Living in each other’s pockets: Nucleotide variation inside a genomic island harboring Pan I and its neighbors in Atlantic cod

Abstract

The Pan I locus in Atlantic cod lies in a genomic island of divergence extending over a large genomic region. The locus has two divergent alleles, defined by a single DraI restriction site, that have been related to behavioral differences of habitat selection by depth and temperature. The Pan I locus is known to be under an unusual mix of balancing selection and selective sweeps within the functional types. Here we study nucleotide variation in a 12.5 kb region inside the genomic island harboring Pan I and neighboring loci for sortilin 1 (Sort1) and ataxin 7-like 2 (Atxn7l2) which we partially covered. Variation of the 31 gene copies throughout the region falls into two divergent haplogroups that correlate with the 25 copies of A and six copies of B alleles of Pan I. The unfolded site frequency spectrum for the part with Pacific cod used as the outgroup is trimodal with a mode at singletons and two high frequency modes at 6/31 and 25/31 representing the two genealogical lineages. The folded site frequency spectrum for the entire region similarly has a high frequency mode of mutations that have accumulated on the two lineages. The high frequency of singletons is accounted for by multiple merger coalescent models. Parameter estimates using these models indicate sweepstakes reproduction. The high frequency modes of the spectrum is evidence for balancing selection. Analysis of non-synonymous changes shows that Pan I is at least one focus of selection within the genomic island. There may be multiple sites of selection and epistatic interactions. There is extensive linkage disequilibrium throughout the region. We suggest that the genomic island of divergence is a supergene of co-adapted complexes possibly locked together by structural variation.