) De Block 691 (Madagascar) Rakotonasolo 41 (Madagascar) Davis 2326 (Madagascar) Billiet 3034 (DRCongo) Van Caekenberghe 52 (Madagascar) Van Caekenberghe 17 (Nigeria) Dessein 1448 (Cameroon) Van Caekenberghe 62 (Zimbabwe) Van Caekenberghe 49 (China) Van Caekenberghe 100 (cultivated at Kalmthout) Degreef 95 (Gabon) Van Caekenberghe 79 (Zimbabwe) De Block 1072 (Madagascar) Van Caekenberghe 348 (Sierra Leone) Van Caekenberghe 349 (Zambia) Sequences obtained from GenBank Van Caekenberghe 57 (Japan) Van Caekenberghe 317 (Ecuador) Tosh 30 (Madagascar) Ingroup Cinchonidinae Ixoridinae Continued Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Ceropegia linearis E.May Strychnos potatorum L.f. Luculia pinceana Hook. 1742 Tosh et al. — Evolutionary history of the Afro-Madagascan Ixora species (Rubiaceae) AP P E N D I X 1. Continued Species Dessein 1455 (Cameroon) Van Caekenberghe 42 (Mauritius) Walters 1437 (Gabon) Delprete (Brazil) Tosh 400 (Madagascar) Van Caekenberghe 316 (China) Mouly 236 (New Caledonia) Groeninckx 80 (Madagascar) De Block 1773 (Madagascar) De Block 1977 (Madagascar) De Block 1788 (Madagascar) Merello 1716 (Caribbean) Mouly 241 (New Caledonia) Tosh 408B (Madagascar) Tosh 107 (Madagascar) Dessein 1404 (Cameroon) Dessein 1449 (Cameroon) Friedmann 2631 (Mauritius) Tosh 232 (Madagascar) De Block 773 (Madagascar) Tosh 85 (Madagascar) Tosh 389 (Madagascar) Billiet 7327 (Malaysia) Luke 9304 (Tanzania) Van Caekenberghe 193 (Mozambique) Van Caekenberghe 44 (Mozambique) Van Caekenberghe 74 (China) Van Caekenberghe 198 (Cameroon) Van Caekenberghe 252 (Costa Rica) Dessein 1597 (Cameroon) Dessein 1612 (Cameroon) Sequences obtained from GenBank Van Caekenberghe 166 (Costa Rica) Dessein 1422 (Cameroon) Billiet 53054 (Ivory Coast) De Block 1409 (Ghana) Van Caekenberghe 279 (Gabon) Van Caekenberghe 344 (Venezuela) Van Caekenberghe 58 (Zambia) Van Caekenberghe 7 (Japan) De Block 1313 (Madagascar) Tosh 11 (Madagascar) Dessein 1283 (Zambia) Tosh 322 (Madagascar) Tosh 349 (Madagascar) Tosh 398 (Madagascar) De Block 405 (Kenya) Degreef 86 (Gabon) Van Caekenberghe 82 (Mozambique) Van Caekenberghe 212 (Cameroon) Van Caekenberghe 165 (Hawaii) All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately. Downloaded from http://aob.oxfordjournals.org/ at Stockholms Universitet on March 3, 2014 Ixora batesii Wernham Ixora borboniae Mouly & B.Bremer Ixora brachypoda DC. Ixora brevifolia Benth. Ixora capuroniana De Block Ixora chinensis Lam. Ixora collina (Montrouz.) Beauvis. Ixora crassipes Boivin ex De Block Ixora densithyrsa De Block Ixora elliotii Drake ex De Block Ixora emirnensis Baker Ixora ferrea (Jacq.) Benth. Ixora francii Schltr. Ixora guillotii Hoch. Ixora homolleae De Block & Govaerts Ixora macilenta De Block Ixora nematopoda K.Schum. Ixora nitens (Poir.) Mouly & B.Bremer Ixora perrieri De Block Ixora platythyrsa Baker Ixora quadrilocularis De Block Ixora siphonantha Oliv. Ixora sp. Ixora tanzaniensis Bridson Kraussia floribunda Harv. Mitriostigma axillare Hochst. Mussaenda pubescens Dryand. Oxyanthus unilocularis Hiern Pentagonia tinajita Seem. Petitiocodon parviflora (Keay) Robbr. Petitiocodon parviflora (Keay) Robbr. Pinckneya bracteata (Bartram) Raf. Posoqueria latifolia (Rudge) Schult. Pseudomussaenda sp. Psilanthus ebracteolatus Hiern Psilanthus mannii Hook.f. Rothmannia longiflora Salisb. Rosenbergiodendron formosum (Jacq.) Fagerl. Sabicea venosa Benth. Tarenna gracilipes (Hayata) Ohwi Tricalysia ambrensis Randriamb. & De Block Tricalysia analamazaotrensis Homolle ex Randriamb. & De Block Tricalysia coriacea (Benth.) Hiern Tricalysia cryptocalyx Baker Tricalysia dauphinensis Randriamb. & De Block Tricalysia leucocarpa (Baill.) Randriamb. & De Block Tricalysia microphylla Hiern Tricalysia pedunculosa (N.Hallé) Robbr. Vangueria madagascariensis J.F.Gmel. Virectaria procumbens (Sm.) Bremek. Warszewiczia coccinea (Vahl) Klotzsch Voucher information (kept at BR) and accession origin