A 2024 invertebrate data extract lacks the CO2-only means needed to test CO2 by ozone interactions

We audited treatment support for CO2 by ozone interactions in the public extract from Ryalls et al. (2024). All 66 combined-exposure rows from 11 publications match an ozone-only row by study and shared control. Of these, 63 also match on biological and measurement fields. All have equal control means and control sample sizes. The extract contains no CO2-only rows. None supplies all four means for a factorial interaction. This limit occurs in both US publication periods and the one Chinese publication. A rank check and a constructive calculation show that either interaction sign fits the same three supplied cells. Two primary papers contain the fourth treatment, so this is a limit of the extract, not proof that the experiments are absent. Selective extraction, shared experiments, mixed outcomes and publication years limit wider use. We do not estimate warming effects or infer no biological interaction. Code, aggregate results, source hashes and independent checks are included. External records remain outside the package.

air pollution · ozone · CO2 · invertebrates · factorial interaction · identification · observation bias · data reuse