The satyrine butterfly Coenonympha tullia (Nymphalidae: Satyrinae) displays a deep split between two mitochondrial clades, one restricted to northern Alberta, Canada, and the other found throughout Alberta and across North America. We confirm this deep divide and test hypotheses explaining its phylogeographic structure. Neither genitalia morphology nor nuclear gene sequence supports cryptic species as an explanation, instead indicating differences between nuclear and mitochondrial genome histories. Sex-biased dispersal is unlikely to cause such mito-nuclear differences; however, selective sweeps by reproductive parasites could have led to this conflict. About half of the tested samples were infected by Wolbachia bacteria. Using multilocus strain typing for three Wolbachia genes, we show that the divergent mitochondrial clades are associated with two different Wolbachia strains, supporting the hypothesis that the mito-nuclear differences resulted from selection on the mitochondrial genome due to selective sweeps by Wolbachia strains.