American Journal of Botany 96(4): 816–852. 2009. PHYLOGENY OF THE TRIBE INDIGOFEREAE (LEGUMINOSAE–PAPILIONOIDEAE): GEOGRAPHICALLY STRUCTURED MORE IN SUCCULENT-RICH AND TEMPERATE SETTINGS THAN IN GRASS-RICH ENVIRONMENTS1 Brian D. Schrire,2,6 Matt Lavin,3 Nigel P. Barker,4 and Félix Forest5 2The Herbarium, Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3AB, UK; 3Plant Sciences and Plant Pathology, 119 Plant Bioscience Building Montana State University, Bozeman, Montana 59717 USA; 4Molecular Ecology and Systematics Group, Department of Botany, Rhodes University, Grahamstown 6140, South Africa; and 5The Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3DS, UK This analysis goes beyond many phylogenies in exploring how phylogenetic structure imposed by morphology, ecology, and geography reveals useful evolutionary data. A comprehensive range of such diversity is evaluated within tribe Indigofereae and outgroups from sister tribes. A combined data set of 321 taxa (over one-third of the tribe) by 80 morphological characters, 833 aligned nuclear ribosomal ITS/5.8S sites, and an indel data set of 33 characters was subjected to parsimony analysis. Notable results include the Madagascan dry forest Disynstemon resolved as sister to tribe Indigofereae, and all species of the large genus Indigofera comprise just four main clades, each diagnosable by morphological synapomorphies and ecological and geographical predilections. These results suggest niche conservation (ecology) and dispersal limitation (geography) are important processes rendering signature shapes to the Indigofereae phylogeny in different biomes. Clades confined to temperate and succulent-rich biomes are more dispersal limited and have more geographical phylogenetic structure than those inhabiting tropical grass-rich vegetation. The African arid corridor, particularly the Namib center of endemism, harbors many of the oldest Indigofera lineages. A rates analysis of nucleotide substitutions confirms that the ages of the oldest crown clades are mostly younger than 16 Ma, implicating dispersal in explaining the worldwide distribution of the tribe. Key words: biogeography; character evolution; dispersal limitation; Fabaceae; Indigofera; Indigofereae; Leguminosae; molecular phylogeny; rates analysis. Indigofera with a worldwide distribution and c. 750 species is the third largest genus in legumes (Schrire, 2005a) and 19th (Mabberley, 1997) to 26th (Stevens, 2001) in size of all angiosperm genera. Centers of diversity occur primarily in Africa and Madagascar (c. 550 species), Asia, especially the temperate Sino-Himalayan region (c. 105 species), Australia (c. 50 species), and the New World (c. 45 species). Although the tribe Indigofereae has been subjected to phylogenetic analysis (Barker et al., 2000; Schrire et al., 2003), only about 12% of the tribe and 12% of the genus Indigofera were sampled for molecular and morphological data, leaving many taxonomic and biogeographical findings in need of validation. Furthermore, Indigofereae is now well resolved as sister to a large clade 1 comprising tribes Millettieae, Abreae, and Phaseoleae (Wojciechowski et al., 2004; Lewis et al., 2005). As such, more appropriate outgroups can now be included in phylogenetic analyses of the tribe. Also, age estimates derived from a comprehensive rates analysis of the legume family (e.g., Lavin et al., 2005) are now available with which to fix the age of the Indigofereae stem node. Such estimates are necessary for understanding the time frame in which the ecological and geographic phylogenetic structure of the Indigofereae phylogeny, detected even during the initial phases of this study (e.g., Schrire et al., 2003), was attained. The combined nuclear ribosomal sequence (ITS/5.8S) and morphological data set of Schrire et al. (2003) was clearly able to resolve monophyletic subgroups of Indigofereae that showed ecological and geographical integrity. For example, the analysis included three separate dispersal events from the Old into the New World (one in dry succulent-rich and two in grass-rich environments), one diversification into the temperate Cape region of southern Africa, and a large and very recent dispersal event into temperate Asia. Also notable among these findings was that each of the over 750 species of Indigofera could belong potentially to just one of four subclades, each imprinted with a unique geographical range and subset of habitat preferences. Taxonomic progress in such a large group as Indigofereae can only proceed when monophyletic subgroups are circumscribed and targeted for study. It was important to ascertain, therefore, if the preliminary finding by Schrire et al. (2003), that all species of Indigofera form only four main subclades, was an artifact of undersampling. The aims of this study are to conduct a phylogenetic analysis of a greatly expanded Indigofereae data Manuscript received 3 June 2008; revision accepted 3 December 2008. The authors thank P. Linder for his much valued support since the start of the Indigofereae project and for subsequent advice on the KöppenGeiger climate system; P. Wilson for material, advice, and information on the Australian species; M. Thulin for material and useful discussion; N. Veitch for clarifying the taxonomy inherent in phytochemical nomenclature; T. Pennington and P. Coley for ecological insights into the role of plant defense strategies; D. Bellstedt for very helpful discussion on Zygophyllum and African arid corridor disjunctions; P. Craven for clarifying aspects of Namibian phytogeography; A. Tucker for much help with phytochemical bibliographies; S. Howis and S. Ramdhani for additional ITS sequences of Cape species; G. Lewis, R. Polhill and C. Stirton for their longstanding support, and the curators of K, GRA, NBG and MONT for use of specimens and analyzing leaf material for DNA sequence variation. The authors also thank M. Simmons for constructive review comments. 6 Author for correspondence (e-mail: b.schrire@kew.org) doi:10.3732/ajb.0800185 816 April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae set, together with an evolutionary rates analysis, to validate the monophyly of Indigofereae subgroups; to investigate how morphology, ecology, and geography have imposed phylogenetic structure on Indigofera; and how this structure reveals significant new data on the evolution of one of the most species rich genera of legumes. This comprehensive review provides a context for ongoing and future evolutionary studies of this legume group and in general to species-rich pantropical genera like Indigofera. MATERIALS AND METHODS Taxon sampling—The sample of 92 accessions of Indigofera, nine other Indigofereae (representing Phylloxylon, Cyamopsis, Indigastrum, Microcharis, and Rhynchotropis), and 12 outgroup genera in Schrire et al. (2003) was expanded to 274 accessions of Indigofera (266 species along with duplicate accessions of eight of these), 27 species of other Indigofereae (representing the same five genera), and 18 outgroups (Appendix 1). This represents a threefold increase in taxon sampling over the Schrire et al. (2003) analysis. Sampling among the ingroup Indigofereae was guided by including the most divergent species from a taxonomic, ecological, and geographical perspective. Sampling was further designed to greatly expand each of the four main clades of Indigofera that were resolved in Schrire et al. (2003). For example, the least sampled palaeotropical clade was increased from 13 to 51 taxa, a fourfold increase in sampling. Among the outgroups, sampling was guided by the comprehensive phylogenetic analyses of Wojciechowski et al. (2004), which resolved a well supported Millettioid-Phaseoloid clade as sister to Indigofereae. Early-branching genera of the Millettioid-Phaseoloid clade were sampled as outgroups including Abrus Adans., Aganope Miq., Austrosteenisia Geesink, Canavalia DC., Clitoria L., Craibia Dunn, Craspedolobium Harms, Dalbergiella Baker f., Disynstemon R. Vig. Fordia Hemsl., Leptoderris Dunn, Millettia Wight & Arn., Ophrestia Forbes, Platycyamus Benth., Platysepalum Welw. ex Baker, Schefflerodendron Harms, Sylvichadsia Du Puy & Labat, and Xeroderris Roberty. The genus Disynstemon has never been allied with Indigofereae, but the results of preliminary ITS sequence analyses and trnK/matK sequence data (M. Lavin, unpublished data) prompted a close inspection of its morphology and biogeography. It was then determined to be a strong candidate sister group of Indigofereae. Morphological data—The 31 character data set of Schrire et al. (2003) has been refined and expanded to 80 binary and multistate characters (a nearly threefold increase in the number of characters delimited), which include vegetative, inflorescence and floral, pollen, and cytological characters (Appendix 2). All multistate characters were treated as unordered. Character numbers in Appendix 2 are cross-referenced in the text where they provide additional supporting data that should be read in conjunction with the discussion. Characters were scored for the species that were also sampled for DNA data. Herbarium specimens were the primary source of the morphological data, while field and cultivated specimens provided a secondary source. Literature reports were used to validate morphological character scorings taken from the primary and secondary sources. DNA data—DNA sequences from the nuclear ribosomal ITS and 5.8S region were sampled because this locus was shown by Schrire et al. (2003) to provide robust resolution at higher taxonomic levels within Indigofereae, including especially the resolution of intergeneric relationships and the monophyly of constituent genera and primary subclades within genera. The ITS/5.8S DNA sequence data of 833 aligned nucleotide sites were augmented in this analysis with the addition of 33 insertion-deletion (indel) characters (TreeBase accession S2193; http://gemini.oscs.montana.edu/~mlavin/data/indigo.txt), data not included in Schrire et al. (2003). Indels scored were unequivocally alignable and coded as binary (presence/absence) following Simmons and Ochoterena (2000). Sequences were aligned manually with the program Se-Al (Rambaut, 1996) using the similarity criterion of Simmons (2004). Relations were only reported where they were consistently resolved with high support, regardless of alternative alignments within only the midportion of the ITS-1 region and at the very 5′ end of the ITS-2 region. Such alternative alignments pertain mostly to just the outgroup sequences. No nucleotide sites were omitted from analysis of any alternative alignment. 817 Because of the large taxonomic size of the Indigofereae and the detailed resolution obtained from the ITS/5.8S region, we decided to increase taxon sampling rather than to sample additional genetic loci. With respect to inferring species phylogenies, the phylogenetic utility of nuclear ribosomal repeat sequences potentially can be compromised by the presence of incompletely homogenized paralogs and pseudogenes within individuals (e.g., Bailey et al., 2003). The entire ITS/5.8S region was thus PCR amplified and sequenced as reported in Schrire et al. (2003). An annealing temperature of 50°C and direct sequencing of PCR products was intended to detect when and where intraindividual ITS/5.8S sequence variation might arise. As in many papilionoid clades, all PCR products from Indigofereae and outgroups sequenced cleanly in both the forward and reverse directions (Northwoods DNA, Solway, Minnesota, USA). The 5.8S region of every sequence included no indels or divergent sequences. ITS/5.8S sequences were subjected to Bayesian Markov chain Monte Carlo (MCMC) analysis (Yang and Rannala, 1997) such that base frequencies and among-site substitution rates were estimated separately for each of the ITS1, 5.8S, and ITS2 regions. Highly similar estimates of frequencies of nucleotide bases and substitution types among these three regions suggested little evidence of pseudogenes, which have been detected in a few other subgroups of legumes, most notably the subfamily Mimosoideae (Bailey et al., 2003). Phylogenetic and evolutionary rates analysis—Parsimony analysis using the program PAUP* version beta 10 (Swofford, 2002) followed search strategies reported in Schrire et al. (2003) designed to detect the globally optimum solutions (e.g., maximum trees set at 10 000 and invoking steepest descent and retention of multiple parsimonious trees). A total of 10 000 bootstrap (Felsenstein, 1985) replicates was analyzed with less stringent search options (e.g., invoking neither steepest descent nor retention of multiple parsimonious trees) such that clade support values were biased low. Conflict among data sets was evaluated by an incongruence length difference test (Farris et al., 1995) using the settings as described for the bootstrap analysis (and excluding uninformative characters), and by comparing bootstrap trees from the individual data partitions, identifying incongruence only when conflicting clades were marked by bootstrap supports of over 80%, a conservative threshold. Bayesian analysis included two separate chains run for 107 generations, estimating topology, branch length, and substitution parameters separately for the ITS1, 5.8S, and ITS2 region and initially from uniform or random priors, and sampling parameter estimates every 105 generations. The final 50 trees from each of the two runs (which were at likelihood stationarity) were then combined and subjected to an evolutionary rates analysis (Appendix S1, see Supplemental Data with the online version of this article). Estimates of rates of substitution were converted to absolute ages using penalized likelihood (Sanderson, 2002) implemented in version 1.71 of the program r8s (Sanderson, 2004). The root age was fixed with both the maximum (55 Ma) and minimum (50 Ma) age estimates of the Indigofera stem clade reported in Lavin et al. (2005; node 56, the most recent common ancestor of Xeroderris and Indigofera). Assessing the influence of ecology and geography on phylogeny—Terminal taxa in this analysis are generally narrowly distributed within one of the four global biomes delineated by Schrire et al. (2005a, b; Appendix 2, character 84). The Indigofereae phylogeny was thus analyzed for ecological and geographical structure including altitude (Appendix 2, character 81), climate (character 82), continent (character 83) and biome. These four biogeographical and ecological characters where then optimized onto one of the most parsimonious cladograms with the program MacClade (Maddison and Maddison, 2005; Figs. 1–6). The ecological and geographical structure of the Indigofereae phylogeny was quantified using a Mantel regression approach with Euclidean geographical distances between pairwise comparisons of terminal taxa as the predictor variable and phylogenetic distance as the response (Legendre, 1990). Biome (grass, succulent, and temperate) served as an indicator variable. Terminal taxa in the speciose CRIM (Cyamopsis to Microcharis) and Indigofera clades were treated as samples and scored for longitude and latitude to derive Euclidean distances (Appendix 1; online Appendices S2 and S3). Species numbers in the outgroup genera and Phylloxylon were too low to give significant results and were thus not referenced with coordinates. Widespread species were scored for just the samples used in this analysis. Because geographical phylogenetic structure was determined to be low when the entire Indigofereae phylogeny was analyzed, pairwise comparisons were restricted to samples from the same continent. All regression analyses were conducted with the statistical program R (R Development Core Team, 2007) using the Base and Ecodist (Goslee and Urban, 2007) packages. American Journal of Botany 818 RESULTS The results of an incongruence length difference test suggested conflict between the molecular and morphological data sets (p = 0.000171). However, the bootstrap majority rule consensus trees generated for individual data partitions (Appendices S4 and S5, see Supplemental Data with the online version of this article) showed no conflicting clades that were supported by values of over 80%. Only the results of the combined analysis are therefore presented and discussed. Combined parsimony analysis— Analysis of the combined data set produced the maximum number of trees each with a length of 6164, an ensemble consistency index (CI; Kluge and Farris, 1969) of 0.213, and an ensemble retention index (RI; Farris, 1989) of 0.753. Of the 833 nucleotide sites, 33 indels, and 80 morphological (946 total) characters, all of the indel and morphological characters and 448 of the nucleotide sites (561 total) were parsimony informative. The strict consensus is fairly well resolved (Figs. 1–6) and inspection of the Adams consensus (Adams, 1972; online Appendix S6) reveals few if any terminal taxa or clades that are placed in highly variable positions among the most parsimonious set of trees. The morphological data alone provided a total length of 982, a CI of 0.108, and an RI of 0.774. This data is comparable to the molecular data (ITS/5.8S sequence and indel variation) having a total length of 5182, CI = 0.233, and RI = 0.748, with the nucleotide substitution variation (indels excluded) providing a total length of 5089, CI = 0.228, and RI = 0.738. Morphological characters with a relatively high retention index were very common, well distributed among the vegetative and reproductive traits, and had strong grouping power throughout the tree, including within and among all genera of tribe Indigofereae (Appendix 2). Rooting the tree with respect to Clitoria and Schefflerodendron resulted in the bulk of the remaining rainforest-centered outgroup genera (Figs. 1, 2) forming a weakly supported monophyletic group. The exception is the Madagascan Disynstemon, which forms the well-supported sister to the Indigofereae. In addition to the morphological synapomorphies that reveal this relationship (Fig. 1), a setting in Madagascan dry succulentrich forests is also shared by Disynstemon and Phylloxylon, the latter being the well-supported earliest branching clade within Indigofereae (Figs. 1, 2). This ecological and geographical setting is thus optimized as the ancestral condition of the tribe Indigofereae, representing a major switch in biome predilection from the millettioid outgroups. All the recently recognized Indigofereae genera were resolved as monophyletic with strong bootstrap support, as were generic groups such as the CRIM clade and its sister relationship to the genus Indigofera (Figs. 1, 2). The four main subclades of Indigofera, the palaeotropical, pantropical, Cape, and Tethyan clades (Figs. 1, 3–6), were resolved generally with high support. The exceptions are the succulent biome-centered Cape and Tethyan clades. Indigofera nudicaulis is the moderately supported sister to the rest of the well-supported Cape clade (Fig. 5). The monophyly of the Tethyan clade (Fig. 6) has been consistently resolved in all analyses up to this most comprehensively sampled one, but only with moderate support. The morphological character states (Appendix 2) mapped to the phylogeny (Figs. 1–6) are those with the highest consistency and retention indices and are fixed or nearly so in the indicated clade. Although these have various levels of homoplasy, [Vol. 96 deep-branching clades can often be well diagnosed with morphological apomorphies (Figs. 1–6). Ecology and geography, as much as morphology, tend to be phylogenetically structured. For example, species of Phylloxylon and the CRIM clade (Fig. 2) cluster according to seasonally dry succulent-rich (indicated with red branches) or grass-rich savanna (brown branches) habitats. The palaeotropical clade (Fig. 3) predominantly inhabits the grass biome (and is optimized as such with MacClade), as does the pantropical clade (Fig. 4). The pantropical clade, however, comprises notable clusterings of species inhabiting the temperate biome (indicated with blue branches) and the seasonally dry succulent biome (red branches). The Cape clade (Fig. 5), largely limited to the Greater Cape Floristic Region of southern Africa, is mostly temperate inhabiting (blue branches) but with early-branching lineages from succulent-rich vegetation in the adjacent KarooNamib region to the north. Finally, the Tethyan clade (Fig. 6) is optimized as inhabiting succulent-rich vegetation (red branches) but with three distinct clusters of grass-rich (brown branches) or succulent- and grass-rich vegetation (black branches) inhabiting species. Each of these grass biome inhabiting clades has a pantropical distribution unlike any of the other red-branched Tethyan subclades. Geographical setting can thus be unequivocally mapped on internal branches (Figs. 1–6), revealing that the Indigofereae phylogeny is structured as much geographically as it is ecologically. Bayesian and evolutionary rates analysis of the ITS/5.8S nucleotide sequence variation— The Bayesian analysis revealed that the substitution and base frequency parameter estimates are highly similar among the ITS1, 5.8S, and ITS2 regions (online Appendices S7 and S8). Because the 5.8S region has many more invariant sites, the relative substitution rate in this region is higher than in the ITS1 and ITS2 regions (Appendix S7). The PL rate smoothed Bayesian consensus (Fig. 7, depicting only two outgroup taxa, Xeroderris and Disynstemon, as well as Indigofereae) resolves very similar relationships to those of the parsimony consensus (Figs. 1–6) and mostly with well-supported relationships (i.e., posterior probabilities of 0.95 or higher). The Bayesian analysis of just the ITS/5.8S nucleotide data differs from the parsimony analysis of combined data mainly in resolving a trichotomy of Phylloxylon, the CRIM clade, and Indigofera, and by not placing I. nudicaulis as belonging to any one of the four main Indigofera subclades (Fig. 7). The evolutionary rates analysis using the 50 Ma fixed root age (i.e., the most recent common ancestor of Xeroderris and Indigofereae) reveals that age and rate estimates differ only slightly from those derived with a 55 Ma fixed root age, especially for nodes high in the tree (Fig. 8; online Appendix S9). With either of these two fixed root ages, much of the extant diversification of the tribe took place within about the last 16 Ma (Fig. 7; Appendix S9). This time frame encompasses all transoceanic sister clades (e.g., pantropical species and sister clades on different continents or continental fragments; see Figs. 2–6). The herbaceous, almost entirely African CRIM clade has the oldest age estimate of any crown clade within Indigofereae, whereas the shrubby Cape clade has by far the youngest age estimate when Indigofera nudicaulis is excluded (Appendix S9, nodes 4 and 39). Inclusion of I. nudicaulis as part of the Cape clade, however, renders its age estimate coeval with the Indigofera crown clade, or over 20 Ma (Fig. 7, nodes 5–6; Appendix S9). The Tethyan crown clade, predominantly of the succulent April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 819 Fig. 1. One of 10 000 most parsimonious phylograms derived from the analysis of the combined Indigofereae data set. The primary branches not resolved in the strict consensus are dashed. Green branches signify clades confined or nearly so to tropical wet forest settings (names in green signify taxa 820 American Journal of Botany biome, has by a slight margin the oldest age estimate (or next oldest after the Cape clade including I. nudicaulis) at about 15 Ma (compare nodes 10, 13, 14, and 39 in Fig. 7 and Appendix S9). Effects of ecology and geography on phylogeny— The Indigofereae phylogeny shows geographical structure more in temperate and succulent-rich inhabiting subclades than in grass-rich or savanna inhabiting ones. That is, geographical setting is generally a better predictor of phylogenetic relatedness within these two biomes than in the grass biome. This geographical structure is evinced by the results of the Mantel regressions (Fig. 9A–F). In general, the correlation between geographic and phylogenetic distance is weak when pairwise comparisons come from the grass biome (indicated in brown, Fig. 9). In contrast, the correlations for those comparisons within succulent (indicated in red) or temperate biomes (indicated in blue) are often stronger. For example, the Asian subclades of the pantropical clade (Fig. 9F) of Indigofera show much less variance for comparisons from the temperate biome than comparisons from the grass biome, which renders a higher correlation in the former. The notable exception to such geographic phylogenetic structure lies with the Cape clade (Fig. 9B), where pairwise comparisons of terminal taxa within the succulent biome show no correlation between geographic and phylogenetic distance. This exception is because much of the succulent biome in the Greater Cape Floristic Region and the Karoo-Namib region to the north comprises a number of narrowly confined, geographically and ecologically distinct vegetation types closely juxtaposed to each other and harboring very divergent species groups belonging to the Cape clade (Fig. 5). Only the succulent biome has been found to harbor such phylogenetically divergent taxa within narrow geographic confines, as exemplified by species such as I. nudicaulis and I. merxmuelleri. When the 77 nodes listed in Appendix S9 are classified according to their ecological predilection (e.g., inhabiting the grass, succulent, or temperate biomes, or a combination), the age distributions of clades reveal that the succulent biome harbors a much broader variance of clade-ages compared to the other two biomes (Fig. 10). Indeed, the succulent biome includes the oldest clades (Fig. 10D). This analysis reveals that the age distribution of clades strictly confined to either the grass (Fig. 10B) or the temperate biome (Fig. 10F) does not exceed 14 Ma. In contrast, clades strictly confined to the succulent biome have an age distribution that exceeds 14 Ma, including two clades that exceed 30 Ma (i.e., Disynstemon + Indigofereae, and Phylloxylon + the other five Indigofereae genera; see Fig. 7 and online Appendix S9, nodes 1 and 2). The Tethyan clade inhabits primarily the succulent biome (Fig. 6) and has relatively older constituent subclades compared to the other three Indigofera clades (note the generally lower numbered subclades within the Tethyan clade, as reported in Fig. 7 and online Appendix S9, where subclades are listed in order of estimated age). [Vol. 96 DISCUSSION The decision to greatly expand taxon sampling with respect to just the ITS/5.8S and morphological data has proved successful on several grounds. First, herbarium samples very commonly yielded ITS/5.8S products that sequenced cleanly. Second, the substitution rates estimated during this analysis (online Appendix S9; averaging 3.6 × 10−9 substitutions/site/ year for the 50 Ma fixed root age, and 3.0 × 10−9 substitutions/ site/year for the 55 Ma fixed root age) fit well within the expected range of values reported for other plant groups, i.e., 0.38–8.34 × 10−9 substitutions/site/year (Kay et al., 2006). Third, no pseudogenes or other paralogous ITS/5.8S sequences were detected in this analysis, as suggested by the homogeneity of substitution and base frequency estimates across the ITS1, 5.8S, and ITS2 region (online Appendix S8). According to Sanderson and Doyle (1992) and Bailey et al. (2003), this homogeneity would not be expected if concerted evolution among the nuclear ribosomal repeats was incomplete. Finally, our greatly expanded taxon sampling of morphology and ITS/5.8S sequences validates the main findings in Schrire et al. (2003), especially with regard to the generally high support values for the principle clades within the tribe Indigofereae. Phylogenetic relationships— The monophyly of tribe Indigofereae and all of its constituent genera and the sister relationship of the CRIM clade (Cyamopsis, Rhynchotropis, Indigastrum, and Microcharis) to Indigofera (Figs. 1, 2) endorse the relationships at the highest hierarchical levels posited by Schrire et al. (2003). The increase in outgroup sampling stabilized the position of the root node and the interrelationships of the main ingroup subclades, in agreement with Halanych (1998). The added sampling also led to the newly detected sister group relationship of the tribe with Disynstemon. Disynstemon— This monospecific genus endemic to Madagascar was placed in tribe Millettieae (Geesink, 1981, 1984; Du Puy and Labat, 2002b). Restricted to succulent-rich deciduous forests along dry river valleys in southwestern Madagascar, D. paullinioides is a woody liana bearing digitately trifoliolate leaves with stipellate leaflets, a character combination suggestive of genera placed in the basal Millettioid-Phaseoloid group (Du Puy and Labat, 2002b; Schrire, 2005b). The relationship is biogeographically notable because Phylloxylon, the first branching lineage within tribe Indigofereae, is also endemic in Madagascar (Du Puy and Labat, 2002a). Both Disynstemon and Phylloxylon were optimized as succulent biome taxa (Figs. 1, 2), and by extension the tribe Indigofereae is optimized as arising within this biome from Africa and Madagascar. This optimization represents a major switch in biome predilection from the rainforest biome inhabited by most millettioid outgroups. Because Disynstemon is well supported as sister to Indigofereae, the question of whether it should be included within the tribe must also be considered. Although Disynstemon + Indigo- ← confined to these settings); red branches (and names) signify clades (taxa) in succulent-rich settings, brown branches (and names) signify clades (taxa) in grass-rich settings, and black branches signify clades inhabiting a combination of biomes (here both succulent- and grass-rich settings). Morphological and ecological characters optimized to the ancestral node of the marked primary clades are indicated, and these generally have high retention indices (character and state numbers as listed in Appendix 2). Node-optimized characters are arranged in the following order: ecological, morphological (general), breeding system, defense, and dispersal-related. – , synapomorphies; =, parallelisms; ×, reversals; ε, possible synapomorphies based on equivocal parsimony optimizations of character states; *, nodes not supported in the Bayesian analysis. The patristic lengths were obtained with the default fast optimization “acctran” (Swofford, 2002). April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 821 822 American Journal of Botany fereae share the synapomorphies of an explosive flower tripping mechanism (Appendix 2, character 38; hereafter numbers in brackets refer to character numbers in Appendix 2, unless otherwise stated), the presence of pearl bodies (character 8) and a similar biogeographical affinity to the succulent biome, the presence of biramous hairs (character 2) is an iconic synapomorphy of tribe Indigofereae as well as other characters noted in Fig. 1. Until a more thorough analysis of outgroups is undertaken in the Millettioid-Phaseoloid alliance we do not alter the circumscription of the tribe. Phylloxylon— Similar to the case of Disynstemon, the phylogenetic relationship of Phylloxylon as sister to the rest of the tribe has strong bootstrap support in the combined analysis (Fig. 1), in contrast to Schrire et al. (2003). Analysis of just the ITS/5.8S sequence data resolved Phylloxylon in a trichotomy with the CRIM clade and Indigofera (Fig. 7). The increase in both character and taxon sampling has further strengthened this potential sister group relation of Phylloxylon, which is in agreement with the results of limited trnK/matK sequence data analyses (M. Lavin et al., unpublished data). Morphological data unequivocally suggest that Phylloxylon occupies the earliest branching lineage within Indigofereae because this genus harbors so many plesiomorphies with respect to Indigofereae (Fig. 1). These include a woody habit (Appendix 2, character 1), flowers bearing bracteoles (character 30), anthers lacking an apiculum but having a halo-like connective (53), and the presence of fusiform pods (63) that lack seed chambers (71) and bear large seeds (75) with a darkly pigmented testa (76). The small size of the pollen and the thin exine distinguishes Phylloxylon pollen from the rest of the tribe and is scarcely different from a generalized papilionoid type (Ferguson and Strachan, 1982). Phylloxylon comprises seven species of which at least five are narrowly restricted to succulent-rich deciduous forests in western and northern Madagascar. Two species occur at higher altitudes on the Central Plateau and the most widespread species in the genus, P. xylophylloides, occupies the more humid forest in the grass biome along the eastern margin of the Plateau. The presently well-resolved position of Phylloxylon belies its convoluted history (Du Puy et al., 1995; Du Puy and Labat, 2002a). The genus was first collected by Flacourt toward the end of the 17th century, and Jussieu (1789) referred it to a species of Xylophylla (Euphorbiaceae). Phylloxylon was eventually described by Baillon (1861) from this material, which had very immature flower buds, but he provisionally placed it in the Euphorbiaceae (Du Puy et al., 1995; Du Puy and Labat, 2002a). Baker (1883) initially described specimens of the genus as Exocarpus xylophylloides in the Santalaceae, but later from fertile material established the genus Neobaronia in the Leguminosae (Baker 1884), suggesting that it belonged in the tribe Dalber- [Vol. 96 gieae because of the indehiscent, one-seeded fruits. Subsequently, Harms (1900) united Phylloxylon and Neobaronia in the Leguminosae. Taubert (1894) and Hutchinson (1964) placed Phylloxylon in the Andira group in tribe Dalbergieae near Geoffroea (Du Puy et al., 1995). Phylloxylon was finally placed in tribe Indigofereae by Peltier (1967), a decision followed by Polhill (1981) and subsequent authors. CRIM clade— The two genera Indigastrum and Microcharis were treated as subgenera of Indigofera by Gillett (1958), but Schrire (1995) reinstated them as genera most closely related to Cyamopsis and Rhynchotropis in what was subsequently termed the CRIM clade (Barker et al., 2000). This clade comprises a distinct and relatively old diversification (online Appendix S9, node 4) confined almost entirely to Africa with the exception of Cyamopsis tetragonoloba (native to the succulent-rich vegetation of northwestern India), the pantropical Indigastrum parviflorum (Heyne ex Wight. & Arn.) Schrire, and two Madagascan endemics, Microcharis aphylla (R. Vig.) Schrire, Du Puy & Labat and M. phyllogramme (R. Vig.) Schrire, Du Puy & Labat. Indigofera— The fact that all species sampled belong to one of four monophyletic subgroups of Indigofera is remarkable. The resolution of four subclades of Indigofera (palaeotropical, pantropical, Cape, and Tethyan; Figs. 1, 3–6) with only the monophyly of the Tethyan clade supported at less than a strong level (Figs. 1, 6) agrees with the results of Schrire et al. (2003). It was strongly felt, however, that further extensive sampling that intentionally included morphologically, ecologically, and biogeographically divergent species would show that these few Indigofera subclades were an artifact of undersampling. Contrary to this, the additional extensive sampling in this analysis only reinforced these four Indigofera subclades as likely to be real. Any further sampling at this point is not predicted to break this pattern because only Indigofera species with very similar morphologies, ecologies, and geographies to those already in the data matrix remain to be sampled. Genera once segregated from Indigofera, including Bremontiera DC. and Vaughania S. Moore, are now firmly established as being derived from within Indigofera. The former monotypic genus Bremontiera comprises Indigofera ammoxylum, which is endemic to the island of Réunion (Polhill, 1990). Vaughania is an endemic Madagascan radiation of Indigofera species (Du Puy et al., 1994; Du Puy and Labat, 2002a). Remarkably, these two taxa are resolved as closely related, sometimes sister to each other, within the Tethyan clade, in the I. leucoclada– Vaughania humbertiana subclade (Fig. 6; online Appendix S9, node 19). This latter subclade contains the most morphologically distinctive species in the genus, which are all associated with endemic radiations on islands of the Indian Ocean (Schluter, 2000). The increased sampling of Vaughania species ← Fig. 2. Phylogram of the outgroups and the non-Indigofera clades of tribe Indigofereae; a portion of the phylogram from Fig. 1 comprising Phylloxylon and the CRIM clade (Cyamopsis, Rhynchotropis, Indigastrum, and Microcharis). The branches not resolved in the strict consensus are dashed. Green branches signify clades confined or nearly so to tropical wet forest settings (names in green signify taxa confined to these settings); red branches (and names) signify clades (taxa) in succulent-rich settings; brown branches (and names) signify clades (taxa) in grass-rich settings; blue branches (names) signify clades (taxa) in temperate settings, and black branches signify clades inhabiting a combination of biomes (mostly succulent- and grass-rich settings). Morphological, ecological, and geographical characters optimized to the ancestral node of the marked clades are indicated, and these generally have high retention indices (character and state numbers as listed in Appendix 2). Node-optimized characters are arranged in the following order: ecological, morphological (general), breeding system, defense, and dispersal-related. –, synapomorphies; =, parallelisms; ×, reversals; ε, possible synapomorphies based on equivocal parsimony optimizations of character states; *, nodes not supported in the Bayesian analysis. The patristic lengths were obtained with the default fast optimization “acctran” (Swofford, 2002). April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 823 Fig. 3. Phylogram of the palaeotropical clade of Indigofera; a portion of the phylogram from Fig. 1. The branches not resolved in the strict consensus are dashed. Brown branches signify clades confined or nearly so to grass-rich settings (names in brown signify taxa confined to these settings), red branches 824 American Journal of Botany has shown much stronger bootstrap support for its position embedded within Indigofera (Fig. 6) than in the analysis of Schrire at al (2003). The 11 species of Vaughania have been formally transferred to or reinstated within Indigofera (Schrire, 2008). Palaeotropical clade— So named for its distribution which is entirely absent from the New World, this clade is optimized as being restricted to Africa-Madagascar, in the grass biome (Fig. 1), in lowland areas less than 800 m in altitude, and to arid climates (Köppen-Geiger zones BSh, BWh and BWk; Kottek et al., 2006). With a crown age of c. 13.2 Ma (Appendix S9, node 14), the palaeotropical clade comprises some 190 species including 173 in Africa, nine endemic to Madagascar, about seven in Asia and one species in Australia. Synapomorphies of this clade (Figs. 1, 3) are the presence of hydathodes (Appendix 2, character 9), densely arranged pearl bodies (8), pods held erect (68), and pollen with a high tectal perforation density (77). In addition, parsimony character optimizations are equivocal about the presence of unequally biramous hairs (2), stipels (25) and red petals (33) also being synapomorphies of this clade (denoted by ε before the character states in Figs. 1, 3). In the African–Asian I. argentea–I. nyassica subclade (Appendix S9, node 46), I. argentea, restricted to the SaharaSindian region (White and Léonard, 1991), is an early-divergent succulent biome species sister to the rest of the grass-biomecentered diversification. This subclade is one of the most speciose in the genus with c. 80 species, occurring mainly in Africa, but one species, I. colutea, is pantropical, and three are endemic to Madagascar, two to Asia (Myanmar to China), and one to Australia. A switch in defenses to multicellular gland-tipped trichomes occurs in this subclade from dense pearl bodies or hydathodes (Fig. 3; Appendix 2, characters 6, 8, 9). The I. compressa–I. tetrasperma subclade (Appendix S9, node 25) also comprises the two early-branching succulent biome species, I. compressa endemic to Madagascar and I. phymatodea endemic to the northeast Horn of Africa, as well as the poorly resolved Asian (largely Indian) I. mysorensis–I. uniflora subclade. These elements point to the succulent biome underpinning diversification in this otherwise grass-biome-centered clade. Well-resolved groups within the I. compressa–I. tetrasperma subclade include the succulent and grass-biome-centered I. tanganyikensis–I. bainesii group (Appendix S9, node 58) and the I. microcalyx–I. tetrasperma subclade (Fig. 3). In the I. microcalyx–I. tetrasperma subclade, a further switch in defenses occurs with loss of hydathodes and a reversal to sparse pearl bodies (characters 8, 9). In the I. congesta–I. tetrasperma (and I. inhambanensis–I. strobilifera) subclades (Fig. 3), an imperceptible gradation of leaves that merge into foliar bracts (29) is notable, accompanied by a truncation in shoot extension leading to the aggregation of flowers (27) into clusters, capitula or paniculate inflorescences (see Appendix 2, characters 27, 69). [Vol. 96 Pantropical clade— This clade is optimized as being restricted to Africa-Madagascar, in the grass biome, in lowland areas less than 800 m in altitude, and to lowland equatorial climates (Köppen-Geiger zone Aw; Kottek et al., 2006), giving way to upland and montane warm temperate climates (KöppenGeiger zones Cw and Cf) in the I. socotrana–I. verruculosa subclade (Fig. 4). A reversal to lowland and upland tropical equatorial and arid climates characterizes the I. caloneura–I. verruculosa subclade. At a crown age of c. 13.3 Ma (Appendix S9, node 13), the pantropical clade is the most speciose of the four major Indigofera clades, with about 310 species, including c. 142 in Africa, 11 endemic to Madagascar, one to Socotra, about 26 to the New World, some 85 to Asia, and about 45 to Australia. Synapomorphies for the clade (Fig. 1) are the presence of brown, biramous hairs on the stems and inflorescences (character 3), often correlated with growing in upland and montane habitats; a hyaline, pubescent indumentum on the dorsal surface of the standard (43) and the pollinator orientation and tripping mechanism characters of a dense, keeled, upper margin fringe (46) and proximal wing petal crests with hairs (45). Pearl bodies densely scattered over the plant (8) is plesiomorphic for this clade. In addition, parsimony optimizations are equivocal about whether states 1 or 2 of character 35 (calyx lobe length) is a synapomorphy of the pantropical clade, although given the context of character state evolution state 2 appears to be more likely. Major trends in character evolution within this clade (Fig. 4) include a reversal from herbs to suffrutices in the first three branching subclades and to shrubs in the I. socotrana–I. verruculosa subclade (Fig. 4; Appendix 2, characters 1, 35, 58; Appendix S9, node 31). This subclade is also distinguished by a marked shift in character evolution associated with its diversification in the temperate biome. The global distribution of the pantropical clade is achieved initially through a phylogenetic grade of warm temperate (i.e., tropical montane to extratropical latitude) African diversifications from I. socotrana to I. roseocaerulea (Fig. 4), occurring mostly from the Afromontane regions of southern, eastern, and northeastern Africa (including Socotra) and upland Madagascar. Nested in this group is a Himalayan to montane western China and warm temperate eastern Asian radiation of c. 70–75 species distributed as far north as Korea and Japan. These species comprise the monophyletic I. cylindracea–I. atropurpurea subclade with a remarkably young crown age of c. 3.2 Ma (Appendix S9, node 69). In this subclade (and with tendencies in the tropical I. swaziensis–I. verruculosa subclade), the standard petal is persistent (38), folding over the ovary. The I. frutescens–I. langebergensis subclade (crown age c. 7.8 Ma; Appendix S9, node 31) represents the smallest and only Afromontane-derived element of Indigofera to diversify in the Cape region, with many characters in these species converging on other species in the Cape clade (Fig. 5). Early-divergent species in this Afromontane phylogenetic grade (Fig. 4) are I. socotrana (endemic to high altitudes in Socotra) and I. jucunda ← (names) signify clades (taxa) in the succulent-rich biome, and black branches signify clades inhabiting a combination of biomes (here both succulent- and grass-rich settings). Morphological, ecological, and geographical characters optimized to the ancestral node of the marked clades are indicated, and these generally have high retention indices (character and state numbers as listed in Appendix 2). Node-optimized characters are arranged in the following order: ecological, morphological (general), breeding system, defense, and dispersal-related. –, synapomorphies; =, parallelisms; ×, reversals; ε, possible synapomorphies based on equivocal parsimony optimizations of character states. *, nodes not supported in the Bayesian analysis. The patristic lengths were obtained with the default fast optimization “acctran” (Swofford, 2002). April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 825 826 American Journal of Botany (from eastern South Africa). Indigofera jucunda is also sister, among others, to the temperate eastern Asian and South African I. natalensis–I. frutescens subclades. While this grade extends to lower altitudes than the typical Afrotemperate distribution of Galley et al. (2007), a north to south migration is indicated for the origin of the Cape I. frutescens–I. langebergensis subclade, a direction considered rare in Cape clades compared with northward diversifications (Galley and Linder, 2006; Galley et al., 2007). The I. caloneura–I. verruculosa subclade, centered in the tropical grass and succulent biomes, is marked by a reversal of nearly all characters associated with temperate and higher altitude conditions (Fig. 4). The I. caloneura–I. sootepensis subclade has a largely lowland tropical Asian distribution. Although the sister clade to this is not well resolved, two subclades are apparent within it indicated by a difference in biogeographical predilection. A predominantly succulent biome I. arrecta–I. platycarpa subclade is sister to a tropical African and Australian grass biome subclade, I. swaziensis–I. verruculosa. The glycoside indican (34), which darkens leaves upon drying, is a synapomorphy of the commercial indigo-bearing I. arrecta–I. platycarpa subclade. Nested within this subclade, the I. cavallii–I. coerulea group in the northeast African–west Asian succulent biome is sister to the Neotropical I. conzattii–I. platycarpa subclade. This neotropical subclade represents the second oldest of three amphi-Atlantic disjunctions in the genus with a crown age of 6.4 ± 0.5 Ma (Fig. 7; online Appendix S9, node 31). The Australasian I. boviperda–I. verruculosa subclade has a crown age of c. 5.3 Ma (Appendix S9, node 62) and includes an endemic New Caledonian species yet to be described, sister to the Australian species I. australis (crown age c. 4.6 Ma; Appendix S9, node 65). Cape clade— This clade is optimized as being restricted to Africa, largely to the Greater Cape Floristic Region (Born et al., 2007; Appendix 3), in the succulent biome only and to arid zone climates (Köppen-Geiger zones BSh, BWh, BSk, and BWk; Kottek et al., 2006), in lowland areas less than 800 m in altitude (Fig. 5). The I. cytisoides–I. heterophylla subclade marks a significant shift to the temperate biome only and warm temperate zone summer dry climates (Köppen-Geiger zone Csb) at lowland and upland altitudes (to 2000 m). The I. meyeriana–I. heterophylla subclade is optimized by arid zone climates (Köppen-Geiger zones BSh, BWh, BSk, and BWk) and warm temperate zone, summer dry climates (Köppen-Geiger zone Csb). The Cape clade comprises some 86 species, of which c. 70– 75 are endemic within the Greater Cape Floristic Region. These species can be divided into two main groups. About 42 species from the phylogenetic grade between I. cytisoides to I. gifbergensis (Fig. 5) are restricted to the nutrient-poor soils of the fynbos region (Rebelo et al., 2006). Some 44 species in diverse groups including I. nudicaulis, I. merxmuelleri, and the I. [Vol. 96 Olygophyllae sp. nov.2–I. denudata and I. Digitatae sp. nov.–I. heterophylla subclades tend to occur on nutrient-rich soils both within and to the north and east of the Cape Floristic Region (e.g., Goldblatt and Manning, 2000). The relationship of Indigofera nudicaulis (Fig. 5), like that of Disynstemon (Figs. 1, 2), is well supported only in the combined analysis with 76% bootstrap support. A parsimony analysis of just the ITS/5.8S data places this species in a position similar to that depicted in the Bayesian consensus (Fig. 7). A parsimony constraints analysis of just the ITS/5.8S sequence data implementing the Templeton and Kishino–Hasegawa tests (Templeton, 1983; Kishino and Hasegawa, 1989) reveals that I. nudicaulis is not strongly resolved as sister to any of the four main Indigofera clades, suggesting it may be the only species of Indigofera not belonging to one of the four main subclades of this genus. A sister relationship to the rest of the Cape Clade, however, is bolstered not only by the support in the strict consensus tree of the combined analysis (Fig. 5), but also by being restricted biogeographically to the same narrow Huns–Orange group subcenter of endemism in southern Namibia as the next branching species I. merxmuelleri (Craven and Vorster, 2006; P. Craven, University of Stellenbosch, unpublished data; Appendix 3). In addition, I. nudicaulis shares the following Cape clade apomorphies (Fig. 1): a shrubby habit (1), calyx lobes shorter than the tube (35), presence of a long ovary/short style (58), keel petals with an acute apex (50), a broad keel beard (51), and the presence of spherical seeds (72). Harvey (1862) placed I. nudicaulis in his section Simplicifoliae, an artificial assemblage of species with simple leaves, including I. ovata (found much higher in the Cape clade in this analysis, Fig. 5) and I. obcordata (now placed in the Tethyan clade, Fig. 6). Indigofera nudicaulis, I. merxmuelleri, and the I. Olygophyllae sp. nov. 2–I. denudata subclade comprise arid-adapted shrubs largely restricted to the xeric shrublands of southern Namibia to the karroid shrublands and the Albany thicket vegetation types of southern Africa (Cowling et al., 2005; Hoare et al., 2006; Mucina et al., 2006). In the Cape I. cytisoides–I. heterophylla subclade, a major shift in morphology (Fig. 5) is also associated with becoming locked into the nutrient-poor, acidic soils and fire disturbance regime of the fynbos region (see Appendix 2, characters 1, 17, 26, 31, 59, 74, 76). Further modifications in the Cape I. ovata–I. heterophylla subclade (Appendix S9, node 39) are correlated with a range expansion to montane fynbos environments (characters 46, 49). A significant switch in morphology occurs in the Cape I. Brachypodae sp.nov. 9–I. brachystachya subclade (Appendix S9, node 72) to mass flowering with persistent petals and the development of dendritic hairs (Appendix 2, characters 4, 38). A reversal in sclerophylly (17) marks the Cape I. declinata–I. heterophylla subclade as does a general tendency for reduction in hairiness of flowers, correlated to a tolerance for a wider range of (particularly wet) habitats in the I. filifolia–I. gifbergensis subclade (characters 25, 36, 74; Fig. 5). The I. ← Fig. 4. Phylogram of the pantropical clade of Indigofera; a portion of the phylogram from Fig. 1. The branches not resolved in the strict consensus are dashed. Blue branches signify clades confined or nearly so to the temperate biome (names in blue signify taxa confined to these settings), red branches (names) signify clades (taxa) in succulent-rich settings, brown branches (names) signify clades (taxa) in grass-rich settings, and black branches signify clades inhabiting a combination of biomes (either temperate and grass-rich or succulent- and grass-rich biomes). Morphological, ecological, and geographical characters optimized to the ancestral node of the marked clades are indicated, and these generally have high retention indices (character and state numbers as listed in Appendix 2). Node-optimized characters are arranged in the following order: ecological, morphological (general), breeding system, defense, and dispersal-related. –, synapomorphies; =, parallelisms; ×, reversals; ε, possible synapomorphies based on equivocal parsimony optimizations of character states. The patristic lengths were obtained with the default fast optimization “acctran” (Swofford, 2002). April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 827 828 American Journal of Botany Digitatae sp. nov.–I. heterophylla subclade (Fig. 5) is adapted to a wide range of more nutrient-rich substrates in the Greater Cape Floristic Region (Appendix 2, characters 1, 17). A tendency toward denser pearl bodies on the stems (8) indicates further ant associations most probably enhancing protection. The I. cuneifolia–I. alpina subclade of the temperate biome extends eastward into grassy fynbos and beyond the Cape Floristic Region into montane grassland up to the Drakensberg, with outliers as far north as the Nyika Plateau in Malawi. Contrary to the case of the northern Afromontane-derived Cape I. frutescens–I. langebergensis subclade noted in the pantropical clade (Fig. 4), the I. cuneifolia–I. alpina subclade is a source of Afromontane taxa that have migrated and radiated northward from the Cape region. Tethyan clade— This clade is optimized as being restricted to Africa-Madagascar in the succulent biome only and to arid climates (Köppen-Geiger zones BSh, BWh, BSk, and BWk; Kottek et al., 2006), in lowland areas less than 800 m in altitude. Within the I. anabibensis–Vaughania humbertiana subclade (Fig. 6), the I. microcarpa–I. squalida group most notably extends to the succulent and grass biomes together, and to equatorial winter dry climates (Köppen-Geiger zone Aw). In the I. fanshawei–I. arabica subclade a temperate biome, Cape-centered I. depressa–I. glaucescens group is optimized by warm temperate summer dry and fully humid climates (Köppen-Geiger zones Csb and Cfa). Within the I. hiranensis–I. bongensis subclade, the succulent and grass biome-centered I. asperifolia–I. bongensis group extends into largely equatorial winter dry climates (Köppen-Geiger zone Aw). With a crown age of c. 15.5 Ma (Appendix S9, node 10), the Tethyan clade is morphologically the most diverse of the four main clades of Indigofera, comprising some 163 species, including c. 110 species in Africa, four endemic or near endemic to Socotra, 14 to Madagascar; one to Réunion, c. 15 to Asia, two to Australia and c. 17 to the New World. Morphological synapomorphies of the Tethyan clade (Fig. 1) are reflexed pods (67) that are tetragonous in cross section (62) and the loss of endocarp tannins (70), characters principally involved with a shift in seed defenses. General trends in morphology include simple or trifoliolate leaves (15) arising from a pinnate condition, alternate leaflets (16), a character restricted only to the Tethyan clade in Indigofera and to the succulent biome subclade in Indigastrum, stipules with a prominent midrib (22), a reduction in calyx lobe length (35), a shift from pink to redcarmine petals (34) usually with anther hairs absent in the latter (54), flatter or angular pods (62) where endocarp tannin loss (70) is substituted in part by the presence of pearl bodies (65), probably enhancing ant associations for seed protection, and a shift to wings, flanges or spines on the pods (64) to aid fruit dispersal. So called for its distribution that most closely tracks that of the global succulent biome (Schrire et al., 2005a), the Tethyan clade (Fig. 6) was renamed from the earlier boreotropical clade [Vol. 96 (e.g., Schrire et al., 2003) to avoid confusion with the boreotropics hypothesis (Lavin and Luckow, 1993; Schrire et al., 2005b). The neotropics, Madagascar, the Sahara-Sindian region, and African arid corridors (Van Zinderen Bakker, 1969; Appendix 3), which link southern Africa with the northeast Horn, harbor most of the species of the Tethyan clade. The predilection of Tethyan clade species for the succulent biome is violated only for widespread species and some secondary diversifications in the grass biome (e.g., the I. antunesiana–I. bongensis subclade) or temperate biome (e.g., the I. depressa–I. glaucescens subclade). Although relationships among the four main Indigofera clades are not well resolved, the Tethyan clade could be sister to the Cape clade. First, they are resolved together in the Bayesian analysis (Fig. 7). Second, they might have an ancestral affinity to the succulent biome given that early-branching lineages in the Cape clade (e.g., I. nudicaulis and I. merxmuelleri) have the same predilection (Fig. 5). The weak bootstrap support for this clade could in large part be due to the affinity of the constituent species for seasonally dry, succulent-rich vegetation. The evolutionary persistence of lineages in this particular vegetation (Lavin, 2006; Pennington et al., 2006b) would potentially render a small difference between the age or depth of the crown clade and the age or depth of the corresponding stem clade; hence a possible explanation for the low support for this stem-to-crown-node branch. Within the first major dichotomy in the Tethyan clade, which is represented by I. anabibensis–Vaughania humbertiana (Fig. 6), four distinctive subclades have very divergent evolutionary trends. First, the I. anabibensis–I. trigonelloides subclade is an African arid corridor diversification having a crown age of c. 11.7 Ma (Appendices 3 and S9, node 23), with early-diverging lineages centered in that area of the KarooNamib region (White, 1983) to the north of the Greater Cape Floristic Region (Born et al., 2007; Fig. 6). Indigofera trigonelloides is unusual at species level in having a disjunct African distribution between the northeast Horn and the southwest. Second, the sister I. fanshawei–I. arabica subclade has many early-divergent species also associated with African arid corridor disjunctions, particularly in the southwest, in that part of the Karoo-Namib region to the north of the Greater Cape Floristic Region. For example, with a crown age of c. 5.4 Ma (Appendix S9, node 61), I. pungens is sister to the temperate biome, Cape-centered I. depressa–I. glaucescens diversification (Fig. 6). The karroid vegetation, Cape Floristic Region, I. obcordata is an early-branching species sister to the largely Albany thicket vegetation I. disticha–I. sessilifolia subclade and also to a more widespread dry African–Asian I. praticola–I. arabica subclade. Third, the I. microcarpa–I. squalida subclade is sister to the endemic Madagascan–Réunion I. leucoclada–Vaughania humbertiana diversification. In the pantropical succulent and grass biome I. microcarpa–I. squalida subclade, with a crown age of c. 12.9 Ma (Appendix S9, node 19), I. microcarpa is widespread in the neotropics and Africa, while I. glandulosa–I. ← Fig. 5. Phylogram of the Cape clade of Indigofera; a portion of the phylogram from Fig. 1. The branches not resolved in the strict consensus are dashed. Blue branches signify clades confined or nearly so to the temperate biome (names in blue signify taxa confined to these settings) and red branches (names) signify clades (taxa) in succulent-rich settings. Morphological, ecological, and geographical characters optimized to the ancestral node of the marked clades are indicated, and these generally have high retention indices (character and state numbers as listed in Appendix 2). Node-optimized characters are arranged in the following order: ecological, morphological (general), breeding system, defense, and dispersal-related. –, synapomorphies; =, parallelisms; ×, reversals; ε, possible synapomorphies based on equivocal parsimony optimizations of character states; *, nodes not supported in the Bayesian analysis. The patristic lengths were obtained with the default fast optimization “acctran” (Swofford, 2002). April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 829 830 American Journal of Botany squalida represents an endemic Asian–Australian section of the genus. Fourth, the shrubby I. leucoclada–Vaughania humbertiana subclade, also with a crown age of c. 12.9 Ma (Appendix S9, node 19), contains the most morphologically distinctive species in the genus associated with endemic radiations on ocean islands. Relationships within this subclade remain poorly resolved (Fig. 6), but the Madagascan I. leucoclada–I. bemarahaensis group, the Réunion endemic tree species I. ammoxylum and the Madagascan species formerly placed in Vaughania are grouped together with the I. microcarpa–I. squalida subclade for the first time in this analysis. The second major dichotomy in the Tethyan clade represented by the I. hiranensis–I. bongensis subclade has a crown age of c. 13.1 Ma (Appendix S9, node 16). While early branching elements are poorly resolved (Fig. 6), it is likely that two main diversifications can be identified, each with a basal grade of succulent biome-centered species. The I. marmorata–I. jamaicensis subclade is concentrated in the northeast Horn of Africa extending to northwestern India and with an endemic species in Australia. This clade is sister to a pantropical succulent and grass biome-centered I. trita–I. jamaicensis subclade with an amphi-Atlantic crown age of c. 3.7 Ma (Fig. 7; Appendix S9, node 51), representing the first of two Tethyan clade Old World–New World disjunctions. The crown age of the neotropical I. guaranitica–I. jamaicensis element is c. 0.7 Ma (Appendix S9, node 77). The second I. schimperi–I. bongensis subclade includes the I. nephrocarpa–I. cordifolia group from the northeast Horn of Africa to the Sahara-Sindian region, Asia, and Australia, and the I. diphylla–I. diversifolia group from the northeast Horn of Africa to Madagascar and Réunion, which is sister to the largely grass biome, suffrutescent pyrophyte I. asperifolia–I. bongensis subclade. This subclade represents the second Tethyan Old World–New World disjunction with a crown age of c. 7.9 Ma (Appendix S9, node 45). The New World element, I. asperifolia–I. miniata has a crown age of c. 3.7 Ma (Appendix S9, node 68). Early-branching nodes are poorly resolved in the sister group to the New World element, but two subclades are resolved, the I. semitrijuga–I. spicata group is distributed in the succulent and grass biome and with an early-divergent species restricted to the succulent biome in the northeastern Horn of Africa to west Asia, and the African suffrutescent pyrophyte I. antunesiana–I. bongensis subclade distributed in the grass biome. A high level of exceptional morphology relative to the rest of the Tethyan clade marks the suffrutescent pyrophyte I. asperifolia–I. bongensis subclade. Such characters include loss of the keel beard (51) but retention of the keel upper margin fringe (46), the occurrence together of many-seeded pods (69) and calyx lobes (35) longer than twice the length of the tube (shorter calyx lobes, in often shrubby species, are associated with longer, many-seeded pods elsewhere in Indigofera), vestigial leaves at the base of the plant (23), and either simple leaves (15) or leaflets reduced in number on basal [Vol. 96 leaves (this combination of fire adapted characters is unique to this subclade), anther hairs (54) present in carmine-red petals (34), and the co-occurrence of pod pearl bodies (65) and endocarp tannins (70), except for loss of pearl bodies in the I. semitrijuga–I. spicata subclade. Significantly different selection pressures appear to have constrained this fire-adapted subclade with its extensive diversification in the grass biome. In summary at least three dispersal events of Indigofera have occurred from the Old World to the New World, comprising the I. cavallii–I. platycarpa (Fig. 4; Appendix S9, node 31), I. asperifolia–I. bongensis (Fig. 6; Appendix S9, node 45), and I. trita–I. jamaicensis subclades (Fig. 6; Appendix S9, node 51). The I. cavallii–I. platycarpa subclade contains all the commercial indigo-producing species. Three diversifications into the temperate Cape Floristic Region (two more than reported in Schrire et al., 2003) have been identified from three separate Indigofera clades: the I. cytisoides–I. heterophylla (Fig. 5; Appendix S9, node 39), I. depressa–I. glaucescens (Fig. 6; Appendix S9, node 61) and I. langebergensis–I. frutescens subclades (Fig. 4; Appendix S9, node 31). The single temperate Asian I. cylindracea–I. atropurpurea diversification (Fig. 4; Appendix S9; node 69) is resolved, and so is a separate tropical Asian diversification, I. caloneura–I. sootepensis (Fig. 4; Appendix S9, node 56). Most Australian species are resolved as a subclade (Appendix S9, node 62) that includes a species yet to be described from New Caledonia. Morphology as a predictor of phylogenetic relationships— The phylogenetic structure of morphological character states is indicated by an RI of 0.766 (Appendix 2), compared to an overall RI of 0.754 for the combined data analysis. This similarity in RI reveals that morphological data has the same ability to group subclades as molecular data, in agreement with Donoghue and Sanderson (1992) and much of the legume literature where such comparisons have been made (e.g., Lavin et al., 2001). Characters associated with breeding systems and plant defenses in Indigofera are discussed further to explore morphological phylogenetic structure in the four major subclades of the genus. Breeding systems— An explosive floral tripping mechanism (Arroyo, 1981) is the single most important functional feature dominating the architecture of the Indigofera flower (Appendix 2, characters 37, 38, 47, 53, and 54). In the palaeotropical clade (Fig. 3), distinguishing characters pertaining to the breeding system include the presence of red-carmine flowers (33) that lack anther hairs (54) and have a high density of perforations on the tectal surface of the pollen (77). This suite of characters is correlated with calyx lobes being longer than twice the length of the tube (35). It is not clear if the plesiomorphic pink petals in the I. kirkii–I. hermannioides subclade are homologous to those of other clades because such flowers lack anther hairs, ← Fig. 6. Phylogram of the Tethyan clade of Indigofera; a portion of the phylogram from Fig. 1. The branches not resolved in the strict consensus are dashed. Red branches signify clades confined or nearly so to succulent-rich settings (names in red signify taxa confined to these settings) blue branches (and names) signify clades (taxa) in temperate settings, brown branches (and names) signify clades (taxa) in grass-rich settings, and black branches signify clades inhabiting a combination of biomes (either temperate and grass-rich or succulent- and grass-rich biomes). Morphological, ecological, and geographical characters optimized to the ancestral node of the marked clades are indicated, and these generally have high retention indices (character and state numbers as listed in Appendix 2). Node-optimized characters are arranged in the following order: ecological, morphological (general), breeding system, defense, and dispersal-related. –, synapomorphies; =, parallelisms; ×, reversals; ε, possible synapomorphies based on equivocal parsimony optimizations of character states; 䊏, nodes supported only in the Bayesian analysis. The patristic lengths were obtained with the default fast optimization “acctran” (Swofford, 2002). April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 831 Fig. 7. A penalized likelihood rate-smoothed Bayesian consensus phylogeny of Indigofereae and selected outgroups (Xeroderris and Disynstemon). Time scale is calibrated with a fixed root age of 50 Ma. Numbers associated with selected subclades indicate the nodes listed in online Appendix S9. These subclades were selected because of their ecological or geographical integrity, and within Indigofera, the Tethyan clade centered mostly in the succulent-rich biome has the generally older subclades. and coding in these subclades is uncertain due to limited availability of material of some species. Unique shifts in breeding system characters occur in the I. congesta–I. tetrasperma and I. inhambanensis–I. strobilifera subclades (Fig. 3) with a marked truncation in shoot extension leading to the aggregation of flowers into clusters, capitula, or paniculate inflorescences (see Appendix 2, characters 27, 28, 29, 58, and 69). In the pantropical clade (Fig. 4), distinguishing breeding system characters include the presence of anther hairs (54) in pink flowers (33) having a low pollen tectal perforation density (77); calyx lobes about equaling or up to twice as long as the tube (58) and many-seeded pods (69). Interesting parallels in breeding system characters were noted when comparing the phylogenetic grade from I. longimucronata to I. roseocaerulea in the 832 American Journal of Botany [Vol. 96 Fig. 8. Penalized likelihood rate-smoothed (PL) age and rate estimates showing the estimates derived from the 55 Ma fixed root age as a function of those derived from a 50 Ma fixed root. Ages are given in Ma and rates in substitutions per site per Ma. pantropical clade (Fig. 4) with the I. cytisoides–I. heterophylla subclade in the Cape clade (Fig. 5), both of which have largely evolved within the context of temperate conditions. Convergent synapomorphies are the shrubby habit (1); the presence of pink petals (34) with a dense keel upper margin fringe of hairs (46), anther hairs (54) and a standard that is sometimes persistent (38) and has a pubescent rather than strigose dorsal surface (43); the presence of a distal wing beard (44) and a broad keel beard (51); a long ovary/short style with long pods and mostly short calyx lobes (58); and the occurrence of often long fruiting pedicels (31) and a glabrous ovary (59). Many of these characters such as the standard and wing indumentum, the dense keel fringe, longer fruiting pedicels, and glabrous ovary are almost entirely restricted in Indigofera to these subclades, indicating that they were most likely adaptive within the context of temperate environments. In the pantropical clade, dense and showy keel fringes occur together with often large flowers (32) and wing petal proximal crests (45) with hairs. The floral characters noted have the probable role of enhancing tripping mechanism efficiency (46) and pollinator attraction and orientation, possibly within environments of greater competition for pollinators (see character 26). The long fruiting pedicels are possibly adaptive for improved dispersal of the many-seeded, long pods and the glabrous (usually dark-colored) ovary has been discussed (59) as possibly enhancing seed development and maturation. In the pantropical clade, this suite of characters is lost in the largely tropical I. caloneura–I. verruculosa subclade (Fig. 4). A close correlation is apparent throughout the tribal phylogeny between the flower color and presence of anther hairs (see Appendix 2, characters 33, 54). In the Cape clade (Fig. 5), the distinguishing breeding system characters include the presence of anther hairs (54) in pink flowers (33) having a low pollen tectal perforation density (77), a dense upper keel margin fringe (46), and a long ovary/short style (58). Unique developments, which are mostly associated with enhanced competition for pollinators in the Cape fynbos environment, are peduncles longer than twice the length of the subtending leaf, persistent petals indicating a shift to massflowering from the plesiomorphic syndrome of a few flowers opening per raceme per day, which is correlated with a trap- lining method of pollinator visitation, and the presence of darkcolored keel tips (see Appendix 2, characters 26, 38, 49). A reversal to carmine-red petals (33) occurs in the I. cuneifolia–I. alpina subclade (Fig. 5). In the Tethyan clade (Fig. 6), the distinguishing breeding system features include retention of a low density of pollen tectal perforations (77) in species with carmine-red petals (33), with some exceptions noted in Fig. 6, and a short ovary/long style (58), which transforms to the long ovary/short style state, e.g., in the Madagascan–Réunion I. leucoclada–Vaughania humbertiana subclade. The many morphological apomorphies of this latter subclade (Fig. 6) are exemplified by the remarkable shift in flower morphology and pollination mechanism distinguishing, e.g., V. depauperata to V. cereghellii. The corolla becomes asymmetrical and distorted (40) and the spirally twisted keel curls upward in front of the standard, with the wing petals no longer forming a level platform (see Fig. 6; Appendix 2, characters 40, 55, 56, 61). Also in the Tethyan clade, a recurrent grouping of characters is found in the I. anabibensis–I. trigonelloides, I. microcarpa–I. squalida, I. nephrocarpa–I. cordifolia and the I. semitrijuga–I. spicata subclades (Fig. 6), usually concentrated in arid areas linked to the succulent biome. These are annual or perennial herbs (1) having small flowers in dense inflorescences with the wing and keel petals ± glabrous, i.e., the upper margin fringe (46) and keel beard (51) have been lost. Also, the standard petal is narrow, elongated, and hairy often only in the distal half, with a tapering claw and usually acute apices, and fruits are reduced to short, 1–4-seeded pods (69). These characters are also set in the broader context of densely hairy calyces with lobes generally more than twice the length of the tube (35) almost enclosing the flowers, and keels with acute apices (50). Loss of the keel beard and reduction of the area of hairs on the standard, and loss of the keel upper margin fringe are all associated with a shift to protection of these small and narrow flowers within dense inflorescences and hairy calyces. Alternative defenses against herbivores and microorganisms— Phylogenetic structure of traits within each of the four clades of Indigofera is exemplified by different suites of mor- April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 833 Fig. 9. Mantel regressions depicting phylogenetic distances between all pairwise comparisons of terminal branches in a specified subset of Indigofereae as a function of geographical distance. Specified subsets represent a combination of clade and continent affiliations of terminal taxa. Mantel correlations (and the 95% confidence interval) are reported in the upper left corner of each graph. Red signifies pairwise comparisons confined to the succulent-rich biome, brown signifies pairwise comparisons confined to the grass-rich biome, and blue signifies pairwise comparisons confined to the temperate biome. 834 American Journal of Botany [Vol. 96 Fig. 10. Age distributions of selected Indigofereae subclades that have ecological or geographic integrity (i.e., those indicated in Figs. 2–6 and 7 and online Appendix S9). The age distributions are parsed according to ecological affiliation. April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae phological and chemical characters associated with plant defenses. The ecological interactions between plants and phytophagous organisms or pathogens have produced a bewildering array of plant defenses including physical (morphological), biotic (e.g., extrafloral nectaries and ant associations) and chemical (secondary metabolite) characters, as well as phenological escape and reduced nutritional quality (Ehrlich and Raven, 1964; Koptur, 1985; Schrire, 1989; Coley et al., 2005; Brenes-Arguedas et al., 2008). The large divergence of defensive strategies evident among the four major clades in Indigofera suggests that herbivores and pathogens have imposed strong selection pressures on Indigofera species. In the palaeotropical clade (Figs. 1, 3), physical defenses include a combination of dense pearl bodies attracting ants (8); calyx lobe and leaflet hydathodes modified as extrafloral nectaries attracting ants (9); multicellular gland-tipped trichomes (with sticky secretions) scattered over the plant surfaces (6); an often dense, spreading, sericeous indumentum (2); calciumoxalate crystals often occurring together with thick bundle sheaths around the primary to tertiary veins of leaflets (Schrire, 1995); broad keel beards (51); and endocarp tannins around the seeds (70). Chemical defenses include the widespread presence of the nonprotein amino acid canavanine and the indole alkaloid indican, which is a glycoside precursor to the dye indigo (Bell, 1981; Hegnauer and Hegnauer, 2001), and a range of phenolic acids (i.e., benzoic acid and protocatachuic acid, Adinarayana and Sarada, 1987), flavonoids (i.e., flavone glycosides; Adinarayana and Sarada, 1987; Rajkapoor et al., 2007) and isoflavanoids (i.e., pterocarpans and isoflavans; Selvam et al., 2004). Nitropropionic acid esters, alkaloids, and toxic nonprotein amino acids and cyanogenic glycosides reported in the other three clades appear to be absent in the palaeotropical clade. It is noteworthy that this clade has developed a greater range of physical and biotic defenses compared to the more chemical-rich deterrents in the other clades. For example, of the 29 palaeotropical clade species of Indigofera surveyed for uses by Burkill (1995) in West Africa, nearly all were recorded as good quality fodder, forage and browse for stock with no toxic species listed and relatively few having medicinal properties. In the pantropical clade (Figs. 1, 4) physical defenses comprise dense pearl bodies (8), an often dense spreading indumentum (2), the presence of broad keel beards (51) and wing petal beards (44), persistent petals folding over the developing ovaries (38), and endocarp tannins present around the developing seeds. Chemical defenses include the glycoside indican (although in higher proportions [34] in the indigo producing I. arrecta–I. platycarpa subclade) and the nonprotein amino acid canavanine, and a range of nitropropionic acid esters (Garcez et al., 2003; Zhang et al., 2006), cyanogenic glycosides (Siegler et al., 1989), guanidine alkaloids (i.e., stizolamine; Yoshida and Hasegawa, 1977), phenolic acids (Hasan et al., 1989), phenolic glycosides (Aziz-ur-rehman et al., 2005), flavonoids (i.e., flavonols, flavones and anthocyanins; Bisby et al., 1994; Hegnauer and Hegnauer, 2001), isoflavonoids (i.e., rotenoids [Kamal and Mangla, 1993], which have otherwise been recorded in Phylloxylon [Boiteau, 1938] and the millettioid genera Lonchocarpus, Millettia, Chadsia, Tephrosia, Mundulea, and Derris [Du Puy et al., 1995]), tannins, lignans (Aziz-ur-rehman et al., 2005) and terpenoids (diterpenoids, Thangadurai et al., 2002; saponins and sterols, Burkill, 1995). Of the 15 pantropical clade species of Indigofera listed by Burkill (1995), this clade has by far the highest percentage of taxa with medicinal uses, likely indicative of its more complex chemistry. Many taxa produce the dye 835 indigo, including all of the commercial dye species. Very few species are used for forage or are grazed by stock. Isoflavonoid reports are the first records of these compounds occurring in the Indigofereae because they were recorded as absent in the tribe by Gomes et al. (1981). In the Cape clade (Figs. 1, 5), physical defenses include branch and/or inflorescence tips ending in spines (11); the presence of leaf sclerophylly (17); sparse (occasionally dense) pearl bodies scattered over the plant (8); a sometimes dense, spreading, sericeous indumentum (2); calcium oxalate crystals (Schrire, 1995); broad wing and keel petal beards (44 and 51); and endocarp tannins present around the developing seeds (70). Chemical characters include the widespread presence of the glycoside indican and the nonprotein amino acid canavanine, and a range of nitropropionic acid esters (Hegnauer and Hegnauer, 2001). Cape species have mostly been overlooked for biochemical surveys. In the Tethyan clade (Figs. 1, 6), physical defenses comprise branch and/or inflorescence tips ending in spines (11), the presence of sparse pearl bodies sometimes present on the lower leaf surfaces (7) and pods (65), calyx lobes longer than twice the length of the tube (35), which are correlated with small flowers protected by densely hairy calyces almost enclosing the flowers, calcium-oxalate crystals present in the leaves but with thickened bundle sheaths absent (Schrire, 1995), tetragonous or flattened pods (62) correlated with a loss of endocarp tannins around the developing seeds (70), and loss of the keel beard (51). Chemical deterrents include the widespread presence of the glycoside indican and the nonprotein amino acids canavanine and indospicine (Hegnauer and Hegnauer, 2001), and a range of nitropropionic acid esters (Hegnauer and Hegnauer, 2001), phenolic acids (Sharif et al., 2005), flavonoids (i.e., flavanones; Lima et al., 2003), isoflavonoids (i.e., 2-arylbenzofuran; Bisby et al., 1994) and terpenoids (saponins and sterols; Lima et. al., 2003). Burkill (1995) records the greatest proportion of toxic species occurring in Indigofera among the 14 Tethyan clade taxa listed, as well as a significant number of relatively low-grade indigo-yielding plants. Some species provide at best low-quality fodder, especially in the I. anabibensis–I. arabica subclade (Fig. 6). Sampling of Indigofera species for biochemical assays has barely included 40 species to date and has focused largely on the few known poisonous species (e.g., I. hendecaphylla Jacq. and I. linnaei) or the economically important indigo-producing I. arrecta–I. platycarpa clade (Fig. 4). The different suites of defensive strategies in the four Indigofera clades most likely represent trade-offs in resource allocation and constraints imposed by community (biome) traits (Koptur, 1985; Brenes-Arguedas et al., 2008). Indigofera falls into the “defense syndrome” of Coley et al. (2005), having a slow expansion of young leaves with low synchrony of leaf production, normal chloroplast development (i.e., not delayed until after a “colorful” leaf flush), and a high level of ant defenses owing to a more reliable food source for ants. The widespread occurrence of sparsely distributed pearl bodies (O’Dowd, 1982; Schrire, 1995), usually concentrated on new growth in the axils of leaves and between the leaflets, is the plesiomorphic condition in Indigofera for this ant–plant association. Densely arranged pearl bodies also scattered over the stems and inflorescences are synapomorphies for each of the two grass-biomecentered pantropical and palaeotropical clades (Fig. 1). The presence of specialized pearl bodies on the lower leaflet surfaces and pods is a synapomorphy of the I. microcarpa– 836 American Journal of Botany Vaughania humbertiana subclade, and pearl bodies on the pods only is a synapomorphy of the amphi-Atlantic I. asperifolia–I. bongensis subclade; both subclades occur in the Tethyan clade (Fig. 6). In the palaeotropical clade (Fig. 3), an association with ants is also developed secondarily with actively secreting hydathode extrafloral nectaries (9), making this feature an additional ant–plant relationship to pearl bodies. The presence of such extrafloral nectaries, whether stalked (as defines the I. kirkii–I. hermannioides subclade) or sessile (as defines the I. compressa–I. tetrasperma subclade) is correlated with the presence of dense pearl bodies on the plant surfaces, so both systems appear to reinforce each other. Ant associations with Indigofera are poorly documented and are in need of further research. Ecology and geography as predictors of phylogenetic relatedness— With respect to the biomes characterized by Schrire et al. (2005a, b), i.e., the dry tropical succulent biome, the savanna-type grass biome, and the temperate biome, visual inspection of the phylogenies with biome affinity superimposed reveal that species are likely to inherit their ancestral biome predilection and general geographical setting (niche conservatism; Harvey and Pagel, 1991). The Mantel regression analysis reveals that geographic phylogenetic structure is stronger in the succulent biome as compared to the grass biome (Fig. 9), most likely because of the long-term stability of the tropical dry succulent-rich biome (e.g., Lavin, 2006; Pennington et al., 2006a). The reasoning here is that the richness of succulent taxa (e.g., in the Cactaceae, Euphorbiaceae, or Aizoaceae) is a signature of little disturbance, particularly from fires (e.g., Schwilk et al., 1997). Although plants are able to disperse widely, immigrants into succulent-rich vegetation have to contend with persistent resident plants. Persistence of residence over evolutionary time allows endemic lineages to become common even in narrow geographical confines via ecological drift (Hubbell, 2001). Dispersal limitation is also a characteristic of the temperate biome in the Cape Floristic Region (e.g., Latimer et al., 2005; Etienne et al., 2006; Fig. 9B), which comprises fragmented islands of fynbos on nutrient-poor soils interdigitating with drier succulent-rich vegetation on nutrientrich soils. Under the fragmented island condition, a correlation between geographic and phylogenetic distance is expected to be higher in the nutrient-poor-soil-type metacommunity than in the savanna-type vegetation to the east of the Cape Floristic Region, where immigration is enhanced by the large continuous area that is prone to occasional and local eradication of resident plants. A characteristic of grass biome savanna communities relative to the more dispersal limited succulent and temperate biomes is that they harbor a wide range of Indigofereae clades. Coexisting Indigofera species are more likely to be distantly related in the grass biome. For example, the floras of Malawi, Zimbabwe and the Limpopo Province of South Africa (c. 633 000 km2) are almost entirely restricted to the Zambezian savanna (grass) biome of southern Africa (White, 1983). The flora comprises some 121 Indigofera and six Microcharis species that are not close relatives and represent most subclades from across the Indigofereae phylogeny, i.e., six Tethyan, five palaeotropical, and five pantropical subclades and one species from the Cape clade (Figs. 2–6). Species are thus represented from the full range of variation in the genus. The succulent-rich and temperate biomes, in contrast, harbor closely related species within narrow geographical localities. [Vol. 96 For example, the flora of Somalia (c. 638 000 km2) is almost entirely restricted to Somalia–Masai, succulent-rich vegetation (White, 1983; Schrire et al., 2005a, b) and comprises c. 50 Indigofera and eight Microcharis species (Thulin, 1993). Representation of Indigofera here is in four Tethyan subclades (but with 30% of all Somalian species in one subclade in a closely related group), two palaeotropical subclades (with a further 20% of all species in one subclade of closely related species), and one pantropical subclade containing a further 16% of all species clumped in a closely related group. Two-thirds of the species of Indigofera in Somalia are thus restricted to only three subclades in the genus. In Microcharis all the grass biome species are widely scattered within one of the two sister clades, while the Somalian species form a closely related group restricted to the other sister clade. The many distantly related sympatric Indigofera species inhabiting grass-rich savanna communities is analogous to the many coexisting species of the genus Inga in the rainforest biome (Richardson et al., 2001; Coley et al., 2005; Brenes-Arguedas et al., 2008), except that Inga represents an even more recent diversification with as great a diversity of defense strategies. Many legume rainforest genera also have substantial grass biome diversifications (Schrire et al., 2005a, b), indicating an often common dispersal-prone history in the evolution of such genera in both biomes. Although Inga is concentrated in rainforests while Indigofera is distributed almost exclusively outside this biome, niche-space partitioning based on the defense features could be driving species coexistence in Indigofera in the grass biome in similar ways to that proposed for Inga (Brenes-Arguedas et al., 2008). The occurrence of Indigofereae in the arid corridors of Africa— A minimum of 20 of the 30 Indigofereae clades that are largely restricted to the succulent biome (Fig. 10; Appendix S9), are directly associated with African arid corridor disjunctions (details provided in Appendix 3). Some of the oldest arid corridor crown ages also involve Africa–Asia disjunctions; for example in Cyamopsis, the disjunction between C. tetragonoloba in northwestern India and the African species (Appendix S9, node 8) is c. 18–19 Ma. Clades having a northeastern Horn of Africa center of diversity in the African arid corridor are marked be a range of discontinuous crown ages, i.e., c. 13.1 Ma for the Tethyan subclade I. hiranensis–I. bongensis (Appendix S9, node 16), c. 11.1 Ma for Microcharis (Appendix S9, node 26), and c. 1–9 Ma in, e.g., the pantropical I. cavallii–I. platycarpa subclade (Appendix S9, node 31) and the Tethyan I. semitrijuga–I. spicata subclade (Appendix S9, node 63). Similarly, clades with a southwestern Karoo-Namib center in the African arid corridor are marked by a comparable range of discontinuous crown ages. The Cape clade (I. nudicaulis–I. heterophylla; Appendix S9, node 6) has a crown age of c. 20.3 Ma, the Indigastrum crown is c. 14.4 Ma (Appendix S9, node 12) and the Tethyan subclade, I. anabibensis–I. arabica, is c. 11.5 Ma. The range of clades with crowns c. 1–9 Ma includes the Cape subclade I. Olygophyllae sp.nov.2–I. denudata (Appendix S9, node 66) and the Tethyan subclades I. obcordata–I. arabica (Appendix S9, node 50) and I. pungens–I. glaucescens (Appendix S9, node 61). Other legume examples dated for the northeastern center of the African arid corridor and also fitting this pattern include the Chapmannia and Zygocarpum crowns in Dalbergieae, at 10 ± 2.6 and 7.2 ± 1.3 Ma respectively (Lavin et al., 2000, 2004) and Wajira in the Phaseoleae at 5.7 ± 0.7 Ma (Thulin et al., 2004). April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae The existence of the African arid corridor also has important implications for the evolution of southern African, particularly the Greater Cape Floristic Region, taxa (Appendix 3). The anomalous position of Indigofera nudicaulis as an early-divergent sister to the rest of the Cape clade was highlighted earlier. This species is restricted within the Namib center of endemism (Born et al., 2007; Appendix 3) to the narrow Huns-Orange group subcenter (P. Craven, University of Stellenbosch, unpublished data), a similar distribution to that of I. merxmuelleri, the next branching species within the Cape clade (Fig. 5). In the Tethyan clade (Fig. 6), I. pungens is endemic more broadly to the Namib center, including the transitional northern Richtersveld region, and it is sister to the second temperate Cape diversification (Appendix S9, node 61). The early-branching I. Digitatae sp. nov. (Fig. 5) is another endemic from the northern Richtersveld to the southwest of the Namib center and the I. Digitatae sp. nov.–I. heterophylla subclade (Appendix S9, node 57) may, with further sampling, prove to be a third succulent biome-centered dispersal into the Cape with an earlydivergent sister species associated with the Namib center of the African arid corridor (Appendix 3). Two (possibly three) independent diversifications of Cape Indigofera species thus share similar origins in the Namib center. Other angiosperm examples that reiterate this pattern are Gazania and the core Arctotis group in the Asteraceae, with respective crown ages of c. 6.5 Ma and c. 4.6 Ma (Howis et al., 2009; R. J. McKenzie and N. P. Barker, Rhodes University, unpublished manuscript). Both have species associated with the Namib center sister to clades in the Greater Cape Floristic Region. A similar pattern is evident in the Geraniaceae, in Monsonia including Sarcocaulon (Touloumenidou et al., 2007) and in Pelargonium (Bakker et al., 2004). One clade in Pelargonium, for example, has elements typical of an overall succulent biome distribution (Schrire et al., 2005a, b), with one subclade comprising an early divergent grade of Namib center species and some later diverging Cape Floristic Region species, while in the other subclades species occur from the northeastern Horn of Africa (in Ethiopia and Socotra) and Madagascar. A number of examples of this pattern also occur throughout subfamily Zygophylloideae (Zygophyllaceae). The first branching Melocarpum and Fagonia clades in the analyses of Bellstedt et al. (2008) show arid corridor links between the southwestern Karoo-Namib region (White, 1983) and the northeastern Horn of Africa (e.g., Beier et al., 2004). Early-divergent elements sister to a monophyletic Zygophyllum (van Zyl, 2000; Bellstedt et al., 2008) also have a predilection for the Namib center (Born et al., 2007), as does subgenus Agrophyllum of Zygophyllum. Species of subgenus Agrophyllum evince a typical succulent biome distribution with disjunct species in southwestern and northeastern Africa, west Asia, and Madagascar (Bellstedt et al., 2008). Some vertebrate examples with this pattern are given by Pickford (2004), and the scorpion genus Parabuthus (Prendini, 2001, 2004) has centers of diversity within southwestern Africa and the northeastern Horn and Arabia, with Namib center, early-divergent species sister to Cape species. This repeated pattern argues for the southern African arid corridor Namib subcenters being most likely refugia not only for the oldest extant lineages in Indigofera (Fig. 7, node 6) but for many other taxa as well (Midgley et al., 2005). The occurrence of Indigofereae in Madagascar— Many Indigofereae species inhabit Madagascar, and these represent lineages scattered throughout the Indigofereae phylogeny. They 837 must have colonized this island many times independently. Succulent-rich vegetation in northern, western, and southern Madagascar is phylogenetically linked to the African (and West Asian) arid corridors (Appendix 3; Lavin et al., 2000, 2004; Meve and Liede, 2002; Schrire et al., 2003). In general, Madagascar has been colonized by or has served as a source area for mostly African and Asian lineages via oceanic dispersal at various times throughout the Cenozoic (Yoder and Nowak, 2006). The Madagascan Ormocarpopsis (Dalbergieae) represents a fairly recent divergence from African Ormocarpum (Lavin et al., 2004). In contrast, the age of Madagascan lemurs is about 62 Ma, which is similar to its African–Asian sister lineage (Yoder et al., 1996). Madagascan muroid rodents diverged from Asiatic rodent lineages probably later than Madagascan lemurs (Jansa et al., 1999). The only generalization that seems to emerge is that on average the ages of transcontinental animal lineages are older than such plant lineages (Pennington et al., 2006b). The diversity of species and phylogenetic lines within Africa suggest that the Indigofereae have been in residence on this continent since the origin of the tribe. Indeed, the number of independent Madagascan lineages alone suggests a long African residence. The evolutionary rates analysis points to this residency extending back to the early Tertiary (Fig. 7; Appendix S9) considering that the Madagascan Disynstemon is sister to the tribe. Within Indigofereae, the earliest branching Phylloxylon (mean crown age of 9.3 Ma; node 29 in Fig. 7 and Appendix S9) is endemic to Madagascar. The genus Microcharis is the only genus in the CRIM clade with an endemic Madagascan diversification comprising two species (M. phyllogramme sampled here; Fig. 2) putatively arising from a single dispersal event less than 5–6 Ma in the African M. galpinii–M. spathulata clade. Indigofera has 55 species in Madagascar resulting from a minimum estimate of 30 dispersal events, with these from three of the four clades in the genus. Only the Cape clade lacks Madagascan species. The majority of these events, 20–26, dispersed from Africa to Madagascar and the balance from Asia. Nineteen Madagascan-inhabiting species also occur in Africa (and elsewhere for a few of these widespread species). A total of 36 species are endemic to Madagascar, and these arose from an estimated 15 of the 30 total Madagascan dispersal events. The abundance of succulent-rich and grass-rich environments shared between Madagascar and Africa could explain the shared Indigofereae flora in these two regions. The Tethyan clade has the highest percentage of Madagascan taxa with 14% of the overall c. 163 species in the clade, compared to an average of 6% of the totals of species in the palaeotropical and pantropical clades (c. 190 and 310 species, respectively). The Tethyan clade also has two endemic Madagascan crown clades comprising 13 species, with the oldest at about 8 Ma including species formally placed in Vaughania (Appendix S9, node 44). The Tethyan clade comprises 11 Madagascan dispersal events, compared to 12 events in the pantropical clade, which has twice the number of species relative to the Tethyan clade. The I. compressa–I. wituensis crown in the palaeotropical clade and I. bosseri–I. mangokyensis crown in the pantropical clade are c. 5–6 Ma, and all other Madagascan diversifications in the genus are less than 5 Ma. The footprint of Madagascan Indigofera diversifications is thus evident more deeply within the Tethyan clade, which also has the largest number of succulent biome taxa linked phylogenetically with the African arid corridor and Sahara-Sindian region. The above findings indicate that the many clades of neoendemic Madagascan Indigofereae have 838 American Journal of Botany their closest sister group relationships to African taxa, and there is overwhelming evidence that this is due to transoceanic dispersal assembly from the Upper Miocene to the present. This scenario fully supports the floral and faunal trends found elsewhere in Madagascar (e.g., Yoder and Nowak, 2006). Synthesis— Disynstemon for the first time has been established as the sister genus to Indigofereae. Other intergeneric and interspecific relationships resolved in previous but less well-sampled phylogenetic studies (e.g., Schrire et al., 2003) have also been fully validated through more detailed analysis. Perhaps unsurprisingly, Disynstemon, along with other earlybranching lineages of Indigofereae that are estimated to be some of the oldest lineages in the tribe, share a similar ecology (seasonally dry succulent-rich biome) and geography (Africa and Madagascar). The high retention indices for the morphological characters reveal high levels of phylogenetic structure equivalent to that found for the molecular data. The many linked characters related to the breeding system and diversity of defense strategies exemplify the morphological and biochemical integrity of tribe Indigofereae and each of the four main Indigofera clades. Examples are the many characters integral to the distinctive flower tripping mechanism that constrain floral morphology in the tribe. Such characters have dominated the morphological phylogenetic structure of the different genera and segregates within Indigofera. Also notable have been the closely convergent suites of breeding system and dispersal mechanism morphologies arising independently in the temperate biome diversifications of two widely divergent Indigofera subclades. A remarkable finding from the analysis of defense characters in Indigofera was that the Tethyan and pantropical clades listed the largest number of toxic and medicinally used species in the genus, respectively, owing to their complex chemistry. The palaeotropical clade, however, with its relatively simple chemistry and reliance on ant-association defenses, hosted the largest number of good quality animal feed species. Using different lines of evidence, we have scrutinized the four main clades of Indigofera for phylogenetic structure in morphology (e.g., traits noted), geography (focusing on the African arid corridor and Madagascar regions) and ecology (comparing similarly sized floras in the grass and succulent biomes in Africa). This evidence has provided insights into how geographical and ecological setting imposes structure on phylogenies. Such structure arises from the role played by ecological drift (the random walk of abundances of taxa in a community through time) on immigration and resident speciation within a biome. Low levels of successful immigration result in endemic diversifications having low extinction rates, thus higher levels of resident speciation and greater phylogenetic structure. This is more likely to occur when a biome is restricted in size, fragmented, and dominated by harsh climatic (e.g., arid) or edaphic (e.g., nutrient poor soil) conditions. A signature of phylogenetic structure in clades is when geography and ecology are very predictive of phylogenetic relatedness. The results of the Mantel regressions (Fig. 9) show that in the succulent and temperate biomes, ecological and geographical setting is generally a better predictor of phylogenetic relatedness than in the grass biome. Indigofereae subclades mostly confined to temperate or succulent-rich tropical habitats are more narrowly constrained and fragmented geographically (i.e., more dispersal limited) than those primarily inhabiting tropical grass-dominated settings, strongly suggesting that ecology has determined the geographical structure of the Indigofer- [Vol. 96 eae phylogeny. This is best illustrated in the Tethyan clade where the only three predominantly grass biome diversifications are pantropical in distribution compared to all the other succulent biome diversifications which are more geographically constrained. A further example is the almost entirely succulent biome flora of Somalia that harbors only seven of the c. 28 subclades of the Indigofera phylogeny, with nearly two thirds of Somalian species representing closely related groups within only three of these subclades. In contrast, the equivalently sized, predominantly grass biome southern African flora encompasses 17 subclades from all four Indigofera clades. In addition, a diversity of defense strategies identified for these grass-biome-inhabiting clades is seen as driving coexistence or sympatry of Indigofera species in grass-rich environments. At least 30 of the 77 Indigofereae clades (Fig. 10; Appendix S9) have diversifications largely restricted to the succulent biome, compared with only 16 to the grass biome and 14 to the temperate biome. The age distributions of clades (Fig. 10) reveals that the succulent biome harbors a much greater range of clade ages compared to the other two biomes, as well as comprising the oldest clades in the genus. The Tethyan and Cape clades are remarkable in containing a number of relatively old succulent biome lineages that possibly explain the relatively moderate clade support of these two main Indigofera clades. It is evident that the evolution of tribe Indigofereae has occurred substantially within the context of the succulent biome, associated particularly with the African arid corridor axis and its two arid poles, the northeastern Horn and southwestern African centers. Recognition of the succulent biome also highlights the phylogenetic relatedness and hence metacommunity affiliation of the African arid corridor with Madagascar and the SaharaSindian regions, centers that, in addition, underpin diversification in the palaeotropical clade. Arid corridor links to the Greater Cape Floristic Region of southern Africa are exemplified through the repeated biogeographical pattern shown of early divergent southern Namib center lineages giving rise to more derived mesic and temperate Cape diversifications. At least two and possibly three temperate Cape biome diversifications have early-divergent elements restricted to the Namib center and transitional Richtersveld region of northwestern South Africa. The greater persistence of vegetation in the succulent biome, e.g., in the southwestern African Karoo-Namib region, explains why its various subcenters are refugia for the oldest extant lineages of many genera, particularly Indigofera in the narrow Huns-Orange group subcenter. It is likely that this also explains why Indigofera nudicaulis is the only species of 266 sampled that was not resolved within any of the four Indigofera clades in the parsimony bootstrap and Bayesian analyses. This scenario does not inform on the origin of the genus Indigofera, except that the group probably arose in the succulent and grass biomes of Africa and Madagascar. Grass biome clades are not expected to have such old crown ages because this biome is larger in size with greater potential for immigration and it has a higher water requirement than the succulent biome. Grass biome clades are thus more prone to biome switching and are not as persistent as succulent biome clades. Despite evidence of some late Tertiary disjunctions in the tribe being associated with occasionally continuous African arid corridors, the predominant picture of relatively young crown diversification ages in Indigofera (Appendix S9), particularly those of the three Old World– New World disjunctions, emphasizes the overall importance of distribution patterns being explained by long-distance dispersal. April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae We suggest that Indigofereae is not exceptional among the species-rich clades of Leguminosae, all of which evince a high level of allopatric diversification within biomes, often at intercontinental scales. Very diverse legume genera such as Astragalus and Inga show little tendency to switch ecological setting, and in Astragalus there is even a greater tendency to switch continents than ecology. The species richness of clades like Indigofera and Acacia, however, also appears due to an ability to shift between biomes and disperse at intercontinental scales. This dispersal rate is much slower than speciation in tropical succulent-rich and temperate environments, than in tropical grass-rich or wet forest settings. This glimpse into patterns of legume diversity applies to legumes in general and suggests that causes of speciation are in large part as much geographical as ecological. LITERATURE CITED Ackerly, D. D., D. W. Schwilk, and C. O. Webb. 2006. Niche evolution and adaptive radiation: Testing the order of trait divergence. Ecology 87 (supplement): 50–61. Adams, E. N. III. 1972. Consensus techniques and the comparison of taxonomic trees. Systematic Zoology 21: 390–397. Adinarayana, D., and M. Sarada. 1987. Phytochemical study of Indigofera mysorensis Rottl. leaves. Journal of the Indian Chemical Society. 64: 648–649. Alvin, K. L. 1987. Leaf anatomy of Androstachys johnsonii Prain and its functional significance. Annals of Botany 59: 579–591. Arroyo, M. T. K. 1976. The systematics of the legume genus Harpalyce (Leguminosae: Lotoideae). Memoirs of the New York Botanical Garden 26: 1–80. Arroyo, M. T. K. 1981. Breeding systems and pollination biology in Leguminosae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 2, 723–769. Royal Botanic Gardens, Kew, UK. Axelrod, D. I. 1992. Climatic pulses, a major factor in legume evolution. In P. S. Herendeen and D. L. Dilcher [eds.], Advances in legume systematics, vol. 4, The fossil record, 259–279. Royal Botanic Gardens, Kew, UK. Aziz-ur-rehman, A. U., A. Malik, N. Riaz, H. Ahmad, S. A. Nawaz, and M. I. Choudhary. 2005. Lipoxygenase inhibiting constituents from Indigofera het[e]rantha. Chemical & Pharmaceutical Bulletin 53: 263–266. Bailey, C. D., T. G. Carr, S. A. Harris, and C. E. Hughes. 2003. Characterization of angiosperm nrDNA polymorphism, paralogy and pseudogenes. Molecular Phylogenetics and Evolution 29: 435–455. Baillon, H. E. 1861. Species Euphorbiacearum. A. Euphorbiacées Africaines. Deuxième partie. Adansonia 2: 27–55. Baker, J. G. 1883. Contributions to the flora of Madagascar, part 1, Polypetalae. Journal of the Linnean Society., Botany 20: 126–127. Baker, J. G. 1884. Further contributions to the flora of central Madagascar. Journal of the Linnean Society, Botany 21: 336. Bakker, F. T., A. Culham, P. Hettiarachi, T. Touloumenidou, and M. Gibby. 2004. Phylogeny of Pelargonium (Geraniaceae) based on DNA sequences of three genomes. Taxon 53: 17–28. Barker, N. P., B. D. Schrire, and J.-H. Kim. 2000. Generic relationships in the tribe Indigofereae (Leguminosae: Papilionoideae) based on sequence data and morphology. In P. S. Herendeen and A. Bruneau [eds.], Advances in legume systematics, vol. 9, 311–337. Royal Botanic Gardens, Kew, UK. Breckle, S.-W. 2002. Walter’s vegetation of the Earth: The ecological systems of the geo-biosphere, 4th ed., 527. Springer-Verlag, Berlin, Germanhy. Beier, B. A., J. A. A. Nylander, M. W. Chase, and M. Thulin. 2004. Phylogenetic relationships and biogeography of the desert plant genus Fagonia (Zygophyllaceae) inferred by parsimony and Bayesian model averaging. Molecular Phylogenetics and Evolution 33: 91–108. 839 Bell, E. A. 1981. Non-protein amino acids in the Leguminosae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 2, 489–499. Royal Botanic Gardens, Kew, UK. Bellstedt, D. U., L. van Zyl, E. M. Marais, B. Bytebier, C. A. de Villiers, A. M. Makwarela, and L. L. Dreyer. 2008. Phylogenetic relationships, character evolution and biogeography of southern African members of Zygophyllum (Zygophyllaceae) based on three plastid regions. Molecular Phylogenetics and Evolution 47: 932–939. Bisby, F. A. 1981. Genisteae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 1, 409–425. Royal Botanic Gardens, Kew, UK. Bisby, F. A., J. Buckingham, and J. B. Harborne [eds.]. 1994. Phytochemical dictionary of the Leguminosae, vol. 1, Plants and their constituents, 1051. Chapman & Hall, London, UK. Boiteau, P. 1938. Les légumineuses à roténone de la flore Malgache. Bulletin économique de Madagascar 3: 111–129. Bond, W. J., and P. Slingsby. 1983. Seed dispersal by ants in shrublands of the Cape Province and its evolutionary implications. South African Journal of Science 79: 231–233. Born, J., H. P. Linder, and P. Desmet. 2007. The greater Cape Floristic Region. Journal of Biogeography 34: 147–162. Brenes-Arguedas, T., P. D. Coley, and T. A. Kursar. 2008. Divergence and diversity in the defensive ecology of Inga. Journal of Ecology 96: 127–135. Bruneau, A., and G. J. Anderson. 1988. Reproductive biology of diploid and triploid Apios americana (Leguminosae). American Journal of Botany 75: 1876–1883. Burkill, H. M. 1995. The useful plants of west tropical Africa, vol. 3, Families J–L, 857. Royal Botanic Gardens, Kew, UK. Calow, P. 1983. Evolutionary principles, 108. Blackie and Son, Glasgow, UK. Coley, P. D., J. Lokvam, K. Rudolph, K. Bromberg, T. E. Sackett, L. Wright, T. Brenes-Arguedas, D. Dvorett, S. Ring, A. Clark, C. Baptiste, R. T. Pennington, and T. A. Kursar. 2005. Divergent defensive strategies of young leaves in two species of Inga. Ecology 86: 2633–2643. Cowling, R. M. 1983. Phytochorology and vegetation history in the south eastern Cape, South Africa. Journal of Biogeography 10: 393–419. Cowling, R., and D. Richardson. 1995. Fynbos: South Africa’s unique floral kingdom. Fernwood Press, Cape Town, South Africa. Cowling, R. M., and S. M. Pierce. 2001. Namaqualand: A succulent desert. Fernwood Press, Cape Town, South Africa. Cowling, R. M., . Proche , and J. H. J. Vlok. 2005. On the origin of southern African subtropical thicket vegetation. South African Journal of Botany 71: 1–23. Craven, P., and P. Vorster. 2006. Patterns of plant diversity and endemism in Namibia. Bothalia 36(2): 175–189. Dahlgren, R. 1963. Studies on Aspalathus and some related genera in South Africa. Opera Botanica 9: 53–57. de Winter, B. 1971. Floristic relationships between the northern and southern arid areas of Africa. Mitteilungen der Botanischen Staatsammlung München 10: 424–437. Donoghue, M. J., and M. J. Sanderson. 1992. The suitability of molecular and morphological evidence in reconstructing plant phylogeny. In P. S. Soltis, D. E. Soltis, and J. J. Doyle [eds.], Plant molecular systematics, 340–368. Chapman and Hall, New York, New York, USA. Du Puy, D. J., and J.-N. Labat. 2002a. Tribe Indigofereae. In D. J. Du Puy [ed.], The Leguminosae of Madagascar, 450–509. Royal Botanic Gardens, Kew, UK. Du Puy, D. J., and J.-N. Labat. 2002b. Tribe Millettieae. In D. J. Du Puy [ed.], The Leguminosae of Madagascar, 371–443. Royal Botanic Gardens, Kew, UK. Du Puy, D. J., J.-N. Labat, and B. D. Schrire. 1994. Révision du genre Vaughania S.Moore (Leguminosae). Bulletin du Muséum national d’Histoire naturelle, série 4, 16, section B, Adansonia 1: 75–102. Du Puy, D. J., J.-N. Labat, and B. D. Schrire. 1995. A revision of Phylloxylon (Leguminosae: Papilionoideae: Indigofereae). Kew Bulletin 50: 477–494. 840 American Journal of Botany Dutra, H. P., A. V. L. Freitas, and P. S. Oliveira. 2006. Dual ant attraction in the neotropical shrub Urera baccifera (Urticaceae): The role of ant visitation to pearl bodies and fruits in herbivore deterrence and leaf longevity. Functional Ecology 20: 252–260. Ehrlich, P. R., and P. H. Raven. 1964. Butterflies and plants: A study in plant coevolution. Evolution 18: 586–608. Etienne, R. S., A. M. Latimer, J. A. Silander Jr., and R. M. Cowling. 2006. Comment on “Neutral Ecological Theory Reveals Isolation and Rapid Speciation in a Biodiversity Hot Spot.”. Science 311: 610b. Farris, J. S. 1989. The retention index and the rescaled consistency index. Cladistics 5: 417–419. Farris, J. S., M. Källersjö, A. G. Kluge, and C. Bult. 1995. Testing significance of incongruence. Cladistics 10: 315–319. Felsenstein, J. 1985. Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39: 783–791. Ferguson, I. K. 1984. Pollen morphology and biosystematics of the subfamily Papilionoideae (Leguminosae). In W. F. Grant [ed.], Plant biosystematics, 377–394. Academic Press, Don Mills, Ontario Canada. Ferguson, I. K., and R. Strachan. 1982. Pollen morphology and taxonomy of the tribe Indigofereae (Leguminosae: Papilionoideae). Pollen et Spores 24: 171–210. Garcez, W. S., F. R. Garcez, and A. Barison. 2003. Additional 3-nitropropanoyl esters of glucose from Indigofera suffruticosa (Leguminosae). Biochemical Systematics and Ecology 31: 207–209. Geesink, R. 1981. Tephrosieae (Benth.) Hutch. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 1, 245–260. Royal Botanic Gardens, Kew, UK. Geesink, R. 1984. Scala Millettiarum. A survey of the genera of the tribe Millettieae (Legum.-Pap) with methodological considerations. Leiden Botanical Series, vol. 8, 1–131. E.J. Brill/Leiden University Press. Leiden, Netherlands. Galley, C., B. Bytebier, D. U. Bellstedt, and H. P. Linder. 2007. The Cape element in the afrotemperate flora: From Cape to Cairo? Proceedings of the Royal Society of London, B, Biological Sciences 274: 535–543. Galley, C., and H. P. Linder. 2006. Geographical affinities of the Cape flora, South Africa. Journal of Biogeography 33: 236–250. Gillett, J. B. 1958. Indigofera (Microcharis) in tropical Africa: With the related genera Cyamopsis and Rhynchotropis. Kew Bulletin, Additional Series 1: 1–166. Goldblatt, P. 1978. An analysis of the flora of southern Africa: Its characteristics, relationships and origins. Annals of the Missouri Botanical Garden 65: 369–436. Goldblatt, P. 1981. Cytology and the phylogeny of Leguminosae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 2, 427–463. Royal Botanic Gardens, Kew, UK. Goldblatt, P., and J. Manning. 2000. Cape Plants: A conspectus of the Cape flora of South Africa. Strelitzia 9: 708. Gomes, C. M. R., O. R. Gottlieb, R. C. Gottlieb, and A. Salatino. 1981. Phytochemistry in perspective: Chemosystematics of the Papilionoideae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 2, 465–488. Royal Botanic Gardens, Kew, UK. Goslee, S., and D. Urban. 2007. The ecodist package, version 1.1.3. The R project for statistical computing. Website http://www.r-project.org. Grammatikopoulos, G., and Y. Manetas. 1994. Direct absorption of water by hairy leaves of Phlomis fruticosa and its contribution to drought avoidance. Canadian Journal of Botany 72: 1805–1811. Halanych, K. M. 1998. Lagomorphs misplaced by more characters and fewer taxa. Systematic Biology 47: 138–146. Harms, H. 1900. Leguminosae. In A. Engler and K. Prantl [eds.], Die Natürlichen Pflanzenfamilien, Nachträge, vol. 2, 33. W. Englemann, Leipzig, Germany. Harvey, W. H. 1862. Leguminosae. In W. H. Harvey and O. W. Sonder, Flora capensis, vol. 2, 163–203. Hodges, Smith and Co., Dublin, Ireland. Harvey, P. H., and M. D. Pagel. 1991. The comparative method in evolutionary biology. Oxford Series in Ecology and Evolution, vol. 1. Oxford University Press, Oxford, UK. Hasan, A., P. G. Waterman, and N. Iftikhar. 1989. The use of new chromatographic techniques for the isolation and purification of phe- [Vol. 96 nolic acids from Indigofera heterantha. Journal of Chromatography 466: 399–402. Hegnauer, R., and M. Hegnauer. 2001. Chemotaxonomie der Pflanzen, vol. XIb–2, Leguminosae, part 3, 276–288. Birkhauser-Verlag, Basel. Switzerland. Heinrich, B. 1983. Insect foraging energetics. In C. E. Jones and R. J. Little [eds.], Handbook of experimental pollination biology, 187–214. Van Nostrand Reinhold, New York, New York, USA. Heyn, C. C. 1981. Tribe Trifolieae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 1, 383–385. Royal Botanic Gardens, Kew, UK. Hilton-Taylor, C. 1994. Western Cape domain (succulent Karoo): Republic of South Africa and Namibia. In S. D. Davis, V. H. Heywood, and A. C. Hamilton [eds.], Centers of plant diversity: A guide and strategy for their conservation, vol 1, 204–217. World Wildlife Fund and International Union for Conservation of Nature, Cambridge, UK.. Hoare, D. B., L. Mucina, M. C. Rutherford, J. H. J. Vlok, D. I. W. Euston-Brown, A. R. Palmer, L. W. Powrie, et al. 2006. Albany thicket biome. In L. Mucina and M. C. Rutherford [eds.], The vegetation of South Africa, Lesotho and Swaziland. Strelitzia 19: 541–567. Howis, S., N.P. Barker, and L. Mucina. 2009. Globally grown, but poorly known: species limits and biogeography of Gazania Gaert. (Asteraceae) inferred from chloroplast and nuclear DNA sequence data. Taxon 58: in press. Hubbell, S. 2001. The unified neutral theory of biodiversity and biogeography. Princeton University Press, Princeton, New Jersey, USA. Hutchinson, J. 1964. Order 7, Leguminales. In The genera of flowering plants (Angiospermae), I. Dicotyledones, 221–489. Clarendon Press, Oxford, UK. Janis, C. M. 1993. Tertiary mammal evolution in the context of changing climates, vegetation, and tectonic events. Annual Review of Ecology and Systematics 24: 467–500. Jansa, S. A., S. M. Goodman, and P. K. Tucker. 1999. Molecular phylogeny and biogeography of the native rodents of Madagascar (Muridae: Nesomyinae): A test of the single-origin hypothesis. Cladistics 15: 253–270. Janzen, D. H. 1971. Euglossine bees as long-distance pollinators of tropical plants. Science 171: 203–205. Janzen, D. H. 1981. The defenses of legumes against herbivores. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 2, 951–977. Royal Botanic Gardens, Kew, UK. Johnson, S. D. 1992. Plant–animal relationships. In R. Cowling [ed.], The ecology of fynbos: Nutrients, fire and diversity, 175–205. Oxford University Press, Cape Town, South AFrica. Jürgens, N. 1997. Floristic biodiversity and history of African arid regions. Biodiversity and Conservation 6: 495–514. Jussieu, A. L. de. 1789. Genera plantarum. Herissant and Barrois, Paris, France. Kamal, R., and M. Mangla. 1993. In vivo and in vitro investigations on rotenoids from Indigofera tinctoria and their bioefficacy against the larvae of Anopheles stephensi and adults of Callosobruchus chinensis. Journal of Biosciences 18: 93–101. Kay, K. M., J. B. Whittall, and S. A. Hodges. 2006. A survey of nuclear ribosomal internal transcribed spacer substitution rates across angiosperms: An approximate molecular clock with life history effects. BMC Evolutionary Biology 6: 36. Kishino, H., and M. Hasegawa. 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. Journal of Molecular Evolution 29: 170–179. Kluge, A. G., and J. S. Farris. 1969. Quantitative phyletics and the evolution of anurans. Systematic Zoology 18: 1–32. Koptur, S. 1985. Alternative defenses against herbivores in Inga (Fabaceae: Mimosoideae) over an elevational gradient. Ecology 66: 1639–1650. Kottek, M., J. Grieser, C. Beck, B. Rudolf, and F. Rubel. 2006. World map of the Köppen-Geiger climate classification updated. Meteorologische Zeitschrift 15: 259–263. April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae Kürschner, H. 1998. Biogeography and introduction to vegetation. In S. A. Ghazanfar and M. Fisher [eds.], Geobotany: Vegetation of the Arabian Peninsula, vol. 25, 63–98. Kluwer, Dordrecht, Netherlands. Lackey, J. A. 1981. Tribe Phaseoleae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 1, 301–327. Royal Botanic Gardens, Kew, UK. Latimer, A. M., J. A. Silander Jr., and R. M. Cowling. 2005. Neutral ecological theory reveals isolation and rapid speciation in a biodiversity hot spot. Science 309: 1722–1725. Lauter, N., C. Gustus, A. Westerbergh, and J. Doebley. 2004. The inheritance and evolution of leaf pigmentation and pubescence in teosinte. Genetics 167: 1949–1959. Lavin, M. 2006. Floristic and geographical stability of discontinuous seasonally dry forests explains patterns of plant phylogeny and endemism. In R. T. Pennington, G. P. Lewis, and J. A. Ratter [eds.], Neotropical savannas and seasonally dry forests, 433–447. Taylor and Francis, Boca Raton, Louisiana, USA. Lavin, M., P. S. Herendeen, and M. F. Wojciechowski. 2005. Evolutionary rates analysis of Leguminosae implicates a rapid diversification of lineages during the Tertiary. Systematic Biology 54: 575–594. Lavin, M., and M. Luckow. 1993. Origins and relationships of tropical North America in the context of the boreotropics hypothesis. American Journal of Botany 80: 1–14. Lavin, M., R. T. Pennington, B. B. Klitgaard, J. I. Sprent, H. C. De Lima, and P. E. Gasson. 2001. The dalbergioid legumes (Fabaceae): Delimitation of a pantropical monophyletic clade. American Journal of Botany 88: 503–533. Lavin, M., B. D. Schrire, G. P. Lewis, R. T. Pennington, A. Delgado Salinas, M. Thulin, C. E. Hughes, et al. 2004. Metacommunity process rather than continental tectonic history better explains geographically structured phylogenies in legumes. Philosophical Transactions of the Royal Society of London, B, Biological Sciences 359: 1509–1522. Lavin, M., M. Thulin, J.-N. Labat, and R. T. Pennington. 2000. Africa, the odd man out: Molecular biogeography of dalbergioid legumes (Fabaceae) suggests otherwise. Systematic Botany 25: 449–467. Le Maitre, D. C., and J. J. Midgley. 1992. Plant reproductive ecology. In R. Cowling [ed.], The ecology of fynbos: Nutrients, fire and diversity, 135–174. Oxford University Press, Cape Town, South Africa. Legendre, P. 1990. Quantitative methods and biogeographic analysis. In D. J. Garbary and R. R. South [eds.], Evolutionary biogeography of the marine algae of the North Atlantic. NATO AS1 series, vol. G 22, 9–34. Springer-Verlag, Berlin, Germany. Lewis, G., B. Schrire, B. Mackinder, and M. Lock. 2005. Legumes of the world, 577. Royal Botanic Gardens, Kew, UK. Lima, A. K., E. L. C. Amorim, T. M. Aquino, C. S. A. Lima, R. M. M. Pimentel, J. S. Higino, and U. P. Albuquerque. 2003. Estudo farmacognóstico de Indigofera microcarpa. Revista Brasileira de Ciências Farmacêuticas 39: 373–379. Linder, H. P., M. E. Meadows, and R. M. Cowling. 1992. History of the Cape flora. In R. Cowling [ed.], The ecology of fynbos: Nutrients, fire and diversity, 113–134. Oxford University Press, Cape Town, South Africa. López, J., T. Rodríguez-Riaño, A. Ortega-Olivencia, J. A. Devesa, and T. Ruiz. 1999. Pollination mechanisms and pollen–ovule ratios in some Genisteae from SW Europe. Plant Systematics and Evolution 216: 23–47. Lyshede, O. B. 1977. Structure and function of trichomes in Spartocytisus filipes. Botaniska Notiser 129: 395–404. Mabberley, D. J. 1997. The plant book, 2nd ed. Cambridge University Press, Cambridge, UK. Maddison, D. R., and W. P. Maddison. 2005. MacClade, version 4.08. Sinauer, Sunderland, Massachusetts, USA. Maley, J. 1996. The African rain forest—Main characteristics of changes in vegetation and climate from the Upper Cretaceous to the Quaternary. Proceedings of the Royal Society of Edinburgh 104B: 31–73. 841 Marloth, R. 1903. Results of experiments on Table Mountain for ascertaining the amount of moisture deposited from the south-east clouds. Transactions of the South African Philosophical Society 14: 403–408. Marloth, R. 1910. On the absorption of water by aerial organs of plants. Transactions of the Royal Society of South Africa 1: 429–433. Mauseth, J. D. 1988. Plant anatomy. Benjamin/Cummings, Menlo Park, California, USA. McKey, D. 1989. Interactions between ants and leguminous plants. In C. H. Stirton and J. L. Zarucchi [eds.], Advances in legume biology. Monographs in Systematic Botany from the Missouri Botanical Garden 29: 673–718. Meve, U., and S. Liede. 2002. Floristic exchange between mainland Africa and Madagascar: A case study of Apocynaceae-Asclepiadoideae. Journal of Biogeography 29: 865–873. Midgley, G. F., G. Reeves, and C. Klak. 2005. Late Tertiary and Quaternary climate change and centers of endemism in the southern African flora. In A. Purvis, J. L. Gittleman, and T. Brooks [eds.], Phylogeny and conservation, 230–242. Cambridge University Press, Cambridge, UK. Mucina, L., N. Jürgens, A. Le Roux, M. C. Rutherford, U. Schmiedel, K. J. Esler, L. W. Powrie, et al. 2006. Succulent Karoo biome. In L. Mucina and M. C. Rutherford [eds.], The vegetation of South Africa, Lesotho and Swaziland. Strelitzia 19: 221–299. O’Dowd, D. J. 1982. Pearl bodies as ant food: An ecological role for some leaf emergences of tropical plants. Biotropica 14: 40–49. Ohashi, H., R. M. Polhill, and B. G. Schubert. 1981. Desmodieae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 1, 292–300. Royal Botanic Gardens, Kew, UK. Peltier, M. 1967. La position systématique du genre Phylloxylon. Adansonia, série 2, 7: 255–257. Pennington, R. T., D. E. Prado, and C. A. Pendry. 2000. Neotropical seasonally dry forests and Quaternary vegetation changes. Journal of Biogeography 27: 261–273. Pennington, R. T., G. P. Lewis, and J. A. Ratter. 2006a. An overview of the plant diversity, biogeography and conservation of neotropical savannas and seasonally dry forests. In R. T. Pennington, G. P. Lewis, and J. A. Ratter [eds.], Neotropical savannas and seasonally dry forests, 1–20. Taylor and Francis, Boca Raton, Louisiana, USA. Pennington, R. T., J. E. Richardson, and M. Lavin. 2006b. Insights into the historical construction of species-rich biomes from dated plant phylogenies, neutral ecological theory and phylogenetic community structure. New Phytologist 172: 605–616. Pennington, R. T., C. H. Stirton, and B. D. Schrire. 2005. Tribe Sophoreae. In G. Lewis, B. Schrire, B. Mackinder, and M. Lock [eds.], Legumes of the world, 227–249. Royal Botanic Gardens, Kew, UK. Pickford, M. 2004. Southern Africa: A cradle of evolution. South African Journal of Science 100: 205–214. Polhill, R. M. 1976. Genisteae (Adans.) Benth. and related tribes (Leguminosae). In V. H. Heywood [ed.], Botanical systematics, vol. 1, 143–368. Academic Press, London, UK. Polhill, R. M. 1981. Tribe Indigofereae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, vol. 1, 289–291. Royal Botanic Gardens, Kew, UK. Polhill, R. M. 1990. Légumineuses. In J. Bosser, Th. Cadet, J. Guého, and W. Marais [eds.], Flore des Mascareignes: La Réunion, Maurice, Rodrigues, vol. 80, 1–235. Sugar Industry Research Institute, Port Louis, Mauritius; ORSTOM, Paris, France; Royal Botanic Gardens, Kew, UK. Prendini, L. 2001. Phylogeny of Parabuthus (Scorpiones, Buthidae). Zoologica Scripta 30: 13–35. Prendini, L. 2004. The systematics of southern African Parabuthus Pocock (Scorpiones, Buthidae): Revisions to the taxonomy and key to the species. Journal of Arachnology 32: 109–186. Quézel, P. 1978. Analysis of the flora of Mediterranean and Saharan Africa. Annals of the Missouri Botanical Garden 65: 479–534. Rambaut, A. 1996. Se-Al, version 2.0a11, sequence alignment editor. University of Oxford, Oxford, UK. Website http://evolve.zoo.ox.ac. uk/software.html. 842 American Journal of Botany R Development Core Team. 2007. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, website http://www.R-project.org. Rajkapoor, B., N. Murugesh, K. Lalitha, and D. R. Rama. 2007. Cytotoxic activity of [a] flavone glycoside from the stem of Indigofera aspalathoides. Journal of Natural Medicine 61: 80–83. Rebelo, A., C. Boucher, N. Helme, L. Mucina, and M. C. Rutherford. 2006. Fynbos biome. In L. Mucina and M. C. Rutherford [eds.]. The vegetation of South Africa, Lesotho and Swaziland. Strelitzia 19: 53–219. Reynolds, B. D., W. J. Blackmon, E. Wickremesinhe, M. H. Wells, and R. J. Constantin. 1990. Domestication of Apios americana. In J. Janick and J. E. Simon [eds.], Advances in new crops, 436–442 Timber Press, Portland, Oregon, USA. Richards, A. J. 1986. Plant breeding systems. George Allen and Unwin, London, UK. Richardson, J. E., R. T. Pennington, T. D. Pennington, and P. M. Hollingsworth. 2001. Recent and rapid diversification of a species-rich genus of neotropical trees. Science 293: 2242–2245. Rundel, P. W. 1982. Water uptake by organs other than roots. Encyclopedia of Plant Physiology, New Series 12B(2): 111–134. Sanderson, M. J. 2002. Estimating absolute rates of molecular evolution and divergence times: A penalized likelihood approach. Molecular Biology and Evolution 19: 101–109. Sanderson, M. J. 2004. r8s, version 1.71, User’s manual. University of California, Davis Website http://loco.biosci.arizona.edu/r8s. Sanderson, M. J., and J. J. Doyle. 1992. Reconstruction of organismal and gene phylogenies from data on multigene families: Concerted evolution, homoplasy, and confidence. Systematic Biology 41: 4–17. Schluter, D. 2000. The ecology of adaptive radiation. Oxford University Press, Oxford, UK. Schrire, B. D. 1984. A taxonomic revision of the tribe Desmodieae (Leguminosae-Papilionoideae. MSc thesis, University of DurbanWestville, Durban, South Africa. Schrire, B. D. 1989. A multidisciplinary approach to pollination biology in the Leguminosae. In C. H. Stirton and J. L. Zarucchi [eds.], Advances in legume biology. Monographs in Systematic Botany from the Missouri Botanical Garden 29: 183–242. Schrire, B. D. 1991. Systematic studies in African Indigofereae (Leguminosae-Papilionoideae). PhD thesis, University of Natal, Pietermaritzburg, South Africa. Schrire, B. D. 1995. Evolution of the tribe Indigofereae (Leguminosae– Papilionoideae). In M. D. Crisp and J. J. Doyle [eds.], Advances in legume systematics, vol. 7, Phylogeny, 161–244. Royal Botanic Gardens, Kew, UK. Schrire, B. D. 2005a. Tribe Indigofereae. In G. Lewis, B. Schrire, B. Mackinder, and M. Lock [eds.], Legumes of the world, 361–365. Royal Botanic Gardens, Kew, UK. Schrire, B. D. 2005b. Tribe Millettieae. In G. Lewis, B. Schrire, B. Mackinder, and M. Lock [eds.], Legumes of the world, 367–387. Royal Botanic Gardens, Kew, UK. Schrire, B. D. 2008. The Madagascan genus Vaughania is reduced to synonymy under Indigofera (Leguminosae–Papilionoideae-Indigofereae). Kew Bulletin 63: 477–479. Schrire, B. D., M. Lavin, N. P. Barker, H. Cortes-Burns, I. Von Senger, and J.-H. Kim. 2003. Towards a phylogeny of Indigofera (Leguminosae–Papilionoideae): Identification of major clades and relative ages. In B. B. Klitgaard and A. Bruneau [eds.], Advances in legume systematics, vol. 10, Higher level systematics, 269–302. Royal Botanic Gardens, Kew, UK. Schrire, B. D., M. Lavin, and G. P. Lewis. 2005a. Global distribution patterns of the Leguminosae: insights from recent phylogenies. In I. Friis and H. Balslev [eds.], Plant diversity and complexity patterns: local, regional and global dimensions. Biologiske Skrifter 55: 375–422. Schrire, B. D., G. P. Lewis, and M. Lavin. 2005b. Biogeography of the Leguminosae. In G. Lewis, B. Schrire, B. Mackinder, and M. Lock [eds.], Legumes of the world, 21–54. Royal Botanic Gardens, Kew. Schrire, B. D., and J. R. Sims. 1997. A re-evaluation of pollen morphology and taxonomy in the tribe Indigofereae (LeguminosaePapilionoideae). Kew Bulletin 52: 841–878. [Vol. 96 Schutte, A. L., J. H. J. Vlok, and B.-E. Van Wyk. 1995. Fire-survival strategy—A character of taxonomic, ecological and evolutionary importance in fynbos legumes. Plant Systematics and Evolution 195: 243–259. Schwilk, D. W., J. E. Keeley, and W. J. Bond. 1997. The intermediate disturbance hypothesis does not explain fire and diversity pattern in fynbos. Plant Ecology 132: 77–84. Scotese, C. R. 2001. Atlas of Earth history, vol. 1, Paleogeography, PALEOMAP Project, 52. Arlington, Texas. Website http://www.scotese.com. Seely, M. K., M. P. De Vos, and G. N. Louw. 1977. Fog imbibition, satellite fauna and unusual leaf structure in a Namib desert dune plant Trianthema hereroensis. South African Journal of Science 73: 69–72. Selvam, C., S. M. Jachak, R. G. Oli, R. Thilagavati, A. K. Chakraborti, and K. K. Bhutani. 2004. A new cyclooxygenase (COX) inhibitory pterocarpan from Indigofera aspalathoides: Structure elucidation and determination of binding orientation in the active sites of the enzyme by molecular docking. Tetrahedron Letters 45: 4311–4314. Sharif, A., E. Ahmed, A. Malik, N. Riaz, N. Afza, S. A. Nawaz, M. Arshad, M. R. Shah, and M. I. Choudhary. 2005. Lipoxygenase inhibitory constituents from Indigofera oblongifolia. Archives of Pharmacal Research 28: 761–764. Siegler, D. S., B. R. Maslin, and E. E. Conn. 1989. Cyanogenesis in the Leguminosae. In C. H. Stirton and J. L. Zarucchi [eds.], Advances in legume biology. Monographs in Systematic Botany from the Missouri Botanical Garden 29: 645–672. Simmons, M. P. 2004. Independence of alignment and tree search. Molecular Phylogenetics and Evolution 31: 874–879. Simmons, M. P., and H. Ochoterena. 2000. Gaps as characters in sequence-based phylogenetic analyses. Systematic Biology 49: 369–381. Small, E., and B. Brookes. 1984. Reduction of the geocarpic Factorovskya to Medicago. Taxon 33: 622–635. Small, E., P. Lassen, and B. Brookes. 1987. An expanded circumscription of Medicago (Leguminosae, Trifolieae) based on explosive flower tripping. Willdenowia 16: 415–437. Smith, R. J. 1980. Rethinking allometry. Journal of Theoretical Biology 87: 97–111. Stevens, P. F. 2001 [onward]. Angiosperm Phylogeny Website, version 8 [more or less continuously updated]. Website http://www.mobot. org/MOBOT/research/APweb/. Missouri Botanical Garden, St. Louis, Missouri, USA. Stirton, C. H. 1981. Petal sculpturing in papilionoid legumes. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, 771– 788. Royal Botanic Gardens, Kew, UK. Stock, W. D., F. Van Der Heyden, and O. A. M. Lewis. 1992. Plant structure and function. In R. Cowling [ed.], The ecology of fynbos: Nutrients, fire and diversity, 226–240. Oxford University Press, Cape Town Swofford, D. L. 2002. PAUP*. Phylogenetic analysis using parsimony (*and Other Methods). Version 4. Sinauer, Sunderland, Massachusetts. Taubert, P. 1894. Leguminosae. In A. Engler and K. Prantl [eds.], Die Naturlichen Pflanzenfamilien, ed. 1, 3: 184–384. Wilhelm Engelmann, Berlin. Templeton, A. R. 1983. Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the evolution of humans and apes. Evolution 37: 221–244. Thangadurai, D., M. B. Viswanathan, and N. Ramesh. 2002. Indigoferabietone, a novel abietane diterpenoid from Indigofera longiracemosa with potential antituberculous and antibacterial activity. Die Pharmazie 57: 714–715. Thulin, M. 1993. Fabaceae (Leguminosae). In M. Thulin [ed.], Flora of Somalia, vol. 1, 341–465. Royal Botanic Gardens, Kew. Thulin, M. 1994. Aspects of disjunct distributions and endemism in the arid parts of the Horn of Africa, particularly Somalia. In J. H. Seyani and A. C. Chikuni [eds.], Proceedings of the XIIIth Plenary Meeting April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae of the Association pour l’Étude taxonomique de la Flore d’Afrique Tropicale (AETFAT), 2: 1105–1119, Zomba, Malawi. National Herbarium and Botanic Gardens of Malawi, Zomba, Malawi. Thulin, M., and A. N. B. Johansson. 1996. Taxonomy and biogeography of the anomalous genus Wellstedia. In L. J. G Van der Maesen, X. M. van der Burgt and J. M. van Medenbach de Rooy [eds.], The biodiversity of African plants, 73–86. Proceedings of XIVth congress of the Association pour l’Étude taxonomique de la Flore d’Afrique Tropicale (AETFAT). Kluwer, Wageningen. Netherlands. Thulin, M., M. Lavin, R. Pasquet, and A. Delgado Salinas. 2004. Phylogeny and biogeography of Wajira (Leguminosae): A monophyletic segregate of Vigna centered in the Horn of Africa region. Systematic Botany 29: 903–920. Touloumenidou, T., F. Bakker, and F. Albers. 2007. The phylogeny of Monsonia (Geraniaceae). Plant Systematics and Evolution 264: 1–14. Van der Pijl, L. 1982. Principles of dispersal in higher plants, 3rd ed. Springer-Verlag, New York. Van Wyk, A. E., and G. F. Smith. 2001. Regions of floristic endemism in southern Africa: A review with emphasis on succulents. Umdaus Press, Hatfield, South Africa. Van Zinderen Bakker, E. M. 1969. The arid corridor between south western Africa and the horn of Africa. Palaeoecology of Africa 4: 139–140. Van Zinderen Bakker, E. M. 1975. The origin and the palaeoenvironment of the Namib desert biome. Journal of Biogeography 2: 65–73. Van Zyl, L. 2000. A systematic revision of Zygophyllum (Zygophyllaceae) in the southern African region. PhD dissertation, University of Stellenbosch, Stellenbosch, South Africa. Verdcourt, B. 1969. The arid corridor between the North East and South West areas of Africa. In E. M. van Zinderen Bakker [ed.], Palaeoecology of Africa, vol. 4, 140–144. A. A. Balkema, Cape Town, South Africa. Vermeij, G. J. 1978. Biogeography and adaptation: Patterns of marine life. Harvard University Press, Cambridge, Massachusetts, USA. 843 Ward, J. D., and J. Corbett. 1990. Towards an age for the Namib. In M. K. Seely [ed.], Namib ecology: 25 years of Namib research, 17–26. Transvaal Museum Monograph No. 7. Transvaal Museum, Pretoria, South Africa. Werger, M. J. A. 1978. The Karoo-Namib region. In M. J. A. Werger [ed.], Biogeography and ecology of southern Africa, 233–299. W. Junk, The Hague, Netherlands. White, F. 1983. The vegetation of Africa. A descriptive memoir to accompany the UNESCO/AETFAT/UNSC vegetation map of Africa. UNESCO, Paris, France. White, F., and J. Léonard. 1991. Phytogeographical links between Africa and Southwest Asia. Flora et Vegetatio Mundi 9: 229–246. Wilkinson, H. P. 1979. The plant surface (mainly leaf). In C. R. Metcalf and L. Chalk [eds.], Anatomy of the dicotyledons, 2nd ed., vol. 1, 97–165. Clarendon Press, Oxford, UK. Wojciechowski, M. F., M. Lavin, and M. J. Sanderson. 2004. A phylogeny of legumes (Leguminosae) based on analysis of the plastid matK gene resolves many well-supported subclades within the family. American Journal of Botany 91: 1846–1862. Yang, Z., and B. Rannala. 1997. Bayesian phylogenetic inference using DNA sequences: A Markov chain Monte Carlo method. Molecular Biology and Evolution 14: 717–724. Yoder, A. D., M. Cartmill, M. Ruvolo, K. Smith, and R. Vigalys. 1996. Ancient single origin for Malagasy primates. Proceedings of the National Academy of Sciences, USA 93: 5122–5126. Yoder, A. D., and M. D. Nowak. 2006. Has vicariance or dispersal been the predominant biogeographic force in Madagascar? Only time will tell. Annual Review of Ecology Evolution and Systematics 37: 405–431. Yoshida, T., and M. Hasegawa. 1977. Distribution of stizolamine in some leguminous plants. Phytochemistry 16: 131–132. Zhang, X.-X., Z.-X. Zhang, L. Chen, and Y.-F. Su. 2006. New aliphatic nitro-compounds from Indigofera carlesii. Fitoterapia 77: 15–18. Appendix 1. Voucher information and GenBank accession numbers for taxa used in this study. Voucher specimens are deposited in the following herbaria: CANB = Australian National Herbarium, Canberra; DAV = John M. Tucker Herbarium, Davis, California; E = Royal Botanic Garden, Edinburgh; FHO = Forest Herbarium, Oxford; GRA = Selmar Schonland Herbarium, Grahamstown; K = Royal Botanic Gardens, Kew; IUI = University Herbarium, Seoul, Korea; MONT = University Herbarium, Bozeman, Montana; NBG = Compton Herbarium, Cape Town; P = Muséum national d’Histoire naturelle, Paris; RB = Jardim Botânico do Rio de Janeiro; UPS = Botanical Museum, Uppsala. Taxon—Voucher (Herbarium), Origin, Coordinates (in decimal degrees latitude then longitude; not recorded in outgroups and Phylloxylon), GenBank accession(s). When two accessions of the same taxon were sampled, one of the vouchers is distinguished by its unique code in the data matrix. Clitoria ternatea L.—USDA PI 322364, Hu 1068 (DAV), Brazil, Sao Paulo, AF467038. Schefflerodendron usambarense Harms—Semsei 3165 (K), Tanzania, Tanga Distr., EU752495. Aganope thyrsiflora (Benth.) Polhill—Sidiyasa 640 (K), Indonesia, AF467018. Xeroderris stuhlmannii (Taub.) Mendonça & Sousa— Corby 2162 (K), Zimbabwe, AF467485. Dalbergiella nyassae Baker f.—Schrire 2650 (K), Zambia, Lusaka Distr., AF521795. Craibia brevicaudata (Vatke) Dunn—Polhill & Robertson 5296 (K), Kenya, AF467039. Austrosteenisia blackii (F. Muell.) Geesink—Pedley 5005 (K), Australia, Queensland, Brisbane, AF467020. Platycyamus regnellii Benth.—Lima s.n. (RB), Brazil, Minas Gerais, AF467491. Craspedolobium schochii Harms—A. Henry 9241c (K), China, Yunnan, Mengtzi, EU729480. Ophrestia radicosa (A. Rich.) Verdc.—USDA PI 255748, Hu 1104 (DAV), Zambia, AF467484. Canavalia brasiliensis Mart. ex Benth.—USDA PI 319336, Hu 1069 (DAV), Mexico, AF467034. Fordia splendidissima (Blume ex Miq.) Buijsen—Tangah s.n. (K), Malaysia, Sabah, AF467048. Sylvichadsia grandifolia (R. Vig.) DuPuy & Labat—R. Capuron 24962 SF (K), Madagascar, Andrapengy, EU729481. Abrus precatorius L.—Hu 1136 (DAV), Taiwan, AF467015. Platysepalum hirsutum (Dunn) Hepper—P. Adames 720 (K), Liberia, Mt Bele road, EU729482. Leptoderris fasciculata Dunn—J. Fay 4128 (K), Central African Republic, EU729483. Millettia grandis (E. Mey.) Skeels—Schrire 2597 (K), South Africa, KwaZulu-Natal, Vernon Crookes N.R., AY 009139. Disynstemon paullinioides (Baker) M. Peltier—P. Phillipson 3077 (K), Madagascar, Toliara, EU729484. Phylloxylon decipiens Baill.—Serv. Forestier 126 R-6 (K), Madagascar, Sahafay, EU729485. P. xylophylloides (Baker) Du Puy, Labat & Schrire— Du Puy et al. M880 (K), Madagascar, AF274684. P. perrieri Drake—Labat & Coute 2663 (K), Madagascar, AF274685. P. spinosa Du Puy, Labat & Schrire—Schrire et al. 2531 (K), Madagascar, Ambilobe, AF521715. P. arenicola Du Puy, Labat & Schrire—Capuron 24691 (K), Madagascar, Sables, EU729486. Cyamopsis tetragonoloba (L.) Taub.—Reid s.n. (K seed voucher ex G. Reid, University of Stirling, UK), AF274687. C. dentata (N.E.Br.) Torre—Smith 3215 (K), Botswana, –20.93, 21.43, AF274686. C. serrata Schinz—Germishuizen 2618 (K), Namibia, –21.45, 17.82, EU729487. C. senegalensis Guill. & Perr.—Giess 8529 (K), Namibia, –21.13, 14.85, AF521717. C. senegalensis Guill. & Perr. [AF467040]—USDA 263525 Hu 1099 (DAV), Senegal, 14.9, –16.3, AF467040. Indigastrum argyroides (E.Mey.) Schrire—Ramdhani & Konje 631 (GRA), South Africa, Richtersveld Nat. Park, –28.15, 16.97, EU729488. I. argyraeum (Eckl. & Zeyh.) Schrire—Schrire & Barker 2616 (K), South Africa, Carlisle Bridge, –33.08, 26.23, AF274691. I. candidissimum (Dinter) Schrire—Volk 12738 (K), Namibia, Maltahöhe Distr. Dowisib, –25.23, 18.52, EU729489. I. costatum (Guill. & Perr.) Schrire subsp. macrum (E. Mey.) Schrire—Burrows 6624 (K), South Africa, Mpumalanga, –25.33, 30.48, AF521716. I. fastigiatum (E. Mey.) Schrire—D. Edge 531 (K), South Africa, Mpumulanga Prov., –25.3, 30.13, EU729490. 844 American Journal of Botany Microcharis tenuirostris (Thulin) Schrire—M. Thulin et al. 7559 (K), Somalia, 2.87, 43.97, EU729491. M. gyrata (Thulin) Schrire—M. Gilbert et al. 7646 (K), Ethiopia, Sidamo Region, 4.43, 41.82, EU729492. M. stipulosa (Chiov.) Schrire—M. Gilbert et al. 8270 (K), Ethiopia, Negele, 5.23, 39.62, EU729493. M. tritoides (Baker) Schrire—J. Beckett 1395 (K), Somalia, 9.95, 45.18, EU729494. M. karinensis (Thulin) Schrire—Thulin et al. 10469 (UPS), Somalia, 11.3, 50.38, EU729495. M. sessilis (Thulin) Schrire—Thulin & Warfa 6263 (UPS), Somalia, 11.15, 49.78, EU729496. M. latifolia Benth.—Schrire 2571 (K), Kenya, K7, Malindi, –3.3, 39.95, AF274690. M. galpinii N.E.Br.—Schrire 2611(K), South Africa, Mpumalanga, Witklip Forest R., –25.22, 30.83, AF274689. M. remotiflora (Taub. ex Baker f.) Schrire—Brummitt et al. 14850 (K), Malawi, –15.68, 35.17, EU729497. M. phyllogramme (R. Vig.) Schrire, Du Puy & Labat—Humbert 28742 (K), Madagascar, Isalo, –22.8, 45, EU729498. M. spathulata (J.B. Gillett.) Schrire—Pope et al. 2163 (K), Zambia, –9.2, 29.32, EU729499. Rhynchotropis marginata (N.E.Br.) J.B. Gillett—Fanshawe F2458 (K), Zambia, Chingola, –12.52, 27.85, EU729500. R. poggei (Taub.) Harms [2]—Bingham 11708 (K), Zambia, –15.48, 28.47, AF274688. R. poggei (Taub.) Harms—Pawek 10791 (K), Malawi, –11.47, 33.42, AF521714. Palaeotropical clade: Indigofera kirkii Oliv.—Robertson 7111 (K), Kenya, Arabuko Sakoke Forest Res., –3.28, 39.98, AF521718. I. erythrogramma Welw. ex Baker—Bidgood et al. 1696 (K), Tanzania, Lindi Distr., –17.42, 32.23, EU729501. I. inhambanensis Klotzsch—Schrire 2588 (K), South Africa, Durban, Bluff, –29.93, 31, EU729502. I. strobilifera (Hochst.) ex Baker subsp. strobilifera—Schrire et al. 2572 (K), Kenya, Malindi Distr., –3.3, 39.95, EU729503. I. glabra L.—Rudd 3266 (K), Sri Lanka, Anuradhapura Distr., 8.57, 80.48, EU729504. I. biglandulosa J.B. Gillett—Bidgood & Vollesen 3072 (K), Tanzania, Kigoma Distr., –4.87, 29.63, EU729505. I. hermannioides J.B. Gillett—E. Milne-Redhead & Taylor 10126 (K), Tanzania, Songea Distr., Gumbiro, –10.27, 35.65, EU729506. I. gairdnerae Baker f.—M. Bingham 10056 (K), Zambia, Mongu Distr., –15.17, 23.1, EU729507. I. congesta Welw. ex Baker— Schrire et al. 2578 (K), Kenya, Kwale Distr., –4.22, 39.43, EU729508. I. macrocalyx Guill. & Perr.—J.E. Madsen et al. 22 (K), Burkina Faso, 11.68, –4.12, EU729509. I. heudolottii Benth.—Sharland 369 (K), Nigeria, Kaduna State, 9.6, 8.15, EU729510. I. paniculata Vahl. ex Pers.—Schrire 2574 (K), Kenya, Gongoni N.R., –4.38, 39.47, AF521720. I. tetrasperma Vahl ex Pers.—H. Ern 3205 (K), Togo, 8.13, 1.52, EU729511. I. pulchra Willd.—H. Ern 3083 (K), Togo, Dapango, 10.97, 0.13, EU729512. I. bracteolata DC.—J.E. Madsen 5866 (K), Burkina Faso, Boulgou Prov., 11.72, –0.28, EU729513. I. nigritana Hook. f.—H. Ern et al. 1348 (K), Togo, Savavas Region, 1.77, 0.2, EU729514. I. phymatodea Thulin— Thulin et al. 7545 (K), Somalia, Bay Prov., 2.8, 44.07, EU729515. I. compressa Lam.—Du Puy et al. M139 (K), Madagascar, Toliara, –25.5, 45.32, EU729516. I. wituensis Baker f.—Schrire et al. 2570 (K), Kenya, Malindi Distr., –3.3, 39.95, EU729517. I. mysorensis Rottl. ex DC.—van der Maesen 3980 (K), India, Andhra Pradesh, 16.08, 78.92, EU729518. I. wightii Graham ex Wight & Arn.—Chantaranothai et al. 90/422 (K), Thailand, Khonkaen, 16.5, 103, EU729519. I. uniflora Buch.-Ham. ex Roxb.—B.K. Vijay-Kumar 1285 (K), India, Maharastra, 15.9, 74.5, EU729520. I. microcalyx Baker—Baker et al. 3179 (K), Tanzania, Kigoma Distr., –4.92, 29.65, EU729521. I. demissa Taub.—Bidgood & Vollesen 3071 (K), Tanzania, Kigoma Distr., –4.87, 29.63, EU729522. I. simplicifolia Lam.—Eimunjeze & Latilo FHI 65635 (K), Nigeria, Kwara Borgu Distr., 10.4, 10.1, EU729523. I. omissa J.B. Gillett—Porembski & Biedinger 1352 (K), Liberia, Niemére, 8.58, –4.67, EU729524. I. tanganyikensis Baker f. var. tanganyikensis—Schrire & Stirton 2581 (K), Kenya, Machakos Distr., –1.5, 37.05, EU729525. I. monantha Baker f.—Bidgood et al. 2925 (K), Tanzania, Kigoma Distr., –5.68, 29.92, EU729526. I. nebrowniana J.B. Gillett—Schrire 2351 (K), South Africa, Pretoria, University campus, –25.75, 28.27, AF521791. I. bainesii Baker—C. Skarpe S 130 (K), Botswana, Ghazi, –23.12, 20.53, EU729527. I. vohemarensis Baill.—Schrire 2568 (K), Kenya, Mida Creek, –3.3, 39.95, AF274697. I. eremophila Thulin—Thulin & Warfa 5999 (K), Somalia, Bari Region, 11.52, 50.33, EU729528. I. Brevipatentes sp. nov.—Bidgood et al. 6347 (K), Tanzania, Kondoa Distr., –4.95, 35.8, EU729529. I. basiflora J.B. Gillett—H.M. Richards 23730 (K), Tanzania, Masai Distr., –2.8, 36.9, EU729530. I. suaveolens Jaub. & Spach—Gillett & Hemming 24078 (K), Somalia, 2.08, 44.18, EU729531. I. brevicalyx Baker f.—P. Rwaburindore 3299 (K), Uganda, W Ankole, –0.42, 30.48, EU729532. I. argentea Burm. f.—Miller et al. M0137 (E), Socotra, Ras [Vol. 96 Hawlaf, 12.7, 54.08, AF521785. I. rubroglandulosa Germishuizen— Schrire 2582 (K), South Africa, KwaZulu-Natal, Umtamvuna N.R., –30.9, 30.15, AF521761. I. hilaris Eckl. & Zeyh.—Schrire 2607 (K), South Africa, Mpumalanga, Sudwala’s Kraal, –25.38, 30.7, AF274694. I. rothii Baker—Ensermu & Petros E.1956 (K), Ethiopia, Hararge Prov., 9.42, 41.68, EU729533. I. mildbraediana J.B. Gillett—Jayeola 198 (K), Nigeria, 9.87, 8.88, AF521762. I. colutea (Burm. f.) Merr.—Miller et al. 19201A (E), Socotra, Haggeher Mtns, 12.62, 54.13, AF521776. I. podophylla Benth.—Groenendijk et al. 2092 (K), Mozambique, Maputo, –25.67, 32.58, EU729534. I. heterotricha DC.—Schrire 2433 (K), South Africa, Mpumulanga, Groblersdal, –25.35, 29.67, EU729535. I. poliotes Eckl. & Zeyh.—Schrire & Barker 2618 (K), South Africa, E Cape, Grahamstown, –33.32, 26.42, EU729536. I. grata E. Mey.—Schrire 2596 (K), South Africa, KwaZulu-Natal, Vernon Crookes N.R., –30.28, 30.6, AF521790. I. mimosoides Baker var. mimosoides—Brummitt et al. 16074 (K), Malawi, Lilongwe Distr., –14.33, 33.48, EU729537. I. quarrei Cronquist—J. Mutimushi 1913 (K), Zambia, Luano, –12.58, 27.93, EU729538. I. atriceps Hook.f. subsp. kaessneri (Baker f.) J.B. Gillett—Schrire 2565 (K), Kenya, Thika Distr., –0.97, 37.08, EU729539. I. trachyphylla Oliv.—Q. Luke et al. 8426 (K), Tanzania, Udzungwa, –7.68, 36.52, EU729540. I. mooneyi Thulin—Gilbert et al. 556 (K), Ethiopia, Sidamo Reg., 6.82, 37.83, EU729541. I. nyassica Gilli—J. Pawek 5084 (K), Malawi, Mzimba Distr., –11.47, 33.98, EU729542. Pantropical clade: Indigofera longimucronata Baker f.—Schrire 2573 (K), Kenya, Mida Creek, –3.3, 39.95, AF274695. I. deightonii J.B. Gillett— Porembski & Biedinger 1454 (K), Liberia. Mt Nianjbo, 8.82, –5.18, EU729543. I. karnatakana Sanjappa—Rudd & Balakrishnan 3142 (K), Sri Lanka, Polonnaruwa Distr., 8, 80.9, EU729544. I. astragalina DC.— J. Pawek 6538 (K), Malawi, –11.62, 34.3, EU729545. I. hirsuta L.—G. Prance 30489 (K), French Guiana, Illes du Salut, –5.3, 52.6, EU729546. I. sanguinea N.E.Br.—Schrire 2606 (K), South Africa, Mpumalanga, Buffelsklook N.R., –25.33, 30.48, AF521721. I. setiflora Baker—A. Strid 2847 (K), Zambia, C. Prov. Serenje Distr. Kundalila Falls, –13.17, 30.75, EU729547. I. melanadenia Benth. ex Harv.—Schrire 2383 (K), South Africa, Badplaats-Lake Chrissie, –26.07, 30.53, EU729548. I. leprieurii Guill. & Perr.—J.E. Madsen 6200 (K), Burkina Faso, Boulgou Prov., 11.68, –0.33, EU729549. I. stenophylla Guill. & Perr. [1910]—Friis et al. 7779 (K), Ethiopia, Gojjam Region, 11.18, 35.92, EU729550. I. stenophylla Guill. & Perr.—Miehe 407a (K), Sudan, 13, 24.2, EU729551. I. longibarbata Engl.—E.A. Banda 647 (K), Malawi, Mzimba, Vipya, –11.83, 33.8, EU729552. I. schlechteri Baker f.—Schrire 2354 (K), South Africa, Mpumalanga Prov., Verloorenvallei N.R., –25.23, 30.15, EU729553. I. laxiracemosa Baker f.—P. Schäfer 7189 (K), Mozambique, Maputo Distr. Marracuene, –25.67, 32.58, EU729554. I. varia E. Mey.— Gilbert 6250 (K), Kenya, –2.53, 37.85, AF521730. I. vicioides Jaub. & Spach—G. Germishuizen 2459 (K), Namibia, Rehoboth Distr., –23.7, 17.4, EU729555. I. dendroides Jacq.—Schrire et al. 2579 (K), Kenya, Kwale Distr. Vuga, –4.18, 39.67, EU729556. I. bojeri Baker—, Madagascar, Nanokely, –19.55, 47.15, J. & M. Peltier 1944 (K), EU729557. I. frondosa N.E.Br.—Schrire 2613 (K), South Africa, Mpumalanga, Verlooren Vallei N.R., –25.32, 30.15, AF521722. I. tristis E. Mey.—Schrire 2591 (K), South Africa, KwaZulu-Natal, Silverglen N.R., –29.93, 30.88, AF521724. I. hedyantha Eckl. & Zeyh.—Barker 1788 (GRA), South Africa, Grahamstown, Featherstonekloof, –33.37, 26.53, EU729558. I. verrucosa Eckl. & Zeyh.—Barker 1548 (GRA), South Africa, Hogsback, –32.6, 26.87, AF521723. I. zeyheri Spreng. ex Eckl. & Zeyh.— Barker 1431a (GRA), South Africa, Port Alfred, –33.6, 26.9, AF274698. I. socotrana Vierh.—Thulin & Gifri 8866 (K), Socotra, Tinire, 12.58, 54, EU729559. I. frutescens L.f.—Barker 1738 (GRA), South Africa, W Cape, Gifberg Pass, –31.75, 18.03, EU729560. I. frutescens L.f. [2]—Barker & McKenzie s.n. (GRA), South Africa, W Cape, Goudini Spa, –33.65, 19.77, EU729561. I. langebergensis L. Bol.—L. Bolus 19514 (K), South Africa, W Cape, Garcia’s Pass, –33.95, 21.28, EU729562. I. jucunda Schrire—Schrire 2592 (K), South Africa, KwaZulu-Natal, Umgeni R., –29.82, 31.02, AF521728. I. natalensis H. Bol.—Schrire 2599 (K), South Africa, KwaZulu-Natal, –30.97, 30.15, New Germany, AF521726. I. Psiloceratiae sp. nov. (= I. braamtonyi nom. nud.)—Schrire 2584 (K), South Africa, KwaZulu-Natal, Umtamvuna N.R., –30.97, 30.15, AF521729. I. roseocaerulea Baker f.—P. Lovett & Kayombo 410 (K), Tanzania, Mbeya Distr. Itimba, –8.83, 33.33, EU729563. I. sutherlandoides Welw. ex Baker—S. Hooper & Townsend 651 (K), Zambia, C Prov. Kapiri Mposhi, –13.87, 28.63, EU729564. I. April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae fulgens Baker—Clarke 62 (K), Tanzania, –9.97, 39.45, EU729565. I. baumiana Harms—Robinson 6736 (K), Zambia, Mongu Distr., –15.27, 23.13, EU729566. I. mangokyensis R. Vig.—Goldblatt & Schatz 8984 (K), Madagascar, Fianarantsoa, –22.18, 46.92, EU729567. I. lyallii Baker—Goldsmith 56/68 (K), Zimbabwe, Melsetter, –19.98, 32.95, EU729568. I. bosseri Du Puy & Labat—J. & M. Peltier 3017 (K), Madagascar, Isalo, –22.4, 45.28, EU729569. I. himalayensis Ali—YUYU 10941 (K), India, Himalayas, Kew 1990 – 8017, 32.3, 77.2, AF534792. I. cylindracea Graham ex Baker—R. Bedi 552 (K), Bhutan, Motithang, 27.4, 90.3, EU729570. I. heterantha Wall. ex Brandis—Billiet & Leonard 6709 (K), India, Kashmir, 34.1, 74.7, EU729571. I. hebepetala Benth. ex Baker—Howick & McNamara (HOMC) 1864 (K), India, Himalayas, Kew 1996 – 12, 32.3, 76.9, AF534793. I. pendula Franch.—Alpine Expedition to China (ACEX) no. 1148 (K), China, Yunnan, Kew 1996 – 13, 23.87, 100.13, AF521766. I. amblyantha Craib—SICH 90 (K), China, Kew 1988 – 8607, 31.7, 103.9, AF534791. I. nigrescens Kurz ex King & Prain—Kingdom Ward 8517 (K), India, Assam, Delei Valley, 28.33, 96.58, EU729572. I. dosua Buch.-Ham. ex D. Don—Schilling & Sayers 349 (K), Nepal, Kathmandu valley, 27.7, 85.2, AF534790. I. cassioides Rottl. ex DC.—K. Mathew RHT 52911 (K), India, Tamil Nadu, Kodaikanal, 10.4, 78, EU729573. I. lacei Craib—K. Larsen 34199 (K), Thailand, N Maehongson, 18.25, 98, EU729574. I. atropurpurea Buch.-Ham. ex Horn.—A. Bullock 622 (K), India, Manipur, 25, 93.9, EU729575. I. kirilowii Maxim. ex Palibin—Choi 369 (IUI), Korea, 37.4, 126.7, AF534796. I. grandiflora B.H. Choi & S.K. Choi—Park 151 (IUI), Korea, 35.8, 128.1, AF534795. I. koreana Ohwi—Choi et al. 9113 (IUI), Korea, 37, 127.3, AF534794. I. decora Lindl.—Ohashi 20679 (IUI), Japan, 38.2, 140.7, AF534797. I. venulosa Champ. ex Benth.—Endo 2018 (IUI), Taiwan, 24, 120.8, AF534798. I. rhynchocarpa Welw. ex Baker—Schmidt et al. 1133 (K), Tanzania, –8.48, 31.5, AF521787. I. podocarpa Baker f. & Martin—Hooper & Townsend 53 (K), Zambia, Solwezi Distr., 12, 25.4, EU729576. I. swaziensis H. Bol.—Schrire 2609 (K), South Africa, Mpumalanga, Witklip Dam, –25.22, 30.83, AF521788. I. macrophylla Schum.—De Wilde 953 (K), Ivory Coast, nr. Bouaké, 9.6, –7.4, EU729577. I. emarginella A.Rich.—P. Kasper 31 (K), Sudan, Loka West, Kajelu, 4.1, 30.6, EU729578. I. binderi Kotschy—Moyale et al. 4118 (K), Ethiopia, Sidamo, 3.55, 39.05, EU729579. I. homblei Baker f. & Martin—Brummitt et al. 16288 (K), Malawi, Chitapa Distr., –9.98, 33.35, EU729580. I. subcorymbosa Baker—B.J. Coetzee 1333 (K), South Africa, Wyliespoort, –22.9, 29.93, EU729581. I. sedgewickiana Vatke & Hild.—Gillett & Watson 23455 (K), Somalia, 10.92, 49.43, EU729582. I. boranica Thulin—Gilbert et al. 8090 (K), Ethiopia, 5.18, 48.18, EU729583. I. caloneura Kurz (= I. oblonga Craib)—Larsen et al. 31501 (K), Thailand, Phukhieo, 16.42, 102.08, AF521765. I. galegoides DC.—K. & S. Larsen 34356 (K), Thailand, Maehongson, 19.25, 98, EU729584. I. zollingeriana Miq.—Furuse 3837 (K), Japan, Yoshino, 24.4, 124.2, AF521772. I. sootepensis Craib—K. Larsen (K. Boonyamalik) 34072 (K, P), Thailand N, Maehongson, 18.25, 98, EU729585. I. australis Willd.—Story 6641 (K), Australia, Cessnock, –32.9, 151.4, AF521674. I. New Caledonia sp. nov.—H.S. Mackee 23622 (K), New Caledonia, Tieta, –20.9, 164.7, EU729586. I. ixocarpa Peter G. Wilson & Rowe—P. Wilson & Rowe 1080 (K), Australia, WA, Fortescue, –22.72, 117.73, EU729587. I. verruculosa P.Wilson—Dunlop 4943 (K), Australia, N Territory, Little Nourlangie Rock, –12.87, 132.8, EU729588. I. rugosa Benth.—Wilson & Rowe 1024 (K), Australia, Hamersley Range, –22.05, 117.68, AF521773. I. haplophylla F. Muell.—N. Byrnes 2508 (K), Australia, NT, Daly Waters, –16.35, 135.3, EU729589. I. Australia Grp. sp. nov. 1 (= I. decipiens nom. nud.)—Wilson & Palmer 1777 (K), Australia, WA Fortescue, Seven Mile Creek, –23.33, 116.95, EU729590. I. pratensis F. Muell.—Clarkson 4213 (K), Australia, Queensland, –15.63, 143.97, AF521763. I. georgii E. Pritzel—P. Wilson & Rowe 507 (K), Australia, Queensland. Warrego, –25.95, 144.68, EU729591. I. boviperda Morrison subsp. boviperda— P.G. Wilson 1804 (K), Australia, WA Fortescue, –23.93, 114.85, EU729592. I. Australia Grp. sp. nov. 2 (= I. cuspidata nom. nud.)—Wilson & Palmer 1776 (K), Australia, WA Fortescue, Turee Creek, –23.33, 116.95, EU729593. I. arrecta Hochst. ex A. Rich.—Schrire 2563 (K), Kenya, Karatina, –0.48, 37.13, AF521727. I. amorphoides Jaub. & Spach— Gilbert et al. 7373 (K), Ethiopia, Shewa, Awash R., 8.83, 40.03, EU729595. I. tinctoria L.—Miller et al. 19224 (E), Socotra, Haggeher Mts., 12.58, 54.05, AF521775. I. longiracemosa Baill.—Schrire 2567 (K), Kenya, Arubuko Sokoke F., –3.3, 39.95, AY124764. I. cavallii Chiov.—Thulin et al. 6867 (K), Somalia, Bay, 1.8, 42.8, EU729596. I. articulata Gouan— Miller et al. 19052 (E), Socotra, Samha, 12.22, 53.03, AF521782. I. 845 coerulea Roxb. var. occidentalis J.B. Gillett & Ali—Miller et al. DA22 (E), Socotra, Qeyso-Maale, 12.62, 53.52, AF521783. I. conzattii Rose— Hughes 2075 (FHO), Mexico, Puebla, 19, –97.9, AF521585. I. thibaudiana DC.—Hughes 2105 (FHO), Mexico. Oaxaca, 15.7, –96.5, AF521586. I. blanchetiana Benth.—Thomas et al. 9641 (K), Brazil, Bahia, Juazeiro, –10, –42.2, EU729597. I. cuernavacana Rose—Gallardo et al. 182 (K), Mexico, Guerrero, 17.7, –101.52, EU729598. I. platycarpa Rose—Lavin et al. 5118 (K), Mexico, Michoacun, 18.1, –102.2, EU729599. I. caroliniana Miller—A.B. Pittman 07100210 (MONT), USA, South Carolina, Florence Co., 34, –79.7, EU729600. I. byobiensis Hosok.— Ohashi & Endo 20337 (K), Taiwan, Maopitou, Pingtung, 22.7, 120.5, EU729601. I. suffruticosa Miller—Hu 1102 (DAV), Brazil, –23.5, –46.2, AF467051. I. truxillensis Kunth—Lewis & Klitgaard 2198 (K), Ecuador, La Loja, –8.73, –77.93, AF521777. I. truxillensis Kunth [1677]—Hughes 2223 (FHO), Peru. –3.97, –79.45, EU729602. Cape clade: Indigofera nudicaulis E.Mey.—Giess & Muller 12253 (K), Namibia, Warmbad Distr., –28.67, 18.22, EU729603. I. nudicaulis E.Mey. [2080]—J.P. Rourke 2080 (NBG), South Africa, Oranjemund Distr., Richtersveld, –28.15, 16.88, EU729604. I. merxmuelleri Schreiber— Merxmüller & Giess 3445 (K), Namibia, Luderitz Distr., Witputz, –18.55, 16.65, EU729605. I. Digitatae sp. nov.—Barker 1709 (K), South Africa, Richtersveld N.P., De Koei camp, –28.28, 17.02, EU729606. I. cuneifolia Eckl. & Zeyh.—Schrire & Barker 2636 (K), South Africa, E Cape, Suurberg Pass, –33.32, 25.77, AF521749. I. dimidiata Vogel ex Walp.—Barker 1773 (GRA), Lesotho, Molimo Nthuse, –29.43, 27.78, AF521792. I. mollis Eckl. & Zeyh.—R. Clarke & G. Coombs 395 (GRA), South Africa, Graaff Reinet, “Asante Sana”, –32.27, 24.87, EU729607. I. alpina Eckl. & Zeyh.—R. Clarke et al. 10 (GRA), South Africa, Graaff Reinet, “Asante Sana”, –32.27, 24.87, EU729608. I. meyeriana Eckl. & Zeyh.—F. Forest 413a (K), South Africa, Laingsburg Distr., –33.37, 21.45, EU729609. I. digitata Thunb.—F. Forest 447b (K), South Africa, W Cape, –32.13, 18.95, EU729610. I. burchellii DC.—R. Clarke & G. Coombs 14 (GRA), South Africa, Graaff Reinet, “Asante Sana”, –32.27, 24.87, EU729611. I. amoena Ait.—Barker 1739 (GRA), South Africa, Gifberg Pass, –31.75, 18.03, AF521789. I. psoraloides (L.) L.—Barker 1861 (GRA), South Africa, Clanwilliam, Pakhuis Pass, –32.15, 19.05, EU729612. I. porrecta Eckl. & Zeyh.—Schrire & Barker 2626 (K), South Africa, Knysna, Brenton, –34.08, 23.03, AF521750. I. heterophylla Thunb.—Schrire & Barker 2622 (K), South Africa, W Cape, Misgund, –33.75, 23.47, AF521751. I. Olygophyllae sp. nov. 1—Taylor 365 (K), South Africa, Montague, Kogmanskloof Point, –33.82, 20.08, EU729613. I. Olygophyllae sp. nov. 2 (= I. salteri nom. nud.)—Schrire 2490 (K), South Africa, Namaqualand, Doornriver bridge, –31.87, 18.65, EU729614. I. Olygophyllae sp. nov. 3—Barker 1720 (GRA), South Africa, NW Cape, Spektakel Pass, –29.67, 17.63, EU729615. I. aff. nigromontana Eckl. & Zeyh.—Barker 1732 (GRA), South Africa, N Cape Prov., N7 turnoff to Soebatsfontein, –30.23, 17.88, EU729616. I. denudata L.f.—Schrire & Barker 2619 (K), South Africa, nr. Jeffrey’s Bay, –33.93, 24.98, AF521753. I. denudata L.f. [2635]—Schrire & Barker 2635 (K), South Africa, E. Cape, Highlands to Alicedale, –33.33, 26.15, EU729617. I. angustata E. Mey.—Schrire & Barker 2617 (K), South Africa, Grahamstown, –33.32, 26.42, AF521752. I. filifolia Thunb.—Schrire & Barker 2645 (K), South Africa, Vogelgat Kloof R., –34.4, 19.32, AF521760. I. ionii J.K. Jarvie & C. H. Stirton—Schrire & Barker 2648 (K), South Africa, Betty’s Bay, –34.37, 18.87, AF521748. I. gifbergensis Stirton & J.K. Jarvie—F. Forest 446 (K), South Africa, W Cape, –32.15, 19.05, EU729618. I. cytisoides (L.) L.—Schrire & Barker 2644 (K), South Africa, Fernkloof N.R., –34.4, 19.27, AF521754. I. capillaris Thunb.—Schrire & Barker 2638 (K), South Africa, Ceres, Mitchells Pass, –33.4, 19.28, AF274692. I. concava Harv.—D. McDonald 1177 (K), South Africa, Boosmansbos, –34, 20.8, EU729619. I. declinata E, Mey.—Barker 1783 (GRA), South Africa, Tsitsikamma National Park, –33.87, 23.88, EU729620. I. filicaulis Eckl. & Zeyh.—Schrire & Barker 2639 (K), South Africa, Ceres, –33.37, 19.3, AF521759. I. hispida Eckl. & Zeyh.—Schrire & Barker 2620 (K), South Africa, Kareedouw Pass, –33.97, 24.27, AF521755. I. flabellata Harv.— Schrire 2623 (K), South Africa, Noetsie, –34.08, 23.12, EU729621. I. brachystachya (DC.) E. Mey.—F. Forest 427a (K), South Africa, W Cape, –34.42, 20.63, EU729622. I. Brachypodae sp. nov.—Barker 1794 (GRA), South Africa, Knysna, Montagu Pass, –33.88, 22.45, EU729623. I. sulcata DC.—Barker 1913 (GRA), South Africa, Grahamstown, –33.37, 26.53, EU729624. I. ovata Thunb.—Vlok et al. 32 (K), South Africa, Kleinmond, –34.3, 19, EU729625. I. angustifolia L.—Schrire & 846 American Journal of Botany Barker 2643 (K), South Africa, Fernkloof N.R., –34.4, 19.27, AF521756. I. sarmentosa L.f.—Schrire & Barker 2646 (K), South Africa, Vogelgat N.R., –34.38, 19.32, AF521758. I. glomerata E. Mey.—Schrire & Barker 2640 (K), South Africa, Shaw’s Pass, –34.32, 19.42, EU729626. I. alopecuroides (Burm. f.) DC.—Schrire & Barker 2641 (K), South Africa, Shaw’s Pass, –34.32, 19.42, AF521757. I. superba C.H. Stirt.— Schrire 2507 (K), South Africa, Hermanus, –34.38, 19.32, EU729627. I. candolleana Meisn.—Crisp 9069 (CANB), South Africa, Cape Town, –33.93, 18.43, AF287641. I. mauritanica (L.) Thunb.—Schrire & Barker 2649 (K), South Africa, Table Mtn., –33.95, 18.42, AF521794. Tethyan CLADE: Indigofera ammoxylum (DC.) Polhill (= Bremontiera ammoxylum DC.)—Strasberg s.n. (K), La Réunion, –21.2, 55.6, AF274699. I. ammoxylum (DC.) Polhill (= Bremontiera ammoxylum DC.) [1665]—Friedmann 2141 (K), La Reunion. Grande Chaloupe, –21.1, 55.5, EU729628. I. linifolia (L.f.) Retz—M. Evans 3569 (K), Australia, N.T., Edith Falls, –14.17, 139.15, EU729629. I. dalzellii T. Cooke—Vijay Kumar 1278 (K), India, 18, 75, AF521793. I. cordifolia Heyne ex Roth— Edwards et al. 3700 (K), Eritrea, 15.58, 29.33, AF521741. I. microcarpa Desv.—Hooper & Townsend 1120 (K), Kenya, –2.4, 40.2, EU729630. I. nummulariifolia (L.) Livera ex Alston—Bidgood et al. 617 (K), Tanzania, Iringa Distr., –8.22, 35, EU729631. I. drepanocarpa Taub.—Kahurananga et al. 2776 (K), Tanzania, Kahama Distr., –3.8, 32.6, EU729632. I. trifoliata L.—J.S. Beard 8326 (K), Australia, W.A., Mitchell Plateau, –14.87, 125.77, AF521746. I. squalida Prain—C.F. von Beusekom & C. Phengkhlai 1071 (K), Thailand, Tak Distr, 17, 99.1, EU729633. I. glandulosa Wendl.—van der Maesen 2826 (K), India, Hyderabad, 17.5, 78.3, EU729634. I. leucoclada Baker—Du Puy et al. M270 (K), Madagascar, –12.22, 49.17, AF521743. I. bemarahaensis Du Puy & Labat—Du Puy et al. 2653 (K), Madagascar, –18.68, 44.7, AF521744. I. exellii Torre—Vollesen MRC 2270 (K), Tanzania, Selow Res., –8.47, 38.47, EU729635. I. exellii Torre [1721]—Thulin et al. 6798 (K), Somalia, Bay, Buur Heybo, 2.98, 44.27, EU729636. Vaughania pseudocompressa Du Puy, Labat & Schrire—Du Puy et al. M873 (K), Madagascar, –25.02, 46.38, AF274701. V. depauperata Du Puy, Labat & Schrire—Du Puy et al. M1032 (K), Madagascar, –24.67, 43.93, AF521737. V. cloisellii (Drake) Du Puy, Labat & Schrire—P. Phillipson et al. 3484 (K), Madagascar, Toliara, –23.3, 44.28, EU729637. V. dionaeifolia S. Moore— A. Richard 41 (K), Madagascar, Amipijonoa, –15.98, 46.93, EU729638. V. longidentata Du Puy, Labat & Schrire—Du Puy et al. M145 (K), Madagascar, Toliara, –24.87, 56.58, EU729639. V. cerighellii Du Puy, Labat & Schrire—Du Puy et al. M684 (K), Madagascar, –22.62, 45.35, AF521740. V. interrupta Du Puy, Labat & Schrire—P. Phillipson et al. 3427 (K), Madagascar, Toliara, –24.6, 44.68, EU729640. V. mahafalensis Du Puy, Labat & Schrire—Labat et al. 2451 (K), Madagascar, –23.48, 43.77, AF521736. V. humbertiana Du Puy, Labat & Schrire—Du Puy et al. M687 (K), Madagascar, –23.42, 43.78, AF274700. Indigofera anabibensis Schreiber—de Winter & Leistner 5704 (K), Namibia, Kaokoveld, Otjihu, –18.33, 13.33, EU729641. I. sessiliflora DC. —B.V. Shetty 2327 (K), Sind, Jodhpur, Ramika, 26.3, 73.1, EU729642. I. trigonelloides Jaub. & Spach—Liebenberg 4947 (K), Namibia, Fransfontein, –20.22, 15.02, EU729643. I. auricoma E. Mey.—G. Germishuizen 2513 (K), Namibia, Karibib Distr., –21.7, 16, EU729644. I. hololeuca Benth. ex Harv.—Thompson & Le Roux 166 (K), South Africa, NW Cape, –28.38, 17.17, EU729645. I. daleoides Benth. ex Harv.— Brown 6021 (K), Botswana, Ghanzi, –22.4, 21, EU729646. I. alternans DC.—Schrire 2440 (K), South Africa, Free State Province, –28.17, 27.1, EU729647. I. fanshawei J.B. Gillett—Brummitt et al. 16951 (K), Zambia, –13.3, 30.2, EU729648. I. aspera Perr. ex DC.—Lock 43970 (K), Ghana, Srogboa Wot Anloga, 5.8, 0.9, EU729649. I. pungens E. Mey.—Barker 1710 (GRA), South Africa, Richtersveld National Park, –28.28, 17.03, EU729650. I. depressa Harv.—C.H. Stirton 10264 (K), South Africa, S Cape, Platbos, –34.1, 21.2, EU729651. I. leptocarpa Eckl. & Zeyh.— Schrire & Barker 2631 (K), South Africa, E Cape, –33.83, 24.92, AF521742. I. glaucescens Eckl. & Zeyh.—Schrire 2462 (K), South Africa, Port Elizabeth, –34.05, 25.57, EU729652. I. obcordata Eckl. & [Vol. 96 Zeyh.—Acocks 17133 (K), South Africa, Laingsburg, –33.2, 20.9, EU729653. I. disticha Eckl. & Zeyh.—Schrire & Barker 2615 (K), South Africa, Hellpoort Pass, –33.08, 26.23, AF264793. I. torulosa E. Mey.— Schrire 2359 (K), South Africa, Mpumalanga, Potloodspruit, –25.03, 30.47, AF521774. I. sessilifolia DC.—P. Herman 395 (K), South Africa, FreeState, Verwoerd Dam, –30.5, 21.2, EU729654. I. praticola Baker f.—Smith 4437 (K), Botswana, Boro R., –19.85, 23.52, EU729655. I. hochstetteri Baker—Miller et al. M10182 (E), Socotra, Hammaderoh, 12.62, 54.23, EU729656. I. arabica Jaub. & Spach—Gilbert & Thulin 96 (K), Ethiopia, Harar, 9.58, 41.55, EU729657. I. asperifolia Bong.—A. Delgado Salinas 2001 (MONT), Peru, Cajamarca, Llacanora, –7.2, –78.4, EU729658. I. tephrosioides Kunth—Lewis & Klitgaard 3003 (K), Ecuador, La Loja, –3.97, –79.45, AF521781. I. tephrosioides Kunth [1760]—A. Delgado-Salinas 2105 (MONT), Peru, Cajamarca, San Pablo, –7.13, –78.83, EU729659. I. hartwegii Rydb.—M. Sousa et al. 5870 (K), Mexico, Oaxaca, 17.3, –96.5, EU729660. I. lespedezioides Kunth— Pennington 908 (E), Bolivia, –17.8, –63.2, AF521780. I. bongardiana (Kuntze) Burkart—Sakuragui et al. 419 (K), Brazil, Sáo Paulo, Itararé, –24.1, –49.3, EU729661. I. leptosepala Nutt. ex Torrey & A. Gray— Brenan et al. 14327 (K), Mexico, Oaxaca, 17, –96.3, EU729662. I. miniata Ort.—Lavin 5165 (K), Mexico, La Guerrero, 17.5, –99.5, EU729663. I. conjugata Baker—Brummitt et al. 16236 (K), Malawi, Chitipa Distr., Mafinga Mtns., –9.98, 33.35, EU729664. I. thomsonii Baker f.—H.M. Richards 13086 (K), Tanzania, Ufipa Distr. Mbisi Forest, –7.87, 31.67, EU729665. I. achyranthoides Taub.—R. Germain 709 (K), D.R. Congo, Dungu, Basape-River, 3.67, 28.53, EU729666. I. bongensis Kotschy & Peyr.—J. Myers 7001 (K), Sudan, Mt. Nakbi, 5, 28.08, EU729667. I. antunesiana Harms—Brummitt 11528 (K), Malawi, –13.12, 33.95, EU729668. I. semitrijuga Forssk.—M. Thulin & Warfa 6014 (K), Somalia, Bari Region, 11.37, 50.05, EU729669. I. linnaei Ali—Phengklai et al. 4249 (K), Thailand, Kanchanaburi, 14.4, 99.3, EU729670. I. spicata Forssk.—A. Delgado-Salinas 2051 (MONT), Peru, Amazonas, Bagua, –5.6, –78.5, EU729671. I. spicata Forssk. [1899]—I. Friis et al. 9736 (K), Ethiopia, Kefa Reg. 20 km W of Dimma, 6.57, 34.97, EU729672. I. diphylla Vent.—C. Pase 3132 (K), Niger, Dakoro, 14.58, 6.67, EU729673. I. volkensii Taub.—I. Friis et al. 8322 (K), Ethiopia, Sidamo Reg. N of Yavello, 4.9, 38.15, EU729674. I. diversifolia DC.— Du Puy et al. M66 (K), Madagascar, Toliara, –23.68, 44.63, EU729675. I. nephrocarpa Balf. f.—Miller et al. 14171 (E), Socotra, Abd al-Kuri, 12.32, 52.63, EU729676. I. nephrocarpoides J.B. Gillett—Miller et al. 19202B (E), Socotra, E Haggeher Mts., 12.52, 54.2, EU729677. I. hiranensis Thulin—Gillett et al. 22625 (K), Somalia, 4.03, 45.75, EU729678. I. lupatana Baker f.—Schrire 2561 (K), Kenya, Jaratina, –0.48, 37.13, EU729679. I. schimperi Jaub. & Spach—Schrire 2564 (K), Kenya, Tana R., –0.67, 37.2, AF274696. I. oblongifolia Forssk.—Miller et al. 19139 (E), Socotra, Nogad Plain, 12.33, 54.02, AF521778. I. marmorata Balf. f. [1718]—Miller et al. 19221 (E), Socotra, Haggeher Mts., 12.58, 54.05, AF521779. I. marmorata Balf. f.—Miller et al. 10370 (K), Socotra, Wadi Daneghari, 12.62, 54.1, EU729680. I. trita L.f.—K. Mathew RHT 6638 (K), India, Madras, Tiruchi, 10.8, 78.7, EU729681. I. subulata Vahl ex Poir.—Schrire 2569 (K), Kenya, Arubuko Sokoke F., –3.3, 39.95, AF521745. I. angulosa Edgw.—Remanandan 4561 (K), India, M. Pradesh, Palasner, 21.53, 75.03, EU729682. I. gypsacea Thulin—Thulin et al. 7307 (K), Somalia, Galguduud, 4.7, 46.62, EU729683. I. ewartiana Domin—G. Chippendale 7088 (K), Australia, NT, Brunchilly, –18.02, 134.48, EU729684. I. kelleri Baker f.—Thulin et al. 3781 (K), Ethiopia, Bale Region, Web R., 6.92, 40.8, EU729685. I. spiniflora Hochst. & Steud. ex Boiss.—Miller et al. DA23 (E), Socotra, Qeyso-Maale, 12.58, 53.52, EU729686. I. spinosa Forssk.—Edwards et al. 3686 (K), Eritrea, 15.58, 39.33, AF521784. I. guaranitica Hassl.— T.M. Pedersen 7724 (K), Argentina, Corrientes, –28.4, –58.2, EU729687. I. guaranitica Hassl. [1634]—Irwin et al. 13978 (K), Brazil, Brasilia DF, –15.6, –47.9, EU729688. I. jamaicensis Spreng.—P. Lemus s.n. (Berendsohn 1179) (K), El Salvador, Bejuco, 13.67, –89.25, EU729689. I. jamaicensis Spreng. [1629]—Berendsohn et al. 1402 (K), El Salvador, 13.82, –89.93, EU729690. April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 847 Appendix 2. Morphological and biogeographical characters and character states, with examples given where states are synapomorphies of clades. After the list of character states, the range of the tree length (tl), consistency index (ci), and retention index (ri), respectively, over 10 000 most parsimonious trees are reported in square brackets for each character. For 51 of the 80 following morphological characters, a report of a single number in these three categories signifies no range of values. 1. Habit. 0 = annual to short-term perennial herbs from a taproot; 1 = suffrutices sprouting from a woody rootstock; 2 = resprouting shrubs, trees, lianas; 3 = obligate reseeding woody shrublets or shrubs with fire-cued seed germination and recruitment [47–48, 0.067–0.063, 0.769–0.753]. The switch from trees or lianas to annual or short-term, perennial herbs (state 0), a synapomorphy of the CRIM + Indigofera clade (Fig. 1), is considered a paedomorphic developmental event (Calow, 1983), possibly arising as a response to the arid or open environments constraining evolution in the tribe. The development of a suffrutescent habit (state 1) in the grass biome genus Rhynchotropis (Fig. 2) is associated with the production of precocious flowering, before the leaves are produced, in fire-prone environments, thus likely avoiding competition with the developing grass sward in immediate post fire conditions. Later, normal shoots emerge together with leaves within the regenerated herbaceous layer. In Indigofera, state 1 is a synapomorphy for the Cape I. Digitatae sp. nov.–I. heterophylla (Fig. 5) and pantropical I. verrucosa–I. tristis and I. arrecta–I. platycarpa (Fig. 4) subclades, while parsimony character optimizations are equivocal about state 1 being a synapomorphy for the Tethyan subclade I. asperifolia–I. bongenis (Fig. 6). Reversals to a shrubby habit (state 2) are apparently correlated with either arid to semi-arid environments associated with succulent biome clades, e.g., the palaeotropical I. vohemarensis–I. bainesii (Fig. 3) and Tethyan I. hiranensis–I. bongensis and I. leucoclada–Vaughania humbertiana subclades (Fig. 6), or cooler high altitude tropical or temperate latitude environments associated with the temperate biome, e.g., in the pantropical I. socotrana–I. verruculosa subclade (Fig. 4) and the Cape clade (Fig. 5). Obligate reseeding (state 3) is a synapomorphy marking the monophyly of the Cape I. cytisoides–I. heterophylla subclade (Fig. 5). Like character 17, this state represents a major switch to species becoming locked into the nutrient-poor, acidic soils of the fynbos region (specifically the sandstone and limestone fynbos vegetation types) within the Cape (Rebelo et al., 2006). Obligately reseeding fire ephemerals, or more long-lived shrubs, invest all their resources in seed production instead of regeneration because they are killed by fire. Fire ephemerals appear during a brief period of one to several years after fire, then perish and persist as dormant seeds in the soil (Cowling and Richardson, 1995). Within the fire-controlled sandstone or limestone fynbos environment, reproduction is largely driven by recurrent fires, and species of Indigofera, whether fire ephemerals or more longer lived reseeding shrubs, are often habitat generalists that are relatively widely distributed, at least within major subunits of the Cape Floristic Region (Le Maitre and Midgley, 1992; Schutte et al., 1995; Rebelo et al., 2006). Many fynbos shrubs have a distinctive morphology with an open branch arrangement rather than a dense canopy, allowing effective heat dissipation from leaves (Cowling and Richardson, 1995) and a flexible response to the strong winds that are an integral part of the fynbos environment. Plants have reduced lateral branch development with leaves clustered around the stems giving a tailor rod-like appearance, or the lateral branches are one to many, about equally long and ± parallel to the main stems (Dahlgren, 1963). 2. Biramous hairs. 0 = absent, 1 = arms all ± equal except on calyx lobes, 2 = some hairs with arms very unequal in length [45–49, 0.044–0.041, 0.602– 0.565]. The most readily diagnostic synapomorphy of the Indigofereae (Fig. 1) is the presence of biramous hairs (Polhill, 1981). This development parallels that in other dry-adapted, largely shrubby or herbaceous and speciose legume groups, e.g., Astragaleae (Hutchinson, 1964), and a number of genistoid tribes, e.g., Australian Brongniartieae, Crotalarieae, and Genisteae (Polhill, 1976). Such hairs have anatomical features fitting them for water absorption (Lyshede, 1977; Schrire, 1991), and foliar uptake of water has been demonstrated in many plant species (e.g., Seely et al., 1977; Rundel, 1982; Alvin, 1987; Grammatikopoulos and Manetas, 1994). State 2 is a proxy for a spreading indumentum of variable thickness arising from overlapping hairs where one arm is much longer than the other. 3. Hairs. 0 = all hyaline, 1 = mixed hyaline and orange, brown, or black [23, 0.043, 0.815]. Loss of brown hairs is a synapomorphy for the CRIM + Indigofera clade (Fig. 1), and parallel regains within Indigofera characterize the pantropical clade (Fig. 4), the Tethyan I. leucoclada–V. humbertiana subclade (Fig. 6) and the palaeotropical I. quarrei–I. nyassica and I. bracteolata–I. tetrasperma subclades (Fig. 3). 4. Biramous hairs (sometimes dendritic) appearing apically from epidermal papillae. 0 = absent, 1 = present [1, 1.000, 1.000]. State 1 is a synapomorphy of the Cape subclade I. flabellata–I. brachystachya (Fig. 5). Marloth (1903, 1910) discussed the role of plant hairs in water uptake from mists off the Cape Mountains. State 1 is thus seen as further enhancing this function, providing larger surfaces for condensation both on the hair branches and in the collar surrounding the hair bases. The role of biramous hairs and active water absorption is mentioned in character 2. The distribution of this clade is restricted to the ocean-facing slopes and summits along the southern, west-to-east-running mountain ranges of the Cape. The epidermal papillae in this clade have a chimney-like appearance, forming a collar through which the long-stalked biramous or sometimes dendritic hairs project. Hairs become dendritic where the biramous branches have further subdivided. 5. Subsimple or simple spreading bristle-like hairs. 0 = absent, 1 = present [10–11, 0.100–0.091, 0.550–0.500]. Bristle-like hairs often without an apparent second arm at the base occur sporadically in the pantropical clade (Fig. 4). Parsimony character optimizations are equivocal about state 1 being a synapomorphy of the palaeotropical I. glabra–I. hermannioides subclade (Fig. 3). 6. Multiseriate glandular capitate trichomes with a multicelled stalk and head. 0 = absent, 1 = present [2, 0.500, 0.909]. State 1 is a synapomorphy of the palaeotropical I. argentea–I. nyassica subclade (Fig. 3), enhancing plant protection (Schrire 1991, 1995). In many taxa, the gland-tipped trichomes produce a sticky, often yellowish, exudate with a strong odor, and their presence is variable in some species, being dense in some individuals and almost absent in others. The sticky exudate is likely to act as both a physical impediment and chemical deterrent to herbivores (Schrire, 1991). 7. Pearl bodies on lower leaflet surfaces. 0 = absent except perhaps on juvenile foliage, 1 = present always on mature leaflets [9, 0.111, 0.600]. State 1, which is identified by numerous, often microscopic pearl bodies covering the lower leaflet surfaces, is a synapomorphy of the Tethyan I. microcarpa–V. humbertiana subclade (Fig. 6). The pearl bodies discussed in character 8 are independent of this character. 8. Pearl bodies on stems and inflorescences. 0 = absent, 1 = stalked pearl bodies sparse and mostly restricted on the stems to leaf axils and also between leaflets, 2 = stalked pearl bodies densely scattered on the stems, 3 = sessile verrucate pearl bodies densely scattered, 4 = sessile globose pearl bodies densely scattered [33–35, 0.121–0.114, 0.701–0.680]. Plant defenses in Indigofereae are enhanced by the presence of food pearl bodies (O’Dowd, 1982), which is a synapomorphy of Disynstemon + tribe Indigofereae (Fig. 1). These are symbiotically associated with ant protection against herbivores (Janzen, 1981; McKey, 1989; Schrire, 1995; Dutra et al., 2006). Warty outgrowths (state 3) cover the two Australian species I. verruculosa and I. ixocarpa (Fig. 4), and state 4, in the outgroup Schefflerodendron (Fig. 2), appears similar to the sessile glands in tribe Phaseoleae subtribe Canjaninae (Lackey, 1981). 9. Hydathode extrafloral nectaries on tips of stipules, leaflets, or calyx lobes. 0 = absent, 1 = sessile, tuberculate, 2 = stalked [8–9, 0.250–0.222, 0.727–0.682]. This character is not to be confused with multicellular, gland-tipped trichomes, character 6. Passive hydathodes appear to be prevalent at vein terminations in leaf apices of most species in the tribe and on the tooth apices of the dentate leaflet margins of Cyamopsis (Wilkinson, 1979; Schrire, 1995). Within the two major palaeotropical subclades, I. kirkii–I. hermannioides and I. compressa–I. tetrasperma (Fig. 3), however, two types of specialized actively secreting hydathode extrafloral nectaries have developed. These include (1) stalked capitate to crateriform “glands” occurring most predominantly on juvenile growth, on the tips of leaflets, stipules, and calyx lobes in the I. kirkii–I. hermannioides subclade (Schrire, 1991, 1995), and (2) sessile, inflated calyx lobe tips, and sometimes similar “glands” on the leaflet apices, which actively secrete a clear liquid particularly in juvenile growth in the I. compressa–I. tetrasperma subclade (Schrire, 1991, 1995). In those species with active leaflet apical hydathodes, the exudate often spreads along the leaflet margins leaving them glistening and sticky. Variations on the theme are blister-like glands on the calyx lobes and leaflet apices and, in the case of I. brevicalyx (Fig. 3), also on the lateral leaflet margins. In some cases, species appear to have entirely glandular leaflet margins (see character 21). Wilkinson (1979) notes the similarity of hydathodes to extrafloral nectaries, with the main difference being the nature of the vascular supply (phloem in nectaries and xylem in hydathodes). Hydathodes may be actively secreting various ions, and in some cases sugars, out of the leaf or circulating nutrients to immature, expanding shoots (Wilkinson, 1979; Mauseth, 1988). An association with ants probably developed secondarily with actively 848 American Journal of Botany secreting extrafloral nectaries, making this feature an additional protective ant– plant relationship to pearl bodies. Extrafloral nectaries, whether stalked or sessile, are correlated with the presence of dense pearl bodies on the plant surfaces, so both systems appear to reinforce each other. 10. Vegetative brachyblasts. 0 = absent, 1 = present [15, 0.067, 0.533]. This trait refers to the often ± fasciculate leaves on short shoots and represents convergent synapomorphies of Disynstemon (Fig. 2), Phylloxylon (Fig. 2), the Cape I. Olygophyllae sp. nov.2–I. denudata subclade (Fig. 5) and the Tethyan I. leucoclada–V. humbertiana subclade (Fig. 6). 11. Branch and inflorescence rachises ending in spines. 0 = absent, 1 = present [9, 0.111, 0.556]. State 1 is a synapomorphy of Phylloxylon (Fig. 2) and the Cape I. Olygophyllae sp. nov.2–I. denudata subclade (Fig. 5). This state is often associated with arid environments. 12. Cladodes, phyllodes, or phyllodinous leaf rachises. 0 = absent, 1 = present [5–7, 0.200–0.143, 0.733–0.600]. Parsimony character optimizations are equivocal about state 1 being a synapomorphy of Phylloxylon (Fig. 2), while state 1 is a synapomorphy of the Tethyan I. ammoxylum–V. humbertiana subclade (Fig. 6). Nearly all species with such leaves are from the islands of Madagascar and Réunion. 13. Leaves or leaflets. 0 = persistent, 1 = ephemeral or absent [6, 0.167, 0.722]. State 1 represents convergent synapomorphies in Phylloxylon (Fig. 2), the Cape I. filifolia–I. gifbergensis subclade (Fig. 5) and the Tethyan Vaughania subclade (Fig. 6). In some cases, the terminal leaflet persists after the laterals have fallen. 14. Leaves glaucous, glabrescent, and subsucculent, 0 = absent, 1 = present [3, 0.333, 0.818]. All three conditions typically define state 1, being convergent synapomorphies of the Cape subclades I. Olygophyllae sp. nov.2–I. denudata and I. declinata–I. concava (Fig. 5), and I. frutescens-I. langebergensis (Fig. 4). 15. Leaves. 0 = pinnate, 1 = digitately to pinnately trifoliolate, 2 = simple or unifoliolate [42–44, 0.048–0.045, 0.518–0.494]. The simple or unifoliolate leaf condition (state 2) is a synapomorphy for the Madagascan endemic Phylloxylon (Fig. 2), the Microcharis–Rhynchotropis clade (Fig. 2), and the Tethyan subclade Vaughania (Fig. 6). State 1 represents convergent synapomorphies in the Tethyan I. angulosa–I. jamaicensis subclade (Fig. 6), the Cape subclades I. Digitatae sp. nov.–I. heterophylla and I. nigromontana–I. denudata (Fig. 5), and in Microcharis, the M. tritoides–M. sessilis subclade (Fig. 2). The state 2 condition of the Microcharis–Rhynchotropis clade is linked to a neotonous growth form. Many grass biome species of Microcharis (M. galpinii to M. remotiflora) are small, ephemeral herbs in transiently inundated areas, and are reduced almost to flowering seedlings with persistent seedling leaves at the base of the stems. In contrast, many species from the succulent biome are perennials (or shrubby) with trifoliolate leaves. This state and the shrubby or scandent habit are associated in often arid areas. 16. Leaflets. 0 = opposite, 1 = alternate [8–9, 0.125–0.111, 0.767–733]. State 1 represents convergent synapomorphies of the succulent biome Indigastrum argyroides–Indigastrum candidissimum subclade (Fig. 2) and within Indigofera, the Tethyan I. anabibensis–I. trigonelloides and I. schimperi–I. bongensis subclades (Fig. 6). 17. Leaflets sclerophyllous (coriaceous, often darker above than below). 0 = absent, 1 = present [2, 0.500, 0.923]. State 1 is a synapomorphy of the Cape subclade I. cytisoides–I. heterophylla (Fig. 5). Sclerophylly is associated with a major switch from the arid, succulent-rich biome areas surrounding the Cape Floristic Region (Goldblatt and Manning, 2000), to a substantial diversification of temperate biome species that have become locked into the nutrient-poor, acidic soils of the fynbos region (specifically the sandstone and limestone fynbos vegetation types) within the Cape (Linder et al., 1992; Stock et al., 1992). Sclerophyllous leaves in plants with mostly an ericoid (or heath-like) growth form are an adaptation in systems where low nutrients limit the options of drought deciduousness. Long-lived, tough, low-nutrient leaves capable of resisting desiccation and herbivory are advantageous (Rebelo et al., 2006). High levels of tannins in such leaves are one of a number of unpalatable secondary compounds that effectively deter herbivores. Such morphological phylogenetic structure that co-occurs with ecology is known for other mediterranean-type environments (e.g., Ceanothus; Ackerly et al., 2006). 18. Leaflet margins. 0 = entire, 1 = dentate [2, 0.500, 0.500]. The dentate leaflet condition, unusual in Leguminosae, is a synapomorphy for Cyamopsis (Fig. 2). 19. Leaflet margins revolute. 0 = absent, 1 = present [2, 0.500, 0.857]. State 1, in which leaflets with revolute margins probably represent a further development in sclerophylly (see character 17), is a synapomorphy of the Cape subclade I. ovata–I. brachystachya (Fig. 5). 20. Leaflet margins involute. 0 = absent, 1 = present [2, 0.500, 0.500]. Leaflets with involute margins do not co-occur with the sclerophyllous condition. [Vol. 96 Parsimony character optimizations are equivocal about state 1 being a synapomorphy of the Cape I. declinata–I. concava subclade (Fig. 5). In the context of reducing water loss, state 1 represents an alternative to the revolute-margined or sclerophyllous leaf. This state is also usually associated with glaucous, subsucculent leaves (see character 14). 21. Leaflet margins. 0 = not viscid, 1 = viscid [4, 0.250, 0.000]. This character does not group any taxa although it does appear among very closely related taxa, such as I. wituensis and I. tanganyikensis in the palaeotropical clade (Fig. 3). This group of species is distributed between East Africa, India, and Madagascar. 22. Stipule shape. 0 = linear, lanceolate to narrowly triangular, 1 = broadly ovate or triangular, 2 = lanceolate-ovate with central midrib [10–11, 0.200– 0.182, 0.818–0.795]. State 2, which refers to stipules (and bracts) that are supported by a prominent midrib, is a synapomorphy of the Tethyan subclades I. schimperi–I. bongensis and I. auricoma–I. trigonelloides (Fig. 6). These stipules are likely to provide additional protection to developing buds. State 1 represents convergent synapomorphies of the palaeotropical I. kirkii–I. hermannioides (Fig. 3) and the Cape I. cuneifolia–I. alpina subclades (Fig. 5). 23. Leaves becoming vestigial or much reduced in size and leaflet number toward stem base. 0 = absent, 1 = present [3–4, 0.333–0.250, 0.778–0.667]. Stem bases with much reduced trilobed leaves that are scarious and decurrent on the stems occur sporadically in the fire-prone, suffrutescent Tethyan I. asperifolia–I. bongensis subclade (Fig. 6). These often precociously flowering species have vestigial (stipular) leaves on the basal parts of shoots, while true leaves occur higher on vegetative shoots. 24. Stipule fusion. 0 = not fused, 1 = fused into a single triangular, entire or bilobed scale between the stem and petiole base, 2 = fused on leaf-opposed margin, leaving a collar-like scar around stem [2, 1.000, 1.000]. Synapomorphies are state 1 of the Tethyan Vaughania subclade (Fig. 6) and state 2 of the Cape I. cuneifolia–I. alpina subclade (Fig. 5). 25. Stipels. 0 = absent, 1 = paired, 2 = single between leaflets [45–47, 0.044– 0.043, 0.686–0.672]. Stipel loss (state 0) is a synapomorphy for tribe Indigofereae (Fig. 1), and parsimony character optimizations suggest an independent regain (state 1) is a synapomorphy for the palaeotropical clade (Fig. 1). Other regains occur in parallel in the pantropical I. jucunda–I. verruculosa subclade (Fig. 4) and minor subclades within the Cape (Fig. 5) and Tethyan (Fig. 6) clades. State 2 is a synapomorphy of the largely wet habitat Cape I. filifolia–I. gifbergensis subclade. 26. Peduncle length of inflorescences. 0 = shorter than to ± twice as long as subtending leaf or bract, 1 = more than twice as long as subtending leaf or bract [8, 0.125, 0.588]. State 1 is a synapomorphy of the Cape I. sarmentosa–I. heterophylla subclade (Fig. 5). Peduncle length is likely related to a showy floral display. Most fynbos plants rely on insect pollination, and intense competition arises between fynbos species for the attention of pollinators, whose low numbers often result in many flowers not being pollinated (Cowling and Richardson, 1995). Well-exposed inflorescences with an excess of flowers that do not develop into fruits may still contribute to plant fitness by increasing attractiveness of the floral display (Johnson, 1992). Species in the Cape subclade, I. meyeriana–I. heterophylla (Fig. 5), have a habit of clambering through other shrubs to flower above the canopy so many suffrutices may become quite tall and woody. This habit is considered opportunistic for enhanced display and longevity (through support and protection). 27. Inflorescence type. 0 = simple, open racemes, 1 = racemes with 1–4 flowers crowded in axils of leaves or bracts, often secondarily aggregated into heads or congested panicles, 2 = large, loosely branched panicles or pseudoracemes [7, 0.286, 0.783]. Most outgroups are characterized by state 2, with the notable exception of Disynstemon (Fig. 2), the sister to tribe Indigofereae. In the palaeotropical subclades I. congesta–I. tetrasperma and I. inhambanensis–I. strobilifera (Fig. 3), state 1 is the result of a progressive series of neotonous events involving the truncation of leaf development, with normal leaves below becoming gradually reduced and bract-like above, and the truncation of inflorescence growth, with reduced, 1–4-flowered racemes hidden within the axils of leafy bracts. Such a break from the constraints of an archetypal, simple axillary raceme in the rest of Indigofera represents a radical shift in inflorescence morphology in these subclades. In the I. congesta–I. tetrasperma subclade (Fig. 3), for example, this shift has led to secondary aggregations of racemes into clusters, dense capitula, and ultimately to congested panicles, where the primary inflorescence branch comprises one to two flowers with a joint below and sometimes a simple secondary bract at the joint, indicating that the lower part is probably a reduced raceme. This trend is seen as further maximizing seed production in an opportunistic, r-selected life strategy (Vermeij, 1978), with all paniculate species being annual herbs (including those with woody bases). April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 28. Bracts. 0 = caducous, 1 = persistent (at least to juvenile fruiting) [16, 0.063, 0.615]. State 1 represents convergent synapomorphies of the CRIM clade (Fig. 2) and the palaeotropical I. congesta–I. tetrasperma subclade (Fig. 3). Persistent bracts in the palaeotropical subclades I. inhambanensis–I. strobilifera and I. microcalyx–I. tetrasperma are probably related to the neotonous events described in the character 27. 29. Bract shape. 0 = linear, lanceolate to ovate, abruptly distinguished from adjacent leaves subtending inflorescences, 1 = foliar 1–3-fid grading indistinguishably into the adjacent leaves subtending the inflorescence [4, 0.250, 0.667]. State 1 represents convergent synapomorphies of the palaeotropical subclades I. congesta–I. tetrasperma and I. inhambanensis–I. strobilifera (Fig. 3). Congested inflorescences (character 27) occur elsewhere in Indigofera without foliar bracts, suggesting the evolution of the two characters is independent. 30. Bracteoles. 0 = absent, 1 = present [1, 1.000, 1.000]. Loss of bracteoles is a synapomorphy for the CRIM + Indigofera clade (Fig. 1). This trait also acts as a proxy in this group for the development of an extended flowering regime of ± continuous raceme production during the growing season, where each raceme reliably produces a few high rewarding flowers per day and elicits a trapline behavioral response by high-energy-demanding bees (Janzen, 1971; Heinrich, 1983; Richards, 1986). 31. Fruiting pedicels. 0 = c. 2(2.5) mm long or less, 1 = longer than 2.5 mm [25–26, 0.040–0.038, 0.707–0.695]. State 0 is a synapomorphy of the Indigofereae clade (Fig. 1). Reversals to long pedicels mark the Microcharis–Rhynchotropis clade (Fig. 2), the pantropical I. dendroides–I. verruculosa subclade (Fig. 4), and the Cape subclade I. filifolia– I. gifbergensis (Fig. 5). Higher altitude and temperate clades tend to have longer pedicels. 32. Flower size. 0 = small (staminal sheath 1–7 (8) mm long), 1 = large (staminal sheath > 8 mm long) [21–23, 0.048–0.043, 0.630–0.593]. The persistent nature of the staminal sheath allows it to serve as a proxy for flower size, which has a bimodal distribution. State 0 is a synapomorphy of the Disynstemon + Indigofereae clade (Fig. 1), with independent regains principally characterizing the pantropical subclade I. socotrana–I. verruculosa (Fig. 4). 33. Petal colors at anthesis (not including nectar guide). 0 = weakly bicolored: pink, magenta, and mauve to white; 1 = ± uniformly colored: carmine red to orange; 2 = strongly bicolored: keel greenish-white and other petals red to orange [22–23, 0.091–0.087, 0.861–0.854]. The plesiomorphic flower color in Indigofereae is state 0. Parsimony character optimizations are equivocal about state 1 being a synapomorphy of the palaeotropical clade; otherwise, state 1 is a synapomorphy for, among others, the Tethyan subclades I. nephrocarpa–I. bongensis and I. aspera–I. arabica (Fig. 6) and the Microcharis–Rhynchotropis clade (Fig. 2). State 2 represents convergent synapomorphies of the pantropical I. caloneura–I. verruculosa subclade (Fig. 4) and the Tethyan I. hiranensis–I. bongensis subclade (Fig. 6). 34. Dried foliage. 0 = not dark colored, 1 = dark grey-green to bluish black [4, 0.250, 0.800]. State 1 is a synapomorphy of the pantropical I. arrecta–I. platycarpa subclade (Fig. 4). This character is probably based on the higher than average accumulation of the glycoside indican being converted to indigo upon drying. Species with this trait include all the commercial species from which indigo is extracted. 35. Calyx lobe length. 0 = shorter than the tube, 1 = ± equal to twice the length of the tube, 2 = lobes longer than twice the length of the tube [48–50, 0.042–0.040, 0.720–0.707]. The intermediate calyx lobe length (state 1) is a synapomorphy for the CRIM + Indigofera clade (Fig. 1) and state 0 the Cape clade (Fig. 2) and the pantropical I. socotrana–I. verruculosa subclade (Fig. 4). Parsimony character optimizations are equivocal about state 0 being a synapomorphy of the Tethyan I. ammoxylum–V. humbertiana subclade (Fig. 6), although given the context of character state evolution, this alternative is more likely. State 2 is a synapomorphy for subclades within all four Indigofera clades. A general tendency is for species with short calyx lobes (state 0) to be woody and produce relatively few long elastically dehiscent pods per inflorescence and for species with long calyx lobes (state 2) to be annual and produce many short pods per inflorescence. Thus a correlation is evident between state 2, where calyx lobes are longer than twice the tube in herbs and suffrutices, and states 1 or 0, which occur frequently in shrubby subclades. It is possible that long calyx lobes are adaptive for protection of developing fewer-seeded ovaries. An allometric linkage between increasing pod size and decreasing calyx size (Smith, 1980) may result in shorter calyx lobes associated with manyseeded pods. 36. Calyx indumentum. 0 = hairy, 1 = glabrous (not including ciliate margins) [2, 0.500, 0.500]. State 2 is a synapomorphy of the largely wet habitat restricted Cape I. filifolia–I. gifbergensis subclade (Fig. 5). 37. Calyx vexillary lobes. 0 = broadly separated from each other by a Ushaped sinus, 1 = partially fused and only separated distally by a narrow V- 849 shaped sinus [1, 1.000, 1.000]. A uniquely derived character state marking the monophyly of the CRIM + Indigofera clade (Fig. 1) is the U-shaped sinus broadly separating the two vexillary calyx lobes and pushing them into more lateral positions. The state of broadly separate vexillary lobes serves as a proxy for a level landing platform provided by the wing petals because of the mechanical support provided by the laterally positioned vexillary calyx lobes. 38. Petals. 0 = caducous, 1 = persistent (marcescent) [11–12, 0.091–0.083, 0.750–0.725]. Petals that are immediately caducous after flower tripping (state 0) serve as a proxy for the explosive tripping mechanism, which marks the monophyly of the Disynstemon + Indigofereae clade (Fig. 1). In such flowers, a tension is developed between the reproductive column tending to curve upward and the enclosing petals tending to curve downward (Polhill, 1976). Depression of the interconnected wings and keel petals by a pollinating bee splits the upper keel suture, and the reproductive column, with a mass of released pollen, shoots upward, making contact with the bee. Explosive tripping mechanisms occur in just a few other groups of legumes, e.g., in Harpalyce in the Brongniartieae (Arroyo, 1976) and in Spartium, Ulex, and some Cytisus and Genista species in the derived Genisteae (Polhill, 1976; Arroyo, 1981; Bisby, 1981; López et al., 1999). The recent transfer of Brya from the Desmodieae to Dalbergieae (Lavin et al., 2001) marks an independent evolution of this mechanism in the Dalbergieae. Within the Old World clade of Lewis et al. (2005), explosive tripping has evolved at least five times; in the Baphioid group of Pennington et al. (2005), in the Millettioid s.l. clade in the genus Apios (Reynolds et al., 1990; Bruneau and Anderson, 1988, in most of tribe Desmodieae except the Lespedeza group (Ohashi et al., 1981; Schrire, 1984), and all of tribe Indigofereae (Polhill, 1981; Schrire, 1995; Schrire et al., 2003). Another occurrence is in tribe Trifolieae, in Medicago s.l. including the medicagoid species of Trigonella (Small and Brookes, 1984; Small et al., 1987). The rapidly caducous corolla occurs in only two of these most species rich taxa, i.e., the tribes Desmodieae and Indigofereae. Early corolla loss may enhance speciation rates by somehow diminishing gene flow (e.g., by limiting generalist pollinators), thus enhancing the initial stage of divergence via mutation and drift. Explosive pollen release is often associated with species-rich taxa (M. Lavin, unpublished data). These include Medicago (83 species), Desmodieae (c. 460 species), Genisteae (c. 100 species) and Harpalyce (24 species). Indigofereae with c. 750 species is the most species rich of this group. In Indigofera, a significant shift occurs in the pattern of flowering in the Cape I. Brachypodae sp. nov. 9–I. brachystachya subclade (Fig. 5), with a reversal to short peduncles and the development of ± rigid, persistent petals that remain attractive after explosive flower tripping. Inflorescences thus appear to compete for display through mass flowering rather than having the typical few flowers open per raceme per day. This syndrome is associated with the production of strong floral scents and a shift to a mid or late summer to winter flowering phenology, compared to a winter to summer period of flowering for most other Cape species of Indigofera. In the pantropical I. himalayensis–I. atropurpurea subclade (Fig. 4), and with apomorphic tendencies elsewhere in the pantropical clade, the standard petal is retained and folds over the developing ovary, possibly providing additional protection until the pod is well developed. Marcescent petals mark clades elsewhere in Leguminosae, e.g., Trifolium (Heyn, 1981) and various genera in Dalbergieae (Lavin et al., 2001). 39. Petal pearl bodies. 0 = absent, 1 = present [3, 0.333, 0.333]. The unusual occurrence of pearl bodies on the petals marks the monophyly of the tropical Asian I. glandulosa–I. squalida subclade nested within the Tethyan clade (Fig. 6). 40. Petal symmetry. 0 = bilateral, 1 = spirally twisted [2, 0.500, 0.833]. State 1 is a synapomorphy of the Tethyan Vaughania depauperata to V. cereghellii subclade (i.e. excluding the earliest branching V. pseudocompressa, Fig. 6). This morphology refers to the spirally twisted keel curving upward in front of the standard and one wing being generally pointed forward and the other aligned with the keel and becoming suberect such that the two together cannot serve as a level landing platform. 41. Wing petals twisted outward, similar in appearance to standard, and rendering a radially symmetric 3-petaled arrangement. 0 = absent, 1 = present [2, 0.500, 0.947]. State 1 is a synapomorphy of the clade containing Indigastrum, Rhynchotropis, and Microcharis (Fig. 2). The keel has lateral pockets, not spurs, and is abruptly constricted toward the apex into a long rostrum, which is crinkled, dark-colored, and appears staminode-like in the centers of the flowers. Cyamopsis flowers are plesiomorphic and similar to those of Indigofera in having wings that form a level platform over the keel. One Cyamopsis species has spur appendages, and the upper margins of the keel of a few species have isolated hairs. The rest of the CRIM clade has entirely glabrous petals. 42. Standard dorsal surface. 0 = glabrous, 1 = hairy with hyaline hairs only, 2 = hairs mixed with hyaline, orange, brown, or black hairs [31–33, 0.066– 850 American Journal of Botany 0.061, 0.724–0.705]. State 0 represents convergent synapomorphies of the CRIM clade (Fig. 2) and the Cape I. filicaulis–I. heterophylla subclade (Fig. 5). The presence of dark hairs on the dorsal surface of the standard (state 2), however, marks the higher altitude palaeotropical I. quarrei–I. nyassica (Fig. 3) and pantropical I. dendroides–I. tristis subclades (Fig. 4). One possible hypothesis for this dark coloring is that it may be due to anthocyanin accumulation, often associated with higher altitudes (Lauter et al., 2004). These authors have proposed that anthocyanin accumulation only occurs in high altitude forms of teosinte (Zea mays) and may improve metabolism and translocation by helping plants absorb and retain radiant energy. Fixation of this character in the largely lowland pantropical I. roseocaerulea–I. verruculosa subclade (Fig. 4), however, indicates a more complex evolution of this character. 43. Standard indumentum. 0 = coarsely to finely ± appressed strigose, 1 = spreading pubescent, hirsute, or tomentose [22, 0.045, 0.691]. This character was scored as inapplicable in the CRIM clade because of its glabrous petals. Character 42 refers to the incidence and color of standard hairs, while this character refers to hair type. State 1 is a synapomorphy of the pantropical clade (Fig. 1) and arises in parallel many times throughout the Cape clade (Fig. 5). Shrubby higher altitude or more temperate species tend to have the state 1 condition. 44. Distal wing beard. 0 = absent, 1 = present [23–25, 0.043–0.040, 0.690– 0.662]. State 1, a beard on the distal upper wing surfaces, represents convergent synapomorphies in the Cape I. Olygophyllae sp. nov.2–I. heterophylla (Fig. 5) and pantropical I. socotrana–I. verruculosa subclades (Fig. 4). Both clades are associated with higher altitude or more temperate habitat diversifications. 45. Proximal crest of wing petals. 0 = glabrous, 1 = hairy [11, 0.091, 0.796]. State 1 (a rugose boss with pink hairs on the proximal upper surface of the wings) is a synapomorphy of the pantropical clade (Fig. 1). The proximal crests are actually part of the nectar guide located at the center base of the keel (Schrire 1995), and they often have minute folds, which like petal sculpturing in other tribes (Stirton, 1981) appear to serve as pollinator footholds. The tripping mechanism is activated by the pollinator forcing apart the wing bases in search of the nectary. 46. Fringe on upper keel margin. 0 = absent, 1 = present, but of inconspicuous whitish hairs, 2 = dense, prominent and often with pink to red hairs [28–29, 0.071–0.069, 0.845–0.839]. States 1 and 2, described in Schrire (1995), are distinct Indigofera character traits that could well be related to a level landing platform provided by the wing petals and the expansion of the nectar guide provided by the proximal wing crests. State 1, a sparse fringe of hairs along the upper margin of the keel, visible between the wings and raised above the wing platform in the untripped flower, is a synapomorphy of Indigofera (Fig. 1). State 2, where the fringe is dense, ornamental, either white or pink to red, and erect or forming a flattened circle around the entrance to the nectary, is a convergent synapomorphy of the pantropical clade (Fig. 1) and Cape I. ovata–I. heterophylla subclade (Fig. 5). The keel fringe appears to act as an orientation cue and possible tactile guide for pollinators (Schrire, 1991) but is also likely to be an interlocking mesh of adhesion tissue along the upper keel suture helping to maintain the downward petal tension of the explosive tripping mechanism (Small et al., 1987). 47. Keel appendages. 0 = absent, 1 = lateral folds (broad and shallow invaginations), 2 = lateral spurs (deep and narrow invaginations) [4, 0.500, 0.950]. The Disynstemon + Indigofereae clade (Fig. 1) synapomorphies of simple racemes and smaller flowers with explosive pollen release are further canalized in tribe Indigofereae (Fig. 1) by the development of keel spurs (state 2). A distinctive feature (although not exclusive to explosive tripping) is that the wings and keel have an interlocking boss and socket at the base of the laminas, with the keel spur housed in a corresponding wing petal invagination. Unique to explosive tripping, however, is that the wings and keel do not return to their original position after flowers have been tripped (Small et al., 1987), Such flowers are therefore usually visited only once compared to the plesiomorphic passive (or valvular) and pump or brush tripping mechanisms that allow repeated pollinator visits. Both keel spur and corresponding wing petal surfaces possess adhesion tissue so that the wings and keel interlock, thus strengthening and forming a wing “platform” for pollinator orientation. Basally from the wing invagination, the limb dilates to an auricle, and together they both block the throat of the corolla, preventing the bee’s tongue from reaching the nectary. An appropriately sized pollinator spreads the wing auricles, which triggers the explosive mechanism (Small et al., 1987); the auricles act as levers, transmitting the spreading motion through to the keel petals via the keel spur–wing invagination connection, thus pulling apart the adhesion tissue that holds the keel petals together (e.g. character 46) and rapidly releasing the reproductive column. The pronounced state 2 condition of Indigofereae transformed into the state 1 condition, which is a synapomorphy for the clade of Indigastrum, Rhynchotropis, and Microcharis (Fig, 2). [Vol. 96 48. Keel abruptly constricted toward the apex into a long rostrum. 0 = absent, 1 = present [1, 1.000, 1.000]. State 1 is a synapomorphy of the Indigastrum, Rhynchotropis, and Microcharis clade (Fig. 2), in which the distinctly rostrate keel superficially resembles a staminode (Schrire, 1995). This character might be considered a redundant description of the floral morphology encapsulated by character 41, but because the overall floral morphologies found in Indigastrum, Rhynchotropis, and Microcharis are so extremely modified compared to the rest of the Indigofereae, we treat characters 41 and 48 as representing two divergent aspects of this morphology. 49. Inside surfaces of keel tips with a dark coloration in the region of the anthers. 0 = absent, 1 = present [4, 0.250, 0.929]. State 1 is a synapomorphy of the Cape I. ovata–I. heterophylla subclade (Fig. 5). Dark keel tips are possibly linked either to an accumulation of flavonoids toxic to phytophagous insects (Schrire, 1991) or to promotion of pollen maturation and desiccation through enhanced heat absorption before flower tripping in cooler temperate conditions. This dark coloration could thus aid in drying pollen released by the anthers ± 12 h before its explosive release the following morning (Schrire, 1995). 50. Keel tip. 0 = obtuse or blunt, 1 = gradually tapering to an acute apex [22–23, 0.045–0.043, 0.696–0.681]. Keels with an acute apex are a synapomorphy of the Cape clade (Fig. 1). 51. Keel beard. 0 = absent, 1 = thinly present along distal lower margin, 2 = broadly present over distal areas of lateral surfaces [47–48, 0.043–0.042, 0.767–0.762]. A thin beard of hairs (state 1) along the distal margin of the keel is a synapomorphy for Indigofera (Fig. 1). The dorsal standard indumentum and the keel beard together ensure that the bud is entirely covered and protected by hairs. A broad keel beard (state 2) is associated largely with higher altitude or temperate clades and is a synapomorphy for the Cape clade (Fig. 1), the pantropical I. dendroides–I. verruculosa subclade (Fig. 4) and two small palaeotropical subclades (Fig. 3). 52. Scales at base of anthers. 0 = absent, 1 = present [1, 1.000, 1.000]. State 1 is a synapomorphy of the CRIM clade (Fig. 2). Cyamopsis, Indigastrum, and Microcharis have small scales below the anthers, and the dense hairs beneath the anthers of Rhynchotropis are treated as homologous with scales (Schrire, 1995). 53. Anther apiculum. 0 = absent, 1 = present [3, 0.333, 0.909]. State 1 is a synapomorphy of the CRIM + Indigofera clade (Fig. 1). The specificity of pollen placement through the tripping mechanism is enhanced by the presence of extended apical connectives on the anthers. These appear to effectively separate the drying pollen mass (arising from anthers dehiscing some 12 h before anthesis) from the stigma (Schrire, 1991). For the tripping mechanism to be most effective, a relatively dry “pollen cloud” must be directed on to the pollinator. 54. Anther hairs. 0 = absent, 1 = present on at least some anthers [26–28, 0.038–0.036, 0.838–0.825]. Hairs scattered above and/or below the anthers is a synapomorphy of the node combining the pantropical, Cape, and Tethyan clades (Fig. 1); however, the node is not consistently resolved among the set of most parsimonious trees. A close correlation occurs in Indigofera between the presence of anther hairs in plesiomorphically pink flowers and the loss of anther hairs in the apomorphic states of red-carmine and greenish-orange flowers. The red-flowered palaeotropical clade lacks anther hairs (state 0) except for isolated species from southern Africa, and the loss of anther hairs (state 0) in the pantropical I. caloneura–I. verruculosa (Fig. 4) and Tethyan I. hiranensis–I. bongensis and I. aspera–I. arabica subclades (Fig. 6) is associated with these two apomorphic color states. 55. Ends of staminal filaments. 0 = free ≤1 mm, 1 = free c. 1.5 mm or more [4, 0.250, 0.893]. State 0 is a synapomorphy of the Indigofereae (Fig. 1), where greater precision of pollen delivery through the tripping mechanism is achieved by a robust arrangement of alternately longer and shorter stamens being fused nearly their entire length into a sheath. Reversals to state 1 occur in the highly modified flowers of the Vaughania (Fig, 6) and Rhynchotropis (Fig. 2) subclades. The species, formerly placed in Vaughania (Tethyan clade), are distinguished by a remarkable shift in flower morphology and pollination mechanism (see character 40). A reversal occurs to long filaments where the free ends exceed 1.5 mm, and the sheath is fused for only about two-thirds the length of the stamens. Flowers of Rhynchotropis have an L-shaped keel, with the stamens sharply bent upward and free distally for 2–3 mm. The anthers have dense tufts of hairs above and below, as in the Vaughania subclade. 56. Shape of the staminal sheath. 0 = straight, 1 = curved [4, 0.250, 0.625]. A curved staminal sheath (state 1) generally co-occurs with character 40 and refers to the unusual floral morphology found in most species of the Vaughania subclade (Fig. 6). 57. Staminal sheath length. 0 = 2/3 or more the length of the keel, 1 = <2/3 as long [3, 0.333, 0.600]. In contrast to most legumes, a relatively short staminal sheath (compared to the keel petals) is distinctive of Disynstemon and Phylloxylon (Fig. 1). April 2009] Schrire et al.—Phylogeny of the tribe Indigofereae 58. Ovary to style length. 0 = style a third or more the length of the gynoecium (i.e., a short ovary–long style syndrome), 1 = style less than a third the length of the gynoecium (i.e., a long ovary–short style syndrome) [20–21, 0.050–0.048, 0.848–0.840]. State 1 represents convergent synapomorphies of the Cape clade (Fig. 1), the pantropical I. socotrana–I. verruculosa subclade (Fig. 4) and two minor Tethyan subclades (Fig. 6). Schrire (1991) discussed the switch from a short ovary/long style syndrome, which is largely associated with calyx lobes being longer than the tube, to a long ovary/short style state, with calyx lobes generally equaling or shorter than the tube. The change to this latter apomorphic state was considered by Schrire (1991) as possibly being involved in a trade-off between producing better (i.e., more diversely) protected, fewerseeded pods, in r-selected (Vermeij, 1978) annual or perennial herbs vs. improved longer-term dispersal in the many-seeded pods of predominantly k-selected woody species. Pods arising from a short ovary–long style state may be protected by a combination of longer calyx lobes, pearl bodies, multicellular gland-tipped hairs, endocarp tannins, secondary chemical compounds, or a dense indumentum. Longer pods are usually protected by dense endocarp tannins, perhaps a dense indumentum in juvenile pods, and by chemical defenses. Dehiscence by longer valves that can build up greater tangential stresses, and may be better regulated by apical beaks (often visible as an entasis at the base of the style in the developing ovary), are likely to disperse seeds a greater distance than fewer-seeded pods. The trade-off between calyx lobe length and ovary to style length is independent of character 35, which is focused on calyx lobe length in relation to pod size and growth habit. 59. Ovary. 0 = hairy, 1 = glabrous [26–28, 0.038–0.036, 0.609–0.578]. State 2 represents convergent synapomorphies of Phylloxylon (Fig. 2), the pantropical I. jucunda–I. verruculosa subclade (Fig. 4), and the Cape I. sarmentosa–I. heterophylla subclade (Fig. 5). The dark-colored glabrous pods in Indigofera co-occur with cooler temperate or higher altitude environments and may play a role in enhancing seed maturation by promoting absorption of solar radiation. 60. Style laterally compressed. 0 = absent, 1 = present [1, 1.000, 1.000]. State 1 is a synapomorphy of the clade containing Rhynchotropis and Microcharis (Fig. 2). Microcharis is distinguished by short and thickened styles, while Rhynchotropis flowers have an L-shaped keel, with the gynoecium sharply bent upward, and the style is long, curved, variously flattened, twisted, and constricted. 61. Stigma. 0 = capitate, 1 = discoid, oblique or crateriform [2, 0.500, 0.969]. State 1 represents convergent synapomorphies for the CRIM clade and the Tethyan Vaughania subclade. Stigma morphology is discussed and illustrated in Schrire (1995). 62. Pods in transverse section. 0 = terete to oval, 1 = tetragonous (trigonous), 2 = laterally compressed or flattened (often with raised seed eminences) [13–14, 0.154–0.143, 0.911–0.903]. Oval or terete pods (state 0) is a synapomorphy of the Disynstemon + Indigofereae clade (Fig. 1) with the outgroups having state 2. Tetragonous pods (state 1) are convergent synapomorphies of the Tethyan clade (Fig. 1) and the palaeotropical I. Brevipatentes sp. nov.–I. bainesii subclade (Fig. 3). Species with trigonous pods are uncommon and embedded in clades distinguished by tetragonous pods. Independent regains of state 2 mark the CRIM clade (Fig. 2) and the Tethyan I. fanshawei–I. arabica subclade (Fig. 6). Parsimony character optimizations are equivoval about state 2 being a synapomorphy of the palaeotropical I. congesta–I. tetrasperma subclade (Fig. 3). 63. Pod valves. 0 = not fusiform, dehiscent, 1 = fusiform, very tardily dehiscent [1, 1.000, 1.000]. State 1 is a synapomorphy of Phylloxylon, which refers to the woody, turgid, tardily dehiscent pods tapering at the base and to a beaked apex, containing 1–2 darkly colored seeds, 5–15 mm in diameter. 64. Pod surface embellished either with lamellae or with sutures thickened, winged, flanged or spiny. 0 = absent, 1 = present [3, 0.333, 0.778]. State 1 is a synapomorphy of the Tethyan I. microcarpa–I. squalida subclade (Fig. 6), indicating a shift to pods playing a role in seed dispersal. Pod surface embellishments appear to be correlated with a reduction to short, 1–4-seeded, often tardily dehiscent pods, and an adaptation to epizoochorous dispersal is hypothesized for the single-seeded, indehiscent pods with spiny processes along the sutures in I. nummulariifolia. 65. Pearl bodies on pod valves. 0 = absent, 1 = present [12–13, 0.083–0.077, 0.744–0.721]. State 1 occurs almost entirely within the Tethyan clade (Fig. 6) and is a synapomorphy of the I. microcarpa–V. humbertiana subclade. Parsimony character optimizations are equivocal about state 1 also being a synapomorphy of the Tethyan I. asperifolia–I. bongensis, I. subulata–I. jamaicensis, and I. anabibensis–I. trigonelloides subclades. This character is independent of characters 7 and 8, which pertain to pearl bodies located elsewhere on the plant. 66. Pod position relative to pedicel. 0 = in line with pedicel, 1 = held at c. 45°–90° to the pedicel [1, 1.000, 1.000]. State 1 is a synapomorphy of the Microcharis–Rhynchotropis clade. 851 67. Pods reflexed. 0 = absent, 1 = present [12–14, 0.083–0.071, 0.902– 0.884]. State 1 represents convergent synapomorphies of the Tethyan clade (Fig. 1), the Cape I. Digitatae sp. nov.–I. heterophylla subclade (Fig. 5) and the pantropical I. sedgewickiana–I. verruculosa subclade (Fig. 4). 68. Pods erect. 0 = absent, 1 = present [12–14, 0.083–0.071, 0.703–0.649]. Erect pods diagnose the palaeotropical clade (Fig. 1) and Cyamopsis (Fig. 2). This state could be considered an additional state to character 67 without any changes to the results presented. 69. Pod seed number. 0 = 1–4(5), 1 = (5)6–many-seeded [39, 0.026, 0.525]. State 1 is a synapomorphy of the CRIM + Indigofera clade (Fig. 1), with reversals to state 0 scattered throughout Indigofera. Species in the palaeotropical I. inhambanensis–I. strobilifera and I. congesta–I. tetrasperma subclades (Fig. 3) produce large numbers of short pods mostly with only 1–2 seeds, maximizing the production of well-protected seeds in opportunistic environments. A further development in two species within the I. congesta–I. tetrasperma subclade (represented by I. macrocalyx) is the accrescent calyx lobes that entirely surround the thin-walled, single-seeded pods. The role of seed protection and dispersal appears to have switched from pod wall to calyx, which is densely covered by pearl bodies. 70. Endocarp tannins. 0 = absent or diffuse, 1 = distinct spots or stripes [13–14, 0.077–0.071, 0.874–0.863]. The presence of endocarp tannins is a synapomorphy of Indigofera (Fig. 1). They are turgid vesicles when fresh, but dry to dark brown, scattered spots or stripes in the endocarp, particularly in the transverse septa. Both the tannin vesicles and the septa isolating individual seeds provide chemical and physical deterrents to herbivores, particularly to larvae burrowing into the juvenile pods. The loss of endocarp tannins (state 0) is a synapomorphy for, e.g., the Tethyan clade, and is often associated with arid habitat species developing either short, few-seeded pods or flattened pods with pearl bodies. 71. Septa between seeds. 0 = thin, translucent (cellophane-like), 1 = broad, papery (opaque), 2 = absent, 3 = seeds embedded in spongy or papery endocarp [8, 0.375, 0.891]. State 0 is a synapomorphy of the CRIM clade (Fig. 2) and state 1 a synapomorphy of the Indigofera clade (Fig, 1). Various outgroup taxa (excluding Disynstemon) are characterized by state 3. 72. Seed shape. 0 = subcylindrical, quadrate or ellipsoid, 1 = trigonous, 2 = ± spherical, 3 = flattened and either reniform or lenticular [9, 0.333, 0.902]. State 0 is a synapomorphy of the Disynstemon–Indigofereae clade (Fig. 1). Some members of the Tethyan I. microcarpa–I. squalida subclade (Fig. 6) are characterized by state 1, state 2 (see character 76) is a synapomorphy for the Cape clade (Fig.1), and state 3 is limited to the outgroups. 73. Seed surface sculpture. 0 = smooth, 1 = tuberculate [9, 0.111, 0.467]. State 1 represents convergent synapomorphies in Cyamopsis (Fig. 2) and the succulent biome subclade of Indigastrum (Fig. 2). The character has a strong association with arid environments. 74. Aril or funicle of seed. 0 = not persistent, 1 = persistent [8, 0.125, 0.708]. Loss of the seed aril (state 0) marks the monophyly of tribe Indigofereae (Fig. 1). Convergent regains (state 1) occur in the pantropical I. dendroides–I. tristis and I. natalensis–I. frutescens subclades (Fig. 4) and in the largely wet-habitatrestricted Cape I. filifolia–I. gifbergensis subclade (Fig. 5). Aril development is associated with seed dispersal (Van der Pijl, 1982). A feature characteristic of nutrient-poor heathland systems is the common occurrence of ant dispersal of seeds (Bond and Slingsby, 1983; Johnson, 1992; Le Maitre and Midgley, 1992). A typical ant dispersed (myrmecochorous) propagule in the Cape Floristic Region has a firm and durable pale-colored food body (elaiosome) attached to the hard, dark-colored often shiny seed. These characters are often combined with a ballistic (explosive) mechanism of seed dispersal (Bond and Slingsby, 1983). The pantropical, Cape-centered I. frutescens–I. langebergensis subclade (Fig. 4) and Cape I. filifolia–I. gifbergensis subclade (Fig. 5), representing independent incursions into the Cape flora, both have seeds with a pale, exserted hilum and a persistent, dried funicle that contrasts against the uniformly dark testa (character 76). Seeds of I. filifolia are carried by ants, but it is uncertain whether the dried funicle and hilum are providing food, or mimicking an elaiosome (B. Schrire, personal observation). 75. Seed size. 0 = diameter ≤ 4 mm long, 1 = diameter > 4 mm long [2, 0.500, 0.952]. State 0, subcylindrical pods with small seeds, is a synapomorphy of the CRIM + Indigofera clade (Fig. 1). Pods with a few large seeds (state 1) occur throughout the outgroups, Disynstemon and Phylloxylon. 76. Seed color. 0 = pale (cream, khaki, green to orange-brown), 1 = dark (olive to dark brown, red, purple mottled or black) [4, 0.250, 0.943]. State 1 is a synapomorphy of the Cape I. cytisoides–I. heterophylla subclade (Fig. 5). A dark seed color is also associated with the relatively larger, spherical seed character marking the Cape Clade (Fig. 1, character 72 state 2). The hard, shiny or dull, dark or mottled seeds of Indigofera are likely to contrast against the pale sandstone- and limestone-derived soils and perhaps are picked up 852 American Journal of Botany rapidly by ants (Bond and Slingsby, 1983). Johnson (1992) considers that myrmecochory has evolved numerous times in such nutrient-poor environments where seed predation is strong. The mottled nature of many paler seeds in the fire ephemerals, which lack dried funicular remains, are probably camouflaged within the soil seed bank against all predators until the next fire (see character 74). 77. Pollen tectal perforation density (per µm2 averaged over 10 µm2). 0 = 3–9, 1 = 10–45, 2= 50–150 [10, 0.300, 0.941]. State 2 is a synapomorphy of the palaeotropical clade, (Fig. 3) and state 1 a synapomorphy of the Indigastrum, Microcharis, and Rhynchotropis clade (Fig. 2). The trend in Indigofera from a finely perforate to finely microperforate tectum is unique in the Papilionoideae (Ferguson and Strachan, 1982; Schrire and Sims, 1997). This trend is possibly linked to the harmomegathic prerequisite for delivering a discharge of dry pollen to the pollinator by means of the explosive tripping mechanism (Ferguson, 1984). 78. Pollen tectal surface. 0 = finely to coarsely rugulate or reticulate, 1 = smooth [1, 1.000, 1.000]. State 1 is a synapomorphy of the Indigastrum, Microcharis, and Rhynchotropis clade (Fig. 2). 79. Pollen footlayer thickness (average in µm2). 0 = up to 0.2, 1 = 0.4–0.6 [1, 1.000, 1.000]. A thick pollen footlayer is a synapomorphy of the Microcharis and Rhynchotropis clade (Fig. 2). 80. Chromosome number. 0 = 2n = 22, 20, 1 = 2n = 16, 2 = 2n = 14 [3, 0.667, 0.962]. The 2n = 14 chromosome number is a synapomorphy for both Cyamopsis and Indigastrum (Fig. 2). Outgroups of the tribe have 2n = 20 or 22 (chromosome numbers are unknown for Disynstemon and Phylloxylon), while the base number of 2n = 16 is fixed in Indigofera and has also been found in some species of Microcharis (Goldblatt, 1981; Schrire, 1991, 1995). 81. Altitude. 0 = lowland (0–800 m a.s.l.), 1 = upland (800–2000 m), 2 = montane (over 2000 m). The intervals are based on ranges of altitude determined for all representative species in the analysis. 82. Köppen-Geiger climate classification. 0 = equatorial monsoon [Am], 1 = equatorial savanna with dry winter [Aw], 2 = Am and Aw, 3 = arid [B], 4 = Aw and B, 5 = warm temperate [C], 6 = Aw and C, 7 = B and C, 8 = Aw and B and C. This system follows the World Map of Köppen-Geiger climate classification of Kottek et al. (2006). 83. Continent. 0 = Africa-Madagascar, 1 = Africa-Madagascar and Asia, 2 = Asia, 3 = Asia and Australasia-Pacific, 4 = Australasia-Pacific, 5 = Africa-Madagascar and Asia and Australasia-Pacific, 6 = Americas, 7 = Africa-Madagascar and Americas, 8 = pantropical. 84. Biome affinity. 0 = succulent, 1 = grass, 2 = temperate, 3 = succulent and grass, 4 = grass and rainforest, 5 = grass and temperate. Of the four biomes, the succulent biome occupies the smallest area, is the most fragmented, and comprises semi-arid, nonfire-adapted (or intolerant to fire), succulent-rich and grass-poor seasonally dry tropical forests (Pennington et al., 2000), thickets, and bushland (Zonoecotone II/III and Zonobiome III, Breckle, 2002). The temperate biome occupies the largest area with distribution centers in the Mediterranean area and warm and cold temperate regions of the northern and southern hemispheres, as well as montane tropical regions (Zonobiomes IV–VIII, Breckle, 2002). The grass biome occupies the second largest region and comprises fire-adapted (i.e. prone to, or tolerant of fire) succulent-poor and grass-rich seasonally dry tropical forests, woodlands, savannas, and grasslands (Zonobiome II but excluding Zonoecotone II/III, Breckle, 2002). Although scored for most outgroups, the rainforest biome is essentially not inhabited by the tribe Indigofereae and is thus not discussed further. Appendix 3. Indigofereae in the arid corridors of Africa. A phylogenetic and geographical pattern repeated throughout the entire legume phylogeny links often remarkably distant regions of succulent-rich vegetation within a global succulent biome (Schrire et al., 2005a, b). In the Old World, such areas occur in the northeastern Horn of Africa in the Somalia-Masai regional center of endemism (White, 1983); the arid south, west, and north of Madagascar; and southern and southwestern Africa in the Karoo-Namib regional center of endemism (White, 1983) and Albany thicket region (Cowling, 1983; Cowling et al., 2005; Hoare et al., 2006). These southern regions have been linked to the north by various dry forest and thicket “arid corridors” (Van Zinderen Bakker, 1969; Verdcourt, 1969; de Winter, 1971; Thulin, 1994; Jürgens, 1997; Schrire et al., 2003). The Sahara-Sindian region (White and Léonard, 1991) is a floristic continuation of the Somalia-Masai regional center of endemism to Arabia, west Asia and northwestern India. A northeastern–southwestern African arid corridor floristic disjunction pertains to taxa that occur in the arid zones of the northeastern Horn of Africa and have their nearest relatives in southern, particularly southwestern Africa. Verdcourt (1969), Goldblatt (1978) and Werger (1978), among others, postulated the existence of arid corridors emerging during the drought periods of the Pleistocene, which stretched across Africa connecting the older established northeastern and southwestern arid floras. Alternatively, a northeast–southwest continuous arid flora could also have existed during the Tertiary (Quézel, 1978; Kürschner, 1998). Thulin (1994) argued that in general, northeast–southwest disjunctions involving the same species are either Pleistocene relics or results of fairly recent long-distance dispersal, but that more confined northeast–southwest disjunctions such as that of Wellstedia (Thulin and Johansson, 1996) often involve disjunct genera or genera with vicariant species in both regions. This latter pattern could reflect a more ancient (e.g., late Tertiary) link between the two floras, confirming that such arid corridors have existed at various times throughout this period. The relatively stable arid biodiversity hotspots and source areas at either end of these arid corridors, i.e., the northeastern Horn and parts of the southwestern Karoo-Namib regions, are of considerable age (Jürgens, 1997). The northeastern center has had dry climates from c. 50–60 Ma (Axelrod, 1992; Janis, 1993; Maley, 1996; Scotese, 2001), while the southwestern Namib region has been arid from c. 55 Ma with brief semi-arid phases, particularly from c. 20–14 Ma (Van Zinderen Bakker, 1975; Ward and Corbett, 1990; Cowling and Pierce, 2001), with full desert conditions prevailing for the last 14–10 Ma (Ward and Corbett, 1990). The southern part of the Karoo-Namib regional center of White (1983), which abuts the Greater Cape Floristic Region (Born et al, 2007), i.e., that area of southern Namibia and northwestern South Africa on either side of the lower Orange River, has broadly been treated as the Gariep center of endemism (Hilton Taylor, 1994; Van Wyk and Smith, 2001). Hilton Taylor (1994) circumscribed the center as a smaller unit than Van Wyk and Smith (2001), and both concepts are geographically delimited rather than being derived using phytogeographical or floristic criteria. The broader Gariep center defined by Van Wyk and Smith (l.c.) corresponds largely to the Namib phytogeographical center described by Born et al. (2007) with the exception that the Richtersveld region is considered transitional between the latter’s Greater Cape Floristic Region and Namib center, while it is included within the Gariep center of Van Wyk and Smith (l.c.). The analysis of Born et al. (2007) showed the Greater Cape Floristic Region to comprise the Cape Floristic Region (e.g., Goldblatt, 1978; Goldblatt and Manning, 2000) and the Hantam-Tanqua-Roggeveld and Namaqualand regions, considered by, e.g., Mucina et al. (2006) as part of the Succulent Karoo vegetation type largely to the north of the Cape Floristic Region. Craven and Vorster (2006) and P. Craven (University of Stellenbosch, unpublished data) delimited phytogeographical regions within Namibia, one of which is the narrow Huns-Orange group restricted to the Hunsberg and Huib highlands and arid mountains to the east along the Orange River, largely to the north of the Orange River bulge (Hilton Taylor, 1994). The Huns-Orange group is a subcenter within the Namib center of Born et al. (2007). The various group subcenters after P. Craven (University of Stellenbosch, unpublished data) within the Namib center and transitional parts of the Richtersveld region to the south, are increasingly being implicated as important areas for the subsequent evolution of many southern African plant taxa (e.g., Howes et al., 2009). Ward and Corbett (1990) noted a pluvial interlude to desert conditions of their extended Namibian tract (including the HunsOrange group) occurring from c. 20–14 Ma, which is reflected in a change from arid to more mesic semi-arid climates prevailing. This date corresponds closely with the crown clade age estimate of the Cape clade (Appendix S9) indicating that its initial diversification likely occurred in more mesic conditions than today. The present distribution of the two earliest branching species of the Cape clade, Indigofera nudicaulis and I. merxmuelleri is restricted to the narrow Huns-Orange group within the Namib center. The Huns-Orange group and other subcenters to the north of the Greater Cape Floristic Region are likely refugia not only for the oldest extant lineages of Indigofera but also for many other taxa as well.