Detection and Mitigation of Spatial and Temporal Bias in Species Distribution Modeling for Seasonal Populations

This project examines evidences about the role of environmental and socioeconomic factors to the presence of Aedes albopictus – a vector species of Zika and other infectious diseases – in the Southeast Pennsylvania. Previous studies analyzing the invasion of A. albopictus in temperate regions, such as the Northeastern United States have predicted an increase in mosquito distribution associated with increased temperatures and changing climate. However, there is limited knowledge about how human activities and in particular, the growing patterns of urbanization in temperate areas can influence the distribution of the Aedes genus mosquitos in these latitudes. The study region includes one of the largest metropolitan areas in US (Philadelphia) with diverse population as well as large areas of rural landscapes characterized through variation in microclimate and terrain parameters, giving optimal conditions for examining interactions between human activities, natural and build environments and mosquito presence. Our analysis combines state-of-the-art machine learning methods with data from the US Census, spatial data derived from satellite observations and field information in order to evaluate most influential site factors on A. albopictus presence. Additionally, we develop a strategy for bias mitigation in mosquito collection data, which are essential for effective and accurate species distribution modeling.