Empirical mapping of suitability to dengue fever in Mexico using species distribution modeling

Abstract

Dengue is considered the most important vector borne virus disease worldwide placing some 2.5 billion people at risk globally. Despite the public health concern about dengue fever, spatially explicit suitability assessments for this disease are limited due to data restrictions and the challenges posed by the complexity of the interactions among its risk factors, which involve social, economic, and ecological processes. This paper demonstrates an empirical approach to identify suitable areas for dengue fever using species distribution modeling and evaluates the relative contribution of climatic and socio-economic factors as dengue fever suitability determinants. Several models showing the potential distribution of dengue fever within all the Mexican municipalities are produced using different sets of predictor variables. The results suggest that at the scale of this study the climatic variables were more important determinants of suitability for dengue fever than the socio-economic variables considered in this study. All the models perform well (average testing AUC about 0.8) and show similar patterns. The model with the least number of variables and best performance includes the variables minimum temperature of the coldest month, mean temperature of the wettest quarter, and annual precipitation. However, there is not a high variability of AUC scores among the models generated.