Abstract
Microsatellites have gained wide application for elucidating population structure in nonmodel organisms. Since they are generally noncoding, neutrality is assumed but rarely tested. In Atlantic cod ( Gadus morhua L.), microsatellite studies have revealed highly heterogeneous estimates of genetic differentiation among loci. In particular one locus, Gmo 132, has demonstrated elevated genetic differentiation. We investigated possible hitch‐hiking selection at this and other microsatellite loci in Atlantic cod. We employed 11 loci for analysing samples from the Baltic Sea, North Sea, Barents Sea and Newfoundland covering a large part of the species’ distributional range. The ‘classical’ Lewontin–Krakauer test for selection based on variance in estimates of F ST and (standardized genetic differentiation) revealed only one significant pairwise test (North Sea–Barents Sea), and the source of the elevated variance could not be ascribed exclusively to Gmo 132. In contrast, different variants of the recently developed ln Rθ test for selective sweeps at microsatellite loci revealed a high number of significant outcomes of pair‐wise tests for Gmo 132. Further, the presence of selection was indicated in at least one other locus. The results suggest that many previous estimates of genetic differentiation in cod based on microsatellites are inflated, and in some cases relationships among populations are obscured by one or more loci being the subject to hitch‐hiking selection. Likewise, temporal estimates of effective population sizes in Atlantic cod may be flawed. We recommend, generally, to use a higher number of microsatellite loci to elucidate population structure in marine fishes and other nonmodel species to allow for identification of outlier loci that are subject to selection.