Independent reproduction: CalCOFI zooplankton volume at 66 stations was 21–26% below a 1951–1993 temperature relation in 1994–2021/2023, and the 1951–1993 decrease did not continue

Roemmich and McGowan (1995) related a large 1951–1993 decrease of CalCOFI zooplankton displacement volume off southern California to surface warming. We tested if the decrease continued and if the 1951–1993 relation to temperature predicts 1994–2023 volume. We used public NOAA ERDDAP volumes (15,945 tows at 66 core stations, 69 years), Scripps bottle temperatures (to May 2021) and ERSSTv5. Annual indices came from one regression on year, station, month and night. We wrote the plan before we calculated post-1993 results. Later we found SCI publication cffa5a7f, which did a similar test; we did not use its code, and our index, predictors and benchmarks differ. Results (natural log): the 1951–1993 trend was −0.020 per year (95% block bootstrap −0.034 to −0.004; 57% decrease). The 1994–2023 trend was +0.022 per year (−0.003 to +0.036); the change of trend excludes zero. The 1951–1993 relation to 10 m temperature overpredicted volume: mean residual −0.29 (−0.57 to −0.09) in 1994–2017 and −0.24 (−0.52 to −0.05) in 1994–2021, with training and test years resampled together. With ERSST the residual for 1994–2023 was −0.28 (−0.59 to −0.06). The result is the same with the Ohman–Smith gear correction, for small plankton, and with T10 − T200 stratification. The year-to-year slope on 10 m temperature was −0.49 per °C in 1951–1993 and −0.17 in 1994–2021; the difference is not resolved for this primary combination. The 1984–1993 mean had the lowest out-of-sample error in 1994–2017 (RMSE 0.33 against 0.44 for the trend and for the temperature relation), but this advantage is not resolved. In 2018–2023 total volume agreed with the ERSST relation (+0.12, −0.17 to +0.40), but small plankton stayed low (−0.39). Our estimates agree with cffa5a7f within their intervals. Thus a level decrease near 1977–1980 and a weaker temperature response fit the data better than a continuous warming-driven decrease. Limits: displacement volume is not carbon and is strongly affected by salps; the data cover southern California; sampling was sparse in 1969–1984 and after 2017; gear factors come from few paired tows; both studies use the same public data.

CalCOFI · zooplankton displacement volume · California Current · out-of-sample test · reproduction · ocean warming · stratification · regime shift · gear correction · long-term ecological data · ERSSTv5