This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy Molecular Phylogenetics and Evolution 49 (2008) 843–866 Contents lists available at ScienceDirect Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev The phylogenetic utility of chloroplast and nuclear DNA markers and the phylogeny of the Rubiaceae tribe Spermacoceae Jesper Kårehed a,*, Inge Groeninckx b, Steven Dessein c, Timothy J. Motley d, Birgitta Bremer a a Bergius Foundation, Royal Swedish Academy of Sciences and Botany Department, Stockholm University, SE-106 91 Stockholm, Sweden Laboratory of Plant Systematics, K.U. Leuven, Leuven, Belgium c National Botanic Garden of Belgium, Meise, Belgium d Old Dominion University, Department of Biological Sciences, Norfolk, VA, USA b a r t i c l e i n f o Article history: Received 31 March 2008 Revised 17 September 2008 Accepted 30 September 2008 Available online 10 October 2008 Keywords: 5S-NTS Accuracy atpB-rbcL Bayesian inference ETS ITS petD rps16 Phylogenetic utility Partition metric Phylogeny Rubiaceae Spermacoceae trnL-F a b s t r a c t The phylogenetic utility of chloroplast (atpB-rbcL, petD, rps16, trnL-F) and nuclear (ETS, ITS) DNA regions was investigated for the tribe Spermacoceae of the coffee family (Rubiaceae). ITS was, despite often raised cautions of its utility at higher taxonomic levels, shown to provide the highest number of parsimony informative characters, in partitioned Bayesian analyses it yielded the fewest trees in the 95% credible set, it resolved the highest proportion of well resolved clades, and was the most accurate region as measured by the partition metric and the proportion of correctly resolved clades (well supported clades retrieved from a combined analysis regarded as ‘‘true”). For Hedyotis, the nuclear 5S-NTS was shown to be potentially as useful as ITS, despite its shorter sequence length. The chloroplast region being the most phylogenetically informative was the petD group II intron. We also present a phylogeny of Spermacoceae based on a Bayesian analysis of the four chloroplast regions, ITS, and ETS combined. Spermacoceae are shown to be monophyletic. Clades supported by high posterior probabilities are discussed, especially in respect to the current generic classification. Notably, Oldenlandia is polyphyletic, the two subgenera of Kohautia are not sister taxa, and Hedyotis should be treated in a narrow sense to include only Asian species. Ó 2008 Elsevier Inc. All rights reserved. 1. Introduction When inferring phylogenies from molecular data it is important to use DNA regions with an evolutionary rate suitable for the taxa under study (Soltis and Soltis, 1998). A slowly evolving region may provide too little information to recover a fully resolved phylogeny. On the other hand, regions evolving too quickly will have their phylogenetic signal masked by homoplasy due to multiple hits, i.e., aligned nucleotides are the same by chance and not by common ancestry. In the present paper we will compare the phylogenetic utility of chloroplast and nuclear DNA sequences for resolving the phylogeny of the species-rich tribe Spermacoceae of the coffee family (Rubiaceae). The DNA regions (e.g., rbcL, atpB, 18S rDNA) utilized in the numerous studies synthesized in the high-level classifications of the Angiosperm Phylogeny Group (APG, 1998; APGII, * Corresponding author. Fax: +46 0 8 16 55 25. E-mail address: jesper@bergianska.se (J. Kårehed). 1055-7903/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2008.09.025 2003) would be of limited value in resolving the phylogeny of Spermacoceae, due to their conservative nature. For phylogenetic studies at lower taxonomic levels noncoding chloroplast regions have been used frequently and successfully. The rationale behind using noncoding regions is the assumption that they are phylogenetically more informative because they are under less functional constraints (Gielly and Taberlet, 1994). Shaw et al. (2005) investigated the relative utility of 21 noncoding chloroplast regions and divided them into three tiers. They found that five of the regions (rpoB-trnC, trnD-trnT, trnS-trnfM, trnS-trnG, trnT-L; tier 1) generally have enough phylogenetic signal for studies at the lower taxonomic levels. Five additional regions (psbM-trnD, rpl16, rps16, trnG, ycf6-psbM; tier 2) were identified as potentially useful, but less likely to provide full resolution. The remaining regions (3’trnK-matK, 5’rpS12rpL20, matK-5’trnK, psbA-3’trnK, psbB-psbH, rpS4-trnT, trnC-ycf6, trnH-psbA, trnL, trnL-trnF, trnS-rpS4; tier 3) were all considered to provide too little phylogenetic information to be recommended. More recently, Shaw et al. (2007) identified nine newly explored noncoding chloroplast regions (rpl32-trnL(UAG), trnQ(UUG)-5’rps16, Author's personal copy 844 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 3’trnV(UAC)-ndhC, ndhF-rpl32, psbD-trnT(GGU), psbJ-petA, 3’rps16– 5’trnK(UUU), atpI-atpH, and petL-psbE), which provided even higher levels of variation than the ones identified in their previous study (Shaw et al., 2005). Mort et al. (2007) further investigated the utility of the regions in tier 1 and 2 and compared them to the internal transcribed spacer (ITS) region (ITS1, 5.8S gene, and ITS2; White et al., 1990; Baldwin, 1992) of the nuclear ribosomal DNA region. As implied from its use in numerous studies, ITS is a useful marker for resolving phylogenetic relationships at various taxonomic levels, in particular infrageneric. When analysing ITS caution has to be taken to avoid problems resulting from concerted evolution on the ribosomal DNA arrays. Concerted evolution may homogenize different paralogous gene copies in a genome leading to the loss of all but one of the copies, i.e., different copies may be present in different organisms by chance and consequently the gene trees and species trees will not agree (Alvarez and Wendel, 2003). Compared to the noncoding chloroplast regions tested by Shaw et al. (2005), ITS was not consistently more informative (Mort et al., 2007). Mort et al. (2007) raised the concern of how well organismal phylogeny is inferred by ITS data. They concluded that, despite incongruences with chloroplast data (as measured by the ILD test; Farris et al., 1995), including ITS data may still result in an increase in both resolution and support suggesting that a phylogenetic signal masked by homoplasy in chloroplast data could be strengthened by a combined analysis (Nixon and Carpenter, 1996; Wenzel and Siddall, 1999; Mort et al., 2007). The studies by Shaw et al. (2005, 2007) and Mort et al. (2007) aimed at identifying molecular markers useful at lower taxonomic levels across a wide taxonomic range. We want to explore how useful different markers are for resolving the phylogeny of a large clade where the taxonomic sample is intended to infer both infrageneric and intrageneric relationships (this terminology is admittedly unfortunate since the current generic classification of Spermacoceae is known to include both para- and polyphyletic genera; Groeninckx et al., in press). We will identify the regions with the highest number of (parsimony) informative characters or the regions providing most resolution or the highest proportion of well resolved clades. We will also compare how effective different DNA regions are in obtaining the ‘‘true phylogeny.” To make these comparisons a region should provide potentially phylogenetic informative characters or produce well supported, resolved phylogenies. It must also provide accurate topologies, i.e., provide congruent results with other regions and provide data not susceptible to analytical assumptions, for example long branch attraction. Measures of accuracy often include comparing the number of bipartitions not shared between topologies. The partition metric, often called the Robinson–Foulds distance, is twice the number of internal branches that differ in the number of bipartitions between two topologies (Robinson and Foulds, 1981; Penny and Hendy, 1985; Steel and Penny, 1993). Rzhetsky and Nei (1992) adjusted the partition metric to account for multichotomies. Other methods used to measure how much deviation exists between two trees are the number of nearestneighbor exchanges needed to transform one tree to the other (Waterman and Smith, 1978), the number of correctly resolved taxon bipartitions divided by the number of possible bipartitions (Hillis, 1995), and comparisons of geometric distances in tree space between differing topologies (Billera et al., 2001). We have chosen Spermacoceae, a tribe of the coffee family (Rubiaceae), to investigate the utility of four chloroplast regions (atpB-rbcL, rps16, trnL-F, petD) and three nuclear regions (ITS, the external transcribed spacer, ETS, and the 5S non-transcribed spacer, 5S-NTS). Rubiaceae are currently classified in two or three subfamilies and more than 40 tribes (Bremer et al., 1995; Robbrecht and Manen, 2006; Bremer, in press). Spermacoceae with over 1000 species have a mainly pantropical distribution, but a few genera extend into temperate regions. The majority of the species are herbaceous. Four-merous flowers together with fimbriate stipules characterize the tribe, but these characters are not omnipresent. The delimitations of Spermacoceae have varied from a long recognized morphologically well-defined tribe (Spermacoceae sensu stricto, s.s.; Hooker, 1873; Bremekamp, 1952, 1966; Verdcourt, 1958; Robbrecht, 1988) to a wider interpretation stemming from molecular analyses and including the traditionally recognized tribes Hedyotideae, Knoxieae, Manettieae, and Triainolepideae (Bremer, 1996; Bremer and Manen, 2000). Here Spermacoceae are treated to comprise 60 genera and include Manettieae and most genera of Hedyotideae, but not their probable sister tribe Knoxieae, which recently is expanded to include Triainolepideae and a few genera from Hedyotideae (Andersson and Rova, 1999; Dessein, 2003; Kårehed and Bremer, 2007). The only previous phylogeny with a global perspective of Spermacoceae (Groeninckx et al., in press) revealed a number of well supported clades. The relationships between these clades are less resolved. In addition, the relationships within some of the clades are not strongly supported. The Spermacoceae sequence data, thus, provide an interesting mixture of early branching and late branching clades, clades with apparent different rates of evolution, as well as a combination of well supported clades and unresolved taxa. The study of Groeninckx et al. (in press) used only chloroplast data (atpB-rbcL, rps16, trnL-F). Since their study was intended for tribal relationships, the combination of relatively slower and faster markers (cf. Shaw et al., 2005) was rational. We chose to investigate the chloroplast region, petD (the petD intron and the petB-petD spacer; Löhne and Borsch, 2005), which according to Löhne and Borsch (2005) has less length variation than trnT-F, but has about the same number of informative characters. The petD region is comparatively untested in Rubiaceae. Although used less extensively than ITS, ETS seems to provide phylogenetic information at about the same magnitude as ITS (Baldwin and Markos, 1998; Musters et al., 1990). Within Rubiaceae, ETS and ITS have been utilized together at the generic level by Negrón-Ortiz and Watson (2002) for Erithalis, Markey and de Lange (2003) for Coprosma, Nepokroeff et al. (2003) for Psychotria, and Razafimandimbison et al. (2005) for Neonauclea. 5S-NTS (Cox et al., 1992; Sastri et al., 1992) has been even less used than ETS. For Rubiaceae, 5S-NTS was reported to be about twice as informative as ITS for the Alibertia group (Persson, 2000). It has also been used for Randia (Gustafsson and Persson, 2002) and most recently for inferring the phylogeny of the Kadua clade (Motley, in press), one of the well supported clades of Spermacoceae (Groeninckx et al., in press). The nuclear data were added as faster evolving DNA regions, which could provide better resolution of relationships within younger taxa or taxa with slow evolutionary rates, and strengthen phylogenetic signals masked by homoplasy in the chloroplast data. We also investigated the extent the nuclear data can resolve early branching clades of Spermacoceae, where previous chloroplast studies have failed (Groeninckx et al., in press). This will determine if the nuclear data only contribute information on closely related taxa, being too homoplastic at higher levels, or if the nuclear data also are useful for resolving relationships at somewhat higher taxonomic levels. Topological comparisons of chloroplast-trees to nuclear trees will provide evidence of suitable DNA regions to use for large, species-rich, tribal and/or family lineages. In our evaluations of the relative utility of chloroplast (atpBrbcL, rps16, trnL-F, petD) and nuclear (ITS, ETS, 5S-NTS) DNA regions we want to investigate if the eight well supported clades discussed by Groeninckx et al. (in press; Section 4.2) remain monophyletic with the additional data, examine if other clades retrieved in both studies have increased support with the addition of data, and Author's personal copy J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 determine if additional data will resolve additional monophyletic clades. In addition, we will present a total-evidence phylogeny of Spermacoceae and discuss to what extent the phylogeny agrees with the current generic classification. 2. Materials and methods 2.1. Taxon sampling and molecular data We used the chloroplast data set of Groeninckx et al. (in press) as the basis for our sampling. Their data set included 128 species. Our chloroplast data set was enlarged by the addition of three atpB-rbcL sequences (Gomphocalyx hernarioides, Hydrophylax maritima, and Pentanopsis gracilicaulis; the latter two species were not previously sampled), two rps16 sequences (Oldenlandia biflora, Pentanopsis gracilicaulis), four trnL-F sequences (Gomphocalyx hernarioides, Oldenlandia rosulata, Pentanopsis gracilicaulis, Phylohydrax carnosa), and petD sequences for the newly sampled Manettia luteorubra and 101 of the species included by Groeninckx et al. (in press; Appendix). Before sequencing nuclear data for the entire taxon sample, a test sample was selected. The test sample included 12 species (Arcytophyllum thymifolium, Dibrachionostylus kaessneri, Hedyotis quinquenervis, Kadua littoralis, Kohautia amatymbica, Kohautia virgata, Mitrasacmopsis quadrivalvis, Oldenlandia angolensis, Oldenlandia echinulosa, Pentodon pentandrus, Spermacoce hispida, and Thecorchus wauensis) representing the major clades recovered by Groeninckx et al. (in press). The sample was used to investigate if different nuclear regions could be amplified, and if the resulting sequences could be aligned to each other. The nuclear regions investigated were ITS, ETS, and 5S-NTS. Both the ITS and ETS sequences were straightforward to amplify and possible to align. Therefore, the two regions were chosen to be sequenced for the larger sample of Spermacoceae. In total we sampled 139 taxa for ITS and 98 for ETS (Appendix). New taxa added to the nuclear data but not present in the taxon sample of Groeninckx et al. (in press) include: two genera (Diodella and Psyllocarpus), 13 species (Bouvardia sp. Torres & Torino 3637 (BR), Galianthe sp. Persson & Gustavsson 298 (GB), Hedyotis capitellata, H. effusa, H. megalantha, Kohautia longifolia, K. cf. longifolia, Oldenlandia cf. galioides, O. monanthos, O. sp. C of Flora Zambesiaca, Richardia brasiliensis, Spermacoce filifolia, S. verticillata), and one subspecies Oldenlandia mitrasacmoides subsp. nigricans. Extra individuals for a few taxa already included are also added, either because those sequences were already available (the outgroup taxa) or, for example, to confirm the placement of a taxon (Amphiasma luzuloides, Conostomium natalense, Dibrachionostylus kaessneri, Kadua cordata, Oldenlandia corymbosa, O. goreensis, O. herbacea var. goetzei, O. herbacea var. herbacea, O. lancifolia, and Pentodon pentandrus). To be able to compare the phylogenetic utility of the different DNA regions without taking the effect of unequal taxon sampling into account, we also prepared a reduced taxon sample with those 49 taxa that are present in all separate data sets (referred herein as the reduced Spermacoceae data set). The 5S-NTS region was easy to amplify, all test taxa worked. The PCR (polymerase chain reaction) products resulting in the strongest bands (as measured by the amount of staining with Ethidium bromide when separated on an agarose gel) were, however, of quite different lengths. All the test taxa produced a single predominant band, although fainter bands occurred. These might be a result of unspecific annealing or due to multiple copies. The sequenced PCR products differed from 244 base pairs (bp; Spermacoce hispida) to 899 bp (Kohautia obtusiloba). Homology between the different bands was, thus, impossible to ascertain. Even the se- 845 quences of the bands of equal lengths, for which homology could be postulated, were not possible to align. Since 5S-NTS seemed to be easy to amplify for taxa of the Spermacoceae, but was too variable to be used for the entire tribe, we wanted instead to test its usefulness for more restricted, well defined clades. For that purpose we chose Hedyotis s.s. (Groeninckx et al, in press). 5S-NTS was sequenced for nine Hedyotis species present in the ITS and ETS data sets. This matrix (the Hedyotis data set) was used to determine how informative and accurate the 5SNTS region is compared to the other two nuclear data sets (see below). For the Hedyotis data set we made no attempts to cut out the predominant band or clone the PCR product, because the obtained sequences showed no indications of multiple copies (e.g., multiple peaks in the chromatograms). The low copy-number nuclear marker Tpi (Strand et al., 1997) was also investigated. We were not able to obtain PCR products for any of the 12 test taxa. Nevertheless, the use of this and other low copy-number nuclear markers with specifically designed primers may prove to be valuable in the future for exploring clades that remain unresolved with other DNA regions. 2.2. Sequencing DNA was extracted from fresh, silica-gel dried material or herbarium specimens using the CTAB method (Doyle and Doyle, 1987). The petD region was amplified with the forward primer PIpetB1365F and the reverse primer PIpetD738R (Löhne and Borsch, 2005). Polymerase chain reactions and the sequencing of the new chloroplast data were performed as described by Groeninckx et al. (in press). Polymerase chain reactions for the nuclear data sets were run on an EppendorfÒ MastercyclerÒ gradient (Bergman & Beving Instrument, Stockholm, Sweden). The 50-ll reactions included 5 ll reaction buffer, 5 ll MgCl2, 5 ll TMACL (Chevet et al., 1995), 4 ll DNTP, 0.25 ll Taq (5 U/ll), 0.5 ll 50 primer (20 lM), 0.5 ll 30 primer (20 lM), 0.5 ll BSA 1%, and 1–2 ll of DNA templates and sterilized H2O adding up to 50 ll. The amplifications consisted of an initial denaturation for 1 min at 95 °C, followed by 37 cycles of 1 min at 95 °C, 1 min 30 s at 50 °C, 1 min 30 s at 72 °C (usually +1 s/cycle), and a final extension phase of 7 min at 72 °C. The PCR products were purified with the MultiScreenÒ Separations System (Millipore, USA). The purified products were subsequently sequenced with the BigDyeTM terminator cycle sequencing kit (Applied Biosystems, Stockholm, Sweden) on a GeneAmp PCR System 9700 (Applied Biosystems) and analyzed on an ABI PRISMÒ 3100 Genetic Analyzer (Applied Biosystems). Primers used for both PCR and sequence reactions were for ITS the forward primer P17 and the reverse primer 26S-82R (Popp and Oxelman, 2001) or P25 (Oxelman, 1996). For ETS the primer pair ETS-Erit-F (Negrón-Ortiz and Watson, 2002) and 18S-E (Baldwin and Markos, 1998) was used. The 5S-NTS region was amplified using the primers PI and PII (Cox et al., 1992) and attempts to amplify Tpi were performed with primers TPIX4FN and TPIX6RN (Strand et al., 1997). 2.3. Phylogenetic analyses All alignments were made by eye. Simple indel coding (Simmons and Ochoterena, 2000) of insertion/deletion (indel) events was performed with the computer program SeqState (Müller, 2005). The indels were subsequently included in the analyses. Bayesian analyses to estimate the phylogeny of the entire Spermacoceae data set and the reduced Spermacoceae data set were done for all regions independently, the combined four chloroplast regions, the combined nuclear ITS and ETS regions, and a total evidence analysis. For the Hedyotis taxon sample the ITS, ETS, and 5SNTS data were analyzed both separately and combined. Author's personal copy 846 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 The choice of nucleotide-substitution models for the Bayesian analyses of the molecular data was determined based on the corrected Akaike criterion (Burnham and Anderson, 2002) calculated using the computer program MrAIC (Nylander, 2004) in conjunction with the program PhyML (Guindon and Gascuel, 2003). The general time reversible model with a gamma distribution of substitution rates (GTR + G) was chosen for all the DNA regions of the Spermacoceae data set, except ETS. For ETS the model (GTR + I + G) includes a proportion of invariant sites. In the reduced Spermacoceae data set the GTR + G model was used for all regions. The Hedyotis data set was analysed under the GTR + G model for ETS and 5S-NTS, and the HKY model (with different substitution rates for transitions and transversions) was used for ITS. The insertion/ deletion data for all data sets were analyzed under the standard discrete (morphology) model. In the combined analysis the data set was partitioned and the partitions were unlinked and, consequently, had their own set of parameters. The Bayesian analyses were performed using the computer program MRBAYES (v3.1.2; Huelsenbeck and Ronquist, 2001; Ronquist and Huelsenbeck, 2003). For the Spermacoceae data set the Markov chain was run for 5,000,000 generations with three additional ‘‘heated” chains (Metropolis-coupled Markov chain Monte Carlo; Huelsenbeck et al., 2001). Trees were sampled every 100th generation. Two separate runs were run for all data sets. When the average standard deviation of split frequencies between the separate runs was <0.05, it was taken as an indication that the Markov Chains had converged on the stationary distribution. The first 500,000 generations were, consequently, discarded as a burn-in period for all analyses except for the combined nuclear data and for the combined chloroplast and nuclear data, for which a burn-in period of 1,000,000 generations was used. The reduced Spermacoceae and the Hedyotis data sets were run for 1,000,000 generations with a burn-in period of 100,000 generations and every 100th tree was sampled. 2.4. Comparison of phylogenetic utility of the seven DNA regions In order to investigate the phylogenetic utility of the seven DNA regions we compared the information content, i.e., how much of the information that could be contributed to sequence length, the number of trees in the 95% credible set of trees, the proportion of well resolved nodes, and the phylogenetic accuracy of the respective regions. The number of parsimony informative characters excluding and including indels are shown in Table 1 and Fig. 1(a–c). Linear regressions of the number of parsimony informative characters against the aligned length of the data matrices are shown in Fig. 1(d–f). We also present the 95% credible set of trees (the smallest set of trees whose cumulative posterior probability sum to 0.95; Felsenstein, 1968; Table 1, Fig. 1g–i). The 95% credible set gives an indication of the precision of the maximum posterior probability estimate of the phylogeny (the most probable tree). A DNA region which produces a 95% credible set including few trees would indicate that the region better discriminates between conflicting topologies than another region which produces a larger 95% credible set. The size of the 95% credible set could, thus, be seen as an approximation of the strength of the phylogenetic signal in a data set (with more phylogenetically informative data the likelihood function can update the prior distribution more strongly). In Table 2 and Fig. 2(a–c) the proportion of clades resolved with PP 0.95 divided by the number of possible bipartitions (i.e., number of bipartitions in a fully dichotomous tree, which equals the number of terminal taxa minus three) are shown. We choose to compare only clades with PP 0.95, since these clades could be considered ‘‘true” given the data and the models used in the Bayesian analyses (Huelsenbeck et al., 2002). As a measure of phylogenetic accuracy we used the partition metric (PM; Penny and Hendy, 1985) modified to account for multichotomies (Rzhetsky and Nei, 1992). The partition metric ranges from zero (identical phylogenies) to twice the number of possible clades resolved (number of taxa minus three). We only compared phylogenetic accuracy for the reduced Spermacoceae data set and the Hedyotis data set. Table 2 and Fig. 2(d and e) show the partition metric for the phylogenetic tree from the Bayesian analyses of each of the DNA regions compared to the phylogenetic tree from the Bayesian analysis of the combined data set, the latter taken as a reasonable estimation of the ‘‘true” phylogeny. When calculating the partition metric only nodes with PP 0.95 were considered. The calculations were made using the function dist.topo in the R package APE (Paradis et al., 2004; R Development Core Team, 2007). As another approximation of accuracy, we calculated the proportion of the number of clades with PP 0.95 in the phylogenetic tree from the separate analysis of a given marker, which also has a PP 0.95 in the tree from the combined analysis, divided by the total number of clades with PP 0.95 in the combined tree (i.e., ‘‘correctly” resolved clades assuming the combined tree to be the ‘‘true” tree; Table 2; Fig. 2f and g). Table 1 Number of taxa, aligned length of the data matrices, number of parsimony informative characters, and the number of trees in the 95% credible set for the different data sets and DNA regions. N, number of taxa; Al. l., aligned length; Pars. inf. – indels, number of parsimony informative characters excluding indel characters; Pars. inf. indels, number of parsimony informative indel characters,% Pars. inf., percentage of the total number of aligned characters that are parsimony informative. Data set N Al. l. (bp) Pars. inf. - indels Pars. inf. indels % Pars. inf. No. of trees in 95% cred. set Spermacoceae atpB-rbcL rps16 trnL-F petD ITS ETS 103 109 115 102 139 98 1,366 715 1,003 1,799 2,173 1,055 142 153 138 276 452 329 46 26 62 90 308 100 13.8% 25.0% 19.9% 20.3% 35.0% 40.7% 85,420 85,373 85,486 79,697 60,098 66,519 Reduced Spermacoceae atpB-rbcL 49 rps16 49 trnL-F 49 petD 49 ITS 49 ETS 49 1,366 715 1,003 1,799 2,173 1,055 71 102 88 163 294 253 19 10 17 42 181 76 6.5% 15.7% 10.5% 11.4% 21.9% 31.2% 17,001 17,020 17,066 14,812 7,670 12,410 Hedyotis ITS ETS 5S-NTS 424 799 334 33 27 55 1 9 14 8.0% 4.5% 20.7% 3 19 9 9 9 9 Author's personal copy 847 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 a d 800 g 90000 700 80000 600 Spermacoceae 70000 500 60000 50000 400 40000 300 30000 200 20000 100 10000 0 0 atpB-rbcL rps16 trnL-F petD No. of parsimony informa tive characters excluding indels b ITS ETS atpB-rbcL e 500 Reduced Spermacoceae trnL-F petD ITS ETS ITS ETS No. of trees in 95 % cred. set h 450 18000 400 16000 14000 350 12000 300 10000 250 8000 200 6000 150 4000 100 2000 50 0 atpB-rbcL 0 atpB-rbcL rps16 trnL-F petD No. of parsimony informative characters excluding indels c rps16 No. of parsimony informa tive indels ITS ETS rps16 trnL-F petD No. of trees in 95 % cred. set No. of parsimony informative indels f 80 i 20 70 18 60 16 14 Hedyotis 50 12 40 10 8 30 6 20 4 2 10 0 0 ITS ETS 5S-NTS ITS ETS No. of trees in 95 % No. of parsimony informative characters excluding indels 5S-NTS cred. set No. of parsimony informative indels Fig. 1. (a–c) Number of parsimony informative characters in: (a) each of the six regions included in the analysis of the Spermacoceae data set, (b) the three nuclear regions included in the analysis of the reduced Spermacoceae data set, (c) the three nuclear regions included in the analysis of the Hedyotis data set. (d–f) Linear regression of number of parsimony informative characters (including indels) against aligned length (bp) of: (d) the separate DNA regions of the Spermacoceae data set (black circles = chloroplast regions, white circles = nuclear regions, dashed line = regression line of the chloroplast data, r2 = 0.70, black line = regression line of all data combined, r2 = 0.61), (e) the separate DNA regions of the reduced Spermacoceae data set (black circles = chloroplast regions, white circles = nuclear regions, dashed line = regression line of the chloroplast data, r2 = 0.43, black line = regression line of all data combined, r2 = 0.51), (f) the Hedyotis data set (r2 = 0.37). (g–i) Number of trees in the 95% credible set for: (g) the Spermacoceae data set, (h) the reduced Spermacoceae data set, (i) the Hedyotis data set. 3. Results 3.1. Phylogenetic utility In the analysis of the Spermacoceae data the nuclear regions provide most information (Table 1, Fig. 1a). ITS is the most informative region and provides 1.8 times as many parsimony informative characters as ETS and is more than twice as informative as petD, the most informative chloroplast region. The petD region is in turn about twice as informative as the other chloroplast regions. Linear regressions of the number of parsimony informative characters (including indels) against the lengths of the separate matrices for both the chloroplast data separately (r2 = 0.70, p = 0.161) and in combination with the nuclear data (r2 = 0.61, p = 0.067) are shown in Fig. 1d. When the nuclear data are added, less of the information content can be explained by just sequence length, i.e., the nuclear data are indicated to be more informative than the chloroplast data for a given sequence length. A similar pattern is retrieved when comparing the data from the reduced Spermacoceae data set (Table 1, Fig. 1b). ITS provides 1.4 times as many parsimony informative characters as ETS and 2.3 times as many as petD, the latter region being about twice as informative as the other chloroplast data. The linear regressions of parsimony informative characters relative to aligned length of the chloroplast and combined data for the reduced Spermacoceae data set are shown in Fig. 1e. The regressions indicate for this data set, i.e., with reduced taxon sampling and no missing data, that the length of the sequences explains less of the information content (r2 = 0.50, p = 0.289 for the chloroplast and r2 = 0.43, p = 0.155 for the combined data), than for the entire Spermacoceae data set. For the Hedyotis data ETS and ITS provide about the same number of informative characters, but the proportion of informative in- Author's personal copy 848 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Table 2 Proportion of clades with posterior probability  0.95 (PP  0.95) divided by the number of possible bipartitions for the different data sets and DNA regions, the partition metric (PM), and proportion of correctly resolved clades for the reduced Spermacoceae and the Hedyotis data sets. Data set Clades with PP  0.95/ possible bipartitions PM (PP  0.95) Correctly resolved clades 36.0% 45.3% 39.3% 61.6% 65.7% 63.2% n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. Reduced Spermacoceae atpB-rbcL 37.0% rps16 47.8% trnL-F 41.3% petD 63.0% ITS 69.6% ETS 58.7% 21 20 19 15 12 14 44.7% 55.3% 52.6% 71.1% 81.6% 71.1% Hedyotis ITS ETS 5S-NTS 2 4 1 40.0% 80.0% 80.0% Spermacoceae atpB-rbcL rps16 trnL-F petD ITS ETS 83.3% 50.0% 66.7% dels is larger for ITS. The 5S-NTS region provides the most parsimony informative characters (Table 1, Fig. 1c). It is about twice as informative as the other two nuclear regions. A linear regression of parsimony informative characters against the length of the matrices is shown in Fig. 1f (r2 = 0.37, p = 0.582). Since the shortest marker is the most informative, there is a negative correlation between parsimony informative characters and sequence length for this data set. For all data sets it seems that there is a variation in the information content among the DNA regions, more variability than can be explained by sequence length alone. In particular, the nuclear data are more informative than the chloroplast regions and the 5S-NTS is the most informative nuclear region despite its short length. Examination of DNA region variability using the number of trees in the 95% credible sets (Table 1, Fig. 1g–i) from the separate analyses of the Spermacoceae and the reduced Spermacoceae data sets give nearly the same picture as the number of parsimony informative characters: ITS had the fewest trees, followed by ETS and petD. The other three chloroplast regions had about the same number of trees. For the Hedyotis data set the ITS data, not the 5S-NTS, had the fewest number of trees in the 95% credible set. Additionally, the number of well resolved nodes in the resulting phylogenies can be an indication of the phylogenetic utility of a DNA region. Here we consider a node to be well resolved if the posterior probability of that node is 0.95. To account for the different sized taxon samples in the Spermacoceae data set, we compared the number of nodes with a posterior probability 0.95 divided by the maximum number of internal nodes in an unrooted, fully dichotomous tree (Table 2; Fig. 2). For the Spermacoceae data, the ITS region resolved the highest proportion of well resolved nodes (66%; 70% for the reduced data set). The petD region performed comparatively better in this evaluation and resolved almost as many nodes as ETS (62% and 63%, respectively). In the reduced Spermacoceae data set, petD resolved more nodes than ETS (63% vs. 59%). The other three chloroplast regions as predicted retrieved less nodes (36% for atpB-rbcL to 45% for rps16; 37% to 48% for the reduced data set). For the Hedyotis data set ITS resolved the highest number of nodes (83%) in spite of the fact that the 5S-NTS provided the highest number of informative characters. The most accurate region for the reduced Spermacoceae data set (Table 2; Fig. 2d) was ITS (PM = 12) followed by ETS (PM = 14), and the petD region (PM = 15). For the Hedyotis data set (Table 2; Fig. 2e), 5S-NTS was more accurate than ITS (PM = 1 vs. PM = 2, respectively), which was in turn more accurate than ETS (PM = 4). When comparing the number of correctly resolved clades, the same picture emerges (Table 2; Fig. 2f and g). ITS correctly resolved the highest proportion of nodes for the reduced Spermacoceae data set and is followed by ETS and petD, which performed equally well. For the Hedyotis data set both ITS and 5S-NTS resolved the same proportion of the clades supported in the combined analysis. 3.2. Phylogenetic analyses Spermacoceae are monophyletic in all analyses and all independent and combined analyses retrieved a number of well supported clades (Fig. 3). The separate analyses were naturally less resolved than the combined. The topologies from the separate analyses, even if less resolved, agree with the topology of the combined analysis for all clades discussed here. No well supported relationships in any of the separate analyses are contradicted by the other data. The separate analyses of the Spermacoceae data will, thus, not be discussed in detail. Neither will the phylogenies resulting from the analyses of the reduced Spermacoceae data set, since these were merely performed to enable us to compare the number of parsimony informative characters and to calculate the partition metric on congruent data sets. We regard the combined analysis as the best estimate of the phylogeny (total evidence) of Spermacoceae and, if not otherwise indicated, it is the result on which we will focus our discussions (Fig. 3). We also present both the combined and separate analyses of the Hedyotis data set (Fig. 4), because the 5S-NTS was not included in the combined analysis of the Spermacoceae data set. Adding the petD data to the previously published chloroplast data (Groeninckx et al., in press) increased the proportion of well resolved nodes from 52% to 56%. The addition of nuclear data further increased the proportion to 70%. In fact the nuclear data by themselves resolved 67% of the possible nodes. The combined analyses showed a marked decrease in the number of trees in the 95% credible set compared to the separate analysis (33,152 for the Spermacoceae data set compared to 60,098 for ITS). In the reduced Spermacoceae data the combined analyses yielded 553 credible trees as compared to 7670 for ITS. The combined Hedyotis data set produced three trees as did the ITS analysis. Apart from the difference in resolution between the DNA regions, we do not see any bias for any of the regions towards better resolving either early or late branching taxa. 4. Discussion 4.1. Phylogenetic utility Inferring the phylogeny of a group naturally implies finding one or several gene regions which are preferably single copy and are informative. In a phylogenetic study the actual number of potentially informative characters are of interest, i.e., a longer but less variable region is of more interest than a highly variable but short region, which despite a high information content per sequenced base pair may not provide enough characters to resolve the phylogeny. The 21 noncoding chloroplast regions studied by Shaw et al. (2005) varied in the extent (22–83%) that length per se could explain the variation in number of potentially informative characters. Mort et al. (2007) found a much lower correlation between length and information content (5.2% for chloroplast data and 0.9% for chloroplast plus ITS data), but since they included only rapidly evolving regions this result is not unanticipated. The linear regres- Author's personal copy 849 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 a 0.70 Spermacoceae 0.60 0.50 0.40 0.30 0.20 0.10 0.00 atpB-rbcL rps16 trnL-F petD Proportion of clades b ITS ETS 0.95/possible bipartitions f d 0.80 Reduced Spermacoceae 25 0.70 0.60 0.60 20 0.50 0.50 0.40 15 0.40 0.30 10 0.30 0.20 0.20 5 0.10 0.10 0 0.00 atpB-rbcL rps16 trnL-F petD Proportion of clades c ITS 0.00 atpB-rbcL ETS rps16 trnL-F petD ETS atpB-rbcL rps16 trnL-F petD ITS ETS Proportion correctly resolved clades e 0.90 ITS PM 0.95/possible bipartitions g 0.80 0.9 4.5 0.70 0.8 4 0.7 0.60 Hedyotis 0.90 0.80 0.70 3.5 0.50 3 0.40 2.5 0.6 0.5 0.4 2 0.30 1.5 0.3 1 0.2 0.10 0.5 0.1 0.00 0 0.20 ITS ETS 5S-NTS 0 ITS ETS PM Proportion of clades 0.95/possible bipartitions 5S-NTS ITS ETS 5S-NTS Proportion correctly resolved clades Fig. 2. (a–c) Proportion of the possible number of clades with a PP  0.95 for: (a) the Spermacoceae data set, (b) the reduced Spermacoceae data set, and (c) the Hedyotis data set. (d and e) The partition metric (PH-85) for: (d) the reduced Spermacoceae data set, (e) the reduced Hedyotis data set. (f and g) Proportion of correctly resolved clades in: (f) the reduced Spermacoceae data set, (g) the Hedyotis data set. sion of the chloroplast regions of our Spermacoceae data (Fig. 1) showed that more than two-third of the parsimony informative variation could be explained by the length of the sequences (r2 = 0.70, p = 0.161). With the inclusion of the nuclear data the correlation decreased (Fig. 1d; r2 = 0.61, p = 0.067). The greater number of informative characters provided by the nuclear data are, thus, indicative of the nuclear data being more informative than the chloroplast data and not merely an effect of sequence length. For the nuclear regions examined using the Hedyotis data set, where the shortest region provided the highest number of informative characters, the linear regression naturally shows a negative correlation (Fig. 1f; r2 = 0.37, p = 0.582). These results indicate that nuclear data generally can be expected to provide more information than chloroplast data (Fig. 1). The highly variable chloroplast regions identified by Shaw et al. (2005, 2007) should be further explored, especially in cases where gene trees are suspected to differ from the species tree (Alvarez and Wendel, 2003). However, for Spermacoceae the chloroplast and nuclear phylogenies are essentially identical (only the position of taxa with low posterior probabilities differ, and this is here interpreted as a lack of phylogenetic signal and not due to incongruent data sets; see Section 4.2. for details of the Spermacoceae phylogeny). When inves- tigating more rapidly evolving chloroplast regions with ITS, Mort et al. (2007) did not find ITS to be either universally the most informative region or the region providing the highest clade support, although it generally was one of the best choices for resolving their low taxonomic level phylogenies. It is noteworthy that the nuclear data not only resolve nodes between closely related taxa, but also support some of the early branching events in Spermacoceae. Finding relatively fast evolving DNA regions useful also at higher taxonomic levels (less homoplastic than is often assumed) has also been shown previously in the Asterids (Bremer et al., 2002), Rubiaceae (Motley et al., 2005), and basal angiosperms (Müller et al., 2006). In order to obtain a well supported phylogeny of a group of organisms it is, however, not enough to analyze a high number of potentially phylogenetic informative characters. The included characters should naturally reflect our best estimate of the true phylogeny. According to our results, the regions providing the most information are in fact also the most accurate. The proportion of well resolved nodes is generally also a good approximation, but with two exceptions. Apparently, the number of parsimony informative characters and the number of trees in the 95% credible set actually give a valuable insight into the accuracy of a DNA region, i.e., the Author's personal copy 850 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 a 1 1 1 1 1 clade A 1 .89 1 1 1 1 1 1 1 1 1 1 1 1 .96 1 1 1 1 1 1 1 .91 1 1 1 1 1 1 1 .97 .53 1 1 .99 clade B .87 .83 1 .99 1 .53 1 1 clade G 1 .89 1 clade D 1 1 1 .53 1 .98 .69 1 1 clade E .96 1 .52 .60 1 .79 .56 .59 .61 Arcytophyllum-Houstonia clade F Agathisanthemum-Hedyotis s.s. clade clade C Pentanopsis clade 1 .94 Kohautia 1 .98 Batopedina pulvinellata Outgroup Carphalea madagascariensis Pentanisia parviflora Kohautia cynanchica Kohautia subverticillata Kohautia longifolia Kohautia amatymbica Kohautia senegalensis Kohautia coccinea Kohautia cf. longifolia Kohautia caespitosa Oldenlandia rosulata Manostachya ternifolia Gomphocalyx herniarioides Phylohydrax madagascariensis Phylohydrax carnosa Oldenlandia affinis Pentanopsis gracicaulis Pentanopsis fragrans Amphiasma benguellense Amphiasma luzuloides Tanzania Amphiasma luzuloides Zambia Oldenlandia herbacea var. herbacea Dessein et al. 1041 Oldenlandia herbacea var. herbacea Dessein et al. 463 Oldenlandia herbacea var. goetzei Dessein et al. 1218 Oldenlandia herbacea var. goetzei Dessein et al. 442 Conostomium zoutpansbergense Conostomium quadrangulare Conostomium natalense Dahlstrand 1346 Conostomium natalense Bremer et al. 4341 Pentodon pentandrus Zambia Pentodon pentandrus Zanzibar Dentella repens Dentella dioeca Agathisanthemum globosum Lelya osteocarpa Agathisanthemum bojeri Oldenlandia uniflora Oldenlandia goreensis Dessein et al. 455 Oldenlandia angolensis Oldenlandia goreensis Richards & Arasululu 25910 Oldenlandia goreensis Dessein et al. 1335 Oldenlandia goreensis Dessein et al. 1286 Hedyotis macrostegia Hedyotis effusa Hedyotis consanguinea Hedyotis megalantha Hedyotis korrorensis Hedyotis swertioides Hedyotis rhinophylla Hedyotis fruticosa Hedyotis lawsoniae Hedyotis quinquenervis Hedyotis lessertiana var. marginata Hedyotis lessertiana var. lessertiana Dibrachionostylus kaessneri Strid 2598 Dibrachionostylus kaessneri Strid 2564 Oldenlandia fastigiata Mitrasacmopsis quadrivalvis Hedythyrsus spermacocinus Oldenlandia nervosa Oldenlandia geophila Oldenlandia echinulosa var. pellucida Oldenlandia echinulosa Oldenlandia microtheca Stenaria nigricans Houstonia longifolia Houstonia caerulea Arcytophyllum thymifolium Arcytophyllum setosum Arcytophyllum nitidum Arcytophyllum lavarum Arcytophyllum ericoides Arcytophyllum rivetii Arcytophyllum ciliolatum Arcytophyllum macbridei Arcytophyllum muticum Arcytophyllum aristatum FIGURE 3B Fig. 3. (a and b) Phylogenetic tree of the combined analysis with posterior probabilities indicated below the branches. more potentially phylogenetic informative characters a region provides the better one would expect it to estimate the correct species phylogeny. For data sets such as our Spermacoceae tribe data, with a wide range of taxa representing different taxonomic levels, it seems that one does not have to be too concerned for nuclear data to be either Author's personal copy 851 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 b .54 1 clade H 1 1 .85 .75 1 1 .77 1 1 1 1 1 1 .95 1 .56 .66 1 1 1 1 1 1 1 1 clade I 1 .91 1 1 .98 .52 .59 1 1 .59 1 1 .73 .97 1 .74 .90 clade J 1 1 .56 1 1 .51 1 1 .90 1 K .96 .85 .96 1 1 .61 .56 .69 .83 .97 .51 1 L .96 1 .94 .84 .55 Spermacoce clade 1 Pachystigma Oldenlandia s.s. 1 1 Kadua 1 Hedyotis capitellata Oldenlandia mitrasacmoides ssp. trachymenoides Oldenlandia mitrasacmoides ssp. nigricans Synaptantha tillaeacea Oldenlandia tenelliflora Oldenlandia cf. galioides Oldenlandia galioides Oldenlandia lancifolia Dessein et al. 1356 Oldenlandia lancifolia Dessein et al. 1256 Oldenlandia biflora Kadua affinis Kadua fosbergii Kadua axillaris Kadua fluviatilis Kadua acuminata Kadua rapensis Kadua coriacea Kadua foggiana Kadua centranthoides Kadua parvula Kadua degeneri Kadua littoralis Kadua laxiflora Kadua cordata Kadua flynnii Kadua elatior Kohautia obtusiloba Kohautia virgata Kohautia microcala Dessein et al. 1149 Oldenlandia wiedemannii Oldenlandia corymbosa Gabon Oldenlandia corymbosa Australia Oldenlandia monanthos Oldenlandia wauensis Oldenlandia corymbosa Zambia Oldenlandia sp. C Fl. Zamb. Oldenlandia taborensis Oldenlandia nematocaulis Oldenlandia densa Oldenlandia capensis var. pleiosepala Oldenlandia capensis var. capensis Manettia luteo-rubra Manettia lygistum Manettia alba Arcytophyllum serpyllaceum Bouvardia sp. Bouvardia ternifolia Bouvardia glaberrima Nesohedyotis arborea Oldenlandia tenuis Oldenlandia salzmannii Galianthe sp. Galianthe eupatorioides Galianthe brasiliensis Diodia spicata Emmeorhiza umbellata Crusea megalocarpa Crusea calocephala Spermacoce flagelliformis Spermacoce prostrata Spermacoce erosa Richardia stellaris Richardia brasiliensis Richardia scabra Spermacoce confusa Spermacoce verticillata cult. Spermacoce verticillata Madagascar Spermacoce capitata Spermacoce remota Spermacoce ocymifolia French Guiana Spermacoce ocymifolia Ecuador Mitracarpus microspermus Mitracarpus frigidus Psyllocarpus laricoides Diodella teres Ernodea littoralis Spermacoce ruelliae Spermacoce hispida Spermacoce filituba Spermacoce filifolia Hydrophylax maritima Diodia sarmentosa Diodia aulacosperma Fig. 3 (continued) too homoplastic to provide a phylogenetic signal or to be indicative of other gene trees not congruent with the chloroplast tree. Both ITS and ETS, turn out to be the most useful regions. Perhaps with a denser sample within some of the smaller well-defined clades caution must be taken to caveats of nuclear data in general, and ITS and ETS in particular (see, e.g., Alvarez and Wendel, 2003). Author's personal copy 852 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 ETS Hedyotis effusa Hedyotis megalantha 1.00 combined Hedyotis korrorensis Hedyotis rhinophylla 0.90 Hedyotis fruticosa Hedyotis effusa 1.00 Hedyotis lawsoniae Hedyotis lessertiana var. marginata 0.95 Hedyotis lessertiana var. lessertiana 0.94 Hedyotis megalantha Hedyotis quinquenervis 1.00 Hedyotis korrorensis ITS Hedyotis effusa Hedyotis korrorensis Hedyotis rhinophylla 1.00 Hedyotis megalantha Hedyotis rhinophylla 1.00 0.82 Hedyotis lawsoniae Hedyotis fruticosa 1.00 Hedyotis fruticosa Hedyotis lessertiana var. lessertiana 0.99 0.99 1.00 0.75 Hedyotis quinquenervis 1.00 Hedyotis lawsoniae Hedyotis lessertiana var. marginata Hedyotis lessertiana var. lessertiana 5S-NTS Hedyotis effusa 1.00 Hedyotis korrorensis Hedyotis quinquenervis 1.00 Hedyotis megalantha Hedyotis lessertiana var. marginata 0.95 Hedyotis quinquenervis 1.00 Hedyotis lessertiana var. marginata Hedyotis lessertiana var. lessertiana 0.95 Hedyotis lawsoniae 0.58 Hedyotis rhinophylla 1.00 Hedyotis fruticosa Fig. 4. Phylogenetic trees from the combined and separate analyses of the Hedyotis data set with posterior probabilities indicated below the branches. For example hybridization events between closely related taxa might become an important issue to account for when studying lower level relationships of particular clades. The highly variable 5S-NTS region seems, however, promising for such studies. This region is very informative, but should preferably best be used in combination with some of the fast evolving chloroplast regions recently identified (Shaw et al. 2005, 2007; Mort et al., 2007) to ascertain that the information it provides does infer the species phylogeny. 4.2. Phylogeny of Spermacoceae Our analysis combining nuclear and chloroplast data (Fig. 3) agrees strongly with the chloroplast phylogeny presented by Groeninckx et al. (in press). For example, Spermacoceae are monophyletic, Hedyotis and Oldenlandia are polyphyletic, and Kohautia is biphyletic. Some genera are paraphyletic (e.g., Agathisanthemum and Spermacoce) and others are monophyletic (e.g., Dentella and Kadua) in both studies. Groeninckx et al. (in press) specifically discussed eight well supported clades: Kohautia, the Pentanopsis clade, the Agathisanthemum–Hedyotis s.s. clade, Kadua, the Arcytophyllum-Houstonia clade, Oldenlandia s.s., Pachystigma, and Spermacoceae s.s. These clades are also retrieved in this study. We will refer to Spermacoceae s.s. as the Spermacoce clade, since we do not want to imply that the clade could be recognized at tribal level. If such a tribe is acknowledged several other clades would also have to be recognized as tribes and we currently do not have the appropriate knowledge to delimit all such taxa. Furthermore, we also question the value of a split of Spermacoceae into smaller entities, especially considering that many taxa outside the Spermacoce clade often have been treated as congeneric (see below). Our analyses provide several new insights into the phylogeny of Spermacoceae. Spermacoceae in the combined analysis are resolved as a basal dichotomy. Clade A (PP 1.00) constitutes Kohautia + the Pentanopsis clade and includes Conostomium, which is retrieved as monophyletic in contrast to previous results of Author's personal copy J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Groeninckx et al. (in press). Clade B was not recovered in the study by Groeninckx et al. (in press), but it is not supported (PP 0.83). Clade B constitutes two clades, clades C (PP 0.91) and D (PP 1.00). Clade C contains two subclades, Pentodon + Dentella and their sister group, the Agathisanthemum–Hedyotis s.s. clade. This sister relationship is a novel finding. In the analyses by Groeninckx et al. (in press) Pentodon + Dentella were resolved as the possible first branching clade of the entire tribe. Clade D constitutes the remaining taxa of Spermacoceae. The Spermacoce clade is for the first time retrieved as a well supported monophyletic clade. The relationships within Spermacoceae will be dealt with in following detailed discussions of each major clade. 4.2.1. Kohautia Kohautia has 36 species distributed from Africa via the Arabian Peninsula to India and Australia (Govaerts et al., 2008; number of species and distribution for genera are in the following taken from this work, if not otherwise stated). It is well characterized by its flowers, which always have both anthers and stigmas included (Bremekamp, 1952). The stigmas are usually situated well below the anthers, but occasionally they are nearly equal in height. This flower type is so rare in Rubiaceae that the generic status of Kohautia never has been questioned. Groeninckx et al. (in press), nevertheless, showed that the genus was not monophyletic. Their result is confirmed by our analyses. The Kohautia species form two well supported clades corresponding to the two subgenera of Kohautia, i.e., Kohautia and Pachystigma. Subgenus Kohautia is the sister group (PP 1.00) of the so-called Pentanopsis clade and subgenus Pachystigma is well supported as the sister group to Oldenlandia s.s. (PP = 1.00; clade I). The subgenera Kohautia and Pachystigma are easily distinguished by their stigmas. The former has a style ending with two filiform stigmas and the latter has a single ovoid or cylindrical stigma (Bremekamp, 1952). Indications from chromosomal and palynological data (Lewis, 1965a) as well as from seed shape (Mantell, 1985) support the division into two separate clades. Pachystigma is restricted to Africa and Madagascar, while Kohautia extends to Tropical Asia and Australia. The two subgenera should be treated as separate genera. Alternatively, the nine species of Pachystigma could be included in Oldenlandia s.s., but considering the distinct flower type we suggest the recognition of a new genus. 4.2.2. The Pentanopsis clade The Pentanopsis clade as identified by previous chloroplast studies (Thulin and Bremer, 2004; Dessein et al., 2005; Groeninckx et al., in press) is also supported by our results (PP 1.00). It is an Afro-Madagascan clade. In addition to Pentanopsis it consists of Amphiasma, Conostomium, Gomphocalyx, Manostachya, Phylohydrax, and three species of Oldenlandia. Pentanopsis itself consists of two species distributed from Ethiopia to Northern Kenya. Pentanopsis gracilicaulis was recently transferred from Amphiasma. It agrees with Pentanopsis fragrans in having larger flowers than the species of Amphiasma, persistent, more or less woody stipules, and four- versus three-colporate pollen grains (Thulin and Bremer, 2004). Amphiasma, here represented by two of its seven species, is distributed from southern Tanzania to Namibia. Our data together with morphological support (e.g., tubular stipular sheaths, very short corolla tube with a hairy throat, flattened seeds, and threecolporate pollen; Bremekamp, 1952) suggest that Amphiasma is monophyletic (PP 1.00). Oldenlandia affinis has been shown to be closely related to Amphiasma by Andersson and Rova (1999). Pentanopsis was not included in that study and Oldenlandia affinis is in fact well supported as the sister group to Pentanopsis (PP 1.00; this study; Groeninckx et al., in press). A detailed morphological study should address whether additional Oldenlandia species also 853 belong to this clade and if there are morphological characters shared by Oldenlandia affinis, Pentanopsis, and Amphiasma other than general characters such as sessile, linear leaves, only indistinctly beaked capsules, and seeds with non-punctate testa cells (Bremekamp, 1952). Conostomium is a genus of five species distributed from Ethiopia to South Africa. It was initially described based on capsules that are loculicidally dehiscent at the apex. This capsule type is, however, also observed in other genera of Spermacoceae. Conostomium is further characterized by its seeds with granulate testa cells and characteristic pollen grains (Bremekamp, 1952). Other studies (Lewis, 1965a) have shown that the distinctiveness of the pollen grains was overemphasized. Conostomium was not supported as monophyletic by Groeninckx et al. (in press); Oldenlandia herbacea was nested within it. In the present analyses we sequenced additional specimens (another specimen of Conostomium natalense, Oldenlandia herbacea var. goetzei, and O. herbacea var. herbacea, i.e. two of the three varieties) in order to retest these results. Our data support Conostomium as monophyletic (PP 0.96) with Oldenlandia herbacea as its sister group (PP 1.00). Oldenlandia herbacea is similar to Conostomium in having coarsely granulate testa cells (Bremekamp, 1952; Dessein, 2003) and comparatively large pollen grains (Bremekamp, 1952; Scheltens, 1998). Bremekamp (1952) placed Oldenlandia herbacea in the subgenus Euoldenlandia (the subgeneric classification of Bremekamp, 1952, only includes African species, but non-African species are sometimes referred to). He mentioned that together with O. pumila it differs from the other species of the subgenus by having longpedicellate flowers and granulate testa cells. Oldenlandia pumila is distributed from India to Java (and is mentioned as introduced elsewhere; Tanzania and Jamaica according to Bremekamp, 1952). Bremekamp (1952) suggested, consequently, that Oldenlandia herbacea, which is widespread in Africa and present in India and on Sri Lanka, originated in Asia and spread into Africa. Considering that it is nested within the African Pentanopsis clade, this biogeographic hypothesis seems reversed. Since Oldenlandia herbacea is related to Conostomium and not to Oldenlandia s.s., it should be transferred out of Oldenlandia. Phylohydrax with two species and the monotypic Gomphocalyx are sister taxa (PP 1.00). They were formerly included in the Spermacoce clade (Robbrecht, 1988) based on their habit, uniovulate ovary locules, and pluri-colporate pollen grains. Their placement in the Pentanopsis clade is, however, supported both by their distribution (East Africa and Madagascar) and by several morphological characters (e.g., heterostylous flowers, filiform stigma lobes, placenta attached at the base of the septum; Thulin and Bremer, 2004; Dessein et al., 2005). Manostachya and Oldenlandia rosulata (PP 0.89) also belong to the Pentanopsis clade as the sister group (PP 1.00) to Phylohydrax and Gomphocalyx. Oldenlandia rosulata occurs in Tropical and Southern Africa and is closely related to Oldenlandia microcalyx from Cameroon and Angola, the only other member of Oldenlandia subgenus Trichopodium (Bremekamp, 1952). Since these two species of Oldenlandia cannot be kept in Oldenlandia a new genus should possibly be erected. The three species of Manostachya from Central and Eastern Tropical Africa are characterized by their testa cells with thick outer walls and a network of ridges on the inside and by large pollen grains (Bremekamp, 1952). Manostachya notably has a basic chromosome number of x = 11, while most other African Spermacoceae have x = 9 (Lewis, 1965a). Stephanococcus (not investigated here due to lack of material) with a single winding species from Cameroon, Gabon, and D.R. Congo has been considered an isolated genus, but was suggested as the closest relative of Manostachya based on similarities in the axillary flower clusters, stipular sheath, Author's personal copy 854 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 and in the flattened seeds (Bremekamp, 1952). It might consequently also belong to the Pentanopsis clade. 4.2.3. Clade C The Pentodon and Dentella clade (PP 1.00) is the sister (PP 0.91) to the Agathisanthemum–Hedyotis s.s. clade (PP 1.00). This sister relationship is not found in the chloroplast analyses. Groeninckx et al. (in press) reported Pentodon + Dentella as the first branching clade of Spermacoceae (BS < 50%, PP 0.84). According to our chloroplast data the two genera are indicated as the sister group to clade D, albeit with very low posterior probability (PP 0.56). Because Pentodon and Dentella are supported as sister taxa in all independent analyses (PP 1.00), a second individual of Pentodon pentandrus was sequenced for the nuclear data. It reconfirms the position and identity of the first individual. Pentodon has two African species, one of which is extending to the Arabian Peninsula and Madagascar and is also introduced to the New World. Pentodon is atypical among the Spermacoceae in having five-merous flowers and peltate placentas with a bilobed apex (Bremekamp, 1952). Dentella with its eight species (sometimes only one or two recognized; e.g., Ridsdale, 1998) has a wide distribution from Tropical and Subtropical Asia to the Southwest Pacific. It shares with Pentodon the character five-merous flowers. 4.2.4. The Agathisanthemum–Hedyotis s.s. clade Agathisanthemum forms a well supported clade (PP 1.00) together with Lelya and three species of Oldenlandia (O. angolensis, O. goreensis, O. uniflora). This clade is sister to Hedyotis s.s. (PP 1.00). This result supports Bremekamp’s (1952, p. 5) notion that Agathisanthemum is ‘‘in aspect not unlike the Indian Hedyotis fruticosa L and its nearest allies.” The four species of the African Agathisanthemum are characterized by capsules which split into two mericarps. This capsule structure is similar to Hedyotis s.s., but Agathisanthemum differs from Hedyotis s.s.by havingmorenumerous,smaller,angularseeds and shorterstigmas (Bremekamp, 1952). Verdcourt (1976) suggested that Agathisanthemum possibly would be better classified as a section of Hedyotis. Agathisanthemum is, however, not monophyletic (Groeninckx et al., in press; this study). The monotypic Lelya falls within Agathisanthemum. Both genera have the same pollen aperture type (Lewis, 1965a). Lelya differs from all other Spermacoceae in having a thick-walled capsule with a solid beak. This unique capsule was the reason for the generic status for Lelya, which otherwise resembles Oldenlandia (Bremekamp, 1952). In habit it specifically resembles Oldenlandia goreensis (Bremekamp, 1952), one of the three Oldenlandia species which constitute the sister group (PP 1.00) to Agathisanthemum + Lelya (PP 1.00). Two of the three Oldenlandia species are classified in Oldenlandia subgenus Anotidopsis (O. angolensis, O. goreensis). This subgenus includes other African species and Asian species that are characterized by distinctly beaked capsules and sheathing stipules with a bifid or bipartite lobe on each side of the stem (Bremekamp, 1952). The New World taxon Oldenlandia uniflora (central and eastern North America, Caribbean, Brazil, Paraguay, N. Argentina), is the sister to O. angolensis and O. goreensis (PP 1.00). More detailed studies with extended sampling are needed to evaluate if a new genus should be described to encompass all or part of subgenus Anotidopsis and O. uniflora or if these species are better treated as members of Agathisanthemum. Hedyotis has traditionally been treated either in a narrow sense (Hedyotis s.s.; Bremekamp, 1952; Hallé, 1966; Terrell, 1975, 1991, 2001c; Andersson et al., 2002) restricted to tropical and subtropical Asia to the north-western Pacific or has been given a wide circumscription including also American and Polynesian taxa (e.g., Fosberg, 1943; Fosberg and Sachet, 1991; Merrill and Metcalf, 1942; Hsienshui et al., 1999; Dutta and Deb, 2004). Hedyotis s.s. is a well supported clade (PP 1.00). There is no support for a Hedyotis s.l. (Andersson et al., 2002; Groeninckx et al., in press). All of the North American species should, thus, be recognized as Houstonia or other segregate genera (Terrell, 1991, 2001a,b) and the Polynesian taxa are to be treated as Kadua (Terrell et al., 2005), as done here and by Govaerts et al. (2008). One of the sampled species currently accepted as Hedyotis (Govaerts et al., 2008) do not group with the Hedyotis s.s. clade (Hedyotis capitellata; clade H in Fig. 3b). On the other hand, two Asian species of Oldenlandia, O. consanguinea and O. effusa, belong to Hedyotis s.s. and consequently the names Hedyotis consanguinea Hance and H. effusa Hance should be used for the two species. 4.2.5. Clade D The first branching taxon of clade D (PP 1.00) is the Kenyan Dibrachionostylus. It is rather distinct within Spermacoceae in having a style divided into two branches. Bremekamp (1952) considered the monotypic Dibrachionostylus as a close relative of Agathisanthemum and he identified additional differences (e.g., corolla tube glabrous inside, entirely glabrous style, and non-punctate testa cells). Verdcourt (1976) suggested affinities also to Hedythyrsus. Clade D constitutes, apart from Dribrachionostylus, a clade (clade E; PP 0.96) with two lineages: clade F (PP 0.53) and the remaning taxa of clade D (Fig. 3b; PP 0.66). Clade F contains Mitrasacmopsis + Hedythyrsus + four species of Oldenlandia (clade G; PP 0.89) and the Arcytophyllum–Houstonia clade (PP 1.00). 4.2.6. Clade G The monotypic Mitrasacmopsis from Central and Eastern Tropical Africa and Madagascar was originally placed in Loganiaceae because of its semi-inferior to superior ovary (Jovet, 1941). Bremekamp (1952) described a new genus Diotocranus within Hedyotideae, which was later reduced to the synonymy of Mitrasacmopsis. According to Bremekamp (1952), Mitrasacmopsis is morphologically similar to Hedythyrsus in the type of capsule dehiscence (loculicidal followed by septicidal dehiscence), the small number of seeds per capsule, and the strongly undulating walls of the testa cells. They also share beaked capsules (although the beak is much more pronounced and the capsule base is bilobed in Mitrasacmopsis) and a similar placentation type (distinctly stalked placentas with the ovules at a peripheral position of the placenta; Groeninckx et al., 2007, in press). Hedythyrsus includes two species with a similar distribution to Mitrasacmopsis, but is absent from Madagascar. According to our data Hedythyrsus is sister to Mitrasacmopsis (PP 1.00). The monotypic genera Pseudonesohedyotis and Nesohedyotis are allied to Hedythyrsus according to Verdcourt (1976) and, consequently, they probably also belong to this clade. The rps16 sequence of Nesohedyotis (Andersson and Rova, 1999) does according to our analysis, however, belong to clade J (Fig. 3b). Nesohedyotis with unisexual flowers is unusual among Spermacoceae. It is endemic to St. Helena. There is, thus, some biogeographical indication that its position in the largely American clade J might be correct. The hypothesis that the Tanzanian Pseudonesohedyotis with hermaphroditic flowers also belongs to clade J seems less likely from a biogeographical and morphological standpoint and it has more likely affinities to Hedythyrsus. The sister group to Mitrasacmopsis and Hedythyrsus is Oldenlandia fastigiata (PP 1.00), which is distributed from Ethiopia to Mozambique. It has a similar placentation type as the other two genera (Groeninckx et al., in press), but has testa cells with straight walls (Bremekamp, 1952). Oldenlandia fastigiata is classified in subgenus Oldenlandia (Bremekamp, 1952), a classification that is unsupported phylogenetically. The sister group (PP 0.89) of Mitrasacmopsis, Hedythyrsus, and Oldenlandia fastigiata is a clade (PP 1.00) of three other species of Author's personal copy J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Oldenlandia: O. echinulosa from Tropical Africa, O. geophila from Zambia, and O. nervosa distributed from West Central Tropical Africa to Angola. Clade G, thus, seems to be an entirely African clade. Oldenlandia echinulosa and O. nervosa are classified in subgenus Hymenophyllum and O. geophila in subgenus Orophilum (Bremekamp, 1952). Both subgenera are characterized by distinctly petiolate leaves. Hymenophyllum comprises annuals with rather large and thin leaves, a short, fimbriate stipular sheath, glabrous style, and testa cells with undulating walls. It differs from subgenus Orophilum, which includes perennial species with smaller, often leathery leaves, a stipular sheath drawn out into a triangular lobe, hirtellous style, and testa cells with straight walls (Bremekamp, 1952). The type species of subgenus Orophilum, Oldenlandia monanthos, belongs to Oldenlandia s.s. (Fig. 3b). Further studies are needed to properly investigate how to classify the taxa of clade G. If the current phylogenetic relationships remain with a denser taxon sampling, three possibilities should be investigated: (1) lumping all taxa into Mitrasacmopsis (generic name with priority), (2) merging Hedythyrsus and Oldenlandia fastigiata into Mitrasacmopsis and erecting a new genus for the other species of Oldenlandia, or (3) keeping Mitrasacmopsis and Hedythyrsus as distinct genera and erecting two new genera to encompass Oldenlandia fastigiata and the other species of Oldenlandia, respectively. 4.2.7. The Arcytophyllum–Houstonia clade The Arcytopyllum–Houstonia clade is well supported (PP 1.00). Arcytophyllum with 16 species (excluding A. serpyllaceum; see below) distributed from Mexico to western South America is well supported as monophyletic (PP 1.00; Andersson et al. 2002). The sister group of Arcytophyllum is a clade (PP 1.00) comprising Houstonia (PP 0.69) + Stenaria nigricans (PP 0.98), and one species of Oldenlandia (PP 1.00): O. microtheca from Mexico. This clade as well as the entire Arcytophyllum–Houstonia clade seems to be restricted to the New World. That Oldenlandia microtheca belongs to the clade seems reasonable considering both its distribution and the fact that it has a basic chromosome number of x = 11, which is in contrast to most species of Oldenlandia (Lewis, 1965a), and agrees with several counts made for species of Houstonia (Lewis, 1962, 1965b). The relationships within Houstonia, which consists of 30 North and Central American species, has recently been closer studied by Church (2003). Stenaria contains five species distributed from Central and Eastern USA to Mexico and Bahamas. It was previously included in Hedyotis, but was elevated to generic status by Terrell (2001a). In the study by Church (2003), Stenaria was nested within Houstonia and, consequently, suggested to be merged with Houstonia. Two closely related genera from Baja California, Stenotis (Terrell, 2001b) with seven species and the monotypic Carterella (Terrell, 1987; possibly congeneric with Stenotis; Church, 2003), most likely also belong to the Arcytophyllum–Houstonia clade. 4.2.8. Clade H; Asian, Australian, and Pacific Spermacoceae Clade H (Fig. 3b) is a well supported clade (PP 1.00), but its basal branches are poorly supported. Hedyotis capitellata, distributed from North Eastern India (Assam) to the Philippines, and the two Australian species Oldenlandia mitrasacmoides and Synaptantha belong here, but their relationships to the other taxa of clade H are uncertain. Synaptantha with two species from Australia might be the sister (PP 0.75) to the remaining taxa of clade H, the majority of which have an Australian-Asian-Pacific distribution. Synaptantha has long been recognised as a separate genus (Hooker, 1873). It differs from other Spermacoceae by having corolla with the lobes slightly connate, stamens with filaments attached to both the ovary and the corolla, and semi-inferior ovaries (Halford, 1992). Synaptantha does not only have a distinct morphology, it also has many autapomor- 855 phies in the ITS tree (one of the longest branches in the ITS tree; not present in the ETS data). The sequence does, however, not seem to represent a pseudogene since 5.8S is conserved; in a pseudogene one would expect also 5.8S to have an elevated substitution rate (Bailey et al., 2003). Synaptantha also seems to have an elevated rate of evolution in the chloroplast regions. In the trnL-F analysis it has the longest branch. It also has long branches in the other three chloroplast regions, but in these regions the rate is not as conspicuously elevated as for ITS and trnL-F. The remaining taxa of clade H (PP 0.77) include a clade with three Oldenlandia species (O. galioides, O. lancifolia, O. tenelliflora [syn. Scleromitrion tenelliflorum; Hedyotis tenelliflora]; PP 1.00), Kadua (PP 1.00), and its sister species O. biflora (PP 1.00). Both Oldenlandia galioides, which is distributed from New Guinea to the South West Pacific, and O. tenelliflora, from Tropical and Subtropical Asia to Northern Queensland, have obconic seeds (in contrast to, e.g., O. mitrasacmoides, which has scutelliform seeds; Halford, 1992). Oldenlandia lancifolia has more angulate seeds (Bremekamp, 1952) and is widespread in Africa and naturalized in Tropical South America. Bremekamp (1952) placed it in the subgenus Aneurum together with the North American Oldenlandia boscii and the Asian O. diffusa. One or several new genera should better be recognised to acknowledge the early branching members of clade H, since the species clearly do not belong to either Hedyotis s.s. or Oldenlandia s.s. Oldenlandia biflora is the sister group to Kadua (PP 1.00) as in the chloroplast study by Groeninckx et al. (in press). It is distributed from tropical and subtropical Asia to the West Pacific. That the sister group to the Polynesian Kadua would have such a distribution seems reasonable. Oldenlandia biflora should be transferred from Oldenlandia, but whether there is morphological support to include it in Kadua will have to await further studies. Kadua with presently 28 species (Terrell et al., 2005; Govaerts et al., 2008) is monophyletic (PP 1.00). This supports the conclusion from previous molecular and seed anatomical studies that the Polynesian taxa formerly included in Hedyotideae (as Hedyotis, Gouldia, or Wiegmania) should be regarded as the separate genus Kadua (Motley, 2003; Terrell et al., 2005). Motley (in press) has studied Kadua with a broader sampling and found that Kadua is monophyletic and subclades for the most part reflect former classifications. This study also showed that the Hawaiian Kadua species are paraphyletic with respect to the French Polynesian species and provide strong evidence for migration of a plant lineage out of the archipelago. The nested position of K. rapensis in the Kadua clade in this study supports this finding. 4.2.9. Clade I, the true Oldenlandia and Pachystigma Oldenlandia is clearly polyphyletic. The clade including the type species, O. corymbosa, is here referred to as Oldenlandia s.s. (PP 1.00). All species of Oldenlandia not belonging to this clade should be transferred to other genera or assigned to new genera. Some suggestions are already made above, but we refrain from making the formal taxonomic changes pending studies with better sampling. The monotypic genus Thecorchus distributed from Ethiopia to Senegal should, however, no longer be recognized (Kårehed and Bremer, 2007; Groeninckx et al, in press). It is nested within Oldenlandia s.s. and the use of the name Thecorchus wauensis (Schweinf. ex Hiern) Bremek. should be abandoned in favor of its basionym Oldenlandia wauensis Schweinf. ex Hiern. As mentioned above, the sister group (PP 1.00) of Oldenlandia s.s. is Kohautia subgenus Pachystigma (PP 1.00), which should be recognized as a distinct genus. 4.2.10. Clade J and the origin of the Spermacoce clade Clade J (PP 1.00) is the sister group (PP 1.00) of clade I. It is almost entirely restricted to the New World. All taxa of clade J, ex- Author's personal copy 856 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 cept Nesohedyotis from St. Helena, are from the New World. A South American origin of the Spermacoce clade as suggested by Dessein (2003) seems very probable, since the Spermacoce clade is a predominantly New World taxon. Only Diodia s.l., Hydrophylax, and Spermacoce s.l. (pantropical) extend into other tropical regions. Manettia (PP 1.00) and Bouvardia (PP 0.74) are sister genera (PP 0.97). If Arcytophyllum serpyllaceum (rps16 data only) is included in Bouvardia (PP 1.00) as suggested by Andersson et al. (2002) they appear monophyletic. Manettia with 124 species from Tropical America has been recognised as a separate tribe, Manettieae (Bremekamp, 1934). Based on its winged seeds, Manettieae were previously placed in the subfamily Cinchonoideae (Schumann, 1891). Manettia was later moved to the Rubioideae and Hedyotideae because of the presence of raphides (Bremekamp, 1966). Robbrecht (1988) included both genera in Cinchoneae, but because of the shared presence of raphides the Hedyotideae were suggested as a possible alternative placement. Manettia is morphologically similar to Bouvardia and Bremer (1996) found the two genera to be sister taxa in Rubioideae based on molecular data. Bouvardia comprises 41 species from Southern USA to Central America. Manettia differs from Bouvardia mainly in its viney habit and hard endosperm. Nesohedyotis (rps16 data only) and two Oldenlandia species (O. salzmanii + O. tenuis; PP 1.00) belong to clade J, but their position as successive sister taxa to the Spermacoce clade is very weekly supported. 4.2.11. The Spermacoce clade The Spermacoce clade, although long recognized as a separate tribe (Spermacoceae; Berchtold and Presl, 1820; Hooker, 1873; Bremekamp, 1952, 1966; Verdcourt, 1958; Robbrecht, 1988, 1994), was not recovered as a well supported monophyletic group in the chloroplast study by Groeninckx et al. (in press). However, with the addition of nuclear data, the Spermacoce clade is well supported (PP 1.00), provided that Gomphocalyx and Phylohydrax are excluded. Both genera belong to the Pentanopsis clade (Thulin and Bremer, 2004; Dessein et al., 2005; see above). The Spermacoce clade differs from the rest of the tribe in having ovaries with a single ovule per locule attached near the middle of the septum and often pluriaperturate pollen grains, in contrast to few to many ovules per locule and often tricolporate pollen grains. The genera of the Spermacoce clade are often recognized based on the type of fruit dehiscence. A detailed overview of the Spermacoce clade is given by Dessein (2003). Galianthe (PP 1.00; including Diodia spicata) is the first branching taxon of the Spermacoce clade. It constitutes 45 species from Mexico to South America and is characteristic among Spermacoceae s.s. in having terminal lax inflorescences, heterostylous flowers, and a stigma with two distinct lobes (Cabral, 1991). Diodia spicata (treated under the name Spermacoce spicata (Miq.) in ed. by Govaerts et al., 2008) makes Galianthe paraphyletic. Diodia spicata differs by having terminal spicate inflorescences with isostylous flowers and fruits with two cocci separating from the base. Nevertheless, since it approaches Galianthe with its two-armed stigma and 7-zonocolporate pollen with relatively long colpi and double reticulum, it seems reasonable to include Diodia spicata in Galianthe (Dessein, 2003). The Central American Crusea, here represented by two of 14 species, is monophyletic (PP 1.00). Emmeorhiza, a South American monotypic genus, is the sister to Crusea (PP 0.90). This relationship was not found by Groeninckx et al. (in press). The position of Emmeorhiza was not well supported in their study. In the strict consensus tree of the parsimony analysis it was the sister to Galianthe + Diodia spicata (JK and BS <50%) and in the Bayesian inference it was the sister to Nesohedyotis arborea (PP 0.67). The addition of petD data to the other chloroplast data places Emmeorhiza in the same larger clade as in the combined analysis (clade K; PP 0.96), but as the first branching taxon (PP 0.96) and not as the sister to Crusea. Emmeorhiza is somewhat similar in habit to the genus Denscantia. Both genera are climbers with thyrsoid-like inflorescences with isostylous flowers (Dessein, 2003). Denscantia (first published as Scandentia, but later changed to Denscantia; Cabral and Bacigalupo, 2001a,b) was segregated from the probably closely related Galianthe. Denscantia has isostylous flowers, lateral fusion of the stipular bases, and pollen with multiple endoapertures whereas Galianthe often has heterostylous flowers, stipules fused only with the leaf base or the petiole, and pollen with a simple endocingulum (Cabral and Bacigalupo, 2001a; Dessein, 2003). Further study is needed to determine if Denscantia belong to the Crusea + Emmeorhiza clade or if the morphological resemblance to Emmeorhiza is due to convergence. The genus Spermacoce is not monophyletic in the present study. Several smaller genera are intermingled with Spermacoce species. Richardia (PP 1.00) has 16 species from Tropical and Subtropical America and is naturalized elsewhere. It is a well defined genus characterized by mainly 3–4-carpellate ovaries and schizocarpous fruits splitting into indehiscent mericarps (Dessein, 2003). Spermacoce ocymifolia has been treated as Hemidiodia ocymifolia in a genus of its own (Schumann, 1888) characterized by its fruits. They consist of two indehiscent mericarps. At maturity the mericarps are only partially joined by their bases. Hemidiodia was sunk into Borreria (a genus previously recognized to encompass American Spermacoce) because, apart from the fruit type, they are morphologically similar (Bacigalupo and Cabral, 1996). Our data show Spermacoce ocymifolia as sister to S. remota (PP 0.83). Mitracarpus (PP 1.00; represented by two out of the 49 species) also belongs to the Spermacoce clade as the first branching taxon of a well supported subclade (clade L; PP 0.96). Mitracarpus is from Tropical America and is naturalized elsewhere. The genus is distinguished by its cirumscissile opening of the fruits and in having two large and two small calyx lobes and seeds with an X-shaped ventral groove (Dessein, 2003). Diodella teres and Psyllocarpus laricoides, two species only represented by nuclear data, are each others closest relatives (PP 1.00) and the second branching clade (PP 0.94) of clade L. If their sister group relationship is due only to the present taxon sampling or a true relationship will have to await further studies. No obvious morphological characters are shared by the two genera. Psyllocarpus with its nine Brazilian species is distinct from the other genera of the Spermacoce clade by having a capsule strongly compressed parallel to the septum. Govaerts et al. (2008) accepted 30 species in Diodia. In contrast, Bacigalupo and Cabral (1999) only recognized five species in the genus. These authors transferred 16 species provisionally to the genus Diodella. Nine of these are now formally published under Diodella. Diodella is distributed from Central USA to Tropical America and is also present in Africa. It is characterized by fruits dehiscing into two, indehiscent, one-seeded mericarps. Dessein (2003) suggests that species assigned to Diodella fall into two groups: one group is related to Diodella teres, the type species, the other one to Diodella sarmentosa. This is confirmed by our analysis as Diodella sarmentosa forms a clade (PP 0.84) with Diodia aulacosperma, Ernodea, Hydrophylax, Spermacoce hispida, S. ruelliae, S. filifolia, and S. fillituba. The African Diodia aulacosperma (Socotra, S. Somalia to E. Tanzania (including Zanzibar) is not regarded as a member of a restricted Diodia (Bacigalupo and Cabral, 1999; Dessein, 2003). Ernodea, a genus with nine species distributed from Florida south to Central America and in the Caribbean Islands, shares some seed and pollen morphological characters with Diodella sarmentosa (Dessein, 2003), but are in habit and fruit morphology quite different. Diodella sarmentosa has dry fruits splitting into two mericarps, while Ernodea has drupaceous fruits, a rare condition in Spermacoceae. Hydrophylax maritima, the sole species in the genus, is a sea shore plant from India, Sri Lanka, and Thailand. Dessein (2003) pointed to the strong similarity in habit between Hydrophylax Author's personal copy 857 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 and the West African Diodia vaginalis (another of the species probably to be excluded from Diodia) and species of Diodia s.s. However, pollen grains are very different in these taxa. Further studies, especially with a larger sampling of Spermacoce species, will reveal if Spermacoce should be split or if several of the other genera of the Spermacoce clade should be merged. Note added in proof The DNA here sequenced as Spermacoce filifolia was extracted from Spermacoce flagelliformis, De Block et al. 794 (BR). The accession numbers AM939538 and AM933010 consequently refer to sequence data from the latter species. Acknowledgments Anbar Khodabandeh has been very helpful in the laboratory. Sylvain Razafimandimbison and Johan Nylander have contributed with valuable suggestions and discussions. The study was financed by a research grant to BB from The Swedish Research Council. Appendix List of taxa used in the phylogenetic analyses with voucher information (geographic origin, collector, collector number, herbarium), and accession numbers. New taxa/specimens not included in by Groeninckx et al. (in press) are indicated in bold. Missing sequences are marked with. Literature citations for previous published sequences: (1)=Andersson and Rova 1999, (2)=Andersson et al. 2002, (3)=Dessein et al. 2005. Taxon Voucher information Agathisanthemum Klotzsch A. bojeri Klotzsch Zambia: Dessein et al. 671 (BR) A. globosum (Hochst. Zambia: Dessein ex A. Rich.) Klotzsch et al. 201 (BR) Amphiasma Bremek. A. benguellense (Hiern) Bremek. A. luzuloides (K. Schum.) Bremek. Angola: Kers 3350 (S) Zambia: Dessein et al. 1167 (BR) Tanzania: Iversen et al. 87694 (UPS) Arcytophyllum Willd. ex Schult. & Schult. f. A. aristatum Standl. Ecuador: Hekker & Hekking 10335 (GB) A. ciliolatum Standl. Unknown: Oligaard et al. 58395 (NY) A. ericoides (Willd. ex Unknown: Edwin et Roem. & Schult.) al. 3624 (S) Standl. A. lavarum K. Schum. Unknown: Cronquist 8827 (NY) A. macbridei Standl. Unknown: Wurdack 1073 (NY) A. muticum (Wedd.) Colombia: Standl. Andersson et al. 2195 (GB) A. nitidum (Kunth) Unknown: Pipoly Schltdl. et al. 6467 (GB) A. rivetii Danguy & Ecuador: Harling & Cherm. Andersson 22232 (GB) A. serpyllaceum Mexico: Stafford (Schltdl.) Terrell et al. 203 (MO) A. setosum (Ruiz & Unknown: Pav.) Schltdl. Andersson et al. 2196 (GB) atpB-rbcL rps16 ITS ETS 5S-NTS trnL-trnF petD EU542917 EU543018 EU543077 EU557678 AM939424 / / EU542918 EU543019 EU543078 EU557679 AM939425 / / EU542919 AF002753(1) EU543079 EU557680 AM939426 AM932918 / EU542920 EU543020 EU543080 EU557681 AM939428 AM932919 / / / / / / AF333348(2) AF333349(2) / / / / / AF333350(2) AF333351(2) / / / / / AF333352(2) AF333353(2) / / / / / AF333354(2) AF333355(2) / / / / / AF333356(2) AF333357(2) / / / / EU557682 AM939429 / / / / / AM939430 / / / / / / AF333365(2) / / / / EU542921 AF002754(1) / EU543081 AF333359(2) / EU542922 AF333362(2) AF333363(2) / / AF333364(2) / / AF002755(1) AM939427 AM932920 / / (continued on next page) Author's personal copy 858 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Appendix Table 1 (continued) Taxon Voucher information A. thymifolium (Ruiz & Ecuador: Ståhl Pav.) Standl. 4481 (GB) Batopedina Verdc. (outgroup) B. pulvinellata Zambia: Dessein Robbrecht et al. 264 (BR) D.R. Congo: Malaisse 7695 (UPS) Bouvardia Salisb. B. glaberrima Engelm. B. ternifolia (Cav.) Schltdl. B. sp. atpB-rbcL rps16 Crusea Cham. & Schltdl. C. calocephala DC. Guatemala: Gustafsson et al. 215 (GB) C. megalocarpa (A. Mexico: Pringle Gray) S. Watson 3852 (S) Diodella Small D. sarmentosa Sw. French Guiana: Anderson et al. 2071 (GB) ETS 5S-NTS EU542924 EU543021 EU543083 EU557684 / / / / EU543084 / EU557685 AM939432 AM932922 / / / / / / / / AM266989 / / / / s.n. / / / / / / AM939433 AM932923 / / EU557686 / / / EU542927 AF002760(1) EU543085 EU557687 AM939435 AM932925 / / / / EU542928 EU543024 EU543086 EU557688 AM939436 AM932926 / EU542929 / EU543087 EU557689 AM939437 AM932927 / EU542930 / EU543088 EU557690 AM939438 AM932928 / EU542931 EU543025 EU543089 EU557691 AM939439 AM932929 / EU543090 EU557692 / EU543091 EU557693 AM939440 AM932930 / EU542933 AF002761(1) / EU557694 AM939442 AM932932 / / / / / AM939441 AM932931 / / AF002762(1) / / / / Dentella J. R. Forst & G. Forst. D. dioeca Airy Shaw Australia: Harwood / / 1559 (BR) D. repens (L.) J. R. Forst. Australia: EU542932 AF333370(2) & G. Forst. Andersson 2262 (GB) Dibrachionostylus Bremek. D. kaessneri (S. Moore) Kenya: Strid 2598 Bremek. (GB) Kenya: Strid 2564 (UPS) ITS EU557683 AM939431 AM932921 / Carphalea Juss. (outgroup) C. madagascariensis Madagascar: De EU542926 EU543023 Lam. Block et al. 578 (BR) / / Madagascar: Razafimandimbison 524 (UPS) Conostomium (Stapf) Cufod. C. natalense (Hochst.) South Africa: Dahlstrand 1346 Bremek. (GB) South Africa: Bremer et al. 4341 (UPS) C. quadrangulare Ethiopia: Puff & (Rendle) Cufod. Kelbessa 821222 2/ 2 (UPS) C. zoutpansbergense South Africa: (Bremek.) Bremek. Bremer et al. 4331 (UPS) petD EU542923 AF333366(2) EU543082 Cult.: Forbes s.n. (S) EU542925 EU543022 Cult.: S2928 (BR) / AF002758(1) Mexico: Spencer et al. 363 (NY) Mexico: Torres & Torino 3637 (BR) trnL-trnF / / / AM266995 / / AM939434 AM932924 / / / / / Author's personal copy 859 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Appendix Table 1 (continued) Taxon Voucher information atpB-rbcL rps16 trnL-trnF petD ITS D. teres (Walter) Small Madagascar: De Block et al. 793 (BR) / / / / AM939443 AM932933 / Kenya: Luke 9029 (UPS) French Guiana: Anderson et al. 1961 (GB) EU542934 EU543026 EU543092 EU557695 AM939444 AM932934 / EU542935 EU543027 EU543093 EU557696 AM939535 AM933008 / EU542936 AY764289(3) EU543094 EU557697 AM939445 AM932935 / Cuba: Rova et al. 2286 (GB) EU542937 AF002763(1) EU543095 EU557698 AM939446 AM932936 / Argentina: Vanni & Radovancick 996 (GB) Argentina: Schinini & Cristobal 9811 (GB) Bolivia: Persson & Gustavsson 298 (GB) EU542938 AY764290(3) EU543096 EU557699 AM939447 AM932937 / EU542939 EU543028 EU543097 EU557700 AM939448 AM932938 / / / / AM939449 AM932939 / EU567461 / / / Diodia L. s.l. D. aulacosperma K. Schum. D. spicata Miq. (Spermaoce spicata (Miq.) in ed.; Govaerts et al., 2008) Emmeorhiza Pohl ex Endl. E. umbellata (Spreng.) Trinidad: Hummel K. Schum. s.n. (GB) Ernodea Sw. E. littoralis Sw. Galianthe Grieseb. G. brasiliensis (Spreng.) E. L. Cabral & Bacigalupo G. eupatorioides (Cham. & Schltdl.) Cabral G. sp. Gomphocalyx Baker G. herniarioides Baker Hedyotis L. H. capitellata Wall. / Madagascar: De EU542940 AY764291(3) EU567466 Block et al. 569 (BR) Burma: Meebold 17373 (S) H. consanguinea Hance Hong Kong: Shiu (syn. Oldenandia Ying Hu 10821 (S) consanguinea (Hance) Kuntze) H. effusa Hance (syn. China: Tsang Oldenlandia effusa 21044 (S) (Hance) Kuntze) H. fruticosa L. Sri Lanka: Larsson & Pyddoke 22 (S) H. korrorensis The Caroline (Valeton) Hosok. Islands: Fosberg 47697 (S) H. lawsoniae Wight Sri Lanka: Wambeek & Wanntorp 2996 (S) H. lessertiana var. Sri Lanka: lessertiana Thwaites Klackenberg 413 (S) H. lessertiana var. Sri Lanka: Fagerlind marginata Thwaites 3668 (S) & Trimen H. macrostegia Stapf. Sabah:Wallander 6 (GB) / ETS 5S-NTS / / / AM939452 / / EU542941 / / EU557701 AM939450 AM931939 / / / AM939491 AM932940 AM931935 EU542942 / EU543098 EU557702 AM939453 AM932941 AM931929 EU542943 / EU543099 EU557703 AM939454 AM932942 AM931937 EU542944 / / EU557704 AM939455 AM932943 AM931931 EU542945 EU543029 EU543100 EU557705 AM939466 AM932944 AM931932 EU542946 EU543030 EU543101 EU557706 AM939456 AM932945 AM931934 EU542947 AF002767(1) EU543102 / / / AM942768 / / (continued on next page) Author's personal copy 860 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Appendix Table 1 (continued) Taxon Voucher information H. megalantha Merr. atpB-rbcL rps16 trnL-trnF petD ITS ETS 5S-NTS / / / / AM939457 AM932946 AM931936 Marianas (Guam): Andersson 07 (S) H. quinquenervis Sri Lanka: Bremer Thwaites et al. 163 (S) H. rhinophylla Sri Lanka: Fagerlind Thwaites ex Trimen 5082 (S) H. swertioides Hook. f. South India: Klackenberg & Lundin 03 (S) EU542948 / EU543103 EU557707 AM939458 AM932947 AM931933 EU542949 / EU543104 EU557708 AM939459 AM932948 AM931930 EU542950 EU543031 EU543105 EU557709 AM939460 / AM931938 Hedythyrsus Bremek. H. spermacocinus (K. Schum.) Bremek. Zambia: Dessein et al. 1017 (BR) EU542951 EU543032 EU543107 EU557711 AM939461 AM932950 / USA: Vincent & Lammers s.n. (GB) USA: Yatskievych 96-49 (MO) USA: Weigend 9963 (NY) EU542953 AF333379(2) EU543109 EU557713 AM939464 / / EU542954 AF002766(1) / EU567462 AM939465 / / / s.n. / / / / EU567457 / / / / / / EU542955 / EU543110 EU557714 AM939467 AM932952 / / s.n. s.n. / AM942769 / / / AF002765(1) / / / / / / s.n. s.n. / AM942770 / / EU543111 EU557715 AM939468 / / EU542957 AF333376(2) EU543112 EU557716 / / / / / / AM939469 / / / s.n. s.n. / AM942771 / / Houstonia L. H. caerulea L. H. longifolia Gaertn. Hydrophylax L. f. H. maritima L. f. Sri Lanka: Lundqvist 8945 (UPS) Kadua Cham. & Schltdl. K. acuminata Cham. & Hawaii: cult. at BR Schltdl. K. affinis Cham. & Hawaii HI: Motley Schltdl. 1733 (NY) K. axillaris (Wawra) W. Hawaii: HarrisonL. Wagner & Gagne s.n. (GB) Lorence Maui HI: Motley 1724 (NY) K. centranthoides Hawaii: Skottsberg 6788 (S) Hook. & Arn. K. cordata Cham. & Cult.: Lorence 8021 (PTBG) Schltdl. Hawaii HI: Fagerlind 6863 (S) K. coriacea (J. E. Smith) Hawaii HI: Motley W. L. Wagner & 1703 (NY) Lorence K. degeneri (Fosberg) Cult.: Wood 5062 W. L. Wagner & (PTGB) Lorence K. elatior (H. Mann) W. Kauai HI: Wagner 6350 (BISH) L. Wagner & Lorence K. fluviatilis C. N. Oahu HI: Motley Forbes 1747 (NY) K. flynnii (W. L. Kauai HI: Perlman Wagner & Lorence) 15631 (BISH) W. L. Wagner & Lorence K. foggiana (Fosberg) Hawaii: Sparre 27 W. L. Wagner & (S) Lorence / EU542956 EU543033 EU542958 AF333371(2) EU543113 EU557717 AM939470 AM932953 / / s.n. s.n. / AM942772 / / / s.n. s.n. / AM942773 / / / s.n. s.n. / AM942774 / / EU543114 EU557718 AM939471 / / EU542959 / Author's personal copy 861 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Appendix Table 1 (continued) Taxon Voucher information atpB-rbcL rps16 trnL-trnF petD ITS K. fosbergii (W. L. Wagner & D. R. Herbst) W. L. Wagner & Lorence K. laxiflora H. Mann Oahu HI: Motley 1677 (NY) / s.n. s.n. / AM942775 / / Molokai HI: Perlman 6677 (BISH) Cult. at WU: Kiehn & Luegmayr 920823-2/1 Cult.: Perlman 12783 (GB) Rapa Is.French Polynesia: Perlman 17953 (NY) / s.n. s.n. / AM942776 / / EU543115 EU557719 AM939472 AM932954 / EU542961 AF333375(2) EU543116 EU557720 AM939473 AM932955 / / s.n. / EU542962 EU543035 EU543117 EU557721 AM939484 AM932956 / EU542963 EU543036 EU543118 EU557722 AM939474 AM932957 / EU542964 EU543037 EU543119 EU557723 AM939476 AM932959 / EU542965 EU543038 EU543120 EU557724 AM939477 AM932960 / / / / / AM939478 AM932961 / / / / / AM939475 AM932958 / EU542966 EU543039 EU543121 EU557725 AM939479 AM932962 / / / / EU542967 EU543040 EU543122 EU557726 AM939481 / / / s.n. / / EU542968 EU543041 EU543123 EU557727 AM939482 AM932964 / EU542969 / EU543124 EU557728 AM939483 AM932965 / EU542970 / EU543125 EU557729 AM939485 / / / / EU567463 / EU543126 EU557730 AM939487 AM932967 / EU543127 EU557731 / K. littoralis Hillebr. K. parvula A. Gray K. rapensis F. Br. Kohautia Cham. & Schltdl. K. amatymbica Eckl. & South Africa: Zeyh. Bremer et al. 4307 (UPS) K. caespitosa Schnizl. Zambia: Dessein et al. 432 (BR) K. coccinea Royle Zambia: Dessein et al. 751 (BR) K. cynanchica DC. Zambia: Dessein et al. 469 (BR) K. longifolia Klotsch Zambia: Dessein et al. 462 (BR) K. cf. longifolia Zambia: Dessein Klotsch et al. 790 (BR) K. microcala Bremek. Zambia: Dessein et al. 1149 (BR) Zambia: Dessein et al. 1321 (BR) K. obtusiloba Schnizl. Kenya: Luke 9035 (UPS) K. senegalensis Cham. Burkina Faso: & Schltdl. Madsen 5940 (NY) K. subverticillata (K. Zambia: Dessein Schum.) D. Mantell et al. 432 (BR) K. virgata (Willd.) Madagascar: De Bremek. Block et al. 539 (BR) Lelya Bremek. L. osteocarpa Bremek. Tanzania: Gereau 2513 (BR) EU542960 s.n. s.n. / / Manettia Mutis ex L. M. alba (Aubl.) Wernh. French Guiana: EU542971 AF002768(1) Andersson et al. 1917 (GB) M. luteorubra (Vell.) Unknown: Bremer / / Benth. 2716 (UPS); cult. at Stockholm University M. lygistum (L.) Sw. Colombia: EU542972 AF002769(1) Andersson et al. 2128 (GB) Manostachya Bremek. M. ternifolia E. Zambia: Dessein Sampaio Martins et al. 265 (BR) EU542973 EU543042 / ETS / 5S-NTS / AM939480 AM932963 / / / / AM939486 AM932966 / / / AM932968 / (continued on next page) Author's personal copy 862 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Appendix Table 1 (continued) Taxon Voucher information ETS 5S-NTS petD EU542974 AF002770(1) EU543128 EU567464 AM939488 / EU542975 EU543044 / EU557732 AM939489 AM932969 / Mitrasacmopsis Jovet M. quadrivalvis Jovet Zambia: Dessein et al. 1273 (BR) EU542976 EU543045 EU543129 EU557733 AM939490 AM932970 / Nesohedyotis (Hook. f.) Bremek. N. arborea (Roxb.) Cult.: Chase 2915 Bremek. (K) / / / EU542977 EU543046 EU543130 EU557734 AM939492 AM932971 / EU542978 EU543047 EU543131 EU557735 AM939493 AM932972 / EU542979 EU567459 EU543132 EU557736 AM939494 AM932973 / EU542980 EU543048 EU543133 EU557737 AM939496 AM932974 / EU542981 EU543049 EU543134 EU557738 AM939497 AM932975 / EU542982 EU543050 EU543135 EU557739 AM939502 AM932979 / / / / / AM939500 AM932977 / / / / / AM939501 AM932978 / / EU543061 EU543147 EU557751 AM939503 AM932980 / EU542983 EU543051 EU543136 EU557740 AM939504 AM932981 / EU542984 / EU543137 EU557741 AM939505 AM932982 / EU542985 EU543052 EU543138 EU557742 AM939506 AM932983 / EU542986 EU543053 EU543139 EU557743 AM939507 / / / / / AM939498 / / EU542987 EU543054 EU543140 EU557744 AM939508 / / EU542988 EU543055 EU543141 EU557745 AM939510 AM932985 / / / / / AM939509 AM932984 / / / / / AM939511 / / / / / / AM939495 / / Oldenlandia L. O. affinis (Roem. & Zambia: Dessein Schult.) DC. et al. 627 (BR) O. angolensis K. Schum. Zambia: Dessein et al. 932 (BR) O. biflora (L.) Lam. Unknown: cult. at BR O. capensis L. f. var. Zambia: Dessein capensis et al. 843 (BR) O. capensis L. f. var. Tanzania: Kayombe pleiosepala Bremek. et al. s.n. (BR) O. corymbosa L. Zambia: Dessein et al. 487 (BR) Australia: Andersson 2260 (GB) Gabon: Andersson & Nilsson 2263 (GB) O. densa in ed. (O. Zambia: Dessein et al. 346 (BR) robinsonii Verdc., nom. illeg.) O. echinulosa K. Zambia: Dessein Schum. et al. 928 (BR) O. echinulosa K. Tanzania: Kayombo Schum. var. & Kahemela 1993 pellucida (Hiern) (BR) Verdc. O. fastigiata Bremek. Zambia: Dessein et al. 1019 (BR) O. galioides (F. Muell.) Australia: Harwood F. Muell. 1511 (BR) O. cf. galioides (F. Australia: Muell.) F. Muell. Harwood 1519 (BR) O. geophila Bremek. Zambia: Dessein et al. 935 (BR) O. goreensis (DC.) Zambia: Dessein Summerh. et al. 1286 (BR) Zambia: Dessein et al. 455 (BR) Zambia: Dessein et al. 1335 (BR) Tanzania: Richards & Arasululu 25910 (BR) rps16 ITS trnL-trnF Mitracarpus Zucc. ex Schult. & Schult. f. M. frigidus (Willd. ex French Guiana: Roem. & Schult.) K. Andersson et al. Schum. 1995 (GB) M. microspermus K. Guiana: JansenSchum. Jacobs et al. 4785 (GB) atpB-rbcL AF003607(1) / / / / / Author's personal copy 863 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Appendix Table 1 (continued) Taxon Voucher information atpB-rbcL O. herbacea (L.) Roxb. var. goetzei (DC.) Summerh. Zambia: Dessein et al. 442 (BR) Zambia: Dessein et al. 1218 (BR) Zambia: Dessein et al. 463 (BR) Zambia: Dessein et al. 1041 (BR) Zambia: Dessein et al. 1356 (BR) Zambia: Dessein et al. 1256 (BR) Mexico: Frödeström & Hultén 681 (S) Australia: Harwood 1520 (BR) O. herbacea (L.) Roxb. var. herbacea O. lancifolia (Schumach.) DC. O. lancifolia (Schumach.) DC. O. microtheca (Cham. & Schltdl.) DC. O. mitrasacmoides (F. Muell.) F. Muell. subsp. nigricans Halford O. mitrasacmoides (F. Muell.) F. Muell. subsp. trachymenoides (F. Muell.) Halford O. monanthos (Hochst. ex A. Rich.) Hiern O. nematocaulis Bremek. O. nervosa Hiern rps16 Zambia: Dessein et al. 924 (BR) Gabon: Andersson & Nilsson 2326 (GB) O. rosulata K. Schum. Zambia: Dessein et al. 1197 (BR) O. salzmannii (DC.) Brazil: Harley Benth. & Hook. f. ex 15514 (UPS) B. D. Jacks. O. sp. C Fl. Zamb. Zambia: Dessein et al. 716 (BR) O. taborensis Bremek. Tanzania: Bidgood et al. 4015 (BR) O. tenelliflora (Blume) Unknown: cult. at Kuntze BR O. tenuis K. Schum. Guyana: JansenJacobs et al. 41 (UPS) O. uniflora L. USA: Godfrey 57268 (GB) O. wauensis Schweinf. Ethiopia: Friis et al. ex Hiern (syn. 2560 (UPS) Thecorchus wauensis (Schweinf. ex Hiern) Bremek.) O. wiedemannii K. Kenya: Luke & Luke Schum. 8362 (UPS) Pentanisia (Paraknoxia) Bremek. (outgroup) Zambia: Dessein P. parviflora (Stapf ex et al. 678 (BR) Verdc.) Verdc. ex Bremek. ETS 5S-NTS petD EU542989 EU543056 EU543142 EU557746 AM939550 AM932986 / / / / EU542990 EU543057 EU543143 EU557747 AM939552 AM932988 / / / / AM939553 AM932989 / EU542991 EU543058 EU543144 / AM939512 AM932990 / / / / AM939499 AM932976 / EU542992 EU543059 EU543145 EU557749 AM939513 AM932991 / / / / EU543146 EU557750 AM939515 AM932992 / / / AM939516 / / / AM939517 AM932994 / / / / / Australia: Harwood EU542993 / 1516 (BR) Ethiopia: Friis et al. 276 (BR) ITS trnL-trnF / / EU542994 EU543060 AM939551 AM932987 / AM939514 AM932993 / / / AF333382(2) / XX999999 AM939518 AM932995 / / EU543043 EU567465 AM939519 / EU567467 / EU542995 AY764294(3) EU543148 EU557752 AM939520 AM932996 / / / / EU542996 / EU543149 EU557753 AM939522 / EU542997 EU543062 EU543106 EU557710 AM939451 AM932949 / / AM939549 AM932997 / / EU542998 AY764293(3) / EU557754 AM939523 / / EU542999 AY764295(3) EU543150 EU557755 AM939524 / / EU543017 EU543076 EU543168 EU557774 AM939548 AM933018 / EU543000 EU543063 EU543151 EU557756 AM939525 AM933001 / EU543001 EU543064 EU543152 EU557757 / / / (continued on next page) Author's personal copy 864 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Appendix Table 1 (continued) Taxon Pentanopsis Rendle P. fragrans Rendle P. gracilicaulis (Verdc.) Thulin & B. Bremer Pentodon Hochst. P. pentandrus (K. Schum. & Thonn.) Vatke Phylohydrax Puff P. carnosa (Hochst.) Puff P. madagascariensis (Willd. ex Roem. & Schult.) Puff Voucher information atpB-rbcL rps16 trnL-trnF petD ITS Kenya: Verdcourt 2454 (S) / / / / AM267020 / Ethiopia: Gilbert et al. 7458 (UPS) Somalia: Thulin et al. 10512 (UPS) / EU543065 EU543153 EU557758 AM939526 AM933002 / EU567458 EU567460 EU567468 / Zambia: Dessein et al. 598 (BR) EU543002 EU543066 EU543154 EU557759 AM939528 AM933003 / Zanzibar: Sundström 2 (GB) / / / EU543003 EU543067 XX999999 South Africa: Bremer 3783 (UPS) Madagascar: De EU543004 AY764292(3) EU543155 Block et al. 640 (BR) Psyllocarpus Mart. & Zucc. P. laricoides Mart. & Brazil: Andersson Zucc. et al. 35750 (UPS) Richardia L. R. brasiliensis Gomes R. scabra L. R. stellaris L. Spermacoce L. S. capitata Ruiz & Pav. / Madagascar: De Block et al. 904 (BR) Colombia: Andersson et al. 2073 (GB) Australia: Egeröd 85343 (GB) French Guiana: Andersson 1908 (GB) S. confusa Rendle ex Colombia: Andersson et al. Gillis 2074 (GB) S. erosa Harwood Australia: Harwood 1148 (BR) S. filifolia (Schumach. Zambia: Dessein & Thonn.) J.-P. et al. 881 (BR) Lebrun & Stork S. filituba (K. Schum.) Kenya: Luke 9022 Verdc. (UPS) S. flagelliformis Poir. Madagascar: De Block et al. 794 (BR) S. hispida L. Sri Lanka: Wanntorp et al. 2667 (S) S. ocymifolia Willd. ex French Guiana: Roem. & Schult. Andersson et al. (Hemidiodia 2040 (GB) ocymifolia (Willd. ex Roem. & Schult.) K. Schum.) / ETS / 5S-NTS / / AM939527 AM933004 / EU557760 AM939529 / / EU557761 AM939530 / / / / / / AM939531 AM933005 / / / / / AM939533 AM933007 / EU543005 AF003614(1) EU543156 EU557762 AM939532 AM933006 / EU543006 EU543068 EU543157 EU557763 AM939534 / / EU543007 EU543069 EU543158 EU557764 AM939536 / / / / / EU543008 EU543070 EU543159 EU557765 AM939537 AM933009 / / / / EU543009 EU543071 EU543160 EU557766 AM939539 AM933011 / EU543010 EU543072 EU543161 EU557767 / EU543011 EU543073 EU543162 EU557768 AM939540 AM933017 / EU542952 / EU543108 EU557712 AM939463 / / AF003619(1) / / / AM939538 AM933010 / / / / Author's personal copy 865 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Appendix Table 1 (continued) Taxon Voucher information Ecuador: Bremer 3340 (UPS) S. prostrata Aubl. Colombia: Andersson et al. 2078 (GB) S. remota (Lam.) French Guiana: Bacigalupo & Cabral Andersson et al. 2016 (GB) S. ruelliae DC. Gabon: Andersson & Nilsson 2296 (GB) S. verticillata L. Madagascar: De Block et al. 632 (BR) cult. BR; no voucher Stenaria (Raf.) Terrell S. nigricans (Lam.) USA: Yatskievych Terrell 96-92 (MO) Synaptantha Hook. f. S. tillaeacea (F. Muell.) Hook. f. atpB-rbcL rps16 trnL-trnF petD ITS / / / / AM939462 AM932951 / EU543012 / EU543163 EU557769 AM939541 AM933012 / EU543013 / EU543164 EU557770 AM939542 AM933013 / EU543014 EU543074 EU543165 EU557771 AM939543 AM933014 / / / / / AM939544 AM933015 / / / / / AM939545 AM933016 / EU543015 AF333373(2) EU543166 Australia: Lazarides EU543016 EU543075 & Palmer 272 (K) References Alvarez, I., Wendel, J.F., 2003. Ribosomal ITS sequences and plant phylogenetic inference. Mol. Phylogenet. Evol. 29, 417. Andersson, L., Rova, J.H.E., 1999. The rps16 intron and the phylogeny of the Rubioideae (Rubiaceae). Pl. Syst. Evol. 214, 161–186. Andersson, L., Rova, J.H.E., Guarin, F.A., 2002. Relationships, circumscription, and biogeography of Arcytophyllum (Rubiaceae) based on evidence from cpDNA. Brittonia 54, 40–49. APG, 1998. An ordinal classification for the families of the flowering plants. Ann. MO Bot. Gard. 85, 531–553. APG, 2003. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG II. Bot. J. Linn. Soc. 141, 399–436. Bacigalupo, N.M., Cabral, E.L., 1996. Infrageneric classification of Borreria (Rubiaceae-Spermacoceae) on the basis of American species. Op. Bot. Belg. 7, 294–308. Bacigalupo, N.M., Cabral, E.L., 1999. Révision de las especies Americanas del género Diodia (Rubiaceae, Spermacoceae). Darwiniana 37, 153–165. Bailey, C.D., Carr, T.G., Harris, S.A., Hughes, C.E., 2003. Characterization of angiosperm nrDNA polymorphism, paralogy, and pseudogenes. Mol. Phylogenet. Evol. 29, 435–455. Baldwin, B.G., 1992. Phylogenetic utility of the internal transcribed spacers of nuclear ribosomal DNA in plants: an example from the Compositae. Mol. Phylogenet. Evol. 1, 3–16. Baldwin, B.G., Markos, S., 1998. Phylogenetic utility of the external transcribed spacer (ETS) of 18S–26S rDNA: congruence of ETS and ITS trees of Calycadenia (Compositae). Mol. Phylogenet. Evol. 10, 449–463. Berchtold, F., Presl, J.S., 1820. O přirozenosti Rostlin. Krala Wiljma Endersa, Prague. Billera, L.J., Holmes, S.P., Vogtmann, K., 2001. Geometry of the space of phylogenetic trees. Adv. Appl. Math. 27, 733–767. Bremekamp, C.E.B., 1934. Notes on the Rubiaceae of Surinam. Recueil Trav. Bot. Néerl. 31, 248–308. Bremekamp, C.E.B., 1952. The African species of Oldenlandia L. sensu Hiern et K. Schumann. Verh. Kon. Ned. Akad. Wetensch., Afd. Natuurk., Sect. 2 48, 1–298. Bremekamp, C.E.B., 1966. Remarks on the position, the delimitation and the subdivision of the Rubiaceae. Acta Bot. Neerl. 15, 1–33. Bremer, B., Andreasen, K., Olsson, D., 1995. Subfamilial and tribal relationships in the Rubiaceae based on rbcL sequence data. Ann. MO Bot. Gard. 82, 383–397. Bremer, B., 1996. Phylogenetic studies within Rubiaceae and relationships to other families based on molecular data. Op. Bot. Belg. 7, 33–50. Bremer, B., Manen, J.-F., 2000. Phylogeny and classification of the subfamily Rubioideae (Rubiaceae). Pl. Syst. Evol. 225, 43–72. Bremer, B., Bremer, K., Heidari, N., Erixon, P., Olmstead, R.G., Anderberg, A.A., Källersjö, M., Barkhordarian, E., 2002. Phylogenetics of asterids based on 3 coding and 3 non-coding chloroplast DNA markers and the utility of non-coding DNA at higher taxonomic levels. Mol. Phylogenet. Evol. 24, 274–301. Bremer, B., in press. A review of molecular phylogenetic studies of Rubiaceae. Ann. MO Bot. Gard. EU543167 ETS 5S-NTS EU557772 AM939546 / / EU557773 AM939547 / / Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference, A Practical Information-theoretic Approach. Springer, New York. Cabral, E.L., 1991. Rehabilitación del género Galianthe (Rubiaceae). Bol. Soc. Argent. Bot. 27, 234–249. Cabral, E.L., Bacigalupo, N.M., 2001a. Scandentia, nuevo género de RubiaceaeSpermacoceae. Darwiniana 39, 29–41. Cabral, E.L., Bacigalupo, N.M., 2001b. Denscantia, nuevo nombre en reemplazo de Scandentia (Rubiaceae-Spermacoceae. Darwiniana 39, 353. Chevet, E., LeMaître, G., Katinka, M.D., 1995. Low concentrations of tetramethylammonium chloride increase yield and specificity of PCR. Nucleid Acids Res. 23, 3343–3344. Church, S.A., 2003. Molecular phylogenetics of Houstonia (Rubiaceae): descending aneuploidy and breeding system evolution in the radiation of the lineage across North America. Mol. Phylogenet. Evol. 27, 223–238. Cox, A.V., Bennett, M.D., Dyer, T.A., 1992. Use of the polymerase chain reaction to detect spacer size heterogeneity in plant 5S-rRNA gene clusters and to locate such clusters in wheat (Triticum aestivum L.). Theor. Appl. Genet. 83, 684–690. Dessein, S., 2003. Systematic studies in the Spermacoceae (Rubiaceae). D. Phil. Thesis, Katholieke Universiteit Leuven. Dessein, S., Andersson, L., Geuten, K., Smets, E., Robbrecht, E., 2005. Gomphocalyx and Phylohydrax (Rubiaceae): sister taxa excluded from Spermacoceae s.s., featuring a remarkable case of convergent evolution. Taxon 54, 91–107. Doyle, J.J., Doyle, J.L., 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem. Bull. Bot. Soc. Am. 19, 11–15. Dutta, R., Deb, D.B., 2004. Taxonomic revision of Hedyotis L. (Rubiaceae) in Indian subcontinent. Botanical Survey of India, Kolkata, India. Farris, J.S., Källersjö, M., Kluge, A.G., Bult, C., 1995. Testing significance of incongruence. Cladistics 10, 315–319. Felsenstein, J., 1968. Statistical Inference and the Estimation of Phylogenies. D. Phil. Thesis, University of Chicago, Chicago, USA. Fosberg, F.R., 1943. The Polynesian species of Hedyotis (Rubiaceae). Bernice P. Bishop Mus. Bull. 147, 1–102. Fosberg, F.R., Sachet, M.-H., 1991. Studies in Indo-Pacific Rubiaceae. Allertonia 6, 191–278. Gielly, L., Taberlet, P., 1994. The use of chloroplast DNA to resolve plant phylogenies: noncoding versus rbcL sequences. Mol. Biol. Evol. 11, 769–777. Govaerts, R., Ruhsam, M., Andersson, L., Robbrecht, E., Bridson, D., Davis, A., Schanzer, I., Sonké, B., 2008. World Checklist of Rubiaceae. The Board of Trustees of the Royal Botanic Gardens, Kew, Kew. Available from: <http:// www.kew.org/wcsp/rubiaceae/> accessed September 2008. Groeninckx, I., Vrijdaghs, A., Huysmans, S., Smets, E., Dessein, S., 2007. Floral ontogeny of the Afro-Madagascan genus Mitrasacmopsis with comments on the development of superior ovaries in Rubiaceae. Ann. Bot. 100, 41–49. Groeninckx, I., Dessein, S., Ochoterena, H., Persson, C., Motley, T.J., Kårehed, J., Bremer, B., Huysmans, S., Smets, E., in press. Phylogeny of the herbaceous tribe Spermacoceae (Rubiaceae) based on platid DNA data. Ann. MO Bot. Gard. Guindon, S., Gascuel, O., 2003. A simple, fast and accurate method to estimate large phylogenies by maximum-likelihood. Syst. Biol. 52, 696–704. Author's personal copy 866 J. Kårehed et al. / Molecular Phylogenetics and Evolution 49 (2008) 843–866 Gustafsson, C., Persson, C., 2002. Phylogenetic relationships among species of the neotropical genus Randia (Rubiaceae, Gardenieae) inferred from molecular and morphological data. Taxon 51, 661–674. Halford, D.A., 1992. Review of the genus Oldenlandia L. (Rubiaceae) and related genera in Australia. Austrobaileya 3, 683–722. Hallé, N., 1966. Famille des Rubiacées, pt. 1. III. Hedyotidées. In: Aubréville, A. (Ed.), Flore du Gabon. Muséum National d’Histoire Naturelle, Paris, pp. 75–124. Hillis, D.M., 1995. Approaches for assessing phylogenetic accuracy. Syst. Biol. 44, 3– 16. Hooker, J.D., 1873. Rubiaceae. In: Bentham, G., Hooker, J.D. (Eds.), Genera Plantarum. Lovell Reeve and Co., Williams and Norgate, London, pp. 7–151. Hsienshui, L., Wancheung, K., Weichiu, C., Hsianghao, H., Hen, W., 1999. Angiospermae. Dicotyledoneae. Rubiaceae (1). In: Hsienshui, L. (Ed.), Flora Reipublicae Popularis Sinicae, Tomus 71 (1). Science Press, Beijing, pp. 1–427. Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17, 754–755. Huelsenbeck, J.P., Ronquist, F., Nielsen, R., Bollback, J.P., 2001. Bayesian inference of phylogeny and its impact on evolutionary biology. Science 294, 2310–2314. Huelsenbeck, J.P., Larget, B., Miller, R.E., Ronquist, F., 2002. Potential applications and pitfalls of Bayesian inference of phylogeny. Syst. Biol. 51, 673–688. Jovet, P., 1941. Aux confins des Rubiacées et des Loganiacées. Notulae Systematicae 10, 39–56. Kårehed, J., Bremer, B., 2007. The systematics of Knoxieae (Rubiaceae)—molecular data and their taxonomic consequences. Taxon 56, 1051–1076. Lewis, W.H., 1962. Phylogenetic study of Hedyotis (Rubiaceae) in North America. Am. J. Bot. 49, 855–865. Lewis, W.H., 1965a. Cytopalynological studies of African Hedyotideae (Rubiaceae). Ann. MO Bot. Gard. 52, 182–211. Lewis, W.H., 1965b. Pollen morphology and evolution in Hedyotis subg. Edrisia (Rubiaceae). Am. J. Bot. 52, 257–264. Löhne, C., Borsch, T., 2005. Molecular evolution and phylogenetic utility of the petD group II Intron: a case study in basal angiosperms. Mol. Biol. Evol. 22, 317–332. Mantell, D.E., 1985. The Afro-South-West Asiatic Genus Kohautia Cham. & Schlechtd. (Rubiaceae—Rubioideae—Hedyotideae): Morphology, Anatomy, Taxonomy, Phytogeography and Evolution. D. Phil. Thesis, Universität Wien, Vienna, Austria. Markey, A.S., de Lange, P.J., 2003. A new species of Coprosma Sect. Acerosae (Rubiaceae) endemic to the South and Stewart Islands of New Zealand. NZ J. Bot. 41, 459–473. Merrill, E.D., Metcalf, F.P., 1942. Hedyotis Linnaeus versus Oldenlandia Linnaeus and the status of Hedyotis lancea Thunberg in relation to H. consanguinea Hance. J. Arnold Arbor. 23, 226–230. Mort, M.E., Archibald, J.K., Randle, C.P., Levsen, N.D., O’Leary, T.R., Topalov, K., Wiegand, C.M., Crawford, D.J., 2007. Inferring phylogeny at low taxonomic levels: utility of rapidly evolving cpDNA and nuclear ITS loci. Am. J. Bot. 94, 173–183. Motley, T.J., 2003. Phylogeny of Hawaiian and Pacific Hedyotis (Rubiaceae): fruit evolution and the implications for conservation and genomics. Abstracts of Annual Meeting, Mobile, Alabama, Botany 2003, pp. 88–89. Motley, T.J., Wurdack, K.J., Delprete, P.G., 2005. Molecular systematics of the Chiococceae–Catesbaeeae complex (Rubiaceae): flower and fruit evolution and systematic implications. Am. J. Bot. 92, 316–329. Motley, T.J., in press. Phylogeny of Kadua (=Hedyotis): origins, biogeography, and subgeneric classification. Ann. MO Bot. Gard. Musters, W., Boon, K., Van Der Sande, C.A.F.M., Van Heerikhuizen, H., Planta, R.J., 1990. Functional analysis of transcribed spacers of yeast ribosomal DNA. EMBO J. 9, 3989. Müller, K., 2005. SeqState: primer design and sequence statistics for phylogenetic DNA datasets. Appl. Bioinformatics 4, 65–69. Müller, K.F., Borsch, T., Hilu, K.W., 2006. Phylogenetic utility of rapidly evolving DNA at high taxonomical levels: contrasting matK, trnT-F, and rbcL in basal angiosperms. Mol. Phylogenet. Evol. 41, 99–117. Negrón-Ortiz, V., Watson, L.E., 2002. Molecular phylogeny and biogeography of Erithalis (Rubiaceae), an endemic of the Caribbean Basin. Pl. Syst. Evol. 234, 71– 83. Nepokroeff, M., Sytsma, K.J., Wagner, W.L., Zimmer, E.A., 2003. Reconstructing ancestral patterns of colonization and dispersal in the Hawaiian understory tree genus Psychotria (Rubiaceae): a comparison of parsimony and likelihood approaches. Syst. Biol. 52, 820–838. Nixon, K.C., Carpenter, J.M., 1996. On simultaneous analysis. Cladistics 12, 221–241. Nylander, J.A.A., 2004. MrAIC.pl. Program Distributed by the Author. Evolutionary Biology Centre. Uppsala University, Uppsala, Sweden. Oxelman, B., 1996. RAPD patterns, nrDNA ITS sequences and morphological patterns in Silene section Sedoideae (Caryophyllaceae). Pl. Syst. Evol. 201, 93– 116. Paradis, E., Claude, J., Strimmer, K., 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20, 289–290. Penny, D., Hendy, M.D., 1985. The use of tree comparison metrics. Syst. Zool. 34, 75– 82. Persson, C., 2000. Phylogeny of the neotropical Alibertia group (Rubiaceae), with emphasis on the genus Alibertia, inferred from ITS and 5S ribosomal DNA sequences. Am. J. Bot. 87, 1018–1028. Popp, M., Oxelman, B., 2001. Inferring the history of the polyploid Silene aegaea (Caryophyllaceae) using plastid and homoeologous nuclear DNA sequences. Mol. Phylogenet. Evol. 20, 474–481. R Development Core Team, 2007. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Razafimandimbison, S.G., Moog, J., Lantz, H., Maschwitz, U., Bremer, B., 2005. Re-assessment of monophyly, evolution of myrmecophytism, and rapid radiation in Neonauclea s.s. (Rubiaceae). Mol. Phylogenet. Evol. 34, 334. Ridsdale, C.E., 1998. Rubiaceae. In: Dassanayake, M.D. (Ed.), A revised Handbook to the Flora of Ceylon. A.A. Balkema, Rotterdam/Brookfield, pp. 141–343. Robbrecht, E., 1988. Tropical woody Rubiaceae. Op. Bot. Belg. 1, 1–271. Robbrecht, E., 1994. Supplement to the 1988 outline of the classification of the Rubiaceae. Index to genera. Op. Bot. Belg. 6, 173–196. Robbrecht, E., Manen, J.-F., 2006. The major evolutionary lineages of the coffee family (Rubiaceae, angiosperms). Combined analysis (nDNA and cpDNA) to infer the position of Coptosapelta and Luculia, and supertree construction based on rbcL, rps16, trnL-F and atpB-rbcL data. Syst. Geogr. Pl. 76, 85–146. Robinson, D.F., Foulds, L.R., 1981. Comparison of phylogenetic trees. Math. Biosci. 53, 131. Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19, 1572–1574. Rzhetsky, A., Nei, M., 1992. A simple method for estimating and testing minimumevolution trees. Mol. Biol. Evol. 9, 945–967. Sastri, D.C., Hilu, K., Appels, R., Lagudah, E.S., Playford, J., Baum, B.R., 1992. An overview of evolution in plant 5S DNA. Pl. Syst. Evol. 183, 169–181. Scheltens, A., 1998. Pollenmorfologische studie van de Afrikaanse Hedyotideae (Rubiaceae). D. Phil. Thesis, Katholieke Universiteit Leuven, Leuven, Belgium. Schumann, K.M., 1888. Rubiaceae iia. In: von Martius and auct. suc., C.F.P. (Eds.), Flora Brasiliensis 6. R. Oldenburg, Munich, Wien, and Leipzig, pp. 1-124. Schumann, K.M., 1891. Rubiaceae. In: Engler, A., Prantl, K. (Eds.), Die natürlichen Pflanzenfamilien, vol. 4 (4). Engelmann, Leipzig, pp. 1–156. Shaw, J., Lickey, E.B., Beck, J.T., Farmer, S.B., Liu, W., Miller, J., Siripun, K.C., Winder, C.T., Schilling, E.E., Small, R.L., 2005. The tortoise and the hare II: relative utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis. Amer. J. Bot. 92, 142–166. Shaw, J., Lickey, E.B., Schilling, E.E., Small, R.L., 2007. Comparison of whole chloroplast genome sequences to choose noncoding regions for phylogenetic studies in angiosperms: the tortoise and the hare III. Amer. J. Bot. 94, 275–288. Simmons, M.P., Ochoterena, H., 2000. Gaps as characters in sequence-based phylogenetic analyses. Syst. Biol. 49, 369–381. Soltis, D.E., Soltis, P.S., 1998. Choosing an approach and an appropriate gene for phylogenetic analysis. In: Soltis, D.E., Soltis, P.S., Doyle, J.J. (Eds.), Molecular Systematics of Plants II—DNA Sequencing. Kluwer Academic Publishers, Boston, Dordrecht, London, pp. 1–42. Steel, M.A., Penny, D., 1993. Distributions of tree comparison metrics-some new results. Syst. Biol. 42, 126–141. Strand, A.E., Leebens-Mack, J., Milligan, B.G., 1997. Nuclear DNA-based markers for plant evolutionary biology. Mol. Ecol. 6, 113–118. Terrell, E.E., 1975. Relationships of Hedyotis fruticosa L. to Houstonia L. and Oldenlandia L. Phytologia 31, 418–424. Terrell, E.E., 1987. Carterella (Rubiaceae), new genus from Baja California, Mexico. Brittonia 39, 248–252. Terrell, E.E., 1991. Overview and annotated list of North American species of Hedyotis, Houstonia, Oldenlandia (Rubiaceae), and related genera. Phytologia 71, 212–243. Terrell, E.E., 2001a. Taxonomy of Stenaria (Rubiaceae: Hedyotideae), a new genus including Hedyotis nigricans. Sida 19, 591–614. Terrell, E.E., 2001b. Stenotis (Rubiaceae), a new segregate genus from Baja California, Mexico. Sida 19, 899–911. Terrell, E.E., 2001c. Taxonomic review of Houstonia acerosa and H. palmeri, with notes on Hedyotis and Oldenlandia (Rubiaceae). Sida 19, 913–922. Terrell, E.E., Robinson, H.E., Wagner, W.L., Lorence, D.H., 2005. Resurrection of genus Kadua for Hawaiian Hedyotidinae (Rubiaceae), with emphasis on seed and fruit characters and notes on South Pacific species. Syst. Bot. 30, 818–833. Thulin, M., Bremer, B., 2004. Studies in the tribe Spermacoceae (RubiaceaeRubioideae): the circumscriptions of Amphiasma and Pentanopsis and the affinities of Phylohydrax. Pl. Syst. Evol. 247, 233–239. Verdcourt, B., 1958. Remarks on the classification of the Rubiaceae. Bull. Jard. Bot. État. 28, 209–281. Verdcourt, B., 1976. Rubiaceae (part 1). In: Polhill, R.M. (Ed.), Flora of Tropical East Africa. Crown Agents for Oversea Governments and Administrations, London, pp. 1–414. Waterman, M.S., Smith, T.F., 1978. On the similarity of dendrograms. J. Theor. Biol. 73, 789–800. Wenzel, J.W., Siddall, M.E., 1999. Noise. Cladistics 15, 51–64. White, T.J., Bruns, T., Lee, S., Taylor, J., 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: Innis, M., Gelfand, D., Sninsky, J., White, T. (Eds.), PCR protocols: a guide to methods and applications. Academic Press, San Diego, pp. 315–322.