Dominance reversal maintains large-effect resistance polymorphism in temporally varying environments.

A central challenge in evolutionary biology is to uncover mechanisms maintaining functional genetic variation ¹ . Theory suggests that dominance reversal, whereby alleles subject to fluctuating selection are dominant when beneficial and recessive when deleterious, can help stabilize large-effect functional variation in temporally varying environments ^2,3 . However, empirical evidence for dominance reversal is scarce because testing requires both knowing the genetic architecture of relevant traits and measuring the dominance effects on fitness in natural conditions ⁴ . Here, we show that large-effect, insecticide-resistance alleles at the Ace locus in Drosophila melanogaster ^5,6 persist worldwide at intermediate frequencies and exhibit dominance reversal in fitness as a function of the presence of an organophosphate insecticide. Specifically, we use laboratory assays to show that the resistance benefits driven by these alleles are dominant, while the associated costs are recessive (or codominant). Further, by tracking insecticide resistance and genome-wide allele frequencies in field mesocosms, we find that resistance and resistant alleles increase and then decrease rapidly in response to an insecticide pulse but are maintained at low frequencies in the absence of pesticides. We argue that this pattern is only consistent with beneficial reversal of dominance. We use Wright's theory of dominance ⁷ to hypothesize that dominance reversal should be common in general if the environmental shifts that make alleles deleterious also make them behave effectively as loss-of-function and thus recessive alleles. We also use the known haplotype structure of mesocosm populations to establish that the insecticide pulse generates chromosome-scale genomic perturbations of allele frequencies at linked sites. Overall, our results provide compelling evidence that this mechanism can maintain functional genetic variation and enable rapid adaptation to environmental shifts that can impact patterns of genomic variation at genome-wide scales via linked fluctuating selection.