A theoretical and numerical assessment of genetic variability.

The equilibrium behavior of one-locus viability selection models is studied numerically. The selection schemes include randomly chosen viabilities, viabilities chosen to measure a hypothetical distance between the alleles making up the genotype and viabilities that obey various allelic dominance relations. From 3 to 8 alleles are considered. Among the key conclusions are (1) equilibria that are most polymorphic do not usually have the highest mean fitness, (2) the more structure there is in the choice of the viability model, the greater is the level of polymorphism at equilibrium, and (3) for the numbers of alleles chosen here, the equilibrium reached by iteration from the centroid of the allele frequency simplex is the best predictor of the equilibrium attainable from randomly chosen starting vectors. Preliminary evidence shows that this is not the case for 16 alleles.