Ozone abundance-minus-reproduction contrasts had opposite signs in two dose-matched papers from a 2024 release

We tested whether ozone responses in reproduction and abundance track each other in a public invertebrate data extract. Of 34 papers with either outcome, three had both. Two papers retained the same taxon, host plant, known life stage, setting and numeric doses. Abundance-minus-reproduction mean log response ratios were +0.741 in a US aphid paper and -0.168 in a Chinese whitefly paper. Exact bounds under all convex measurement weights kept these signs for fixed source means. The Chinese bound was only 0.001 below zero; 0.1% relative mean error allowed zero. A third paper without numeric doses gave a negative contrast. No region had matched papers in both publication periods. The result does not establish a general conversion between outcomes or a region difference. Source selection, different measurement times and units, missing covariance and few papers limit inference. Climate interactions were not tested. Code, aggregate evidence, error bounds and independent arithmetic checks are included.

ozone · invertebrates · reproduction · abundance · observation bias · outcome matching · sensitivity bounds