INTEGRATING EMPIRICAL RESULTS INTO EPIDEMIOLOGICAL MODELS
When considering effects of stress on infected host fitness and infectivity, responses varied depending on stressor type. Environmental stress decreased host survivorship and increased infection intensity, pollution decreased host survival and pathogen prevalence, and limiting resources decreased host reproduction and pathogen intensity.
We integrated the best-supported relationships from our meta-analysis into mathematical models to evaluate the net impact of these simultaneous effects of stressors on host-pathogen interactions. We built two dynamic Susceptible-Infected (SI) models. An SI-Resource model following the framework of Civitello et al. (2018) where key processes (i.e., host reproduction and pathogen transmission) could depend on resource availability (Box 1). And an SI-Environmental gradient model following the framework of Lafferty & Holt (2003) where key processes (i.e., host survivorship and pathogen transmission) could depend on an abiotic environmental factor (Box 2). Because our meta-analysis suggested no proportional difference between uninfected and infected hosts for survival or reproduction, we incorporated this result by including a common parameter for the strength of these effects on both groups (Box 1 and Box 2).
We used the models to determine equilibria of disease prevalence as a function of resource availability and environmental stress gradients, using the numerical integration function “lsoda” in the R packagedeSolve (Soetaert et al. 2010). We examine different scenarios in which fecundity and infectivity, or background death and infectivity, had different sensitivities to either resource (Box 1) or environmental stress gradients (Box 2), respectively. We simulated epidemiological dynamics of each model across a gradient of either resource availability or environmental stress, then plotted equilibrium infection prevalence and host density against such gradients for each model (Fig. 5).