It’s been 60 years since Dr. Dan Janzen discovered the ant-plant mutualism. It was Veracruz, Mexico in 1961. Janzen, then a graduate student at UC Berkeley, noticed an ant chasing a beetle off an acacia branch, and out of pure curiosity, he decided to dissect it. 

“I discovered that there were ants all over the surface,” he said. Remarkably, they were inside the plant, too. A colony of ants with a single queen were packed into its swollen, hollow thorns, roaming in and out of chewed-out holes like a busy tunnel during rush hour. 

That field season, Janzen would down several more acacias to dissect. Most of the time, he would bring them home successfully, but on one ambitious trip when he tried to take two at once, he decided he couldn’t take both without getting swarmed by angry ants. He left one plant behind, and what he would find a month later became a lightbulb moment. 

The stump of the plant he brought to the lab to study had been decimated by herbivores. But just a meter and a half away, the one left behind was regenerating, with verdant green shoots climbing towards the sky. The ants, Janzen thought, must have returned to the stump and defended it after he dropped the branch. “It began to sort of click in my brain,” he said. 

Over the next several years, Janzen would remove ant colonies from some 5,000 acacias—some in heavy shade, others in open sun, some by the ocean, others near riverbeds. And each time, the plants, stripped of their defenses, became prime pickings for insects, mice, deer, and any other grazer out there. “Everybody eats the acacias when there’s no colony… it’s basically defenseless,” he said. 

Janzen’s discovery would become one of the classic examples of mutualistic symbiosis—symbiosis meaning the species live in close association, and mutualistic meaning they benefit from each other. Today, there are at least 681 plant species and 113 ant species known to be mutualists. In the most extreme cases, called obligate mutualisms, the two can’t survive without each other. But even then, they’re not doing it out of the kindness of their hearts. It’s for their own benefit.

Dr. Megan Frederickson said that the ant-plant mutualism began as a “happy accident.” 

Millions of years ago, an ancient carnivorous ant began hunting in trees for insects like beetles and caterpillars. By happenstance, plants benefited because it culled hungry herbivores, said Frederickson, an evolutionary biologist at the University of Toronto. Plants became pressured to attract and retain ants. Some began feeding them directly, leading to nectaries that grow outside of their flowers. Then, they hollowed out stems and thorns or curled up leaves to produce housing structures called domatia. Others even evolved fat- and protein-rich structures called food bodies to provide a whole, nutritious diet to their colonies.

In time, this “happy accident” became more specialized—more codependent—because in general, ants and plants have similar interests. Ants want larger plants to accommodate larger colonies, and plants want more guards. “It was a gradual process over many millions of years,” Frederickson said. 

But while their interests were largely aligned, they weren’t entirely so. An undercurrent of self-interest bled under the veneer of total cooperation. Take, for example, Acacia cornigera. This plant feeds its ant colonies a peculiar nectar that contains no sucrose—an oddity among plants. Most ants wouldn’t feed on this, but these mutualistic ants only do—or more precisely, it’s the only nectar they can feed on. The plant makes sure of that. 

This odd nectar contains an enzyme called chitinase, which disrupts sucrose digestion in the ant’s gut. Once they have one sip, they become entirely sucrose intolerant. Heil compared it to an oat milk company selling a drink that inhibits lactase, the enzyme that helps humans digest dairy milk. “You would have to go back to that company and buy that milk again because you can’t digest lactose milk. That is what the plant is doing with the ants,” he said.

Betrayal goes both ways, and ants aren’t innocent. Frederickson, for example, has discovered that some ant species chew and kill their host’s flowers, preventing them from reproducing. It seems convoluted, but freeing resources from seeds funnels them into ant-friendly resources like nectar and housing. The ants don’t care if the plant reproduces, after all. They just need it to grow. 

Although full-on cheating is rare, Frederickson said, mutualistic partners are certainly incentivized to get the better end of the deal, even if it’s just incremental improvement. These negotiations seem trivial—inconsequential dealings taking place behind the closed doors of hollow thorns—but the impact spreads far and wide. 

In fact, it can make or break entire ecosystems. 

Enter the devil’s garden.

Here, in the western Amazon rainforest, the forest floor is bare. There’s no understory. Ants run far and free. This patch of land is a far cry from the diverse, lush greenery that envelops it. “It’s like a hole in the tropical rainforest,” a void of diversity where one species of tree, Duroia hirsuta, dominates, said Frederickson. 

“Devil’s garden” is a poor translation of the Quechua word, supaychacra. A better translation might be “gardens created by an evil forest spirit,” Frederickson said.

“They’re very unusual places,” she added. “It feels like you’re walking in an apple orchard or pine plantation or something human made.” 

But it’s not spirits or humans that tend to the trees. It’s ants. 

D. hirsuta hosts colonies of ants in its swollen, hollow stems. The insects defend their hosts from encroaching plants of other species by stinging them and injecting formic acid into their veins. The result is an eerie patch of forest up to half a hectare, about the size of an American football field, all created by a single colony.

The fierce defense from tiny ants shapes whole landscapes, and it’s not just in devil’s gardens. Across the Atlantic Ocean, in Kenyan savannas, large patches of land similarly consist of a single species: the whistling thorn, Acacia drepanolobium.

“We’re talking about thousands of square kilometers. You can see these monocultures from space,” said Dr. Jake Goheen, an ecologist at the University of Wyoming.

In part, these single species stands occur because the soil is harsh. It expands and contracts as it wets and dries, winnowing the potential species that could survive there. But it’s also because of symbiotic ants, which defend whistling thorns from massive megaherbivores—namely, elephants. 

“Elephants are the only herbivore that can kill a tree in one meal,” Goheen said. “[They] can walk up to a tree, snap it in half at the trunk, and kill it.” But the ants aren’t deterred. 

In the rare case that an elephant tries to feed on a whistling thorn, it’s met with an unwelcome greeting. Thousands of ants climb into its nostrils and up its trunk, stinging and latching to the sensitive inner skin. “It’s not a deadly mistake,” said Dr. Todd Palmer, an ecologist at the University of Florida. But because a four-meter tree can have 100,000 ants, it’s undoubtedly unpleasant. 

Even though 98% of the plants at Palmer’s field sites are A. drepanolobium, elephants won’t touch them, even if there’s a drought—even if they’re starving. And that affects everything from fire patterns to soil nutrients to animal movement. “It’s kind of crazy,” Palmer said. 

Entire ecosystems are built on this mutualism, which itself is built upon a careful negotiation of costs and benefits, of manipulations and self-interest. These ecosystems therefore rest precariously on a paper-thin margin between friend and foe. And now, human change is causing it to wobble on that edge. Homo sapiens are shuttling species into new habitats and driving others into extinction. That changes the math for the ants and plants; it changes whether paying the price of the mutualism is worth it. And if it topples, the ecosystems they build will, too.

According to the IUCN, African savanna elephant populations have declined by at least 60% in the last 50 years. Without elephants, there’s little benefit to hosting ants. That “changes the calculus” for ant-plants, Palmer said. 

Supporting an ant colony is costly. Plants expend valuable carbohydrates and nitrogen to feed and house them. One plant that Heil studies spends 20% of its nitrogen on the mutualism, and Palmer’s whistling thorns actually grow better without ants. Despite that, it’s worth it. “Twenty percent of your nitrogen to survive is a good deal,” Heil said. 

But it might not be for long, at least for whistling thorns. In savannas, as elephants disappear, the benefit of hosting plants is dwindling too, and the cost is becoming harder to justify. 

Palmer and Goheen recently simulated a worst-case scenario where elephants went entirely extinct. In a decade-long experiment, they excluded elephants from patches of savanna for 10 years. The changes were sweeping. 

Without elephants, whistling thorns invested less into feeding and housing ants; they produced less active nectaries and fewer swollen thorns. “That has a cascading impact on the acacia [and ant] community,” Palmer said. The change in plant investment altered which species of ant occupied the plant—and, ultimately, how many trees survived. 

The most protective species that occupies whistling thorns is Crematogaster mimosae. They raise their brood, house their workers, and even farm scale-insects inside swollen thorns. Normally, they occupy 52% of trees at Palmer’s field site, but without elephants and with fewer swollen thorns and nectaries needed to sustain their colonies, that percentage decreased by nearly a third. 

Meanwhile, Crematogaster cjostedti isn’t very protective, and it doesn’t live in swollen thorns. It dwells in the tunnels excavated by a beetle pest that infests the plant. Normally, they occupy 16% of host plants, but without elephants, that percentage doubled. The now dominant C. cjostedti doesn’t kill the beetles, allowing the pests to spread across the landscape and doubling the rate of tree mortality. C. cjostedti is “almost parasitic,” Palmer said. “The [beetle] larvae burrow their way through the tree, sort of like a cancer.”