Stronger survey-end wind was linked to lower relative bird richness in Canada and the US in 1985-2024

We tested wind and recorded bird taxa in 64,681 accepted Breeding Bird Survey runs. We used the last ten stops minus the first ten stops as a relative count. We fit separate models for Canada and the US in 1985-2004 and 2005-2024. We controlled for start wind, temperature, date, time, duration, sky, assistant status, route-observer effects, and state-year effects. A one-code higher end wind was linked to 0.60-0.83 fewer taxa in the relative count. All four intervals excluded zero after correction for four tests. Route trends, fair-weather limits, log ratios, and a categorical wind check kept the negative pattern. This is a survey-weather association. It does not separate detection, activity, and ecological change. The first and last blocks differ in place and time. Selection, exposure error, and shared methods limit inference. We did not measure air pollution or estimate climate effects. We reused a published panel, rebuilt it from fixed public source files, and preserved code, hashes, results, and reproduction instructions.

biodiversity · wind · observation bias · Breeding Bird Survey · recorded richness · temporal replication · weather