Abstract

Models that purport to explain the maintenance of MHC polymorphism must be able to explain a variety of phenomena. (1) The range of MHC allele frequencies at some of the loci is very large, with some alleles quite common and many others rare, while at others the range of allele frequencies is far narrower. (2) MHC alleles and their frequencies often have long persistence times, in some cases tens of millions of years. (3) Random-mating populations appear to be in Hardy-Weinberg equilibrium for MHC. (4) There is no obvious, strong and consistent selection pressure yet detected that acts differentially on different MHC genotypes. (5) Because the allelic composition of the MHC polymorphism does change over evolutionary time, the MHC system must be capable of accommodating new alleles with similar properties without destruction of the equilibria that permit the maintenance of the older alleles. In this review I examined the degree to which a large number of models that have been proposed fit these criteria. These include heterosis, marginal overdominance, conditional heterosis, assortative mating, maternal-fetal incompatibility, molecular mimicry, minority advantage, pathogen adaptation, and optimum allele frequency models. Most of the models do poorly at accounting for a number of the above phenomena. The last class, optimum allele frequency models, have the most satisfactory set of properties. However, optimum allele frequency models require mechanisms that somehow "feed back" from the frequency of an allele in the population to the fitness of an organism carrying that allele. Thus, these models require that MHC polymorphisms be maintained by some type of group selection. Evidence for an against optimum allele frequency selection, and ways in which this type of selection might be detected experimentally, are presented.