Role of Climate, Land Use/Cover and Water Quality on West Nile Virus Incidence: A Modeling Approach

Abstract

West Nile virus (WNV), a vector-borne infectious disease, has been a major public health concern in North America since 1999. This virus is transmitted to susceptible mosquitoes when they feed on infected birds. Infected mosquitoes spread the virus to humans and other animals when they bite. To control mosquito-borne diseases, it is necessary to identify the locations of mosquito breeding sites and to monitor changes in mosquito population under different environmental conditions. The focus of this study was to investigate the impacts of different risk factors on Culex quinquefasciatus population in the central north part of the State of Georgia and particularly in the Atlanta metropolitan area. The main risk factors considered in this study were climate variability, Land use/cover (LULC) types and their impact on water quality and streamflow. To demonstrate which specific components of water chemistry are conducive to breeding Culex mosquitoes, a mesocosm experiment was designed. The emergence pattern of Culex mosquitoes was found to be strongly related to certain nutrients, and results showed that breeding sites with higher PO4 or NO3 concentrations have higher survival rate of larvae. High NO3 concentrations favor the development of male mosquitoes and suppress the development of female mosquitoes, but those adult females that do emerge, develop faster in containers with high NO3 levels compared to the reference group. Also, the addition of PO4 in the absence of nitrogen sources to the larval habitat slowed larval development, however, it took less days for larvae to reach the pupal stage in containers with combination of NO3 and PO4 or NH4 and PO4 nutrients. In addition, short term effects of climate conditions on seasonal variation of Culex mosquito abundance and their infection rate in the central north part of Georgia from 2002 to 2009 were assessed. The Poisson regression model and Artificial Neural Network (ANN) model were used for the prediction purposes. Statistical analysis revealed that increasing temperature and PET and decreasing surface moisture in preceding late winter and preceding spring increased Culex quinquefasciatus female mosquitoes abundance in summer/early fall about 2 times as many and also increased the number of infectious mosquitoes about 3.5 times. Also low precipitation in late winter decreased mosquito abundance in summer. However, above average temperature in late winter and early spring coupled with below average precipitation favors the incidence of WNV in mosquitoes. Both ANN and regression models predicted the seasonal cycle of mosquito abundance fairly accurate. Addition of antecedent mosquito count data or infection rate as predictors improved the prediction power of both models by increasing ENASH values and decreasing RBIAS values. To examine the relationship between LULC and various water quality parameters and to predict water quality in unmonitored watersheds in Atlanta area, an ANN-based model was applied. Streamflow and water quality data from neighboring USGS stations in the Atlanta area with leave-one-site-out jackknifing technique were used to build the predictive models for PO4, NH4 and NO3 loading values. NO3, NH4 and PO4 predictive models with best performance had ENASH values of 0.99, 0.89, and 0.66 respectively and RBIAS values of 8%, -6% and -7% respectively. No general trend was observed between percent imperviousness or percent forest cover or watershed size and the model performances. Also, a lumped model (ANN) and semi-distributed watershed model, Soil & Water Assessment Tool (SWAT), were combined to improve ANN performance for predicting flow during warm and cool. 62% of runs for predicting flow during cool season and 83% of runs for predicting flow during warm season had “good” to “very good” performance ratings. The developed predictive models can be used for a more accurate warning of high-risk periods for WNV and could have important implications for the control of West Nile Virus spread by Culex mosquito species. Also, the findings of this study can help reduce the costs and efforts required for effective mosquito vector control by focusing on areas with higher risk.