Document Type



Classical isolation by distance models predict a negative exponential relationship between genetic similarity and geographic distance. Furthermore, under random spatial differentiation all alleles should reflect the same pattern because migration affects all loci identically. The effect of nonrandom composition of migrant groups (kin-structured migration) on these predictions was investigated using computer simulation. Spatial autocorrelation analysis of the simulated gene frequencies showed a decline in similarity with distance in the kin-structured experiments, as predicted by the isolation by distance model. However, the spatial autocorrelations were lower and the variation among correlograms of different alleles was greater than what would be expected under random migration. These findings suggest caution in inferring specific evolutionary conclusions from empirical spatial patterns.