Movement One · The Colony
Chapter 2
What the Ants Are Doing
At dawn in the Chihuahuan Desert, before the sun has cleared the ridgeline, a harvester ant colony is already awake. The temperature on the sand is still bearable — perhaps twenty degrees Celsius, climbing fast. The nest entrance is a hole the width of a thumb, surrounded by a cleared disc of bare ground a meter or so across, swept of debris by the ants themselves over many days. Nothing moves at the entrance for a while. Then a few ants come up, hesitate at the rim, and go out.
These are not foragers. They are patrollers, and they always go first.
Watch what they do, and what the colony does in response, and the day unfolds with a logic that has no author.
The patrollers spread out across the foraging area in the directions the colony's foragers will later take. There are not many of them — a few dozen, in a colony of perhaps ten thousand. They move over the ground, antennae down, and then they come back. This is the whole of their task in the early morning: go out along a direction, and return. What they are testing is not visible from above. A patroller that goes out and meets trouble — a predator, a competing colony's workers, a stretch of ground that is simply too hot — does not come back, or comes back slowly, or comes back having met too few of its own kind along the way. A patroller that goes out and finds the way clear comes back promptly.
The species here is the red harvester, Pogonomyrmex barbatus, and the colony at the entrance is older than it looks. A colony of this kind takes about five years to mature, and once mature it can live fifteen to twenty more, the whole arc of it governed by a single queen who mated once, on a single nuptial flight, and stored enough sperm that morning to lay eggs for the next two decades. The thumb-wide hole leads down into a structure of branching chambers and tunnels reaching a meter or two into the desert, dug and redug by the ants over the colony's life. Most of what the colony is, the observer never sees. What surfaces at dawn is the thin outer edge of it.
And here is the thing the colony does with that, which is the first thing worth slowing down for.
The foragers do not leave the nest until the patrollers return.
Deborah Gordon spent decades establishing this, colony by colony, season by season, in a study population near Rodeo, New Mexico, that has been observed continuously since 1985. The foragers wait underground. They are ready. But the foraging day does not begin on a schedule, and it does not begin on a command. It begins when patrollers come back into the nest in sufficient numbers. The returning patrollers are themselves the signal. A forager that has been touched, in the dark of the corridor, by enough returning patrollers in a short enough time begins to leave. If the patrollers do not return — if Gordon quietly removed them with forceps as they came up, which she did — the foragers stay home. The food is out there. The foragers are willing. But without the returning patrollers, the colony does not send them.
No ant decided this. No ant assessed the morning and judged it safe. The patrollers tested the ground simply by walking on it and surviving the walk. Their return is the assessment. The colony reads the safety of the desert in the rate at which its own scouts come home.
There is an order to the morning, too, that no one set. The nest-maintenance workers come up first of all, in the dim cool, clearing debris and carrying out the previous day's refuse, before the heat arrives. Then the patrollers, settling the directions. Only then the foragers, in the window between the patrollers' return and the moment the sand turns lethal — a window that may be only a few hours wide. The sequence repeats every day, but it is not a schedule any ant consults. Each group's activity helps trigger the next. The maintenance work and the patrolling change the traffic at the entrance, and the changed traffic is what tips the foragers into leaving. The day assembles itself, one task handing off to the next, the way a line of dominoes is not a plan but still falls in order.
The sun is higher now. The patrollers have come back. Foraging has begun.
A harvester ant colony eats seeds. The desert scatters them unevenly — a wash of them here where the wind piled up last season's grasses, almost none there on the baked flat. The colony's problem, every morning, is to find the seeds and bring them home before the heat makes the surface lethal, which it will, by late morning, when the sand passes sixty degrees and an ant caught in the open cooks in minutes. The colony has a few hours. It cannot afford to send everyone everywhere.
And there is a real cost to being wrong, which is what makes the regulation matter. A forager out on the surface is losing water with every minute, in air that pulls moisture out of a small body fast. Foraging is, in plain terms, the colony spending water and risking lives to gather seeds. Send too many out on a poor morning and the colony burns more than it brings home; the lizard takes some, the heat takes others, and the survivors return with little. Send too few on a good morning and the seeds go ungathered and a neighbor takes them. The colony is running, every morning, a calculation about whether the desert is worth the water — and it runs that calculation without any term in it for water, or seeds, or risk. It runs it as a rate of contact at a door, and the rate happens to encode the answer. The economics are real. The arithmetic is nowhere.
So watch a single forager. It leaves the nest along one of several trails — faint paths worn and chemically marked by the ants that went before — and travels out, sometimes ten or fifteen meters, which for an animal a few millimeters long is an enormous distance, the equivalent of a person walking several kilometers. A forager will tend to leave by the same trail it used the day before, and to search the same direction; the colony's effort is not scattered evenly around the compass but committed, for stretches of days, to particular sectors of the desert. It searches. It is not efficient in any way a person would recognize; it casts about, doubling back, following the ground, keeping its bearings off the angle of the sun and the look of the landmarks around it. When it finds a seed it picks it up and carries it home, in a path far straighter than the wandering search that found the seed — the ant having kept a running tally of the twists of its outbound trip and collapsed them into a direct line home. When it gets home it goes inside, drops the seed, and waits.
The seed it dropped does not get eaten right away. Harvester colonies store seeds in chambers near the surface, granaries the colony builds and tends, and a colony may hold a great many of them against the lean stretches between rains. Some seeds the ants mill into a soft mash to feed the brood; some sprout in the warm chambers and are carried out before they take root. The colony is not foraging for tonight's meal. It is foraging for a buffer, a hedge against a desert that gives generously for a few weeks and then gives nothing for months. The single seed in a single ant's mandibles is a deposit into a store that no ant manages and no ant can see the size of. The store is the colony's memory of good mornings, kept in a currency it can eat.
Then it goes out again. But not immediately, and this is the second thing worth slowing down for.
A forager that has just returned waits inside the nest before going out for its next trip. How long it waits depends on one thing: how quickly other foragers are coming back. If returning foragers are streaming in — antenna brushing antenna in the entrance chamber, contact after contact in quick succession — a waiting forager leaves again soon. If returns have slowed, if the ant inside is touched only now and then by an incoming sister, it waits longer. It may not go out at all.
Gordon measured this directly. She would drop extra foragers, or seeds, near the entrance to raise the rate of returns, and watch outgoing traffic climb. She would intercept returning foragers to lower the rate, and watch outgoing traffic fall. In one version of the experiment she replaced the returning foragers with tiny glass beads coated in the colony's own scent and dropped them into the entrance at controlled intervals; the colony sent out foragers in response to the beads, as if the beads were sisters coming home. It was the rate of arrival the colony was reading, not the food, not the foragers themselves — the rhythm of contact at the door. The relationship was steady. The rate at which foragers leave the nest is set by the rate at which foragers come back.
It is worth being precise about the interval involved, because the precision is the surprise. A returning forager that meets the next incoming forager within something like ten seconds is likely to head out again soon. Stretch the gap to half a minute and it tends to stay. The ant is not timing anything in any sense it could report. It is holding, in its quarter-million neurons, something like a brief running average of how recently it was last touched, and acting when that average crosses a line. The line differs from ant to ant. Some leave readily; some are reluctant. The spread is not a flaw to be corrected. It is what lets the colony respond smoothly across a range of conditions rather than lurching all at once — a few eager ants probe when returns are thin, the cautious majority commit only when returns are strong.
Consider what that accomplishes, with no one accomplishing it.
A forager comes back faster when food is close and plentiful — it finds a seed quickly and returns. It comes back slower when food is scarce or far — it searches longer before it finds anything to carry. So a high rate of returns means the foraging is going well, and the colony responds by sending more ants out to exploit it. A low rate of returns means the seeds are gone, or the patch has been picked clean, and the colony responds by easing off, keeping its workers safe inside, spending less effort for less reward. The colony tunes its entire foraging force to the productivity of the desert that morning, continuously, without any ant knowing the productivity of the desert that morning. Each ant knows only how often it has been touched.
There is no forager in charge of foraging. There is no count kept anywhere of how many ants are out. The number of ants foraging at any moment is a quantity the colony regulates precisely and that no single ant represents. It lives in the rate of contact in a dark chamber. It lives in the substrate, not in the ants.
The harvester does most of this without laying much of a chemical trail at all; it runs largely on contact, ant against ant, at the door. Other species lean harder on the chemistry, and laboratory work on them shows the same logic in a different medium. A forager that finds a rich food source and returns lays a chemical mark on the way home, and other ants follow the mark out. The richer the source, the more ants return successful, the more marks are laid, the stronger the trail, the more ants it pulls — a path that strengthens itself the better it works. When the source runs dry the returning ants stop marking it, the chemistry is volatile and fades within minutes, and the trail simply disappears for lack of reinforcement. Researchers have set out two feeders of unequal richness and watched a colony, with no individual comparing them, concentrate almost entirely on the better one within the hour, purely because the better feeder generated more successful returns and therefore more marking. The mechanism differs — contact in one species, chemistry in another — but the shape is identical. Success reinforces the path that produced it. Failure is the absence of reinforcement, and the path fades. No one tallies the result.
Now disturb it.
This is where the colony shows what it is, because a system that only works when nothing goes wrong is not intelligent — it is merely arranged. The desert does not leave colonies undisturbed. A summer storm floods a nest. A horned lizard sits by an entrance and eats foragers one after another. A neighboring colony pushes into a productive patch. A researcher, deliberately, makes a mess.
Consider the flood first, because it is the harshest test. The Chihuahuan summer brings sudden, violent rain; a storm can drop water faster than the ground can take it, and a nest dug into a meter of soil can fill. The colony does not have a flood plan. What it has is a structure that drains and a workforce that responds to where the water and the digging are. Ants caught below move upward and outward; the chambers and tunnels, shaped over the colony's years for exactly the soil it lives in, shed much of the water on their own; the maintenance work that any other day clears debris now clears mud and reopens collapsed passages. The colony does not recover because it understands the flood. It recovers because the same local rules that run an ordinary morning — go where the work is, respond to the ants around you — happen to add up to flood recovery when there is a flood. A system robust to disturbance is not a system that anticipated the disturbance. It is a system whose ordinary behavior already covers the case.
Watch what happens when the mess is made.
In one kind of experiment, the disturbance is a small obstruction or disarrangement near the nest entrance that the colony must clear — a scattering of toothpicks, a barrier of debris. The work that needs doing has suddenly changed. The colony has nest-maintenance workers, but not enough of them for a sudden mess, and foragers it would normally send out for seeds. Within a stretch of minutes to an hour or two, the allocation shifts. Ants that had been doing one task take up another. Foragers become nest workers. The colony does not have a manager who notices the mess and reassigns the staff. The mess changes the pattern of encounters each ant experiences near the entrance, and the changed encounters change what each ant does next.
The reallocation runs on the same currency as everything else: an ant's recent history of brief contacts. An ant that keeps meeting other ants carrying debris, in a place where debris needs carrying, takes up carrying debris. The task is in the air, in the traffic, in the rate and kind of the touches. Gordon ran exactly this kind of disturbance and watched the four outdoor tasks — foraging, patrolling, nest maintenance, refuse work — bleed into one another. Pile work onto one task and ants are drawn out of the others to meet it; the groups are not sealed compartments but pools that drain toward wherever the encounters say the work has gone. An ant moved from foraging to maintenance was not reassigned. It met a different pattern of ants at the entrance, and the different pattern is the instruction.
When the disturbance is severe — a patch the colony depended on suddenly raided and emptied — the foragers working that direction come home slow and empty, the rate of productive returns from that direction falls, and over the following hours the colony's effort drains away from the dead patch and pools toward the directions still paying off. Within hours, not days, the colony's workforce has been redistributed to match a world that changed that morning. Nobody redistributed it.
There is a subtlety here that took years of fieldwork to see, and it changes how the disturbance should be read. The colony does not respond as a fixed machine would, the same way every time. Its readiness shifts with its recent history. A colony that has been through a stretch of hard mornings — disturbance after disturbance — grows more conservative, slower to pour foragers out onto uncertain ground, as if the run of bad experience had lowered the whole colony's appetite for risk. None of the ants remember the hard mornings; ants live about a year and many of those present did not witness the earlier trouble. What carries the memory is the state of the substrate — the worn or unworn trails, the standing population, the prevailing rhythm of contact the colony has settled into. The disturbance does not just change what the colony does today. It leaves a mark on how the colony will behave tomorrow.
What is striking, watching this happen, is the timescale. The colony is not slow about it. A patch goes dead in the morning and by early afternoon the colony has already pulled away from it. The reallocation does not wait for any ant to grasp the situation, because no ant ever will. It moves at the speed of the contacts themselves — each ant's behavior updating with every touch it gets, the whole pattern shifting as fast as the traffic shifts. There is no delay for a decision to be reached, because there is no point at which a decision is reached. The response and the disturbance are nearly the same event, separated only by however long it takes for the changed traffic to propagate from one ant to the next.
There is one more thing the colony does, out at its edges, that is worth watching, because it looks the most like negotiation and contains the least.
A mature harvester colony holds a foraging territory, and its neighbors hold theirs, and the boundaries between them are real — stable, often, for years. But no ant has ever walked the perimeter of the territory. No ant carries a map. No ant has any idea that a territory exists. The boundaries are held in place by encounters at the edges. Where foragers from two colonies meet, each tends to turn back toward home; the meeting itself is mildly repellent. A direction in which one's own colony's foragers are dense and a neighbor's are sparse is a direction worth pushing into; a direction thick with the neighbor's traffic is a direction to ease away from. Each colony's foragers, responding only to whom they bump into, settle collectively onto a line that neither colony drew. The boundary is an agreement that no party negotiated, written in a medium — the pattern of who meets whom, how often — that no party can read.
The strangeness is recognized by smell. An ant wears its colony's identity on its body — a layer of waxy compounds on its shell, a particular blend shared across nestmates and slightly different in every colony. When two foragers' antennae touch, each reads the other's blend in an instant. Nestmate: carry on. Stranger: a different reaction entirely. No ant was taught the difference and no ant holds the boundary in mind; the recognition is as immediate and as local as a touch, and the boundary is only the sum of millions of such touches across a season.
A neighbor weakens, sends out fewer foragers, and the line drifts toward it. A neighbor grows and the line drifts back. The territory breathes with the relative strength of the colonies around it, year after year, and the whole arrangement is maintained by ants that turn around when they bump into a stranger. The map exists. It is accurate, it is current, it adjusts itself to the rise and fall of the neighbors, and it is written nowhere that any single ant could ever read.
It is worth holding, for a moment, how little is inside any of these ants.
A harvester ant carries on the order of a few hundred thousand neurons — call it a quarter of a million, in a brain that occupies a fraction of a cubic millimeter. It cannot count. It has no representation of the colony, no sense of the morning's foraging total, no concept of the territory it helps defend, no awareness that a flood has reorganized its work. It can do a small number of things. It can move. It can carry. It can detect, through its antennae, the chemistry of a trail and the brush of another ant, and it can keep a rough running sense of how often that brush has been happening lately. It has a threshold, somewhere in that quarter-million neurons, and when the rate of contact crosses it, the ant changes what it is doing. That is most of the ant.
Take one of these ants out of the colony and put it alone on a patch of clean sand, and it does almost nothing worth the name. It walks. It does not forage usefully, because foraging is not a thing one ant does — it is a thing that only means anything against a stream of returning sisters that no longer exists. It does not patrol, because there is nothing to report a clear path to. It does not maintain a nest, defend a border, or regulate anything. Stripped of the traffic that told it what to do, the ant has almost no behavior left. The rules inside it are rules about other ants, and with no other ants the rules have nothing to act on. The single ant is not a small colony. It is a component with its inputs cut.
And yet, stacked up — ten thousand of these, several thousand trips a day, every day for the life of the colony — they produce a system that decides when the desert is safe to enter, matches its labor to the food available hour by hour, redistributes its workforce within hours of a disaster, and holds a stable border with its neighbors across years. Each of those is a hard problem. Engineers are paid well to solve versions of them. The colony solves all of them at once, continuously, with no part of it understanding any of them.
It is tempting, watching this, to look harder at the ant — to assume the cleverness must be in there somewhere, hidden, if only one could read a quarter-million neurons. But the harder one looks at the ant, the simpler it gets. The threshold is simple. The contact-counting is simple. The rule that turns a rate into an action is about as simple as a rule can be. The complexity is not in the ant and getting hidden. It is not in the ant at all.
The work is done by the ants. The deciding is done somewhere else — in the rate of contact in a corridor, in the chemistry fading on a trail, in the pattern of who meets whom at the edge of the world. The ants supply the labor and the touches. The substrate supplies the intelligence.
There is a longer rhythm underneath the daily one, worth naming before the day closes. None of these decisions were learned by any ant, and none were designed by any ant, but they were shaped — over a great deal of time. A colony that reads the morning badly, that sends foragers out when it should hold them or holds them when it should send, brings home fewer seeds, raises fewer young, sends out fewer new queens on the nuptial flights of later years. A colony that reads the mornings well does the opposite. The thresholds inside the ants, the readiness to leave at this rate of contact rather than that one, are inherited; the daughters of a well-tuned colony tend to be tuned the same way. Over generations the desert keeps the settings that work and quietly discards the ones that do not. The intelligence on display this morning is not the cleverness of any ant. It is the residue of a very long run of mornings, most of which ended in colonies that no longer exist.
By late morning the surface has become an oven. The seeds that were going to be found have mostly been found. The rate of returning foragers, which set the whole machine in motion at dawn, has been falling for an hour, and as it falls the outgoing traffic falls with it. Fewer ants leave. The ones outside finish their trips and come home and are not replaced. The trails, no longer walked, no longer refreshed with chemistry, begin to fade in the heat. By the hottest part of the afternoon the cleared disc around the entrance is empty. The colony has pulled itself inside, not by decree, but because the conditions that sent it out have quietly reversed.
In the cool of evening a little activity may resume, and at dusk it tapers again, and the entrance goes still. Deep in the nest the queen lays eggs she has been laying all day, unaware that any of this took place. She did not order the morning's foraging or learn of its outcome; she is the colony's ovary, not its head, and the colony around her ran the day without consulting her once.
The colony has eaten. It has lost a few foragers to the lizard and the heat, and made a few more in the brood chambers to replace them. It decided, this morning, when the desert was safe to enter, and matched its labor to the seeds available hour by hour, and shrugged off a disturbance, and held a border with colonies it will never know exist. Every one of those decisions was real, and not one of them was made by anyone. They were made in the rate of contact at a door, in the chemistry fading on a trail, in the smell of a stranger met at the edge of the world. It has held its ground.
Tomorrow the patrollers go out first, and it happens again.