Remember the headlines that claimed bottom trawling released as much carbon as all of air travel? We thought those claims were probably bogus when first reported, but Hiddink et al. 2023, a response paper published May 2023, now makes those claims Officially Bogus.
The original headlines came from Sala et al. 2021, a paper published in Nature that garnered more media coverage than any marine science paper of the past decade. We’ve covered the science and follow-ups over the last few years, but here’s a quick summary:
Sala et al. 2021 advocated for increasing the number of marine protected areas (MPAs) that restrict fishing. The paper used three different models that claimed benefits from MPAs:
- Food security – MPAs would increase food availability via more abundant fisheries.
- Biodiversity – MPAs would enhance ocean biodiversity.
Carbon/climate change – claimed that bottom trawling released as much carbon as all air travel, thus selling MPAs as a climate change solution via carbon sequestration. The paper suggested selling carbon credits from MPAs to fund the creation of more MPAs.
These three claims have quickly fallen apart, however. The original food security model was retracted, and the modified one by Sala et al. 2021 had similar issues. A response published last year points out that the biodiversity and carbon claims were based on the assumption that fishing disappears rather than being displaced. And now, Hiddink et al. 2023 demonstrates that the carbon model overestimated carbon benefits by 2-3 orders of magnitude, i.e., 100-1000 times.
Hiddink et al. 2023 notes two main reasons why the model in Sala et al. 2021 misfired:
- The fundamental assumptions of the carbon cycle were incorrect.
- The validation of those assumptions was also incorrect.
Here, we explain the carbon cycle in ocean sediment and discuss the potential for bottom trawling to contribute to carbon emissions. We also break down the carbon model from Sala et al. 2021 and show why it was incorrect based on Hiddink et al. 2023’s analysis.
The carbon cycle in ocean sediments
When an organism in the ocean dies, no matter how big or small, it starts to sink. On its way down, and after it has landed on the bottom, the organic carbon that formed the organism is broken down by microbes in seawater and other organisms feeding off it. The carbon is mineralized from solid organic carbon into its more basic form, CO2, dissolved in seawater to be used again by alive and growing organisms.
Only a tiny portion of the sinking organic carbon reaches the ocean floor, but when it does, bacteria and benthic invertebrates quickly consume most of it and respirate CO2 back into the water column; nearly all organic carbon on the seafloor is mineralized within weeks. However, some carbon slips through and is eventually buried further down in the sediment, where less oxygen is present, and mineralization is significantly slower. This buried layer may serve as a carbon storage facility that helps regulate global carbon cycles and, thus, climate change. This is the key mechanism explored in Sala et al. 2021: how much buried organic carbon is exposed and remineralized into CO2 because of trawling?
The critical flaw in Sala et al. 2021’s model was treating all sediment the same, disregarding the significant differences between buried carbon and the fresh, top layer.
The mechanics of trawling suggest that the top layer is significantly more affected. Trawl nets skim along the bottom; when they contact the seafloor, they dig about an inch into the sediment (the model assumed 2.4 cm). Consequently, the organic carbon disturbed by trawling would come from the top layer and likely would have mineralized naturally. However, the model assumed that the carbon disturbed by trawling was comparable to the carbon stored deep within the buried layer.
Dr. Gordon Holtgrieve, a biogeochemist at the University of Washington not involved in Sala et al. 2021 or Hiddink et al. 2023, told me, “Sala et al. 2021 certainly took a very aggressive approach to modeling organic carbon mineralization from the sediments. Their assumptions of a single pool [treating all sediment the same] with high carbon lability certainly stacks the deck in favor of a high-impact result.”
Dissecting the Sala et al. 2021 carbon model
The Sala et al. 2021 model was relatively simple; it amounted to the amount of available carbon resuspended in the water due to trawling (called labile carbon) times the rate at which it remineralizes (a derived constant, k). It assumed a labile carbon rate of p = 0.7, meaning 70% of carbon in the sediment is able to be remineralized and a k value ranging between 0.3 – 17 per year.
Labile carbon (p) × mineralization rate (k)
0.7 × 0.3 – 17 yr-1
Hiddink et al. 2023 point out that the labile carbon assumption of 0.7 and the high k values are only acceptable for a ‘fresh’ top layer of sediment that would remineralize naturally, not the stored, buried carbon as the model was intended for.
Hiddink et al. 2023 reports actual samples from the North Sea that show the k value of buried sediment between 100 and 1,000 times lower than those used by Sala et al. 2023. They argue that a more realistic k assumption would be 0.001 – 0.01 per year. From the response:
We argue that incorporating the role of composition would require lowering the k value to around 0.01 yr−1, which is representative of sub-surface sediments, and applying it to the bulk of the sediment (fraction of reactive material, p = 1) or, alternatively, using the original high k values (k = 0.3–17 yr−1) and applying them to the fraction of reactive material p present in historically buried organic carbon (p = 0.001–0.01).
Sala et al. 2021 validated their model by comparing their modeled data to four empirically measured locations, but Hiddink et al. 2023 points out that the empirically measured locations did not separate the natural mineralization of fresh organic material from the mineralization of buried carbon caused by trawling. The mineralization of fresh organic carbon will always dominate the total amount of CO2 produced, but the modeled data of carbon remineralized due to trawling was compared to all forms of mineralization (including fresh carbon naturally mineralized) at the empirically measured sites. This is not a valid comparison nor a way to validate assumptions.
If Sala et al. 2021 were correct in their assumptions, there would be much less carbon on the seafloor in trawled areas vs. untrawled areas, but a review paper published last year Epstein et al. 2022, found contradictory results. The review directly addressed what the Sala et al. 2021 model would have predicted: less carbon in trawled areas vs. untrawled. However, of 49 papers examined by the review, 61% showed no significant difference, 29% reported less organic carbon storage in trawled areas, and 10% reported higher amounts of carbon stored after trawling—this is plausible if benthic organisms that would otherwise remineralize organic carbon and pump oxygen into the sediment are removed via trawl. Regardless, the review strongly suggests that the Sala et al. 2021 model is critically flawed and supports the assertion by Hiddink et al. 2023 that the model is astronomically off-base (2-3 orders of magnitude).
In response to the critique by Hiddink et al. 2023, the authors of Sala et al. 2021 acknowledged that their model was dependent on assumptions but maintained that their conclusion was valid. They recognized Epstein et al. 2022, but did not address the validation critique.
Dr. Jack Middelburg, a leading ocean biogeochemist and co-author of Epstein et al. 2022, told me, “I agree with Hiddink et al. regarding the inconsistent, unrealistic premises underlying Sala et al. sediment respiration enhancement due to trawling.” I asked him how the original model got through peer review and if he thought a biogeochemist had peer-reviewed it; he said, “likely not a sediment biogeochemist.”
How was Sala et al. 2021 peer-reviewed?
The model used in Sala et al. 2021 was inaccurate by multiple orders of magnitude due to poor assumptions about the remineralization rate of carbon. This makes three out of three models used in Sala et al. 2021 to come under severe and legitimate criticism. The assumption that the top layer of sediment mineralizes fresh carbon like deeper sediment is nearly indefensible and should have been caught in peer review.
The peer review process is confidential, so we don’t know how the process played out or who was chosen to review Sala et al. 2021 before it was published, but I think it is fair to say that it was botched—and not necessarily for the fault of Nature or the editors.
Reviewing complex computer models is hard, and the pool of scientists who can adequately review them shrinks with increasing complexity.
Sala et al. 2021 had three complicated computer models! In a perfect scenario, there would be at least one reviewer for each model with extensive knowledge of the model assumptions. The only peer reviewer to sign his name does not appear to have any quantitative modeling experience based on his Google Scholar page.
The original food security model published in the Proceedings of the National Academy of Science was found to be flawed because perhaps the only person with intimate knowledge of the model it was based on took a closer look and found problems. That eventually led to further digging and eventual retraction.
Nature and dozens of scientists have spent over two years trying to clean up the mess created by the flawed peer-review process. Hiddink et al. 2023, a five-paragraph critique, was submitted nearly two years ago, while the official response to the Sala et al. 2021 food security model has still not been published.
All the while, Sala et al. 2021 has been used to drive policy and advocate for benefits that don’t seem to exist.
A potentially harmful myth
No model demonstrates the potential harm of Sala et al. 2021 more than their flawed carbon model. The unrealistic expectations about the number of potential carbon credits generated through MPAs could lead to increased carbon emissions. I worry that governments or NGOs would consider counting an imaginary reduction in CO2 emissions by trawling to offset real CO2 emissions from other activities. Imagine if the airline industry paid for an end to trawling to make airline travel “carbon neutral” based on Sala et al. 2021. Giving a pass to real emissions to fund more MPAs would be terrible for the ocean and the planet.
Holtgrieve sums it up well:
The authors draw a straight line between a decrease in sediment carbon storage and an increase in atmospheric CO2. That line is not at all straight and involves numerous untested assumptions. It was reported in the popular press that “Bottom trawling releases as much carbon as air travel, landmark study finds.” This framing is highly disingenuous because air travel emits a large and well-known amount of CO2 straight to the atmosphere, where it alters the climate. The amount of CO2 released from trawling is highly uncertain, and it’s possible that only a small fraction gets to the atmosphere. Making an equivalence between air travel and trawling is dangerous as it takes our attention away from real solutions that work, reducing fossil fuel emissions. The authors know this and are complicit in this deception.
Max Mossler
Max is the managing editor at Sustainable Fisheries UW.
Read the full response:
