Sexual selection drives speciation in an Amazonian frog

Abstract

One proposed mechanism of speciation is divergent sexual selection, whereby divergence in female preferences and male signals results in behavioural isolation. Despite the appeal of this hypothesis, evidence for it remains inconclusive. Here, we present several lines of evidence that sexual selection is driving behavioural isolation and speciation among populations of an Amazonian frog (Physalaemus petersi). First, sexual selection has promoted divergence in male mating calls and female preferences for calls between neighbouring populations, resulting in strong behavioural isolation. Second, phylogenetic analysis indicates that populations have become fixed for alternative call types several times throughout the species' range, and coalescent analysis rejects genetic drift as a cause for this pattern, suggesting that this divergence is due to selection. Finally, gene flow estimated with microsatellite loci is an average of 30 times lower between populations with different call types than between populations separated by a similar geographical distance with the same call type, demonstrating genetic divergence and incipient speciation. Taken together, these data provide strong evidence that sexual selection is driving behavioural isolation and speciation, supporting sexual selection as a cause for speciation in the wild.

Figure 1

(a) Spectrograms of simple calls of Physalaemus petersi from La Selva and Yasuní, and a complex call from Yasuní. Simple calls consist of one component, the whine (electronic supplementary material, audio files 1–3). Complex calls have an additional second component, the ‘squawk’. (b) Map of study sites. Open circles represent sites in which males only produce simple calls and solid circles are sites in which males produce complex calls. Rivers are shown with solid lines and country borders with dashed lines. Elevation is represented by shading. (c) Maximum-likelihood tree inferred from the analysis of approximately 2.4 kb of the mtDNA 12S and 16S genes and the intervening valine tRNA. Numbers shown by nodes are Bayesian posterior probabilities from 180 000 sampled trees (asterisks indicate posterior probabilities of 100%). Open bars indicate sites in which males only produce simple calls and solid bars indicate sites in which males produce complex calls. Outgroups are not shown.

Figure 2

Hypothesis for how sexual selection for complex calls at Yasuní incidentally drove behavioural isolation between La Selva and Yasuní Physalaemus petersi. (a) In phonotaxis experiments, Yasuní females strongly preferred calls with squawks to those without squawks, whereas La Selva females did not. The numbers of females that responded to each call stimulus are shown next to arrows pointing to the stimuli. Asterisks indicate significant preferences. Males from Yasuní and other populations with complex calls have larger body sizes and proportionally larger larynges (ventral views of larynges shown in (b)) and larger fibrous masses than males from La Selva and other populations that only produce simple calls, suggesting that selection for squawks has caused an increase in the size of these morphological features. Larger body and larynx size is the probable morphological basis for the lower frequency whines at Yasuní and other populations with complex calls. (c) In a separate set of phonotaxis experiments, La Selva females strongly preferred simple calls of La Selva males to those of Yasuní males, and Yasuní females unanimously preferred simple calls of Yasuní males to those of La Selva males. These strong local mate preferences are probably generated by the large and significant differences in the dominant frequencies of whines.