Molecular population genetics of the sex-specific mitochondrial DNA of the California sea mussel, Mytilus californianus

Mussels in the genus Mytilus have an unusual mode of inheritance of mitochondrial DNA known as doubly uniparental inheritance (DUI). In most animals the mitochondrial genome is inherited maternally, but in Mytilus and a handful of other bivalve mollusks an F genome is passed down to daughters and an M genome is passed down to sons, creating two distinct mtDNA lineages. This unusual system presents research opportunities not available in standard systems of mtDNA inheritance. Marine species with long larval durations, such as M. californianus, are expected to disperse widely and show little population structure. If patterns among different kinds of markers---nuclear or mitochondrial---are discordant, this is can be evidence of natural selection acting on one or both markers. Most studies are limited by having only one mtDNA marker to work with. In Mytilus it is possible to compare two independently inherited haploid mtDNA markers, F and M, to nuclear data. In this dissertation, the F and M mtDNA of M. californianus was surveyed for population genetic structure over more than 2000 km of coastline on the west coast of North America by direct DNA sequencing of four protein-coding mitochondrial genes sampled from four locations. Reasons for the lack of detected structure are explored in the context of the results of previous allozyme studies on M. californianus and compared to published studies that surveyed other species over parts of the same geographic region. In addition, given that recent research has shown that mtDNA does not evolve in a strictly neutral manner as was previously thought, Mytilus affords the opportunity to investigate the roles of selection, migration, and genetic drift in the evolution of mtDNA without the complication of comparing species that have different demographic histories. The same M. californianus sequences were used to investigate the importance of selection in maintaining the high levels of polymorphism found in both genomes. Recently developed Markov chain Monte Carlo methods were employed along with more traditional comparisons of silent and replacement polymorphism. The factors affecting the large divergence between the two genomes was also investigated. The results support the hypothesis that purifying selection is relaxed in M relative to F, but that the relaxation of selective constraint is not uniform over the genome.