Passing vehicles hid low-pitched, warm-adapted birds at North American Breeding Bird Survey stops in 1997-2025, but traffic grew too little to change the bird temperature index trend by over 1%

Question: bird community temperature indices (CTI) of the North American Breeding Bird Survey (BBS) follow only a small part of the spatial CTI-temperature slope. Known detection artefacts make the CTI trend larger. Can roadside traffic, which masks low-pitched songs, make it smaller? Data: BBS 2026 release, 1997-2025: 3.1 million stop counts in 62,179 runs on about 4,700 routes with the number of vehicles that passed during each 3-minute count; song frequencies of Sagar et al. (2024, CC0); GHCN_CAMS May-July temperature. Method (plan written before outcomes): Poisson pseudo-likelihood and linear models with stop x observer and run fixed effects, so that each stop is compared with itself on other mornings; 334 per-species models and a random-effects meta-regression; vehicle and CTI trends with route x observer effects; 100-draw route bootstrap; a traffic-standardized CTI. Results: each passing vehicle reduced counts by 0.83%, by 1.99% below 2 kHz and by 0.50% at 4-5 kHz. Species loss decreased by 0.40 percentage points per kHz (95% CI 0.26-0.57; inside families 0.26). Warm-adapted species sing lower (-0.81 C per kHz), and each vehicle lowered the stop CTI by 0.0019 C (CI 0.0012-0.0027). The observer noise flag reduced counts by about 13%. Traffic per stop increased by only 0.072 vehicles per decade (about +27% since 1998). Thus the bias of the within-observer CTI trend (0.071 C per decade; temperature 0.328) was -0.00014 C per decade (CI -0.00021 to -0.00007), 0.19% of the trend (0.26-0.30% with concave forms) and about 0.15% of the gap to the spatial expectation (0.165). The traffic-standardized trend differed by +0.00003 C per decade, and the spatial slope did not change (0.504). Vehicles remove low-pitched birds, but long observer tenure removes high-pitched birds; this opposite signature supports a hearing mechanism for the tenure drift. Conclusion: traffic masking is real but does not hide climate tracking; the known detection artefacts do not explain weak tracking. Limitations: constant noise from distant roads and long-term road avoidance are not measured; no vehicle data before 1997; 17 species lack song frequency.