In 1977, two years after declaring independence from Portugal, Mozambique erupted into civil war. Over the next 15 years, the violent conflict claimed at least a million lives—and that was just the humans.

Government troops and resistance fighters also slaughtered their way through the wildlife in the nation’s renowned Gorongosa National Park, once touted as a natural paradise. Thousands of elephants were hunted for their ivory, which was sold to buy arms and supplies. Zebras, wildebeest, and buffalo were killed for meat. Around 90 percent of the park’s large mammals were shot or died of starvation.

“They caused almost total collapse of the wildlife there,” says Joshua Daskin, an ecologist at Yale University who started working at Gorongosa in 2013.* “I wondered if that was a one-off, or emblematic of a wider trend.”

Spoiler: it’s the latter. Together with Rob Pringle, from Princeton University, Daskin compiled 65 years’ worth of data on the abundance of large mammals across all of Africa. These populations, they found, were stable during peacetime, but almost always fell during periods of war. And in explaining declines in wildlife, nothing mattered more than war—not human population density, the presence of towns or cities, protected reserves, or droughts.

“This speaks to the pervasive nature of conflict,” says Daskin. “It affects the ability, accountability, and motivation of governments to fulfil their conservation duties. It disturbs the fabric of local societies by increasing poverty, and displacing people into protected areas where they may harvest wildlife. It leads to withdrawal of NGOs. It increases problems with law enforcement, which might lead to increases in poaching.”

To make matters worse, as others have shown, war happens most often in places where wildlife otherwise flourishes. Between 1950 and 2000, 80 percent of major armed conflicts took place in biodiversity hotspots, where animal life is at its richest and most diverse. That, says Daskin, is because the same factors that cause peril for wildlife—climate change, the harvesting of natural resources, and fast-growing human populations—can also heighten tensions between people. And so, when people declare war on each other, they inadvertently declare war on the natural world.

That seems intuitive, but several case studies have shown that war can be a boon for the wild. The Rhodesian Bush War, in what is now called Zimbabwe, created an environment so hostile that poachers couldn’t operate—and elephant populations rebounded to decadal highs. The Korean Demilitarized Zone, which snakes between North Korea and South Korea, has become a de facto national park, since the absence of people means sanctuary for red-crowned cranes, Amur leopards, and other endangered species.

But for every tale of hope, there’s also one of doom. Animals can be collateral damage, sources of meat and money, or even political leverage. In Vietnam, the chemical weapons that the U.S. military used to denude the forests of vegetation left a toxic legacy for the nation’s jungle-dwellers. In Ethiopia, black-market weapons that were sold during the country’s civil war eventually made their way into the hands of poachers. In the Democratic Republic of Congo, rebels recently threatened to kill protected gorillas if the government took action against them.

It’s hard to work out the balance of these positive and negative anecdotes because, “as you might expect, ecologists tend to work in peaceful areas, so most of the wildlife counts haven’t been done in conflict zones,” says Daskin. But he and Pringle found whatever data they could: 253 time-trends, showing changes in the populations of 36 species in 126 protected areas. They then paired these counts with information on human fatalities from organized conflicts.

Protected areas vary a lot, but on average, Daskin and Pringle found that peaceful ones were stable and their animals self-sustaining. Warfare changed everything—and it didn’t really matter how violent the conflicts were. In fact, while the frequency of conflict was the most important of the factors the duo studied, the intensity of conflict was the least important. “It may not matter whether this is a small-scale battle or a large-scale war,” says Daskin. “The onset of conflict disrupts the ability to protect wildlife.”

Since people in war-torn areas “often rely heavily on wildlife, directly as food or as part of a healthy ecosystem,” says Kaitlyn Gaynor, from University of California, Berkeley, who has studied the issue, “wildlife declines may prolong the wars that caused them.”

At first glance, these seems like deeply defeatist findings. Conservationists are already overwhelmed and underfunded without having to add peace-keeping to their dockets. But Daskin and Pringle offer some good news, too: Even in places where people warred furiously, very few large animals actually went extinct. They took a pounding, but they were rarely knocked out. So, when conflicts cease, it’s possible to save the creatures that were harmed.

Consider Gorongosa National Park. “Most of the wildlife populations that were hanging on have recovered to incredible levels, some cases beyond pre-war sizes,” Daskin says. That’s largely because of a partnership between the government and the Gorongosa Restoration Project, a U.S.-based nonprofit.

In the post-war decades, the partnership has recruited and trained rangers to fight poachers. They deployed teams to boost the health of people who live near the park, through vaccinations, prenatal check-ups, family planning advice, and bed nets to block out malarial mosquitoes. They built four schools. They bring in more than 2,500 children a year to learn how to view wildlife as more than just sources of meat. “It’s about instilling a sense of ownership, and enabling the socioeconomic conditions for conservation to be possible,” says Daskin.

“There is incredible diversity in the nature of armed conflict and its effects on wildlife, and there is no one-size-fits-all solution,” says Gaynor. But the study clearly shows that “war-torn protected areas are not necessarily a lost cause for conservation. We ignore these areas to the peril of wildlife and, critically, to the detriment of vulnerable human populations that rely on healthy ecosystems.”

* This article previously misstated that Daskin is at Princeton and that Pringle is at Yale. We regret the error.