Appendix 2—figure 2. Stabilizing selection during the equilibration phase causes turnover in the genetic basis of adaptation.

The cartoons depict the trajectories of alleles with opposing effects of given magnitudes and initial MAFs. For the purpose of illustration, we focus on alleles with large (A) and moderate (B) effects, with initial MAFs in the tail of the corresponding equilibrium MAF distribution (the 99.5^th percentile), and a shift size of $\mathrm{\Lambda }=2\cdot \sqrt{V_A(0)}$ with $\sqrt{V_A(0)}=17\delta$ . Directional selection during the rapid phase increases the frequency of aligned alleles relative to those with opposing effects, and these frequency differences underlie short-term phenotypic adaptation. (A) The initial MAF of large effect alleles, even those in the 99.5^th percentile, is sufficiently low such that both aligned and opposing alleles still have low MAFs at the end of the rapid phase. Consequently, they are both strongly selected against during the equilibration phase and almost certainly go extinct, thereby erasing their short-term contribution to phenotypic adaptation. (B) Moderate effect alleles start at much higher initial MAFs. In the extreme, this initial frequency is sufficiently high for directional selection during the rapid phase to push aligned alleles above frequency 1/2, thereby reversing the direction of (under-dominant) selection on them, but not on the opposing alleles, during the equilibration phase. Consequently, the expected contribution of moderate effect alleles with sufficiently high initial MAF to phenotypic adaptation is amplified during the equilibration phase.