The nature of parallel evolution

Abstract

When organisms face similar ecological conditions, they often evolve similar phenotypic solutions. When this occurs independently and repeatedly in closely related taxa, it is referred to as parallel evolution. The correlation that arises between phenotype and environment in systems of parallelism provides some of the most compelling evidence for the role of natural selection in driving evolution. This is because it is unlikely that similar phenotypes would have evolved purely by chance. Systems of parallel evolution can thus be viewed as natural replicates of the evolutionary process, enabling researchers to study the genetics and ecology of how adaptation proceeds in nature. However, we still lack a comprehensive understanding of the processes that govern parallel divergence within natural systems, including how demographic history affects the likelihood of parallel evolution, how the extent of genotypic parallelism manifests at the level of the phenotype, and how divergent selection operates across multiple instances of repeated evolution. Here, I use the Australian wildflower species complex, Senecio lautus, as an emerging model system to study the dynamics of repeated evolution within nature. I focus on the Dune and Headland ecotypes of S. lautus, which inhabit contrasting environments along the coast of Australia and often occur in parapatry.I first provide an in-depth discussion of the field of parallel evolution in Chapter I. In Chapter II I demonstrate that S. lautus Dune and Headland populations have evolved multiple times independently. Using genotyping-by-sequencing data I observed phylogenetic clustering by geography, with strong genetic structure between populations, isolation by distance, and surprisingly minimal gene flow between parapatric pairs, as well as in the system as a whole. I further confirmed this analytically by demonstrating that gene flow is not high enough in S. lautus to obscure the observed phylogenetic relationships between populations. My work also highlights the necessity for researchers of parallel evolution to directly demonstrate that populations with similar adaptive phenotypes have evolved multiple independent times, which is surprisingly lacking in studies of parallel evolution. Next, in Chapter III I further contribute to the field of parallel evolution by creating a framework to quantify both genotypic and phenotypic parallelism within empirical systems. My framework is applicable to non-model organisms and provides a common set of analyses to measure parallel evolution. Using this framework, I demonstrate that S. lautus populations inhabiting similar environments have evolved strikingly similar phenotypes. However, these phenotypes have arisen via different mutational changes in different genes, although most contribute to the same biological functions. This suggests that adaptation within S. lautus might be constrained at the level of the biological function, where replicate populations have recruited different mutations in different genes. In Chapter IV I characterise the genomic landscape of parallel divergence across replicate populations within the system. I observed that genomic landscapes were highly heterogeneous, and populations that were more closely related contained landscapes that were more similar, which is consistent with the random accumulation of changes due to genetic drift. Surprisingly, divergent loci did not reside within regions of reduced recombination, and there was no evidence for a pervasive role of linked selection in shaping the observed genomic landscapes within S. lautus. I also identified a highly parallel region that is a candidate barrier locus thought to be resistant to gene flow and involved in the initial stages of population divergence. Finally in Chapter V, I summarise the findings of my three data chapters and provide future directions for the fields of parallel evolution and population genetics.Overall, my work has furthered our understanding of natural selection in the repeated evolution of similar phenotypes within nature. My results pave the way for identifying the causal genes for adaptation by directly linking them to adaptive phenotypes and further demonstrating that these phenotypes confer a fitness advantage within natural populations. Ultimately, my work allows us to begin to understand whether evolution within nature is repeatable and predictable.