Segregation variance after hybridization of isolated populations.

We develop a model to predict the increase in genetic variance of a quantitative character in a hybrid population produced by crossing two previously isolated populations of the same species. The increase in variance in the F2 hybrids, the 'segregation variance', is caused by differences in the average allelic effects at each locus and by linkage disequilibrium among loci. We focus on the case in which the character is additively based and the average value of the character does not differ in the two populations. In that case the predicted segregation variance depends strongly on what is assumed about the genetic basis of the character. If the genetic variance of the character in each population is attributable to loci with numerous alleles of small effect that are in moderate frequency, as in Lande's (1975) model, the segregation variance should increase linearly with time since the populations were isolated, at a rate determined by the inverse of the effective population size. If the genetic variance is attributable to loci with alleles in very low frequency, as in Turelli's (1984) house-of-cards model or in Barton's (1990) model of pleiotropic, deleterious alleles, then the segregation variance in the hybrid population increases at a much lower rate.