Is there selection on RFLP differences in mitochondrial DNA?

Experimental populations of Drosophila simulans were established for the purpose of detecting the presence or absence of selection on a restriction fragment length polymorphism in mitochondrial DNA (mtDNA). It was then discovered that the founding strains differed with respect to the Rickettsia-mediated incompatibility system in this species, which is maternally transmitted together with the mtDNA differences. A population model was constructed using the known fitness effects of the incompatibility system, with the result that the population trajectories can be completely explained by the effects of the microorganism with no need to invoke selection on mtDNA. The strong conclusion is that in this case we can rule out the strong selection proposed by MacRae and Anderson to explain the "dramatic mtDNA changes" in their Drosophila pseudoobscura populations. The population theory used for the experiments is discussed in the context of natural populations. Estimated parameters include the possibility that with two populations, one with the organism and one without it, there may be no bias as to which will invade the other, which in turn suggests no global tendency for the infection to spread or decline.