Evolutionary Interactions Between Sexual Selection, Song Elaboration, and Song Learning: A Combined Empirical and Theoretical Approach

Abstract

Song is a key aspect of oscine songbird behavior. Males that do not sing a species-typical song are unlikely to attract mates. Interestingly, young songbirds must listen to other conspecific males to learn an appropriate adult song. The time spent learning varies widely across species, with some limited to a set developmental window, and others learning throughout life. Furthermore, the degree of song elaboration across species ranges from chipping sparrows, which only know one syllable, to nightingales, which know over a thousand syllables. Of note, females prefer males with larger syllable repertoires over males with smaller syllable repertoires in some species. Finally, there is some evidence that longer learning may be correlated with larger species-level syllable repertoire sizes. Taken together, this suggests that syllable repertoire size, female preferences, and the length of the song-learning window interact to affect the evolutionary dynamics of one other. However, little empirical research has been done to address this possibility directly and there were no computational models developed to examine how these factors interact on an evolutionary timescale. For my dissertation, I mined the literature for published field data. I performed a meta-analysis, which showed that species with larger average syllables repertoire sizes exhibit stronger correlations between individual song elaboration and reproductive success (Chapter 1). The length of the song-learning window did not predict the strength of this correlation (Chapter 1). Additionally, I found evidence for correlated evolution between the length of the song learning window and two metrics of song complexity, syllable and song repertoire size (Chapter 2). Finally, I developed an agent-based model, and tested how different male song-learning strategies interacted with female preferences to affect syllable repertoire size. Of note, this model predicts that different learning strategies are more advantageous in different selection contexts and that longer learning can be lost during evolution even if it is not tied to a survival cost (Chapter 3). Overall, my dissertation provided possible answers to long-standing controversies and proposed novel hypotheses that can now be tested in the field.