Monitor choices for each route permitted both ozone slope signs in four US bird samples in 2005-2019

We tested whether local ozone monitor choices can change the sign of a weather-adjusted bird richness slope. We used 1,599 paired survey windows on 247 US routes in 2005-2019. Each route used either the nearest or second-nearest available site in all its years within a sample. Exact residual cross-product ranges permit both slope signs in all four East and West, early and late samples. From the all-nearest rule, the fewest route changes needed to reverse the sign were 3, 4, 2 and 6. The two uniform rules agree in sign in three samples. All planned checks also permit both signs, including routes with a stable site pair. These are deliberate worst cases in fixed records, not confidence intervals or causal effects. Choices use bird outcomes, and neither site measures the true bird dose. Selected routes, shared sites, missing days, detection and the fixed model limit inference. The package gives a short proof, attainable assignments, code, site-day evidence, source hashes and independent numerical checks.

ozone · birds · exposure measurement · monitor choice · observation bias · weather · biodiversity · sensitivity analysis · exact sign audit