In February 2019, the 60-meter-long South Korean fishing vessel Oyang 77 slipped into Argentinian waters and deployed its trawl nets, hauling in more than 140 tons of hake, skate, and squid. The ship did not have permission to fish those waters, according to Argentine officials, and to avoid detection the crew turned off a beacon that sends a vessel’s precise location via satellite to maritime authorities. But the coast guard caught the Oyang 77, confiscated the catch, and destroyed its nets.
Now, researchers have used artificial intelligence to help authorities more easily decipher what vessels like the Oyang 77 might be doing when they go dark and whether they might be fishing illegally. The approach is already guiding some enforcement agencies in planning their patrols.
Illegal fishing accounts for hauls that are worth about $25 billion per year and include endangered species such as sharks. Some vessels have been caught with enslaved crews. Although there is increasing political awareness of the problem, governments have not taken enough action, says Rashid Sumaila, an ocean and fisheries economist at the University of British Columbia, Vancouver. “This paper is showing some of the possibilities that we can achieve,” says Sumaila, who was not involved. “I think there’s hope.”
To prevent collisions, international law requires many large boats to carry beacons that broadcast their location. These signals, called the automatic identification system (AIS), are displayed in real time on websites as a navigational aid.
To researchers, the records of these locations yield telltale signs of activity: Vessels will travel in one direction to lay out a drifting long line of baited hooks, then double back to haul in the catch. Ships pursuing tuna make looping patterns as they encircle a school of fish with a giant net. Researchers analyzing these patterns have created global maps of commercial fishing, showing these vessels are regularly fishing more than half of the world’s oceans.
But sometimes the beacons go silent, rendering vessels invisible to other boats and to researchers. This can result from poor satellite reception or technical issues with the devices. Other times, researchers suspect, captains disable the devices to avoid detection.
Would it be possible to tell innocent problems from nefarious activities? “It seemed like a really ripe question,” says Heather Welch, a spatial ecologist at the University of California, Santa Cruz, “and a huge problem to tackle.”
To do that, Welch and colleagues turned to a kind of artificial intelligence called machine learning. These computer programs can make sense of vast amounts of data, such as the billions of AIS locations stored in public databases.
First, the researchers identified case studies that clearly looked like deliberate beacon deactivation, for example where the signal abruptly ceased and later turned on again at full strength. They also found examples where the problem seemed to be technical, such as a weakening and intermittent signal. These examples helped train the machine learning programs to recognize signs of ships trying to evade detection.
The analysis detected more than 55,000 instances in which crew seemed to deliberately turn off their AIS beacons between 2017 and 2019. All told, 5269 vessels disappeared for about 5 million hours while fishing, the team reports today in Science Advances. Some vanish for days or even weeks.
“Some vessels go out there and just disappear, hiding from any accountability,” says study co-author Tyler Clavelle, a data scientist at Global Fishing Watch, a nonprofit environmental organization. The new method “is a way to fill in blind spots,” he adds. “Literally.”
Much of this activity was concentrated in regions known to have problems with illegal fishing, such as the Northwest Pacific Ocean. Other hot spots included the rich fishing grounds off Argentina and West Africa, where foreign vessels catch fish without permission. The researchers also found many fishing vessels disabled their beacons when they approached refrigerated cargo vessels, a way of selling illegal catches at sea to avoid being caught at ports.
Pramŏd Ganapathiraju, a consultant in fisheries enforcement, calls the research “an excellent piece of work,” because the analysis shows which country’s fleets are most likely to turn off their AIS beacons. Cases near an Exclusive Economic Zone frequently involved Chinese- and Spanish-flagged ships. Details of locations, times, and vessels will help coastal states take action against the suspected illegal activities, says Ganapathiraju, who was not involved in the research.
Not all evasion is criminal. Another hot spot of hidden vessels was in Alaskan waters, where U.S. vessels must also carry a beacon that can’t be turned off and is visible only to authorities. That means these trawlers can’t hide from law enforcement, so they probably aren’t fishing illegally, Welch says. Instead, they are likely turning off their AIS beacons so competitors can’t follow them to good fishing spots, she notes. In other areas, such as off East Africa, vessels may be concealing their locations from pirates.
Welch is continuing her work. She has funding from the U.S. National Oceanographic and Atmospheric Administration’s Office of Law Enforcement to use AIS data to improve protection of sharks from illegal tuna fishing. For the past 3 years, Global Fishing Watch has been sharing its AIS analyses with the U.S. Coast Guard to help it plan patrols in the North Pacific, which has increased detections of violations. They are also developing an analytical tool to assist port authorities to prioritize which fishing vessels to inspect. Knowing that a vessel had turned off its beacon while at sea is a hint it might have fished illegally, Clavelle says.
Tracking fishing vessels through AIS can help reduce the cost of fisheries enforcement, Sumaila notes. But it’s easier for developed countries to enforce the law by sending patrol ships to catch offenders, whereas poor countries continue to suffer the burden of illegal fishing. The key, he says, is to make it unprofitable by increasing the risk of being apprehended and suffering a significant penalty: In 2019, for example, the Indonesian fisheries ministry blew up vessels caught illegally fishing.