An ongoing public health priority is to understand ecological drivers of vector-borne diseases. Dengue fever and other emerging and reemerging mosquito-borne diseases are rapidly spreading across continents, as a potential consequence of climate change, land-use change, and globalization. Previous studies suggest that ecological factors, such as climate and population dynamics, affect vector ecology and disease transmission. However, such hypotheses are difficult to corroborate from observational data at the population level, due to confounder effects, nonlinear responses, and state dependence of other environmental and socioeconomic factors. To account for these challenges, we used methods based on nonlinear time series analysis, collectively referred to as empirical dynamic modeling (EDM), to analyze a 19-year long time series dataset of dengue incidence, climate, and estimated susceptible population size in San Juan, Puerto Rico.
Results/Conclusions
We identified ecological drivers of dengue and their interactive effects on dengue dynamics in San Juan, Puerto Rico. Estimated susceptible population size was the strongest driver of dengue incidence, and climatic forcing became important above a certain susceptible population size. In particular, temperature and rainfall had net positive and negative effects, respectively. Further, we present the first empirical mechanistic model that includes explicit climatic drivers and uses estimates of the susceptible population to predict dengue outbreaks successfully. The EDM protocol presented here for measuring and predicting how climate and population dynamics interact to drive epidemics adds to a growing body of empirical studies of complex, nonlinear ecological systems embedded in changing environments.