Linking anthropogenic resources to wildlife-pathogen dynamics: a review and meta-analysis

Abstract

Urbanisation and agriculture cause declines for many wildlife, but some species benefit from novel resources, especially food, provided in human-dominated habitats. Resulting shifts in wildlife ecology can alter infectious disease dynamics and create opportunities for cross-species transmission, yet predicting host-pathogen responses to resource provisioning is challenging. Factors enhancing transmission, such as increased aggregation, could be offset by better host immunity due to improved nutrition. Here, we conduct a review and meta-analysis to show that food provisioning results in highly heterogeneous infection outcomes that depend on pathogen type and anthropogenic food source. We also find empirical support for behavioural and immune mechanisms through which human-provided resources alter host exposure and tolerance to pathogens. A review of recent theoretical models of resource provisioning and infection dynamics shows that changes in host contact rates and immunity produce strong non-linear responses in pathogen invasion and prevalence. By integrating results of our meta-analysis back into a theoretical framework, we find provisioning amplifies pathogen invasion under increased host aggregation and tolerance, but reduces transmission if provisioned food decreases dietary exposure to parasites. These results carry implications for wildlife disease management and highlight areas for future work, such as how resource shifts might affect virulence evolution.

Keywords: Aggregation; agriculture; foraging ecology; host-parasite interactions; immune defence; infectious disease ecology; mathematical models; supplemental feeding; urbanisation.

Figure 1

Predicted relationships between provisioning and R₀ (where R₀ = 1 is the pathogen invasion threshold). Aggregation around resources could increase host contact rates and infectious stage build-up in the environment (a; orange), an effect illustrated by increased flocking of house finches at bird feeders and associated increases in conjunctivitis prevalence (b; Altizer et al. 2004). Provisioning can also improve host vital rates and increase host population sizes (a; green), which was suggested to explain higher pathogen prevalence among bumblebees in urban versus rural gardens (c; Goulson et al. 2012). Positive effects of provisioning on R₀ could be countered by improved host condition and immune defence (a; purple). Such an effect is suggested by kit foxes showing lower nutritional stress, higher body condition, and improved immune function in urban areas where food and water was more plentiful (d; Cypher & Frost 1999). Images are provided by Wikimedia Commons.

Figure 2

Distribution of effect sizes for observed relationships between provisioning and infection outcomes (points ± 95% confidence intervals) alongside the mean effect size estimate (diamond) from the bias-corrected REM (a). Each point is a particular host–pathogen interaction. Points above the horizontal line demonstrate cases where provisioning increased infection prevalence, intensity or seroprevalence; points below the horizontal line demonstrate reduced infection outcomes. (b) Estimated mean effect size of predictors on infection outcomes, denoted through diamonds alongside 95% confidence intervals. Sample size (n) refers to the number of host–pathogen interactions corresponding to each level. Positive effect sizes indicate increases in infection outcomes (measures of prevalence, seroprevalence and intensity are pooled).

Figure 3

Visualisation of the MEM explaining the most variation in infection outcomes from the meta-analysis. Data points represent the predicted outcome of provisioning for each combination of food source (see legend) and pathogen type, where the horizontal line represents no influence of supplemental feeding on infection. Asterisks represent means significantly different from zero after adjusting for multiple comparisons (*P < 0.05, **P < 0.01). Effects based on agricultural food and fungal pathogens are not shown owing to limited data.

Figure 4

General modelling framework for how provisioning affects infectious disease dynamics of a microparasite (Box 1). In this compartmental framework (a–b), provisioning causes key parameters to increase (shown in blue) or decrease (shown in red). Varying the response of immune parameters to provisioning generates a range of outcomes on R₀ (c). An increasingly saturating effect of provisioning is shown through line width (dashed indicates no effect on immunity), and this approach can generate outcomes ranging from amplifying prevalence to driving R₀ below the invasion threshold (grey line). Figure is adapted from Becker & Hall (2014), and further model details and parameter definitions are provided in Box 1.

Figure 5

Meta-analysis-guided re-assessment of provisioning effects on pathogen invasion via mathematical models. Simulations examine net effects of resource-mediated processes on R₀ by considering two independent behavioural mechanisms supported by our analysis, in which provisioning either elevates contact rates (a) or decreases dietary exposure to pathogens (b). Along with incorporating the above effects and those of resource-altered resistance and tolerance, the model includes potential influence of resource-altered demography, where line width indicates how strongly birth and mortality parameters respond to provisioning (shown in the legend). Simulations follow the parameterisation given in Becker & Hall (2014), and the analytic expression for R₀ is provided in the Supplemental Material.