The genetic consequences of spatially varying selection in panmixia: more transient than stable polymorphisms in the American eel (Anguilla rostrata)

Spatially varying selection is often inferred for explaining levels of potentially adaptive genetic variation observed in heterogeneous environments. Within a randomly mating population, however, theory predicts that locally selected polymorphisms may only maintain at equilibrium under restrictive conditions. Yet, these predictions have rarely been empirically tested. In order to evaluate the extent to which spatially varying selection favors the maintenance of locally adaptive polymorphisms, we investigated spatiotemporal allelic frequency patterns in coding genes of Anguilla rostrata, a panmictic species distributed over a highly heterogeneous habitat. Using 454 sequencing of individually tagged cDNA libraries, we discovered and typed about 13000 SNPs, and identified candidate variants based on their level of genetic differentiation between young glass eels collected in two ecologically contrasting habitats. Candidate SNPs were then genotyped in about 1000 individuals at different life stages of the same cohort (larvae from the Sargasso Sea, glass eels, 1 year old individuals), as well as in a cohort replicate. Evidence for spatially varying selection was detected at 13 loci by means of regression analyses between allelic frequencies and environmental variables, the most important being the mean sea-surface temperature encountered by glass eels during recruitment. Simulations under a multiple-niche Levene’s model predicted stable equilibrium for only 25% of the loci involved in genetic-by-environment interactions, while incomplete selective sweeps of derived alleles were predicted for the remaining loci. This study shows that significant patterns of genetic variation can arise within a single generation due to spatially varying and repeatable selection in an otherwise panmictic species.