Numerous models now exist that describe how weather and other drivers are expected to influence West Nile virus (WNV) risk within and across seasons in the United States. The dominant predictors vary substantially depending on region, methodology and spatial extent of the analysis, making it challenging to determine the most appropriate model to use for predicting risk. This uncertainty in the most important predictors to use in a model is compounded by variation in the values of temperature and precipitation across the small spatial scales (1m – 1 km) that are the most relevant to mosquitoes. Improving our capacity to predict how climate change may influence WNV risk requires a better understanding of the scale dependence of these relationships and greater collaboration across disciplines
Results/Conclusions
Here we show that the dominant associations between weather and West Nile virus risk vary with the spatial extent of the analysis (e.g., national, state, county). This is placed within the context of similar patterns observed in the literature. We also use data from two years of field collections in Jefferson and Larimer county Colorado to demonstrate that the temperature across microclimates available at a single wetland site can vary by as much as 5°C between full sun and shade. We show that both Culex tarsalis and Culex pipiens mosquitoes were found resting more frequently in cooler microhabitats shaded by overstory trees rather than in reedy or open grass microclimates (P < 0.01), but that there was no difference in the distribution of host-seeking mosquitoes across microclimates. The relationship between temperature and WNV risk is strongly non-linear such that the temperature-driven relative R0 can vary from near 0 in the shade to optimum in the sun. Taken together these results highlight the importance of the spatial scale of predictors and the spatial extent of the analysis for making vector-borne disease predictions.