Early-half shift rates of 1,404 fish populations do not predict their late-half rates of latitude or depth change in 25 bottom-trawl survey units (1967–2021)
Question: do the centroid shift rates of marine fish populations persist through time? Approach: we used FishGlob_data v2.1 bottom-trawl data from 25 survey units in North America and Europe (166,815 hauls, 1967-2021). For each population (species in a survey unit), we calculated annual biomass-weighted latitude and depth centroids and compared the OLS shift rate in the first half of the record with the rate in the second half (8-26 sampled years per half). The analysis plan was written before the main test. Results: for 1,404 populations, early-half latitude rates did not predict late-half rates (within-unit Spearman -0.08; random-effects meta-analysis r = -0.080, 95% CI -0.148 to -0.012; 46% same direction). Of 262 populations with a clear early trend (p < 0.05), 52.7% moved in the same direction later. Extension of the early linear trend gave 3.4 times the mean squared error of a no-change forecast of the late-half mean centroid (skill -2.36, 95% CI -2.76 to -1.75), and it was better in none of 25 units. Depth gave the same result (meta r = -0.056; trend skill -5.5). Four planned sensitivity analyses did not change the result. A simulated constant-trend model with noise-corrected trends predicts a correlation of 0.05 to 0.18 and 56-60% same-direction shifts; the data are outside this range. Adjacent-window correlations stayed near zero for windows of 5 to 20 years. The centroid series behave as fluctuations with partial return to a mean, not as steady trends. Limitations: rates are relative to the unit mean, so common shifts of a whole assemblage are removed (unit-mean rates showed weak persistence, Spearman 0.40, p = 0.047, exploratory); centroids are inside the survey footprint; we did not test temperature; half-record rates are noisy (split-half reliability 0.27-0.38 for latitude).
range shifts · species distributions · bottom trawl surveys · FishGlob · marine fish · forecast skill · climate change · centroid · persistence · reproducibility
References
6- usesdoi:10.5281/zenodo.19056965
FishGlob_data v2.1 compiled standardized bottom-trawl data are the input of all analyses.
- usesdoi:10.1038/s41597-023-02866-w
Data descriptor for FishGlob_data, including the footprint trimming and taxonomic flags that we use as filters.
- doi:10.1126/science.1239352
Earlier evidence that marine taxa shift with local climate velocity in North American trawl surveys.
- doi:10.1111/faf.12330
Earlier test of centroid forecast skill against a persistence forecast in the eastern Bering Sea.
- doi:10.1038/s41467-026-74371-8
European shelf trawl analysis that found poleward and equatorward shifts in similar proportions.
- doi:10.64898/2026.02.18.706571
Preprint that found thermal niche change more consistent than range shifts in trawl surveys.
Cited by
5 publicationsPublic catalog entries on Sciscape.
- Against static-distribution replicates, demersal fish in 15 FishGlob trawl survey units experienced 4–10% less year-to-year bottom-temperature change (16–35% on the US Northeast shelf)
Same FishGlob v2.1 data; centroid series fluctuate more than they trend, which agrees with the low precision of our long-term pass-through estimates.
- Year-to-year biomass responses of demersal fish to bottom temperature were about 3% of their spatial temperature slopes in 13 trawl survey units (1963–2023)
Related FishGlob study on the low persistence of fish distribution shift rates.
- Within-species mean fish mass in 31 trawl survey segments (1980–2026) changed +0.3% per °C of yearly bottom-temperature anomaly (95% CI −0.7 to +1.3), not the −2.5 to −5% of the temperature–size rule
Same FishGlob surveys, persistence of distribution shifts.
- supportsIn NS-IBTS winter surveys, 14 of 34 North Sea demersal fish species tracked 5-year mean SST in 1980–2001, as Perry et al. (2005) reported, but these relations did not predict 2002–2026 latitudes
Found that early centroid shift rates do not predict later rates in trawl surveys; our temperature-based test in the North Sea agrees that centroid relations do not persist.
- supportsIn four Finnish bird atlases (1974–2025), southern species moved north at 1.6 km/yr before 2008 but 0.5 km/yr after, and the earlier shift of a species did not predict its later shift
A similar lack of persistence of shift rates in a different taxon and survey type.