A quantitative model of the relationship between phenotypic variance and heterozygosity at marker loci under partial selfing.

Negative relationships between allozyme heterozygosity and morphological variance have often been observed and interpreted as evidence for increased developmental stability in heterozygotes. However, inbreeding can also generate such relationships by decreasing heterozygosity at neutral loci and redistributing genetic variance at the same time. I here provide a quantitative genetic model of this process by analogy with heterozygosity-fitness relationships. Inbreeding generates negative heterozygosity-variance relationships irrespective of the genetic architecture of the trait. This holds for fitness traits as well as neutral traits, the effect being stronger for fitness traits under directional dominance or overdominance. The order of magnitude of heterozygosity-variance regressions is compatible with empirical data even with very low inbreeding. Although developmental stability effects cannot be excluded, inbreeding is a parsimonious explanation that should be seriously considered to explain correlations between heterozygosity and both mean and variance of phenotypes in natural populations.