Patterns of mosquito and arbovirus community composition and ecological indexes of arboviral risk in the northeast United States

Background In the northeast United States (U.S.), mosquitoes transmit a number of arboviruses, including eastern equine encephalitis, Jamestown Canyon, and West Nile that pose an annual threat to human and animal health. Local transmission of each arbovirus may be driven by the involvement of multiple mosquito species; however, the specificity of these vector-virus associations has not been fully quantified. Methodology We used long-term surveillance data consistently collected over 18 years to evaluate mosquito and arbovirus community composition in the State of Connecticut (CT) based on land cover classifications and mosquito species-specific natural histories using community ecology approaches available in the R package VEGAN. We then used binomial-error generalized linear mixed effects models to quantify species-specific trends in arbovirus detections. Primary results The composition of mosquito communities throughout CT varied more among sites than among years, with variation in mosquito community composition among sites explained mostly by a forested-to-developed-land-cover gradient. Arboviral communities varied equally among sites and years, and only developed and forested wetland land cover classifications were associated with the composition of arbovirus detections among sites. Overall, the avian host arboviruses, mainly West Nile and eastern equine encephalitis, displayed the most specific associations among mosquito species and sites, while in contrast, the mammalian host arboviruses (including Cache Valley, Jamestown Canyon, and Potosi) associated with a more diverse mix of mosquito species and were widely distributed throughout CT. Conclusions We find that avian arboviruses act as vector specialists infecting a few key mosquito species that associate with discrete habitats, while mammalian arboviruses are largely vector generalists infecting a wide diversity of mosquito species and habitats in the region. These distinctions have important implications for the design and implementation of mosquito and arbovirus surveillance programs as well as mosquito control efforts. Author summary The numbers and types of mosquito species in a community vary across space and time, and this variability may be linked to variation in risk of human exposure to mosquito-borne diseases. In this report, the authors use a long-term mosquito-borne virus (i.e., arbovirus) surveillance dataset from the State of Connecticut (CT), United States to first examine whether or not the same land cover classes explain variation in mosquito and arbovirus community composition; they then used statistical models to determine which mosquito species were the most likely and unlikely to test positive for an infection for one of seven arboviruses circulating in CT. Multiple mosquito species tested positive for each arbovirus examined, and blood feeding behaviors were the strongest predictor of whether or not a mosquito species tested positive for a particular arbovirus. Isolations of bird-host arboviruses aggregated in discrete habitat types, while isolations of mammalian-host arboviruses showed no specific habitat associations. The authors conclude that risk of bird-host arbovirus exposures in humans is driven by only a few key mosquito species, whereas risk of mammalian-host arbovirus exposure in humans could be attributed to the general presence and abundance of mammal-feeding mosquito species in the community.