Propagation of genotypic variation across phenotypic levels: a model based on concave relationships unifies disparate observations on phenotypic buffering, dominance, heritability, and inbreeding depression

The way genetic variation propagates across successive phenotypic levels up to the fitness components is central to the issue of the genotype-phenotype relationship. The processes involved are highly non-linear, and exhibit a large diversity of behaviors depending on the genes, organization levels, and traits considered. Nevertheless, the shape of the relationship between traits from adjacent levels is predominantly concave, which is probably due to global constraints on matter and energy in the cell. Based on this observation, we used the properties of concave functions to model how phenotypic differences and inheritance vary across increasingly integrated levels of organization. We show that the more integrated the phenotypic levels, the closer the phenotypic values and the larger the positive deviation from additivity (i.e. dominance or heterosis). These results may explain various observations such as the low heritability and high inbreeding depression of fitness components, and the phenotypic buffering of molecular polymorphisms. Furthermore, the introduction of a cost/crowding factor in the model may explain why overdominance is so rare while heterosis is so common. To our knowledge, this systems biology approach to the genotype-phenotype relationship is the first to be based on a theoretical model of the propagation of genetic variation across phenotypic levels.