Genetic interactions in evolutionary processes

Mutation is the origin of biological variation but how that variation is expressed and selected depends on the mapping from genetic variation to phenotypic variation, the effect on fitness, and selection. Until recently, there was little empirical knowledge about molecular mechanisms underlying these factors and to simplify theoretical and mathematical models, evolutionary genetic theory represented these processes through additive or average effects. New molecular and phenotypic technologies have shown that the development and maintenance of most traits involves an underlying system of complex genetic interactions, and that the pattern of interactions influences the dynamics of mutations and trait evolution.;In this dissertation, I analyzed the significance of genetic interactions in evolutionary processes. In Chapter 2, I used analytical and computational modeling to explore the evolution of reproductive isolation through a polygenic trait with an underlying epistatic architecture. I quantified the relationship between epistatic interactions and the accumulation of hybrid incompatibilities, and found that simulated populations experienced suites of compensatory allelic changes that facilitated genetic divergence. Strong epistatic interactions prevented allelic divergence and reproductive isolation. These results indicate that genetic drift is a plausible hypothesis for the evolution of reproductive isolation.;In Chapter 3, I modeled the evolution of a sexually dimorphic trait controlled by a genetic regulatory network to explore the relationship between network-level regulation, phenotypic identity and evolvability. Sexually dimorphic characters often change rapidly and while this is thought to be due to the strength of sexual selection acting on the trait, a dimorphic character with an underlying pleiotropic architecture may also influence the evolution of the regulatory network that controls the character and result in higher evolvability. I tested this with a computational model of a genetic regulatory network and found that sexually dimorphic characters had higher robustness to mutation, evolvability, and conditional evolvability. These results indicate that producing two pleiotropically linked characters does not constrain either the production of a robust phenotype or adaptive potential. Instead, the genetic system evolves to maximize both quantities.;In Chapter 4, I extended the sexually dimorphic regulatory network model to include sexual selection on the male trait. I studied four sexual selection scenarios---sensory bias, sexual conflict and two representations of the Fisher process---to determine the evolutionary dynamics under each of these hypotheses when the genetic architecture is a complex regulatory network, and to assess the influence of the sexual selection scenario on the resulting robustness and evolvability of the underlying network. When the male trait and female preference were determined by a single network, all sexual selection scenarios decreased evolvability and robustness. When the male trait and female preference were determined by different networks, the type of sexual selection determined the link between evolvability and robustness. Under the sensory bias and sexual conflict scenarios, robustness and evolvability were not affected while the two Fisher scenarios resulted in evolved genotypes with reduced evolvability and robustness. These results indicate that sexual selection scenarios have implications for genetic as well as phenotypic evolution.