Abstract
Predicting risk of zoonotic diseases, i.e., diseases shared by humans and animals, is often complicated by the population ecology of wildlife host(s). We here demonstrate how ecological knowledge of a disease system can be used for early prediction of human risk using Puumala hantavirus (PUUV) in bank voles (Myodes glareolus), which causes Nephropathia epidemica (NE) in humans, as a model system. Bank vole populations at northern latitudes exhibit multiannual fluctuations in density and spatial distribution, a phenomenon that has been studied extensively. Nevertheless, existing studies predict NE incidence only a few months before an outbreak. We used a time series on cyclic bank vole population density (1972-2013), their PUUV infection rates (1979-1986; 2003-2013), and NE incidence in Sweden (1990-2013). Depending on the relationship between vole density and infection prevalence (proportion of infected animals), either overall density of bank voles or the density of infected bank voles may be used to predict seasonal NE incidence. The density and spatial distribution of voles at density minima of a population cycle contribute to the early warning of NE risk later at its cyclic peak. When bank voles remain relatively widespread in the landscape during cyclic minima, PUUV can spread from a high baseline during a cycle, culminating in high prevalence in bank voles and potentially high NE risk during peak densities.