Evaluating loci for use in the genetic analysis of population structure

Loci that show unusually low or high levels of genetic differentiation are often assumed to be subject to natural selection. We propose a method for the identification of loci showing such disparities. The differentiation can be quantified using the statistic F-ST. For a range of population structures and demographic histories, the distribution of F-ST is strongly related to the heterozygosity at a locus. Outlying values of F-ST can be identified in a plot of F-ST vs. heterozygosity using a null distribution generated by a simple genetic model. We use published data-sets to illustrate the importance of the relationship with heterozygosity. We investigate a number of models of population structure, and demonstrate that the null distribution is robust to a wide range of conditions. In particular, the distribution is robust to differing mutation rates, and therefore different molecular markers, such as allozymes, restriction fragment length polymorphisms (RFLPs) and single strand conformation polymorphisms (SSCPs) can be compared together. We suggest that genetic variation at a discrepant locus, identified under these conditions, is likely to have been influenced by natural selection, either acting on the locus itself or at a closely linked locus.