Ophiocordyceps unilateralis, a fungus that creeps along the forest floors of Thailand, Brazil, and the Amazon basin, gets into a carpenter ant through a single spore that lands on its cuticle, drills inward with enzymes, and then spends the next two to three weeks doing something biologists still cannot fully explain: it takes the wheel. The ant keeps foraging, keeps grooming, keeps responding to nestmates. Then, on a schedule the fungus appears to dictate, it leaves the trail, climbs down from the canopy to a leaf roughly 25 centimeters above the ground on the north side of a sapling, bites into the central vein with a grip that will not release, and dies. A stalk grows out of the back of its head within days and begins raining spores onto the foraging trail directly below.
The fungus has been doing this for at least 48 million years. Fossilized leaves in the Messel Pit in Germany carry the unmistakable scars of a death grip, the oldest known evidence of this exact behavior preserved in stone.
The hijack begins in the cuticle
A spore lands on a foraging Camponotus leonardi worker — the species most often studied in Thailand’s Khao Chong rainforest. The spore germinates within hours. It secretes enzymes that dissolve the chitin armor and pushes hyphae into the ant’s hemocoel, the open body cavity where insect blood sloshes around organs.
From there it grows. Slowly.
The fungus does not kill the ant outright because a dead ant cannot climb. Instead it begins multiplying as single yeast-like cells that travel through the hemolymph, colonizing muscle tissue first. By the time the ant starts behaving strangely, the fungus has converted somewhere around 40 percent of its body mass into fungal cells. The ant is, by weight, almost half mushroom.
The strangest part: the brain is mostly spared
One of the most counterintuitive findings came from tissue reconstruction using serial block-face electron microscopy. The fungal cells form dense interconnected networks around the muscle fibers of the mandibles and legs but largely avoid the brain itself.
The hijack is peripheral, not central. The ant’s brain is, in a sense, still trying. The body has simply stopped listening.
Quanta Magazine’s reporting on how these pathogens commandeer behavior describes the fungus secreting a cocktail of bioactive compounds — ergot alkaloids, enterotoxins, and molecules with no known analog — that appear to lock the mandible muscles in a sustained contraction the ant cannot override.
Climb at noon, bite at noon-thirty
Field observations in Thailand showed the behavioral sequence is remarkably consistent. Infected ants leave the colony in the late morning. They wander erratically, falling off branches, climbing back up. They do not return to the nest. Healthy nestmates detect infected workers and carry them away from the colony, sometimes dropping them off cliffs — a sanitary behavior that limits but does not eliminate transmission.
By solar noon, the surviving infected ants converge on understory vegetation between 20 and 30 centimeters off the ground. They almost always choose the north-northwest side of stems. They almost always select leaves where the temperature stays between 20 and 30 degrees Celsius and humidity hovers above 94 percent. These are the conditions the fungus needs to fruit.
Then they bite. And they do not let go.
The death grip leaves a scar
The mandible lock is mechanical and irreversible. Atrophy of the muscles that open the mandibles, combined with sustained contraction of the closing muscles, fuses the jaw to the leaf vein. The ant dies within hours, still attached. Its body becomes a platform.
Over the next four to ten days, a stroma — the reproductive stalk — emerges from the soft tissue at the back of the ant’s head, just behind where the head joins the thorax. It grows two to three centimeters long, thicker than the ant itself, and curves toward the ground. A bulbous structure forms near the tip and begins releasing spores in a sticky rain that falls onto the foraging trails below.
This is why the height matters. Too low and the spores fall onto leaf litter where no ants forage. Too high and they desiccate before reaching the ground. The 25-centimeter sweet spot puts the spore shower directly into the path of Camponotus foragers using the same scent trails the infected ant once walked.
The fungus carries stolen genes
How a mushroom evolved the chemistry to puppet an insect remains one of evolutionary biology’s open questions, but research on related systems suggests parasites manipulate hosts using genes they apparently stole from the hosts themselves — picking up insect DNA over evolutionary time and repurposing it to speak the insect’s own biochemical language.
The implication is unsettling. The fungus may be using molecular signals the ant’s nervous system was already built to obey.
It is one of many hijackers
Behavioral manipulation by parasites turns out to be common once biologists started looking. Toxoplasma gondii makes infected rodents lose their fear of cat urine. Hairworms drive crickets into water to drown. Lancet flukes climb into the brains of ants and steer them onto grass blades at dusk to be eaten by grazing sheep. A growing body of work on parasite-induced behavioral manipulation catalogs dozens of such cases, many involving arthropods.
Strepsiptera — twisted-wing parasites — burrow into wasps and alter both their flight behavior and their reproductive anatomy. Research on Strepsiptera host dynamics shows infected wasps gather at sites where the parasites can mate, drawn there by chemistry they did not choose.
Ophiocordyceps is unusual mainly in how complete the takeover is, and how precisely engineered the final act looks.
Ants fight back
Carpenter ant colonies are not passive victims. Workers groom each other obsessively, removing spores before they can germinate. They produce antifungal compounds in their metapleural glands. They carry sick nestmates away from the brood chamber. They avoid foraging directly under known graveyards of fungus-killed ants — and in lab studies, scouts have been observed rerouting trails around stalk-bearing corpses.
The fossil record at Messel suggests this arms race has been running for tens of millions of years. The 48-million-year-old leaves with death-grip scars mean the fungus already knew how to hijack ants when early primates were still scrambling through Eocene canopies.
And the conflict still shapes evolution today. Recent research on climate-driven ant-parasite arms races found that temperature and humidity patterns determine which side is winning in any given region. Warmer, wetter forests favor the fungus. Cooler, drier ones favor the ants.
The forest is full of stalks
In Khao Chong, researchers have mapped what they call graveyards — patches of understory where dozens of infected ants had bitten leaves within a few square meters. The density can reach more than 25 corpses per square meter in places. Each one a small brown ant, jaws clamped, a slender dark stalk arcing from the head, spores still drifting down weeks after death.
The fungus has been doing this since before there were rainforests in their current form. It will keep doing it, in the same precise way, on the same north-facing leaves, at the same noon hour, for as long as carpenter ants walk beneath them. The ant climbs. The mandibles close. The stalk grows. The spores fall.
Somewhere in a Thai forest right now, an ant is making the climb.
