At the very beginning of his book The Song of the Dodo, the author David Quammen invites us to imagine a fine Persian carpet, which we then slice into 36 equal pieces. “What does it amount to?” he writes. “Have we got 36 nice Persian throw rugs? No. All we’re left with is three dozen ragged fragments, each one worthless and commencing to come apart.” He wrote that almost two decades ago, and it’s still the perfect metaphor for the state of the world’s forests.

Humans chop down an estimated 13 million hectares of trees every year, but even that huge number doesn’t fully capture the destruction we inflict. We don’t just destroy forests—we fragment them, turning unbroken stretches of green into ragged patches fraying at the edges. There are only two places on Earth—the Amazon and the Congo—where forests have retained their old, continuous glory. Everything else has been partitioned into green islands, separated not by water but by roads and farmland.

It matters, this slicing of the world. Just as Quammen’s 36 swatches of fabric don’t add up to a full carpet, fragmented forests are less conducive to life than their total area might suggest. Each green island is limited in how many species it can sustain. And since these populations are cut off from their neighbors, they’re uniquely vulnerable to disease, disaster, or dumb, bad luck. For these reasons, fragmented habitats tend to lose half their animal and plant species within two decades.

Many of the survivors are, quite literally, living on the edge. About 70 percent of the world’s remaining forests lie within a kilometer of an edge, and about half lie within 500 meters. And in a new and unprecedentedly detailed study, a team of scientists led by Marion Pfeifer at Newcastle University has shown that 85 percent of forest animals are affected by the presence of an edge. Some benefit, others suffer. Either way, the results show that humans, by fragmenting forests, have radically restructured them—and that the edge of a forest is almost a completely different world than its core.

Many researchers have tried to assess edge effects in simple ways, such as measuring how the abundance of a species changes as it gets closer to an edge. This ignores the fact that a patch of forest might be within reach of many edges, each of which compounds the hazards of the others. It also treats land simplistically, as if it’s either forest or nonforest; in reality, the land surrounding a group of trees can vary a lot in its usefulness and attractiveness to local species.

To improve on these methods, Pfeifer’s colleague Véronique Lefebvre from Imperial College London developed two new metrics for making sense of fragmented landscapes. The first—edge influence—reflects the variations in tree cover around a given point; roughly speaking, it tells you how “edgy” a particular bit of forest is. The second metric—edge sensitivity—reflects how edgy a habitat has to get before a species avoids it. They then calculated edge sensitivities for 1,673 vertebrate animals around the world. “The study sets a new standard of how edges should be measured,” says Nick Haddad from Michigan State University, who was not involved in the work.

They found that around 39 percent of these species steer clear of edges while 46 percent seem to gravitate toward them. This doesn’t mean that edges are a net good, though. The edge-seeking animals are more likely to be generalists or invasive species, such as green iguanas and boa constrictors. By contrast, the species that stick to the forest core, such as the Sunda pangolin and Baird’s tapir, are almost four times as likely to be on the endangered list. They’re the creatures most at risk of extinction, and their shrinking habitat is even smaller than it first seemed. There’s a lot of them too. About 57 percent of mammals decline in number as you get closer to edges, as do 41 percent of birds, 30 percent of amphibians, and 11 percent of reptiles.

“Edge effects have long been studied but not with the level of detail in this study,” says Isabel Rosa from the German Centre for Integrative Biodiversity Research. “It highlights the importance of considering not only the habitat amount that is available for the species, but also its quality.”

For example, the team’s calculations show that forest-core species only thrive at least 200 to 400 meters away from an edge. To a long-billed black cockatoo, a circular clump of trees that’s 800 meters in diameter is not 200 hectares of inviting greenery—it’s nothing. To a Sunda pangolin, a strip of forest that’s extremely long but just 800 meters wide might as well be no forest at all. “We need to manage patches in a way that maximizes the forest core,” Pfeifer says. Her team has developed a freely available tool called BioFrag to help them, by calculating the edginess of their forests and the edge sensitivities of the local animals.

Conservationists do already think about edge effects, Haddad says, but typically on the scale of tens of meters—not hundreds. Pfeifer and her colleagues have shown that edges are more pernicious, and that their influence extends far deeper into a forest. Only half the world’s forests lie more than 500 meters from an edge, and conservationists should focus on the species that persist in this invisible corral. If they ignore these edge effects, Haddad says, “they may be missing the forest for the trees.”