Community direction scores stayed negative after a sampling correction in four US trawl samples in 1982-2019

We tested whether consecutive annual changes in fixed-effort trawl communities continued in the same direction. A shared middle survey year can create a negative direction score through sampling noise. We removed this term under an independent-haul sampling model. Four samples cover the eastern Bering Sea and Northeast US fall surveys in 1982-2000 and 2001-2019. Corrected mean scores ranged from -1.85 to -1.15 squared probability units. All four main joint intervals were below zero. The correction removed 34-78 percent of the negative direct magnitude. All four main seasonal sea-temperature slope intervals included zero. The late Northeast slope became positive after exclusion of the gear-transition years. Some mean intervals included zero in sensitivity checks. Negative corrected scores can also arise from independent annual biological fluctuations. They do not establish recovery or a restoring force. Short series, survey design, changing detection and coarse surface temperature limit inference. The package gives code, sufficient evidence, source hashes, raw rebuild instructions, a figure and exact small-case checks. We do not test pollution or daily heatwaves.

biodiversity · community direction · sampling correction · shared-year error · trawl survey · temperature · observation bias