A 1983-2008 sea-temperature slope did not forecast 2009-2026 adult length-at-age of North Sea cod, haddock, whiting and saithe; before 2009, warmer years gave longer adults

Earlier studies (Baudron et al. 2014; Ikpewe et al. 2021) linked North Sea warming to shorter adult and longer juvenile gadoids. We did a forward test with ICES NS-IBTS survey data (Q1 1983-2026, 16,371 hauls; Q3 1992-2025, 10,055 hauls). We calculated length-at-age with an age-length key weighted by survey length frequencies in fixed areas, which removes a 2-4 cm bias of raw otolith means. Exposure was the lifetime mean North Sea SST. In the training period, adult length-at-age decreased by 7.6% per 1 C (Q1, 95% CI -10.5 to -4.5) and 6.1% (Q3). This slope predicted a 4.2% decrease for 2009-2026, but the observed change was +0.3%. The SST forecast was worse than a no-change forecast (skill -0.30 in Q1, -0.31 in Q3; ERSSTv5 gave -0.41 and -0.44), and northern and central-southern series gave the same or a null result. After removal of linear trends, warmer years before 2009 gave longer adults (+8.5% per C in Q1, +7.7% in Q3), so the training link came from common trends. Models with survey density, ICES fishing mortality or both also did not beat no change for gadoid adults. Juveniles became longer after 2008 in Q1 (+5.9%), more than the SST forecast (+1.1%); the Q3 juvenile forecast had no skill, and the juvenile SST slope was zero with fishing mortality in the model. Herring, Norway pout and plaice adults followed survey density better than SST. Adult gadoids became shorter in 2018-2026 (-10.9% per decade), but nine years cannot identify the cause. Limitations: surface SST, simple density and fishing terms, possible otolith-reading changes, and short test periods.