How Ants Farm Fungi: The Strange Agricultural Mutualism of Leafcutters
A leafcutter ant colony of Atta or Acromyrmex can contain 8 million workers organized into a strict caste system with millimeter-scale minor workers, centimeter-scale soldiers, and a single queen that lives for 15-20 years and produces every worker in the colony. The colony's underground chambers contain a fungus garden that the ants cultivate, harvest, and depend on for nearly all their nutrition. The ants do not eat the leaves they cut. They use the leaves as substrate to grow the fungus, and the fungus is what they eat. The system has been running continuously since approximately 60 million years ago, when the agricultural mutualism between the ancestral ants and the ancestral fungus first stabilized. Human agriculture is 12,000 years old. The ants beat us to it by a factor of 5000.
The leafcutter system is one of three known cases of evolved agriculture in non-human animals. The other two are the fungus-growing termites of the African genus Macrotermitinae, which independently evolved a similar mutualism around 30 million years ago, and the ambrosia beetles, several lineages of which farm fungus in galleries they bore into wood. Each of the three cases evolved the same general solution to the same problem: how to obtain nutrition from plant material that is mostly cellulose, which animals cannot digest. The answer in all three cases is to recruit a fungus that can digest cellulose, cultivate it on the plant material, and eat the fungus.
The four-way mutualism
The leafcutter ant agricultural system is not a two-way mutualism between the ant and the fungus. It is a four-way mutualism involving the ant, the fungus, an antibiotic-producing bacterium, and (depending on how you count) the leaf substrate the ants supply. The full architecture is one of the most thoroughly co-evolved cooperative systems known.
The fungus is one of several species in the family Lepiotaceae, most commonly Leucoagaricus gongylophorus. The fungus produces specialized structures called gongylidia, which are nutrient-dense swellings that the ants harvest and feed to the larvae. The gongylidia are the equivalent of the ear of corn or the head of wheat: they are the part of the crop that is eaten, and they exist primarily because the ants have selected for them over millions of years. A fungus removed from the ants and cultivated in a laboratory eventually stops producing gongylidia, which is the fungal equivalent of an agricultural cultivar reverting to wildtype when the breeders stop selecting for the trait.
The fungus has been propagated clonally by the ants for at least the last 30 million years. New colonies are founded by a winged virgin queen who takes a piece of the parental colony's fungus garden with her in a specialized buccal pouch before flying out to mate. After mating, she founds a new colony and inoculates her first fungus garden from the carried piece. The continuity of clonal propagation across geologic time means that the cultivated fungus has accumulated genetic differences from any wild relative substantial enough that Leucoagaricus gongylophorus is now considered a separate species that exists only in association with the ants. There is no wild population of the cultivated fungus, only the populations being maintained in millions of ant colonies across Central and South America.
The pathogen problem
Like any monoculture, the leafcutter fungus garden is vulnerable to disease. The principal pathogen is a parasitic mold called Escovopsis, which appears to be specific to leafcutter fungus gardens and which can destroy a colony's entire food supply if it gets out of control. Escovopsis has been co-evolving with leafcutter colonies for as long as the system has existed, and it is functionally the equivalent of wheat rust or potato blight in human agriculture: a host-specific parasite that periodically devastates the crop.
The ants' defense against Escovopsis is a third partner in the mutualism: bacteria of the genus Pseudonocardia, which the ants culture on specialized structures on their bodies called crypts. The bacteria produce antibiotics that suppress Escovopsis growth, and the ants groom the bacterial cultures onto their fungus gardens as a kind of biological pesticide application. The bacteria-ant relationship is itself a mutualism: the ants provide the bacteria with food (via specialized glandular secretions) and a controlled habitat, and the bacteria provide the ants with the chemical weapons that protect their crop.
The antibiotic production has been running for tens of millions of years without the Escovopsis populations evolving complete resistance, which is a fact that has interested antibiotic researchers since it was first appreciated in the early 2000s. Cameron Currie's work at the University of Wisconsin, particularly Currie et al's 2003 Science paper, demonstrated that the bacteria produce a diverse cocktail of antibiotics rather than a single compound, and that the cocktail composition varies across colonies and across geographical regions. The diversity of antibiotic production keeps the parasite from evolving resistance fast enough to overwhelm the system, which is the same general principle behind modern antibiotic stewardship programs that try to slow resistance evolution by combining drugs and rotating treatments.
The bacterial antibiotics have been investigated as potential sources of new antibiotics for human medicine, with mixed success. A few compounds identified from Pseudonocardia have shown activity against human pathogens in vitro, but none have yet made it to clinical use. The bacteria are difficult to culture outside the ant context, and the antibiotics are part of a complex chemical ecology rather than single-compound treatments, which has slowed the pharmaceutical development.
The agricultural sophistication
The agricultural behavior of leafcutter ants is more sophisticated than the schoolroom version suggests. The ants do not just cut leaves and pile them on the fungus. They process the leaves through a multi-step preparation: the leaves are cut, transported, further trimmed inside the colony, chewed into a paste, mixed with the ants' own fecal droplets (which contain enzymes that help break down the plant material), and then incorporated into the fungus garden in carefully arranged layers.
The ants select leaves for the garden with discrimination. They reject leaves that contain compounds toxic to the fungus, including many plants that defend themselves with secondary metabolites. The selection pressure has shaped plant defenses in the Neotropics: many tropical plants have evolved chemistry specifically targeted at suppressing leafcutter fungus growth, which is the agricultural-equivalent of crop pests evolving herbicide resistance. The ants in turn have evolved to detect and avoid these defended plants, and the fungus has evolved partial resistance to some of the compounds.
The ants also weed their gardens. Workers patrol the garden, remove damaged or contaminated fungus, and dispose of waste in specialized chambers far from the active garden. The waste management is essential: a leafcutter colony processes hundreds of kilograms of leaves per year and produces a corresponding volume of spent fungus material that has to be removed from the active garden to prevent contamination. The waste chambers are oriented downwind of the active garden in the colony's natural ventilation system, which Scott Turner's work on termite mound architecture (which we covered in our piece on termite mounds) has shown is itself a sophisticated engineering achievement.
The ants engage in horticultural pruning. They harvest the gongylidia at the appropriate stage of maturity, leaving the rest of the fungus to continue growing. They reseed depleted areas of the garden with fresh fungus from active areas. They selectively prune fungal growth that is producing fruiting bodies (which are reproductive structures of no nutritional value to the ants) and direct fungal growth toward gongylidia production. The cumulative effect is a managed crop with managed harvests and managed disease control, which is recognizably agricultural in a way that goes beyond mere food storage.
The independent reinvention in termites
The macrotermitine termites of Africa evolved a similar agricultural system independently, beginning approximately 30 million years ago. The termites cultivate fungi of the genus Termitomyces on specialized structures called fungus combs, built from chewed plant material that the termites collect and process. The termites eat the fungus and the partially-digested plant material together, which is more efficient than the leafcutter approach because both the fungus's digested plant material and the fungus tissue are nutritive.
The termite system has its own pathogen, a different species of Escovopsis, and its own defensive bacterial associates. The convergent evolution is striking: starting from completely unrelated insect lineages, both systems converged on cultivated fungus, host-specific pathogen, and bacterial-antibiotic defense. The convergence is strong evidence that this combination is the stable equilibrium for the agricultural-mutualism niche, and that any insect lineage that gets onto this path is likely to end up at the same destination.
The termite mound architecture (which we covered in our piece on termite mounds) is partly a climate-control system for the fungus garden. The fungus requires stable temperature and humidity to flourish, and the mound's passive ventilation maintains these conditions despite external temperature swings of 30 degrees Celsius. The ants and termites have converged on similar climate-control solutions: ants in their underground galleries, termites in their above-ground mounds, both producing the constant 28-30 degree Celsius environment that the fungus needs.
The wider significance
The leafcutter system is significant for several reasons that go beyond entomology. The first is that it provides a working model of agricultural mutualism that has been stable for tens of millions of years, which is informative for thinking about the long-term sustainability of human agriculture. Human agriculture is 12,000 years old, has involved repeated near-failures from soil exhaustion and disease, and has only been industrialized for about 150 years. The leafcutter system has been running for 60 million years and has not crashed. Some of the stability is presumably due to the four-way mutualism providing redundant defenses, and some is due to the smaller scale (each colony is its own farm rather than a continuous monoculture), but the underlying lesson is that long-term agricultural sustainability is achievable if the architecture is right.
The second significance is the antibiotic discovery angle. The system has been running antibiotic-mediated pest control for tens of millions of years without the pathogen evolving complete resistance, and there is presumably something to be learned from how it works. The early-2000s excitement about Pseudonocardia antibiotics as a source of new human drugs has not yet produced clinical results, but the underlying mechanism (diverse cocktail, host-controlled application, environmental rather than systemic exposure) is informative for thinking about how to slow resistance evolution in human medical contexts.
The third significance is the evolution-of-cooperation angle. The leafcutter system involves four-way cooperation across three biological kingdoms (animal, fungus, bacterium), with each partner depending on the others for survival. The stability of the cooperation over geologic time has implications for evolutionary biology and for the broader question of how complex cooperative systems remain stable. The lessons are mostly negative: the system is stable in part because the partners are co-adapted in ways that make defection difficult, in part because the partners are physically constrained in ways that limit the options, and in part because the system has had 60 million years to filter out unstable variations. Replicating the stability in human institutional or technological contexts is a much harder problem than the leafcutter case suggests, because human contexts have neither the genetic constraints nor the geological timescale.
What this is and is not
The honest summary: leafcutter ants invented agriculture 60 million years ago, the system involves a four-way mutualism between ant, fungus, antibiotic-producing bacterium, and (less actively) plant substrate, the agricultural behaviors are sophisticated enough to include crop selection and disease management and waste handling, and the system has been stable for far longer than any human agricultural system. The independent reinvention in termites confirms that the basic architecture is the stable equilibrium for this niche.
The story is sometimes told as a simple natural-wonder, which understates the sophistication of the mutualism. It is also sometimes told as a model that humans should imitate, which overstates the relevance: the constraints that keep the ant system stable do not generalize to human technological contexts, and the lessons that do generalize (diverse defenses, host-controlled application, small scales) are mostly about what does not work rather than what does. The right way to read the leafcutter system is as a working example that complex cooperative agriculture is possible, that it can be stable over geologic time, and that the architecture for stability is more constrained than it might appear. The system is also genuinely beautiful in the way that long-running co-evolved systems are beautiful, with each partner shaped by the others in ways that produce a functional whole that no single partner could have designed. The deeper observation is that evolution is a slower and more thorough engineer than any human institution has ever managed to be, and the products of evolution consistently reward attention from the people who notice that the inventory of biological achievement is much larger than schoolroom biology suggests.