Inferring Histories of Adaptive Divergence with Gene Flow: Genetic, Demographic and Geographic Effects.

Abstract

As genomic data is increasingly available even for non-model organisms, the traditional boundaries among fields such as phylogenetics, phylogeography and genetics of adaptation are disappearing. This thesis provides a synthetic framework for studying ecological genomics, which considers selective processes (such as adaptation to new niches) and neutral processes (such as population size changes due to environmental shifts) simultaneously. Conventionally, studies that look for targets of selection on a genome assume a simple demographic model without validations from the species' ecological or phylogeographic histories. The work demonstrates that one cannot reliably identify selection unless realistic demographic histories are inferred for the species or even a specific genomic region. In particular, I investigate the evolutionary history of large polymorphic inversions in Anopheles gambiae, which maintains adaptive divergence among ecologically divergent populations. By modeling the unique origin and introgression histories of each inversion, I am able to identify target regions of selection within inversions through training discriminant functions with pure drift versus selection simulations. The thesis also extends the existing theory of local adaptation model via chromosomal inversions to consider the source of inversion variation, as well as evaluates the likelihood of such adaptations under different parameter spaces. The findings are particularly important for understanding mosaic genomic evolution in the early stages of speciation, where accumulation of divergence is dampened by gene flow. Finally, I examine how historical events, such as habitat contractions or recolonization, influence current genetic pattern and the application of spatially-explicit demographic modeling under Approximate Bayesian Computation statistics to distinguish different phylogeographic scenarios. The work represents a flexible framework for researchers to translate dynamic phylogeographic hypotheses into testable coalescent models by integrating all the available information of the species, such as distribution records, habitat preference, paleo-climate models, and competition between species. In general, with the amount of information as well as inherent heterogeneity of genomic data, this thesis contributes to the ongoing paradigm shift from studying separate evolutionary processes towards a holistic analysis of the interactions of selective and neutral processes under a rigorous statistical framework.