Abstract

Convergent evolution, where independent lineages evolve similar traits when adapting to similar habitats, is a common phenomenon and testament to the repeatability of evolutionary processes. Still, non-convergence is also common, and a major question is whether apparently idiosyncratic, lineage-specific evolutionary changes are reflective of chance events inherent to evolutionary processes, or whether they are also influenced by deterministic genetic or ecological factors. To address this question, we quantified the degree of convergence in genome-wide patterns of gene expression across lineages of livebearing fishes (family Poeciliidae) that span 40 million years of evolution and have colonized extreme environments in the form of toxic, hydrogen-sulfide-rich springs. We specifically asked whether the degree of convergence across lineage pairs was related to their phylogenetic relatedness or the ecological similarity of the habitats they inhabit. Using phylogenetic comparative analyses, we showed that the degree of convergence was highly variable across lineage pairs residing in sulfide springs. While closely related lineages did not exhibit higher degrees of convergence than distantly related ones, we uncovered a strong relationship between degree of convergence and ecological similarity. Our results indicate that variation in the degree of convergence is not merely noise associated with evolutionary contingency. Rather, cryptic environmental variation that is frequently ignored when we employ reductionist approaches can significantly contribute to adaptive evolution. This study highlights the importance of multivariate approaches that capture the complexities of both selective regimes and organismal design when assessing the roles of determinism and contingency in evolution. Significance StatementWhen different species adapt to similar environmental conditions, we frequently observe a mix between shared (convergent) and lineage-specific (nonconvergent) evolutionary changes. Shared changes provide evidence for the repeatability and predictability of evolution. However, it remains unclear whether lineage-specific changes are caused by random forces that limit the predictability of evolution, or whether they reflect deterministic processes shaped by unidentified genetic and environmental factors. By analyzing patterns of gene expression across fishes in extreme environments, we show that the degree of convergence between lineages is related to ecology, indicating that lineage-specific evolutionary changes are not just noise caused by random processes. Thus, acknowledging the complexity of nature in empirical research is critical if we want to predict evolution.