One added species in 5.69% of surveys can reverse an ozone by warmth slope in the western US in 1995-2009

We tested how missed species can change four published ozone by warmth slopes for recorded bird richness in US surveys in 1995-2024. We held samples, survey controls and fixed effects constant. The error model permits one extra species in each of at most k surveys, with no restriction on its link to exposure. Exact sorted regression weights give the minimum k for a strict point-sign reversal. In the early West, 118 of 2,074 surveys (5.69%) change the slope from -0.551 to +0.003 species per 10 ppb ozone and 2 C warmth. The thresholds are 1 and 98 surveys in the two eastern samples and 21 in the later West. Sparse full-model fits and integer optimization confirm all thresholds. A constant one-species correction in every survey changes no slope. These are coordinated worst-case bounds, not observed detection errors, confidence limits or causal climate effects. The other three original sampling intervals already included zero. Selected surveys, coarse exposures and unknown missed-species processes limit ecological use. Code, aggregate results, hashes, a figure and external witness regeneration are provided.

ozone · temperature · birds · species richness · observation bias · integer bounds · sensitivity analysis