Power for detection.

<p>Power of detection for the different 2-locus models. Power was estimated as the proportion of iterations in which the full model F test for the correct model had a value . Panel A shows results for simulations with disparate allele frequencies within each locus (1%–10% for the rare allele and 90% for the other allele); panel B shows results for simulations with allele frequencies within each locus closer to equal (20%–30% for one allele and 70%–80% for the other allele). Sample sizes were 100, 500, or 1000 diploid individuals.