Influence of variables in the null model
Variables in the null model were significant predictors of
FST in all multiple regressions (Table 1) and in model 7
(Table S6). Distance had the highest independent contribution (8%),
compared to genetic marker and mean sample size (4% each). In general,
FST values become larger when the geographic scale of
studies increases. In contrast, FST values decrease with
larger mean sample sizes of individuals per population. Codominant
markers (microsatellites and allozymes) tend to underestimate
FST values, while dominant markers (AFLP and RAPD)
overestimate them. ISSR markers did not differ from others.