Winter warmth was linked to changes between positive bird counts in eastern North America in 1967-2024, with uncertain western results

We tested whether winter temperature links to bird count changes occur only when records change between zero and nonzero. We rebuilt 6,385,153 species pairs from the BBS 2025 release and used signed climate and winter-class tables from prior SCI work. The main test used 94,845 same-observer survey pairs on 4,634 North American routes in 1967-2024. It compared winter-present with winter-absent species, with route and year effects and seasonal temperature controls. Total log-count change was split exactly into a component from two positive counts and a component with one zero. The positive-pair slope was +0.589 (97.5% interval 0.327 to 0.851) and the zero-transition slope was +0.330 (0.151 to 0.509), in 100 log-count units per C. The positive-pair interval was above zero in both eastern periods after four-test correction. Both western intervals included zero. Survey date and weather controls, two class systems and removal of Carolina Wren retained the pooled result. Thus zero transitions did not account for the full pooled or eastern association. Detection intensity, movement and abundance remain unresolved. These are count-index associations, not survival effects or causal climate-change estimates. The package preserves data, code, source hashes, uncertainty, independent numerical checks and reproduction instructions.

winter weather · bird counts · observation bias · Breeding Bird Survey · count decomposition · regional replication · winter presence