Mean sample size per population
The maximum value that FST can take decreases when the
within-population expected heterozygosity increases. Thus, a general
concern is that large sample sizes are required because small samples
can overestimate FST (Holsinger & Weir, 2009;
Kalinowski, 2005; Willing, Dreyer, & van Oosterhout, 2012). We
accounted for this potential bias by including the mean sample size per
population in our null model. Across the studies, this sample size
ranged from 3 to 285 individuals per population, with an overall mean of
40.12 (± 44.9 SD).