Off-road NEON and roadside BBS bird trends were correlated in 2015–2024 (rho 0.54 per site, 0.83 per species), but NEON site indices were noisier and the mean difference was not robust

Question: do off-road NEON breeding-bird point counts and roadside North American Breeding Bird Survey (BBS) routes give the same species trends at the same places? Data: NEON Breeding landbird point counts (Darwin Core Archive, 2026-09-15) and the BBS 2025 release, 2015–2024 without 2020, BBS routes that start within 50 km of each NEON site. Method: for each species and site, Poisson trend models with point (NEON) or route x observer (BBS) intercepts and a first-year-observer term; REML estimate of year-level variance beyond counting noise, added to trend errors; a bivariate measurement-error model for the correlation of true trends (rho); site bootstrap; BBS and NEON split-half benchmarks; 9 sensitivity variants. Results: for 915 species-site pairs (38 sites, 239 species), raw r was 0.19 and rho 0.54 (95% interval 0.42–0.64); for 108 species at 3 or more sites, rho was 0.83 (0.65–0.95). Two halves of the BBS routes near one site agreed at rho 0.88–1.00, but NEON agreed with one half at 0.49–0.52. The SD of year effects beyond counting noise was 0.31 for a NEON site, 0.26 for one BBS route and 0.11 for pooled routes; one species at one NEON site thus needs about 20 years to detect 3.4% per year with 80% power. BBS trends were on average 2.1% per year lower than NEON trends (−3.0 to −1.1), but only 1.0% lower with route instead of route x observer intercepts and not significant in 2017–2024; the difference did not depend on habitat density. Year-effect changes were larger after a change of the main NEON observer with a lenient filter (p = 7e-5), but not within sites or with a strict filter. Limitation: 5–9 years per site, and local dynamics, landscape, roads and observers are confounded.

NEON · Breeding Bird Survey · bird population trends · point counts · roadside bias · observer effects · monitoring design · measurement error · statistical power