2018 Big Sky GeoCon

Predicting the patterns of vector-borne disease background rates and outbreaks not only indicates an understanding of the disease process but has the pragmatic benefit of disease control and prevention. We applied niche modeling tools to presence-only data using historic and ongoing mosquito surveillance for West Nile Virus (WNV) in the state of Montana. With mosquito infection rates varying between non-detection to as high as 15%, Montana represents a region on the periphery of WNV distribution with high spatial and inter-annual variation in WNV infection. Data on climate, bird-host distribution and land cover were used to build a habitat suitability model for the mosquito vector Culex tarsalis, and a WNV risk model for the state. Not surprisingly, Cx. tarsalis habitat suitability was strongly associated with WNV risk but could not explain inter-annual variation in infection rates. Differences in heating degree units (HDUs) best explained ecoregion and inter-annual variation in WNV risk. The Great Plains and intermountain valleys had significantly higher cumulative HDUs than the Rocky Mountain regions and weekly-cumulative HDUs were significantly higher for the 3 years with the greatest mosquito infection rates compared with the 3 years with the lowest mosquito infection rates. We suggest that together with the Cx. tarsalis habitat suitability model, weekly-cumulative HDUs are sufficient to discriminate geographic regions and years that are most susceptible to WNV outbreak. Furthermore, combined with diligent mosquito surveillance efforts, outbreaks may be predicted and possibly prevented weeks before infection of humans or livestock.