Diversification and reproductive isolation: cryptic species in the only New World high-duty cycle bat, Pteronotus parnellii

Abstract

Background: Molecular techniques are increasingly employed to recognize the presence of cryptic species, even among commonly observed taxa. Previous studies have demonstrated that bats using high-duty cycle echolocation may be more likely to speciate quickly. Pteronotus parnellii is a widespread Neotropical bat and the only New World species to use high-duty cycle echolocation, a trait otherwise restricted to Old World taxa. Here we analyze morphological and acoustic variation and genetic divergence at the mitochondrial COI gene, the 7th intron region of the y-linked Dby gene and the nuclear recombination-activating gene 2, and provide extensive evidence that P. parnellii is actually a cryptic species complex.

Results: Central American populations form a single species while three additional species exist in northern South America: one in Venezuela, Trinidad and western Guyana and two occupying sympatric ranges in Guyana and Suriname. Reproductive isolation appears nearly complete (only one potential hybrid individual found). The complex likely arose within the last ~6 million years with all taxa diverging quickly within the last ~1-2 million years, following a pattern consistent with the geological history of Central and northern South America. Significant variation in cranial measures and forearm length exists between three of the four groups, although no individual morphological character can discriminate these in the field. Acoustic analysis reveals small differences (5-10 kHz) in echolocation calls between allopatric cryptic taxa that are unlikely to provide access to different prey resources but are consistent with divergence by drift in allopatric species or through selection for social recognition.

Conclusions: This unique approach, considering morphological, acoustic and multi-locus genetic information inherited maternally, paternally and bi-parentally, provides strong support to conclusions about the cessation of gene flow and degree of reproductive isolation of these cryptic species.

Figure 1

Eleven cranial measures used in morphometric analysis of Pteronotus parnellii following Smith [[53]]. The characters are: condylobasal length (A-A), zygomatic breadth (B-B), breadth of braincase (C-C), mastoid breadth (D-D), zygorostral length (E-E), interorbital breadth (F-F), rostral breadth (G-G), alveolar length of maxillary toothrow (H-H), alveolar length of mandibular toothrow (I-I), breadth of post-palatal extension (J-J), depth of braincase (K-K).

Figure 2

Bayesian (A) and maximum likelihood (B) phylogenetic reconstruction of unique haplotypes of the 5’ region of COI mitochondrial lineages for the bat species Pteronotus parnellii . Branch supports represent posterior probabilities and non-parametric Shimodaira-Hasegawa-like (SH-like) values respectively. Both analyses were performed on a reduced (unique haplotypes only) dataset. Four haplotype networks (C) correspond to the four major lineages in the phylogenetic constructions. Each circle in the network represents a single haplotype with circle size scaled by haplotype frequency. Squares indicate the most common haplotype in the network.

Figure 3

Distribution of sampling sites for genetic groups. Group 1 is restricted to Central America. Groups 3 and 4 occupy sympatric distributions in the lowlands of the Guyana Shield while the few individuals in Group 2 occurred at higher elevations in Guyana and in Venezuela and Trinidad. Measures of the constant frequency component of most energy for echolocation calls are indicated. Photographs by E.L. Clare and M.B. Fenton.

Figure 4

Phylogenetic reconstruction of the combined COI, Dby 7^th intron and RAG2 regions supports a single topology for the bat speciesPteronotus parnellii. Branch supports represent posterior probabilities followed by non-parametric Shimodaira-Hasegawa-like (SH-like) values respectively.

Figure 5

Variable and fixed characters in A) the Dby 7^th intron and B) the RAG2 regions which differentiate mitochondrial lineages. Base pair references are given above the sequences, ~ indicates removed sequence positions which contain no polymorphisms. Actual sample size for each group is indicated next to example sequences.

Figure 6

Bayesian estimates of divergence time using two fixed substitution rates of 2% per million years (A) and 5% per million years (B) using phylogenetic reconstructions of unique COI haplotypes and the topology supported by the multi-gene reconstruction. Estimated divergence dates (MYBP) are indicated, grey bars are scaled to represent 95% confidence intervals for estimated dates.

Figure 7

Canonical discriminant function analysis (DFA) of genetic groups of Pteronotus parnellii shows significant divergence. (A) Sympatric groups 3 and 4 show greater separation when the data are corrected for correlations with latitude. (B) Group centroids are marked with a star.