Climate and urbanization structure West Nile virus vector communities: Tools for surveillance under environmental change

Problem Statement
West Nile virus (WNv) in eastern North America is primarily transmitted by two cryptic mosquito species, Culex pipiens and Culex restuans. While these species are routinely pooled in surveillance programs as a single "Cx. pipiens/restuans" index due to morphological similarities, they exhibit distinct phenological, thermal, and habitat preferences. Culex restuans dominates early in the season and in vegetated habitats, whereas Culex pipiens peaks later in the summer and thrives in highly urbanized environments. This seasonal turnover is epidemiologically significant, as the shift to Cx. pipiens coincides with the period of highest human WNv risk. However, pooled surveillance indices fail to capture these opposing responses to environmental change. As climate warming and urban expansion alter temperature and land-use gradients, the relative abundance of these vectors is shifting, yet agencies lack species-resolved tools to anticipate when and where vector communities will reorganize.

Methodology
The study integrated data from three eastern metropolitan regions: a multi-year, spatially balanced designed study in the Baltimore-Washington (BW) region (2019–2024), and complementary datasets from Philadelphia and Chicago drawn from existing operational collections. A total of 9,789 Culex specimens were molecularly identified.

The analysis employed two primary statistical approaches:

1. Abundance Modeling (Baltimore-Washington only): Negative-binomial generalized linear mixed models (GLMMs) were used to model species-specific abundance against landscape predictors (e.g., impervious surface, tree canopy, elevation) and climatic predictors (e.g., weekly mean temperature, accumulated degree-days). A species-factor model was fitted to test for divergent responses between the two species.
2. Compositional Turnover (All three regions): Gradient Forest (GF) models, a machine-learning approach, were applied to quantify nonlinear compositional turnover along environmental gradients. These models identified thresholds where the community shifts most rapidly from Cx. restuans to Cx. pipiens.
3. Spatial Prediction: In the BW region, GF-derived biological importance values were transformed via Principal Component Analysis (PCA) and used in beta-binomial generalized linear models to map predicted species composition across the landscape for early and late seasonal windows.

Key Results

• Divergent Environmental Responses: The two species responded in opposite directions to both urbanization and temperature. Cx. pipiens abundance increased with higher impervious surface and warmer temperatures, while Cx. restuans abundance declined along both gradients. Both species showed a positive association with elevation, though Cx. pipiens responded more steeply.
• Thermal Thresholds for Turnover: Accumulated degree-days (ADD) and weekly mean temperature were the most consistent drivers of species turnover across all three regions. The transition from a Cx. restuans-dominated community to a Cx. pipiens-dominated community occurred most rapidly within a narrow thermal window of 594 to 610 accumulated degree-days (corresponding to weekly mean temperatures of approximately 21–22°C). This threshold was consistent across Baltimore-Washington, Chicago, and Philadelphia, despite differences in sampling design and landscape context.
• Landscape Drivers: In Baltimore-Washington, landscape predictors (impervious surface, tree canopy, NDVI) were strong drivers of spatial turnover, with the most rapid shifts occurring in suburban areas (roughly 20–50% impervious surface). In Philadelphia, population density served as a proxy for the urbanization gradient, while in Chicago, limited landscape variation meant weather predictors dominated the turnover signal.
• Spatial Patterns: Spatial predictions indicated that highly urbanized areas remained Cx. pipiens-rich throughout the season, whereas suburban landscapes exhibited the largest seasonal shifts in community composition.

Significance and Operational Tools
The authors argue that these findings provide a framework for moving from reactive to anticipatory WNv management under environmental change. The study proposes three specific operational tools for vector control agencies:

1. Climate-Based Forecasting: The identified degree-day window (594–610 ADD) serves as a transferable thermal threshold. Agencies can use temperature data to forecast the impending shift to Cx. pipiens dominance before it is detectable in trap composition, allowing for proactive adjustment of surveillance timing and intensity.
2. Spatial Targeting: The identification of suburban landscapes (intermediate impervious surface) as the zones of greatest seasonal turnover suggests these areas should be prioritized for intensified surveillance and early-season intervention, rather than focusing uniformly on urban cores where composition is static.
3. Species-Resolved Projections: The estimated species-specific environmental responses allow agencies to interpret existing pooled surveillance data in a species-resolved manner. This enables the anticipation of how vector communities may reorganize in response to future climate warming and land-use changes, even without immediate species-level identification in every trap.

The paper concludes that while the pooled index has served surveillance for decades, it is increasingly insufficient for tracking the reorganization of vector communities driven by shifting environmental gradients. The proposed tools offer a path toward spatially targeted, forward-looking management of WNv risk.