They Ate My Garden and I'm Supposed to Be the Grown-Up Here

In Costa Rica they are zompopas, feminine, because every worker in the colony is female. In the rest of Central America, zompopos. In the scientific literature, leaf-cutter ants of the genera Atta and Acromyrmex. Whatever you call them, if you live in the rural tropics, you already know what they are.

The first thing you notice is the trail. A line of green confetti moving across the forest floor, or your patio, or the trunk of the avocado tree you planted last year. Each fragment is a piece of leaf, cut with precision and held overhead by an ant roughly the size of your fingernail. Follow the trail in one direction and you find the source: a tree being methodically disassembled. Follow it the other way and the fragments disappear into a hole in the ground.

That hole leads to one of the most complex structures built by any animal on earth. And the ants are not eating the leaves.

What You're Looking At

The trail you're watching is a highway. Hundreds of ants move in both directions, loaded foragers heading in and empty ones heading out, with no visible coordinator and no collisions.

What you can't see is underground. A mature colony of Atta cephalotes, the more widespread of Costa Rica's two Atta species, occupies a nest up to six meters deep with hundreds of chambers. Over a colony's lifetime, workers excavate up to 40 metric tons of soil. Ventilation turrets on the surface channel fresh air through peripheral tunnels while stale air exits through central passages. The ants monitor CO₂ levels inside and reshape the turrets accordingly: when carbon dioxide rises, they build shorter turrets with wider openings.

On the surface, a mature nest is unmistakable. A large colony can have dozens of openings spread across a clearing the size of a tennis court, and you can tell them apart by what surrounds them. Entry holes have crescent-shaped mounds of pale excavated soil, each roughly the size of a dinner plate; foragers stream in and out carrying leaf fragments. Ventilation openings are smaller, may have the turret structures described above, and see little or no foot traffic. Because A. cephalotes buries its waste underground, you will not see obvious refuse piles around the nest. If you find a leaf-cutter colony with dark, organic waste heaped at the periphery, you are probably looking at A. colombica, the other Costa Rican species, which dumps its refuse on the surface. Underground, the nest has hundreds of chambers distributed through a zone one to six meters deep, connected by a network of tunnels. The queen resides in one of the fungus garden chambers, and workers distribute her eggs to gardens throughout the nest.

In a climate that delivers three to four meters of rain a year, the nest's survival starts with where it was built. At La Selva, colony density peaks on slope tops and drops to zero in valley bottoms. The funnel-shaped geometry channels water downward through the tunnel network, and the excavated soil, pelletized into porous aggregates by the ants' mandibles, drains faster than the undisturbed clay around it. Young colonies take no chances during downpours: workers seal all entrances with twigs and clay crumbs, accepting dangerous CO2 buildup rather than risk flooding. After the storm passes, they reopen the entrances and rebuild any damaged turrets taller than before.

A single queen runs the colony. She mated once, during a nuptial flight, and stored between 200 and 465 million sperm. During founding, the queen lays about two eggs a day. At maturity, she lays more than 25,000, roughly one every three seconds, fertilizing each with one or two sperm. As she ages, she uses more sperm per egg, and when the supply runs out, the colony dies. The math gives her eight to fifteen years. Over that lifespan, she may produce 150 million daughters or more. At any given moment, the standing population peaks at several million workers, but individual ants live only months to a couple of years, with larger castes outliving smaller ones, so the colony is constantly replacing its own workforce.

Those workers come in four size classes. The tiniest, called minims, are about a millimeter long and tend the fungus gardens. Minors patrol the foraging columns. Mediae are the foragers you see carrying leaves. And the majors, the soldiers with heads the size of apple seeds, defend the colony.

If you step on a trail or put your hand near a nest entrance, the soldiers will bite. They cannot sting (unlike the hormiguillas, Wasmannia auropunctata, that make every Costa Rican gardener miserable) but their mandibles are reinforced with zinc at the atomic level, increasing the hardness of the cutting edge threefold, from something like plastic to something like aluminum. A soldier's bite can draw blood. Indigenous peoples across the Americas used this to their advantage: they held a soldier ant to a wound, let it clamp down, and snapped off the body, leaving the locked mandibles as a suture. The mediae and minors can also bite, though less impressively.

What zompopas are not is a household pest. They do not invade kitchens, contaminate food, or nest in walls. Their foraging is entirely focused on living vegetation. If you have lived with pavement ants or crazy ants (the tiny, fast ones that swarm erratically across every surface) invading your countertops, zompopas are a completely different problem: they will strip your citrus tree but ignore your pantry. The annoyance is in the garden, not the house.

And Then It Was Everywhere

When a new queen lands after her nuptial flight, she carries a fragment of the fungus garden in a pouch behind her mouthparts. She digs down about 12 centimeters and builds a chamber the size of a golf ball. She seals the entrance and does not eat for nine weeks, living off stored body fat and her own dissolved flight muscles while tending that fragment into her first crop of fungus.

By two months, she has about 120 workers and the entrance reopens. The nest is invisible from the surface. By six to eight months, with around 660 workers, the first visible sign appears: a small tower of granulated soil, about the size of a fist. The queen is now a meter deep.

If you do not want a colony in your garden, this is your window. In the first eight to twelve months, the nest has one or two chambers and the queen is within a meter of the surface. You can dig it up with a shovel. After that, a feedback loop takes over: more workers harvest more leaves, which grow more fungus, which feeds the queen to lay more eggs. By two and a half years, there are 65,000 workers, 13 chambers, and the nest covers about 10 square meters, roughly the footprint of a large car. The queen is three meters down. Excavation is no longer practical. By five years: hundreds of thousands of workers, more than 100 chambers, 40 square meters, the floor area of a studio apartment. At three years the colony reaches sexual maturity and begins producing its own queens. A mature colony of eight to fifteen years has millions of workers and typically spans 50 to 150 square meters, though the largest South American species can exceed 600. Researchers who attempted to destroy a mature nest with a bulldozer found the colony had rebuilt five meters away within 60 days.

This growth curve explains something that puzzles people who spend time in the forest. You walk the same trails for years and see no leaf-cutter activity. Then you find one large colony. Within months, large colonies seem to be everywhere, some within a hundred meters of each other. The jungle has not changed. All colonies in an area release queens on the same nights during the rainy season. The surviving founders are the same age. For their first two years they are underground and invisible: a coin-sized entrance, a few hundred workers foraging close to the nest at night. Then the feedback loop kicks in and the cleared trunk trails, the excavation mounds, the defoliated branches all appear within the same season. They were always there. You just couldn't see them yet.

In old-growth forest at La Selva Biological Station, nests are regularly spaced rather than randomly distributed, a pattern driven by territorial exclusion. A colony that reaches sexual maturity at three years produces queens every rainy season for the rest of its life. Those queens fly one to two kilometers on their nuptial flights, sometimes farther. Most die. Colony density at any given site rises and falls. The only long-term population dataset, tracking Atta mexicana over 30 years, recorded a 232 percent increase in colony density in a single decade driven by favorable wet seasons. None of the original colonies survived to the final census; the population was dominated by colonies less than seven years old. In the disturbed, fragmented lowlands that now define much of Costa Rica, the conditions that favor leaf-cutters (light gaps, edge habitat, pioneer vegetation) regenerate continuously. Individual colonies come and go. The population persists.

They Are Not Eating the Leaves

The leaves are substrate. Inside the nest, workers chew them into a wet pulp and feed it to a fungus, Leucoagaricus gongylophorus, that grows in underground gardens. The ants eat the fungus.

This is agriculture. It started 66 million years ago, in the aftermath of the asteroid that killed the dinosaurs. The impact temporarily shut down photosynthesis, and fungi proliferated on decaying vegetation. Ants began cultivating them. Leaf-cutting, the most advanced form of this agriculture, evolved around 27 million years ago. Human agriculture is roughly 10,000 years old.

The fungus has evolved inflated hyphal tips called gongylidia, grown in clusters called staphylae, specifically to feed the ants. It can no longer survive as a free-living organism. The ants, in turn, have lost the genetic ability to synthesize arginine, an amino acid essential for protein construction, immune function, and wound healing. The fungus provides it. Without the fungus, the ants starve at the molecular level even if other food is available. Neither organism can survive without the other. This mutual dependence has been locked in for at least 20 million years.

The most detailed excavation of an Atta colony on record, a medium-sized nest of the Texas species, revealed 169 chambers arranged in a funnel shape that narrowed with depth: 97 contained fungus gardens and 45 held waste. An excavation of an A. cephalotes nest in Suriname in 1939 found 373 chambers, 344 of them containing fungus. Each garden is a roughly spherical cavity, typically 20 to 30 centimeters across. Not all chambers are in use at any given time. New ones are constantly being excavated while old ones are abandoned, so the nest is a shifting structure rather than a fixed blueprint. As a colony ages, the proportion of active gardens shrinks. A six-year-old A. sexdens colony in Brazil contained 1,920 chambers, but only 248, about 13 percent, still held fungus. Waste chambers are individually much larger than garden chambers, so in a mature nest most of the underground volume is either empty or filled with decomposing refuse. All of these excavations are from species that bury their waste underground. In A. colombica, which ejects refuse to the surface, the nest would lack that waste volume entirely, but there are no studies on record of confirmatory excavations. Over the years the nest grows wider as workers dig new chambers in undisturbed soil, typically at the margins. Active gardens, empty voids, and waste chambers coexist throughout the nest at any given time. How this spatial mosaic shifts over a colony's full lifespan has not been tracked in any published study.

Inside a garden chamber, the fungus grows as a spongy mass with a clear vertical gradient. Fresh leaf material goes in at the top, where recognizable fragments sit with sparse fungal growth. The middle zone is where mycelium is densest and where most gongylidia are produced. This is also where the colony keeps its brood: the eggs, larvae, and pupae that are the next generation. Workers chew the staphylae and place them where larvae can consume them directly. The bottom layer is exhausted substrate, dark and depleted, ready for removal. The fungus does not physically descend; new material is continuously added on top, and the mycelium migrates toward it, leaving consumed substrate behind. A single garden stays productive for roughly three to four weeks before it is spent, but a mature colony runs dozens to hundreds of gardens simultaneously, each at a different stage. A queen laying 25,000 eggs a day produces brood continuously, and that brood is distributed across the active gardens, not concentrated in one place. When conditions in a chamber deteriorate, workers relocate the brood first. Other workers aggregate around the brood at a new site, and their increased density triggers digging. The fungus follows. The spent chamber is left empty; excavation studies find these voids still open in the nest structure, but their long-term fate is not well documented.

A leaf fragment arriving at the nest passes through roughly ten processing steps, each performed by successively smaller workers. Medium-sized workers hold the fragment while others cut it into smaller pieces. Roughly 90 percent of all cutting happens underground, not at the foraging site. The fragment is licked, scraped, and punctured to break through the waxy cuticle, then reduced from its original half-square-centimeter to a sliver averaging four thousandths of a square centimeter, a 120-fold reduction. Workers then deposit fecal droplets onto the prepared fragments. These droplets contain fungal enzymes, produced in the gongylidia the ants eat, that survive gut passage intact. The fungus is using the ant as an enzyme delivery vehicle: enzymes and reactive compounds in the fecal fluid arrive at fresh substrate before the mycelium does, pre-digesting plant cell walls. The smallest workers then pluck tufts of mycelium from mature parts of the garden and plant them onto the prepared fragments.

The separation between gardens and waste is enforced through overlapping mechanisms rather than a single barrier. When spent substrate is removed from a garden, it usually passes through a caching site where a dedicated waste worker picks it up; in about 94 percent of transfers, the garden worker and the waste worker never make direct contact. In the remaining cases, a garden worker either hands waste off directly at the heap entrance or, in about 3 percent of transfers, enters the waste area herself. Waste workers are permanently assigned to their role, and if one tries to re-enter the nest through a foraging entrance, other workers attack it. The minims that tend the gardens around the clock, weeding alien fungal spores and applying antimicrobial secretions from their metapleural glands, never handle waste. The system leaks, but redundant defenses keep garden and waste microbial communities distinct.

The Trail System

Leaf-cutters forage through a system of permanent trails that radiate outward from the nest. A mature colony maintains roughly 250 meters of cleared trunk trail at any given time and builds over two kilometers of new trail per year. Individual scouts explore outward from these trails. When a scout finds promising foliage, she returns to the nest dragging the tip of her abdomen along the ground, depositing a chemical from her venom gland. The compound, methyl 4-methylpyrrole-2-carboxylate, was the first ant trail pheromone ever identified. It took 3.7 kilograms of dried ants to isolate it in 1971. Each worker carries about one nanogram. One milligram would theoretically be enough to lay a detectable trail three times around the Earth.

The trail triggers a positive feedback loop. Recruited workers follow it to the food source and reinforce it on their return. Within hours, a stream of foragers becomes a column. The quantity of pheromone a returning worker deposits depends on the quality of the food and how hungry the colony is. If the food source is poor, returning workers lay less pheromone, the trail fades, and the column dissolves. If it is good, the trail strengthens and more workers pour out.

Heavily used trails become physically cleared highways. Workers encountering a twig or leaf on the trail remove it with a low, fixed probability, about one in five thousand encounters. A subset, roughly 15 percent of those who remove one obstacle, enter a clearing mode and remove several more in sequence. No communication is needed. The result, over days, is a bare-soil path the width of a human hand, maintained for months or years.

Traffic on a busy trail is self-organized. Leaf-laden ants returning to the nest have right-of-way; outbound ants without cargo yield. At peak density this priority system prevents jams without any central coordination.

What a scout considers promising is shaped by the colony's history. When a scout encounters familiar foliage, she runs back to the nest laying trail without even touching the leaves. When she encounters an unfamiliar species, she makes probing bites at the leaf edge, assessing its physical properties, and cuts a smaller fragment than usual to carry back quickly, prioritizing information transfer over load size. The colony then tests the new species through the fungus. If the fungus rejects it, foragers stop collecting it within days, a process called delayed rejection. Pioneer species and sun-grown leaves are preferred over shade-grown foliage, because they contain more water and fewer defensive compounds. Species rich in saponins, bitter compounds that foam in water and disrupt cell membranes, get avoided. A 1996 study of five Costa Rican tree species found the ants strongly preferred fruta dorada (Virola koschnyii) and pilón (Hyeronima alchorneoides) while ignoring gavilán (Pentaclethra macroloba), the dominant canopy tree in Caribbean lowland forests, whose leaves are loaded with saponins. Over time, a colony builds a working knowledge of which plants in its territory are worth harvesting and which are not, and scouts adjust their behavior accordingly.

The Vibratome, the Hitchhiker, and the Parasitoid

When a forager cuts a leaf, she vibrates her mandibles at roughly one kilohertz. This stridulation works like a vibratome, aiding the mechanical cut, and simultaneously serves as a recruitment signal that brings nearby workers to the cutting site.

Watch closely and you'll see small ants riding on top of the leaf fragments being carried back to the nest. They are bodyguards. Phorid parasitoid flies, members of the genus Neodohrniphora and related genera, hover near foraging columns trying to lay eggs inside the workers' heads. The larvae develop inside the ant, eventually decapitating it. The hitchhiker ants are too small to be suitable hosts, and they snap at approaching flies.

One popular claim: leaf-cutter ants can carry 50 times their body weight. The measured maximum is closer to nine times body mass. Typical foraging loads are two to three times body mass, which is the range where energy efficiency is highest.

Species in Costa Rica

Leaf-cutter ants belong to two genera. Atta are the large ones, the species you notice: colonies of millions, nests the size of a living room, trails crossing your driveway. Acromyrmex are their smaller relatives, with smaller colonies and less conspicuous habits. Costa Rica has two Atta species and at least four Acromyrmex.

Atta cephalotes is the common one, found from the lowlands up to about 1,400 meters. Atta colombica is less common, found in humid lowland areas including the Osa Peninsula and parts of the northern lowlands. It's the species that maintains external refuse dumps on the soil surface rather than burying its waste underground. Among the Acromyrmex, A. coronatus dominates moist montane habitats, A. octospinosus is the dry-forest species found at Palo Verde, and A. volcanus and at least one additional species round out the list.

Leaf-cutters are not the only ants that farm fungi. Approximately 248 species in 20 genera practice fungus agriculture. Only about 55 of them, all in the genera Atta, Acromyrmex, and Amoimyrmex, actually cut fresh leaves. Costa Rica also has Cyphomyrmex and other "lesser attines," tiny ants that cultivate fungi on caterpillar frass, dead insects, and decaying plant material. They cause no crop damage. You have almost certainly never noticed them.

What Keeps Them in Check

Leaf-cutter ants harvest 12 to 17 percent of all foliage within their foraging range. A mature colony processes over 400 kilograms of dry leaf material per year. In parts of the Neotropics, they account for roughly a quarter of all herbivory. If you've watched them strip a young tree overnight, a fair question is why they haven't eaten every forest bare.

The answer is a suite of natural controls that, in intact forests, keep leaf-cutter populations in balance. When those controls break down, the ants become a different kind of animal.

Natural Population Controls

The most important bottleneck is founding mortality. Across species, roughly 2.5 percent of queens that land after a nuptial flight establish colonies that persist. The rest are killed by fungal infections, predation from birds, bats, and toads, or territorial aggression from established colonies nearby. Harold Fowler, an American ecologist working at Brazil's UNESP university, tracked A. bisphaerica queens in the pastures of São Paulo state and found that workers from mature nests killed 89 percent of new queens that landed near them. In open areas away from established colonies, predators still took 59 percent. His overall founding survival figure was 0.09 percent.

Established colonies suppress each other. Mature colonies patrol their borders, and nest spacing increases with colony size, a pattern consistent with competitive exclusion. The territorial aggression Fowler documented makes the ground around a mature colony a dead zone for founders. While a generation of colonies holds an area, far fewer newcomers survive. When the established generation dies off, the suppression lifts and a new cohort of founders breaks through.

Predators take a steady toll. The army ant Nomamyrmex esenbeckii is the only known predator capable of destroying a mature leaf-cutter colony. Powell and Clark documented 19 raids at Barro Colorado Island in Panama, including battles lasting more than 36 hours in which the army ants removed over 60,000 brood items from a single nest. Both species deploy their largest soldiers at the front. Some raids fail. Some overrun the colony completely.

Above ground, phorid parasitoid flies exert constant pressure. At least 13 host-specific species have been documented. They don't kill colonies outright. They suppress foraging activity, force shifts to nighttime operations, and reduce efficiency. At Corcovado National Park, the pressure is intense enough that Atta cephalotes shifts most of its foraging to nighttime. When phorids are present, defensive postures among ant workers nearly double, from 37 percent to 69 percent of observation periods.

Armadillos dig into nests and consume adults and brood. Tamanduas, the arboreal anteaters common across Costa Rica, eat roughly 9,000 ants per day, raiding nests briefly and moving on before the colony can mount a full defense. None of these predators destroy mature colonies on their own, but they impose persistent costs.

Inside the nest, a fungal parasite called Escovopsis wages a quieter war. Earlier research characterized Escovopsis as highly virulent. More recent work shows lower virulence than expected: in a 118-day experiment, all colonies with queens survived Escovopsis exposure. Queenless colonies died within 36 days. Escovopsis appears to be an opportunistic pathogen that finishes off weakened colonies rather than toppling healthy ones. By the time colonies are one to two years old, nearly 60 percent of them harbor the parasite. The ants defend against it with bacteria on their cuticles that produce antifungal compounds, and Escovopsis fights back with its own chemical weapons. The full arms race, which involves four interacting organisms, has been running for millions of years.

When the Controls Break Down

The most important finding in recent leaf-cutter ant ecology: these ants are disturbance specialists. Their populations increase dramatically when forests are fragmented or degraded.

Meyer et al. (2009) documented nest densities in fragments of Brazilian Atlantic Forest that were 11 times higher at forest edges than in interior habitat. Wirth et al. found roughly sixfold increases within 50 meters of edges. In intact Central Amazon forest, average Atta cephalotes density is 0.03 nests per hectare. In fragments, nests can directly affect six percent of total forest area.

Two mechanisms drive this. The first is bottom-up: fragmentation creates more edges, more light gaps, and more pioneer tree species, exactly the kind of vegetation leaf-cutters prefer. The second is top-down: fragmented forests lose armadillos, army ant populations decline, and phorid fly diversity drops.

Palmeirim et al. (2021), studying 34 forest islands in the Central Amazon, found that pioneer plant availability was the stronger driver. Bottom-up effects mattered more than predator release. Armadillo abundance actually increased alongside leaf-cutter density on smaller islands, suggesting the predators were not controlling the prey.

The Poison Problem

Leaf-cutter ants are essential to the forests they inhabit, but they do real damage to crops. Young trees and seedlings can be killed by a single complete defoliation; mature fruit trees survive one stripping but not three consecutive seasons of it. A citrus tree can lose every leaf in under 24 hours. Up to 30 percent of forestry plantation budgets across the Neotropics go to ant control. And many a keen gardener has despaired at a lovingly nurtured tree being stripped to bare wood overnight. This section covers the poisons commonly sold in Costa Rica to kill zompopas, the effectiveness of each, and their documented effects on the wider ecosystem.

What Costa Ricans Actually Buy

Walk into a Colono Agropecuario or any agroveterinaria in Costa Rica and ask for something to kill zompopas.

What Each One Does to the Ecosystem

Sulfluramid (Mirex-S, Zompokill): The Forever Chemical

Sulfluramid is a slow-acting stomach poison. Forager ants carry the bait granules into the colony, distribute them through the social food-sharing system called trophallaxis, and the colony dies over days to weeks. The active ingredient uncouples oxidative phosphorylation in mitochondria, shutting down cellular energy production.

The problem is what sulfluramid becomes. In aerobic soils, it degrades into PFOS (perfluorooctanesulfonic acid), a persistent organic pollutant listed under the Stockholm Convention since 2009. The soil half-life of sulfluramid itself is only about 14 days, but it does not disappear. It converts into PFOS, which does not break down. In one laboratory microcosm study, 85 percent of the sulfluramid applied had become PFOS after 91 days; field soil conversion rates are lower but the endpoint is the same.

PFOS is water-soluble. It leaches into surface water and groundwater. It bioaccumulates through food chains, with concentrations in organisms reaching more than 5,000 times those in the surrounding water. In Brazil, where 30 to 60 tonnes of sulfluramid are used annually in forestry, researchers detected PFOS in 76 percent of surface water samples near treated areas. Between 2009 and 2020, sulfluramid use in Brazil may have released approximately 357 tonnes of PFOS into the environment.

PFOS exposure in humans is linked to low birth weight, immune suppression, liver damage, thyroid dysfunction, and cancer. The US EPA canceled sulfluramid registration in 2008. The EU prohibits it. Brazil obtained a Stockholm Convention exemption for ant control with no phase-out deadline.

Costa Rica imports approximately 100 kilograms of sulfluramid active ingredient per year. No study has measured PFOS levels in Costa Rican soils or waterways near treated areas. Every bag of Mirex-S applied to a zompopa mound is a small, permanent addition to the PFOS load of the local watershed. The green "slightly hazardous" label on the package reflects acute mammalian toxicity. It says nothing about what the chemical becomes in the ground.

Fipronil: Everything Else Dies Too

Fipronil blocks GABA-gated chloride channels in insect nervous systems. It is effective against leaf-cutter ants; colonies stop cutting leaves within four days of bait application. The EU severely restricted fipronil for agricultural use in 2013 after it was linked to mass pollinator die-offs. In Costa Rica, SENASA has confirmed fipronil as the cause of multiple massive bee kills, including more than two million bees in Esparza and five million in Orotina in 2020 alone.

When you apply fipronil bait to a zompopa nest, the chemical does not stay in the nest. Fipronil has a soil half-life of about 125 days. Its breakdown products are worse: fipronil desulfinyl, formed by sunlight, is more toxic than the parent compound, more persistent, and up to ten times more active at mammalian nerve receptors, meaning it loses much of the selectivity between insects and vertebrates. Fipronil sulfide and sulfone, formed in the soil, are at least as toxic to invertebrates as fipronil itself. So the chemical and its metabolites remain active in the surrounding soil for months. Fipronil is marketed as safe for mammals because it binds selectively to insect nerve receptors. That selectivity degrades in the soil along with the chemical. A dog that digs around a treated nest weeks later is exposed to a compound that is now significantly toxic to mammals, capable of causing tremors, seizures, and hyperexcitation.

Fipronil does not distinguish between leaf-cutter ants and other ant species. Studies of fipronil-treated sites show declines in ant abundance, species richness, and diversity across the board. In one study, non-target ant communities that were treated continuously for three years showed little recovery even after treatment stopped. Termites are severely affected; field studies in Madagascar found that fipronil impacts on termites were "very severe and long-lived." Spiders, beetles, and other soil invertebrates shift in community composition, and these changes persist for six months or more.

The effects cascade up the food chain. Army ants that pass through the treated area die, and with them go the antbirds, antthrushes, and other species that depend on army ant swarms to flush their prey. Populations of two lizard species have been linked to declines in termite prey caused by fipronil. Armadillos and anteaters that dig into treated nests consume poisoned ants directly; the desulfinyl metabolite they absorb has far greater vertebrate toxicity than fipronil itself. Guillade and Folgarait (2014) showed that fipronil significantly decreases survivorship and longevity of phorid parasitoid flies, the organisms that naturally suppress leaf-cutter foraging. You kill the colony, but you also kill the biological system that was keeping zompopa populations in check.

Chlorpyrifos: The Neurotoxin

Chlorpyrifos is an organophosphate that permanently disables acetylcholinesterase, the enzyme that turns off nerve signals after they fire. Without it, nerves keep firing, muscles seize, and the organism dies. The same mechanism operates in insects, birds, fish, amphibians, and humans. There is no selectivity. Anything with a nervous system is vulnerable.

Applied to a zompopa nest, chlorpyrifos suppresses earthworm and termite communities in the surrounding soil. At recommended doses, litter decomposition decreases significantly for at least three months. In tropical forest with heavy clay soils, termites are the primary soil engineers: their galleries increase hydraulic conductivity up to 30-fold in soils with more than 50 percent clay (Cheik et al., 2019), building the channels that aerate the ground and let water infiltrate instead of running off. Kill them and the soil compacts, water sheets across the surface, and organic matter accumulates without decomposing. Chlorpyrifos is also volatile. It evaporates from the soil surface and drifts, extending its reach beyond the application site.

Chlorpyrifos is very toxic to birds. Documented wildlife incidents from lawn and soil insect treatments include deaths of robins, starlings, sparrows, geese, and bluebirds. It is highly toxic to fish and bioaccumulates in aquatic organisms; if the treated nest is anywhere near a stream or drainage, chlorpyrifos enters the water. In Costa Rica, Polidoro et al. (2016) documented fish kills in banana-growing watersheds linked to chlorpyrifos. Mena et al. (2021) found toxic effects in the native fish Astyanax aeneus. Amphibians appear particularly sensitive.

In humans, prenatal exposure has been linked to a 6.5-point reduction in psychomotor development scores at age three and structural brain abnormalities visible on MRI. Costa Rica imports over 250,000 kilograms of active ingredient annually. In May 2025, the Stockholm Convention's COP-12 listed chlorpyrifos for global elimination, with 22 specific use exemptions granted across signatory countries. Costa Rica obtained a five-year exemption specifically for pineapple. Chlorpyrifos requires a prescription under Decreto 34142 (2007). It is restricted and continues to be used widely.

Gasoline and Diesel in Nests

The do-it-yourself approach. No agricultural authority recommends this, but it remains common. People pour gasoline or diesel into nest entrances, sometimes igniting it.

A mature Atta nest can extend several meters underground with hundreds of chambers connected by tunnels. Pouring hydrocarbons into the entrance creates contamination pathways deep into the soil profile. The fuel kills soil bacteria, fungi, earthworms, and invertebrates indiscriminately. It seeps into groundwater. Hydrocarbons persist in soil for months to years depending on soil type and drainage. And it usually fails at its stated purpose: the queen chamber is meters below the surface, connected by passages that can be sealed off by worker ants. The colony survives. The soil does not recover quickly. Pumping propane into the nest and igniting it does not improve the odds. The tunnel system branches into hundreds of passages across several meters of depth. The gas disperses before it can fill the system, and the ants seal off tunnels faster than the pressure front moves. The result is a localized underground explosion that collapses a few chambers near the surface while the queen and most of the colony sit untouched below.

Sodium Octaborate (Omitox, Novex): The Boron Option

Sodium octaborate is a boron compound in the same chemical family as boric acid and borax. It works the same way: as a slow stomach poison. Foragers carry the granules back, share them through trophallaxis, and the colony dies over days. Boron disrupts metabolic processes in insects but has low acute toxicity for mammals. Unlike sulfluramid, it does not degrade into persistent organic pollutants. Unlike fipronil, it does not kill bees at nanogram concentrations. Unlike chlorpyrifos, it is not a neurotoxin. Boron is a naturally occurring soil element, and sodium octaborate does not bioaccumulate in food chains.

This makes it the least ecologically damaging commercial option on the list. Whether it actually kills the colony is another question: boron compounds work slowly, and large Atta colonies with millions of workers may not absorb a lethal dose before the bait degrades. But to the extent that it works, it does so without contaminating the watershed or poisoning the organisms around the nest.

The Feedback Loop and the Regulatory Gap

Chemical control kills natural enemies alongside the target ants. Phorid flies exposed to fipronil or chlorpyrifos die. Armadillos and anteaters that consume poisoned ants accumulate toxins. The non-target ant community shifts in composition. When surviving colonies rebound, they do so in an environment with fewer natural controls. The result is increased dependence on repeated pesticide application. The ants come back. The natural enemies do not. Meanwhile, PFOS from sulfluramid accumulates permanently in the soil, the water, and the food chain.

Costa Rica has 68 prohibited pesticides plus specific bans on methyl parathion, monocrotophos, carbofuran, endosulfan, and chlorothalonil. The Servicio Fitosanitario del Estado (SFE) under MAG maintains official prohibited and restricted lists. Enforcement is another matter. Studies document farmers using banned products. Costa Rica failed its national pesticide reduction objective, as reported at COP-16 in 2024. Technically, no agricultural pesticide in Costa Rica is sold without some form of restriction. In practice, Mirex-S sits on open shelves at agricultural supply stores across the country.

The products carry green toxicological bands meaning "slightly hazardous." This classification reflects acute mammalian toxicity on contact. It does not reflect PFOS contamination, aquatic toxicity, pollinator kills, or long-term bioaccumulation. The label tells you almost nothing about what the product does to the ecosystem around your house.

The industry's current candidate to replace sulfluramid is isocycloseram, a novel insecticide that acts on the GABA system through a different binding site than fipronil. Zanetti et al. (2024) found it comparable to sulfluramid in field trials against multiple Atta species. The problem: isocycloseram is itself a PFAS compound. Industry research found it probably reduces testicle size, lowers sperm count, and harms the liver in rats. In January 2026, public health groups sued the US EPA over its approval, arguing the agency failed to consider adverse effects on children and developing fetuses. Replacing one forever chemical with another is not progress.

What Actually Works Without Poison

Say you're a crusty permaculturist whose cacao seedlings are getting stripped to bare stems every night. You've read everything above. You wouldn't touch sulfluramid with a three-meter pole. Even the boron granules feel wrong. You want to kill the nest, or at least make it leave you alone, without poisoning everything around it. Here is what exists, what the evidence actually says, and where it runs out.

What Has Evidence Behind It

Canavalia ensiformis (jack bean, frijol burro): The most promising non-chemical method documented. Dr. Keith Andrews at Zamorano in Honduras found that placing 5 to 15 kilograms of freshly cut jack bean leaves on mound areas for three consecutive nights resulted in complete cessation of colony activity for periods ranging from four months to five years. The mechanism: the ants preferentially harvest the jack bean leaves, but compounds in them, long-chain saturated fatty acids, are fungicidal to their cultivar. The fungus garden dies. The practical requirement: 5 to 15 kilograms of fresh leaves means you need an established patch of jack bean, at least 5 to 15 square meters, planted two or three months before you need it. Jack bean grows in poor soils where almost nothing else will, but you have to plan ahead.

The caveat is significant. No controlled, replicated field trial has been published in peer-reviewed literature. ECHO, the organization that publicized Andrews's work, states there are "too many unanswered questions to recommend with much conviction." Practitioners in other countries report inconsistent results. One farmer in Belize reported that the ants refused to pick up the leaves. And Canavalia brasiliensis, the species common in Costa Rica, may differ in effectiveness from Canavalia ensiformis. This method is worth trying. It is not guaranteed.

Sesame leaves (Sesamum indicum, ajonjolí): Sesamin and sesamolin, lignans found in sesame plants, are toxic to the cultivar fungus. Sesamolin is five times more potent than sesamin. The field method works the same way as jack bean: cut sesame leaves in the evening and place them along foraging trails. The ants carry them in, and the lignans poison the fungus garden. In one trial, sesame leaf-based bait achieved 80 percent inhibition of colony activity at 90 days. Timing matters: this works best in the late dry season, when the colony is desperate for plant material and will harvest aggressively. During the rainy season, the ants are more selective and may take only a small amount before learning the material is harmful to their fungus.

Trap crops and sacrificial planting: Moringa and cranberry hibiscus are reported to be intensely preferred by foragers. Planting them near the garden diverts foraging pressure away from crops you care about. This approach accepts coexistence rather than elimination. It is probably the most realistic strategy for small farms and home gardens.

Shade: Leaf-cutters prefer sun-grown foliage with high water content and low concentrations of defensive compounds. Shade-grown crops suffer less damage. Agroforestry systems with canopy cover are inherently more resistant to leaf-cutter foraging than open-field monocultures.

Physical barriers: Metal collars and cotton-fiber wraps around the trunks of individual plants can prevent foragers from climbing. A band of soft cotton wrapped around the trunk is particularly effective because the hook-shaped claws on ant feet cannot grip the fibers. The cheapest version is a roll of absorbent cotton from any pharmacy (algodón absorbente), wound around the trunk a few times at about knee height. Strips of old t-shirts also work, as does the kapok-like fiber from silk floss tree pods (Ceiba speciosa, known in Costa Rica as ceiba barrigona). Sticky bands also work against ants but trap geckos, tree frogs, and other small animals that climb the trunk, so cotton or metal is preferable. Any barrier protects specific plants but does nothing to the colony.

Harvesting the refuse: Ant nest refuse is extraordinary fertilizer. Organic carbon, nitrogen, potassium, phosphorus, and magnesium concentrations are 20 to 50 times higher than in surrounding soils. Some Costa Rican farmers harvest this material. The Tico Times has profiled the approach under the headline "turning the leaf-cutter ant into an ally."

What Doesn't Work (Despite What You've Heard)

Citrus peels: D-limonene in orange peels disrupts pheromone trails. The effect lasts about 36 hours. Then it evaporates. The ants reroute. Even though orange peel contains roughly 2 to 5 percent limonene by weight, the concentration degrades rapidly once the peel is cut. This is a temporary inconvenience, not a control method.

Diatomaceous earth: Directly tested against leaf-cutter ants in eucalyptus plantations by Ferreira-Filho et al. (2015). Control efficacy was low. Must be reapplied after every rain.

Queen removal: The queen chamber is five to eight meters underground. Mature colonies may have more than one queen. Workers show improved stress tolerance after queen loss. Excavating a mature colony requires heavy equipment. This is only viable for colonies less than a few months old, when the chamber is shallow.

Neem: Lab evidence shows toxicity to ants. Field evidence shows that leaf-cutters defoliate young neem trees. Only mature neem with high azadirachtin concentrations are avoided. Planting neem as a companion crop is counterproductive unless the trees are already well established.

Mint, bay leaves, cayenne pepper, vinegar: Costa Rican fumigation companies recommend these as "natural" alternatives. They are temporary repellents. They redirect ants. They do not suppress colonies.

The Honest Answer

You cannot eliminate a mature leaf-cutter ant colony without poison or extraordinary effort. The queen is meters underground, the colony has millions of workers, and it has been maintaining itself for years.

The practical question for most homeowners and small farmers is whether you can live with them. For most situations, the answer is managing the relationship rather than winning a war. That means understanding what they target and why.

Why They Strip One Plant and Ignore the Next

Zompopas don't eat leaves. Their fungus does. So a plant's "resistance" has almost nothing to do with whether the ants find it palatable and everything to do with whether the fungus can digest it. When foragers bring back leaves that poison or inhibit the fungus garden, the colony learns to avoid that species. The gardener ants inside the nest detect the damage first and stop processing the harmful material. Chemical cues from the rejected substrate then spread through the colony. Contrary to what you might expect, roughly a third of foragers visit the waste chamber, spending an average of 14 minutes there, long enough to pick up the scent of what went wrong. Foragers exposed to dump waste from a particular plant reduce their acceptance of that plant by more than half. The avoidance can last for months. Each colony builds its own blacklist through experience, which is why a colony at your neighbor's house might ignore a plant that yours destroys.

A tempting idea follows from this: blend the leaves of a plant you want to protect and pour them into the nest through an entry hole, tricking the colony into associating that scent with trouble. On its own, this probably would not work. The colony needs two signals together: the scent of the plant and evidence that the fungus was damaged by it. Without actual fungal damage, the chemical cue for avoidance never forms. This is what makes jack bean and sesame effective as biological controls: the ants carry the material into the garden, it genuinely damages the fungus, and only then does the colony learn to avoid it. A more sophisticated version of the idea would be to blend citrus leaves (or whatever you want to protect) with soursop or another species whose compounds damage the cultivar fungus, and introduce the mixture into the nest through entry holes. If the colony's foragers encounter waste that smells like citrus alongside damaged fungus, they might learn to avoid citrus. A simpler approach: chop soursop and citrus leaves together finely enough that foragers cannot separate them, and place the mixture along active trails. If the ants carry both into the garden and the soursop damages the fungus while citrus scent is present, the association might form naturally. None of this has been tested. But the mechanism is understood well enough that a motivated gardener could experiment.

Beyond fungus chemistry, physical leaf traits matter. Ants strongly prefer young, soft leaves with thin cuticles. Tough, mature leaves require more energy to cut and carry. Dense trichomes (the tiny hairs on leaf surfaces) make it hard for ant mandibles to grip. Thick waxy coatings make leaves slippery. This is why seedlings of almost any species are vulnerable but mature trees of the same species may be left alone, and why new growth flushes get hit hardest.

Variegation often correlates with altered leaf chemistry. If zompopas strip your green ginger but leave the variegated one alone, the likely explanation is that the variegated cultivar produces different secondary metabolites that the fungus can't handle. Same genus, different chemistry, different result.

What They Destroy

Citrus. Young citrus trees are among the most devastating targets. A large colony can strip an entire young lime or mandarin overnight. Citrus leaves are soft, thin, and nutritionally ideal for the fungus. Even mature citrus trees get hit, though they can usually survive the defoliation. If one of your citrus trees gets stripped while another is untouched, the explanation is usually leaf age or epiphyll growth: older citrus leaves develop a film of algae and lichen on the surface that repels ants. The clean, fresh leaves on new growth have no such protection.

Mango. Herbivory rates above 90 percent in studies of Atta cephalotes. Mango is one of the most consistently preferred commercial crops across the range of leaf-cutter ants, along with citrus, cacao, and coffee. Mango leaves are relatively soft and apparently lack the terpenoid profiles that deter the fungus.

Hibiscus and moringa. Tropical gardeners routinely describe both as irresistible. Cranberry hibiscus (Hibiscus acetosella) and moringa (Moringa oleifera) are so consistently targeted that they are more useful as trap crops than as plants you expect to keep.

Croton (Codiaeum variegatum). Soft, thin leaves with high water content and minimal physical defenses. The bright pigments that make croton attractive to gardeners (anthocyanins, carotenoids) are not the same compounds that deter the fungus. The leaves offer almost no resistance to cutting, and the fungus apparently thrives on them.

Roses. Cut flowers and buds as well as leaves. The ants harvest both the foliage and the petals. Roses in tropical gardens near active colonies are a losing proposition without barriers.

Cassava / yuca. One subspecies of Atta sexdens is named the "cassava leaf-cutting ant" for a reason. Cassava's large, soft leaves are heavily targeted across Central and South America.

Cacao. Consistently listed alongside citrus as a major crop target across Central America. Cacao plantations in the Caribbean lowlands rely on Azteca ants that live in the canopy and aggressively patrol the leaves, attacking leaf-cutter foragers on contact. Where these guardian ants are absent, damage is severe.

Macadamia. Non-native, Australian origin, no evolutionary exposure to Atta. Attack begins on the youngest leaves and can kill seedlings outright. Macadamia orchards in the Central Valley are a known target.

Ylang ylang (Cananga odorata). The soft, aromatic leaves are heavily targeted. Large mature trees can be killed through repeated defoliation over successive seasons.

Seedlings of almost anything. Young plants have thin cuticles, few trichomes, low concentrations of defensive compounds, and cannot survive complete defoliation. A species whose mature tree is completely ignored can lose every seedling to a nearby colony. This is the single most important thing to understand about zompopa damage: it is disproportionately concentrated on young plants.

What They Tend to Leave Alone

One broad pattern is worth knowing. Flowering plants fall into two groups: dicots (most trees, shrubs, and broadleaf plants, with branching leaf veins) and monocots (grasses, palms, bananas, heliconias, orchids, with parallel veins). Leaf-cutters overwhelmingly prefer dicots. In one experiment, palm leaves placed on foraging trails lost only 3 percent of their area versus 61 percent for dicot leaves. Monocot leaves are tougher, waxier, and chemically less palatable to the fungus. Most of the plants on this list are dicots that happen to have strong chemical defenses; the monocots are called out below.

Mature neem (Azadirachta indica). Young neem trees get defoliated. Mature neem trees are almost never touched. The difference is the concentration of azadirachtin, which increases as the tree ages. Azadirachtin disrupts the fungus garden and also repels foragers directly. This age-dependent resistance is one of the clearest demonstrations that a plant's chemistry, not its identity, is what matters.

Soursop / guanábana (Annona muricata). The Annonaceae family produces acetogenins, alkaloids, and terpenoids that are toxic to a wide range of organisms, including the cultivar fungus. Soursop leaves are used in traditional medicine as insecticides across the tropics. The ants learn quickly that this material kills their garden.

Guava and cas (Psidium spp.). In laboratory bioassays, leaf-cutter workers actively rejected guava leaf discs. Guava leaves contain a dense cocktail of tannins, flavonoids, and phenolic acids, including gallic acid and epigallocatechin gallate, that are antifungal to the cultivar. Cas (Psidium friedrichsthalianum), the sour-fruited relative used for refrescos, is even more chemically aggressive and likely at least as resistant. If you want a fruit tree near an active colony, guava is one of the safer bets.

Pepper (Piper spp.). Costa Rica has dozens of native Piper species. Piperine and related alkaloids inhibit the growth of Leucoagaricus gongylophorus. These are among the plants that researchers describe as "avoided altogether" rather than sampled and rejected.

Heliconia and banana (monocots). Heliconia and banana combine multiple defenses beyond the general monocot advantages. Their fracture toughness is among the highest measured in tropical forests, nearly double the dicot average, achieved through dense fiber strands along the vascular bundles. Both families are loaded with calcium oxalate raphide crystals, microscopic needles that synergize with enzymes to cause high mortality in chewing insects. Both contain tannins and phenolics at concentrations that reduce the cultivar fungus's growth by more than half. Banana leaves add another layer: a thick epicuticular wax that physically blocks fungal colonization of the leaf substrate. The ants that cut grass and other monocots must physically strip this wax from each fragment and discard it as waste before processing.

Coconut and pejibaye (monocots). Palms share the same leaf toughness advantage as banana and heliconia. Coconut does not appear on any leaf-cutter pest list. Pejibaye (Bactris gasipaes) is the same: its documented pests are weevils, not ants.

Mature avocado. Thick, waxy cuticles on mature foliage make the leaves physically difficult to cut and carry. Avocado seedlings and new growth flushes are still vulnerable. The ants also feed from stem wounds on avocado trees, causing secondary infections, even when they leave the foliage alone.

Sesame (Sesamum indicum, ajonjolí). The lignans sesamin and sesamolin are directly fungicidal to the cultivar. The ants will harvest sesame leaves initially, which is exactly why sesame works as a biological control agent: they carry the fungicidal material into their own garden before learning to avoid it.

Coffee (shade-grown). Coffee is a paradox. In laboratory leaf disc tests, workers rejected coffee leaves outright. Caffeine inhibits the cultivar fungus directly. Yet in unshaded monoculture plantations near Turrialba, A. cephalotes is a documented pest of coffee. The resolution is context: in shade-grown, diversified systems, colony density drops and the ants have better options. Shade-grown coffee in agroforestry is largely left alone. Unshaded monoculture coffee, where the colony has fewer alternatives, gets hit.

Plants with dense trichomes. Hairy-leaved species are physically harder for ant mandibles to grip and cut. Tomato plants are a familiar example: the sticky glandular trichomes that make them smell like tomato also gum up ant mandibles. Many Melastomataceae, one of the most common plant families in Costa Rican forests and gardens, have densely fuzzy leaves. Tibouchina (nazareno), the ornamental with purple flowers planted across the Central Valley, is one. So are Miconia and Conostegia, common understory trees in wet forests whose leaves feel like velvet. Geraniums and coleus, both common garden plants, are hairy enough to deter cutting.

Cacti, agave, and dragonfruit (pitahaya). The thick, waxy cuticle that defines succulents is nearly impossible for the ants to grip and cut cleanly. But the real problem is what's inside. Succulent tissue is loaded with mucilage and organic acids, not the thin, moist leaf substrate the cultivar fungus needs. Many cacti and agaves also contain dense calcium oxalate crystals. Agave is a monocot; cacti and pitahaya are dicots whose physical properties happen to mimic monocot resistance. If you grow pitahaya in the lowlands, leaf-cutters are not a concern.

Plants also fight back after being attacked. When leaf-cutters repeatedly harvest from the same individual, the plant's jasmonate defense pathway activates within an hour. Over the following days, the plant increases production of volatile terpenoids that repel foragers. By day five, ants begin avoiding the plant, and the avoidance can last 9 to 18 weeks. Some species respond by producing new leaves with 25 to 100 percent more trichomes. Others increase terpene concentrations in regrown foliage. The induced defense is measurable: there is a strong negative correlation between a plant's volatile emissions and ant preference. The catch is that it requires repeated defoliation to trigger. A single night's harvest may not be enough.

If your green ginger gets stripped but the variegated cultivar next to it is untouched, the likely explanation is that variegation correlates with a different secondary metabolite profile. Same genus, different chemistry. The fungus can handle one and rejects the other. The colony learned this distinction and now routes its foragers accordingly.

Given all these defenses, you might expect that after 27 million years of coexistence, some tropical tree would have evolved complete resistance. None has. The core reason is that the selective pressure is too diffuse. A single colony harvests from dozens or hundreds of plant species. No individual species bears enough of the burden to drive resistance all the way to fixation. A tree that invests heavily in one chemical defense pays a metabolic cost for it, but the colony just shifts its foraging to the neighbor. Meanwhile, the fungus keeps evolving counter-measures of its own. When plants produce phenolic compounds to poison the garden, the cultivar produces an enzyme called laccase (LgLcc1) that breaks those phenolics down. The ants help by depositing fecal droplets containing the enzyme onto fresh leaf fragments as they chew them into substrate. This is a general-purpose counter to the most common class of plant defense. Species like Pentaclethra macroloba, which dominates some Costa Rican lowland forests, use saponins instead and are consistently avoided. But the ants forage elsewhere, and the tree pays for its defense whether or not it is under attack. The arms race continues because neither side can win it outright.

Living with the Colony: A Practical Strategy

Protect seedlings. This is the highest priority. Young plants have soft cuticles, few trichomes, and low concentrations of defensive compounds. They are the most vulnerable and the most valuable, because you've invested time and money in them and they cannot survive complete defoliation. Wrap trunks with absorbent cotton from the pharmacy at about knee height. For small plants without a trunk to wrap, a ring of cotton around the base of the stem, kept off the soil, works. Check weekly for bridges created by debris or dead leaves.

Use trap crops. Plant moringa, cranberry hibiscus, or another known favorite away from your garden, between the nest and the plants you care about. The ants will preferentially harvest the trap crop. This does not stop them, but it redirects foraging pressure. Replace the trap crop as needed; it is a sacrifice you are choosing to make.

Plant resistant species around what you value. Mature neem, soursop, cas, and pepper plants all contain compounds that the cultivar fungus cannot handle. Use them as buffer plantings around fruit trees or garden beds. The ants will learn to avoid them and may route their trails elsewhere.

Recruit guardian ants. In cacao plantations across the Caribbean lowlands, arboreal Azteca ants patrol the canopy and attack leaf-cutter foragers on contact. Cecropia trees with Azteca colonies are attacked by leaf-cutters significantly less often than unoccupied trees. You do not need to introduce these ants. You need to plant their host trees and let them colonize naturally. Azteca queens seek out myrmecophytic saplings after nuptial flights and establish colonies inside hollow stems and swollen branch nodes. The three most useful hosts in Costa Rica are guarumo (Cecropia spp.), which is a fast-growing pioneer with hollow internodes; laurel (Cordia alliodora), a timber tree already common in Talamanca cacao agroforestry whose swollen branch nodes house the specialist Azteca pittieri; and guaba (Inga spp.), a nitrogen-fixing shade tree that hosts Azteca sericeasur. Maintain at least 30 to 40 percent shade canopy over your planting area. Do not spray insecticides near canopy ants. Preserve the understory vegetation that connects shade trees to your crop trees: studies in Chiapas found that cutting these vegetation bridges measurably reduced ant-mediated pest suppression. The approach is conservation biocontrol, designing your planting so that the ants that already exist in the landscape have a reason to move in.

Attack the fungus, not the ants. If jack bean or sesame is available, this is the most promising non-chemical approach. Cut leaves in the evening and place them directly along active foraging trails. The ants do the work of carrying the fungicidal material into their own garden. Time it for the late dry season when the colony is desperate for plant material and will harvest aggressively.

Accept some leaf loss on mature trees. A healthy mature tree can tolerate partial defoliation. The ants rarely kill an established tree; they take what they need and move on. The tree refoliates. This is easier to accept once you understand that the colony is also aerating your soil, cycling nutrients, and dispersing seeds.

Harvest the refuse, if you can find it. Spent fungus garden substrate is rich in nitrogen, phosphorus, and organic matter. It is excellent compost. A. colombica dumps this material on the surface in obvious dark refuse piles, and you can shovel it straight into your garden beds. A. cephalotes, the common species, buries its waste underground, so you cannot easily harvest it. The pale soil mounds around its entry holes are excavated subsoil, not composted waste, and dropped leaf fragments near the nest are raw plant material that has not been through the fungus.

Stop killing what kills them. Leaf-cutter colonies already have natural enemies. Phorid parasitoid flies suppress daytime foraging. Entomopathogenic fungi (Beauveria, Metarhizium, Trichoderma) attack the ants and their fungus garden directly; UCR has two active research projects trying to develop these into field-ready biocontrol. Army ants (Nomamyrmex esenbeckii) are the only predator documented to destroy a mature colony outright. Armadillos excavate young and medium-sized nests. Every one of these organisms is harmed or killed by the same broad-spectrum pesticides sold to control zompopas. Fipronil and chlorpyrifos both reduce phorid fly survival. Sulfluramid contaminates the soil that army ants forage through. The most effective biological control you have is the one that already lives in your landscape. The practical advice is mostly negative: stop doing things that remove it. Keep dogs contained at night so armadillos can work. Do not clear undergrowth to bare soil. Do not spray near canopy ants. Maintain habitat complexity around the margins of your property. The research on entomopathogenic fungi is promising but not yet commercially available. Everything else is already there, if you let it be.

What They Give Back

The conversation about zompopas usually focuses on what they take. Understanding their ecological role requires looking at mechanisms, and the mechanisms are more complex than a simple ledger of costs and benefits. The same colony that defoliates a fruit tree also transforms the soil beneath it, accelerates nutrient cycling across hundreds of square meters, and maintains biological diversity underground that the forest depends on.

The nutrient enrichment described earlier (20 to 50 times more nitrogen, phosphorus, and other elements in nest refuse compared to surrounding soil) is real. At La Selva, refuse holds 2.5 percent nitrogen by mass, five times higher than forest soil, and 877 micrograms of ammonium per gram of dry material, 500 times higher. But this enrichment picture is incomplete. In 2013, Meyer and colleagues measured soil nutrients along transects extending 24 meters from Atta cephalotes nests in Atlantic forest. They found the opposite of enrichment at the surface: topsoil nutrient availability decreased toward the nest. On nest surfaces, leaf litter mass averaged 150 grams per square meter. At 24 meters' distance, in undisturbed forest, it averaged 1,300 grams. The ants had stripped the ground bare. The enrichment exists underground, in refuse chambers, while the surface above is depleted of the litter that normally feeds the topsoil. The nest creates a nutrient inversion: impoverished on top, concentrated below.

Whether neighboring plants actually access those buried nutrients was an open question until 2007, when Sternberg and colleagues traced it directly. They applied nitrogen-15-labeled potassium nitrate to leaves being harvested by the ants. Three to four months later, plants within 11 meters of the nest showed nitrogen-15 enrichment of 78.8 per mil, compared to 1.7 per mil in controls, a 46-fold difference. The pathway is concrete: harvested leaves enter the colony, pass through the fungus garden, exit as refuse, decompose in underground chambers, and re-enter the soil as plant-available nitrogen and minerals. The timeline, three to four months from harvest to plant uptake, is far shorter than the one to two years required for equivalent decomposition of leaf litter on the forest floor.

The refuse decomposes roughly twice as fast as fresh leaf litter. The fungus has already broken down the hemicelluloses and starch during its passage through the garden; what remains has a reduced lignin-to-cellulose ratio and higher nitrogen content, so soil microorganisms process it more rapidly. The colony operates what researchers describe as a biphasic bioreactor: fresh leaf fragments enter the peripheral zone of the fungus garden, where hemicellulase and amylase activity breaks down accessible carbohydrates within hours. The fungus converts these to sugar alcohols, primarily mannitol, that feed the ants through specialized swollen hyphal tips called gongylidia. Material transits the garden over several weeks, with progressively less accessible carbon remaining. What the fungus cannot use is discarded to refuse chambers, where it enters the soil food web already partially digested. The colony accelerates the forest's decomposition cycle by intercepting leaf material, processing it through a biological intermediary, and returning it to the soil in a form that microorganisms can mineralize faster.

Something unexpected happens in the root zone. At La Selva Biological Station, researchers found that the standing crop of roots and mycorrhizal fungi in active nest soil is about the same as in surrounding forest. But the turnover rate is dramatically different. Fine roots in nest soil survive an average of 27.5 days before dying. In non-nest soil, they survive 152 days. Mycorrhizal hyphae survive about 10 days in nests versus 32 days in controls. Yet because roots and hyphae grow faster in the nutrient-rich nest environment, the standing biomass stays constant: old structures die and new ones replace them continuously. Root production in nest soil runs at 31.1 kilograms of carbon per cubic meter per year, compared to 5.6 in non-nest soil. Mycorrhizal hyphal production: 47.3 versus 14.8. The nest is a zone of accelerated biological cycling where roots and fungi grow, die, and decompose at three to six times the background rate, feeding carbon and nutrients into the soil continuously.

What the Colony Builds

Above ground, the colony reshapes the forest canopy. Through excavation, the ants undermine root systems of trees directly overhead, and they clear understory vegetation from the nest surface. Meyer et al. (2011) measured a 40 percent increase in canopy openness above nests compared to controls. These openings function differently from treefall gaps. A tree falls once and the gap begins closing as surrounding vegetation fills the space. A leaf-cutter gap is persistent: the colony suppresses regrowth and continues undermining roots for as long as it lives, which can be 15 to 20 years. Soil temperatures on nests run two degrees warmer than at 24 meters' distance, and the daily temperature range is twice as wide. Light-demanding species recruit into these openings, and the colony's selective herbivory determines which ones survive.

In Costa Rica, Rockwood (1976) found that only 22 percent of available plant species were palatable to Atta cephalotes. Pioneer species, with higher water and nutrient content and lower concentrations of secondary compounds, are harvested at roughly three times the rate of shade-tolerant species. The selectivity is mediated by the fungus garden itself. When foragers bring back leaves containing compounds toxic to the cultivar Leucoagaricus gongylophorus, the fungus rejects them. Workers that visit the waste dump learn to associate the offending plant's odor with failed substrate. This olfactory memory persists at least seven days. The colony learns, through distributed chemical feedback, which plants to avoid. Hymenaea courbaril (guapinol) resists attack because it produces caryophyllene epoxide, a sesquiterpene that is directly antifungal against the cultivar. The defense works because it targets the fungus, not the ant. Over time, selective herbivory at this scale gives chemically defended species a competitive advantage in the colony's foraging range, subtly reshaping the composition of the forest around each nest.

The nest interior is a habitat in its own right. More than 80 arthropod species have been documented inside leaf-cutter ant nests, including beetles, flies, spiders, pseudoscorpions, mites, moths, and annelid worms. Some are obligate inhabitants with no other home. The genus Attaphila includes nine described species of cockroach that live exclusively in leaf-cutter nests, feed on the fungal garden, and disperse to new colonies by climbing onto queen ants during nuptial flights. They evade detection by acquiring cuticular hydrocarbon profiles that match the host colony's chemical signature, rendering them invisible to the ants' nestmate recognition system. Frogs and snakes use the chambers as nurseries. The nest is climate-controlled: fungus garden chambers hold steady between 22 and 26 degrees Celsius, buffered about five degrees above average soil temperature, with humidity maintained at 85 to 95 percent. The ants actively regulate these conditions, opening or closing ventilation turrets in response to temperature and CO₂ fluctuations.

Below the visible fauna, something was discovered at La Selva in 2023 that reframes the colony's relationship with the forest. Allen and colleagues sequenced DNA from nest soils and found them to be diversity hotspots for mycorrhizal fungi. In the surrounding forest, the mycorrhizal community is overwhelmingly dominated by arbuscular mycorrhizal types. In nest soils, the researchers identified 226 distinct arbuscular mycorrhizal lineages, 159 ectomycorrhizal lineages (including Scleroderma sinnamariense, Tomentella, Cenococcum, and Rhizopogon), at least 135 orchid mycorrhizal lineages, and ericoid mycorrhizal Sebacinaceae. Scleroderma sinnamariense sequences appeared in nest samples but not in control soils. The researchers directly observed ants carrying sporocarp material.

Three mechanisms appear to concentrate this diversity. The ants physically transport fungal inocula. The abundant nest fauna, including springtails, symphylans, and midge larvae, consume and redistribute spores. And the tunnel-and-chamber architecture provides physical pathways where roots and fungi encounter each other in configurations that would not occur in undisturbed soil. When a colony dies or is abandoned, roots grow rapidly through the old tunnels into the enriched chambers. But mycorrhizal colonization lags, because roots grow downward while many mycorrhizal fungi do not, creating a temporal window in which novel root-fungus associations form. The obligate ectomycorrhizal plant Gnetum leyboldii was documented establishing within abandoned nest footprints at La Selva, forming ectomycorrhizae with S. sinnamariense on new roots. The nest does not merely house fungi. It creates conditions for mycorrhizal partnerships that would not otherwise exist in that forest.

At the landscape scale, these effects accumulate. At La Selva, active nests average 67 square meters and collectively occupy about 1.2 percent of the land surface. But that snapshot understates their influence: colonies die and new ones establish elsewhere, so over a decade a much larger fraction of the forest floor has been underneath an active colony and received its soil modifications, nutrient enrichment, and mycorrhizal inoculation. During the 2015-2016 El Niño drought at La Selva, nest soils maintained higher moisture and supported microbial communities with greater resilience than surrounding soils. Fewer nest-associated fungi declined in abundance post-drought. The colonies had, without intention, created drought refugia in the soil.

The forest would still exist without them. It would function differently: nutrient turnover would slow, the patchwork of fertility hotspots that drives plant diversity would flatten, canopy gap dynamics would lose their most persistent driver, more than 80 nest-dependent arthropod species would lose their habitat, mycorrhizal diversity in the soil would decline, and decomposition would decelerate. The case for their ecological importance does not require exaggeration. The measured evidence is enough.

The Name and the Season

In Costa Rica, the word is "zompopa," feminine, reflecting that every worker ant in the colony is female. Elsewhere in Central America, "zompopo" is used. The etymology is debated: some sources attribute it to Nahuatl, others propose Mayan origins. The Real Academia Española (RAE), the institution that governs the Spanish language, lists the word as a large leaf-cutting ant of Central America; in Honduras it also means a person who wanders aimlessly. In Costa Rican slang, calling someone a zompopo means they are dim-witted or thick-headed, though this usage is not in the RAE dictionary. In Costa Rican usage, calling someone a zompopo also implies they have a conspicuously large head.

Every rainy season, mature colonies release winged queens and males for nuptial flights. The males die within hours. Their bodies litter streets and sidewalks wherever zompopas live, but the spectacle is most visible in the densely populated Central Valley, where residents from Alajuela to Cartago wake up to find them covering the ground. La Nación periodically runs articles explaining the phenomenon to puzzled newcomers.

In Guatemala, these winged queens are "zompopos de mayo," though shifting rainfall now pushes their emergence into June. A single mature colony can release thousands of alates in one night, and across a region, the swarms number in the hundreds of thousands. They are collected at dawn, sometimes by the bucketful, roasted on clay comals, and eaten with salt and lime. The tradition is pre-Columbian, documented in the Popol Vuh, where ants play a pivotal role in the discovery of maize, and in the 1722 Historia Natural del Reino de Guatemala. In Colombia, the queens of a related species (Atta laevigata) are "hormigas culonas," sold in markets in Santander for roughly 50,000 pesos per pound and exported internationally. In Oaxaca, the Mexican species Atta mexicana is the chicatana, a gourmet ingredient described as smoky with notes of cacao and umami, selling for 500 to 1,000 pesos per kilo.

Costa Rica does not eat its zompopas. La Nación describes insect eating as "todo un reto" (a real challenge) for the Tico palate. The country has 48 documented edible insect species, but the emerging industry focuses on crickets and mealworms.

A Science Hub

Costa Rica is a global center for leaf-cutter ant research and has been for decades. La Selva Biological Station has hosted studies on leaf-cutter ecology since the 1970s. In 2016, a collaboration between UCR, Harvard, and the University of Wisconsin-Madison discovered selvamicin, an antifungal compound produced by bacteria on the ants' skin, from colonies at La Selva. It inhibits Candida albicans, a human pathogen, with reportedly lower toxicity than existing antifungals. UCR's Adrian Pinto-Tomás runs a public education site, www.zompopas.com, launched in 2012. The university's research covers everything from antibiotic discovery to biofuel applications to biological pest control.

At Guayabo National Monument, the country's most important pre-Columbian archaeological site near Turrialba, leaf-cutter ant activity has become a conservation problem of a different kind: the ants promote fungal growth that damages the basaltic-andesite rock of the ancient structures. Living with zompopas means contending with them everywhere, including the places we most want to preserve.