DESCRIPTION AND POWER ANALYSIS OF 2 TESTS FOR DETECTING RECENT POPULATION BOTTLENECKS FROM ALLELE FREQUENCY DATA

When a population experiences a reduction of its effective size, it ge nerally develops a heterozygosity excess at selectively neutral loci, i.e., the heterozygosity computed from a sample of genes is larger tha n the heterozygosity expected from the number of alleles found in the sample if the population were at mutation drift equilibrium. The heter ozygosity excess persists only a certain number of generations until a new equilibrium is established. Two statistical tests for detecting a heterozygosity excess are described. They require measurements of the number of alleles and heterozygosity at each of several loci from a p opulation sample. The first test determines if the proportion of loci with heterozygosity excess is significantly larger than expected at eq uilibrium. The second test establishes if the average of standardized differences between observed and expected heterozygosities is signific antly different from zero. Type I and II errors have been evaluated by computer simulations, varying sample size, number of loci, bottleneck size, time elapsed since the beginning of the bottleneck and level of variability of loci. These analyses show that the most useful markers for bottleneck detection are those evolving under the infinite allele model (IAM) and they provide guidelines for selecting sample sizes of individuals and loci. The usefulness of these tests for conservation biology is discussed.