Sources of variance in protein heterozygosity: the importance of the species-protein interaction

We report on a detailed survey of protein heterozygosity in Canadian freshwater fish and mammals. A simple one-way analysis showed substantial variance among species. However, a two-way analysis of species and proteins showed that there was little if any variance among species or among proteins, but a very large species-protein interaction. We could not remove this interaction by analysing taxa separately by constructing completely balanced datasets, by eliminating study bias or by excluding monomorphic proteins, and we could not decompose the interaction by classifying enzymes according to their form and function. We conclude that most of the variance in protein heterozygosity is attributable to species-protein interaction. This casts some doubt on the interpretation of comparative studies of mean heterozygosity among species or among proteins. Our result seems inconsistent with the neutral theory of protein variation but not with the differential action of natural selection.