Sexual reproduction is expected to facilitate the removal of deleterious mutations from populations because biparental inheritance (i.e., segregation) and recombination during meiosis break down linkage disequilibria (LD), allowing mutations to be selected independently from their genetic background. Accordingly, the absence of recombination and segregation is expected to increase selective interference between loci, translating into reduced efficacy of natural selection. While there now exist multiple lines of evidence demonstrating that asexual lineages do experience accelerated accumulation of putatively harmful mutations, whether these mutations influence phenotype in a manner that could contribute to the maintenance of sex remains almost entirely unevaluated. Here, I use the New Zealand freshwater snail, Potamopyrgus antipodarum, to address these questions. In particular, I take advantage of the fact that the mitochondrial genome is expected to suffer from these mutational effects and interacts extensively with the nuclear genome to evaluate potential harmful effects of mutation accumulation in asexuals on a genome-wide scale. I present evidence that harmful mutations remain extant longer in asexual populations than in sexual populations, that the degree of functional constraint determines the extent of mutation accumulation in asexuals, that there is genetic variation for mitochondrial function in asexual lineages of P. antipodarum, and that phenotypic variation for mitochondrial function is mediated by both genetic and environmental variation. Together, these analyses provide strong evidence that asexual lineages are accumulating deleterious mutations, and that there is genetic variation, structured by lake, for mitochondrial function.

Recent years have witnessed the integration of theoretical advances in population genetics with large-scale analyses of complete genomes. As a result, a growing number of studies suggest the frequent occurrence of deleterious as well as adaptive mutations. Given the evidence for the widespread occurrence of selection, the finite sizes of natural populations, and the limited recombination in every genome, mutations under selection are expected to alter the fate of genetically linked mutations. The consequences of this non-independent behavior of mutations can be described by the Hill-Robertson effect in terms of the reduction in the effective population size (Ne). Reduction in the effective population size has two effects: 1) a reduction in levels of genetic variation and 2) a reduction in the effectiveness of selection that is manifested in an increased probability of fixation of deleterious mutations and a reduced probability of fixation of advantageous mutations. Changes in Ne that have previously been frequently associated with changes in recombination rate can also occur locally, in association with changes in the number of sites under selection even when the recombination rate remains uniform. The main objective of the work presented in this thesis is to investigate these local effects of the non-independent behavior of mutations on patterns of polymorphism and divergence in Drosophila using computer simulation and experimental approaches. A computer simulation approach is developed to investigate the local consequences of linked selection on estimates of selection and the proportion of adaptive substitutions using the McDonald-Kreitman framework. The results suggest that even a high level of recombination is unlikely to remove all the effects of linked selection. Ignoring these local linkage effects leads to misleading estimates of the intensity of selection and the proportion of adaptive substitutions. Two predictions of the Hill-Robertson effect were tested empirically by examining patterns of polymorphism and divergence combined with codon bias estimates in genes with and without introns: 1) the effectiveness of selection and polymorphism levels are expected to be reduced in the center of the long coding sequence of genes without introns (the intragenic Hill-Robertson effect), and 2) introns are expected to function as modifiers of recombination thereby increasing the effectiveness of selection in the central region of the coding sequence of genes containing centrally located introns. The evidence from divergence and codon bias patterns in genes with a long coding sequence supports the presence of the intragenic Hill-Robertson effect. However, polymorphism levels do not show the expected decrease in the center of the coding sequence. With regard to the second prediction, results indicate that intron presence does not increase the effectiveness of selection at synonymous sites in the set of investigated genes. Rather, intron presence is associated with increased levels of adaptation at nonsynonymous sites. Further investigations are necessary to clarify the role of introns in mediating the increase in adaptation.