<p>One of two models of disease transmission is generally assumed. In the density-dependent model hosts interact with one another like molecules in a chemical reaction such that rates of transmission increase with host density. Alternatively, hosts may have a set, density-independent contact rate. In this case transmission increases with frequency of infection in the population. These two forms of transmission have strikingly different consequences for the dynamics and outcome of epidemics, yet predicting which term fits a particular system a priori has been difficult at best. Moreover, transmission in natural systems often seems to follow some very different, intermediate model.
<p>I will present the results of experimental ranavirus epidemics in populations of wood frog tadpoles (<i>Rana sylvatica</i>). I then fit a suite of transmission models to data on the number of initially susceptible tadpoles that became infected in order to determine which model best described the actual transmission dynamics. I will then discuss which aspects of behavior are most likely to influence transmission dynamics, presenting preliminary data on behavioral observations and experiments.
<p><b>Results/Conclusions </b>
<p>The primary conclusion from these experiments was that the commonly assumed density-dependent model of transmission completely fails to describe actual transmission in these populations. The frequency-dependent model performs better, but is ultimately inadequate. Rather transmission followed a saturating function of densities of susceptible and infected hosts. Predator avoidance behaviors and habitat structure seem to alter contact patterns and thus transmission. These results strongly suggest that realistic epidemic models must account for realistic host behaviors.