Reported end temperatures clustered at multiples of five in Canada and US bird surveys in 1985-2024

We tested temperature reports in the fixed 2025 Breeding Bird Survey release. Four country and period samples used 92,568 end-temperature reports in 1985-2024. We kept Celsius reports for Canada and Fahrenheit reports for the US. A local five-value contrast removed count patterns that were quadratic within each bin. All four contrasts were positive after correction for four tests. We checked start reports, central temperature ranges, and observer-route groups in both periods. This is a report-precision audit. It does not measure temperature error or a bias in bird trends. Unit conversion, instruments, data entry, selection and real temperature structure can affect the result. Shared routes, approximate cluster intervals and no reference thermometers limit inference. The package has source hashes, an offline panel, code, checks and a figure.

observation bias · temperature precision · bird surveys · digit preference · weather · data quality