Statistical power for detecting genetic divergence—organelle versus nuclear markers

Abstract

Statistical power is critical in conservation for detecting genetic differences in space or time from allele frequency data. Organelle and nuclear genetic markers have fundamentally different transmission dynamics; the potential effect of these differences on power to detect divergence have been speculated on but not investigated. We examine, analytically and with computer simulations, the relative performance of organelle and nuclear markers under basic, ideal situations. We conclude that claims of a generally higher resolving power of either marker type are not correct. The ratio R = F _ST,organelle/F _ST,nuclear varies between 1 and 4 during differentiation and this greatly affects the power relationship. When nuclear F _ST is associated with organelle differentiation four times higher, the power of the organelle marker is similar to two nuclear loci with the same allele frequency distribution. With large sample sizes (n ≥ 50) and several populations or many alleles per locus (≥5), the power difference may typically be disregarded when nuclear F _ST > 0.05. To correctly interpret observed patterns of genetic differentiation in practical situations, the expected F _STs and the statistical properties (i.e., power analysis) of the genetic markers used should be evaluated, taking the observed allele frequency distributions into consideration.