Bay-Scale Population Differentiation among Anadromous Rainbow Smelt
Mark W. Coulson and Paul Bentzen
Spatial separation of spawning aggregations has long been recognized as a major determinant of population genetic structure. Isolation by distance (IBD) (Wright 1943) predicts that genetic differences accumulate along increasing geographic distances separating populations. However, local factors may also play a role in further structuring populations. Rainbow smelt Osmerus mordax are assumed to remain near the coast and do not undergo lengthy migrations, suggesting a potential for strong population structuring. It has been shown that regional forms may exhibit strikingly different degrees of genetic structure, such as that seen between Newfoundland and the mainland. While Newfoundland populations display a strong signal of IBD, mainland populations seemingly show little effect of distance on population structuring (Bradbury et al. 2006). Therefore, it is possible that structure among mainland samples may be influenced by more local effects (e.g., bay-scale). Factors such as variation in the hydrographic and temperature conditions encountered throughout their range may therefore be a more pronounced determinant of population structure than increasing distance alone. Indeed, member vagrancy has been proposed for smelt and other species whereby the spatial scale across which early life history stages are retained determines the scale at which structuring occurs (Iles and Sinclair 1982).
Anadromous runs of rainbow smelt were sampled from 22 populations by dip netting throughout much of their native range (Figure 1A). Approximately 50–100 individuals per population were sampled and genotyped for nine microsatellite loci (see Coulson et al. 2006 for loci characteristics and procedures). Basic descriptive microsatellite statistics were conducted as in Bradbury et al. (2006). Factorial correspondence analysis (FCA) was carried out to emphasize differences between populations based on their multilocus genotypes. Two approaches were taken to estimate the effects of distance on genetic differentiation. First, pairwise measures of FST were plotted against pairwise geographic distances, and second, an assignment test was carried out on all fish: the resultant proportion of misassigned fish to each location was plotted against the geographic distance of that location from the source (i.e., where the fish was sampled). Finally, a barrier analysis was conducted to determine the principal discontinuities in the spatial pattern of the genetic data.