Figure 3 .

Selection regimes of stationary adaptation. (A) Selection-dependent fixation probability of mutations G(σ), scaled by the population size N. Analytic model solution (red line) and simulation results (circles) show three regimes of selection: (i) effective neutrality regime (white background), where G(σ) takes values similar to the fixation probability of independent sites with reduced selection, $G_0\left(\sigma /2N\tilde{\sigma }\right)$ (short-dashed blue line); (ii) adaptive regime (green), where G(σ) crosses over to the fixation probability for unlinked sites with full selection, G₀(σ) (long-dashed blue line) [the strong-selection part of this crossover is captured by the Gerrish–Lenski model, G_GL(σ) (brown line, see section 5 of File S1); and (iii) strongly deleterious passenger regime (red), where G(σ) is exponentially suppressed, but drastically larger than for unlinked sites (long-dashed blue line) due to hitchhiking in selective sweeps. (B and C) Selection-dependent degree of adaptation α(f) and fitness flux Φ(f), scaled by N. Analytical model solution (red line) and simulation results (circles) show two regimes of selection: (i) effective neutrality regime (white background), where α(f) and Φ(f) take values similar to those of unlinked sites with reduced selection (short-dashed blue line), and (ii) adaptive regime (green), where α(f) and Φ(σ) cross over to values of unlinked sites with full selection (short-dashed blue lines). The strong-selection part of the crossover for Φ is captured by the Gerrish–Lenski model, Φ_GL(f) (brown line, see section 5 of File S1). System parameters are N = 2000, L = 1000, 2Nμ = 0.025, 2Nγ = 0.1, and $2N\overline{f}=50$, and simulation time is 2 × 10⁶ generations.