Sustainable fisheries management relies on scientists’ ability to accurately estimate the size of targeted fish populations and the speed at which they reproduce. This information allows national and regional government agencies to sufficiently restrain fishing—primarily through catch limits and seasonal or area-based closures—so that fish populations have a chance to recover. Maintaining this delicate balance is vital for global food security and the livelihoods of tens of millions of people worldwide.
Yet recent research from a team of authors, including 2021 Pew marine fellows Amanda Bates and Rick Stuart-Smith, raises serious questions about the accuracy of current stock assessment models. These complex mathematical tools combine many sources of information related to fish biology, oceanographic conditions, fishing practices, and ecosystem health to produce recommendations for the managers who are tasked with ensuring the sustainability of fisheries.
The authors of the study, which was published in August 2024 in the journal Science, analyzed 230 stock assessments worldwide from 1980 to 2018 and compared their historical biomass estimates—snapshots of the size of fish populations at the time of the assessment—with updated biomass estimates calculated for those same years using newer models. Estimates produced by newer models should be more accurate than older ones because they include additional data collected over a longer period.
The study found evidence that 66% of the assessments in the sample included inflated biomass estimates—suggesting that the models for many stock assessments have been overestimating the number of fish remaining in the sea, compared with the updated knowledge from later assessments. These biased estimates were particularly prevalent in stocks that are now classified as overfished.
“We find that our models are often too optimistic,” Bates said, noting that modeling across all 230 stocks in the study likely overestimated fish biomass by an average of 10%.
The researchers also found evidence that many of the recovery trends estimated by current stock assessment models were overly optimistic for overfished stocks—with the result that fisheries managers may have approved continued fishing of stocks that were failing to recover.
“A bias toward overestimating fish stocks in overfished fisheries could lock in stock collapse, as the decline in fish may not be noticed until it’s too late to take precautionary measures,” Bates said.
In a co-authored perspective piece about the research, University of British Columbia fisheries professor Daniel Pauly and Rainer Froese, senior scientist at the Helmholtz Center for Ocean Research, wrote, “[The research shows that] rising trends in biomass reported for overfished stocks were often inaccurate, resulting in so-called phantom recoveries for stocks where actual biomass was fluctuating at a low amount or even declining. On the basis of these data, fishery managers could reasonably conclude, albeit incorrectly, that the stock was recovering and able to support even higher catch levels.” Froese is also a 2003 Pew marine fellow.
The study’s findings demonstrate how systematic biases, which can be masked by the complexity of modern stock assessment models, may have significantly affected the sustainability of global fishing. The authors also present evidence showing that it is likely that many more stocks should be classified as “collapsed”—that is, under 10% of their historical maximum biomass.
“The study highlights ways to improve the accuracy of fish stock assessments, such as expanding independent fisheries monitoring and changing stock assessment protocols,” said lead study author Graham Edgar. “This could include establishing a ‘red team’ that looks at potential worst-case scenarios and works to prevent the collapse of fish biomass.”
Despite their findings, Bates and Stuart-Smith argue that avoiding overexploitation in fisheries is possible so long as management decisions are guided by the precautionary principle—particularly when the accuracy of assessment models is uncertain. Stock assessment science is notoriously complex, and there's no consensus on the prevalence of bias in assessments. In addition to using appropriate models that are adjusted for known biases and account for uncertainty, the authors encourage the simplification of models where possible to reduce opaqueness and provide managers a clearer understanding of the modeling results underlying their decision-making.
“I’ve always found it ironic that our best efforts to estimate how many fish are left in the sea tend to rely so heavily on what has been taken out,” said Stuart-Smith. “We didn’t need this study to show how challenging the task of trying to estimate how many fish can sustainably be removed really is, but I hope that the results ram home how important it is to consider the more conservative end of the spectrum of uncertainty in this whole process.”
Critically, these practices are possible only for fisheries that are actively managed and for which managers have the resources to conduct thorough stock assessments. Many global fisheries—particularly those located outside of high-income countries—haven’t ever been assessed.
“Fisheries feed the livelihoods of many coastal people, and our cultures are intimately tied to fisheries and fishing practices,” said Bates. “If we care about future generations and want healthy ocean systems, we need to rethink how we’re managing our fisheries.”
Nathan Fedrizzi works on the Pew Fellows Program in Marine Conservation.