Landsat kelp canopy in 2017–2025 stayed below a 1984–2013 SST and wave model in northern California (7%) and northern Baja (31%), and by total area in central California (53%), but not in Washington

Question: after the 2014-2016 northeast Pacific marine heatwave, did floating kelp canopy return to the level that ocean conditions predict? Data: Santa Barbara Coastal LTER Landsat kelp canopy (knb-lter-sbc.74.34, 30 m pixels, quarterly, 1984 to 2026 Q2, Baja California to Washington), NOAA OISST v2.1 monthly SST and winter wave height from 18 NDBC buoys. Method: we added canopy into 0.25-degree cells and used July-September canopy. For 8 regions we fit cell fixed-effects models of log canopy on three SST anomalies and a regional winter wave anomaly with 1984-2013 data only. We compared the ratio of summed observed to summed expected canopy in 2017-2025 with the 95% range of 9-year means of leave-one-year-out ratios before 2014 (moving-block bootstrap). Results (108 cells): in northern California (bull kelp), canopy was 6% of its 1984-2013 mean, and the ratio to the model was 0.067 (fit bootstrap 0.046-0.12; normal range 0.27-1.30). Canopy was below 25% of the mean in 11 of 12 years in 2014-2025, but never more than 2 years in sequence before 2014, while SST was only 0.28 C above the mean and waves were below the mean. In northern Baja the ratio was 0.31 (normal range 0.55-1.31), and canopy was 5-12% of the mean in 2023-2025, which were cool years. In central California the ratio of sums was 0.53 (normal range 0.69-1.24); three Monterey Peninsula and Big Sur cells had 79% of this deficit. With equal cell weights the central California value was 0.67, inside normal error, which agrees with SCI publication b9ad32fc. Washington (1.19) was not below the model, so a coast-wide sensor artefact is unlikely. Other regions were inside normal error or not robust. In 12 model variants the ratio was 0.04-0.19 in northern California and 0.20-0.45 in northern Baja; central California was 0.44-0.64 and outside normal error in 11 variants. Limitations: the model has low cross-validated skill (-0.26 to 0.38), so the expectation is near the 1984-2013 mean with small SST and wave adjustments; Landsat misses sparse near-shore canopy; grazers and nutrients are not measured, so the study does not identify the cause.

giant kelp · bull kelp · Macrocystis pyrifera · Nereocystis luetkeana · marine heatwave · Landsat · OISST · wave height · NDBC buoys · kelp recovery · out-of-sample prediction · regime shift · urchin barrens · California Current · Baja California · reproduction