Tropical bolitoglossines in MesoAmerica show the widespread hump-shaped pattern of elevational species richness (), with the largest number of species at intermediate elevations and fewer species at the lowest and highest elevations. Almost all of the observed species richness values are outside the upper and lower 95% CIs expected under a null model of stochastic placement of elevational range midpoints, and the fit between the expected and observed values is not significant (r2
=0.155). The elevational ranges of species and clades appear to be more concentrated at mid-elevations than expected under the mid-domain model (Colwell & Hurtt 1994
Figure 1 Hump-shaped pattern in species richness of tropical bolitoglossine salamanders (supergenus Bolitoglossa) in MesoAmerica. Thick grey lines indicate the upper and lower 95% CIs for the number of species at each 500m elevational band expected from (more ...)
Combined, partitioned Bayesian analysis of two mitochondrial genes (cytochrome b
and 16S) reveals strong support for many relationships within bolitoglossines (). Contrary to hypotheses based on morphology (Elias & Wake 1983
does not appear to be the sister taxon of all other tropical bolitoglossines. Instead, Nyctanolis
is the sister taxon of Dendrotriton
, and these two clades are the sister group of a clade containing Nototriton
. There is strong support for a clade containing the species-rich genera Bolitoglossa
, along with several small genera nested inside of Pseudoeurycea
(Lineatriton, Ixalotriton, Parvimolge
). There is also strong support for placing Chiropterotriton
as sister taxon of this large clade, and moderate support for placing Thorius
as sister taxon of that clade. Cryptotriton
is weakly placed as sister taxon of all other tropical bolitoglossines.
Figure 2 Estimated phylogeny and divergence times for tropical bolitoglossine salamanders (outgroup taxa not shown). Numbers adjacent to clades indicate Bayesian posterior probabilities (×100), with asterisks showing clades with Pp0.95. Results (more ...)
A recent analysis of amphibian phylogeny (Frost et al. 2006
) included 10 out of 13 tropical bolitoglossine genera. Their tree concurs with ours in the placement of Thorius
as sister group to a clade including Bolitoglossa
and relatives (Ixalotriton, Lineatriton
). Our hypotheses also agree in placing Oedipina, Nototriton
in a basal clade. Our trees differ only in the placement of Parvimolge
versus with Bolitoglossa
) and Cryptotriton
(sister taxon of all other tropical bolitoglossines versus sister taxon of Dendrotriton
). However, their analysis included only a single representative of each genus, and excluded three genera.
Mapping elevational distributions onto our phylogeny () using generalized least squares shows that mid-elevations (approx. 1000–2000
m) were colonized relatively early in the history of bolitoglossines, and that there have been subsequent invasions of low elevation habitats (in Oedipina
). Major invasions of the highest elevations have occurred within Pseudoeurycea,
but some other taxa (e.g. Bolitoglossa)
also extend into very high elevations. The Bayesian reconstruction for the ancestor of tropical bolitoglossines also suggests that mid-elevations were colonized early, and lower and higher elevations subsequently (Pp
=0.931 for elevational band from 1201 to 1800
m), and shows that this general result is robust to an alternate reconstruction method and to uncertainty in the topology and branch lengths.
Using 19 clades that together include all known tropical bolitoglossine species, we find no relationship between rates of diversification of clades and elevational distribution of clades, regardless of whether we analyse the raw data (a) or phylogenetically independent contrasts (b). Results also are similar (no significant relationship) using different ages for Bolitoglossinae, using equal branch lengths for independent contrasts, and assuming a high extinction rate when estimating diversification rates (results not shown). Note that one clade (adspersa group of Bolitoglossa) has an unusually high diversification rate, but this clade is found primarily in South America (26 out of 31 species) and has relatively little impact on overall species richness patterns in MesoAmerica (deleting this clade gives similar results; r2=0.005, p=0.780). Data on species numbers, estimated age (including assessment of confidence), diversification rate and elevational distribution of each clade are provided in electronic supplementary material 3.
Figure 3 Rate of diversification of clades versus average elevational midpoint of clades for tropical bolitoglossine salamanders, using (a) raw data and (b) phylogenetically independent contrasts (using branch lengths from ). The results shown assume a (more ...)
We identified four branches on which there are significant shifts in diversification rates, based on the 138 sampled species (). However, changes in elevational ranges on these branches are not significantly different from those on other branches (mean difference=−5.478, d.f.=270, p=0.9721), further confirming that there is no significant association between changes in elevational range and diversification rate.
We found a strong relationship between the estimated timing of colonization of elevational bands and the present species richness of these bands, regardless of whether we use the age of the first colonization of each band (a) or the summed ages of each colonization (b). Results also are similar using the lowest elevation of occurrence of each species rather than the midpoint of their elevational ranges (c,d).
Figure 4 The species richness of 500m elevational bands versus the estimated date (Myr ago) when the elevational band was first colonized, for tropical bolitoglossine salamanders in MesoAmerica, using (a,c) the oldest colonization of each elevational (more ...)