where GST is an estimator of FST for loci with any number of alleles (Nei 1973) and n is the number of subpopulations (Latter 1973; Crow & Aoki 1984; Takahata & Nei 1984). This version of GST corrects the expected amount of differentiation among subpopulations for a finite number of subpopulations.

The correction term n-1

is at a maximum of

4 with two subpopulations and approaches one as n gets large. For example, when Nem = 0.1 and n = 10, the expected value of GST is about 94% of that expected for an infinite number of demes. This implies that a given level of gene flow is more effective at homogenizing allele frequencies among fewer subpopulations than among a very large number of subpopulations. For n greater than about 50 the adjustment for a finite number of demes makes little difference and the finite number of subpopulations behave essentially as an infinite number of subpopulations.

Given that the infinite island models leads to an expected level of genetic differentiation among demes for some level of the effective migration rate, it is natural to reverse the relationship:

0 0

Post a comment