Epistasis describes interactions between genes where the effect of one allele depends on alleles at other loci. For quantitative traits controlled by many loci, epistasis can reorganize how genetic variation is expressed and transmitted across generations. H. Allen Orr University of Edinburgh has emphasized that nonadditive interactions alter the effective landscape of adaptation, while Michael Lynch Indiana University Bloomington has highlighted their role in shaping genetic architecture over long timescales. Together these perspectives show that epistasis is not merely a complication for models but a driver of evolutionary trajectories.
Mechanisms: how epistasis alters inheritance
At the mechanistic level, epistasis changes the mapping from genotype to phenotype by converting combinations of alleles into different phenotypic outcomes than predicted by independent effects. This influences the partitioning of genetic variance into additive and nonadditive components. Additive variance is what natural and artificial selection most directly converts into evolutionary change; epistatic interactions can mask additive effects under some genetic backgrounds and reveal them under others. Nick Barton University of Oxford has shown that in large, recombining populations epistatic variance often behaves like additive variance for short-term responses, whereas in small or structured populations it can produce long-lasting, context-dependent effects.
Evolutionary consequences and context-dependence
Consequences of epistasis include modified rates of adaptation, emergent ruggedness of the fitness landscape, and altered maintenance of genetic diversity. When epistasis creates multiple local fitness peaks, populations may become trapped on suboptimal solutions, leading to divergent evolutionary paths among populations facing similar selection. This has practical implications: in crop and livestock improvement, breeders observe that selection outcomes can differ by population and environment when epistasis is present, and in medical genetics epistasis contributes to variable disease risk across human populations. Environmental heterogeneity, cultural practices, and territorial isolation can amplify these patterns by changing which allele combinations are favored and by limiting recombination among populations.
Recognizing epistasis changes both prediction and inference. Models that incorporate interaction terms or use genotype-aware mapping improve accuracy of genomic prediction and illuminate the pathways of adaptation. Empirical and theoretical work by established researchers and institutions underscores that accounting for epistasis is essential for realistic understanding of quantitative-trait evolution, especially when population size, structure, and environment make nonadditive effects persistent and consequential.