Short-term prediction of Culex quinquefasciatus abundance in Central North Georgia, USA, based on the meteorological variability

Culex quinquefasciatus is the main vector of West Nile Virus (WNV) in the southeast USA, and inter-annual variation in this vector abundance is mainly related to meteorological variability. In this study, short-term effects of meteorological conditions on seasonal variation in the vector abundance in the central north part of the State of Georgia, USA, from 2002 to 2009 were assessed. Four weeks moving average temperature, precipitation, potential evapotranspiration, and available moisture in the surface layer of soil were considered as risk factors. Cross-correlation maps were developed to investigate influences of preceding environmental conditions during a time-lagged interval on mosquito count data. The Poisson regression model and Artificial Neural Network (ANN) model were used for prediction purposes. Two sets of predictors were used: (1) the interval lagged climate data with the highest correlation and (2) single time lag antecedent Culex mosquito abundance up to 10 weeks prior to the events combined with lagged climate data. Results revealed that both models predicted the seasonal cycle of vector abundance fairly accurately, with ANN performing better than the regression model. The addition of antecedent mosquito data as input improved the prediction power of both models. The developed predictive models can be helpful in informed decision-making when high WNV activities are anticipated.