Sequential adaptive introgression of the mitochondrial genome in Drosophila yakuba and Drosophila santomea

Abstract

Interspecific hybridization provides the unique opportunity for species to tap into genetic variation present in a closely related species and potentially take advantage of beneficial alleles. It has become increasingly clear that when hybridization occurs, mitochondrial DNA (mtDNA) often crosses species boundaries, raising the possibility that it could serve as a recurrent target of natural selection and source of species' adaptations. Here we report the sequences of 46 complete mitochondrial genomes of Drosophila yakuba and Drosophila santomea, two sister species known to produce hybrids in nature (~3%). At least two independent events of mtDNA introgression are uncovered in this study, including an early invasion of the D. yakuba mitochondrial genome that fully replaced the D. santomea mtDNA native haplotypes and a more recent, ongoing event centred in the hybrid zone. Interestingly, this recent introgression event bears the signature of Darwinian natural selection, and the selective haplotype can be found at low frequency in Africa mainland populations of D. yakuba. We put forward the possibility that, because the effective population size of D. santomea is smaller than that of D. yakuba, the faster accumulation of mildly deleterious mutations associated with Muller's ratchet in the former species may have facilitated the replacement of the mutationally loaded mitochondrial genome of D. santomea by that of D. yakuba.

Keywords: gene flow; positive selection; recurrent adaptation; speciation.

Fig 1

Marginal density for the migration rate parameter (m) obtained by fitting the isolation-with-migration model to a data set with two descendant populations, Drosophila yakuba and Drosophila santomea. 1, 2.5 and 5× correspond to θ per sequence of ∼30, ∼75 and ∼150, respectively (θ per sequence observed in D. yakuba and D. santomea are 29.79 and 21, respectively).

Fig 2

Marginal density for the migration rate parameter (m) obtained by fitting the isolation-with-migration model to a data set with three descendant populations. Marginal density for m between (a) Drosophila santomea–Drosophila yakuba island populations, (b) D. yakuba island–D. yakuba mainland populations and (c) D. santomea –D. yakuba mainland populations. 1, 2.5 and 5× correspond to θ per sequence of ∼30, ∼75 and ∼150, respectively (θ per sequence observed in the D. yakuba mainland, D. yakuba island and D. santomea populations are 23.27, 18.68 and 21, respectively).

Fig 3

Neighbour-joining trees reconstructed using complete sequences of 48 mitochondrial genomes from Drosophila erecta, Drosophila yakuba (yak) and Drosophila santomea (san). D. santomea sequences from the classic hybrid zone in the Obo Natural Reserve are underlined, and asterisks indicate Africa mainland lines of D. yakuba. Bootstrap values were obtained after 1000 replicates.

Fig 4

Marginal density for the migration rate parameter (m) obtained by fitting a two-population isolation-with-migration model to a data set of mtDNA sequences only (mtDNA) or mtDNA and Y chromosome sequences combined (mtDNA + Y chr.).

Fig 5

Schematic representation of the temporal dynamics of mitochondrial introgression in the Drosophila yakuba–Drosophila santomea system. Grey rectangles represent upper and lower 95% highest posterior density of the time to the most recent common ancestor.