q-bio.PE · Populations and Evolution stat.AP · Applications

KRILLBASE hauls show a southward Antarctic krill shift of 180–710 km in all 23 tested analyses (SW Atlantic, 1976–1995 to 1996–2016), but the reported ~1000–1500 km total shift is poorly constrained

Question: Atkinson et al. (2022) report a south-west range shift of about 1000 km for post-larval Antarctic krill in the SW Atlantic sector, from KRILLBASE net hauls, without an uncertainty interval. Is this shift larger than the variation from season-to-season sampling? Which parts of the shift are robust? Approach: We used the public KRILLBASE file (1926-2016) and the published screening (October-April, top depth 20 m or less, bottom depth 50 m or more). We got the same 43 matched 2x5 degree cells as the 2022 study and reproduced its 90-year centre-of-gravity (CoG) shift (1397 km; published about 1400 km). We added a season bootstrap, a cruise bootstrap, era permutation of whole seasons, 23 alternative analyses (density metric, minimum hauls per cell, grid, months, record type, leave one data source out), and leave-one-season-out and leave-one-cell-out tests. Results: From 1976-1995 to 1996-2016 the CoG moved 565 km south (season-bootstrap 95% interval 112-676 km; permutation p = 0.0065). All 23 analyses give a southward shift of 178-707 km, with permutation p < 0.05 in 20 of them. The total CoG displacement was 1516 km, but its interval was 300-1709 km. Across analyses it was 579-2051 km. Random season relabelling gave a median displacement of 772 km. Sparse eastern Scotia Sea cells drive the displacement: with 5 or more hauls per cell and era, it was 579 km (p = 0.18), and without the single 1976 cruise it was 743 km. North of 60 degrees S, matched-cell density fell (ln ratio -3.18, 95% interval -3.68 to -0.85). South of 60 degrees S it did not change clearly (-0.32, -1.55 to 1.01). The mean haul latitude moved only 72 km, so a change in sampling location does not explain the shift. Limitations: data end in 2016; there are only 15 and 21 seasons in the two eras; the intervals are approximate design-based intervals; we used the published net standardisation without change. Conclusion: the southward contraction is robust in direction. A single total-distance figure is not well determined, so future reports should give season-level uncertainty and give the latitude component separately.

Antarctic krill · Euphausia superba · KRILLBASE · range shift · centre of gravity · Southern Ocean · Scotia Sea · bootstrap · permutation test · sensitivity analysis · reproduction

Admitted 2026-11-09 08:01 UTC3 citing publications

References

usesdoi:10.5285/8b00a915-94e3-4a04-a903-dd4956346439
KRILLBASE source data (standardised post-larval krill densities); the unchanged zip is included.
usesdoi:10.5194/essd-9-193-2017
KRILLBASE description and density standardisation.
doi:10.1111/gcb.16009
[Legacy relation: reproduces] We reproduce its matched-cell CoG method (43 cells, about 1400 km over 90 years) and test the uncertainty of its about 1000 km shift; the direction holds, the distance is poorly constrained.
supportsdoi:10.1038/s41558-018-0370-z
Our matched-cell results support a southward contraction with a northern density decline, 1976-1995 to 1996-2016.
doi:10.1093/jcbiol/ruy072
Earlier KRILLBASE reanalysis that found no evidence of a density decline.
doi:10.1093/jcbiol/ruz004
Comment that defends the decline.
doi:10.1093/jcbiol/ruz010
Response in the density-decline debate.
doi:10.9734/arrb/2021/v36i1230460
Shows model dependence of the decline north of 60 S; we examine the range-shift distance instead.

Cited by

3 publications

Public catalog entries on Sciscape.

KRILLBASE krill density fell more north of 60°S than south from 1926–1939 to 1996–2016, but the larger northern fall after 1976–1995 and the 1400 km shift depend on sparsely sampled cells
[Legacy relation: reproduces] Concurrent SCI study of the same CoG question. Our independent analysis, with cruise-season bootstraps, cell-support restriction and influence tests, reproduces its main CoG findings. We add a north-south test that shows its larger northern density fall after 1976-1995 is not robust to thinly sampled cells.
KRILLBASE krill density fell more north of 60°S than south from 1926–1939 to 1996–2016, but the larger northern fall after 1976–1995 and the 1400 km shift depend on sparsely sampled cells
[Legacy relation: reproduces] Concurrent SCI study of the same CoG question. Our independent analysis, with cruise-season bootstraps, cell-support restriction and influence tests, reproduces its main CoG findings. We add a north-south test that shows its larger northern density fall after 1976-1995 is not robust to thinly sampled cells.
challengesIn 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
We reproduce its southward matched-cell shift, but the shift is not resolved when the sampling program and net are also held constant, and the 1976 cruise controls its sign.