Ecological factors driving the long-term evolution of influenza's host range

Abstract

The evolution of a pathogen's host range is shaped by the ecology of its hosts and by the physiological traits that determine host specificity. For many pathogen traits, there is a trade-off: a phenotype suitable for infecting one set of hosts poorly infects another. Introducing and analysing a simple evo-epidemiological model, here we study how such a trade-off is expected to affect evolution of the host ranges of influenza viruses. We examine a quantitative trait underlying host specificity, given by an influenza virus's degree of adaptation to certain conformations of sialic acid receptors, and investigate how this receptor preference evolves in a minimal network of host species, including humans, that differ in life history and receptor physiology. Using adaptive dynamics theory, we establish thresholds in interspecific transmission rates and host population sizes that govern the emergence and persistence of human-adapted viruses. These ecological thresholds turn out to be largely independent of the strength of the evolutionary trade-off, underscoring the importance of ecological conditions in determining a disease's host range.

Figure 1.

(a) Transmission structure of host community, highlighting receptor conformations in three host populations: reservoir hosts (waterfowl; r), intermediate hosts (pigs and chickens; m) and target hosts (humans; t). Population sizes in each class are denoted by N_i, with i = r, m, t. (b) Trade-off for receptor preference. The strength of the trade-off is given by s, with s < 1 characterizing a weak trade-off and s > 1 a strong trade-off. Moving away from the origin, the curves correspond to s = 1.5, 1, 0.75, 0.5, 0.25 and 0.05. Colours indicate the degree of specialization on the nearby receptor: red (high specialization), orange (low specialization) and blue (negligible specialization: generalists).

Figure 2.

Evolutionary outcomes in a neutral host ecology. (a) PIPs for different trade-off strengths s for N_t = N_m= N_r, c = 1, β_rr = β_mm = β_tt = 1/3 day⁻¹, ν_r = ν_m = ν_t = 1/6 day⁻¹ and γ_r = γ_m = γ_t = 1/180 day⁻¹. Black (white) areas indicate where the mutant has a positive (negative) growth rate in the endemic environment determined by the resident. Grey areas indicate regions in which the resident phenotype is not viable. (b) Trait evolution plots for the PIPs in (a). Grey areas indicate phenotype pairs that are mutually invasible and that therefore can coexist and coevolve. Black lines are evolutionary isoclines at which the selection pressure on one phenotype vanishes. Circles correspond to evolutionary attractors if filled and to evolutionary repellors if open. Arrows show the directions, at the quadrant level, of positive selection pressures (for better readability, such arrows are shown here only for the largest bounded regions).

Figure 3.

Conditions that permit the coexistence of perfect specialists, assuming frequency-dependent transmission, realistic ecological parameters of host populations (electronic supplementary material, table S1) and a linear trade-off (s = 1). Parameter combinations that permit specialist coexistence are in grey. Coexistence is evolutionarily stable for higher trade-offs (s = 0.75 and above), but not for weaker trade-offs; however, even at weaker trade-offs, extremely well-adapted viruses are able to coexist (see text, figure 2). (a) Effects of the relative population size N_m/N_r = N_m/N_t of intermediate hosts and of the degree c₁ of mixing between reservoir and intermediate hosts. (b) Effects of the relative population size N_t/N_r = N_t/N_m of target hosts and of the degree c₂ of mixing between intermediate and target hosts.