A bird survey-temperature slope change above 15 C was clear only in the later western US sample in 1990-2024

We tested whether the slope of recorded bird taxa against survey start temperature changes above 15 C. We reused 9,876 accepted surveys in four US states and made 7,068 consecutive-year pairs with the same route and observer. Separate models for two regions and two periods controlled for survey timing, effort, weather, and state-year changes. All four fitted slope changes were negative. Only the 2007-2024 western change excluded zero after correction for four tests: -1.93 taxa per 5 C, with an interval from -3.31 to -0.56. Its fitted cool and warm slopes were +0.76 and -1.17 taxa per 5 C. The change remained clear with end-temperature control but became uncertain with close-date and fair-weather restrictions. This result supports a test of curved observation models. It does not establish a common biological threshold. Selected accepted runs, sparse warm western records, model choices, and unknown detection rates limit inference. We did not test pollution or extreme heat effects.

biodiversity · observation bias · Breeding Bird Survey · temperature · nonlinear response · replication