# How real is it?

The species is *Atta sexdens*, a South American leafcutter ant. Leafcutters build the biggest
underground nests of any ant: mature colonies have thousands of chambers, reach several metres
down, and hold millions of workers. They don't eat the leaves; they grow a fungus on them and eat that.

The farm is a slab because real ant farms are thin glass slabs too. The simulation is 2D (1 cell = 1 mm);
the 3D view shows the same slab with the tunnels carved into its face. The clock is a real clock
(1× means one real second per second).

## What follows the known biology

- **Founding.** After her mating flight the queen sheds her wings, digs a narrow shaft alone, hollows
  a small chamber at the bottom, plugs the shaft from inside, and starts a fungus garden from a pellet
  she carried from her mother's nest. She raises the first brood sealed in, without foraging.
- **First workers are tiny.** The first brood is mostly minims (garden tenders) and small workers.
  The range of worker sizes widens as the colony grows; big soldiers (majors) only appear in large colonies.
- **Workers open the nest**, dig through the plug, and foraging begins.
- **Fungus farming.** Foragers bring leaf fragments in; the garden grows in chambers from the floor up;
  spent fungus is removed. Egg-laying slows when the garden is short.
- **Nest size follows colony size.** Ants dig when space runs short and stop when it doesn't
  (demand-driven excavation), so the nest grows with the colony.
- **Food is finite.** Leaves come from real plants on the surface. Plants regrow logistically: fast when
  healthy, slowly when stripped, a little faster in rain. Leafcutters are choosy, so each plant species
  has a different appeal. A hungry garden means fewer eggs and more worker deaths.
- **Trails and recruitment.** A forager coming back from a good plant recruits nestmates to the same
  plant, which is why leafcutters forage in columns. Busy trails wear a darker path on the surface.
- **Division of labour.** Each ant has its own response thresholds for digging, foraging and defence,
  and takes up a task when the colony's need for it outweighs her threshold (the response-threshold
  model, Bonabeau, Theraulaz & Deneubourg 1996). Task also depends on size, as in *Atta* (Wilson 1980),
  and on age: young workers stay in the garden, older ones go outside.
- **Most workers are idle at any moment**, resting in the garden. That's real, and it's also what
  lets the sim run cheaply: resting ants sleep in a queue until they wake.
- **Rival colonies fight.** Leafcutter colonies are territorial; workers of different colonies that
  meet usually fight, and big-headed majors win far more often than small workers. Fighting raises an
  alarm that sends soldiers out to patrol. A colony whose queen dies can't make new workers and fades out.
- **Rain.** Foraging stops in rain. Downpours send water down the entrances into the deepest tunnels;
  sand drains fast and clay slowly. Brood and fungus in flooded chambers are lost.
- **Escovopsis** is a mould that attacks only attine fungus gardens (Currie et al. 1999). Workers weed
  infected garden out; if they can't keep up they dump the whole chamber's garden.
- **Phorid flies** parasitise leafcutter foragers, and minims ride on leaf fragments to fend them off
  (Feener & Moss 1990). A parasitised forager dies days later.
- **Army ants.** *Nomamyrmex esenbeckii* raids mature *Atta* nests for brood (Powell & Clark 2004). Raids
  last hours here, come in waves, and are fought mostly by the big-headed majors.
- **Nuptial flights.** *Atta sexdens* flies at the start of the rains (spring in SE Brazil), usually
  after rain. Daughters carry their mother's traits.
- **Soil is conserved.** Every grain dug out is carried to the surface and dropped on the mound
  (rolling to a stable slope), so the mound's stripes show which soil layer it came from.

## What is an estimate (marked `EST` in `sim.js`)

These numbers are from memory of the literature, **not yet checked against papers**:
the egg-to-worker time (45 days), worker lifespan (120 days, with individual spread), the egg-laying curve,
digging speed per mm², walking speeds per caste, founding shaft depth (~17–23 cm), chamber sizes, the angle
of repose, how fast the garden is consumed, rain frequency, flood volumes and drainage rates, drowning odds,
mould growth and weeding rates, phorid attack odds, raid frequency and brood taken, and the mutation size.

## What is a simplification, not biology

- **Blueprints.** Real ants have no plan; tunnels and chambers emerge from local cues (pheromone in the
  soil, crowding, gravity). Here the colony picks a meandering tunnel path or chamber shape, then the
  individual ants dig it pellet by pellet. The digging is agent-by-agent; the plan is not.
- **Colony traits** are a game abstraction: six numbers standing for many genes and behaviours.
- **What colonies learn** is also an abstraction. After repeated floods a colony piles soil in rims around its
  entrances, and after a raid it raises more majors and posts guards. Real colonies do shift behaviour and caste
  investment with experience, but these particular responses and their strengths are made up for the farm.
- **Floods** follow simple rules: water runs down the tunnels, pools in the lowest ones, and leaves mud on the
  floors as it drains. When it reaches the queen, she and her nurses move the brood to a dry chamber.
- **Empty nests are taken over** by the nearest colony, which digs through to them.
- **Flights come years too early.** Real *Atta* colonies first produce winged queens at about three years
  old with huge populations. Here a colony of 500 workers can fly, because the 1 m farm fills up in year one.
- **Short trails.** Real *Atta* forage up to ~100 m from the nest; here the plants are within the 1 m farm.
- **Fights are one-on-one** and decided by caste strength, with no injuries and no group pile-ons.
- **Dots stand for several ants** once a colony passes its share of 900 dots (the HUD shows the ratio).
- **Fast-forward** (1 day/s, the day-100 start, and catching up after time away) moves ants along whole
  walks at once instead of millimetre by millimetre, taking the time the walk would have taken.

## Good next steps

1. Check the `EST` numbers against Autuori's founding studies and Hölldobler & Wilson's *The Leafcutter Ants*.
2. Swap blueprints for emergent digging (pheromone-laced pellets, Khuong et al. 2016 style).
3. Waste chambers, Pseudonocardia (the antibiotic bacteria workers carry against Escovopsis), and live
   Amazon weather driving the rain.