Off Peru the fleet fishes the Peruvian anchoveta, Engraulis ringens. In years of prolonged heat the anchoveta catch falls by about 60 per cent at the median. The figure comes from a study published on 1 September 2026 in Nature Communications by William Cheung and Isabella Morgante of the Institute for the Oceans and Fisheries at the University of British Columbia, with Juliano Palacios-Abrantes and with Thomas Frölicher of the University of Bern.
Peru is one case. The study looks at the whole world. The authors took catches from 1993 to 2019 as reconstructed by the Sea Around Us project, which adds industrial, small-scale, subsistence and recreational fishing and discards. The data cover 1,246 species of fish and invertebrates in 254 exclusive economic zones, the strips of sea where a coastal state holds fishing rights. For each species in each zone they set the temperature of the top hundred metres of water, from the reanalyses of the European Copernicus service. They added net primary production, the organic matter that microscopic algae build from sunlight and that feeds the whole food web.

A year counts as extremely hot when temperature exceeds the 90th percentile of the local record, and as extremely unproductive when production falls below the 10th. A catch is extremely low when it falls in the lowest tenth of its series. Odds are the probability that an event happens divided by the probability that it does not. The results are odds ratios, which compare the odds under two conditions. An odds ratio of 1.8 does not mean an event is 1.8 times as likely, although the two converge when the event is rare.
Two years in a row of extreme heat raise the odds ratio of a collapsed catch to 1.40, with a 95 per cent confidence interval of 1.20 to 1.65. The strongest effect comes from the combination. A year of extreme heat followed by a year of minimum productivity raises it to 1.8, between 1.5 and 2.1. Cold years and rich years do not produce the mirror effect. Catches do not rise as far as they fall.
The zones where heat matters most, with odds ratios of 2 or more, are the great upwelling systems off Peru and California, the eastern tropical Pacific, the Pacific islands, the Seychelles and the Maldives. Three enclosed seas join them, the Red Sea, the Persian Gulf and the Mediterranean. For these seas the authors write that extreme heat years bring consistently high odds of a catch collapse. The paper does not break the Mediterranean down by species or by country, so Italy’s own fisheries are not singled out.

Among groups of species the most sensitive are sharks and rays, small pelagic fish such as anchovies and sardines, and commercial invertebrates. In collapse years their median catch falls by more than 40 per cent. For reef fish in the Maldives the typical loss is about a quarter. In the Solomon Islands skipjack tuna responds above all to years of poor productivity.
The authors then project the observed relationships onto future climate, with three models and two emissions scenarios, one low and one high. In the early 1990s about 11 per cent of stocks went through a year of extremely low catch. By 2060 the share rises to about 16 per cent under both scenarios. The two curves stay close because the climate of the next few decades is already partly set by gases already emitted. This is a projection, not a measurement.

The limitations are spelled out in the paper. The models find associations, not mechanisms. Catch is not the biomass of fish in the sea and also reflects markets, fleets and rules. Overfishing, habitat loss and oxygen loss are not in the model. In many data-poor regions the reconstructed catches lean on expert estimates. Where fishing effort data existed and were added, the estimates moved by less than 3 per cent.
A 2023 study led by Fredston had concluded that marine heatwaves do not explain declines in bottom-dwelling fish. Cheung and colleagues reply that it used heatwaves lasting days and northern seas where the link is weak. This study looks at year-long anomalies. Among bottom-dwelling fish, heat alone counts for little, while heat followed by a hungry sea nearly doubles the odds of a bad year.
The work was funded by Canadian research councils and the European Union’s Horizon 2020 programme. The authors declare no competing interests. The code is public on Zenodo and the data come from open archives.

References
Cheung W.W.L., Frölicher T.L., Morgante I., Palacios-Abrantes J., “Large fisheries declines linked to compound and extreme climate events”, Nature Communications, 17, article 8897, published 1 September 2026, open access under CC BY-NC-ND 4.0. doi: 10.1038/s41467-026-77077-z
Cheung W.W.L., Morgante I., Palacios Abrantes J., “coruubc/CompoundExtremes: Version 1.0.1”, code, Zenodo, 14 July 2026, MIT licence. doi: 10.5281/zenodo.21350761