Estimation and interpretation of genetic distance in empirical studies

SUMMARY Linear functions of Nei's genetic-distance statistic are calculated frequently in the literature of population genetics. Variance estimates for these linear functions are either not presented or incorrectly calculated. Part of the problem stems from the common assumption that distance statistics are independent random variables. This assumption is not generally correct. We describe methods for estimating the variance of linear combinations of genetic-distance statistics. We also suggest a method for constructing confidence intervals on genetic-distance statistics when these values are small (< 0·10) and their distribution deviates substantially from normal.