In one mouth

A kept thing carried out of a dying colony’s grounds by one foundress, in a pocket in her own mouth — a pellet of the fungus she was raised on, held through her mating flight and planted as the first act of a colony that will starve without it

R894 closed on a keeping renewed from companions: a termite’s gut partners die before every moult, and any intermolt nestmate can restock them. Its closing turn asked for the opposite distribution of trust — renewal that depends on one irreplaceable carrier at one unrepeatable moment. It exists wherever leafcutter ants fly, and the whole of it fits inside a mouth.

A crop that grows only indoors. The leafcutter ants, Atta and their relatives, do not eat the leaves their columns carry home. They feed the clippings to an underground fungus — for the leafcutters, chiefly Leucoagaricus gongylophorus (long filed as Leucocoprinus) — and eat the fungus, which converts leaf pulp they cannot digest into nutrient-rich swellings, gongylidia, that feed larvae and queen. The partnership is old farming, tens of millions of years older than ours, and it is a monoculture: DNA fingerprinting of Atta gardens surveyed over five years found a single fungal genotype per garden, with no polyculture and no genotype turnover — variation arose only from mutation during clonal propagation. A colony’s garden is one lineage, grown in many chambers. Which raises the only question that matters at founding: how does the lineage get to a new address?

Not from the wild. The cultivar is not gathered. Before her nuptial flight, a virgin queen visits her natal colony’s garden and packs a pellet of its mycelium into her infrabuccal pocket, a small pouch inside her oral cavity. That pellet is the colony’s entire agricultural inheritance, and she is its only vehicle: she mates in the air carrying it, lands, sheds her wings, digs — in Atta sexdens accounts, a vertical tunnel some thirty centimetres down ending in a small chamber — and within the first days expels the pellet as the starter inoculum of her incipient garden. There is no second trip. If the pellet is lost, spoiled or fails to take, the founding fails with it; a queen who reaches her chamber without a living garden has brought sperm for millions of workers and food for none of them.

Farmed alone, on her own body. What follows is claustral founding: the queen tends the pellet alone, in the sealed dark, feeding the young garden with her own faecal material and raising the first brood on it and on trophic eggs, while she herself eats nothing — she runs on stored fat and on her flight muscles, histolysing the engines of the one flight she will ever make to pay for the garden’s first weeks. Only when the first cohort of workers ecloses — some forty to sixty days in, in the A. sexdens accounts — do they take cultivation off her, and she narrows to the one job she keeps for life. Until that handover, farmer, nurse and fertiliser are a single animal, and the crop’s whole world is the inside of her chamber and the inside of her body.

The bottleneck is the point, and the risk. Vertical transmission through one mouth makes the fungus a clone-lineage tied to the queen’s own: garden fidelity across generations, no wild recruitment, no neighbours to borrow a strain from. It also stakes everything twice over — the foundress is already the colony’s single point of failure, and the pellet makes her a single point of failure for two species at once. One honest complication belongs on the page: the domestication is not total. Genetic work on leafcutter cultivars has found signatures of recombination — fungal genes exchanging across colonies and even across water gaps too wide for ants — so the cultivar appears to retain some life of its own, occasionally dispersing independently of its carriers. That is a rare leak in the system, not its plumbing: the road a new colony actually travels is still the pocket, the flight, and the planting.

Why this keeper is different from the others in the run. Buchnera passed inside the egg, never leaving a body. The squid’s partner was sieved from an open sea full of candidates. The koala’s was served once, by a mother who stayed present to serve it. The termite’s is re-drunk from companions at every moult, whoever is near and stocked. The ant’s crop crosses between colonies exactly once per generation, in exactly one body, at exactly one moment when that body is also mating, flying, digging and about to fast — the keeping’s entire future compressed into a mouthful carried by a traveller who cannot go back for more.

One founding, in order: the garden in the natal colony, the pocket filled, the flight and the digging, the pellet planted and farmed alone, the first workers taking over. Schematic, compressed from the published record.

Walk it in order.

Schematic only: the stages compress work across species, studies and decades into one foundress’s flight. Labelled cautions: the thirty-centimetre tunnel and forty-to-sixty-day figures are Atta sexdens life-cycle accounts, not an attine-wide timetable; founding is not single-queen in every case — multi-foundress associations occur and pleometrosis has been confirmed in Atta texana; some accounts allow the foundress occasional plant material, so “eats nothing” describes the claustral pattern, not a law for every species; the five-year monoculture result is for surveyed Atta gardens and does not by itself prove every attine garden is clonal — lower attines show cultivar shifts; and the recombination evidence means the cultivar must not be described as having wholly lost sexual reproduction or all independent dispersal — vertical, clonal carriage by queens is the dominant road, with rare genetic exchange reported alongside it.