Long-serving Breeding Bird Survey observers recorded fewer high-pitched birds; because cold-adapted species sing higher, this drift made 31-77% of the within-observer CTI warming trend (1966-2025)

Question: earlier campaign work found that the North American bird Community Temperature Index (CTI) followed less than 10% of warming, and that the CTI increased with observer tenure. We tested whether age-related loss of high-frequency hearing explains this observer signal. Data: 130,303 standard Breeding Bird Survey runs (2026 release, 1966-2025), GHCN_CAMS May-July temperature, and song frequencies of 323 species from Sagar et al. (2024, CC0). Methods: cross-fitted species temperature indices (STI); 5,740 observer changes on 2,941 routes, where we compared the incoming observer's first 3 runs with the outgoing observer's last 3 runs as a function of outgoing tenure; species x frequency models; CTI trends with route x observer and route fixed effects; a 150-draw route bootstrap. We wrote the plan before we calculated outcomes. Results: cold-adapted species sing higher (STI -0.57 C per kHz, p = 0.0005; within families -0.35, p = 0.09). After long-serving observers, new observers recorded more high-pitched birds. The high-frequency share increased by 0.075 percentage points per year of outgoing tenure (95% CI 0.060-0.089). The count gain increased steadily with frequency, from -0.02% per tenure year below 2 kHz to +1.6% above 7 kHz. The effect was present in East and West, before and after 2000, and with mass, abundance and order controls. The temperature placebo was null. The CTI decreased at observer changes by 0.0019 C per year of outgoing tenure (CI 0.0011-0.0025; event-year fixed effects). Thus within one observer the CTI drifts up by 0.011-0.019 C per decade. Low-frequency species had no drift. Corrections reduced the within-observer CTI trend (0.024 C per decade; tracking ratio 0.089) by 31-77%, to tracking ratios of 0.02-0.06. A CTI of species below 3.97 kHz had no drift and gave 0.091-0.111 tracking in both models. Conclusion: observer hearing-type detection loss is a measurable cause of apparent bird CTI warming. Route x observer fixed effects do not remove it. Weak tracking (about 10%) remains for low-pitched species. Limitations: observer ages are unknown; tenure is a proxy; cohort differences can remain; song frequencies come from xeno-canto recordings; the corrections are linear.