Exploring possible causes of a dilution effect using agent-based models

Abstract

The relationship between infectious diseases and ecosystems has become a much-discussed topic, with many emerging infectious diseases originating in animal reservoirs. A controversial subject within ecological epidemiology is the dilution effect. It supposes a negative relationship exists between a given ecosystem’s biodiversity and the risk of infectious disease transmission. An appealing prospect in theory, the dilution effect provides a public health incentive for conservation. This idea is particularly relevant for infectious diseases caused by pathogens circulated within natural reservoirs, such as influenza and ebolaviruses. The converse phenomenon, where biodiversity increases the risk, is known as the amplification effect.

The debate over the dilution effect has been fervent, particularly surrounding its generalisability and scale dependence. We address these concerns by combining popular Ordinary Differential Equation (ODE) models with two classes of agent-based models, one set on a lattice and the other spatially explicit. Agent-based models are uncommon in ecoepidemiology and have yet to be applied to the topic of dilution. We apply existing methods for quantifying the dilution effect in ODEs to numerical data from simulations and highlight circumstances where observations of dilution and amplification are sensitive to the selected definitions of biodiversity and infection risk. We find that a lattice-based approach is well-suited to capturing spatiotemporal dynamics on longer time scales, while the spatially explicit method effectively describes outbreaks on shorter time scales. We also use a lattice-based model to explore the possible mechanisms of the dilution effect in horticulture.

We also highlight two infectious diseases to analyse in more detail: toxoplasmosis and Lyme disease. For toxoplasmosis, we discuss how pathogen influence on prey behaviour can affect infection risk, showing an amplification effect for an example system. With Lyme disease, we discuss the influence of a vector and show how control strategies associated with long-term mitigation of risk can cause a short-term increase in infection. We also show the diluting influence of adding a less competent host, concluding that the dilution effect is not a general phenomenon but a product of scale and the unique properties of individual infections and ecologies.