Introduction

The genus Bolitoglossa, with 117 described species (AmphibiaWeb 2011), is by far the largest genus within the order Caudata. It has the widest range of any tropical salamander genus, from the lowlands of southern Tamaulipas, Mexico to Brazil and Bolivia in South America. The monophyly of Bolitoglossa is well supported on molecular (Parra-Olea et al. 2004) as well as morphological grounds. The lack of a sublingual fold, short ceratohyals, partially or fully webbed feet, and fused distal tarsal 4 and 5 characterize all the species of the genus (Wake 1966; Wake and Elias 1983).

Parra-Olea et al. (2004) used mtDNA sequence data analyzed phylogenetically to subdivide Bolitoglossa into seven subgenera: Bolitoglossa, Eladinea, Magnadigita, Mayamandra, Nanotriton, Oaxakia, and Pachymandra. The subgenus Nanotriton comprises species previously included in the Bolitoglossa rufescens group: Bolitoglossa occidentalis Taylor 1941, Bolitoglossa rufescens Cope 1869, and the recently described Bolitoglossa nympha Campbell et al. 2010. The species of Nanotriton are small, short-tailed salamanders with small pad-like hands and weakly developed feet, all associated with paedomorphosis (Alberch 1983). These species occur in habitats ranging from sea-level lowland forests, to humid cloud forests up to 2000 m in elevation.

In this paper we describe a new species of the subgenus Nanotriton from the Sierra de Juárez and Sierra Mixe, Oaxaca, based on morphological differences from described species and DNA sequence differences from sympatric Bolitoglossa rufescens and the other two species in the subgenus. The new species is assigned to Bolitoglossa (Nanotriton) (Parra-Olea et al. 2004) based on its relatively small body size, fully webbed, pad-like feet with little digital individuation, short tail, overall morphological similarity to other species in the subgenus, and phylogenetic placement with mitochondrial DNA sequence data.

Materials and methods

External morphology was examined in 17 populations of all known species of Bolitoglossa (Nanotriton) (Table 1). These specimens represent most of the geographic range of the subgenus from the Atlantic coast of Mexico (Veracruz) to western Honduras. We took 14 measurements that reflect size and proportional shape of the salamanders: distance from snout to posterior end of vent (SVL), tail length (TL), snout to gular fold length (SG), head width at angle of jaw (HW), axilla-groin length (AG), forelimb length (FLL), hind limb length (HLL), shoulder width (SW), right foot width (RFW), head depth (HD), interorbital width (IO), internarial width (IN), tip of snout to anterior corner of eye (rostrum length, RL), diameter of eye opening (ED). Measurements were taken to the nearest 0.1 mm using vernier calipers. We also counted the number of costal grooves separating adpressed limbs (limb interval, LI). We counted total numbers of ankylosed vomerine (VT), premaxillary (PMT) and maxillary teeth (MT) under a dissecting microscope.

Species identifications of each population were based on geography and the allozyme results of Larson (1983). Bolitoglossa nympha, recently described from the Sierra de Caral on the Guatemala-Honduras border, is currently known only from the type locality (Campbell et al. 2010). The diagnosis for Bolitoglossa nympha relies largely on molecular characteristics, which are not currently available for most populations of the Bolitoglossa rufescens complex. In Larson’s 1983 allozyme study, populations from Finca El Volcán, Guatemala, the San Pedro Sula area, Honduras and Santa Rosa del Copán, Honduras, which are the geographically closest populations included in our study to the type locality of Bolitoglossa nympha, cluster together in a distance-based phylogeny. For this reason, we tentatively treat these three populations as belonging to Bolitoglossa nympha. Definitive species identification of populations from eastern Guatemala and Honduras will require detailed molecular study of the complex.

Statistical analyses were run with the program JMP 8 (SAS Institute, Cary, NC, USA). Wilcoxon tests were used to test for differences between group means for selected variables. In order to test for sexual dimorphism within species, variables were tested for normality within each species using the Shapiro-Wilk test. For variables whose distribution did not differ significantly from the normal distribution, a t-test was used to compare the mean for males and females. A Wilcoxon test was used to compare means for males and females of each species for variables whose distributions differed significantly from normal. Significant differences in means between males and females of the same species were found for nearly all variables (see results), indicating that sexual dimorphism exists within these species, so all further analyses were performed separately for males and females.

Multivariate statistical analyses used all variables except LI, MT, and VT, which were not measured in the same units (mm) as the other variables. In order to remove the effect of body size, each variable was regressed against SVL, and the residuals from a linear fit with SVL were used in further analyses. Separate linear fits were used for males and females. Normality of the residuals for each variable was tested using the Shapiro-Wilk test. Given that Discriminant Function Analysis (DFA) can still be used when the assumption of multivariate normality is violated, particularly when the percent of correct classification is high (Klecka 1980), we performed a DFA using residuals from SVL of all variables except TL, which was missing for several individuals of the new species that had missing or regenerated tails.

Because one group (the new species) has a smaller sample size than the total number of variables measured (8 females, 6 males), a Principal Components Analysis (PCA) was used to reduce the dimensionality of the data. The PCA was performed using residuals of the following variables, which passed the normality test: males: SG, FLL, HLL, IO, HW, SW, RFW, ED; females: FLL, HLL, RL, IO, RFW, ED. A second DFA was then performed on the first three principal components.

Although a full molecular analysis of the subgenus Nanotriton is beyond the scope of the present work, several mitochondrial sequences were generated in order to compare the new species to other members of its subgenus. We sequenced a specimen (paratype) of the undescribed species (IBH 22535), as well as a specimen of Bolitoglossa rufescens (IBH 22536) from the same locality and an individual of Bolitoglossa occidentalis (MVZ 194259) for the 16S rRNA (16S, 518 bp) and cytochrome b (cyt b, 809 bp) mitochondrial genes using primers MVZ117 and MVZ98 (Palumbi 1996) for 16S and primers MVZ15 and MVZ16 (Moritz et al. 1992) for cytb. Reactions were run at 94 °C for 2 min, 38 cycles of 94 °C for 30 s, 48 °C for 30 s (16S) or 1 min (cytb), 72 °C for 1 min, with a final cycle at 72 °C for 8 min. We aligned these sequences with available sequences for Bolitoglossa rufescens (MVZ 194254) and Bolitoglossa nympha (MVZ 194333) from GenBank using the program MUSCLE 3.6 (Edgar 2004) and concatenated alignments for 16S and cytb. We trimmed cytb sequences to a length of 645 bp to match those from GenBank. Individuals of Bolitoglossa mexicana (MVZ176838) and Bolitoglossa hartwegi (MVZ 263458) were used as outgroups for phylogenetic analysis. We used the program RAxML (Stamatakis 2006) to estimate a phylogeny with maximum likelihood under the GTR+G substitution model in order to determine the relationship of the new Sierra de Juárez species to other members of the subgenus Nanotriton. The data were partioned by gene (16S and cytb), and the cytochrome b gene was partitioned by codon position. One thousand bootstrap replicates were performed to assess nodal support. Pairwise distances between species were calculated using PAUP* v4.0 (Swofford 2003). Additionally, we compared the genetic distance between an individual of the new species from the Sierra Mixe (MZFC 16085), for which only a 16S sequence was available in GenBank, to our sample from the Sierra de Juárez.

Results

Means, standard deviations, and ranges of all measurements and tooth counts are given in Table 2. Our new species showed significant sexual dimorphism for only two variables (HLL, RL), although this result may be partially due to small sample size. Males and females of the other species had significantly different means for the following variables: Bolitoglossa nympha – SVL, SG, FLL, HLL, RL, AG, IO, IN, HW, SW, RFW, HD, and VT; Bolitoglossa occidentalis – SVL, AG, IO, SW, MT; Bolitoglossa rufescens – AG, IO, IN, HW, SW, HD.

The Discriminant Function Analysis (DFA) for females using three principal components constructed from residuals from SVL of variables that passed normality tests correctly classified all but one individual of the new species (Table 3). Five of 14 individuals of Bolitoglossa nympha were misclassified as Bolitoglossa occidentalis, 23 of 46 individuals of Bolitoglossa occidentalis were misclassified (12 as Bolitoglossa nympha and 11 as Bolitoglossa rufescens), and 19 of 36 individuals of Bolitoglossa rufescens were misclassified (1 as Bolitoglossa chinanteca, 8 as Bolitoglossa nympha, 10 as Bolitoglossa occidentalis). For males, four of five individuals of the new species were classified correctly while many individuals of the other species were misclassified (Table 3). The number of individuals classified per group differs from the total number of individuals per group because some individuals lack data for measurements such as TL, and the DFA classifies only individuals with data for all included variables. Using residuals from SVL of all variables except TL, misclassification rates were lower (Table 4). For females, all individuals ofthe new species were classified correctly, and only one individual of Bolitoglossa nympha was misclassified. For males, all individuals of the new species were classified correctly, while misclassification was higher for the other species (Table 4).

The maximum likelihood mitochondrial gene tree places Bolitoglossa chinanteca as the sister taxon of Bolitoglossa occidentalis with strong support (BS=99) (Fig. 2). The GTR distance between individuals of the new species (IBH 22535) and Bolitoglossa rufescens from the Sierra de Juárez is 0.08 for 16S and 0.21 for cyt b. The two samples of Bolitoglossa chinanteca from the type locality and the Sierra Mixe have a GTR distance of only 0.004 for 16S.

Based on the correct classification of nearly all individuals of the new species from the Sierra de Juárez and Sierra Mixe, as well as several differences in external morphology and tooth counts from other species of Bolitoglossa (Nanotriton) and differences in mtDNA sequence data, these individuals represent an undescribed species. Type specimens are deposited in the Colección Nacional de Anfibios y Reptiles, Instituto de Biología, Universidad Nacional Autónoma de México (IBH).

Discussion

The Bolitoglossa rufescens group (subgenus Nanotriton, following Parra-Olea et al., 2004) long included only two species (Bolitoglossa rufescens and Bolitoglossa occidentalis) (the taxon Bolitoglossa bilineata was synonymized with Bolitoglossa occidentalis by Wake and Brame (1969)). Populations assigned to Bolitoglossa rufescens and Bolitoglossa occidentalis showed high levels of genetic divergence from one another (Larson 1983). The diminutive body size of these animals, coupled with variation in traits considered to be diagnostic for species (such as the presence or absence of maxillary teeth (Larson 1983)), hindered the taxonomic recognition of additional species within the complex. While Bolitoglossa occidentalis, Bolitoglossa rufescens, and Bolitoglossa nympha strongly resemble each other in overall external morphology (as evidenced by high misclassification rates of these species in the DFA), Bolitoglossa chinanteca is easily distinguished from these three species by its more robust body, as well as by the combination of characters given above. The occurrence of Bolitoglossa chinanteca and Bolitoglossa rufescens at the type locality of Bolitoglossa chinanteca is the second demonstrated instance of sympatry between two members of the subgenus Nanotriton, which further strengthens the case for the recognition of Bolitoglossa chinanteca as a distinct species. Poglayen and Smith (1958) reported Bolitoglossa occidentalis and Bolitoglossa rufescens from 10 km N San Fernando, Chiapas in the Atlantic drainage, and Larson (1983) showed very close geographic proximity between Bolitoglossa rufescens and Bolitoglossa occidentalis in the vicinity of Berriozabal, Chiapas.

No information is currently available on the population size or status of Bolitoglossa chinanteca, although individuals were found at the type locality on two recent visits. Although the distribution of Bolitoglossa chinanteca is not known precisely, a polygon drawn between the three known localities has an area of approximately 255 km². This extent of occurrence, coupled with a decline in extent of occurrence due to habitat destruction, would classify Bolitoglossa chinanteca as Endangered under IUCN Red List Criterion B1ab(i) (B1. Extent of occurrence < 5000 km², a. known from <5 localities, b(i). continuing decline in extent of occurrence). The fact that Bolitoglossa chinanteca has been taken in banana trees in disturbed habitat, however, suggests that it may tolerate disturbance reasonably well. At this time, it does not appear that Bolitoglossa chinanteca qualifies for any of the threatened IUCN categories (CR, EN, VU). This assessment could change if evidence arises that it cannot live away from forest (the banana trees at the type locality are on the forest edge) or that habitat destruction in the region is adversely affecting the species. Because of this, we believe that Bolitoglossa chinanteca should be classified as Near Threatened (NT).

The Sierra de Juárez is among the areas of highest species richness for Neotropical salamanders, and morphologically distinct species continue to be described from the region (Hanken and Wake 2001; Parra-Olea et al. 2005) despite decades of taxonomic study of its salamanders (Wake et al. 1992). Not including the nearby Sierra Aloapaneca, the Sierra de Juárez was previously known to contain 18 salamander species of five genera (Bolitoglossa, Chiropterotriton, Cryptotriton, Pseudoeurycea, and Thorius) (Hanken and Wake 2001; Parra-Olea et al. 2005; Wake et al. 1992); the description of Bolitoglossa chinanteca brings the total number to 19. Such a high diversity of salamanders is notable even for Mexico, and highlights the need for continued taxonomic study of the salamanders of the region and of southern Mexico in general.