Prior-year cold pool area did not change groundfish northing in eastern Bering Sea trawl surveys in 1982-2026; a northward offset of about 9 km stayed after 2000

Question: do eastern Bering Sea fish and invertebrate distributions keep a northward shift after warm periods (hysteresis), or do they follow the summer cold pool with no memory? Approach: we used the NOAA AFSC bottom trawl survey (1982-2026, 44 surveys) on the 271 stations sampled in every year, with area-swept CPUE and zero catches. For 17 taxa we calculated the biomass-weighted centre of gravity. We modelled northing with the official cold pool area (2 C or less) and log abundance. Specified tests: (1) the effect of the previous-survey cold pool when the current cold pool is held fixed; (2) out-of-sample residuals in the cold years after two warm periods (2006-2013 after 2001-2005; 2022-2026 after 2014-2021), compared with cross-validated training blocks. Results: the previous cold pool had no clear effect (+0.3 km north per 100,000 km2, 95% CI -2.4 to +3.1; current-year effect -3.3 km). In 2022-2026 the taxa were at their predicted positions (community z -0.18, rank p 0.55). Exploratory (post hoc) result: the mean northing residual moved up by about 9 km between 1999 and 2000 (HAC SE 1.4 km; break scan sup-F 21.6 at 2000) and stayed up through both later cold periods, including 2026. This shift was present in 11 of 17 taxa (largest for Pacific cod, walleye pollock, flathead sole, arrowtooth flounder and Pacific halibut). It stayed after controls for survey date, fixed station positions and taxon set. Vessel, gear and route changes do not explain it. Limitations: one survey region; the step was found by inspection; causes such as age structure, prey or unrecorded observation changes are not identified; the memory test cannot exclude effects smaller than about 70% of the current-year cold pool effect.

Bering Sea · cold pool · groundfish · distribution shift · hysteresis · bottom trawl survey · centre of gravity · marine heatwave · climate change · structural break · observation bias