Beyond frequency and density-dependence: An experimental demonstration of the importance of non-linear transmission dynamics in a host-macroparasite system

Background/Question/Methods Understanding the form of pathogen transmission is important for modeling disease impacts on host population dynamics, forecasting disease persistence in host populations and establishment in new populations, and understanding the evolution of virulence. An embedded assumption in contemporary disease ecology is that the form of transmission is either density or frequency dependent. However, there are surprisingly few attempts to empirically test transmission forms, particularly for competing models beyond frequency and density dependent transmission. Here, we used a novel experimental approach in which we vary four different factors (duration of exposure, numbers of parasites, numbers of hosts, and parasite density) influencing transmission. We investigated transmission using the trematode Ribeiroia ondatraeand larval amphibian hosts. This macroparasite system offers several advantages as a model system (e.g., ease of manipulating infective stages and lack of intrahost replication). Furthermore, by manipulating host behavior we were able to isolate the influence of parasite behavior on transmission dynamics. We evaluated seven candidate transmission functions using maximum likelihood methods to identify the best fitting model to the experimental data according to Akaike’s information criterion. 

Results/Conclusions Our results indicated that, among the candidate models considered, non-linear forms of transmission involving either a power law or negative binomial function were the best fitting models and consistently outperformed classical density and frequency dependent functions. The power law function was the best fitting model for experiments varying the duration of exposure and host number, while transmission dynamics in experiments varying parasite number independently of parasite density was best represented by the negative binomial function. These functional forms are consistent with saturating infection with high parasite exposures. The negative binomial function remained the best fitting model when parasite behavior was isolated from host behavior using anesthetized hosts. Upon re-analysis of previous empirical data from other macroparasite systems, we found that non-linear functions were a superior fit to the data relative to density or frequency dependence, suggesting that non-linear transmission dynamics are general across multiple host-parasite systems. These functions highlight important parallels with models of other species interactions, including predator-prey and host-parasitoid dynamics. Suggested mechanisms for non-linear transmission include heterogeneity in susceptibility or distribution or density-dependence in the parasite population. Our results have implications for disease management and provide a basis for conceptually integrating models for pathogen and consumer resource systems.