Influence of vectors' risk-spreading strategies and environmental stochasticity on the epidemiology and evolution of vector-borne diseases: the example of Chagas' disease

Abstract

Insects are known to display strategies that spread the risk of encountering unfavorable conditions, thereby decreasing the extinction probability of genetic lineages in unpredictable environments. To what extent these strategies influence the epidemiology and evolution of vector-borne diseases in stochastic environments is largely unknown. In triatomines, the vectors of the parasite Trypanosoma cruzi, the etiological agent of Chagas' disease, juvenile development time varies between individuals and such variation most likely decreases the extinction risk of vector populations in stochastic environments. We developed a simplified multi-stage vector-borne SI epidemiological model to investigate how vector risk-spreading strategies and environmental stochasticity influence the prevalence and evolution of a parasite. This model is based on available knowledge on triatomine biodemography, but its conceptual outcomes apply, to a certain extent, to other vector-borne diseases. Model comparisons between deterministic and stochastic settings led to the conclusion that environmental stochasticity, vector risk-spreading strategies (in particular an increase in the length and variability of development time) and their interaction have drastic consequences on vector population dynamics, disease prevalence, and the relative short-term evolution of parasite virulence. Our work shows that stochastic environments and associated risk-spreading strategies can increase the prevalence of vector-borne diseases and favor the invasion of more virulent parasite strains on relatively short evolutionary timescales. This study raises new questions and challenges in a context of increasingly unpredictable environmental variations as a result of global climate change and human interventions such as habitat destruction or vector control.

Figure 1. Schematic representation of the one-parasite-strain version of the model.

Because vectors are divided into two stages (juvenile and adult), and that both stages can get and transmit the parasite, we present first a simplified vector-borne epidemiological model with only one stage for vectors (panel a) and then the vector life-cycle (panel b). Hosts are represented with dashed lines and vectors with solid lines (a) Susceptible hosts H_s get infected through contacts with infected vectors V_i with probability Φ_h, and susceptible vectors V_s through contacts with infected hosts H_i with probability Φ_v. Susceptible and infected vectors and hosts give birth to susceptible vectors and hosts. Infected host survival S_hi is a function of parasite virulence α. Environmental stochasticity is applied to vector survival (only adults, see below) with intensity ε_s. (b) Susceptible and infected adult vectors V_as and V_ai give birth to susceptible juvenile vectors V_js. Susceptible and infected juvenile vectors V_js and V_ji remain in the juvenile stage with probability P_j and mature into adults with probability (1-P_j). Only adult survival S_va is submitted to stochasticity with intensity ε_s. See text, Table 1, Appendix S1 in File S1 and Fig. S1 for further details.

Figure 2. Influence of vector life-history traits on vector population dynamics and parasite prevalence.

In each panel, the upper, middle and lower graphs display, respectively: total vector density, parasite prevalence in vectors, and parasite prevalence in hosts, according to the proportion of juvenile vectors prolonging the juvenile stage P_j. Red and black lines correspond, respectively, to simulation results in the deterministic (shown at t = 150,000 weeks) and stochastic (shown at t = 10,000 weeks as median values over the 100 simulations, plotted only if the number of simulations without extinction is ≥5) case; dotted lines (open circles), dashed lines (closed circles) and solid lines (triangles) to a relatively low (S_vj = 0.6), intermediate (S_vj = 0.8) and high (S_vj = 0.95) juvenile survival. Left and right panels correspond, respectively, to a relatively low (S_va = 0.6; panels a, c) and relatively high (S_va = 0.95; panels b, d) adult survival; upper and lower panels to a relatively low (w_v = 1; panels a, b) and relatively high (w_v = 2.5; panels c, d) fecundity. The persistence probability of vector populations (proportion of simulations for which vector density does not collapse before the end of the simulation), is given in grey in the upper graphs showing vector density. For prevalence among hosts, all simulations (including those for which vector populations collapse) are taken into account. For vectors (density and prevalence), only simulations for which vector population persisted until the end of the simulation are considered. Other parameters values are: α_r = 0.008, β = 0.005, c = 0.01, S_hs = 0.994, w_h = 0.05, g = 100, q = 50, b_max = 1, ρ = 0.5, p_b = 0.2, ε_S = 0.1.

Figure 3. Dynamics of mutant invasion (proportion of mutants among infections according to time after mutant introduction).

Proportion of the mutant is calculated as the number of hosts infected by the mutant parasite divided by the total number of infected hosts, and given as a median among all simulations for which the vector density did not collapse at the simulation time considered. Red and black lines correspond, respectively, to results in the deterministic and stochastic case (grey lines: persistence probability of vector populations); solid lines (solid circles) and dashed lines (open circles) to a relatively efficient and less efficient risk-spreading strategy. Panels a, b, c: S_va = 0.6, w_v = 2.5, less efficient risk-spreading strategy: P_j = 0.3, more efficient: P_j = 0.8; panels d, e, f: S_va = 0.95, w_v = 1, less efficient risk-spreading strategy: P_j = 0.4, more efficient: P_j = 0.8. Other parameters values are: β = 0.005, c = 0.01, S_hs = 0.994, w_h = 0.05, g = 100, q = 50, b_max = 1, ρ = 0.5, p_b = 0.2, ε_S = 0.1, S_vj = 0.95, α_m = 0.008.