A factor-of-1.5 CO2-only bound left 46 of 63 ozone interaction signs unresolved in a 2024 invertebrate extract

We measured sensitivity to a missing CO2-only treatment mean in a public invertebrate extract. A prior campaign audit found 63 matched joint-exposure and ozone-only records from 11 papers. We rebuilt these pairs and calculated sharp bounds for each log-scale factorial interaction. If the missing CO2-only mean is between two-thirds and 1.5 times its control, 46 records permit zero interaction. Eight retain a positive sign and nine a negative sign. The unresolved counts are 23 of 38 for USA publications in 2001-2005, 20 of 21 for USA publications in 2010-2012, and three of four for one Chinese paper in 2018. A factor-of-1.2 scenario leaves 25 signs unresolved. The median factor needed to permit zero is 1.246. These bounds are conditional scenarios, not evidence-based limits on CO2 effects or sampling confidence intervals. Equal paper weights and broader matching retain the broad sensitivity result. Mixed endpoints, shared field programmes, selective extraction and one Chinese paper prevent independent biological replication. CO2 enrichment is not warming. Code, aggregate results, provenance and independent numerical checks are included.

ozone · CO2 · factorial interaction · partial identification · observation bias · invertebrates · sensitivity analysis