Biogeography, phylogeny, and morphological evolution of central Texas cave and spring salamanders

Abstract

Background: Subterranean faunal radiations can result in complex patterns of morphological divergence involving both convergent or parallel phenotypic evolution and cryptic species diversity. Salamanders of the genus Eurycea in central Texas provide a particularly challenging example with respect to phylogeny reconstruction, biogeography and taxonomy. These predominantly aquatic species inhabit karst limestone aquifers and spring outflows, and exhibit a wide range of morphological and genetic variation. We extensively sampled spring and cave populations of six Eurycea species within this group (eastern Blepsimolge clade), to reconstruct their phylogenetic and biogeographic history using mtDNA and examine patterns and origins of cave- and surface-associated morphological variation.

Results: Genetic divergence is generally low, and many populations share ancestral haplotypes and/or show evidence of introgression. This pattern likely indicates a recent radiation coupled with a complex history of intermittent connections within the aquatic karst system. Cave populations that exhibit the most extreme troglobitic morphologies show no or very low divergence from surface populations and are geographically interspersed among them, suggesting multiple instances of rapid, parallel phenotypic evolution. Morphological variation is diffuse among cave populations; this is in contrast to surface populations, which form a tight cluster in morphospace. Unexpectedly, our analyses reveal two distinct and previously unrecognized morphological groups encompassing multiple species that are not correlated with spring or cave habitat, phylogeny or geography, and may be due to developmental plasticity.

Conclusions: The evolutionary history of this group of spring- and cave-dwelling salamanders reflects patterns of intermittent isolation and gene flow influenced by complex hydrogeologic dynamics that are characteristic of karst regions. Shallow genetic divergences among several species, evidence of genetic exchange, and nested relationships across morphologically disparate cave and spring forms suggests that cave invasion was recent and many troglobitic morphologies arose independently. These patterns are consistent with an adaptive-shift hypothesis of divergence, which has been proposed to explain diversification in other karst fauna. While cave and surface forms often do not appear to be genetically isolated, morphological diversity within and among populations may be maintained by developmental plasticity, selection, or a combination thereof.

Figure 1

Fifty percent majority-rule consensus phylogram of eastern Blepsimolge based on Bayesian analysis. Posterior probabilities of node support greater than or equal to 95% are indicated by asterisks. Species designations (indicated by colored blocks) follow those given by [10], and we include all of the same populations from their molecular analysis plus numerous new samples. Hollow squares indicate topotypical specimens.

Figure 2

Geographic distribution of eastern Blepsimolge mtDNA clades in relation to species boundaries, habitat and major physiographic features. Squares and circles represent spring and cave localities, respectively. Clade associations are labeled by color for each sampled population; populations with haplotypes from two distinct clades are bicolored. Approximate distributions for Eurycea neotenes complex species (colored lines) are drawn according to designations by [10] although these designations are not entirely consistent with our phylogenetic hypotheses. Similarly, physiographic boundaries also appear to be poor predictors of mtDNA clade distributions. B9 = Bullis springs 9–83 and 9–29; CS = Comal Springs; PS = Pedernales Springs.

Figure 3

Diversity of head morphology and pigmentation within the eastern Blepsimolge. Parallel patterns of morphological evolution are evident in the troglomorphic specimens from clades LT, N and BGP, although all labeled clades contain surface forms (i.e., having fully-developed eyes and dark pigmentation). Localities for individuals pictured are as follows: 1 Honey Creek Cave, 2 Cascade Caverns, 3 Camp Bullis Cave #3, 4 Cascade Caverns, 5 Bullis Bat Cave, 6 Golden Fawn Cave, 7 Preserve Cave, 8 CM Cave, 9 Preserve Cave, 10 Hoffman Ranch Estavelle, 11 Fern Bank Spring, 12 Jacob’s Well, 13 Hector Hole, 14 Lewis Valley Cave, 15 Sharon Spring, 16 Morales Spring, 17 Taylor Springs.

Figure 4

Scatterplots of factor scores from principal components analysis (PCA) of log₁₀-transformed measurements. 4a–c: Ordination of specimens of the eastern Blepsimolge (circles) and Typhlomolge (squares; includes E. rathbuni and E. waterlooensis) clades are shown. Closed and open symbols represent specimens collected from cave and surface localities, respectively. 4d: Ordination of morphological variation within and among cave populations of central Texas Eurycea (PC1 vs. PC2). Populations with N > 3 are shown with colored convex hull polygons (individual specimens removed except for outliers). Light gray circles indicate surface specimens; dark gray circles indicate cave specimens from localities with N ≤ 3.

Figure 5

Congruent patterns of morphological variation for selected surface (a) and cave (b) populations. Localities are shown where specimens cluster in either of two groups or where specimens cluster in both groups. Light gray circles and squares indicate Blepsimolge and Typhlomolge, respectively.

Figure 6

Cave (top) and surface (bottom) morphs of Eurycea from Honey Creek Cave (type locality for Eurycea tridentifera). These individuals were observed only meters apart within the same cave stream: the surface morph was encountered 5 m into the cave while the cave morph was observed approximately 20 m from the entrance.