Abstract

By pooling or removing samples of small size, we investigated how results from microsatellite analyses of spatial genetic structure in darkblotched rockfish (Sebastes crameri) were affected. Genotypes from six and seven microsatellite loci from 1206 specimens collected offshore from Washington to California were employed in the analyses. Sample sizes varied greatly among locations (n = 11-114). When adjacent samples of n < 25 were pooled using an absolute genetic distance (F-ST <= 0), the correlation between genetic and geographic distance found in the original data set increased nearly twofold, and overall F-ST (95% confidence interval) increased from 0.001 (0.000-0.002) to 0.002 (0.001-0.003). Removing samples where n < 25 gave a similar result, yet the correlation increase was smaller. Another pooling strategy based on similarity tests allowed larger sizes in composite samples (n > 100) and further increased the correlation, although this strategy did not raise overall F-ST. These results indicate that under genetic isolation by distance, excessive pooling might not enhance the overall genetic differentiation among populations. The regression slope in isolation by distance plots was robust throughout all treatments, and its value suggests limited dispersal distance on this species.