308 - Environmental and temporal partitioning of Culex mosquitoes and implications for West Nile virus transmission risk

Environmental and temporal variation strongly influence mosquito communities and shape West Nile virus (WNV) transmission risk. Because mosquito species differ in habitat use, seasonal phenology, host contact, which Culex species are present, and where and when they contribute to transmission, is essential for interpreting surveillance data and targeting control. However, in regions where multiple Culex species overlap, species-level identification remains a major barrier to resolving these patterns. This AMCARF-funded project addresses this bottleneck by developing efficient, scalable identification of Culex mosquito vectors of WNV using DNA-validated computer vision–based artificial intelligence (AI). Our long-term goal is to enhance arbovirus monitoring by enabling species-specific surveillance capable of resolving how different Culex species contribute to WNV risk across habitats, localities, and seasons. Here, we focus on the hypothesis that species occupying the same municipal landscape contribute to WNV transmission in distinct places and at distinct times, and that resolving species identity is required to detect these differences. During the 2024 field season, adult mosquitoes were collected through collaborations with the Salt Lake City Mosquito Abatement District and the Connecticut Agricultural Experiment Station and analyzed using morphological identification, molecular assays, and preliminary AI-based classification. In Salt Lake City, Utah, preliminary results indicate spatial and temporal separation between Culex tarsalis and Culex pipiens sensu lato, supporting distinct ecological roles in the WNV transmission cycle. Culex pipiens s. l. appears to drive enzootic amplification, while Cx. tarsalis may function as a key bridge vector contributing to human exposure. Additional resolution within the Cx. pipiens complex is forthcoming as DNA identification is completed and IDX computer-vision models are trained on accurately labeled image data. Collections of adult mosquitoes in CT were conducted at four urban and four rural sites. From the 1232 mosquitoes collected, only 13% could be identified to species level based on morphology alone, underscoring the limitations of conventional surveillance. Species confirmation using molecular methods is in progress. Mosquito abundance peaked in July, with Cx. restuans more abundant early in the season. The proportion of each species varied by location, with higher numbers of Cx. pipiens collected from urban areas. Together, these findings illustrate how species-resolved surveillance can reveal the specific places and times where control efforts can most effectively disrupt the WNV transmission cycle in heterogeneous urban, peri-urban, and natural landscapes. By enabling species-level resolution at scale, this approach provides a foundation for linking ecological context to species-specific vector index, and supports timely, targeted, efficient mosquito control strategies.