Dispersal, gene flow and adaptive differentiation in hierarchically structured populations

The balance between local adaptation and gene flow is rarely quantified when populations are hierarchically differentiated across multiple spatial scales. We addressed this gap in Atlantic salmon (Salmo salar) at its southern range edge, using 1342 individuals from 21 rivers along the French Atlantic and Channel coasts with a 60k SNP array, and combining estimates of dispersal with genome-wide scans for adaptive differentiation across spatial scales. Population structure was hierarchical, resolving five genetic clusters with differentiation among groups five times higher than among populations within them. Assignment tests revealed a marked contrast between clusters, but unassigned individuals reached very high proportions within some clusters, indicating that gene flow has already blurred among-populations differences. Adaptive differentiation was detected among and within clusters, but the loci involved overlapped only partially across scales and among clusters, indicating that adaptation at each hierarchical level relies on distinct sets of variants. Dispersal was a poor guide to where this signal persisted: retaining putative dispersers reduced the candidate set more than threefold, showing that immigrant genotypes could dilute among-populations allele-frequency contrasts within a single generation. Yet differentiation among populations remained strong, as expected if dispersers reproduce poorly. Resolving the balance between local adaptation and gene flow therefore requires accounting for dispersal and adaptive differentiation at different spatial scales, with direct implications for conserving hierarchically structured populations at climatically vulnerable range edges.