Temperature change gave no clear added skill beyond lagged volume for annual zooplankton change off two US coasts

We tested if a mismatch between temperature level and annual ecological change can reduce forecast skill. Under a stable linear temperature-level model with independent stationary temperature and residual processes, the explained variance for change from current temperature is (1-rho)/2 times that from temperature change, where rho is the lag-one temperature correlation. We then used published annual zooplankton indices from CalCOFI and EcoMon. Frozen early-period models predicted later annual log-volume changes, with observed test-year temperature. Temperature-change skill against mean change was 0.169 for CalCOFI and 0.088 for EcoMon. Both 95% block ranges included zero. Added skill beyond lagged volume was 0.151 and 0.050, also uncertain. Lagged volume alone gave EcoMon skill of 0.295 [0.192, 0.388]. Alternate training cutoffs gave negative temperature-change skill. Thus the tests do not establish added climate skill. Weak current-temperature forecasts alone cannot show that a temperature-level relation changed. Limits include two systems, short series, mixed-taxon volume, retrospective index construction and conditional use of observed temperature. This is a new analysis of previously published aggregates, not independent replication.