Molecular Systematics of The Red-Bellied and Golden-Fronted Woodpeckers

Abstract.

The Red-bellied Woodpecker species group (Melanerpes carolinus and relatives) is composed of five morphologically similar species whose limits have been unclear. The relationship of the Golden-fronted Woodpecker (M. aurifrons) to the remainder of the group is particularly uncertain. We used mitochondrial DNA sequences to examine the phylogeny of this group and its close relatives. We sequenced 872 bp, including fragments of the genes for ND2, ND3, COIII, and tRNAmet, of 11 species of Melanerpes. We constructed trees from combined sequences by using the maximum likelihood and Bayesian inference approaches. We found that M. aurifrons is not monophyletic but rather consists of two clades, one comprising tropical populations (M. santacruzi), the other, consisting of northern populations, being sister to M. carolinus. The Caribbean species, M. superciliaris, is sister to the carolinus—aurifrons clade. The group as a whole appears to have diversified into multiple lineages in response to several episodes of vicariance, perhaps associated with glacial—interglacial cycles. As a result of these findings, major taxonomic changes in the group are needed.

Resumen.

El grupo de especies de Melanerpes carolinus está compuesto por cinco especies morfológicamente similares cuyos límites han sido poco claros. Particularmente, la relación de M. aurifrons con el resto del grupo es incierta. Examinames las relaciones filogenéticas de este grupo y sus parientes cercanos usando secuencias de ADN mitocondrial. Para un total de 11 especies del género Melanerpes amplificamos 872 pb que incluían fragmentos de los genes ND2, ND3, COIII, y tRNAmet. Los árboles fueron construidos a partir de las secuencias combinadas usando los métodos de máxima verosimilitud e inferencia Bayesiana. Encontramos que M. aurifrons no es monofilético y se separa en dos clados diferentes, uno que incluye a las poblaciones tropicales (M. santacruzi) y otro formado por las poblaciones norteñas, el cual es el grupo hermano de M, carolinus. La especie caribeña M. superciliaris fue el taxón hermano del clado carolinus—aurifrons, El grupo en general parece haberse diversificado en varies linajes como respuesta a varies eventos vicariantes. De acuerdo con nuestros resultados, es altamente probable que los patrones biogeográficos y de especiación del grupo carolinus estén asociados con los ciclos glaciares— interglaciares. Nuestros resultados sugieren que es necesario realizar una actualización taxonómica del complejo.

Introduction

The genus Melanerpes (Piciformes, Picidae) comprises 17–22 species (Peters 1948, Dickinson 2003) that range from southeastern Canada to northern Argentina and some Caribbean islands. The species inhabit diverse forested habitats, from lowlands to mountains and from desert to rainforests (Short 1982, AOU 1998, Winkler and Christie 2002). Species richness of the genus is highest in Mesoamerica (nine species); Mexico has eight species, two of which are endemic (Table 1; AOU 1998, Dickinson 2003).

Melanerpes contains taxa that in the past have been allocated to other genera, such as Centurus, Tripsurus, Balanosphyra, Leuconerpes, Chryserpes, and Asyndesmus (Selander and Giller 1963). These genera were merged into a more inclusive Melanerpes by Peters (1948) and Short (1982), mainly on the basis of plumage characters and “clinal variation” among species (Short 1982). The Melanerpes carolinus group (a superspecies sensu Short 1982) comprises five morphologically similar species (M. carolinus, M. aurifrons, M. uropygialis, M. superciliaris, and M. hoffmannii; Table 1) that have white and black barring on their backs, solid gray or buff underparts, red crowns, yellow or red napes, red or yellow bellies and nasal tufts, and lack black patches on the auriculars and head (Short 1982). They occupy diverse lowland and submontane habitats from deserts (M. uropygialis) to tropical rain forests (e.g., M. a. santacruzi).

There is considerable morphological variation within some species of the complex. For instance, Short (1982) recognized three subspecies of M. uropygialis, five of M. superciliaris, although M. hoffmannii is monotypic. The Golden-fronted Woodpecker (M. aurifrons) ranges from southern Oklahoma south through eastern Mexico to northern Nicaragua and is notable for its extreme morphological variation (Howell and Webb 1995). Fourteen subspecies have been described, of which eight were recognized by Selander and Giller (1963), ten by Short (1982); one was described after those revisions (Dickerman 1987). Some resemble other forms in the carolinus group (e.g., M. a. dubius of the Yucatan Peninsula is similar to M. carolinus of Florida; see Winkler and Christie 2002; M. Patten, pers. comm.) whereas others resemble species outside the group (e.g., M. a. santacruzi is similar to M. pygmaeus). The distributional patterns of these populations are complex, with several zones of sympatry or parapatry among forms with limited interbreeding, producing unevenness in the taxonomy. Ridgway (1914), for example, recognized several species within the aurifrons complex, whereas Peters (1948) and the AOU (1957, 1983, 1998) united them all in a single species.

Morphological variation in M. aurifrons is most evident in the coloration of the head, underparts, and in the pattern of barring on the back. Generally, males have a red crown separated from a golden-orange nape (M. a. aurifrons)., but males of some populations (e.g., M. a. dubius, M. a. santacruzi, M. a. polygrammus) have color continuous from crown to nape; indeed, some populations have both a red crown and a red nape, while some have a red crown and orange nape. In both sexes, the nasal tufts range from yellow to red. The back and upper-parts are barred black and white, but the width of the white bars ranges from wide (M. a. aurifrons) to very narrow (M. a. dubius). The central rectrices vary from all black (M. a. aurifrons) to barred with white (most strongly in M. a. polygrammus). Finally, the color of breast and flanks varies widely: birds of northern populations (M. a. aurifrons) are light gray, while southern forms (M. a. dubius, M. a. santacruzi, M. a. polygrammus) have grayish-brown to olive brown underparts.

Selander and Giller (1963), in an extensive revision of Centurus, recognized M. aurifrons as a single species. Nevertheless, they pointed out that some subspecies in the complex might be more closely related to other species. They also suggested that M. aurifrons as well as M. carolinus, M. hoffmannii, and M. uropygialis belong to a closely related group. No studies since those of Selander and Giller (1963) and Short (1982) have addressed the taxonomy of the entire group, even though many have highlighted the need of a thorough revision of it (Howell and Webb 1995, Winkler and Christie 2002, Navarro-Sigüenza and Peterson 2004, Benz et al. 2006). Our goal in this study is to analyze molecular characters to evaluate phylogenetic relationships within the M. carolinus species group and with other putatively closely related taxa.

Methods

Taxonomic Sampling

We compared a total of 89 individuals of 11 species of Melanerpes (Figure 1; Table 2). These included the five species of the Red-bellied (or M. carolinus) species group sensu Short (1982). Of these we analyzed 8 of the 12 subspecies of M. aurifrons recognized by Winkler and Christie (2002). No samples of the insular forms leei (Cozumel), turneffensis (Turneffe Island, Belize), canescens (Roatán and Barbareta Islands, Honduras), or insulanus (Utila Island, Honduras) were available. We also examined four other species of Melanerpes from Middle America (M. chrysogenys, M. pygmaeus, M. hypopolius, and M. pucherani) and two from South America (M. candidus and M. cactorum). No samples of the closely related M. rubricapillus were available. As outgroups we used two species of Sphyrapicus (S. nuchalis and S. thyroideus), the likely sister group of Melanerpes (Winkler and Christie 2002, Webb and Moore 2005, Benz et al. 2006), and Dryocopus pileatus, available from GenBank (Gibb et al. 2007), as a representative of nonmelanerpine woodpeckers. All samples are documented by voucher specimens deposited in scientific collections available to the public (Table 2). The monotypic genus Xiphidiopicus, endemic to Cuba, likely represents another closely related taxon (Overton and Rhoads 2006), but we were unable to include samples of it.

Dna Extraction and Sequencing

Following the manufacturer's protocol, we extracted genomic DNA from frozen or ethanol-preserved tissue samples (muscle, heart or liver) with the DNEasy Extraction Kit (Qiagen). We amplified portions of three mitochondrial DNA genes: NADH dehydrogenase subunits 2 and 3 (ND2 and ND3) and cytochrome oxidase subunit 3 (COIII). The 366-bp ND2 fragment comprised tRNA-Met (last 30 bp) and NADH dehydrogenase subunit 2 (first 336 bp), and the 506-bp ND3-COIII fragment comprised the last 59 bp of COIII, 69 bp of tRNA-Gly, 351 bp of NADH dehydrogenase subunit 3, and 27 bp of tRNA-Arg. The ND2 fragment was amplified with primers L5215 (Helm-Bychowski and Cracraft 1983) and H5578 (Hackett 1996), the ND3 fragment with primers L10647 (Mindell et al. 1998) and H11151 (Chesser 1999).

PCR amplification was carried out in 50-µl reactions with a GeneAmp PCR system 9700 (Applied Biosystems). Thermocycling conditions for ND2 included an initial hold of 2 min at 94 °C followed by 10 cycles of 15 sec at 94 °C, 30 sec at 55 °C, and 30 sec at 72 °C, followed by three cycles of 15 sec at 94 °C, 15 sec at 50 °C, and 30 sec at 72 °C, and a final hold of 3 min at 72 °C. For ND3 thermocycling conditions were 35 cycles of 1 min at 94 °C, 1 min at 50 °C, and 1 min at 72 °C, followed by a 7-min final extension period at 72 °C.

Following the manufacturer's procedures, we purified the resulting PCR products with Millipore purifying plates (Millipore Corporation) and visualized the products on 1% agarose gels stained with ethidium bromide. Purified PCR products were amplified and cycle-sequenced in 20-µl reactions with the ABI Prism BigDye v3.1 kit (Applied Biosystems), with the same primers mentioned above for amplification. Cycle-sequenced products were purified with Sephadex G50 columns (Sigma Corporation) or CleanSEQ (Agencourt) magnetic-bead purification and analyzed in an ABI 377 automated DNA sequencer (Applied Biosystems).

Phylogenetic Analyses

We assembled chromatograms of complementary strands by using PreGap4 vl.5 and Gap4 v4.10 from the Staden Package (Bonfield et al. 1995). We used Clustal X 1.81 (Thompson et al. 1997, 1999) with the default settings to align sequences.

We analysed the combined sequences phylogenetically by using two approaches, maximum likelihood (ML) with PAUP 4.0b10 (Swofford 2003) and Bayesian inference (BI) with MrBayes 3.0b4 (Huelsenbeck and Ronquist 2001). We performed tree-space searches with the aid of PAUPRat (Sikes and Lewis 2001), which implements a ratchet subroutine for PAUP (Nixon 1999). For ML analyses, we performed 10 repetitions of the ratchet with 200 iterations per run.

We used ModelTest v. 3.06 (Posada and Crandall 1998) to determine the best-fit model of evolution and estimate parameters for ML and BI analyses. For ML analyses, we estimated nodal support by bootstrap analysis (Felsenstein 1985) with a full heuristic search, tree bisection, and the reconnection branch-swapping option. We estimated clade robustness in ML with 200 full heuristic bootstrap replications, performed with informative sites only.

For BI, four Markov chains, in two independent searches, were run for 7 × 10⁶ generations, sampling trees every 250 generations. Because 224 000 generations were required for the —In L scores to reach stability (burn-in period), we discarded these initial generations (849 trees). On the basis on the remaining trees, we used a majority-rule consensus of trees generated after the burn-in to estimate posterior probabilities for each node (Huelsenbeck et al. 2002). We considered P ≥ 95% to indicate strong branch support.

We estimated dates of key nodes in the phylogeny after computing the time to most recent common ancestor with the software BEAST (Drummond and Rambaut 2007). This program executes Bayesian Markov-chain analysis for molecular sequences to average over tree space, so that each tree is weighted proportionally to its posterior probability. We used the best-fit model obtained from Modeltest v. 3.06 (Posada and Crandall 1998) to define the pattern of nucleotide substitution. Because no fossil calibration is available for Melanerpes, we assumed a fixed sequence-divergence rate of 2% per million years (García-Moreno 2004), which has been used for other taxa (Weir and Schluter 2008) in spite of possible drawbacks pointed out by Lovette (2004) and Pereira and Baker (2006). The relaxed uncorrelated lognormal molecular-clock option (Drummond et al. 2006) was implemented. We obtained the posterior-probabilities distribution after running ten iterations of 10 000 000 generations each. We used the default values declared in BEAST for all other priors and operators.

Results

Sequence Variation

Alignment of the six gene fragments for the ingroup (the Red-bellied species group plus M. chrysogenys) resulted in a total of 872 bp of sequence data. Of these sites, 706 were conserved, 164 were variable, and 148 (17%) were phylogenetically informative. As expected, the overall number of transitions was higher than that of transversions; R = 6.9 (31 vs. 5). Two species of Melanerpes (M. cactorum and M. candidus) and the three outgroups had an extra nucleotide in position 174 of the ND3 gene. This nucleotide apparently represents a bulge in the sequence, detected also in other birds and vertebrates (Mindell et al. 1998). On the basis of Akaike's information criterion (AIC), we selected GTR + 1 + Γ as the model of evolution. Parameter estimates were ts/tv ratio 5.2, Γ distribution shape 1.4776, and proportion of invariable sites 0.5659. We used base-frequency values as estimated by the model: A = 0.2901, C = 0.3048, G = 0.1351, T = 0.2700.

Phylogenetic Analyses

The 89 sequences analyzed yielded 46 distinct haplotypes. Except for some poorly supported nodes close to the base of the tree, the two phylogenetic methods used converged on a single, largely congruent topology (Fig. 2). Relationships within the carolinus group according to the two trees were consistent. Maximum likelihood yielded a single tree (-In L = 4567.6). The ML results differed from the BI tree at two nodes. The analyses concurred in separating M. aurifrons haplotypes into two groups that are not sisters: northern populations were sister to the North American M. carolinus rather than to southern populations of M. aurifrons. The West Indian M. superciliaris was placed as sister to the M. carolinus—M. aurifrons complex, though not with clear branch support from either method. The following branches in the tree were M. uropygialis, then M. chrysogenys. Melanerpes hoffmanni did not appear to be part of the carolinus group.

Divergence time estimates obtained by BEAST place the split between the carolinus—aurifrons clade and the rest of the Melanerpes species we analyzed at about 5.4 mya (95% confidence limit 10.5 to 2.4 mya). The divergence between the clade containing tropical populations of M. aurifrons and the clade containing M. carolinus and northern M. aurifrons is dated at 1.02 mya (95% confidence limit 2.3 to 0.22 mya), whereas the separation between northern M. aurifrons and M. carolinus is estimated to have occurred 0.215 mya (95% confidence limit 0.55 to 0.04 mya).

Discussion

Monophyly of Melanerpes and Taxon Sampling

Our results support the monophyly of Melanerpes, as did Webb and Moore (2005), Benz et al. (2006), and Overton and Rhoads (2006). Those studies sampled the broader diversity of woodpeckers better but did not have the fuller representation within the genus that we assembled. Although we did not examine all species of Melanerpes or the possibly closely related Xiphidiopicus, we were able to compare several species of four genera that have at times been recognized within the current Melanerpes, viz., Centurus (e.g., M. carolinus), Tripsurus (e.g., M. pucherani), Leuconerpes (M. candidus), and Trichopicus (e.g., M. cactorum). Thus our study provides a first molecular view into the validity of merging these taxa. Unfortunately, our examination of Melanerpes suffered to some extent from the lack of the Red-crowned Woodpecker (M. rubricapillus) for comparison. This species, although not considered part of the carolinus group, is very similar in morphology and sometimes is considered conspecific with or as the sister species of M. pygmaeus (Peters 1948, Short 1982, Winkler and Christie 2002).

Monophyly of the Carolinus Group

Our results indicate that the superspecies M. carolinus, as defined by Selander and Giller (1963) and Short (1982), is paraphyletic. Moreover, this clade includes species not previously considered closely related (e.g., M. chrysogenys and probably M. pygmaeus). M. chrysogenys, endemic to the tropical dry forest of western Mexico (Winkler and Christie 2002), was previously considered a close relative of M. pucherani and M. hypopolius, but our analyses place it securely within the carolinus group as sister to a clade containing M. uropygialis, M. carolinus, and M. aurifrons (Fig. 2).

The monophyletic group composed of M. carolinus, M. aurifrons, and M. uropygialis we identified was previously suggested by Selander and Giller (1963). Beside similarities in morphology and coloration, these three species share several unique features, including call structure, a massive hyoid apparatus, naked orbital region, and orange subcutaneous fat. Our results differed from Selander and Giller's (1963), however, by including the large Caribbean species M. superciliaris within the group. Those authors had also suggested a close phylogenetic relationship of this species with the aurifrons— carolinus complex, but their idea that M. superciliaris is sister to M. carolinus is not supported by our results. More thorough analyses of M. superciliaris are needed and will require tissue samples from throughout the Caribbean.

Several authors have suggested that M. hoffmannii is a close relative of M. aurifrons (Short 1982, Selander and Giller 1963, Winkler and Christie 2002). Contrary to this idea, we found M. hoffmannii close to the base of the North American— Mesoamerican clade that resembles Short's (1982) carolinus group (superspecies M. carolinus). Previous hypotheses appear to have been based on ambiguous evidence from convergent morphology and parapatric distributions (Selander and Giller 1963) and on reports of hybridization of this species with M. a. santacruzi in Honduras (Monroe 1968).

Nonmonophyly of Melanerpes aurifrons

We did not recover Melanerpes aurifrons, as currently recognized (AOU 1998), as a monophyletic group. Rather, M. carolinus and samples of M. aurifrons from northeastern Mexico to the southern United States showed well-supported reciprocal monophyly. Tropical populations of M. aurifrons (eastern Mexico to Nicaragua) appeared as sister to a clade formed by the previous two lineages. This result was consistent across all of our analytical approaches, suggesting that M. carolinus is part of the aurifrons complex and that M. aurifrons (as currently recognized) comprises at least two distinct lineages. Similarities between northern M. aurifrons (sensu stricto) and M. carolinus have been noted previously in studies of genetics (Smith 1987), vocalizations (Skutch 1969), and morphology (Bent 1964).

Ecological requirements of M. carolinus and M. a. aurifrons are also similar. Melanerpes carolinus inhabits a variety of habitats, especially deciduous and other types of open forest (AOU 1998). The northern aurifrons is found in open areas, especially under arid and semiarid conditions, and the two taxa occur sympatrically in central Texas, where their sizes, nesting habits, and foraging behaviors are similar. This region has been recognized as the climatic break between the arid zones of Texas and the Mexican Plateau and the humid regions of the eastern United States and southeastern Canada (Smith 1987). In the contact zone, both species defend territories interspecifically and intraspecifically, with little evidence of interbreeding in the area of sympatry (Smith 1987). On the other hand, tropical populations of M. aurifrons range mainly in moist tropical lowlands from eastern Mexico to Nicaragua.

This differentiation of M. aurifrons into two clades also corresponds to marked morphological differences, specifically in plumage coloration and size of the northern and tropical populations. Birds belonging to the northern clade have a light gray breast and flanks (smoke gray of Smithe 1975), like those of M. carolinus, ayellow-to-orange nape separated from a red crown, yellow nasal tufts and belly, and broad white bars in the back, whereas the tropical populations belonging to the southern aurifrons clade (including M. a. dubius, M. a. santacruzi, and M. a. polygrammus) have the nape and crown uniform (completely red in dubius and santacruzi, red and orange in polygrammus) and dark breasts and flanks ranging from grayish horn color to brownish olive (Smithe 1975). The width of the bars on the back also differs from that of the northern clade but varies among tropical forms: very thin in dubius and santacruzi, medium in polygrammus. The nasal tufts and belly are red in dubius, yellow in polygrammus, and orange to red in santacruzi.

The subspecies polygrammus, restricted to the Pacific slope of the Isthmus of Tehuantepec in Oaxaca and southern Chiapas (Miller et al. 1957), is the most morphologically distinct taxon in the tropical clade of M. aurifrons Morphologically similar to northern M. aurifrons., polygrammus occurs in arid and semiarid habitats with vegetation similar in phenology to that on the central Mexican plateau (Gallardo-Cruz et al. 2005). Haplotypes from polygrammus are nonetheless intermingled broadly with those of other members of the tropical clade (e.g., within Oaxaca3 and Oaxacal haplotypes; Fig. 2, Table 2), thus not supporting Navarro-Sigüenza and Peterson's (2004) suggestion that polygrammus may represent a distinct species. Areas of contact between polygrammus and neighboring santacruzi have been documented in coastal Chiapas, a region that constitutes a zone of secondary contact for other species (e.g., Campylorhynchus rufinucha; Selander 1964, Vazquez-Miranda et al. 2009).

Other Findings

Our results set the stage for further phylogenetic studies of South American taxa. For example, they highlight interesting issues regarding the Mexican endemic M. hypopolius, which has been considered closely related to the South American M. cruentatus (Short 1982) or to M. uropygialis (Peters 1948). Although the phylogenetic position of M. hypopolius in our three analyses was consistent (close to the South American taxa we analyzed), expanded taxon sampling will be required for relationships among the remaining Central and South American forms of the genus to be resolved.

Biogeography. The woodpeckers of the M. carolinus group exhibit a complex array of distributional patterns, mainly allopatric but with some areas of parapatry. Our phylogeny suggests that speciation proceeded in a geographically coherent sequence (Fig. 3). Diversification of this group may have begun in western Middle America and continued through eastern Mexico, North America, and the Caribbean, finally isolating species in the deserts and North American deciduous forest from those in the tropical lowlands of Mesoamerica(Fig. 3).

Divergence times estimated in BEAST suggest that speciation within the M. carolinus—aurifrons—santacruzi clade may have occurred in association with the several glacial— interglacial cycles approximately every 100 000 years over the last million years (Paillard 1998). These events have had diverse and wide-ranging effects on bird habitats and species' distributions along altitudinal and latitudinal gradients (Pielou 1991, Johnson and Cicero 2004).

According to Hubbard (1973), biogeographic patterns of several bird lineages were molded during the Pleistocene by the alternating effects of these glacial—interglacial cycles restricting species to refugia, especially in arid regions, and by vicariance during the last two glacial and interglacial cycles, about 0.3 mya. However, according to Zink et al. (2000), and as suggested by our data, these speciations might be older than the last two glaciations.

Taxonomic implications. Our findings also suggest that a taxonomic revision of the Melanerpes aurifrons— carolinus complex is necessary. To reflect the phylogenetic pattern and recognize only natural taxa, we suggest splitting the present M. aurifrons into two species. M. aurifrons (Wagler, 1829), sister to and specifically distinct from M. carolinus (Linnaeus, 1758), encompasses only populations ranging from Texas and southern Oklahoma south through the Mexican Plateau to Zacatecas and Jalisco currently recognized as M. a. aurifrons. The other species, to which Melanerpes santacruzi (Bonaparte, 1838) applies as the oldest name, encompasses the tropical populations from southern San Luis Potosi and northeastern Querétaro south along the Atlantic slope of Mexico to Honduras and along the Pacific coast from eastern-most Oaxaca and Chiapas to north-central Nicaragua (Ridgway 1914, Navarro-Siguenza and Peterson 2004). Table 1 synthesizes the proposal and allocation of subspecies. Considering that several morphological features presented by these taxa are established and fixed (i.e., union/separation of nape and crown, width of white bars on the back, as well as DNA sequence variation), each species is diagnosable by a unique combination of characters (Wheeler and Platnick 2000). Further research on the phylogeny of Melanerpes involving all the species of the genus, and deeper analyses of the closely related taxa (e.g., M. supercilaris), will bring greater understanding of the evolution of this important clade of woodpeckers and elucidate its biogeographic history in the Americas.

Acknowledgments

We thank the following institutions and individuals for supplying tissue samples: American Museum of Natural History (AMNH; Paul Sweet), Field Museum of Natural History (David Willard), University of Nevada, Las Vegas (John Klicka), and Kansas University Natural History Museum (KUNHM; Mark Robbins). Invaluable help in the lab was obtained from Blanca Hernández, Nandadevi Cortés, Gabriela García-Deras, Magali Honey, Anahí Ávila (all Museo de Zoología, Facultad de Ciencias), and Laura Márquez (Instituto de Biología, Universidad Nacional Autónoma de México). Further help was obtained from Elisa Bonaccorso and Town Peterson during the stay of EAG-T at KUNHM, and from Esther Quintero and Paul Sweet at AMNH. We thank the following individuals for invaluable help in the field: Aldegundo Garza de León, Hernán Vázquez, Vicente Rodríguez, Luis Antonio Sánchez, Samuel López, Héctor Olguín, Roberto Sosa, and Oliver Komar. Comments on various versions of the manuscript were obtained from Townsend Peterson, Michael Patten, Philip Unitt, Luis A. Sánchez, Gabriela García Deras, Esther Quintero, Lynna Kiere, and two anonymous reviewers. We especially thank Fred Sheldon for valuable comments and for his great help in improving the English of the final version. Financial support for various stages of this project was obtained from the SEMARNAT-CONACyT Sectorial Funds (C01-0265), CONACyT (R27961), CONABIO (V009), DGAPA-UNAM (IN 208906), a collections study grant from the American Museum of Natural History, and a graduate studies scholarship from CONACyT and DGEP-UNAM to EAG-T.

Literature Cited