A positive wind and temperature term did not repeat across four US bird richness samples in 1990-2025

We tested whether wind changed the survey-temperature slope of recorded bird richness. We used 45,527 consecutive-year survey pairs with the same route and observer in the western and eastern US in 1990-2006 and 2007-2025. Models controlled survey timing, weather, observer experience, and state-year changes. The wind and temperature terms were -0.030, -0.047, +0.107 and +0.162 recorded taxa per 5 C per mean Beaufort code. Only the later eastern interval excluded zero after correction for four tests. That interval included zero with state-cluster uncertainty or start-only weather. The evidence does not give a common interaction across regions and periods. This is a survey-weather audit, not a climate effect estimate. Selected roadside routes, missing fields, coarse wind codes and unmeasured local change limit inference. Code, source hashes, exact model inputs and numerical checks are included.

biodiversity · weather · observation bias · Breeding Bird Survey · temperature · wind · replication · first differences