In US Breeding Bird Survey counts (1990–2024), Blue Jay and House Wren decreased in the 3 years after high state West Nile virus incidence, while American Crow counts also rose before such years

Question: Do years of high West Nile virus (WNV) incidence still decrease North American Breeding Bird Survey (BBS) counts, and do states with intense early epidemics keep a long-run deficit? Approach: We used 73,856 BBS standard surveys on 3,710 routes in the 48 conterminous US states (1990–2024) and CDC ArboNET state-year counts of human WNV disease (1999–2024). For 290 species we fitted Poisson models with route x observer and year effects and state trends. The regressors were leads and lags of x = log10(1 + cases per 100,000). Because the BBS is in June and transmission peaks in August–September, the leads are placebo terms. We calibrated each species estimate against the spread of all other species, and we used a state-permutation test. A second design compared states by early epidemic dose, with event windows to 25 years after onset. Results: Blue Jay counts were lower in the 3 surveys after high-incidence years (sum of lags -0.34, SE 0.07; with weather controls -0.38, SE 0.06), with leads near zero. Blue Jay was 4th of 290 species after calibration (z = -3.0), and no state permutation gave a value as low (p = 0.005). The decrease was present in 1999–2005 and in 2006–2023. House Wren (-0.21 to -0.22) and, less robustly, Tufted Titmouse also decreased. American Crow lags were negative (-0.15 to -0.23; -0.33 in 2006–2023), but crow counts were also higher before high-incidence years (lead sum +0.14 to +0.17, SE 0.04), so the crow result fails the placebo test at state scale. Eastern Bluebird increased after such years, which shows confounding by other year effects. The 12 a-priori susceptible species differed only weakly from the other 278 species (difference -0.08 to -0.12, p = 0.17 and 0.04). In the dose design, crows in high-dose states stayed lower 16–25 years after onset with no recovery (-0.14 per 10x dose, SE 0.07), but many other species had regional slopes of the same size. Compared with a parallel county-scale study (sci:sha256:8e3945c31a4c010a34772eab2b32488192ff42a9f01f7bf2884cb6661a862687), the crow and Black-capped Chickadee estimates agree, and the Blue Jay, titmouse and wren estimates are 2–4 times larger at state scale. Limitations: Human incidence is an indirect measure of bird infection; state scale mixes different areas; state-clustered SEs are about 1.6 times too small for single species; weather controls and the permutation test were added after the first results.

West Nile virus · Breeding Bird Survey · Blue Jay · American Crow · wildlife disease · lead placebo · Poisson fixed effects · empirical null calibration · permutation test · CDC ArboNET · population recovery · event study