DifferInt: compositional differentiation among populations at three levels of genetic integration.

In many fields of study, it is important to know how different populations are genetically. Commonly used measures such as FST and its derivatives based on gene identity probabilities do not reliably reflect difference, as they can be maximal when almost all populations are identical and very small when populations are completely distinct. Compositional differentiation, in contrast, is maximal only for completely distinct populations. Moreover, underlying forces of fragmentation that act on single- and multilocus genotypes may be missed if only the allelic level is viewed. For these reasons, descriptive measures of compositional differentiation based on untransformed distributions of genetic types at different levels of genetic integration (alleles, single- and multilocus genotypes) were derived. Here, two measures of complementary differentiation and two new measures of dispersive differentiation are described. One of each considers genic differences between individuals, ensuring their consistency across integration levels. An increase from one level to the next indicates differences among the populations in their forms of gene association. The computer program DifferInt calculates these measures for one or more codominantly expressed gene loci at the gene-pool, single-locus and multilocus levels. Effects of gene pools and gene association on differentiation are compared by permutation analysis. Snail diagrams depict the contribution of each population to differentiation. Results are demonstrated using a numerical example. The applicability of the measures calculated by this program in conservation, phylogeography, mating system analysis and adaptation studies is discussed.