Evidence for non-neutrality of mitochondrial DNA haplotypes in Drosophila pseudoobscura

Mitochondrial DNA (mtDNA) haplotypes usually are assumed to be neutral, unselected markers of evolving female lineages. This assumption was tested by monitoring haplotype frequencies in 12 experimental populations of Drosophila pseudoobscura which were polymorphic for mtDNA haplotypes. Populations were maintained for at least 10 generations, and in one case for 32 generations, while tests of mtDNA selective neutrality were conducted. In an initial population, formed from a mixture of two strains with different mitochondrial haplotypes, the frequency of the Bogota haplotype increased 46% in 3 generations, reaching an apparent equilibrium frequency of 82% after 32 generations. Perturbation of this equilibrium by addition of the less common haplotype resulted in a rapid, dramatic increase in frequency of the second haplotype, and a return to essentially the same equilibrium frequency as before perturbation. This behavior is not consistent with mtDNA neutrality, nor is the equilibrium consistent with a simple model of constant selection on the haploid mtDNAs. Replicate cage experiments with mtDNA haplotypes did not always generate the same result as the initial cage. Several lines of evidence, including manipulations of the nuclear genome, support the idea that both nuclear and mitochondrial genomes are involved in the dramatic mtDNA frequency changes. In another experiment, strong female viability selection was implicated via mtDNA frequency changes. Although the causes of the dramatic mtDNA frequency changes in our populations are not obvious, it is clear that Drosophila mitochondrial haplotypes are not always simply neutral markers. Our findings are relevant to the introduction of a novel mtDNA variant from one species or one population into another. Such introductions could be strongly favored by selection, even if it is sporadic.