Gene flow-dependent genomic divergence between Anopheles gambiae M and S forms.

Anopheles gambiae sensu stricto exists as two often-sympatric races termed the M and S molecular forms, characterized by fixed differences at an X-linked marker. Extreme divergence between M and S forms at pericentromeric "genomic islands" suggested that selection on variants therein could be driving interform divergence in the presence of ongoing gene flow, but recent work has detected much more widespread genomic differentiation. Whether such genomic islands are important in reproductive isolation or represent ancestral differentiation preserved by low recombination is currently unclear. A critical test of these competing hypotheses could be provided by comparing genomic divergence when rates of recent introgression vary. We genotyped 871 single nucleotide polymorphisms (SNPs) in A. gambiae sensu stricto from locations of M and S sympatry and allopatry, encompassing the full range of observed hybridization rates (0-25%). M and S forms were readily partitioned based on genomewide SNP variation in spite of evidence for ongoing introgression that qualitatively reflects hybridization rates. Yet both the level and the heterogeneity of genomic divergence varied markedly in line with levels of introgression. A few genomic regions of differentiation between M and S were common to each sampling location, the most pronounced being two centromere-proximal speciation islands identified previously but with at least one additional region outside of areas expected to exhibit reduced recombination. Our results demonstrate that extreme divergence at genomic islands does not simply represent segregating ancestral polymorphism in regions of low recombination and can be resilient to substantial gene flow. This highlights the potential for islands comprising a relatively small fraction of the genome to play an important role in early-stage speciation when reproductive isolation is limited.