In 2014–2024 Palmer LTER net tows, a 1993–2013 ENSO model predicted pteropod abundance, but the 5-year krill cycle and four other published relations did not

Question: do the 1993–2013 macrozooplankton relations reported for the Palmer LTER grid west of the Antarctic Peninsula (Steinberg et al. 2015) predict the nine later January–February surveys (2014–2020, 2023, 2024)? Approach: we wrote a protocol with six predictions before we looked at the new abundances. We rebuilt the annual abundance anomalies (1,450 tows; grid lines 200–600 as the primary domain) with the published method. We fitted each model to 1993–2013 only and extrapolated it. Each test used a correlation with a permutation p, a skill score against the 1993–2013 mean, and eight index variants, including an independently harmonised data set. Results: the 1993–2013 fits reproduced the published values (krill ~ primary production 2 years before, R² = 0.73; 5-year krill period, permutation p = 0.002; pteropod ~ winter ENSO, R² = 0.58). Only the pteropod (Limacina rangii) ~ prior-winter MEI model predicted 2014–2024: r = 0.79 (p = 0.008), skill 0.60, and r = 0.76–0.85 in all variants. The fixed 5-year krill cycle (r = −0.25, skill −2.1) and the krill–production model (r = −0.47, skill −6.9, n = 6) failed, although the krill–production fit was stronger than the best of 53 candidate predictors by chance. The salp ~ SAM + ENSO, ice krill ~ SAM + ENSO and Thysanoessa macrura trend models had no skill. Krill recruitment pulses continued (2018, 2024), but later than the 5-year phase predicted; the best period for 1993–2024 recruitment is 5.6 years (p = 0.019). Limitations: the test has only 6–9 years. There were fewer tows after 2009. We used the CPC SAM index and MEI.v2, and we tested the published final predictors, not a full repeat of the stepwise search.

Antarctic krill · Euphausia superba · Limacina · Salpa thompsoni · pteropod · ENSO · Southern Annular Mode · western Antarctic Peninsula · Palmer LTER · out-of-sample test · zooplankton · recruitment cycle · primary production · forecast skill