In KRILLBASE (SW Atlantic, 1976–2016), the krill density decline and the southward shift are not resolved within fixed cells sampled by the same program and net

Question: Do the reported decline and southward contraction of Antarctic krill in KRILLBASE net hauls (SW Atlantic sector) remain when each comparison holds the location and the sampling program and net constant? Most 1976–1989 hauls are from one German RMT8 program, and most later hauls are from US AMLR and Palmer LTER nets. Approach: 6,881 standardised hauls (1976–2016, December–March, 90°W–0°) and 652 hauls from 1926–1939. We fitted two-way season indices with no spatial control, with 1°×2° cell effects, and with cell × program × net effects, for arithmetic means, log10(x+1) and presence. Intervals are season-block bootstraps. We compared programs that sampled the same cell in the same season. We repeated the two-era (1976–1995 vs 1996–2016) matched-cell centroid test with cell × unit strata. Results: The arithmetic index fell from 1976–1989 to 1990–2003 by −0.50 log10 units (95% interval −0.94 to −0.05) without control and by −0.54 (−1.01 to +0.23) with cell effects. With cell × unit effects it changed by +0.63 (−0.27 to +1.54). In 42 shared cell-seasons, German RMT8 standardised densities exceeded AMLR IKMT densities by 0.62 log10 units (4.2 times; +0.26 to +0.98), and the difference was positive in all 5 seasons. No model showed a decline after 1990. With matched cells, the density centroid moved 352–510 km south (intervals exclude zero). With matched cell × unit strata, the shift was −330 to −411 km with intervals that include zero. Without the 1976 cruise, it was +232 or −249 km. On 125 cells sampled in 1926–1939 and 1976–2016, the centroid moved +54 km (−314 to +295), although the haul-weighted value moved 522 km south. Conclusion: with the current standardisation, these data do not separate a krill decline or contraction from the change of programs and nets. Limitations: within-unit designs lose power; the program offsets come from 5 seasons; the data end in 2016.

Antarctic krill · Euphausia superba · KRILLBASE · Southern Ocean · range shift · population decline · sampling bias · net standardisation · matched design · fixed effects · bootstrap · Scotia Sea · Antarctic Peninsula