Different ages of common us rare alleles may account for marked differences in gene flow estimates provided by the F, and private alleles methods (Table 3). In populations that are not at equilibrium gene flow will be overestimated by both indirect methods employed in this study. SLATKIN (1985b) showed that the private alleles method will give the false appearance of high levels of gene flow for
approximately t = N generations after a radiation event. Under the island model, the approach to equilibrium is expected to be considerably faster (on the order of t = l / m generations), but only if populations continue to exchange migrants at a rate m per generation. The
estimate of gene flow provided by the RFLP loci from the infinite island model (Nm= 3.4) approaches that required for differentiation to occur among populations by random drift. Since this is likely to be an overestimate, gene flow may actually be low enough to allow
divergence to occur by drift alone, but a suitable number of generations have not elapsed for this to be detectable (particularly among populations from adjacent regions). This conclusion is similar to that reached in previous studies examining mitochondrial DNA sequence variation in this species (SMITHet al. 1989;hNASON et al. 1992).