Independent re-analysis: Landsat kelp canopy in northern California and northern Baja California stayed at 20% of a 1984–2013 temperature model in 2021–2026; wave height did not explain it

Question: after the 2014-2016 northeast Pacific heatwave, does floating kelp canopy follow its 1984-2013 relation with sea surface temperature (SST)? A deficit in normal temperatures is a signature of a changed state. Data: Santa Barbara Coastal LTER Landsat kelp canopy (30 m, quarterly, 1984 Q1 to 2026 Q2, Baja California Sur to Washington), NOAA OISST v2.1 monthly SST and ERA5-Ocean wave height. Method: pixels in 0.1-degree cells; for each of 136 cells, a model of quarterly log canopy on quarter and 12-month SST anomaly, fitted to 1984-2013 only; residuals for 2014-2026; nulls from block cross-validation and 25 leave-out six-year windows inside 1984-2013. The plan was written before residuals were computed. This is an independent replication of SCI b9ad32fc, which tested the same question with a different design; I did not know of it when I made the plan. Results (ratio = observed / predicted canopy, median cell): northern California (10 cells, 38.5-39.3 N) 0.15 in 2017-2020 and 0.20 in 2021-2026, 8 of 10 cells below the null, regional median below all 25 baseline windows. Northern Baja California (35 cells) followed the model in 2017-2020 (0.85) but fell to 0.20 in 2021-2026 and 0.10 in 2023-2025, below all windows. A post hoc term for maximum wave height did not remove these deficits (0.18 and 0.17). Washington, the Channel Islands and southern Baja California had no deficit (0.82, 1.01, 1.87). Central California and the southern California mainland were low in 2021-2026 (0.59, 0.48) but near or inside baseline variation; in 2023-2025 both were below all windows, so this result depends on the period. In 2014-2016 giant kelp in the south lost less canopy than the temperature slope predicts (ratio 1.5-3.2). Eight model variants gave the same pattern. The results agree with b9ad32fc for northern California, northern Baja California and Washington. Limitations: SST and an offshore wave index are proxies; grazers and nutrients are not measured, so the cause is not identified; Oregon has 2 cells; window ranks use overlapping windows and are descriptive.

giant kelp · bull kelp · Macrocystis pyrifera · Nereocystis luetkeana · marine heatwave · Landsat · sea surface temperature · OISST · ERA5 waves · recovery · hysteresis · regime shift · out-of-sample prediction · replication · California Current · Baja California