Q2: Sensitivity of host fitness and infectivity responses to stress
We contrasted the fitness and infectivity effects of stressors on infected hosts. The full model, including stressor type, response trait, and their interaction, was also the model with the lowest AICc score (Table S3). In this model, the interaction arises not only due to the differential sensitivity of fecundity and survivorship responses to stressor type but also because the direction of infectivity responses only aligned with fitness responses in the case of endogenous environmental stressors (Table S4; Fig. 3). Perturbation of the endogenous environment tended to have negative consequences for hosts, both in terms of survival and pathogen intensity (Table S4; Fig. 3, all infectivity vs. fitness contrasts p > 0.05; however, survivorship vs. prevalence: p = 0.057). In contrast, the effects of resource limitation differed significantly between response variables (fecundity vs. intensity: p < 0.001; fecundity vs. prevalence: p = 0.006; survivorship vs. infection intensity: p = 0.010; and survivorship vs. prevalence: p > 0.05). When resources were limited, not only was host fecundity reduced (as noted in Q1), but infection intensity was also reduced (Table S4; Fig. 3). Finally, the effects of chemical pollution were more negative on survivorship than on either proxy of infectivity (survivorship vs. infection intensity: p = 0.024, survivorship vs. prevalence: p = 0.018). We found that pollution decreased both host survival and pathogen prevalence (Table S4).
We obtained a similar pattern of interaction among stressors and fitness and infectivity responses when the RVE was used to account for the non-independence of sampling errors (Fig. S3). Despite these contrasting effects of moderators, heterogeneity remained high (total I2 = 91.33%), both between (I2 = 55.86%) and within (I2 = 35.47%) experiments.