Model predictions
Using our dynamical models (Box 1 and Box 2), we evaluated whether the patterns of trait sensitivity to stressors we documented in the meta-analysis reduce or increase infection prevalence across stress gradients and how stressors ultimately impact host population densities. The SI-Resource model predicts that a decrease in resource productivity decreases infection prevalence (Fig. 5A), in part because host densities also decrease with limited resources (Fig. 5C). Once a pathogen establishes in a population, there is stabilizing feedback, where pathogens suppress host density, increasing resources, and further increasing transmission. Therefore, in all scenarios of sensitivity of pathogen transmissibility to resources (smaller values of the half-saturation transmission constant (ht) increase the sensitivity of transmission rate (β) to resources), the model reaches the same prevalence equilibrium. However, although population density also stabilizes, the impacts on host density are different for each scenario: populations that are more sensitive to resources available will reach smaller population sizes compared to less sensitive populations (Fig. 5C)
The SI-Environmental stress gradient models show that population density decreases regardless of the effects of stress on hosts susceptibility due to increased mortality. But it exponentially decreases host populations when transmission rate is sensitive to the environmental factor (Fig. 5B and D). Specifically, when stress increases host susceptibility (i.e., greater values of βE), infection prevalence will increase rapidly (Fig. 5B) but at the cost of increasing host mortality (Fig. 5D). Therefore, infection prevalence will have a maximum at intermediate stress level but will drop as population densities are too low to sustain transmission. In contrast, as transmission is more negatively affected by stressors (i.e., pathogens are negatively affected by stressors), infection prevalence will quickly reach zero with increasing environmental stress (Fig. 5B). But as the stress increase and persist, populations will decline after the pathogen is extirpated from the system (Fig. 5D).
Our models illustrate that the consequences of stress gradients on disease can depend on the sensitivity that host traits, such as births and deaths, and shared host-pathogen traits, such as transmission (i.e., β) have to stressors. Interestingly, and consistent with Lafferty & Holt (2003) simulations, our models showed that increased environmental stress generally decreased disease, mainly driven by host density reductions. Although stress can make hosts more likely to become infected at the individual level, at the population level, negative impacts on host survival and reproduction may be driving pathogen and host local extinctions (Lafferty & Holt 2003).