Utilizing routine mosquito control surveillance data to investigate the ecology of eastern equine encephalitis virus and West Nile Virus in northeastern Florida

Long-term mosquito and sentinel chicken surveillance data provide the potential to investigate environmental drivers of disease transmission at multiple scales. In northeastern Florida, two important arboviruses, eastern equine encephalitis virus (EEEV) and West Nile virus (WNV), exhibit heterogeneous transmission patterns. Here, we investigated environmental impacts on sentinel chicken seroconversion for EEEV and WNV at 27 locations in northeastern Florida, using a combination of abiotic and landscape variables. We executed Bernoulli generalized linear mixed effects models that included a spatiotemporal term in the R-INLA environment, and we generated weekly Gaussian random fields (GRFS) for 28 sampling weeks between May 1 and November 12, 2018. Model results indicated that forest edge density within 500 m of coops, percent forest within 5000 m, and percent cypress/tupelo wetland coverage within 3500 m had a positive effect on EEEV seroconversion in the study area, which is consistent with known habitats for Culiseta melanura, the primary vector of EEEV. Weekly GRFs for EEEV showed strong spatiotemporal structure at the beginning of the study period. Results for WNV seroconversion did not identify strong effects from environmental variables, but a clear spatiotemporal pattern underlying seroconversion in the study region occurred late in the season. Visualizing the underlying spatiotemporal structure of seroconversion can provide clues about additional variables to investigate in future modeling efforts. Although we analyzed 28 weeks of data, model outputs demonstrate great potential to maximize information gained from large-scale temporal and spatial data to investigate arbovirus ecology.