Figure 3.

Active pathogens can cryptically accumulate costly latency properties. (a) A deep fitness valley separates active strains from the latent region of the fitness landscape when the regression rate ω = 0.3. (b) The fitness valley is less deep when the regression rate is reduced to ω = 0.05, particularly for those strains where all cases progress quickly to disease (y = 0). This allows active strains to traverse the fitness landscape. (a,b) The population mean phenotype of the respective infection populations are shown here as the trajectories starting at ‘s’, and the contours indicate R₀ = 1. The transmission rate β(x), virulence θ(x) and recovery rate γ(x) which lead to the fitness landscapes shown in (a,b) are displayed as the dotted, dashed and dashed-dotted lines in figure 2b. (c) A density plot of the equilibrium realized infection phenotype distribution for the case when ω = 0.05 and latency evolves (i.e. at t = 5000). The forms of parameters as functions of potential latency x and other parameter values are given in the electronic supplementary material, appendix G and table S1. (Online version in colour.)