How much of the world’s oceans are affected by fishing? In February, a team of scientists led by David Kroodsma from the Global Fishing Watch published a paper that put the figure at 55 percent—an area four times larger than that covered by land-based agriculture. The paper was widely covered, with several outlets leading with the eye-popping stat that “half the world’s oceans are now fished industrially.”

Ricardo Amoroso from the University of Washington had also been trying to track global fishing activity and when he saw the headlines, he felt that the 55 percent figure was wildly off. He and his colleagues re-analyzed the data that the Global Fishing Watch had made freely available. And in their own paper, published two weeks ago, they claim that industrial fishing occurs over just 4 percent of the ocean.

How could two groups have produced such wildly different answers using the same set of data? At its core, this is a simple academic disagreement about scale. But it’s also a more subtle debate that hinges on how we think about the act of fishing, and how to measure humanity’s influence on the planet. “I think this discussion really shows how little we know about the world’s oceans and why making data publicly available is so important for stimulating research,” says Kroodsma.

As ships traverse the oceans, many of them continuously transmit their position, speed, and identity to satellites. This automatic identification system was originally developed to prevent collisions, but by training Google’s machine-learning tools on the data, the Global Fishing Watch (GFW)—a nonprofit founded by Oceana, SkyTruth, and Google—can identify different kinds of fishing vessels, and work out where they’re dropping their lines and nets.

They quantified that activity by dividing the oceans into a huge grid of around 160,000 squares. Each of these squares had sides that span half a degree of latitude, and an area of around 3,100 square kilometers. And in 2016, around 55 percent of them included some kind of fishing activity. “You’re dividing the ocean into 160,000 cells, and you’re seeing fishing activity in half of them,” says Kroodsma. “One reason this resonated with people is that however you slice it, that’s impressive.”

Actually, it’s misleading, says Amoroso. He and his colleagues had spent years trying to measure the impact of trawlers and based on their early results, the GFW’s estimates smelled fishy. The problem is that they divided the ocean into such large squares that if a single boat drops a net in an area the size of Rhode Island, that area would count as “fished” in a given year.

Fortunately, the GFW made all their data freely available, so Amoroso’s team could analyze it at finer resolutions. When they divided the ocean into smaller squares that are 0.1 degrees of latitude wide and take up around 123 square kilometers, they found that fishing activity occurs in just 27 percent of them. And when they used even smaller squares that are 0.01 degrees wide and 1.23 square kilometers in area—the size of a city block—just 4 percent of the ocean is “fished.” 

Consider these two pictures, which show the proportion of the North Pacific region that’s affected by trawlers. The red striped region is a 59,000-square-kilometer area where trawling was completely banned in 2016. If you use big 0.5-degree squares, the entire space is incorrectly classified as “fished.” If you use the finer 0.01-degree squares, it’s not.

This seems like a simple matter of scale. If you look at the world at different resolutions, you’ll see different pictures. So the question becomes: Which is the right resolution to use? In a response to their critics, Kroodsma and his team argue that when it comes to fishing, there’s no right answer.

Consider trawlers. They drag large nets behind them to catch bottom-dwelling species that don’t swim very far. So you can roughly work out how much ocean they affect by looking at the area that’s swept by their nets. For that calculation, small squares seem appropriate.

Longline vessels are very different. They tow fishing lines that are studded with thousands of baited hooks. These lines can extend for kilometers, drift with the currents in any direction, and intersect the paths of open-ocean fish that can themselves travel extensively. A single yellowfin tuna, for example, can roam over more than 250,000 square kilometers. These movements mean that longlines affect an area of ocean that’s far greater than the area beneath the boats—or even the hooks. And that’s why the GFW divided the ocean into such large squares.

Their critics retort that the distance over which a tuna could potentially travel is a red herring. What matters is the distance at which they’d be attracted to a nearby hook, and that’s considerably shorter. Sure, longlines affect a much larger range of ocean than the path of their gear, but “using a grid with an arbitrarily low resolution isn’t the right way to assess that,” says Amoroso.

(Science, the journal that published all these papers, didn’t help matters by appending a summary to the original paper entitled “More than half the fish in the sea.” Even if the 55 percent estimate is right, it relates to the percent of ocean affected by fishing vessels, not the proportion of fish that are targeted.) 

These same contentions also apply when comparing land and sea. In their original study, the GFW claimed that fishing affects 200 million square kilometers of ocean, which is four times greater than the 50 million square kilometers of land that’s affected by agriculture. But Amoroso’s team notes that the latter figure comes from studies that use a much tighter grid. If you analyze land and sea at the same resolution, it’s agriculture that outmatches fishing by a factor of four.

Again, this boils down to how scientists should measure humanity’s influence. Kroodsma says they should look at the slice of the world that’s been appropriated for human consumption. On land, that’s mostly the area grazed by livestock. But in the sea, it’s the area that supports targeted fish, not the area that’s swept by nets and hooks. “You want to know the area the fish are covering not the area the fishing vessels are covering,” he says.

“Even if the actual direct physical impact is closer to 4 percent than 55 percent, the actual impact of fishing on the ecosystem is likely to spread to 55 percent or more of the ocean,” says Rashid Sumaila at the University of British Columbia, who studies the economics of fishing. “Think of this as a bullet. Even though the area that’s directly affected is very small, its impact is felt throughout a person’s body.”

Amoroso and his colleagues get that, but they’re saying you can’t understand those broader effects by simply asking where fishing is happening. Instead, you have to ask different and more refined questions. How much of that fishing is sustainable? Which species are being targeted? How are they faring? Can they bounce back? And there are much better ways of directly answering those questions, from stock counts to ecosystem models, than just looking at the whereabouts of boats. Ultimately, the area that’s fished doesn’t really matter. 

But the way we talk about it does matter, especially in shaping how policy-makers and the general public think about the environment. “When these numbers say that agriculture has a lower footprint than fisheries, consumers might end up thinking that pork or beef is more sustainable than anchovies,” says Amoroso. “But some products that come from the ocean are way more sustainable than the products that come from land.”

On the flipside, claiming that just 4 percent of the ocean is fished could lure people into thinking that fish stocks are in good health, and the seas are largely untouched. That’s clearly not true. Overfishing is rampant, and stopping it has proven to be very hard. Currently, marine protected areas, where fishing is restricted and sometimes completely banned, cover just 7 percent of the world’s oceans, and the goal of extending them to 30 percent by 2030 is seen as an ambitious one. “If only 4 percent of the ocean is fished, you should be able to make 96 percent of it into a marine protected area without anyone complaining,” says Kroodsma. “Good luck with that.”

Still, everyone applauds the data that the Global Fishing Watch has accumulated and freely released. “Their work is great, and having those data publicly available is fantastic,” says Amoroso. Indeed, in just the past few months, researchers have used that data to show that wealthy countries dominate the world of industrial fishing, to identify supply chains where vessels offload catches to each other at sea, and to show that the creation of marine protected areas can lead to a wave of pre-emptive overfishing. The GFW’s partners are also looking at how fishing activity overlaps with shark habitats, how longline fleets affect albatrosses, and more.

“It can revolutionize our understanding of human activity in the world’s oceans,” says Kroodsma.