Marine sediments are thought to be the ultimate long-term carbon store; once buried below the active layer, organic carbon can remain unmineralized for millennia to eons (Burdige, 2007; LaRowe et al., 2020). However, disturbances to the seabed by human activities threaten the permanency of this marine carbon (Levin et al., 2020; Paradis et al., 2021). In the case of bottom trawling, heavy fishing gear that is dragged across the seafloor mixes and resuspends sediments, exposing 0.16-0.40 Pg C yr^-1 of previously buried organic carbon to potential microbial degradation (Sala et al., 2021). However, the ultimate fate of this disturbed organic carbon stock is as yet unquantified, hampering our understanding of the effects that bottom trawling has on the global carbon cycle and the potential implications for climate policies.

The protection of organic carbon stored in marine sediments, plants, and animals has been identified as a powerful tool for tackling climate change (Hoegh-Guldberg et al., 2019). However, the uptake of ocean-based climate solutions has been slow due to prevailing climate policies and carbon markets that only recognize mitigation activities with measurable impacts on atmospheric emissions. The challenge with identifying ocean-based solutions under those current paradigms lies in the complexity of quantifying atmospheric emissions generated by anthropogenic activities that occur below the ocean’s surface (Luisetti et al., 2020). Therefore, research addressing this challenge is crucial for discovering new opportunities that can harness the full potential of the ocean in contributing to mitigating climate change.

Here, we examined the fate of trawling-induced carbon released into the global ocean between 1996-2020 and under future scenarios, as well as estimated the fraction of CO₂ emitted to the atmosphere. To estimate trawling-induced CO₂ emissions, we used assumptions and data from Sala et al. (2021), the only study to date to estimate the global impact of trawling on CO₂ fluxes from marine sediments, and two classes of ocean circulation models: (I) the Ocean Circulation Inverse Model (OCIM; 2° resolution; Holzer et al., 2021) and (II) the NASA Goddard Institute for Space Studies (GISS) ModelE2.1 (1° x 1.25° ocean model resolution; Lerner et al., 2021). The latter was used in coupled climate simulations under two realizations: prescribed atmospheric CO₂ concentrations (GISScon) and prognostic atmospheric CO₂ based on anthropogenic emissions, the land and ocean sink, and benthic trawling (GISSemis; Ito et al., 2020). GISS and OCIM models are used to estimate air-sea fluxes and internal oceanic transport of CO₂ over time by simulating the complex interplay of atmospheric and oceanic processes. These models offer detailed spatial-temporal estimates of CO₂ exchange between the ocean and atmosphere by modeling the movement of CO₂ through currents, advection, vertical mixing, biological processes (GISS only), and surface gas exchange. Depending on the geographic location and water depth of bottom trawling, CO₂ is exposed to the sea surface within months to centuries (Siegel et al., 2021). GISS and OCIM models are systematically appraised against the latest observations, are internationally accepted, and are being used in the CMIP6 to represent ocean processes (e.g., air-sea fluxes) for the 6th Assessment report (IPCC, 2022) and in the Global Carbon Budget to estimate surface pCO₂ (Friedlingstein et al, 2020a).

Our retrospective and prospective analyses showed that 55-60% of the CO₂ released into the water column by bottom trawling impacts on sediment carbon stocks is emitted to the atmosphere within ~9 years of the trawling event ( Figure 1 ). Furthermore, we found that the fraction of CO₂ accumulating in the atmosphere remained at 55-60% until the end of our simulations at 2100, regardless of the magnitude of CO₂ predicted to be released into the water column by trawling ( Figure 1 ). These results are significant in that they imply that the 55-60% fraction can be easily applied to estimate trawling-induced CO₂ emissions to the atmosphere under a variety of historical and future trawling scenarios.

Using Sala et al. (2021) estimates of sediment efflux, our models suggest that trawling could have emitted a cumulative 8.5-9.2 Pg CO₂ into the atmosphere between 1996 and 2020 ( Table S1 ; Figure 2 ), contributing 0.97-1.14 ppm to atmospheric CO₂ concentrations ( Figure 2 ). These emissions would equate to ~0.34-0.37 Pg CO₂ yr⁻¹, which is equivalent to ~9-11% of the global emissions from land-use change in 2020 (Friedlingstein et al., 2020b), or nearly double the estimated annual emissions from fuel combustion for the entire global fishing fleet (Parker et al., 2018). Trawling emissions of this magnitude suggest that the protection of seabed organic carbon from benthic trawling gear could prove to be an impactful climate solution. For example, if we continue to trawl at current intensities and spatial distributions, we estimate that bottom trawling could contribute an additional 0.2-0.5 ppm in atmospheric CO₂ concentrations by 2030 and 1.03-1.36 ppm by 2070 ( Figure 2 ).

Whether or not reductions in trawling could be adapted as a climate solution not only depends on the magnitude of the emission reductions, but also the time frame over which those reductions can be achieved. We found that the release of trawling-induced CO₂ from the ocean to the atmosphere occurred rapidly, with ~99% of the total emissions occurring within 7-9 years post-bottom trawling (OCIM: 7 yrs; GISScon: 9 yrs; GISSemis: 9 yrs ± 5 yrs (standard deviation of ensemble members). When emissions were arbitrarily increased by one order of magnitude, it took slightly less time (OCIM:~5 yrs) for total emissions to be released into the atmosphere. The rapid release of CO₂ from the ocean to the atmosphere suggests that historical trawling has only short-term legacy effects on atmospheric emissions. Thus, policies that eliminate or significantly limit trawling impacts on sedimentary carbon stocks would quickly reduce this industry’s contribution to rising atmospheric CO₂ concentrations with maximum benefits occurring 7-9 years after implementation.

In general, atmospheric CO₂ emission hotspots coincided with areas where trawling had the most significant impact on benthic carbon, mainly the East China Sea, the Baltic and the North Sea, and the Greenland Sea ( Figure 3 ). However, horizontal advection can transport trawling-induced CO₂ and resuspended organic carbon to other locations, leading to cross-boundary effects of bottom trawling on local carbon cycles. This phenomenon likely explains why some areas such as the South China Sea, Norwegian Sea, and off the east coast of Japan in the Pacific Ocean had higher atmospheric emissions than expected based on the local rate of trawling emissions ( Figure 3 ). As a result of these cross-boundary effects, we cannot assume that all the atmospheric emissions within a country’s jurisdictional waters come from trawling activities within that zone.

Our ability to quantify the extent of the global bottom trawling fleet through time and space in this study was somewhat limited. Our estimates do not capture trawling activities before 1996, because the intensity and spatial distribution of bottom trawling before that time are unknown. Yet, large-scale bottom trawling began as early as 1950 and peaked in the 1980s and 1990s (Watson et al., 2006; Watson and Tidd, 2018). Furthermore, our model relies on the AIS vessel tracking database processed by Global Fishing Watch (https://globalfishingwatch.org/) to derive trawling events at the global level. Unfortunately, AIS coverage is poor in some fishing-intensive areas. Thus, we undoubtedly underestimate trawling activity in areas of Southeast Asia, the Bay of Bengal, the Arabian Sea, parts of Europe, and the Gulf of Mexico (Taconet et al., 2019).

There are additional uncertainties in the parameters used to estimate the amount of organic carbon remineralized after trawling due to a lack of rigorous field studies. Though our model is parameterized using the best available empirical data (Sala et al., 2021; Atwood et al., 2023), an alternative line of reasoning argues that first-order degradation rates could be one- to two orders of magnitude lower (Hiddink et al., 2023). Because we find that atmospheric emissions scale linearly with the amount of organic carbon remineralized to aqueous CO₂ by trawling, we can straightforwardly examine the sensitivity of our results to this alternative assumption. Leveraging Atwood et al. (2023) finding that reducing first-order degradation rates by one order of magnitude has a negligible effect on the estimated magnitude of organic carbon remineralized after 10 consecutive years of trawling, we find that retaining this reduction to first-order degradation rates results in an estimated 0.19-0.21 Pg CO₂ yr^-1 emitted to the atmosphere due to bottom trawling between 1996-2020. A two-orders of magnitude reduction to first-order degradation would result in 0.02-0.03 Pg CO₂ yr^-1 emitted over the same time period – comparable to the mitigation potential for managing fires in temperate forests (Griscom et al., 2017).

Currently, climate actions aimed at reducing CO₂ emissions from anthropogenic practices (e.g., carbon markets, renewable energy standards, reforestation efforts, etc.) focus exclusively on atmospheric emissions. However, these frameworks overlook the total impact of ocean-use change activities on the carbon cycle, because they ignore the pool of DIC that remains sequestered by the ocean. In the case of trawling, we found that 40-45% of the cumulative trawling-induced CO₂ emissions remained dissolved in seawater, augmenting the acidification already occurring from the burning of fossil fuels. Using Sala et al. (2021) carbon model, we found that trawling increased the global DIC inventory by ~1.82-1.90 Pg C from 1996-2020 ( Table S1 ). This additional dissolved inorganic carbon from trawling results in increased ocean acidification with a global reduction in pH of 3-5 ×10^-4 by 2020 ( Figure 2 ). At the global scale, a pH reduction of that magnitude by 2020 is not significant compared to the effect of anthropogenic emissions due to fossil fuels. However, our models suggest that some semi-enclosed seas could be highly sensitive to an injection of CO₂ from anthropogenic activities. In particular, our models showed that extensive trawling could lead to increased localized acidification in the East and South China Sea ( Figure 4 ). The decrease in pH in this region due to trawling between 2000 and 2020 (GISSemis: -0.034+/-0.001; OCIM: -0.050) is comparable to that from rising atmospheric CO₂ due to the burning of fossil fuels over the same time period (GISSemis: -0.034+/-0.004; OCIM: -0.020). An important caveat to our pH findings is that our models are limited in resolving coastal processes both due to their coarse resolution and lack of biogeochemical complexity. Nevertheless, considering that ocean chemistry can influence organismal development, physiology, and behavior (Baag and Mandal, 2022), and ultimately can affect a species’ productivity and survival, future studies and policy should consider the potential impacts trawling can have on localized ocean acidification.

Ocean-based solutions offer promise in closing the emissions gap to limit global temperature increases to 1.5°C, while also supporting co-benefits like biodiversity preservation and food security (Hoegh-Guldberg et al., 2019; Sala et al., 2021). However, current climate policies and markets require estimates of avoided atmospheric emissions, posing challenges for identifying and implementing these solutions. Our study, which highlights that 55-60% of CO₂ produced from bottom trawling is released into the atmosphere within nine years, becomes a crucial tool for evaluating the reduction of bottom trawling effort as an effective ocean-based climate solution. To refine atmospheric emission estimates, it is essential for field studies to tackle uncertainties in our understanding of how bottom trawling influences the biological and physical processes that govern carbon remineralization and preservation. Furthermore, the incorporation of high-resolution regional models that resolve small-scale processes, such as local currents, will be pivotal in delivering more precise emission estimates at scales pertinent to local policy considerations. Lastly, our findings emphasize the need for policy to avoid exclusive focus on avoided atmospheric emissions, as our results show that trawling-induced increases in DIC in seawater could have severe implications for local or regional ocean acidification.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) are as follows. Data on marine sedimentary carbon stocks is available at https://figshare.com/articles/dataset/marine_soil_carbon/9941816. Data on swept volume ratios and trawling activity are available by contacting Global Fishing research@globalfishingwatch.org. The code for GISS models is available at the TrawlingExpts2023 at http://simplex.giss.nasa.gov; for access to the simplex.giss.nasa.gov repository, email anastasia.romanou@nasa.gov. OCIM code is available upon email request to TD at tdevries@geog.ucsb.edu. All data used for this study and the figures can be found in the NASA Center for Climate Simulation portal: https://portal.nccs.nasa.gov/datashare/modelE_ocean/Atwood_etal2023_paper_data.

All authors contributed to the conceptualization and design of the study. AR, PL, TD, and JM ran data analyses. TA wrote the original draft of the manuscript and all authors contributed to reviewing and editing. All authors contributed to the article and approved the submitted version.