Repeated randomized selection of genotypes for reliable estimates of population differentiation in data containing siblings

In population genetic studies, a proper sampling design is crucial for reliable population differentiation estimates. For genetic studies on amphibians, fish or insects, larvae are often sampled instead of adults due to their higher accessibility and abundance. However, population genetic parameters derived from larval (sibling) sampling can be biased if adults are represented unevenly in the larval population. The removal of full-siblings from data may improve the quality of the results but entails in a low number of individuals per site, especially in small populations. Using simulated data of ten microsatellite loci for ten populations, we estimated pairwise FST and RST values as well as FIS and RIS values for the F0 generation (parental) and the F1 generation (descendant). We applied a repeated randomized selection of genotypes (RRSG) to investigate the impact of removing full-siblings from the data. We also investigated the impact of reduced sample size on the results generated by RRSG as well as the advantages of RRSG over a single estimate of genetic parameters after removal of full-siblings. The RRSG approach produced pairwise FST values that deviated on average only by 0.0050 (N = 45) from their respective estimates for F0. Estimates for HE deviated less the 5% over all data sets. We henceforth suggest applying the described method to interpretations of genetic differentiations in studies with (i) small effective population sizes and (ii) data containing siblings where (iii) offspring can be assigned to at least one parent due to reproduction practice.