Epidemiological models are vital tools for understanding and managing the spread of parasites among hosts. These models always contain at least one transmission function that describes how susceptible hosts interact with infected hosts or infectious agents in the environment, and this transmission function determines the dynamics and stability of the host-parasite system. Historically, just two canonical transmission functions were used by ecologists: the density dependent (DD) and frequency dependent (FD) functions. However, in the past two decades, many nonlinear transmission functions that fall somewhere between the DD and FD functions have been proposed. To better understand when and why ecologists use each transmission function, we reviewed 222 papers from the recent ecological literature. We quantified how often each function was used to model parasites and pathogens with sexual transmission, vector transmission, environmental transmission, and non-venereal, direct transmission. We also searched for empirical tests of multiple transmission functions to quantify which functions perform best for parasites and pathogens with any given transmission mode. Finally, we performed extensive model simulations to identify the conditions under which a flexible NL transmission function should outperform the canonical DD and FD transmission functions.
Results/Conclusions
In our review of the recent ecological literature, we found that nonlinear transmission functions were rarely used. However, nonlinear transmission functions usually provided more parsimonious fits to empirical data than the canonical transmission functions when multiple transmission functions were compared (19/23 papers). In our simulations, the flexible, nonlinear transmission function could always recover the best possible fits to our simulated epidemic data and always accurately estimated known model parameters. Conversely, the FD and DD functions produced poor model fits and biased parameter estimates for a large region of the parameter space that we considered in our simulations. The limited number of existing empirical tests of multiple transmission functions makes it difficult to know a priori which transmission function should be used for any given host-pathogen system. However, the majority (84%) of papers in the ecological literature used a single transmission function, rather than comparing models with different transmission functions. We recommend that in the future, authors consider multiple transmission functions unless there is a strong a priori justification for a particular transmission function, and we emphasize the utility of flexible, nonlinear transmission functions that can capture a wide range of possible transmission dynamics.