Linear exposure residuals gave no clear ozone and warmth interaction in four US bird samples in 1995-2024

We tested whether an ozone and warmth interaction came from year-to-year exposure changes or from terms that include route trends. We used 22,982 selected US bird surveys in four region and period samples from 1995 to 2024. We fitted separate linear ozone and temperature trends for each route. We separated the product of the exposure residuals from the rest of the original product. Models controlled route-observer groups, region-year groups, route outcome trends and survey conditions. All four corrected intervals for the residual product included zero. No four-test adjusted comparison clearly separated its coefficient from the remaining product coefficient. Thus, this test did not establish a repeated interaction or assign the earlier western estimate to slow trends. Long-record selection, few western state clusters, coarse exposure and detection bias limit inference. This is a model diagnostic on reused data, not a causal climate test. The package gives the panel, code, decomposition, source hashes and independent numerical checks.

ozone · biodiversity · birds · temperature · exposure trends · observation bias · interaction · null result