American Journal of Botany 98(12): 2049–2063. 2011.

P HYLOGENETIC AND POPULATION GENETIC ANALYSES OF DIPLOID L EUCAENA (LEGUMINOSAE; MIMOSOIDEAE) REVEAL CRYPTIC SPECIES DIVERSITY AND PATTERNS OF DIVERGENT ALLOPATRIC SPECIATION 1

Rajanikanth Govindarajulu2,5 , Colin E. Hughes3,4 , and C. Donovan Bailey 2,4

2 Department of Biology, P. O. Box 30001 MSC 3AF, New Mexico State University, Las Cruces, New Mexico 88001 USA; 3 Institute of Systematic Botany, University of Zurich, Zollikerstrasse 107, 8008 Zurich, Switzerland; and 4 Department of Plant Sciences, South Parks Road, University of Oxford, Oxford OX13RB UK

• Premise of the study: Leucaena comprises 17 diploid species, fi ve tetraploid species, and a complex series of hybrids whose evolutionary histories have been infl uenced by human seed translocation, cultivation, and subsequent spontaneous hybridiza- tion. Here we investigated patterns of evolutionary divergence among diploid Leucaena through comprehensively sampled multilocus phylogenetic and population genetic approaches to address species delimitation, interspecifi c relationships, hybrid- ization, and the predominant mode of speciation among diploids. • Methods: Parsimony- and maximum-likelihood-based phylogenetic approaches were applied to 59 accessions sequenced for six SCAR-based nuclear loci, nrDNA ITS, and four cpDNA regions. Population genetic comparisons included 1215 AFLP loci representing 42 populations and 424 individuals. • Results: Phylogenetic results provided a well-resolved hypothesis of divergent species relationships, recovering previously recognized clades of diploids as well as newly resolved relationships. Phylogenetic and population genetic assessments identi- fi ed two cryptic species that are consistent with geography and morphology. • Conclusions: Findings from this study highlight the importance and utility of multilocus data in the recovery of complex evo- lutionary histories. The results are consistent with allopatric divergence representing the predominant mode of speciation among diploid Leucaena. These fi ndings contrast with the potential hybrid origin of several tetraploid species and highlight the importance of human translocation of seed to the origin of these tetraploids. The recognition of one previously unrecognized species (L. cruziana) and the elevation of another taxon ( L. collinsii subsp. zacapana) to specifi c status (L. zacapana) is con- sistent with a growing number of newly diagnosed species from neotropical seasonally dry forests, suggesting these communi- ties harbor greater species diversity than previously recognized.

Key words: allopatric speciation; cryptic species; Leguminosae; Leucaena ; Leucaena cruziana ; Leucaena zacapana; phylog- eny; population genetics.

Patterns of diversifi cation among species can be explained recognized in land plant evolution, in part because it violates by a wide variety of evolutionary mechanisms. Geographic iso- assumptions associated with bifurcating species trees, but more lation leading to divergence among populations is generally importantly because of the evolutionary novelty introduced by considered to be the most common mode of speciation (e.g., such events (e.g., Rieseberg, 1995 ; Rieseberg et al., 2003 ; Grant, 1971), but reticulate evolution and polyploidy following Linder and Rieseberg, 2004 ). As a result, much research fo- hybridization between divergent populations can also prompt cused on recovering the evolutionary history of plant lineages sudden reproductive isolation and speciation in plants (e.g., seeks to distinguish between divergent and reticulate mecha- Rieseberg, 1997; Mav á rez et al., 2006). Hybridization is well nisms and to quantify their relative contributions to the genera- tion of species diversity. At the same time, few studies have investigated speciation in relation to geography, making it dif- 1 Manuscript received 3 June 2011; revision accepted 12 October 2011. fi cult to assess the relative frequency of allopatric vs. sympatric Earlier fi eldwork that laid the foundations and provided much of the speciation or the extent to which speciation is associated with material for this research benefi tted from the support of numerous ecological differences (ecological speciation) (Barraclough et al., colleagues in Mexico and notably, J. L. Contreras, H. Ochoterena, M. 1998 ; Barraclough and Vogler, 2000 ). Sousa, and S. Zá rate, as well as ongoing support from the Instituto de Here, we investigated patterns of diversifi cation among dip- Biologí a of the Universidad Nacional Autonoma de Mé xico. The authors loid members of the mimosoid legume genus Leucaena, which also thank the Oxford Forestry Institute and A. Sing for providing seed of currently comprises 17 diploid species, fi ve tetraploid species, Leucaena , J. Pannell for helpful discussions, and L. Urban for comments and a potentially complex series of putative hybrids ( Hughes, on the manuscript. Components of this project were completed while 1998a) whose evolutionary histories have been infl uenced by C.D.B. was on sabbatical at the Department of Plant Sciences, Oxford. This human translocation of seed, cultivation, and subsequent spon- research was supported by funds from NSF DEB0817033 & EF0542228 (C.D.B.), the Leverhulme Trust (C.E.H.), the Royal Society (C.E.H.), and taneous hybridization (Hughes et al., 2002, 2007 ). All species the United Kingdom Department for International Development (C.E.H.). of Leucaena are small to medium-sized trees growing predomi- 5 Author for correspondence (e-mail:[email protected]) nantly in the seasonally dry tropical forests of Mexico and Cen- tral America and extending north into the dry subtropics of doi:10.3732/ajb.1100259 northern Mexico and Texas, and south into the seasonally dry

American Journal of Botany 98(12): 2049–2063, 2011; http://www.amjbot.org/ © 2011 Botanical Society of America 2049 2050 American Journal of Botany [Vol. 98 forests on both sides of the northern Andes as far south as Peru Schleinitzia novoguineensis were chosen as outgroups based on the results of ( Hughes, 1998a ). Seeds and pods of a subset of Leucaena spe- previous studies (e.g., Hughes et al., 2003). Multiple alleles derived from the cies are widely used as a minor food plant in south-central Mex- same accession are indicated by a numerical suffi x (e.g., 1, 2...). For a few ac- cessions, DNA extractions and silica gel dried materials became depleted dur- ico ( Hughes, 1998b ; Hughes et al., 2007 ), and one species, L. ing the study. These were replaced by DNA from either the same individual tree leucocephala has been extensively introduced throughout the trop- or another individual from the same population depending on availability. ics as a fast-growing agroforestry and forage tree (Brewbaker, 1987; Hughes, 1998b) and is now a pantropically naturalized, PCR, DNA sequencing, and alignment — A total of four cpDNA regions and invasive weed ( Hughes and Jones, 1999 ). seven potentially independent nuclear-encoded loci were sequenced from each Previous phylogenetic studies of Leucaena , primarily apply- accession. Chloroplast regions sequenced included the two trnK introns fl ank- ing data from cpDNA and nrDNA ITS, have failed to fully re- ing the matK gene (primers trnK1L-849R and 1908F-trnK2R, Lavin et al., solve the relationships among diploid species, conclusively 2000 ), the intron between psbA - trnH (primers psbAF and trnHR ; Sang et al., 1997 ), and the rpl32-trnL spacer (primers trnL and -rpl32-F, Shaw et al., 2007 ). identify the parentage of several tetraploid species (Hughes et al., The nuclear-encoded loci included nrDNA ITS and six anonymous SCAR- 2002), or investigate the potential occurrence of homoploid based markers referred to as 23L, 28, A9, A2, PA1213, and A4A5 specifi cally hybridization among diploids. Sequence data from a set of con- developed for Leucaena phylogenetics ( Bailey et al., 2004 ). served low-copy nuclear genes identifi ed specifi cally for le- PCR reactions included 1× PCR buffer (10 mmol/L Tris-HCl, 50 mmol/L gume phylogenetics (Choi et al., 2006), as well as a selection of KCl, 2.5 mmol/L MgCl 2 ), 100 µ mol/L of each dNTP, 0.5 µ mol/L of each for- other low-copy nuclear genes, have proved insuffi ciently vari- ward and reverse primer, 35 mmol/L betaine, 1.5 U of Taq polymerase and 1 µ L of genomic DNA in a 25- µ L reaction. PCR amplifi cations began with a 3 min able within Leucaena to be especially useful (R. Govindarajulu, denaturation at 94° C, followed by 35 cycles of 15 – 30 s denaturation at 94° C, unpublished data; Bailey et al., 2004 ). To overcome these dif- 30– 90 s annealing at 57– 60 °C (see Appendix S1 in Supplemental Data with the fi culties, we developed a set of anonymous nuclear-encoded online version of this article) for primers and annealing temperatures), and loci identifi ed using a sequence-characterized amplifi ed region 60– 90 s extension at 72° C; followed by a fi nal extension for 7 min at 72° C. All (SCAR) technique that has considerable potential for address- of the sequences for chloroplast regions and most of the sequences for SCAR ing evolutionary questions in Leucaena ( Bailey et al., 2004 ). based loci were generated by direct sequencing of PCR amplifi ed products. However, accessions that yielded polymorphic reads from the direct sequenc- These new SCAR markers and an AFLP-based population ing were cloned following previously published methods ( Hughes et al., 2002 ). genetic approach are used here to analyze evolutionary rela- As many as 10 colonies were sequenced for each cloned sample to recover tionships among the diploid species of Leucaena and their pat- discrete variation consistent with the observed polymorphisms in the directly terns of diversifi cation with three main objectives. First, diploid sequenced PCR product. species of Leucaena are the fundamental units from which polyploids were likely derived, suggesting that a “ diploids fi rst ” Sequence assembly and phylogenetics — Individual loci were aligned in the approach (e.g., Brown et al., 2002 ; Beck et al., 2010 ) to resolve program CLUSTAL_X ( Thompson et al., 1997 ) and manually adjusted by eye the evolutionary history of diploids is needed to provide foun- in the program WinClada (Nixon, 2002). For parsimony analyses, indels were scored as gap characters using the simple gap coding method of Simmons and dations for comprehensive downstream assessment of polyploid Ochoterena (2000) implemented in the program SeqState ver. 1.4 ( M ü ller, origins. The relationships and origins of the polyploid species 2006). Parsimony analyses were performed with the program NONA (Goloboff, are investigated in the accompanying paper (Govindarajulu 2000) spawned from WinClada (best tree search, 100 random replications, et al., 2011 in this issue). Second, it has been unclear to what extent holding 10 trees per rep, and applying max* and 1000 strict consensus boot- strap replicates, each comprising 100 mults holding 10 trees each). The best hybrid origins of several polyploid Leucaena ( Hughes et al., Γ 2002 , 2007 ) might extend to the origin and diversifi cation of fi tting maximum likelihood (ML) tree (model GTR+ ) and 500 ML bootstrap analyses (model GTR+CAT) were performed using the program RAxML diploids. To examine these questions, we used densely sampled (Stamatakis et al., 2008). Phylogenetic analyses included investigation of individ- gene tree and species tree phylogenetic approaches to explore ual gene trees and simultaneous analysis of concatenated matrices. First, parsimony levels of hybridization and relationships among diploid popula- analyses were run on each data partition to assess potential gene tree/species tions. Third, this framework also served as a molecular test of tree problems and the utility of each locus to resolve relationships within Leu- the morphologically based species boundaries established by caena . For each gene tree, the positions of multiple accessions from each taxon Hughes (1998a). Resulting diploid species relationships were and individual sequences from heterozygous individuals were characterized as monophyletic, “ unresolved ” (if they were unresolved relative to one another), then used to guide relevant population genetic comparisons to or polyphyletic. Attention was particularly paid to “ polyphyletic ” sequences further investigate hybridization among diploids and to identify from individual accessions because these could indicate potential hybridization genetically distinct and isolated population systems correspond- or gene tree/species tree issues. A series of simultaneous analyses were per- ing to species. formed including and excluding accessions that were polyphyletic in the indi- vidual analyses to evaluate their infl uence on the topology of the inferred species tree. However, these accessions did not have any major impacts on sup- MATERIALS AND METHODS port or resolution among diploid species. For the simultaneous analyses pre- sented here, sequences from heterozygous accessions were fused (using IUPAC DNA extractions —A combination of previously extracted DNA samples ambiguity coding to score polymorphisms) irrespective of whether they were ( Bailey et al., 2004 ; Hughes et al., 2002 , 2007 ) and newly obtained silica-dried monophyletic, unresolved, or polyphyletic in the individual analyses to ensure or fresh leaf materials were used for both the DNA sequencing and AFLP stud- matching of terminals across data partitions prior to concatenation of matrices. ies. DNA recovery applied the chemistry and machinery presented in Alexan- der et al. (2007) , except that 10 mmol/L Tris-HCl in 70% ethanol was used in AFLP studies — Sampling — A total of 424 diploid individuals were subject to place of the 70% ethanol column wash step. DNA quality and quantity were AFLP analysis. Samples for this component included the diploid accessions evaluated by visualizing 3 µ L of each sample alongside a 100-bp DNA mass used in the phylogenetic study (see above) and DNA extracted from at least 10 ladder (NEB-N3231: New England Biolabs, Beverly, Massachusetts, USA) on individuals from each of 42 populations of diploid Leucaena species (Appendix 1% agarose gels. 1). The latter were derived from greenhouse grown seeds collected from each of 10 different trees per wild population ( Hughes, 1998b ). Phylogenetic studies using DNA sequence data — Sampling — Two or more representatives from each of the 17 diploid species and all infraspecifi c taxa AFLP analysis — Five positive control samples were included on each 96- previously recognized for three of these species of Leucaena were sampled, well plate ( L. collinsii subsp. zacapana 57/88/06, L. collinsii subsp. collinsii making 59 ingroup accessions (Appendix 1). Desmanthus fruticosus and 51/88/06, L. lempirana 5/91/05, L. multicapitula 81/87/06, and L. salvadorensis December 2011] Govindarajulu et al. — Cryptic diversity and allopatry 2051

99/90/02). Restriction ligation reactions (RLs) and preselective amplifi cations grid cell were generated from a total of 1652 georeferenced herbarium speci- followed a modifi ed Vos et al. (1995) AFLP approach marketed by Applied men records of diploid Leucaena (examined by Hughes, 1998a) using the pro- Biosystems (“ Plant Mapping Protocol” – P/N 402977 rev. E; Foster City, Cali- gram BRAHMS ver. 6.60 (Filer, 2008) and plotted in the program DIVA ver. fornia, USA). Approximately 50 ng of gDNA was restriction digested using 7.1.7 ( Hijmans, 2010 ). Sympatry was mapped using a variety of grid cell sizes, EcoRI and Mse I (New England Biolabs) and ligated with Eco RI and Mse I but cell size was found to have little impact on the results, and a 5-km grid cell adaptors using T4 DNA Ligase (New England Biolabs) at 37° C for 12 – 16 h. size (25 km2 ) was selected based on estimated pollinator dispersal distances The RLs for each sample consisted of an 11-µ L reaction containing 1 × T4 Li- (see also Hughes et al., 2007 ). gase buffer (NEB), 50 mmol/L NaCl, 0.05 mg/mL BSA, 1 pmol/L Mse I Adapter Pair, 10 pmol/L EcoR I Adapter Pair, 1 U Mse I, 5 U EcoR I, and 67 U of T4 li- gase (NEB). The RLs were subsequently diluted to a fi nal volume of 200 μ L RESULTS with 0.1× Tris-EDTA (TE). Each sample was subject to preselective amplifi ca- tion with a single selective base on each primer ( EcoRI-A and Mse I-C) and three independent selective primer amplifi cations (5′ FAM- EcoR I-AC/Mse I- Phylogenetic analyses of DNA sequences — Extensive PCR, CTA, 5 ′ FAM- EcoR I-AT/ MseI-CTG and 5′ JOE -EcoR I-AA/ Mse I-CTA). Prese- cloning, and sequencing recovered at least one allele for 636 of lective and selective amplifi cations included 1.5 mmol/L MgCl2 , 0.1 mol/L the 649 sequence/accession combinations for the 11 sequenced Tris-HCl pH 8.3, 0.5 mol/L KCl, 0.25 µ mol/L of each primer, and ca. 2 U Taq regions and 59 ingroup samples. In contrast, presumed primer in a 20-µ L reaction containing 4-µ L of dilute RL or preselective amplifi ed prod- site divergence (Bailey et al., 2004) limited the successful PCR uct, respectively. Cycling conditions followed the ABI Plant Mapping Protocol. and sequencing in the outgroups Desmanthus fruticosus and Selective products were capillary electrophoresed on an automated ABI 3100 sequencer (Applied Biosystems) with the Genescan-500 ROX standard Schleinitzia novoguineensis to just the cpDNA/ITS/PA1213 (Applied Biosystems). and cpDNA/ITS loci, respectively. Diploid gene trees lacking appropriate outgroup sequences were rooted with L. cuspidata AFLP data analysis — The program GeneMapper 4.0 (Applied Biosystems) for gene tree comparisons. This rooting derives from L. cuspi- was used to identify and score loci between 75 and 500 bp. The total number of data being the only diploid resolved as sister to other Leucaena alleles amplifi ed in an accession was compared to the mean and standard devia- in a previous nuclear-based phylogeny of the group (e.g., tion of the number of fragments amplifi ed for all individuals for that population Hughes et al., 2002 ). Separate analyses of each nuclear-encoded to identify those that failed to amplify well. In practice, those falling below DNA sequence region and the combined cpDNA matrix were a standard deviation typically failed to amplify more than few alleles per reaction. run to evaluate potential gene tree/species tree confl icts. The The data matrix was subjected to distance-based and model-based analyses length of each alignment, number of accessions lacking se- to identify distinct genetic clusters among populations. First, principal coordi- quence coverage, number of gap characters, percentage parsi- nate analyses (PCO) based on Euclidean distances were generated in the pro- mony informative characters, and the tree statistics for each data gram MVSP ver.3.13m (Kovach Computing Services, Pentraeth, UK). The matrix are presented in Table 1 . Assessments of well-sup ported PCO analyses display the clustering pattern of genotypes indicating genetic nodes confi rmed that all data partitions (Appendix S2A – F, differentiation among populations and have proved very useful in helping to identify hybridization at a variety of taxonomic levels when compared to tree- see online Supplemental Data), except A4A5 and nrDNA ITS based and network-based approaches ( Reeves and Richards, 2007 ). Second, for (online Appendix S2G, H), were generally congruent with each subsequent comparisons of specifi c interest, the number of uniquely supported other and with results from previous studies (e.g., Hughes et al., genetic clusters was estimated using the model based Bayesian statistical analy- 2002 , 2007 ). However, additional sampling for A4A5 and sis of Pritchard et al. (2000). The scoring of AFLP profi les for the program nrDNA ITS revealed deep gene tree/species tree problems structure ver. 2.3.1 treated unobserved alleles as missing data (Evanno et al., manifest by occurrences of highly supported divergent alleles 2005). Structure analyses included 10 000 burn-ins and MCMC replicates for each run, 10 replicate runs for each K value (K = 1 – 10). The admixture model from single accessions, more than two alleles in many acces- and allele frequencies were treated independently. Finally, the number of sions (R. Govindarajulu, unpublished data), and well-supported unique genetic clusters was tested by using Δ K calculation as applied by Evanno incongruence relative to the other loci and one another. These et al. (2005) . two loci were excluded from the simultaneous analysis. Gene trees constructed from separate analyses of chloroplast Geographic distributions and sympatry —To investigate geography and and nuclear DNA sequences (Appendix S2) were generally quantify sympatry in relation to species delimitation and potential reticulate consistent with results from previous studies using two loci evolutionary history, we mapped the geographic distributions of (1) the taxa in (Hughes et al., 2002) and, where support was recovered, these each of three major diploid clades, (2) pairs of well-supported recently derived sister species (sensu Barraclough and Vogler, 2000), and (3) levels of sympatry resolved three major diploid clades within Leucaena . From observed across the native range of Leucaena irrespective of phylogenetic rela- the 295 sequence/accession combinations that are possible for tionship. Coordinate fi les for plotting distributions and number of species per the fi ve nuclear-encoded loci and 59 ingroup accessions, we

Table 1. Summary of data assembled for chloroplast and nuclear gene regions. The concatenated matrix includes all data matrices except A4A5 and ITS.

Gene region Length (bp) Gap char No. of PIC % PIC IC L CI RI cpDNA 2597 104 176 6.5 1 178 0.50 0.76 23L 874 75 160 16.8 0 315 0.60 0.83 28 742 72 108 13.2 0 286 0.44 0.75 A9 1235 156 432 28 3 1159 0.47 0.74 PA1213 810 84 121 13.5 1 326 0.53 0.80 A2 705 34 97 12.9 6 179 0.69 0.90 A4A5 778 77 326 38 NA 825 0.54 0.90 ITS 537 40 328 55 NA 895 0.54 0.79 Concatenated matrix 6960 525 1100 15.8 NA 2355 0.54 0.81 Notes: PIC, parsimony informative characters; IC, number of accessions without sequence coverage; L, tree length; CI, ensemble consistency index; RI, ensemble retention index. 2052 American Journal of Botany [Vol. 98

Fig. 1. Best maximum likelihood (ML) tree recovered from simultaneous analysis of fi ve nuclear loci and cpDNA data analyzed with RAxML. Branch support values represent ML and parsimony derived bootstrap values, respectively. December 2011] Govindarajulu et al. — Cryptic diversity and allopatry 2053

Fig. 2. Summary of results for Leucaena lanceolata s.l. (A) PCO scatter plot for all accessions analyzed using AFLPs. (B) Plot of the geographic distribution of accessions representing divergent groups recovered from phylogenetic analysis (Fig. 1) and PCO (Fig. 2A). (C) Plot of the mean likelihood estimates calculated for K = 1 – 10 in structure ( Pritchard et al., 2000 ). (D) Δ K plot calculated according to Evanno et al. (2005) . identifi ed just four cases were two alleles recovered from the With the exception of multiple accessions representing L. same accession resolved in divergent positions, contradicting lanceolata , ML and parsimony-based simultaneous analyses the monophyly of the species (L. collinsii subsp. zacapana resolved multiple accessions of each diploid or infraspecifi c [PA1213; accession 18/84], L. magnifi ca [PA1213; 19/84], L. taxon into well-supported monophyletic groups (Fig. 1). In lanceolata var. lanceolata [23L; 134/92], and L. salvadorensis contrast, the polyphyletic placement of L. lanceolata accessions [A9; 34/88]). Each case was restricted to a single locus for the from Oaxaca, Veracruz, and western Chiapas (913, 134/92, respective accession, and the inclusion or exclusion of these 43/85, 50/87, 51/87) relative to those from northwestern coastal accessions and sequences had no impact on supported nodes in Mexico (46/85, 44/85, 90/92) contradicts the current division of the combined phylogenetic analysis (R. Govindarajulu, unpub- L. lanceolata into two infraspecifi c varieties proposed by lished data). Hughes (1998a) . Both the ML and parsimony-based simultaneous analyses of the concatenated matrix recovered robustly supported relation- AFLP-based assessments — The potential for interspecifi c ships among lineages and provided stronger support and resolu- hybridization among diploid species exhibiting few genetic tion within clades 2 and 3 ( Fig. 1 ) than individual gene trees crossing barriers ( Sorensson and Brewbaker, 1994 ) and prob- (Appendix S2). The modest support for a few nodes within lems with the current delimitation of L. lanceolata (see above) clade 1 may be attributable to limited character data supporting prompted our investigation of species limits using extensive these nodes or confl icting signal resulting from the potential sampling of individuals and a population genetic approach. The inclusion of hybrid accessions or species. An assessment of un- complete matrix included 1215 loci scored from 363 samples ambiguously optimized character distributions along branches that amplifi ed well for all three selective primer combinations. in the parsimony tree suggests that poorly supported nodes are An extensive series of PCO analyses was carried out using the subtended by short branches relative to other branches in the diploid phylogeny as the framework for relevant focused com- combined phylogeny (Appendix S3), further reducing concerns parisons. These similarity-based comparisons included clade about the inclusion of hybrid terminals. to clade analyses (clade 1 to 2, clade 1 to 3, and clade 2 to 3), 2054 American Journal of Botany [Vol. 98

Fig. 3. Summary of results for Leucaena collinsii s.l. (A) PCO scatter plot for all accessions analyzed using AFLPs. (B) Distribution of accessions representing divergent clades recovered from phylogenetic analysis ( Fig. 1 ) and divergent clusters recovered from PCO ( Fig. 3A ). (C) Plot of the mean likelihood estimates calculated for K = 1 – 10 in structure ( Pritchard et al., 2000 ). (D) Δ K plot calculated according to Evanno et al. (2005) . followed by infraclade comparisons removing the most divergent ered in the phylogenetic analysis (Fig. 1), and the clusters differ taxon(a) sequentially, and ultimately the analysis of accessions by six fi xed alleles in the AFLP data set. Leucaena lanceolata of each species in isolation from other species. These compari- var. lancolata accession 1577 showed evidence of possible hy- sons also recovered distinctive clusters of taxa (online Appen- bridization between these two lineages, as evidence by an ad- dix S4) that were largely consistent with the phylogenetic mixed background in results from structure and its hypothesis. In all 27 comparisons only four of 363 accesssions intermediate position in PCO plots (Fig. 2). However, 1577 be- ( L. esculenta 2143 [Appendix S4E], L. lanceolata 1577 [Ap- haved erratically in each comparison that it was included (Fig. pendix S4A, D, N, U], L. pueblana 125_92 [Appendix S4B, E, 2 and Appendices S4A, D, N, U), even resolving between clades L], L shannonii 1_91_03 [Appendix S4P]) failed to resolve 1 and 3 in broad comparisons. The cause of these anomalous with their respective taxon in one or more result, revealing po- placements is unclear. tential hybrid backgrounds or otherwise unique genotypes. This Separate PCO analyses of multiple populations represent- result along with limited evidence for hybrids in the phyloge- ing L. collinsii subsp. collinsi and L. collinsii subsp. zacap- netic analyses suggests that we recovered few individuals with ana, which formed strongly supported monophyletic sister mixed backgrounds indicative of homoploid hybridization groups in the phylogeny (Fig. 1), resolved these taxa into among diploid species. divergent clusters (Fig. 3A). These were also recovered from For L. lanceolata and L. collinsii , more extensive geographic structure ( ΔK = 2, Fig. 3C, D), and they differed by nine sampling in the AFLP study revealed levels of genetic diversity fi xed alleles. None of these results identifi ed potential ad- potentially consistent with previously overlooked species-level mixed individuals. diversity. AFLP data from eight populations and 39 accessions Genetic diversity consistent with potential overlooked spe- of L. lanceolata were analyzed to investigate the polyphyly of cies was also recovered in PCO analyses of available material L. lanceoata observed in the combined phylogenetic analysis of L. lempirana (Appendix S4Y), L. macrophylla (Appendix (Fig. 1). PCO analyses resolved L. lanceolata accessions into S4W), L. salvadorensis (Appendix S4Z), and L. trichodes (Ap- two distinct groups (44/85, 46/85) and (134/92, 43/85, 51/87, pendix S4X). The single populations of L. trichodes sampled 2166, 2171, 1577) along axis 1 ( Fig. 2A ) and are further sup- from the east and west sides of the Andes in Venezuela and ported by results recovered from structure (Δ K = 2; Fig. 2C, Ecuador, respectively, differed by 36 fi xed allelic differences; D ). These groups correspond to the two divergent clades recov- samples from one population each for the subspecies for L. December 2011] Govindarajulu et al. — Cryptic diversity and allopatry 2055

Fig. 4. Geographic distribution of 1652 herbarium records representing diploid accessions of Leucaena from clades 1 – 3. macrophylla differed by 12 fi xed allelic differences; two popu- volcanic axis across Mexico, taxa from clade 2 restricted to in- lations of L. lempirana from closely adjacent valleys in north- land areas of south-central Mexico in the central Mexican high- ern Honduras differed by seven fi xed allelic differences; and lands and valleys south of the volcanic axis, while taxa belonging populations of L. salvadorensis from northern Nicaragua dif- to clade 1 occur along the western coast of Mexico into south- fered by three fi xed allelic differences from those in southern central Mexico, Central America, and northern South America Honduras. However, all these comparisons are based on sparse (Fig. 4). The geographically widespread clade 1 species L. mac- sampling across the geographic range in both the phylogenetic rophylla generates a slight (Appendix S5), and in some areas and population genetic approaches and may thus be biased. superfi cial (as a result of the scale), overlap in distribution be- tween clade 1 and clade 2 taxa ( Fig. 4 ). Geographic distributions — Mapping of 1652 georeferenced An assessment of sympatry irrespective of phylogenetic rela- wild diploid herbarium specimen records using the three differ- tionship identifi ed very few regions with potential overlap among ent approaches shows a clear pattern of allopatry among diploid noncultivated wild diploid accessions, with just 30 of the 1104 species. First, the geographic distributions of the three major grids (5-km grid cells) containing populations of two diploid spe- clades, with minor exceptions, are almost entirely allopatric cies ( Fig. 5; Appendix S5) and none containing more than two. ( Fig. 4 ) with taxa from clade 3 restricted to northeast Mexico Finally, geographic distributions of each well-supported pair and the southern United States, areas east and north of central of sister species are also generally allopatric. For example,

Fig. 5. Geographic representation of sympatry among all 1652 diploid accessions of Leucaena . Each symbol on the map represents one of 1104 grid cells that had one or more occurrences of Leucaena ; either a single species or two species occurred in that 25-km2 area. 2056 American Journal of Botany [Vol. 98

Fig. 6. Geographic distributions of well-supported sister species pairs recovered from phylogenetic analysis ( Fig. 1 ). (A) Leucaena greggii and L. retusa . (B) L. multicapitula and L. salvadorensis . (C) L. matudae and L. pueblana . (D) L. lanceolata s.s. and L. macrophylla .

L. greggii/L. retusa ( Fig. 6A ), L. multicapitula/L. salvadorensis Species diversity— Inclusion of multiple accessions of all spe- ( Fig. 6B ), L. collinsii subsp. collinsii/L. collinsii subsp. zacap- cies in the phylogenetic analyses and population genetic assess- ana (Fig. 3B), and L. matudate/L. pueblana ( Fig. 6C ) occupy ments, along with detailed distribution maps for all taxa, provide strictly allopatric extant distributions. Only the pair of sister an excellent basis for the re-evaluation of boundaries among species L. lanceolata (s.s.) (see below) and L. macrophylla are diploid species of Leucaena . For 16 of the 17 species recognized found in partial sympatry ( Fig. 6D ). by Hughes (1998a), we found high support for monophyly of mul- tiple accessions in the phylogenetic analyses that is congruent with results from population genetic assessments, morphology, and DISCUSSION geographic isolation among wild populations, supporting the species circumscriptions presented in the monographic treat- Previous investigations of phylogenetic relationships among ment of Hughes (1998a) . These results are consistent with a vari- species of Leucaena and its close relatives (e.g., Luckow, 1997; ety of species concepts, including the phylogenetic species Hughes et al., 2002 , 2003 ) have run into common and recurrent concept (sensu Nixon and Wheeler, 1990 ; Davis and Nixon, 1992 ), problems associated with limited available variation at some monophyletic species concept (e.g., Donoghue, 1985), and bio- universally applied loci (e.g., Shaw et al., 2007) and limited logical species concept ( Dobzhansky, 1935 ; Mayr, 1942 ). concerted evolution with nrDNA ITS (e.g., Á lvarez and These data and analyses also provide strong evidence for pre- Wendel, 2003 ). The application of the SCAR-based approach viously overlooked cryptic species within L. collinsii and L. of Bailey et al. (2004) for locus development, specifi cally lanceolata. The two divergent and well-supported monophyl- implemented for studies involving closely related species, has etic clades of accessions representing L. lanceolata ( Fig. 1 ), proven useful in resolving highly supported divergent diploid which are congruent with population genetic differences ( Fig. 2 ), species relationships in Leucaena . This fi nding is also consis- contradict current and historical species delimitations. How- tent with the application of ( Thorogood et al., 2009 ) and at least ever, these lineages occupy distinct and disjunct geographic one extension of the method ( Gonz á lez, 2010 ) in other plant distributions (Fig. 2), providing further evidence for two dis- groups. Thus this approach, along with conserved orthologous tinct species. Similarly, accesssions of L. collinsii subsp. zaca pana markers (e.g., Fulton et al., 2002; Choi et al., 2006; Lohithaswa are geographically isolated from L. collinsii subsp. collinsii et al., 2007), has helped bridge the gap between sole reliance on ( Fig. 3 ), and the two subspecies formed well-supported mono- standard phylogenetic markers (e.g., nrDNA ITS and certain phyletic groups (Fig. 1) for which fi xed allelic differences were cpDNA markers) and the coming accessibility of massive detected in the population genetic analyses (Fig. 3). These fi nd- marker sets derived from second generation sequencing ap- ings along with morphological and chromosome differences proaches in nonmodel systems (e.g., Gompert et al., 2010; support recognition of these as distinct species (see taxonomic M. M. Koopman [Eastern Michigan University] et al., unpub- treatment later), increasing the number of recognized diploid lished manuscript). species of Leucaena from 17 to 19. December 2011] Govindarajulu et al. — Cryptic diversity and allopatry 2057

In addition to these clear-cut examples of previously under- in plants and animals (e.g., Mayr, 1942 ), entails geographic isola- estimated species diversity, population genetic assessments re- tion between populations and subsequent divergence into distinct veal population structures and fi xed allelic differences indicative lineages. In contrast, sympatric and parapatric speciation is in- of strongly differentiated and geographically structured varia- ferred for species with overlapping or proximate distributions tion among populations of several other species of Leucaena , during speciation, which require the development of reproduc- including L. lempirana , L. macrophylla , L. salvadorensis , and tive isolating mechanisms as part of the speciation process (e.g., L. trichodes. This variation is especially notable within L. trichodes Kondrashov, 1986). Although few studies have tested the rela- where populations from opposite sides of the Andes in Venezu- tionship between geography and speciation (e.g., Barraclough ela and Ecuador differ by 36 fi xed allelic differences, refl ecting and Vogler, 2000; Savolainen et al., 2006; Papadopulosa et al., the likely lack of gene exchange and degree of isolation across 2011 ), where such data are available for non-island systems, cla- the Andes (e.g., Dick et al., 2003 ). However, much denser sam- dogenesis has been found to be predominantly associated with pling in both the phylogenetic and population genetic analyses allopatry, with sympatry interpreted as a consequence of postspe- would be needed to confi dently distinguish whether these pat- ciation range movements ( Perret et al., 2007 ). In contrast, recent terns are the result of sparse sampling or cryptic evolutionary studies that focused on island and island-like systems have begun divergence among populations that could merit recognition of to question predominant views on the relative importance of allo- additional species. patric and sympatric mechanisms of speciation ( Barluenga et al., The discovery of overlooked species diversity prompted by 2006 ; Savolainen et al., 2006 ; Papadopulosa et al., 2011 ). densely sampled (complete or near-complete taxon sampling and Contemporary geographic distributions of diploid species of multiple accessions of species) molecular phylogenetic analyses Leucaena show a high degree of allopatry consistent with allo- that reveal robustly supported reciprocally monophyletic clades patric divergent speciation as the predominant mechanism un- that coincide with other evidence from geography, ecology, and derlying diploid species diversifi cation in Leucaena (e.g., Grant, morphology is increasingly common. Taking examples of le- 1971; Barraclough and Vogler, 2000). Allopatry is evident at gumes from neotropical seasonally dry tropical forests, recent three levels: (1) the early divergence of major clades whose novelties delimited in similar ways include Caesalpinia oyame taxa remain largely allopatric ( Fig. 4 ), (2) overall patterns of (Sotuyo et al., 2007; Sotuyo and Lewis, 2007), Mimosa jaenensis wild populations of diploid species irrespective of their phylo- ( S ä rkinen et al., 2011 ), Coursetia greenmanii (de Stefano et al., genetic relationships ( Fig. 5 ), and (3) four of the fi ve recently 2010 ), Coursetia caatingicola ( de Quieroz and Lavin, 2011 ), and derived well-supported pairs of sister species ( Figs. 3, 6 ). Poissonia eriantha (Pennington et al., 2011). This steady addi- Furthermore, artifi cial crossing experiments in Leucaena tion of new species across different genera suggests that species have shown a low degree of reproductive isolation between diversity of neotropical seasonally dry forests may have been sig- most diploid taxa. The results of 65 diploid-diploid crosses, nifi cantly underestimated. representing 12 of the 19 species analyzed here, show a high degree of crossability (77%) ( Sorensson and Brewbaker, 1994 ). Hybridization and speciation among diploid Leucaena— A t Although crossability remains to be tested among the remain- the diploid level, the evolutionary signifi cance of homoploid ing diploids, currently available data show that crossability is hybridization and introgression in species diversifi cation has retained among distantly divergent lineages across the whole remained controversial (e.g., Anderson and Stebbins, 1954 ). diploid phylogeny. The overwhelming absence of diploid-dip- The role of hybridization in the formation of polyploid species loid hybrids in the context of genus that retains crossability is in and lineages is widely recognized (e.g., Doyle et al., 1990; line with geographical isolation and allopatry underlying dip- Wendel et al., 1995 ); however, there is growing evidence that loid species diversifi cation in the genus. homoploid hybridization and introgression have also been im- portant in many plant and animal groups (e.g., Baack et al., Biogeography — Species of Leucaena are concentrated in the 2005 ; Kane et al., 2009 ). In Leucaena , the important outcomes seasonally dry tropical forest (SDTF) biome, a vegetation type of hybridization in terms of multiple polyploid taxa are clearly that occupies a wide but highly disjunct distribution across the established (Hughes et al., 2002, 2007 ; Govindarajulu et al., neotropics (Pennington et al., 2000, 2009 ) and is characterized 2011 in this issue), and a variety of studies have demonstrated by erratic moisture availability, long periods of seasonal the potential for hybridization among diploids in experiments drought, a general absence of grasses and natural fi re distur- that show high artifi cial crossability between species (e.g., So- bance, high levels of endemism ( β diversity), and an abundance rensson and Brewbaker, 1994) and as a result of human trans- of succulent plants including Cactaceae that has prompted its location and cultivation ( Hughes et al., 2007 ). Nonetheless, our designation as the “ succulent biome” (Schrire et al., 2005). A results derived from individual gene tree hypotheses, combined predeliction for this type of vegetation suggests that Leucaena species tree hypotheses, and an AFLP-based population genetic shows a pattern of phylogenetic niche conservatism (sensu approach revealed little evidence for hybridization or introgres- Donoghue, 2008) to SDTFs, with only minor incursions of a sion between wild-collected diploid individuals or populations, few lineages into mid-elevation seasonal pine – oak forests suggesting that reticulation has been of little importance in the ( L. trichandra and L. macrophylla), more-mesic less-seasonal historical diversifi cation of diploid Leucaena . Furthermore, the lowland forests ( L. multicapitula), and subtropical dry matorral predominantly allopatric geographic distributions (Figs. 4– 6 ) ( L. greggii and L. retusa ). This distribution pattern suggests appear to confi rm that there are few opportunities for diploid hy- that diversifi cation of diploid Leucaena was determined more brids to arise in wild populations. Thus, the available evidence by geographic than ecological isolation, in line with the pre- suggests that diploid Leucaena are predominantly derived from dominance of allopatry among species. divergent, rather than reticulate, mechanisms of speciation. In common with phylogenies of other woody SDTF clades, Such divergent modes of speciation are partitioned into allo- the three major clades within Leucaena occupy distinct and patric or sympatric-parapatric mechanisms. Allopatric specia- largely disjunct geographical areas in northeast Mexico and tion, long considered the most common mechanism of speciation Texas, inland south-central Mexico, and Pacifi c coastal Mexico, 2058 American Journal of Botany [Vol. 98

Central America, and northern South America (Fig. 4). This TAXONOMIC TREATMENT pattern of strong geographical structuring in the diploid Leucaena phylogeny, and across phylogenies for other dry forest groups Leucaena zacapana— The results presented here strongly ( Lavin, 2006 ), has been attributed to limited dispersal and im- support raising L. collinsii subsp. zacapana to species rank, migration across the fragmented SDTF biome ( Lavin, 2006 ; distinct from L. collinsii . Multiple accessions of each taxon Pennington et al., 2009 ), a pattern potentially accentuated by form well-supported monophyletic sister clades ( Fig. 1 ), and the resilient ecology of dry forests and phylogenetic niche con- these groups were also recovered in all the population genetic servatism ( Pennington et al., 2010 ). analyses (e.g., Fig. 3A ). These two population systems are fur- Coalescence of sequences of nuclear loci and the resultant re- ther distinguished by nine fi xed AFLP allelic differences and ciprocal monophyly of multiple accessions of Leucaena species is by cytological studies that suggest L. collinsii subsp. colllinsii also consistent with patterns observed for other seasonally dry is 2n = 52, while L. collinsii subsp. zacapana is 2n = 56 ( Cardoso tropical forest lineages and indicative of long persistence of en- et al., 2000 ; Schifi no-Wittmann et al., 2000 ), and by genome demic populations and allopatric speciation in geographically size measurements (Govindarajulu et al., 2011 in this issue). In isolated and evolutionarily persistent dry forest patches ( Barra- addition, a suite of quantitative morphological differences clough, 2010 ; de Stefano et al., 2010 ; Pennington et al., 2010 ). (Hughes, 1998a) and geographically disjunct and isolated dis- tributions of these two lineages (Fig. 3B), further support rec- Conclusions — We fi nd little evidence for contemporary or ognition as two distinct species. historical hybridization among wild-collected diploids and, as a As recognized here Leucaena collinsii is restricted to the result of limited reticulation and the utility of the markers used, central depression of Chiapas in Mexico and adjacent fringes of recover a well-resolved phylogenetic species-level hypothesis. the Departamento of Huehuetenango in Guatemala, between Population genetic structure and phylogenetic resolution iden- 400 and 900 m a.s.l., while L. zacapana is a narrowly restricted tify two additional morphologically cryptic species that are endemic in the Motagua Valley system in Guatemala between supported by a variety of data. Last, the pattern of diversifi ca- 100 and 800 m above sea level (fi g. 44 in Hughes, 1998a ). The tion across neotropical seasonally dry forests is consistent with intervening mountains of central and northwestern Guatemala a general mechanism of divergent allopatric speciation in the rising to between 2000 and 3000 m a.s.l., effectively isolate formation of most diploid species of Leucaena. these two species. Diploid Leucaena represent the majority of species in a The seasonally dry tropical forests of the Motagua Valley genus known to be complicated by human translocation, poly- system in southeastern Guatemala are known to harbor a num- ploidy, and hybridization (e.g., Harris et al., 1994; Hughes and ber of endemic dry forest plant species (e.g., in legumes Cal- Harris, 1998 ; Hughes et al., 2002 , 2007 ). Through the applica- liandra carcera Standl. & Steyermark, Mimosa canahuensis tion of newly available data, dense sampling strategies, and Standl. & Steyermark, Aeschynomene eriocarpa Standl. & complementary phylogenetic and population genetic ap- Steyermark). However, compared to some other neotropical proaches, we have clarifi ed the evolutionary diversifi cation of seasonally dry valleys such as the central depression of Chiapas diploids. There is little evidence to suggest that reticulate evo- and the Tehuac á n Valley in south-central Mexico, or the Mara ñ ó n lutionary processes have played a signifi cant role in the diversi- Valley in northern Peru, current estimates of endemism for Mot- fi cation of diploid Leucaena . agua are modest. Thus, it seems at fi rst sight somewhat surprising The recognition of L. cruziana and L. zacapana as cryptic that there are two endemic species of Leucaena , L. magnifi ca , a species is in keeping with a renaissance of species discovery narrowly restricted endemic only known from the Guatemalan being driven in part by the development of new tools for DNA Department of Chiquimula (Hughes, 1998a), and L. zacapana barcoding and concerns related to loss of biodiversity (e.g., from this one valley system. These fi ndings suggest that levels Savolainen et al., 2005 ; Smith et al., 2008 ). Furthermore, the of endemism in the Motagua valley may be underestimated. recovery of cryptic species-level diversity in Leucaena is con- Leucaena zacapana (C. E. Hughes) R. Govindarajulu & sistent with densely sampled phylogenies revealing geographi- C. E. Hughes comb. et stat. nov. Leucaena collinsii subsp. cally structured genetic variation with patterns of coalescence zacapana C. E. Hughes, Kew Bull. 46(3): 553, 1991. Type: among conspecifi c accessions for a growing number of neotro- Guatemala. Zacapa: Estanzuela in dry thorn forest, 1 Mar 1988, pical seasonally dry forest plant groups ( Sotuyo and Lewis, Hughes 1102 (holotype: FHO!; isotypes: K! MEXU!). 2007; Pennington et al., 2009, 2010 , 2011 ; de Stefano et al., As circumscribed here, L. zacapana corresponds directly to 2010 ; de Quieroz and Lavin, 2011 ; S ä rkinen et al., 2011 ). L. collinsii subsp. zacapana presented by Hughes (1998a) . Dis- Finally, a general pattern of allopatric divergence among tinguishing features, illustrations, and specimen citations lists diploid Leucaena species is in line with historical opinion previously presented under L. collinsii by Hughes (122 – 128 suggesting that this mechanism is the “ null ” model of spe- and fi g. 43 in Hughes, 1998a ) can be directly applied and are ciation in most sexual lineages (e.g., Coyne and Orr, 2004 ). not further elaborated on here. However, this stands in contrast to a growing body of evi- dence that gene fl ow can be common during and after specia- Leucaena lanceolata s.l. — None of the previous circumscrip- tion in many groups (e.g., Chapman and Burke, 2007; tions of the variable and widely distributed L. lanceolata has Papadopulosa et al., 2011) as well as abundant evidence for proved satisfactory. These treatments range from the recognition allopolyploid origins of several Leucaena species ( Hughes of a single taxon (McVaugh, 1987) to division into either nine et al., 2002, 2007 ). Reconciling this pattern of strictly allopatric separate species ( Britton and Rose, 1928 ) or two infraspecifi c diploid speciation with known allopolyploid speciation in taxa (Z á rate, 1994; Hughes, 1998a). While several distinct mor- Leucaena reinforces the idea that human translocation and phological variants are apparent across the range of L. lancoleata cultivation have been critical in creating artifi cial sympatry s.l., these are often geographically localized and based on rela- and opportunities for hybridization ( Hughes et al., 2007 ; tively minor quantitative differences in leaves and pods, while Govindarajulu et al., 2011 in this issue). overall patterns in these traits show no clear-cut discontinuities December 2011] Govindarajulu et al. — Cryptic diversity and allopatry 2059 that are congruent with geography (fi gs. 58, 59 in Hughes, 1998a). Leu caena brandegeei Britton & Rose, N. Amer. Fl. 23: 128. Notably, populations from the Pacifi c coast of eastern Michoacá n 1928. Type: MEXICO. Baja California Sur: nr La Mesa, and inland populations from Oaxaca show overlapping patterns Cape region, 31 Oct 1902, T.S. Brandegee s.n . (holotype: of pod size and indumentum spanning the boundaries between NY!; isotypes: US!UC!). the two infraspecifi c taxa, L. lanceolata var. lanceolata and L. Leu caena palmeri Britton & Rose, N. Amer. Fl. 23: 123. lanceolata var. sousae recognized by Z á rate (1994) and Hughes 1928. Type: MEXICO. Sonora: nr Alamos, 26° 59 ′ N, (1998a) . We identify two robustly supported lineages that coin- 108 ° 57 ′W, 20 Sep 1890, Palmer 718 (holotype: NY!; cide with the geographical disjunction between populations from isotype: US!). western and northwestern Mexico (western Guerrero and Micho- Leu caena pubsecens Britton & Rose, N. Amer. Fl. 23: 122. ac á n to Sonora and Baja California) and those from Pacifi c 1928. Type: MEXICO. Sinaloa: nr Mazatl á n, 23° 14′ N, coastal Oaxaca, southwestern Chiapas, spanning the Isthmus of 106° 24′ W, 1925, J.G. Ortega 5988 (holotype: NY!; iso- Tehuantepec to Veracruz ( Fig. 2B ). Geographically structured ge- types: GH!US!). netic variation of this sort is a common feature of seasonally dry Leu caena sinaloensis Britton & Rose, N. Amer. Fl. 23: 124. tropical forest plants (e.g., Lavin, 2006 ; Pennington et al., 2009 ) 1928. Type: MEXICO. Sinaloa: vicinity of Palmar, and provides a robust basis for delimitation of two species, here 22 ° 13 ′ N, 105 ° 36 ′W, 15 Apr 1910, Rose et al., 14650 (ho- recognized as L. lanceolata (the typical northwestern lineage) lotype: NY!; isotype: US!). and L. cruziana Britton & Rose for the Oaxaca, Veracruz, and Leu caena sonorensis Britton & Rose, N. Amer. Fl. 23: 122. western Chiapas lineage. 1928. Type: MEXICO. Sonora: Sierra de Alamos, nr Recognition of L. cruziana as a species distinct from L. lan- Alamos, 26° 58 ′ N, 108 ° 57 ′ W, 14 Mar 1910, Rose et al., ceolata is congruent with results from analysis of plastid and 12821 (holotype: NY!; isotype: US!). nuclear DNA, AFLP data, and geography, and strongly sup- Leu caena nitens M. E. Jones, Contr. West Bot. 15: 136. 1929. ported by fi xed differences indicative of isolation and consistent Type: MEXICO. Sinaloa: nr Mazatl á n, 23° 14 ′ N, with the phylogenetic species concept (see above). Re-exami- 106 ° 24 ′W, 20 Nov 1926, Jones 22465 (holotype: POM!; nation of the somewhat complex and overlapping patterns of isotypes: MO!US!). variation in quantitative leaf and pod traits in the light of these For additional material examined, see online Appendix S6. new results, suggests that this new division is satisfactory in Leucaena cruziana Britton & Rose, N. Amer. Fl. 23: 123. comparison to previous classifi cations. The narrower circum- 1928. Type: MEXICO. Veracruz: Barranca de Panoaya, scription of L. lanceolata proposed here to include only popula- 19 ° 18 ′ N, 96 ° 25 ′ W, Dec 1919, Purpus 8387 (holotype: NY!; tions from western Guerrero to Sonora including Baja California isotypes: GH!UC!US!). ( Fig. 2B ), creates a morphologically more coherent species un- Leu caena rekoi Britton & Rose, N. Amer. Fl. 23: 122. 1928. complicated by the variability in pod traits and especially pod Type: MEXICO. Oaxaca, nr Pochutla, close to the Pa- vestiture found in coastal Michoacá n and parts of Oaxaca, cifi c coast, 15 ° 44 ′ N, 96 ° 28 ′W, 28 Sep 1917, Reko 3632 which was highlighted as problematic by Hughes (1998a) . Leu- (lectotype, fl owering shoot and leaves only: US!). caena lanceolata , as circumscribed here, generally has leaves Leu caena purpusii Britton & Rose, N. Amer. Fl. 23: 123. with 3 – 5 pairs of pinnae, 4 – 6 pairs of leafl ets per pinna and 1928. Type: MEXICO. Veracruz: Rim of barranca at Re- leafl ets < 20 mm wide and pods < 18 cm long and < 22 mm wide, ° ′ ° ′ Purpus 10607 while L. cruziana has leaves with 2 – 3( − 4) pairs of pinnae, 3 – mudadero, 19 15 N, 96 34 W, Jan 1926, 4( − 5) pairs of leafl ets per pinna, leafl ets 20 – 35 mm wide, and (holotype: NY!; isotype: US!). pods (16 – )20 – 37 cm long and (16 – )20 – 32 mm wide, although Leu caena lanceolata var. sousae (S. Zá rate) C. E. Hughes, there are no clear-cut discontinuities in any of these traits. Contr. Univ. Michigan Herb. 21: 288. 1997. Leucaena The revised synonymy for L. lanceolata listed below is less lanceolata subsp. sousae S. Z á rate, Anales Inst. Biol. extensive than suggested by Hughes (1998a), but still includes Univ. Auton. Mé xico, BOT. 65: 117. 1994. Type: MEX- ICO. Oaxaca: 17 km WNW of Puerto Escondido, Distr. fi ve species described by Britton and Rose (1928), who tended ° ′ ° ′ to pigeon-hole the minor variants they observed among the lim- Juquila, 15 57 N, 97 13 W, 21 Oct 1976, Sousa 6390 ited material available to them, as distinct species. (holotype: MEXU!; isotype: UC!). Under this new division, pod indumentum, which was previ- Three of the nine species previously recognized by Britton ously used as one character to distinguish typical L. lanceolata and Rose (1928) and subsequently treated as conspecifi c with var. lanceolata from L. lanceolata var. sousae , but for which L. lanceolata by Z á rate (1994) and Hughes ( 1998a ) — L. rekoi , there were several notable and problematic exceptions (Hughes, L. cruziana and L. purpusii — as well as the subsequently de- 1998a ), is confi rmed to be an unreliable and labile character for scribed L. lanceolata subsp. sousae ( Z á rate, 1994 ; Hughes, species delimitation. Pods vary from densely velutinous to gla- 1998a), have type localities (from Pochutla, Oaxaca; Barranca brous (when pods are often lustrous or glossy) within and de Panoaya, Veracruz; Remudadero, Veracruz; and Puerto Es- among populations of both L. cruziana and L. lanceolata , just condido, Oaxaca, respectively) that fall within the distribution as it does within several other species of Leucaena (e.g., L. di- of the southeastern lineage. The three earlier names by Britton versifolia , L. lempirana , L. trichandra ) ( Hughes, 1998a ). and Rose (1928) were published simultaneously in their North Leucaena lanceolata S. Watson, Proc. Amer. Acad. Arts 21: American Flora, but doubt has been cast over the identity of L. 427. 1886. Type: MEXICO. Chihuahua: Batopilas, Hacienda rekoi (Z á rate, 1994), because the type collection is a mixed San Miguel, SW Chihuahua, 27 ° 53 ′ N, 108 ° 26 ′ W, Sep 1885, gathering of leaves and fl owers from Leucaena , which Z á rate Palmer 6 (holotype: GH!; isotype: NY!UC!US!). (1994) suggests are doubtfully distinguishable from L. macro- Leu caena microcarpa Rose, Contr. U. S. Natl. Herb. 5: 141. phylla , and fruits of Caesalpinia (Coulteria) velutina (Britton 1897. Type: MEXICO. Baja California Sur: nr Mira- & Rose) Standl. Of the two remaining names, which are both fl ores, 23° 21 ′ N, 109 ° 47 ′ W, 13 Oct 1890, Brandegee 186 based on types from geographically closely adjacent localities (holotype: US!; isotype: UC!). in Veracruz, we are choosing to use the name L. cruziana , 2060 American Journal of Botany [Vol. 98 which appears before L. purpusii , albeit on the same page in de Quieroz , L. P. , and M. Lavin . 2011 . Coursetia (Leguminosae) from Britton and Rose (1928), to recognize the southeastern lineage eastern Brasil: The monophyly of three caatinga-inhabiting species is as a distinct species. revealed by nuclear ribosomal and chloroplast DNA sequence analy- Denser sampling of accessions, and especially from the sis. Systematic Botany 36 : 69 – 79 . populations in Veracruz where only a single individual is in- Dick , C. W. , K. Abdul-Salim , and E. Bermingham. 2003 . Molecular systematic analysis reveals cryptic Tertiary diversifi cation of a cluded in this study) will be needed to assess the full extent widespread tropical rainforest tree. American Naturalist 162 : of variation across this group and the possibility of further 691 – 703 . subdivision. Dobzhansky , T. H. 1935 . A critique of the species concept in biology. For additional material examined, see Appendix S6. Philosophy of Science 2 : 344 – 355 . Donoghue , M. J. 1985 . A critique of the biological species concept and recommendations for phylogenetic alternative. Bryologist 88 : LITERATURE CITED 172 – 181 . Alexander , P. J. , G. Rajanikanth , C. Bacon , and C. D. Bailey . Donoghue , M. J . 2008 . A phylogenetic perspective on the distribution of 2007 . Rapid inexpensive recovery of high quality plant DNA using plant diversity. Proceedings of the National Academy of Sciences, USA a reciprocating saw and silica-based columns. Molecular Ecology 105 : 11549 – 11555 . Notes 7 : 5 – 9 . Doyle , J. J. , J. L. Doyle , and A. H. D. Brown . 1990 . Analysis of a Á lvarez , I. , and J. F. Wendel . 2003 . Ribosomal ITS sequences and polyploid complex in Glycine with chloroplast and nuclear DNA. plant phylogenetic inference. Molecular Biology and Evolution 29 : Australian Systematic Botany 3 : 125 – 136 . 417 – 434 . de Stefano , R. D. , G. C. Fern á ndez-Concha , L. L. Can-Itza , and Anderson , E. , and G. L. Stebbins . 1954 . Hybridization as an evolution- M. Lavin . 2010 . The morphological and phylogenetic distinctions ary stimulus. Evolution 8 : 378 – 388 . of Coursetia greenmanii (Leguminosae): Taxonomic and ecological Baack , E. J. , K. D. Whitney , and L. H. Rieseberg . 2005 . Hybridization implications. Systematic Botany 35 : 289 – 295 . and genome size evolution: Timing and magnitude of nuclear DNA Evanno , G. , S. Regnaut , and J. Goudet . 2005 . Detecting the number content increases in Helianthus homoploid hybrid species. New of clusters of individuals using the software structure : A simulation Phytologist 167: 623 – 630. study. Molecular Ecology 14 : 2611 – 2620 . Bailey , C. D. , C. E. Hughes , and S. A. Harris. 2004 . Using RAPDs Filer , D. 2008 . BRAHMS, version 6.60 [computer program]. Oxford to identify DNA sequence loci for species level phylogeny re- University Press, Oxford, UK. construction: An example from Leucaena (Fabaceae). Systematic Fulton , T. M. , R. van der Hoeven , N. Eannetta , and S. D. Tanksley . Botany 29 : 4 – 14 . 2002 . Identifi cation, analysis, and utilization of conserved ortholog Barluenga, M., K . N. St ö lting, W. Salzburger, M. Muschick, and set markers for comparative genomics in higher plants. Plant Cell 1 4 : A. Meyer . 2006 . Sympatric speciation in Nicaraguan crater lake 1457 – 1467 . cichlid fi sh. Nature 439: 719 – 723. Goloboff , P. 2000 . NONA: A tree searching program [computer pro- Barraclough , T. G. 2010 . Evolving entities: Towards a unifi ed frame- gram]. Available at website http://www.cladistics.com . work for understanding diversity at the species and higher lev- Gompert , Z. , M. L. Forister , J. A. Fordyce , C. C. Nice , R. J. W. els. Philosophical Transactions of the Royal Society, B, Biological Illiamson , and A. C. Buerkle . 2010 . Bayesian analysis of mo- Sciences 365 : 1801 – 1813 . lecular variance in pyrosequences quantifi es population genetic struc- Barraclough , T. G. , and A. P. Vogler . 2000 . Detecting the geograph- ture across the genome of Lycaeides butterfl ies. Molecular Ecology ical pattern of speciation from species-level phylogenies. American 19 : 2455 – 2473 . Naturalist 155 : 419 – 434 . Gonz á lez , D. 2010 . Optimizaci ó n del m é todo SCAR (sequence charac- Barraclough , T. G. , A. P. Vogler , and P. H. Harvey . 1998 . Revealing terized amplifi ed region) que favorece el aislamiento de loci polim ó rfi the factors that promote speciation. Philosophical Transactions of the cos para estudios fi logené ticos en taxa cercanamente relacionados. Royal Society, B, Biolgoical Sciences 353 : 241 – 249 . Revista Mexicana de Biodiversidad 81 : 183 – 185 . Beck , J. B. , M. D. W. Indham , G. Yatskievych , and K. M. Pryer . Govindarajulu , R. , C. E. Hughes , P. J. Alexander , and C. D. Bailey . 2010 . A diploids-fi rst approach to species delimitation and interpret- 2011. The complex evolutionary dynamics of ancient and recent poly- ing polyploid evolution in the fern genus Astrolepis (Pteridaceae). ploidy in Leucaena (Leguminosae; Mimosidae). American Journal of Systematic Botany 35 : 223 – 234 . Botany 98 : 2064 – 2076 . Brewbaker , J. L. 1987 . Leucaena : A multipurpose tree genus for tropi- Grant , V. E. 1971 . Plant speciation. Columbia University Press, New cal agroforestry. In H. A. Steppler and P. K. Nair [eds.], Agroforestry: York, New York, USA. A decade of development. International Council for Research in Harris , S. A. , C. E. Hughes , R. J. Abbott , and R. Ingram . 1994 . Genetic Agroforestry, Nairobi, Kenya. variation in Leucaena leucocephala (Lam.) de Wit. (Leguminosae: Britton, N. L. , and J. N. Rose . 1928 . North American fl ora, vol. 23 , part Mimosoideae). Silvae Genetica 43 : 159 – 167 . 2, Mimosaceae, 121 – 131 . Hijmans , R. J. 2010 . DIVA-GIS [computer program]. University of Brown , A. H. D. , J. L. Doyle , J. P. Grace , and J. J. Doyle. 2002 . Molecular California, Berkeley, California, USA. phylogenetic relationships within and among diploid races of Glycine to- Hughes , C. E. 1998a . Monograph of Leucaena (Leguminosae-Mimo- mentella (Leguminosae). Australian Systematic Botany 15 : 37 – 47 . soideae). Systematic Botany Monographs 55 : 1 – 244 . Cardoso , M. B. , M. T. Schifino-Wittmann , and M. H. Bodanese- Hughes , C. E . 1998b . Leucaena : A genetic resources handbook. Oxford Zanettini. 2000 . Taxonomic and evolutionary implications of intraspe- Forestry Institute, Oxford, UK. cifi c variability in chromosome numbers of species of Leucaena Benth. Hughes , C. E. , C. D. Bailey , and S. A. Harris. 2002 . Divergent (Leguminosae). Botanical Journal of the Linnean Society 134 : 549 – 556 . and reticulate species relationships in Leucaena (Fabaceae) in- Chapman , M. A. , and J. M. Burke . 2007 . Genetic divergence and hy- ferred from multiple data sources: Insights into polyploid ori- brid speciation. Evolution 61 : 1773 – 1780 . gins and nrDNA polymorphism. American Journal of Botany 8 9 : Choi , H. K. , M. Luckow , J. J. Doyle , and D. R. Cook . 2006 . Devel- 1057 – 1073 . opment of nuclear gene-derived molecular markers linked to legume Hughes , C. E. , C. D. Bailey , S. Krosnick , and M. Luckow . 2003 . genetic maps. Molecular Genetics and Genomics 276 : 56 – 70 . Relationships among genera of the informal Dichrostachys and Coyne , J. A. , and H. A. Orr . 2004 . Speciation. Sinauer, Sunderland, Leucaena groups (Mimosoideae) inferred from nuclear ribosomal ITS Massachusetts, USA. sequences. In B. Klitgaard and A. Bruneau [eds.], Advances in legume Davis , J. I. , and K. C. Nixon. 1992 . Populations, genetic variation, and the systematics, part 10, Higher level systematics. Royal Botanic Gardens, delimitation of phylogenetic species. Systematic Biology 41 : 421 – 435 . Kew, UK. December 2011] Govindarajulu et al. — Cryptic diversity and allopatry 2061

Hughes , C. E. , R. Govindarajulu , A. Robertson , S. A. Harris , and Pennington , R. T. , D. E. Prado , and C. A. Pendry. 2000 . Neotropical C. D. Bailey . 2007 . Serendipitous backyard hybridization and the seasonally dry forests and Quarternary vegetation changes. Journal of origin of crops. Proceedings of the National Academy of Sciences, Biogeography 27 : 261 – 273 . USA 104 : 14389 – 14394 . Perret , M. , A. Chautems , R. Spichiger , T. G. Barraclough , and Hughes , C. E. , and S. A. Harris. 1998 . A second spontaneous hybrid in V. Savolainen. 2007 . The geographical pattern of speciation the genus Leucaena (Leguminosae , Mimosoideae). Plant Systematics and and fl oral diversifi cation in the neotropics: The tribe Sinningieae Evolution 212 : 53 – 77 . (Gesneriaceae) as a case study. Evolution 61 : 1641 – 1660 . Hughes , C. E. , and R. J. Jones . 1999 . Environmental hazards of Leucaena. Pritchard , J. K. , M. Stephens , and P. Donnelly . 2000 . Inference of Leucaena— Adaptation, quality and farmings systems, Hanoi, ACIAR population structure using multilocus genotype data. Genetics 155 : proceedings. Australian Centre for International Agricutlural Research, 945 – 959 . Canberra 86 : 61 – 70 . Reeves , P. A. , and C. M. Richards. 2007 . Distinguishing terminal Kane , N. C. , M. G. King , M. S. Barker , A. Raduski , S. Karrenberg , monophyletic groups from reticulate taxa: Performance of phe- Y. Yatabe , S. Knapp , and L. H. Rieseberg . 2009 . Comparative netic, tree-based, and network procedures. Systematic Biology 5 6 : genomic and population genetic analyses indicate highly porous ge- 302 – 320 . nomes and high levels of gene fl ow between divergent Helianthus Rieseberg , L. H. 1995 . The role of hybridization in evolution: Old wine species. Evolution 63 : 2061 – 2075 . in new skins. American Journal of Botany 82 : 944 – 953 . Kondrashov , A. S. 1986 . Sympatric speciation: When is it possible? Rieseberg , L. H . 1997 . Hybrid origins of plant species. Annual Review of Biological Journal of the Linnean Society 27 : 201 – 223 . Ecology and Systematics 28 : 359 – 389 . Lavin , M. 2006 . Floristic and geographic stability of discontinuous sea- Rieseberg , L. H. , O. Raymond , D. M. Rosenthal , Z. Lai , K. sonally dry tropical forests explains patterns of plant phylogeny and Livingstone , T. Nakazato , J. L. Durphy , et al . 2003 . Major endemism. In R. T. Pennington, J. A. Ratter, and G. P. Lewis [eds.], evolutionary transitions in wild sunfl owers facilitated by hybridiza- Neotropical savannas and seasonally dry forests: Plant diversity, bio- tion. Science 301 : 1211 – 1216 . geographic patterns and conservation, 433 – 447. CRC Press, Boca Sang , T. , D. J. Crawford , and T. F. Stuessy . 1997 . Chloroplast Raton, Florida, USA. DNA phylogeny, reticulate evolution, and biogeography of Paeonia Lavin , M. , M. Thulin , J. N. Labat , and R. T. Pennington . 2000 . (Paeoniaceae). American Journal of Botany 84 : 1120 – 1136 . Africa, the odd man out: Molecular biogeography of dalbergioid S ä rkinen , T. S. , J. L. Marcelo Pe ñ a , A. D. Yomona , M. F. Simon , legumes (Fabaceae) suggests otherwise. Systematic Botany 25 : R. T. Pennington , and C. E. Hughes . 2011 . Underestimated 449 – 467 . endemic species diversity in the dry inter-Andean valley of the Rí o Linder , C. R. , and L. H. Rieseberg . 2004 . Reconstructing patterns Mara ñ ó n, northern Peru: An example from Mimosa (Leguminosae: of reticulate evolution in plants. American Journal of Botany 91 : Mimosoideae). Taxon 60: 139 – 150. 1700 – 1708 . Savolainen , O. , M. C. Anstett , C. Lexer , I. Hutton , J. J. Clarkson , Lohithaswa , H. C. , F. A. Feltus , H. P. Singh , C. D. Bacon , C. D. M. V. Norup , M. Powell , et al . 2006 . Sympatric speciation in Bailey , and A. H. Paterson . 2007 . Leveraging the rice genome palms on an oceanic island. Nature 441 : 210 – 213 . sequence for monocot comparative and translational genomics. Savolainen , V. , R. S. Cowan , A. P. Vogler , G. K. Roderick , and R. Theoretical and Applied Genetics 115 : 237 – 243 . Lane . 2005 . Towards writing the encyclopaedia of life: An intro- Luckow , M. 1997 . Generic relationships in the Dichrostachys group duction to DNA barcoding. Philosophical Transactions of the Royal (Leguminosae: Mimosoideae): Evidence from chloroplast DNA re- Society, B, Biological Sciences 360 : 1805 – 1811 . striction sites and morphology. Systematic Botany 22 : 189 – 198 . Schifino-Wittmann , M. T. , M. Cardoso , T. Boff , and C . Simioni 2000 . Mav á rez , J. , C. A. Salazar , E. Bermingham , C. Salcedo , C. D. Chromosome numbers and unreduced gametes in species of Leucaena Jiggins , and M. Linares . 2006 . Speciation by hybridization in Benth. (Leguminosae): New contributions for the taxonomy, evolu- Heliconius butterfl ies. Nature 441 : 868 – 871 . tionary studies and genetic breeding of the genus. Arbora Publishers, Mayr , E. 1942 . Systematics and the origin of species. Columbia Zvolen, Slovakia. University Press, New York, New York, USA. Schrire , B. D. , M. Lavin , and G. P. Lewis . 2005 . Global distribu- McVaugh , R. 1987 . Flora Novo-Galiciana: A descriptive account of the tion patterns of the Leguminosae: Insights from recent phylogenies. vascular plants of western Mexico. University of Michigan Press, Ann Biologiske Skrifter 55 : 375 – 422 . Arbor, Michigan, USA. Shaw , J. , E. B. Lickey , E. E. Schilling , and R. L. Small . Mü ller , K. 2006 . Incorporating information from length-mutational events 2007 . Comparison of whole genome chloroplast sequences to into phylogenetic analysis. Molecular Phylogenetics and Evolution 3 8 : choose noncoding regions for phylogenetic studies in angiosperms: 667 – 676 . The tortoise and the hare III. American Journal of Botany 94 : 275 – Nixon , K. C. 2002 . WinClada (Beta) version 1.00.08 [computer pro- 288. gram]. Published by author, Ithaca, New York, USA. Simmons , M. P. , and H. Ochoterena . 2000 . Gaps as characters in Nixon , K. C. , and Q. D. Wheeler . 1990 . An amplifi cation of the phylo- sequence-based phylogenetic analyses. Systematic Biology 4 9 : genetic species concept. Cladistics 6 : 211 – 224 . 369 – 381 . Papadopulosa , A. S. , W. J. Baker , D. Craync , R. K. Butlind , R. G. Smith , M. A. , J. J. Rodriguez , J. B. Whitfield , A. R. Deans , D. J. Kynastb , I. Huttone , and V. Savolainen . 2011 . Speciation with Janzen , W. Hallwachs , and P. D. N. Hebert . 2008 . Extreme gene fl ow on Lord Howe Island. Proceedings of the National Academy diversity of tropical parasitoid wasps exposed by iterative integra- of Sciences, USA 108: 13188 – 13193. tion of natural history, DNA barcoding, morphology, and collec- Pennington , R. T. , A. Daza , and M. Lavin . 2011 . Poissonia eriantha tions. Proceedings of the National Academy of Sciences, USA 105 : (Leguminosae) from Cuzco, Peru: An overlooked species underscores 12359 – 12364 . a pattern of narrow endemism common to seasonally dry neotropical Sorensson , C. T. , and J. L. Brewbaker . 1994 . Interspecifi c com- vegetation. Systematic Botany 36 : 59 – 68 . patibility among 15 Leucaena (Leguminosae: Mimosoideae) spe- Pennington , R. T. , M. Lavin , and A. Oliveira-Filho . 2009 . Woody cies via artifi cial hybridization. American Journal of Botany 8 1 : plant diversity, evolution and ecology in the tropics: Perspectives from 240 – 247 . seasonally dry tropical forests. Annual Review of Ecology, Evolution Sotuyo , S. , A. Delgado-Salinas , M. W. Chase , G. P. Lewis , and K. and Systematics 40 : 437 – 457 . Oyama . 2007 . Cryptic speciation in the Caesalpinia hintonii com- Pennington , R. T. , M. Lavin , T. Sä rkinen , G. P. Lewis , B. B. plex (Leguminosae: Caesalpinioideae) in a seasonally dry Mexican Klitgaard , and C. Hughes. 2010 . Contrasting plant diversifi ca- forest. Annals of Botany 100 : 1307 – 1314 . tion histories within the Andean biodiversity hotspot. Proceedings Sotuyo , S. , and G. P. Lewis . 2007 . A new species of Caesalpinia of the National Academy of Sciences, USA 107 : 13783 – 13787 . from the Rí o Balsas Depression, Mexico, and an updated taxonomic 2062 American Journal of Botany [Vol. 98

circumscription of the Caesalpinia complex (Leguminosae: Caesal- Orobanche minor (Orobanchaceae) . Plant Systematics and Evolution pinioideae: Caesalpinieae: Poincianella Group). Brittonia 59 : 33 – 282: 31 – 42 36 . Vos , P. , R. Hogers , M. Bleeker , M. Reijans , T. van de Lee , M. Hornes , Stamatakis , A. , P. Hoover , and J. Rougemont . 2008 . A rapid boot- A. Frijters , et al. 1995 . AFLP: A new technique for DNA fi n- strap algorithm for RAxML Web-Servers. Systematic Biology 5 7 : gerprinting. Nucleic Acids Research 23 : 4407 – 4414 . 758 – 771 . Wendel , J. F. , A. Schnabel , and T. Seelanan. 1995 . Bidirectional Thompson , J. D. , T. J. Gibson , F. Plewniak , F. Jeanmougin , and D. interlocus concerted evolution following allopolyploid speciation in G. Higgins. 1997 . The CLUSTAL_X windows interface: Flexible cotton (Gossypium ). Proceedings of the National Academy of Sciences, strategies for multiple sequence alignment aided by quality analy- USA 92 : 280 – 284 . sis tools. Nucleic Acids Research 25 : 4876 – 4882 . Z á rate , P. S. 1994 . Revisi ó n del g é nero Leucaena Benth. en Mé xico. Thorogood , C. , F. Rumsey , S. Harris , and S. Hiscock . 2009 . Gene Anales de Instituto Biolog í a Universidad Nacional Autonoma de fl ow between alien and native races of the holoparasitic angiosperm M é xico, Bot á nica 65 : 83 – 162 .

Appendix 1. Plant material used for phylogenetic study and AFLP analyses. Seed lot number is provided if DNA was extracted from a seedling raised from a seedlot. Herbarium vouchers are all at FHO, with duplicates variously deposited at CAS, EAP, K, MEXU, NY, US, and MO. For complete locality and GenBank information, see Appendix S7 (in Supplemental Data with the online version of this article).

(A) Phylogenetic study Hughes CE 1421, 4_91 -Erandique, Lempira, Honduras. Hughes CE 1654, 128/92 -Tierra Colorada, Oaxaca, Mexico. Hughes CE 1682 , 137/92 -La Trinitaria, Chiapas, Mexico. Hughes CE 1701 , 140/92 -San Taxon , Voucher , Seed lot (where applicable, e.g., 51/81) -Locality. Marcos, San Marcos, Guatemala. Hughes CE 2121 , -Matatlan, Oaxaca, Mexico. L. trichodes (Jacquin) Bentham , Hughes CE 775, 2/86/07 L. collinsii Britton & Rose , Hughes CE 1187 , 51/88 -Huehuetenango, -Trujillo, Venezuela. Hughes CE 997, 61/88 -Manabi, Ecuador. L. Guatemala. Hughes CE 527, 52/88 -Chiapas, Mexico. L. cruziana, Hughes zacapana (C.E. Hughes) R. Govindarajulu & C.E. Hughes , Hughes CE CE 1672 , 134/92 -Matias Romero, Oaxaca, Mexico. Hughes CE 2180 , -El 1096 , 57/88 -Chiquimula, Guatemala. Hughes CE 1120 , 56/88 -Zacapa, Limon, Palmasola, Veracruz, Mexico. Hughes CE 559 , 43/85 -Oaxaca, Guatemala. Hughes CE 299 , 18/84 -Progreso, Guatemala. Mexico. Hughes CE 913 , -Veracruz, Mexico. Hughes CE 835 , 51/87 -Oaxaca, Mexico. Hughes CE 872 , 50/87 -Oaxaca, Mexico. L. cuspidata Standley , Hughes CE 1580, -Tolantongo Cardonal, Hidalgo, Mexico. (B) AFLP analyses Hughes CE 1583, 89/92 -Mina San Miguel, Hidalgo, Mexico. Hughes CE 1586, 88/92 -Jacala, Hidalgo, Mexico. L. esculenta (Sess é & Moci ñ o ex DC.) Bentham , Hughes CE 2114, -Miahuatlá n, Oaxaca, Mexico. Hughes Taxon , Voucher , Seed lot (where applicable, e.g., 51/81) -Locality, -Number of CE 2143 -Huajuapan de Leó n, Oaxaca, Mexico. Bailey & Ochoterena 216, individuals analyzed per population. -Mexico, Mexico. Hughes CE 894 , 47/87 -Guerrero, Mexico. Hughes CE 903 , 48/87/01 -Michoacan, Mexico. Desmanthus fruticosus Rose , Hughes L. collinsii Britton & Rose , Hughes CE 527, 45/85 -Narcisco Mendoza, CE 1532, 109/92 -La Paz,Baja california, Mexico. L. greggii S. Watson , Chiapas, Mexico -8, Hughes CE 1187, 51/88 -Chacaj, Huehuetenango, Hughes CE 1057, 82/87 -Nuevo Leó n, Mexico. Hughes CE 695 , 19/86 Guatemala -13. L. cruziana Britton & Rose , Hughes CE 1672 , 134/92 -Nuevo Leó n, Mexico. Hughes CE 695, 20/86/02 -Nuevo Leó n, Mexico. -Matias Romero, Oaxaca, Mexico -1, Hughes CE 559, 43/85 -San jon, Hughes CE 695 , 21/86/07 -Nuevo Le ó n, Mexico. L. lanceolata S. Watson , Oaxaca, Mexico -10, Hughes CE 2166 , -Cerro Gordo, Veracruz, Mexico Hughes CE 1577, 90/92 -Alamos, Sonora, Mexico. Hughes CE 631 , 46/85 -1, Hughes CE 2171, -La Mancha, Palmasola, Veracruz, Mexico -1, -Michoac á n, Mexico. L. lempirana C.E. Hughes, Hughes CE 1411 , Hughes CE 835, 51/87 -Puerto Angel, Oaxaca, Mexico -11. L. cuspidata 6/91/03 -Negrito, Yoro, Honduras. Hughes CE 1447, 5/91 -Aguan Valley, Standley , Hughes CE 1851, 83/94 -Camarones, Hidalgo, Mexico -9, Yoro, Honduras. L. macrophylla subsp. istmensis C.E. Hughes, Hughes Hughes CE 1586 , 88/92 -Jacala, Hidalgo, Mexico -1, Hughes CE 1583 , CE 580, 47/85 -Oaxaca, Mexico. L. macrophylla subsp. macrophylla 89/92 -Mina San Miguel, Hidalgo, Mexico -1. L. esculenta (Sess é & Bentham, Hughes CE 1179, 55/88 -Guerrero, Mexico. Hughes CE 2076 , Mociñ o ex DC.) Bentham , Hughes CE 2114 , -Miahuatl á n, Oaxaca, -Coxcatlan, Puebla, Mexico. L. magnifi ca (C.E Hughes) C.E. Hughes , Mexico -1, Hughes CE 2143, -Huajuapan de Leó n, Oaxaca, Mexico -1, Hughes CE 1089 , 58/88 -Chiquimula, Guatemala. Hughes CE 412 , 19/84 Hughes CE 894, 47/87 -San Martí n Pachivia, Guerrero, Mexico -10, -Chiquimula, Guatemala. L. matudae (S. Z á rate) C.E. Hughes, Hughes CE Hughes CE 903 , 48/87 -Michoac á n, Mexico -8. L. greggii S. Watson , 2153, -San Miguel Tecuixiapan, Guerrero, Mexico. Hughes CE 879 , 49/87 Hughes CE 695 , 19/86 -El Barrial, Nuevo Le ó n, Mexico -6. L. lanceolata -Guerrero, Mexico. L. multicapitula Schery , Hughes CE 1024 , 86/87 S. Watson , Hughes CE 603 , 44/85 -Escuinapa, Sinaloa, Mexico -7, -Penas Blancas, Guanacaste, Costa Rica. Hughes CE 1025, 81/87 -Los Hughes CE 631 , 46/85 -Playa Azul, Michoac á n, Mexico -10, Hughes Santos, Panama. Schleinitzia novo-guineensis (Warb.) Verdc. , Chaplin, CE 1577, -Alamos, Sonora, Mexico -1. L. lempirana C.E. Hughes , 57/84 -Munda, Soloman Islands. L. pueblana Britton & Rose , Hughes Hughes CE 1447 , 5/91 -Valle del Agu á n, Yoro, Honduras -16, Hughes CE 1648, 125/92 -Lower Tehuacan Valley, Oaxaca, Mexico. Hughes CE CE 1411 , 6/91 -Cuyamapa, Yoro, Honduras -9. L. macrophylla subsp. 2089 , -Cuicatlan, Oaxaca, Mexico. Hughes CE 2140 , -Santo Domingo macrophylla Bentham , Hughes CE 2076 , -Coxcatlan, Puebla, Mexico -1, Tonala, Oaxaca, Mexico. L. pulverulenta (Schlechtendal) Bentham , Hughes CE 1179, 55/88 -Vallecitos, Guerrero, Mexico -10, Hughes CE Hughes CE 1051 , 83/87/02 -Tamaulipas, Mexico. Hughes CE 1058 , 2156 , -Grutas de Cacahuamilca, Taxco, Guerrero, Mexico -1, Hughes CE 84/87 -Texas, USA. Hughes CE 1593, -Xilitla, San Luis Potosí , Mexico. 2164, -Cerro El Encinal, Iguala, Guerrero, Mexico -1. L. macrophylla Hughes CE 1611 , -Huejutla de Reyes, Hidalgo, Mexico. Hughes CE 1866 , subsp. istmensis C.E. Hughes , Hughes CE 580, 47/85 -San Isidro Llano -Misantla, Veracruz, Mexico. L. retusa Bentham in Gray , Rajanikanth Grande, Oaxaca, Mexico -10. L. magnifi ca (C.E Hughes) C.E. Hughes , & Bailey 2009, 23/09/02 -Eddy Co, New Mexico, USA. Bendeck s.n. , Hughes CE 412, 19/84 -El Rinc ó n, Chiquimula, Guatemala -10, Hughes 23/86 -Coahuila, Mexico. L. salvadorensis Standley ex Britton & Rose , CE 1089, 58/88 -El Carrizal, Chiquimula, Guatemala -8. L. matudae Hughes CE 1407, 7/91 -San Juan de Limay, Esteli, Nicaragua. Hughes (S. Zá rate) C.E. Hughes , Hughes CE 2153, -San Miguel Tecuixiapan, CE 742 , 17/86 -Choluteca, Honduras. Hughes CE 746 , 34/88 -Choluteca, Guerrero, Mexico -1, Hughes CE 879 , 49/87 -Mezcala, Guerrero, Honduras. L. shannonii Donnell Smith , Hughes CE 1417 , 1/91 -Santa Mexico -9, Hughes CE 2148, -San Juan Tetelcingo, Guerrero, Mexico Caterina Mita, Jutiapa, Guatemala. Hughes CE 1676 , 135/92 -Cintalapa -1. L. multicapitula Schery , Hughes CE 1025 , 81/87 -Los Santos, de Figueroa, Chiapas, Mexico. Hughes CE 1714, 141/92 -Santa Rita, Panama -13. L. pueblana Britton & Rose , Hughes CE 1648 , 125/92 Yoro, Honduras. Hughes CE 507, 53/87 -Champoton, Campeche, Mexico. -Tehuacan Valley, Oaxaca, Mexico -1, Hughes CE 2089 , -Cuicatlan, L. trichandra (Zuccarini) Urban , Hughes CE 1106, 53/88 -Guatemala, Oaxaca, Mexico -1, Hughes CE 2140, -Santo Domingo Tonala, Oaxaca, Guatemala. Hughes CE 1130 , 54/88 -Huehuetenango, Guatemala. Mexico -1, Hughes CE 2077, -Teotitlá n del Camino, Oaxaca, Mexico -1, December 2011] Govindarajulu et al. — Cryptic diversity and allopatry 2063

Hughes CE 2092 , -Dominguillo, Oaxaca,Mexico -1, Hughes CE -3, Hughes CE 2166, -El Zamorano, Francisco Morazan, Honduras -1. 2139 , -Santo Domingo Tonala, Oaxaca, Mexico -1. L. pulverulenta L. trichandra (Zuccarini) Urban, Hughes CE 1654, 128/92 -Oaxaca, (Schlechtendal) Bentham , Hughes CE 1051, 83/87 -Tamaulipas, Mexico -1, Hughes CE 1682, 137/92 -Chiapas, Mexico -1, Hughes CE Mexico -4, Hughes CE 1058, 84/87 -Texas, USA -3 L. retusa Bentham 1421 , 4/91 -Erandique, Lempira, Honduras -10, Hughes CE 1106, 53/88 in Gray , Bendeck s.n., 23/86 -Coahuila, Mexico -9. L. salvadorensis -Los Guates, Guatemala -5, Hughes CE 1130, 54/88 -Huehuetenango, Standley ex Britton & Rose, Hughes CE 742 , 17/86 -La Garita, Guatemala -1, Hughes CE 1708 , -Erandique, Lempira, Honduras -1, Choluteca, Honduras -8, Hughes CE 746, 34/88 -Choluteca, Honduras Hughes CE 1709, -Erandique, Lempira, Honduras -1, Hughes CE 1710 , -9, Hughes CE 36/88 , 36/88 -La Garita, Choluteca, Honduras -9, -Erandique, Lempira, Honduras -1. L. trichodes (Jacquin) Bentham , Hughes CE 1407, 7/91 -Esteli, Nicaragua -9, Hughes CE 98/90, 98/90 -La Garela, Choluteca, Honduras -9, Hughes CE 1211 , 99/90 -Namali, Hughes CE 775, 2/86 -Cuicas, Trujillo, Venezuela -9, Hughes CE 997 , Choluteca, Honduras -4. L. shannonii Donnell Smith, Hughes 61/88 -Manabi, Ecuador -11, Hughes CE 1418, 3/91 -Copan, Honduras CE 1417 , 1/91 -Jutiapa, Guatemala -10, Hughes CE 1676, 135/92 -2. L. zacapana (C.E. Hughes) R. Govindarajulu & C.E. Hughes , -Chiapas, Mexico -1, Hughes CE 1714, 141/92 -Yoro, Honduras -1, Hughes CE 299, 18/84 -Puerto de Golpe, El Progreso, Guatemala -1, Hughes CE 1399, 2/91 -La Puerta, Chontales, Nicaragua -4, Hughes Hughes CE 1120 , 56/88 -Vallecitos, Zacapa, Guatemala -8, Hughes CE CE 239 , 22/83 -Comayagua, Honduras -9, Hughes CE 282, 26/84 299, 15/83 -Puerto de Golpe, El Progreso, Guatemala -10, Hughes CE -Comayagua, Honduras -1, Hughes CE 507 , 53/87 -Campeche, Mexico 1096, 57/88 -El Carrizal, Chiquimula, Guatemala -13.