Endemic states.

Plots A) and B) show the results of the simulations for the endemic states of configurations corresponding to the points in Fig 2E, for a bipartite virus (v = 2). In both A) and B), x-axis is the total prevalence of the disease, i.e., the fraction of hosts infected by any configuration of the virus, at equilibrium, and the y-axis is the prevalence of the defective variants, i.e., the fraction of hosts infected by at least one defective segment. The points are numerically recovered endemic states, their type being indicated by the labels. In A) the underlying contact network is homogeneous, so that . The solid vertical lines mark the analytical prediction of the total prevalence when wt is present either alone or together with just one variant. The dashed vertical lines mark the analytical prediction of the prevalence when the segments circulate without wt. The crosses mark the prevalence values of the equilibrium points averaged over the runs not leading to extinction, among the 5000 executed per point. B) focuses on P3 (in Fig 2E): the fixed value is obtained either as (homogeneous network, as in (B)), or with two heterogeneous networks with exponent γ = 3.5, μ = 0.45 and γ = 3.2, μ ≈ 0.81 (and thus 〈k〉/〈k2〉 ≈ 0.56 and 〈k〉/〈k2〉 ≈ 0.31, respectively). For each of the points and the equilibria examined, C) reports the branching ratio, defined as the probability of reaching that particular equilibrium. They are computed by starting all the simulations with all susceptible but one in ⌜all⌟ (infected by wt and all the variants), and counting the fraction of the runs that reach that equilibrium, among the ones that do not go to extinction.