Predictive modeling of mosquito abundance and West Nile virus transmission dynamics: Insights from a decade of surveillance in Arizona

West Nile Virus (WNV) represents a significant public and environmental health concern in the United States. Although the virus is maintained in the environment at low viral levels in some bird species, which remain relatively unaffected by the virus, other bird populations such as corvids are decimated. The virus is transmitted primarily through the bites of infected mosquitoes, particularly those of Culex species. Mosquito abundance plays an important role in the transmission dynamics of WNV, with factors such as weather patterns, habitat, and demographics influencing vector populations. There is a growing need for sophisticated models that can integrate ecological and epidemiological data to better predict mosquito abundance and WNV transmission dynamics. Accordingly, this study uses spatial analysis and machine learning algorithms to predict the abundance of two Culex species, C. quinquefasciatus and C. tarsalis, across a ten-year time span in the state of Arizona. Using a total of 71,416 mosquito surveillance data points collected between 2013 and 2023, we created models that incorporate time lagged weather data, habitat variables, demographics, and building structure data. Additionally, we created a model predicting the occurrence of WNV in surveilled mosquitoes. These models, which had high predictive capabilities, can be used in combination with traditional surveillance methods to better alert public health officials and wildlife conservation scientists to human and animal health risks.