Gene flow influences the genomic architecture of local adaptation in six riverine fish species

Understanding how gene flow influences adaptive divergence is important
for predicting adaptive responses. Theoretical studies suggest that when
gene flow is high, clustering of adaptive genes in fewer genomic regions
would protect adaptive alleles from recombination and thus be selected
for, but few studies have tested it with empirical data. Here, we used
RADseq to generate genomic data for six fish species with contrasting life
histories from six reaches of the Upper Mississippi River System, USA. We
used four differentiation-based outlier tests and three
genotype-environment association analyses to define neutral SNPs and
outlier SNPs that were putatively under selection. We then examined the
distribution of outlier SNPs along the genome and investigated whether
these SNPs were found in genomic islands of differentiation and
inversions. We found that gene flow varied among species, and outlier SNPs
were clustered more tightly in species with higher gene flow. The two
species with the highest overall FST (0.0303 - 0.0720) and therefore
lowest gene flow showed little evidence of clusters of outlier SNPs, with
outlier SNPs in these species spreading uniformly across the genome. In
contrast, nearly all outlier SNPs in the species with the lowest FST
(0.0003) were found in a single large putative inversion. Two other
species with intermediate gene flow (FST ~ 0.0025 - 0.0050) also showed
clustered genomic architectures, with most islands of differentiation
clustered on a few chromosomes. Our results provide important empirical
evidence to support the hypothesis that increasingly clustered
architectures of local adaptation are associated with high gene
flow.