In two separate studies leveraging satellite imagery and artificial intelligence techniques, researchers reveal patterns of industrial fishing in coastal marine protected areas (MPAs) worldwide. Collectively, the findings, which may seem contradictory, show that although industrial fishing vessels are present in many protected areas worldwide, MPAs with the highest levels of protection remain largely unfished. Both studies suggest that proper investment in protected areas will pay off and that synthetic aperture radar (SAR) satellite technology could be one of the key tools used to safeguard the ocean’s future sustainability. Roughly 8% of the global ocean is formally protected, with ambitious international targets aiming to more than triple that coverage by 2030. While such protections can yield substantial long-term benefits, especially when paired with adequate fisheries management, potential gains are often compromised by inadequate regulations. In many cases, destructive, illegal, or unreported fishing practices persist even within designated protected areas due to insufficient safeguards. Global-scale monitoring of industrial fishing, including within MPAs, has been aided by the emergence of automatic identification system (AIS) data, which tracks the activity of individual vessels. However, not all vessels are required to use it. Many disable their transponders to avoid detection, making it difficult to obtain reliable, large-scale estimates of fishing pressure within MPAs. As a result, the true effectiveness of MPAs worldwide remains poorly understood.
In one study, Jennifer Raynor and colleagues analyzed 455 coastal MPAs classified as “fully” or “highly” protected under the MPA guide, an assessment framework that evaluates protections based on both regulations and management practices. These categories fully ban industrial fishing within their bounds. Raynor et al. combined AI methods with a recently published global SAR satellite imagery dataset to directly identify industrial fishing vessels operating within MPAs, regardless of whether their AIS is active. The authors found that, overall, very little unauthorized industrial fishing activity occurs in MPAs that prohibit it, averaging just one vessel detected per 20,000 square kilometers – a rate 9 times lower than in unprotected exclusive economic zones. Although a few MPAs in East and South Asia showed higher vessel densities, these cases were outliers driven by small geographic areas and sporadic detections. Only seven MPAs worldwide had vessels present on more than half of observed days, highlighting how rare such activity is in strongly protected areas. Raynor et al. also demonstrate SAR imagery’s reliability in detecting unauthorized fishing vessels. Not only did the method successfully identify AIS-broadcasting vessels with high accuracy, it also detected vessels in 163 MPAs where AIS data showed none, particularly in regions like Southeast Asia, where AIS is often incomplete.
In another study, Raphael Seguin and colleagues quantified fishing activities across a larger group of 6021 coastal MPAs representing a wide range of protective categories as outlined by the International Union for Conservation of Nature (IUCN) management framework. Using the same SAR dataset and deep learning models, Sequin et al. discovered that nearly half of the MPAs evaluated showed evidence of industrial fishing at levels in many cases matching or exceeding those in nearby unprotected waters. According to the findings, industrial fishing vessels were detected in 47% of the world’s coastal MPAs. While stricter IUCN categories did correlate with reduced fishing, the authors concluded that factors such as MPA size and remoteness were more predictive of fishing presence than official protection category alone. In a Perspective, Boris Worm discusses the possible drivers underlying the differences in the two studies’ findings. “Many MPAs have been established quickly without strong protective regulations, meaningful consultation with local stakeholders, or appropriate management capacity. In some cases, this has resulted in “paper parks” that are recognized as protected areas but do not prevent harmful activities,” writes Worm. “Yet the available data show that where proper investments are made, industrial exploitation is curtailed, and protective measures are comprehensive, long-term benefits will accrue.”
Data is available for the production of data visualizations. For more information, please contact Raphael Seguin at raphaelseguin@protonmail.com
