A sea-ice curve for Adélie penguin colony trends is confounded with region in 1984-2017; after 2017 it held on the Antarctic Peninsula, but Ross Sea colonies did not grow as it predicted

We tested the Iles et al. (2020) habitat-optimum curve, which relates mean March-June sea-ice concentration (SIC) within 500 km to long-term Adélie penguin colony growth. We used MAPPPD v5.0 counts and NSIDC Sea Ice Index v4 data. We recovered the unpublished SIC scale of the curve from the vector points of the paper's Fig. 3. In 1984-2017, the 38 long-series colonies are in three regions with SIC ranges that do not overlap. A refitted curve explained 49% of among-colony variance, region alone 53%, and region plus curve 50%. Leave-one-region-out predictions were worse than zero growth. Within the Peninsula (27 sites, 19-45% SIC), growth increased by 0.037 per year per 10 SIC points (SE 0.009), which agrees with the low-ice part of the curve. Ross Sea sites (51-74% SIC) gave no SIC slope. After 2017, the Peninsula regional index decreased by 4.1 ± 2.3% per year, as the curve predicted (-4.1%). The Ross Sea index changed from +1.3% per year (1985-2017) to -1.4 ± 1.6% per year (2017-2024), while the curve predicted +1.5% per year. This shortfall is probable but uncertain (z ≈ 1.8). All 15 Ross Sea site slopes were negative. No forecast of post-2017 counts was clearly better than a no-change forecast. The change in sea ice did not predict the change in growth (Spearman -0.01, n = 19). Nest-only, high-accuracy, ground/aerial-only and 100 km checks gave the same conclusions. Limitations: we used slopes, not the original state-space model; after 2017 only two regions have data, and colonies share year effects.