Premise The evolutionary response of a trait to environmental change depends upon the level of additive genetic variance. It has been long argued that sustained selection will tend to deplete additive genetic variance as favored alleles approach fixation. Non‐additive genetic variance, due to interactions among alleles within and between loci, does not immediately contribute to an evolutionary response. However, shifts in the allele frequencies within and between interacting loci may convert non‐additive variance into additive variance. Here we consider the possibility that an environmental shift may alter allelic interactions in ways that convert dominance into additive genetic variance. Methods We grew a pedigreed population of Brassica rapa in greenhouse and field conditions. The field conditions mimicked agricultural conditions from which the base population was drawn, while the greenhouse featured benign conditions. We used Bayesian models to estimate the additive, dominance, and maternal components of quantitative genetic variance. We also estimated genetic correlations across environments using parental breeding values. Results Although the additive genetic variance was elevated in the greenhouse condition, no consistent pattens emerged that would indicate a conversion of dominance variance. The unusually low genetic variance and broad confidence intervals for the variance estimates obtained through this analysis preclude definitive interpretations. Conclusions Further studies are needed to determine whether between‐environment changes in additive genetic variance can be traced to conversion of dominance variance.

The persistence of a declining population in the face of environmental change may depend on how fast natural selection restores fitness, a process called "evolutionary rescue". In turn, evolutionary rescue depends on a population's adaptive potential. Fisher's theorem states that a population's adaptive potential equals the additive genetic variance for fitness (VA(W)) divided by mean fitness (Ŵ). Both the numerator and denominator of this rate can differ across environments even when holding allele frequencies constant. However, little is known about how these rates change in wild populations during adaptation, including changes in additive and dominance variance. We assessed the change in adaptive potential and dominance variance in fitness (VD(W)) for a Québec population of wild mustard (Brassica rapa) under climate warming. We also assessed adaptive constraints that could arise from negative genetic correlations across environments. We grew a pedigreed population of 7000 plants under ambient and heated (+4°C) temperatures and estimated the change in Ŵ, VA(W), VD(W), and the cross-environment genetic correlations (rA). As predicted, estimates of VA(W) and adaptive potentials were higher under heated conditions but non-significantly so. This is perhaps because, surprisingly, plants exposed to a warmer climate exhibited greater Ŵ. Nevertheless, increased fitness in the warmer environment suggests a plasticity-based short-term potential for adaptation, and that weak but non-significant genetic correlations across environments will enable slow on-going adaptation to warming. Overall, this population of B. rapa harbours existing genetic architecture to persist under warmer temperatures through pre-adaptation but not through evolutionary rescue.