Explanatory sheets: what really happens behind the ant farm
Before you start
Ants Life is a game, and behind the game there is real myrmecology. These sheets tell, mechanism by mechanism, what happens in a real ant colony, what Ants Life keeps of it, and what it bends to stay playable. Every simplification is a good question to ask.
Each sheet follows the same plan: the real mechanism as biology describes it, what the game keeps of it (with the actual parameters of the code), what the game simplifies or distorts, an experiment to run in the game, and a few questions to think about, with a hint for the answer.
Getting the game ready
For the experiments, pick The yard in the menu (M or Esc), except for sheet 6, which uses The double bridge, and sheet 14, which uses The house. There are no sliders to set, only the weather keys, the speed of time and the share of soldiers.
Need
Where
Start a clean game
R, or the button New game of the panel
Weather
Keys 1 rain, 2 snow, 3 heat, 4 cold, 5 wind; the same key stops it
Show the trails
T
See where the ants walk
K shows the heat map
Follow the numbers of ants, food and temperature
G opens the charts
Change the speed of time
+ and - (1/4× to 8×), Space to pause
Read a creature, a nest, a pile
Hover with the pointer (tooltip) or click an ant to follow it
Look inside a nest
Click it; Esc closes the view
Read what happened
The Journal (J)
Share of soldiers
The slider Soldiers among new ants of the panel (0 to 60%, 20% at first)
The scales of the game
At normal speed the game computes 60 steps per second. The whole farm runs at 64% of its first pace to be calmer to watch, and all the durations of this document are in game time. A year lasts four seasons of 12,000 steps, that is 13 minutes and 20 seconds. An ant lives between 14,000 and 25,000 steps, that is between about 3.9 and 6.9 minutes. A pixel has no fixed size in real units.
These scales cannot be right for everything at once. Sheet 12 shows how much the game compresses: a colony of 130 ants against several thousand, an egg that hatches in 9 seconds against several weeks. It is a good way to understand what a model is: a choice between what is kept right and what is given up.
Goals as a way in
Many of the 24 goals of the yard overlap with the sheets: Found a new colony (sheet 3), Dig a nest with 6 rooms (sheet 4), Bring 300 crumbs home (sheet 5), Live through 3 showers and Save 5 eggs from the water (sheet 7), Get through a winter (sheet 8), Bring down the spider (sheet 9), Defeat the red colonies (sheet 10), Send the yellow ant to look around 3 times (sheet 11), Raise a big ant, Find a herd of aphids and Bring down a ladybird (sheet 13).
The house has 12 goals of its own, such as Hold three rooms or Get through a giant foot (sheet 14).
The results given in the experiments are computed from the constants of the code. Chance matters (the creatures arrive at random times, each nest floods at its own pace): a single run may differ, compare several attempts.
01 Leaving a trace: pheromone trails and stigmergy
The yardCollective behavior
No ant knows where the food is. The ground remembers for them, for a few seconds at a time.
The real mechanism
When a foraging ant finds food, it goes back to the nest and lays a trail pheromone on the ground, from a gland whose place depends on the species. Ants that leave the nest later follow the trail by comparing the smell on their two antennae, and when they find food they lay a trail in their turn. The more ants use a path, the more it is marked, the more ants it attracts: a positive feedback. The trace is not stored in any ant. The French biologist Pierre-Paul Grassé named this indirect coordination through the environment stigmergy in 1959, when studying termites.
The pheromone also evaporates. Depending on the species, the substrate and the temperature, an unrenewed trail stays attractive from a few minutes to many hours. Evaporation is as important as deposit: it makes the colony forget a source that has run dry, or a route that was a mistake. Pheromones are potent: an often-quoted estimate says that one milligram of the trail pheromone of a leaf-cutter ant would be enough to mark a trail circling the Earth several times. Ants also vary their effort: in Lasius niger, an ant that found better food lays more, so a richer source attracts a bigger flow.
In Ants Life
An ant that carries a crumb marks the ground cell (8 pixels wide) it stands on at every step: deposit = 0.64 times its freshness, with a cap of trailMax = 14 per cell. The freshness starts at 1 at the find and loses a thousandth per step (freshness = 0.99904, halved after about 720 steps), so the mark is strongest near the food and weaker towards the nest: the trail is a slope that points to the food. The trail evaporates by keeping evaporation = 0.9962 of itself per step, a half-life of 182 steps, that is 3 seconds; from the cap of 14 it falls under the smallest level an ant can feel (0.05) in about 1,480 steps (25 seconds). Each colony reads only its own trail.
An ant looking for food takes three readings 14 pixels away (senseDistance): ahead, and to the left and right at 0.55 radian (about 31.5 degrees, senseAngle). If the best reading is at least 0.05 it turns towards it by 0.16 radian per step and wanders less (noise 0.16 instead of 0.4). An ant carrying a crumb goes home by path integration, turning towards the nest by 0.077 radian per step, and follows its own trail as well. Neither walks a ruler line: at every step each adds to its heading a random tremble of up to 0.15 radian either way and a slow wobble of its own, a sine shifted by the number of the ant (amplitude 0.07 radian and a period of about 140 steps for an ant carrying food, 0.06 and about 170 steps for a forager on a trail), so a column spreads into a loose stream instead of a single file.
What Ants Life simplifies
Only ants that carry food lay a trail, and only on the way home. Many real species also mark on the way out, and mark more for better food. A crumb is a crumb here.
The half-life of 3 seconds is very short next to the many minutes to hours of real trails. It is the game's scale again: with long-lived trails, the ground would fill up before a player could see anything change.
There is one pheromone and one number per cell. Real trails are blends of compounds, with different lifetimes and some signals that say "do not come", as in Pharaoh's ant.
Ants read three points at once; a real ant compares its two antennae and zigzags along the trail.
Experiment: making a trail fade
Action
In the yard, press T so that the trails are shown. Wait until a line links a pile of crumbs to a nest. Then press 1 (rain) for a few seconds, and press it again to stop. Do the same with 3 (heat) and 4 (cold), at 1× speed.
Explanation
Weather changes the share of trail kept at each step. Rain adds a wash of 0.985 on top of the evaporation. Heat lowers the share kept, cold raises it.
Parameters
Keys 1, 3, 4; speed of time. The code: kept per step 0.9962 normally, 0.9962 × 0.985 = 0.9813 in the rain, 0.993 at full heat, 0.998 at full cold.
Check
The half-life of a mark goes from 3 seconds (normal) to about 0.6 second in the rain: the trail vanishes almost at once, and foragers wander until they meet a crumb again. At full heat it is about 1.6 seconds, at full cold about 5.8 seconds. The temperature moves slowly (about 17 seconds for a full swing), so the heat and cold effects build up.
Common pitfall
The trail layer is only a display: hiding it with T changes nothing for the ants. And the weather keys do nothing on the double bridge.
Food for thought
What share of a trail is left after 10 seconds without any new deposit in normal weather?
10 seconds are 600 steps: 0.9962^600 is about 0.10, a tenth.
Why must a trail evaporate? What would happen if it never did?
Without evaporation the first marks stay for ever and keep attracting ants, even to a source that is empty or a longer route: the colony could not change its mind.
In the game, why does an ant mark more strongly near the food than near the nest?
Because its freshness falls with the distance walked. That makes a slope that points to the food, and a follower that climbs it reaches the source.
Grassé, P.-P. (1959). La reconstruction du nid et les coordinations interindividuelles chez Bellicositermes natalensis et Cubitermes sp. Insectes Sociaux, 6(1), 41-80.
Beckers, R., Deneubourg, J.-L., & Goss, S. (1992). Trails and U-turns in the selection of a path by the ant Lasius niger. Journal of Theoretical Biology, 159, 397-415.
Beckers, R., Deneubourg, J.-L., Goss, S., & Pasteels, J. M. (1990). Collective decision making through food recruitment. Insectes Sociaux, 37, 258-267.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
A colony is a family where almost everyone is a sister, and where the roles are fixed by how a larva was fed.
The real mechanism
An ant colony is a reproductive family. One or a few queens lay the eggs; the workers, which are females that usually cannot mate, forage, nurse the brood, dig and defend; the males appear for a season, only to mate. A queen and a worker generally have the same genes: what makes the difference is mostly what the larva was fed and the hormones that follow, plus temperature and sometimes genetic factors. Males come from unfertilized eggs.
In some genera the workers themselves come in several sizes. Pheidole ants have small workers and big-headed soldiers (called "majors") that defend the nest and, in some species, crush seeds; in leaf-cutter ants the mass of the workers of one nest can differ by more than a hundredfold. Many common species, such as the black garden ant Lasius niger, have no real soldier: the workers all fight when needed. In most species the tasks also depend on age: the young stay in the nest with the brood, the old go out to forage, which is dangerous. Nobody gives the orders. The queen does not lead: she lays eggs.
In Ants Life
Three castes are simulated, plus the princesses (sheet 3) and, in the yard, the big ants that cut plants (sheet 13). A worker has 6 hit points and walks at 0.8 pixel per step; a soldier has 12 and walks at 0.67 (the yellow ant, a worker, walks at 1.1). A soldier stays near the nest (it comes back when farther than 110 pixels), charges any enemy within 80 pixels, a creature in sight, or the spider from 140 pixels, and bites four times harder in a fight (0.077 against 0.019 per contact step). The caste is drawn when an egg hatches: a soldier if a random number in 0 to 100 is below the colony's soldiers setting, the slider of the panel (0 to 60%, 20% at first, the red colony starts at 25%). Each colony begins with 12 workers and 3 soldiers, and lives 14,000 to 25,000 steps per ant, drawn at birth.
The queen is a single hidden individual of the nest, with her own hit points (1). She lays one egg every 140 steps (about 2.3 seconds) if the store allows it (sheet 5), and without her no ant is born.
What Ants Life simplifies
There is no larva: an egg becomes an adult after 560 steps (about 9 seconds), and the caste is a die roll. Real caste fate depends on feeding and hormones, and the colony adjusts it to its needs.
No males, no age-based tasks: the same worker forages whatever its age.
Only two kinds of workers. Real soldiers are not always better fighters than workers: it depends on the genus.
The queen cannot be seen or killed in the yard, except by a flood or an earthworm under the nest.
Experiment: the price of an army
Action
Start a game (R) and at once slide Soldiers among new ants to 60%. Run at 8× for two game minutes and read the row of the black colony in the panel (workers, soldiers, food, eggs). Then R again with the slider at 0% and run the same time.
Explanation
The slider only acts on the ants that hatch afterwards; the colony begins with 12 workers and 3 soldiers in both cases. Soldiers stay near the nest and only pick up what they happen to cross, so they bring back little food.
Parameters
The slider (0%, 20%, 60%), the same duration, the same speed. Leave the red colony alone.
Check
At 60% about six new ants in ten are soldiers: you should see many more soldiers and fewer workers, a smaller store and fewer ants in total. At 0% the colony grows faster but has nobody to charge a woodlouse or a spider and no defence against the red soldiers.
Common pitfall
Moving the slider late changes nothing for the ants already born. Chance matters: compare the share of soldiers, not the exact numbers.
Food for thought
With 20% of soldiers among new ants, what share do you expect after a long time, given that a soldier and a worker live the same duration?
Close to 20%: the colony starts with 3 soldiers out of 15, which is 20% too, and both castes live the same time, so the share among living ants stays near the share drawn at hatching.
Why does a real colony not need a leader to decide how many soldiers it has?
Because the feeding of larvae, and the encounters between workers, adjust the production to the needs, without any ant knowing the total.
The queen is not the boss of the game's colony. Why is it more exact to say that she is its ovary?
She lays the eggs, nothing else: no order, no decision, only the reproduction of the colony.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
Wilson, E. O. (1980). Caste and division of labor in leaf-cutter ants (Hymenoptera: Formicidae: Atta). I. The overall pattern in A. sexdens. Behavioral Ecology and Sociobiology, 7, 143-156.
Gordon, D. M. (2010). Ant Encounters: Interaction Networks and Colony Behavior. Princeton University Press.
03 Founding a colony: the nuptial flight and the princesses
The yardReproduction
Once a year, on a warm day, thousands of winged queens leave their nests. Almost none of them will found a colony.
The real mechanism
When a colony is mature, it rears winged males and virgin queens ("princesses"). On a warm, humid and calm day, often after rain, all the colonies of a region release them together: this is the nuptial flight, the "flying ant day" of European summers for Lasius niger. Synchronisation lets queens meet males of other colonies. Mating takes place in the air or on the ground; the males die soon after. The queen stores the sperm of one or several males in her spermatheca and uses it for the rest of her life, up to decades.
Then the young queen drops her wings and looks for a place. In the black garden ant she digs a chamber and founds the colony alone, closed in, without eating: she lives on her fat reserves and on the muscles of her wings, and raises her first, small workers. This claustral founding is a lottery: only a very small fraction of the queens survive to become a colony. Other species found differently: Argentine ants and many Formica split their nests ("budding"): a queen leaves on foot with workers, and the daughter colony stays friendly with its mother. Winged queens can also join an existing colony.
In Ants Life
Every 60 steps a colony tests whether it can send a princess. It needs a queen alive, swarmAnts = 36 ants, swarmStore = 40 crumbs in the store, a flood level below 20%, no rain, a cold below 0.5, a heat below 0.8 and a snow cover below 0.2, and fewer than maxColonies = 4 colonies in the yard. The first flight is possible swarmFirst = 6,000 steps (100 seconds) after the colony exists, then every swarmEvery = 9,000 steps (150 seconds). The flight costs swarmCost = 18 crumbs.
The site is a free spot within swarmNear = 560 pixels of the mother nest, at least swarmApart = 170 pixels from every nest and swarmRival = 260 pixels from the nests of the rival kind. The princess flies straight to it at swarmSpeed = 1.5 pixels per step (90 pixels per second), over rocks and fights. She lands and founds a colony of the same kind: 24 crumbs in the store, 5 eggs, a queen, no worker. If no site is found, the colony tries again 1,200 steps later. Princesses wait in the "Princess hall" of the nest in the picture.
What Ants Life simplifies
There are no males and no mating in the simulation (three small winged males circle the princess in the picture, but they are only a drawing). The flight is a single queen leaving a nest with a few workers' worth of food, and she cannot fail: no predator, no failed founding, no wing loss.
The new colony has 5 eggs from the start, which makes it closer to budding than to claustral founding, where the queen alone has to raise the first workers out of her own reserves.
The daughter is of the same kind as the mother and never fights it. In real life this is true for budding colonies and supercolonies, not for queens that found independently.
The game wants a calm dry day. Real flights prefer warm, humid, windless days, often right after a rain: here the rain blocks them.
A real colony needs several years before it produces its first winged queens. The game's mother needs 36 ants and 100 seconds.
Experiment: the weather that blocks the flight
Action
Play the yard at 4× and note the clock when the Journal says that a young queen takes off from your nest (the colony needs 36 ants first). Start a new game (R) and this time keep the weather bad: start the rain with 1 from about 90 seconds on, and each time a shower ends press 1 again to start the next one. Watch the Journal, then let the weather clear.
Explanation
The colony checks the weather every second. Rain blocks the flight, and so does a nest flooded above 20% (the water takes a while to drain), a cold above 0.5 (below about -13 degrees), a heat above 0.8 (above about 41 degrees) and a snow cover above 0.2.
Parameters
Key 1 (rain); the Journal (J) and the panel to count the ants and read the flood level.
Check
No princess leaves while it rains or while the nest is still wet, even though the colony has the ants and the food. Once the weather is calm and the water is down she leaves at the next test. A new colony appears in the panel after the landing, and the goal Found a new colony is reached when it holds 5 ants.
Common pitfall
Long chains of showers flood your own nest: stop in time if the water gets high (sheet 7). The rival colony also flies, on the same rules.
Food for thought
Why do all the colonies of a region release their princesses on the same day?
So that queens and males from different colonies meet in the air, which avoids breeding between sisters and brothers.
In the game, why is the daughter colony never an enemy of its mother? What does it look like in real life?
The game uses "kinds" as friend or enemy. A real daughter colony is friendly if it was budded off; a queen that founded alone starts a new, independent colony.
A real founding queen has no workers and no food coming in. What does the game give her that makes things easier?
Five eggs and 24 crumbs from the start, and a mother that paid 18 crumbs of her store.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
Keller, L., & Genoud, M. (1997). Extraordinary lifespans in ants: a test of evolutionary theories of ageing. Nature, 389, 958-960.
04 Building underground: nest architecture, digging and the earth
The nest viewCollective behavior
An ant nest is a building that no ant has drawn. Its plan comes out of thousands of small decisions.
The real mechanism
An ant nest is a network of chambers linked by galleries and shafts, with a few entrances. Scientists look at its shape by pouring molten aluminium, or dental plaster, into the nest and digging out the cast. Walter Tschinkel did this for the Florida harvester ant: the nests of this species go down several meters, with dozens to over a hundred chambers. Chambers are used for the brood, for food stores, for resting and for refuse; where they sit, how deep, and how many depends on the species, the soil, the age of the colony and the weather. In many species the brood is moved up and down following the temperature and the humidity of the soil.
The ants dig with their mandibles and carry the soil out in small pellets. Outside they drop it at the entrance, and the heaps become a mound or a crater with the shape typical of each species. A mound can warm up in the sun, and some species build them on purpose (thatch mounds of wood ants, earth mounds of the yellow meadow ant Lasius flavus, which can stand for decades). No ant has the plan. Experiments on Lasius niger study how digging speeds up with the number of workers and how the size of the nest ends up related to the size of the colony.
In Ants Life
Click a nest to see it in section (720 by 460 units): a main shaft, galleries and nine rooms, dug in a fixed order. Each room has a threshold of dug amount at: queen 0, nursery 0.5, granary 1.1, refuse 1.9, rest 3, princess hall 4.2, second nursery 5.6, second granary 7, second rest 8.4. The two starting colonies are at 3.2 (five rooms); a new colony starts at 0.
Digging is a picture of the colony: each ant (workers and soldiers) adds 0.00002 per step to the dug amount, so 30 ants add 0.0006 per step, one unit in about 28 seconds. It stops while the nest holds more than 25% water. Above 30% water the galleries fall in: the dug amount drops by collapse = 0.0006 times the water level per step, from the last room back, but not below collapseFloor = 1.05 (the shaft and the first chambers hold). The earthworm brings it down faster. The earth is shown on the ground: while the colony digs, a few workers carry pellets out to 7 heaps of fine earth around the door (up to 11 pixels each), and a mound grows over the shaft in the section. The yellow ant can help dig with E at the end of a gallery being dug (+0.1 per press, at most every 0.35 second).
What Ants Life simplifies
The plan is the same for every nest: nine fixed rooms in a fixed order. A real nest takes the shape of its soil, its species and its history, and can be a hundred times bigger.
The ants that walk in the section are a picture. The ants of the yard do not really go underground: they disappear at the door.
The speed of digging depends only on the number of workers. In reality soil moisture, soil type, temperature and the crowding of diggers all matter.
Rooms have a use in the picture (the queen lays in the queen's room), but the simulation does not move eggs or food between rooms.
Experiment: rooms, workers and the flood
Action
Start a game, open your nest (click it) and count the rooms: five. Close it, run at 8× and open it again every half minute: the sixth room (the princess hall) appears at 4.2. Then start rain with 1, keep the section open, and watch the water line rise and then the galleries fall in. Stop the rain with 1 and wait until the water has drained.
Explanation
The dug amount grows with the number of ants; the water stops it above 25% and brings the galleries down above 30%, from the deepest.
Parameters
The number of ants (the panel), the rain key 1, the speed of time.
Check
The sixth room needs 1.0 more than the start (3.2 to 4.2). With the 15 ants of the start that takes about 56 seconds, with 30 ants about 28, with 60 about 14. In the rain, a nest at full exposure and full rain strength gains water at 0.0007 per step: it passes 30% after 430 steps (7 seconds), and more slowly otherwise. The collapse then costs about 0.00036 per step at 60% water, that is a unit in 46 seconds, and the digging starts again below 25%.
Common pitfall
Each nest has its own exposure (0.3 to 1), so a nest high up floods slowly or not at all. Compare the two nests.
Food for thought
Why do real ants drop the soil away from the entrance, spread around, instead of in a heap right at the door?
A big heap would block the entrance and slow traffic; spreading the pellets keeps the door clear. Different species have different dropping habits, which is why mounds and craters look so different.
If thirty ants dig one unit in 28 seconds, how long do sixty take?
About 14 seconds: in the game the rate is proportional to the number of ants.
Why is it useful for a nest to be deep, and what does it cost?
Deep soil is steadier in temperature and humidity, so safer against frost, heat and drought; but it is longer to dig and to walk, and the deepest chambers flood first.
Tschinkel, W. R. (2004). The nest architecture of the Florida harvester ant, Pogonomyrmex badius. Journal of Insect Science, 4, 21.
Rasse, P., & Deneubourg, J.-L. (2001). Dynamics of nest excavation and nest size regulation of Lasius niger (Hymenoptera: Formicidae). Journal of Insect Behavior, 14, 433-449.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
An ant carries two stomachs: one for itself, and one for the others.
The real mechanism
Many ants have a crop, an expandable part of the gut that serves as a "social stomach": a forager fills it with liquid food (nectar, honeydew, the juices of prey), goes back to the nest and passes drops from mouth to mouth to nestmates. This trophallaxis feeds the whole colony, including the queen, the larvae and the ants that never leave the nest. The adult worker's gut cannot take solid particles (a filter, the infrabuccal pocket, spits them out as a pellet); in many species, solid food such as pieces of insect is given to the larvae, which digest it and give liquid back.
Experiments with radioactive or dyed food show how it spreads: in Lasius niger, the food of a few foragers reaches most of the colony within a few days, in a chain of exchanges, with a handful of ants receiving much more than the others. Trophallaxis also carries more than food: the colony odor, hormones and proteins pass in the drops, as shown for carpenter ants. A colony can therefore buffer its needs: in honeypot ants, living "repletes" hang from the ceiling with abdomens swollen with liquid, like barrels. Ant colonies also steer their intake of nutrients as a group: they choose their food to balance protein and carbohydrate.
In Ants Life
Each colony has one store, in crumbs (60 at the start, 24 for a colony founded since). An ant that brings a crumb home adds 1 to it; a plugged door takes nothing in. The store is shared by everybody: every ant eats eatEvery = 3,750 steps' worth, that is one crumb per 62.5 seconds, less in winter (divided by 1 + 1.5 × the hibernation). The queen needs the store to hold eggCost = 4 crumbs plus a reserve of 6 + 0.5 per ant (reserve) to lay. When the store is empty everyone loses 0.0026 hit point per step: a worker lasts about 38 seconds, a soldier 77. A flood above 30% spoils 0.004 per step, the earthworm 0.015. A dead ant leaves 1 or 2 crumbs.
What Ants Life simplifies
There is no crop, no exchange between ants and no larva: the store is a single number that everyone reaches instantly, and one crumb is the same for all.
Hunger is the same for the whole colony. In reality some ants are full and others empty, and who eats first is a question in itself (the foragers, the queen, the larvae).
One kind of food only. Real colonies need sugars (honeydew from aphids, nectar) and proteins (insects), and no aphid farming exists here.
No transfer of odor or hormones through food.
Experiment: how many ants can a store feed?
Action
In the yard, watch the row of the black colony in the panel (workers, soldiers, food, eggs). Wait for the depth of winter (about 8 minutes in, longer with a cold snap on 4), when the ants sleep in the nest and nobody forages. Note the number of ants and the food, wait one game minute, and note them again. Compare the loss with a calculation.
Explanation
Every ant consumes the same, so the store falls by (number of ants) / 3,750 per step when nothing comes in.
Parameters
The number of ants, the season (the clock of the panel), the speed of time.
Check
30 ants eat 0.48 crumb per second, about 29 per game minute, and 130 ants 2.1 per second. In the deepest sleep the consumption is divided by 2.5, so 30 ants eat about 11.5 per minute. To lay, a colony of 100 ants needs at least 4 + 6 + 50 = 60 crumbs in store: if the food row stays under that, the eggs row stops growing.
Common pitfall
Outside winter, foragers keep bringing crumbs in, so you would measure the balance between intake and consumption, not the consumption alone. Ants carrying a crumb still finish their trip home.
Food for thought
What happens to a colony whose door stays plugged, even if there is plenty of food outside?
Nothing enters the store while everybody keeps eating from it: it empties, laying stops once it is under the reserve, nothing hatches out, and in the end the ants starve.
Why can the queen of a real colony, who never leaves the nest, eat?
Because workers feed her mouth to mouth: trophallaxis.
What does a colony gain from a "social stomach" compared with each ant keeping what it finds?
A shared buffer: food is spread to the ants that did not find any, and the colony survives a day with no foraging.
Buffin, A., Denis, D., Van Simaeys, G., Goldman, S., Deneubourg, J.-L., & Pasteels, J. M. (2009). Feeding and stocking up: radio-labelled food reveals exchange patterns in ants. PLoS ONE, 4(6), e5919.
LeBoeuf, A. C., et al. (2016). Oral transfer of chemical cues, growth proteins and hormones in social insects. eLife, 5, e20375.
Dussutour, A., & Simpson, S. J. (2009). Communal nutrition in ants. Current Biology, 19, 740-744.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
Ants find the shorter of two routes, though none of them compares them.
The real mechanism
In 1989 Simon Goss, Serge Aron, Jean-Louis Deneubourg and Jacques Pasteels ran a now-classic experiment. A nest of Argentine ants (Linepithema humile) was joined to a food source by a bridge that forked into two branches, one twice as long as the other. In the majority of trials, nearly all the ants (80 to 100%) ended up on the short branch. When the two branches had the same length, ants also ended up on one of them, picked at random, with a very strong preference (Deneubourg et al., 1990).
No ant measures lengths. At the fork, ants first choose almost at random. Those on the short branch reach the food and come back sooner, so the short branch is marked earlier and more often, attracts more ants and gets marked again: the positive feedback of sheet 1 amplifies a small initial difference. A model fitted to the data says that the chance to choose a branch is (k + A)^n / ((k + A)^n + (k + B)^n), where A and B are the amounts of pheromone and n is about 2. The system has a weak point: if the first path found is a poor one, the colony can stay locked on it; Goss et al. saw this when a shorter branch was opened only after the long one had been adopted.
In Ants Life
In the double bridge scenario, one black colony has no queen and no hunger: 30 ants that never die, their nest on the left, a food chamber on the right (160 crumbs, refilled by 100 each time fewer than 60 remain, checked every 470 steps). Between them is a fork: a short branch, nearly straight, and a long one that goes down and back up. The tunnels are 20 pixels in radius, dug in solid earth. The panel counts the ants that are between the fork and the junction (50 pixels in from each) and splits them above and below the middle of the two branches: this is the traffic, in percent. There is no weather and no creature.
The ants follow the rules of sheet 1 unchanged: they mark the way home from the first crumb, a mark evaporates by 0.9962 per step, a forager steers towards the strongest of three readings. Nobody is told which branch is shorter.
What Ants Life simplifies
The tunnels have walls the ants follow, and the long branch is only about 1.5 to 2 times longer than the short one (the exact ratio depends on the size of the window). The original ratio was 2.
The ants lay only on the way home, whereas Argentine ants mark in both directions, so the feedback is slower.
30 ants only, which never die. The original experiments used several hundred, and a population effect appears (too few ants, no collective choice).
Nothing is measured except the percentage on each branch: no count of choices at the fork.
Experiment: watch a choice emerge
Action
Menu, then The double bridge. Press T to show the trails. Run at 4× for two game minutes and note the Traffic line (short way %, long way %) and the food brought. Press R and repeat five times.
Explanation
At first the ants split on both branches. Those that took the short one are back sooner, and their trail is renewed more often.
Parameters
The speed of time, the display of the trails, the number of runs.
Check
In most runs the short branch ends up with most of the traffic; it does not reach 100% because ants keep exploring. A few runs stay on the long branch longer, because the choice is random at the start: this is the sensitivity to the first marks. The trail on the short branch is brighter.
Common pitfall
The traffic shown at the start is meaningless (a few ants). Wait for the first crumbs to be carried home. The weather keys do nothing in this scenario.
Food for thought
If the two branches were the same length, what would you expect in the percentages?
Most of the traffic on one of the two, chosen at random from one run to the next, not 50/50 (Deneubourg et al., 1990).
What would change if the trails did not evaporate at all?
The mark on the long branch would stay for ever, and the choice would depend even more on the first crumbs than on the length.
Is the ants' solution always the best one? When does it fail?
No: it can lock on the first route found, even a long one, and does not see a shorter route that nobody has explored.
Goss, S., Aron, S., Deneubourg, J.-L., & Pasteels, J. M. (1989). Self-organized shortcuts in the Argentine ant. Naturwissenschaften, 76, 579-581.
Deneubourg, J.-L., Aron, S., Goss, S., & Pasteels, J. M. (1990). The self-organizing exploratory pattern of the Argentine ant. Journal of Insect Behavior, 3, 159-168.
Beckers, R., Deneubourg, J.-L., & Goss, S. (1992). Trails and U-turns in the selection of a path by the ant Lasius niger. Journal of Theoretical Biology, 159, 397-415.
An ant is a small cold-blooded animal living in the ground: the sky decides how fast it goes, and whether its house stays dry.
The real mechanism
Ants are ectotherms: their speed, their metabolism and the development of their brood follow the temperature. Within a range, they walk faster and develop faster when it is warm, and they stop foraging when it is too cold or too hot. The Saharan silver ant (Cataglyphis bombycina) forages on the sand at the hottest hours of the day, when its predators hide, on a "thermal tightrope": it makes short trips and must return to cool off. Pheromones are volatile molecules, so heat makes them evaporate faster. Many ants move their brood up and down in the nest to follow the temperature.
Rain washes the trails and makes foraging dangerous. A downpour floods the lowest chambers first. Ants fight water in several ways: harvester ants close their entrances with soil or pebbles, and fire ants, when the water rises, link their bodies into a floating raft that can hold tens of thousands of ants, and which stays dry thanks to the air trapped by the hairs of their cuticle (Mlot, Tovey and Hu, 2011). A layer of snow is also an insulator: the ground under it stays close to freezing while the air is much colder. Wind dries the soil, blows leaves and sand over the entrances and moves the smells that ants follow.
In Ants Life
Rain lasts 1,200 to 2,400 steps (20 to 40 seconds). It multiplies trails by rainWash = 0.985 per step, slows ants by rainSlow = 0.75, and raises each nest's water level by floodRate = 0.0007 times the rain strength (0.5 to 1) and the exposure of the nest (0.3 to 1) per step. The water drains by floodDrain = 0.0002 per step (2.5 times faster in the heat). Above 30% the stores spoil (0.004 per step); above floodEggs = 45% an egg is carried away every 40 steps; above floodDrown = 60% the ants near the nest lose 0.0035 hit point per step; above 90% the queen loses 0.0005 per step (drowned in about 33 seconds). The yellow ant can carry the eggs up the shaft with E (safe for 900 steps).
Heat and cold move the temperature from -1 (-40 degrees) to 1 (+48 degrees), at 0.0008 per step. At full heat ants walk 10% faster, trails keep 0.0032 less per step, nests drain 2.5 times faster. At full cold ants walk 25% slower, trails keep 0.0018 more per step, and an egg takes 35% longer to hatch (hatchCold). Snow needs the temperature under -0.2 (about 3 degrees): the cover rises by 0.00028 per step (about a minute from bare ground to full cover), covers trails (snowWash = 0.008 times the cover) and slows ants by up to 15%. Wind blows up to 26 leaves; one that passes within a door radius plugs it with a chance of 0.004 per step. A plug above 50% stops everything: nothing goes in, nothing hatches out. Workers at the door and the yellow ant (much faster) clear it.
What Ants Life simplifies
Water rises by one number per nest, with no soil, no drainage and no depth: in the picture the water line simply goes up from the bottom. Real chambers flood depending on the soil, the slope and the slope of the nest tunnels.
Ants do not react to the weather before it comes. Real colonies close their doors and stop foraging when rain is coming.
No rafts, no waterproofing, no plugging of the door by the ants themselves.
Heat and cold only change speed, trails and eggs. A real ant has a critical temperature above which it dies in minutes.
Snow insulating the ground is not modelled: it only covers trails and slows ants.
The puddles that gather in the low places while it rains are only a picture. A real puddle is an obstacle to cross or a place to drink; here nothing changes for the ants.
Experiment: the flood and the queen
Action
In the yard, click on your nest to open the section. Start the rain with 1 and watch the blue percentage in the panel and the water line. As the water reaches the nursery, walk the yellow ant to it with the arrows and press E. Stop the rain with 1 after about 30 seconds.
Explanation
The level rises by rain strength times exposure times 0.0007 per step. The queen only suffers above 90%.
Parameters
Duration of the rain (the key), the nest compared (your nest or the red one: their exposures differ), the use of E in the nursery.
Check
In full rain on an exposed nest, 30 seconds (1,800 steps) add up to 1.26, so the level hits the maximum of 1 well before the end: the stores spoil from 30%, eggs go from 45%, ants near the door drown from 60%. With a low exposure of 0.3 it only reaches 38%. Pressing E in the nursery shields the eggs for 15 seconds and counts toward Save 5 eggs from the water.
Common pitfall
The yellow ant must be at the door of her own nest when you click it, or she does not enter. Keep the rain on for a long time and the queen drowns too: Keep the queen alive through a flood means to stop it in time.
Food for thought
Why does the game flood the lowest nest first?
Because water runs downhill: low ground and deep chambers fill first. The nests have an "exposure" between 0.3 and 1 that stands for that.
Why is a snowfall in the game bad for foragers, but a good thing for a real nest?
In the game it covers trails and slows ants. In real life a snow layer is an insulating blanket: the soil under it stays warmer than the air.
Fire ants survive floods by forming rafts. What does this tell about the colony as an animal?
That it acts as a single body: thousands of ants pull themselves into one waterproof structure without a leader.
Mlot, N. J., Tovey, C. A., & Hu, D. L. (2011). Fire ants self-assemble into waterproof rafts to survive floods. PNAS, 108(19), 7669-7673.
Wehner, R., Marsh, A. C., & Wehner, S. (1992). Desert ants on a thermal tightrope. Nature, 357, 586-587.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
In a temperate winter, an ant colony does not die and does not live: it waits.
The real mechanism
In temperate regions, an ant colony has a yearly rhythm. In spring the nests wake up and the brood starts to grow; in summer the colony is at its peak and rears winged queens and males; in autumn the ants fill their reserves; in winter they enter diapause, a state of rest triggered by the shortening of the days and by the cold, in which the queen stops laying, the workers cluster deep in the nest and metabolism drops. Soil temperature changes much less than air temperature as one goes down, so a nest deep enough stays above the worst frost. In several temperate species, a cold rest is even needed before the queens can lay again in spring and before the colony can rear new queens.
Hibernation is not freezing: ants that stay out in a hard frost, far from the nest, die. Winter also stops the other animals (spiders, woodlice and so on), and the plants go dormant, which changes the food available in the next spring.
In Ants Life
A year is four seasons of seasonLength = 12,000 steps (200 seconds): 13 minutes and 20 seconds, and the yard begins in the middle of spring. The temperature of the season is -0.02 + 0.5 × cos of the phase of the year, peaking at the middle of summer: about 14 degrees in spring and autumn, 30 in summer, -14 in the depth of winter. Spells ride on it: heat, cold or snow come by themselves between 14,000 and 30,000 steps apart, with weights per season (for example, in winter 5% heat, 35% cold, 60% snow).
The hibernation level goes from 0 to 1 as the temperature falls below -0.25 (0.5 degree) to -0.6 (-18 degrees). Above 0.3 the queen stops laying (below about -5 degrees) and the woodlouse, spider, beetle and worm stay away. Above 0.6 (about -11 degrees) the ants that carry nothing go home and stay there. Everybody eats less (divided by 1 + 1.5 × the level). Above 0.85, an ant farther than 32 pixels from its nest loses 0.003 hit point per step (about 33 seconds for a worker). Without any spell, an ordinary year puts the ants to sleep for about 126 seconds of its 800 and keeps the queen from laying for about 213 seconds. In autumn a leaf falls every 45 steps.
What Ants Life simplifies
The trigger of rest is the temperature alone. Real diapause is triggered by the length of the day and by temperature together, in advance of winter.
The year lasts 13 minutes. A season is a few minutes, so a queen that skips 213 seconds of laying has not lost a season but a fifth of her year.
No depth: in reality a nest escapes the frost by going down, here an ant is either at home or out. The shaft depth does not protect anyone.
No fat reserve and no stored food for winter: the store is only the number of crumbs, eaten more slowly.
Plants and creatures react to the season without a cause in the simulation (they are timed effects, not populations).
Experiment: a winter in the yard
Action
Start the yard, set the speed to 8× and watch the clock of the panel: the year runs from Spring, year 1. Autumn begins 5 minutes in, winter at 8:20 and the second spring at 11:40. In autumn look at the stores. During the winter watch the ants go home, then note the number of eggs. Come back to the spring.
Explanation
The hibernation level follows the temperature. The sleeping ants return to the nest, the queen stops laying, and consumption falls.
Parameters
The speed of time, the weather keys (4 to bring the cold forward, 2 for snow), the clock in the panel.
Check
The seasons change every 200 seconds. The ants are asleep when the temperature is under about -11 degrees: a bit more than two minutes of an ordinary winter, and more with a cold snap (4) or a snowfall. The goal Get through a winter is reached when spring comes with a living queen.
Common pitfall
Ants left outside, far from the nest, freeze in deep cold; bring your workers home before the winter. Creatures do not come in the cold: it is the quietest season.
Food for thought
How long is a year in the game, and what is it in seconds of game time?
13 minutes and 20 seconds, that is 800 seconds (4 × 200).
Why does the game forbid new eggs in the deep cold?
Because real queens stop laying in diapause, and a nest that is asleep has nobody to look after a brood.
Why does a real nest benefit from being deep in winter, and from being shallow in spring?
Deep soil is steadier and avoids frost; shallow soil warms up first in spring, which helps the brood to develop.
Kipyatkov, V. E. (2001). Seasonal life cycles and the forms of dormancy in ants (Hymenoptera: Formicoidea). Acta Societatis Zoologicae Bohemicae, 65, 211-238.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
09 Spider, woodlouse, mower, worm, beetle, fly: the dangers of the yard
The yardEcology
Ants are small, many and tasty, and they live among people and other animals who do not care about them.
The real mechanism
Ants are both predators and prey. Spiders hunt them: a few specialise in ants (the Zodariidae), others mimic them to be left alone, such as the ant-mimicking jumping spiders. But most spiders avoid ants, which bite, sting and spray acid, and a swarm of ants can overwhelm a spider. Woodlice are not insects but crustaceans that breathe on land and need damp places. Some roll into a ball when threatened; one species, Platyarthrus hoffmannseggi, lives only inside ant nests, where it eats refuse. Earthworms are ecosystem engineers: as Charles Darwin showed in 1881, they turn over the soil, just as ants do, and the two share the ground with little harm to each other. Dung beetles roll balls of dung and compete with other insects, among them ants, for this food; they steer in a straight line by the sky, even by the Milky Way at night (Dacke et al., 2013).
The fly is food. Ants eat dead insects, which are a main source of protein, and cut up what they cannot carry. Live flies are another story: parasitic phorid flies (Pseudacteon) lay an egg in a fire ant, the larva grows in its head, and the ant's head falls off; ants freeze when they see the fly, which is used as a biological control. The mower has no natural equivalent: nests are deep, so a mower mostly hits the foragers on the surface and the plants; the worst human dangers for ants are concreting, tillage and pesticides.
In Ants Life
All the creatures appear only in the yard, in a random time drawn in a range, and none comes in deep cold (hibernation above 0.3). The woodlouse (first at 4,000 to 9,000 steps, then every 10,000 to 22,000) walks at 0.38, has 14 hit points, curls up for 90 steps when bitten, and gives a feast of 60 crumbs. The spider (first at 14,000 to 24,000 steps, then every 40,000 to 70,000, staying 4,000 to 7,000 steps) walks at 0.7, has 80 hit points, and kills the nearest ant within 11 pixels, of any kind, every 45 steps; soldiers charge it from 140 pixels, and the reward is 150 crumbs. When at least mobQuorum = 3 soldiers of a kind are within 140 pixels of it, the free workers of that kind (those that carry nothing) within mobReach = 200 pixels also turn towards it, by 0.2 radian per step. The dung beetle (first at 12,000 to 20,000, then 30,000 to 50,000) walks at 0.42, has 45 hit points, takes two crumbs every 20 steps into its ball and leaves with 40; if it falls, the ball and 60 crumbs become food. Hit points lost per step in a bite: woodlouse 0.05 (soldier), 0.02 (worker), 0.08 (yellow ant); spider 0.1, 0.04, 0.08; beetle 0.04, 0.012, 0.05.
The lawnmower (first at 20,000 to 36,000 steps, then every 60,000 to 100,000, never in rain or snow) gives a noise, then 240 steps (4 seconds) later crosses the yard at 3 pixels per step along one row (one or two passes), killing every ant within 38 pixels of its line except those within 0.7 of a nest radius, who dive in; it kills a woodlouse and a beetle and hurts a spider by 3 per step. The earthworm (first at 30,000 to 50,000, then 80,000 to 140,000) passes under one living nest for 900 steps (15 seconds): the store loses 0.015 per step, an egg is lost every 60 steps, galleries come down, and during the middle of its way (35% to 65%) ants within 30 pixels of the nest lose 0.02 per step and the queen 0.0003. The dead fly is a pile of 140 crumbs.
What Ants Life simplifies
The spider is a stylised "apex predator" that wins against one ant at a time. A real spider hunting ants is much more cautious, and a colony swarming a spider is more usual than the reverse. The game keeps only a trace of the swarm: from 3 soldiers close to the spider, the workers within 200 pixels join in. The house has a smaller spider, tied to its web (sheet 14).
Woodlice, beetles and worms are neighbours, not usual enemies of ants. The game turns them into events to play with.
The mower is a human danger put in as a game event. It kills only surface ants.
The fly is only a corpse. There are no living flies, no parasitoid flies, no antlion pits, no birds.
The creatures come at random times, from the side of the yard, with no ecology behind them.
Experiment: how much does it cost to bring down a spider?
Action
Play the yard at 4× and wait for the toast "A spider stalks into the yard" (between about 4 and 7 minutes). Hover the spider to read its strength, and pause (Space) to look. Keep your soldiers near it (E or the order Bring the colony here) and count the ants lost, which are given in the Journal.
Explanation
The spider has 80 hit points and loses 0.1 per soldier per step in reach (0.04 for a worker). It kills one ant per 45 steps.
Parameters
The share of soldiers (the slider, before the spider comes), the order Bring the colony here (14 ants at most), the speed of time.
Check
80 / 0.1 = 800 soldier-steps, about 13 seconds of one soldier: ten soldiers in reach need 1.3 second, but the spider kills one ant every 0.75 second, so the colony loses a couple of ants. With workers only (0.04) it takes 2,000 worker-steps, 2.5 times more.
Common pitfall
The spider does not come in rain or in the cold, and the yellow ant is also a target (she hits for 0.08). Winning gives a feast of 150 crumbs and the goal Bring down the spider.
Food for thought
How long does a single worker need to bring down a woodlouse of 14 hit points, at 0.02 per step?
14 / 0.02 = 700 steps, about 11.7 seconds; a soldier (0.05) needs 280 steps, 4.7 seconds.
What does the woodlouse that lives in ant nests teach us about the idea of "enemy"?
That an animal that looks like prey can be tolerated, even useful, if it smells right; the relation depends on the chemistry, not on the shape.
Why do dung beetles roll their ball in a straight line?
To get away from the dung pile quickly, where other beetles and ants would steal the ball; they hold the line using the sun, the moon or the Milky Way.
Darwin, C. (1881). The Formation of Vegetable Mould through the Action of Worms, with Observations on their Habits. John Murray.
Dacke, M., Baird, E., Byrne, M., Scholtz, C. H., & Warrant, E. J. (2013). Dung beetles use the Milky Way for orientation. Current Biology, 23, 298-300.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
Ants wage the most organised wars of the animal world, and numbers often count more than strength.
The real mechanism
Ant colonies compete for food and for ground, and neighbouring colonies of the same or of different species fight at their borders. An ant knows a nestmate by its smell, a blend of hydrocarbons on the cuticle that the colony shares: a stranger is attacked. Some species fight to kill with bites, stings or sprays of formic acid, others with ritual tournaments, where rivals stand and drum on each other without hurting (honeypot ants, Myrmecocystus). Some species raid others for their brood (slave-making ants such as Polyergus) and some, like Argentine ants, form huge cooperative supercolonies: in Europe a single one reaches along the Mediterranean coast for about 6,000 kilometres (Giraud et al., 2002).
Who wins a battle of ants? The British engineer Frederick Lanchester modelled battles in 1916: when all the fighters can engage the enemy at once, the strength of an army grows like the square of its number (the "square law"): every extra fighter both kills more and is shot at less. This was tested on ants: in fire ants, the mortalities observed fit the square law (Plowes and Adams, 2005). Quality counts too: a soldier is worth more than a worker only in species that have real soldiers.
In Ants Life
Only the two kinds fight, black against red (colonies of the same kind never do). Two ants of opposite kinds fight when each is within 7 pixels of the other on both axes. Each step of contact, the black ant loses 0.077 against a red soldier and 0.019 against a red worker; the red ant loses 0.077 against a black soldier, 0.019 against a black worker, 0.064 against the yellow ant. A worker has 6 hit points, a soldier 12: a soldier kills a worker in about 78 steps (1.3 seconds) and a soldier in 156 steps (2.6 seconds); a worker needs 316 steps for a worker. Soldiers keep within 110 pixels of their nest, charge any enemy within 80 pixels, and the red colony starts with 25% soldiers against 20% for the black. The yellow ant also bites when the player holds F: 1.1 hit points to the nearest enemy within 24 pixels, at most once every 24 steps (1.1 / 24 = 0.046 per step on average), multiplied by 0.4 + 0.6 × strength. Her strength, from 0 to 1, falls by 0.08 per bite and rises by 0.2 per crumb she picks up; the same factor scales her passive blow of 0.064 per step.
When a colony has no ant and no egg left, it is dead. When the last colony of a kind is gone, a message says so: "The red colony is gone: the yard is yours" or "Your colony has fallen".
What Ants Life simplifies
The enemy is the other colour, not the other smell. There is no recognition, no mistakes, no tolerance of neighbours (real "dear enemy" effect), and no peace between kinds.
No chemical weapons, no stings, no ritual fights, no raid, no flight: ants fight to the death where they meet.
Only two kinds exist, and a daughter colony is always a friend. Real supercolonies are also an exception, not the rule.
Fights are decided by contact, not by groups: Lanchester's law is not built in. If the square law appears, it comes from many pairs fighting at once.
Almost no tactics: soldiers do not call for help, workers do not flee. The one decision taken as a group is against the spider (sheet 9). The strength of the yellow ant, her bite and her healing by the plants are inventions of the game.
Experiment: numbers against quality
Action
In the yard, wait until red ants forage near a pile of crumbs. Send the yellow ant there with a click and read the report: it tells how many enemies and soldiers are near. Choose Bring the colony here and watch the fight at 1/2× speed. Next time, do not call the colony and compare.
Explanation
Each pair in contact fights independently; the side with more ants in the area has more pairs and wins sooner.
Parameters
The order Bring the colony here (14 ants at most, within 260 pixels), the share of soldiers (slider), the speed of time.
Check
In the code, a red ant touched by three black ants takes the damage of all three in the same step. The more ants you bring to one spot, the more of them touch each enemy at once, and the faster the enemy falls: this is the mechanism behind the square law. Soldiers make each contact about four times as damaging.
Common pitfall
Calling ants far from the nest also pulls them from foraging, and the store may fall. The report only counts what is within a short radius.
Food for thought
By Lanchester's square law, how much stronger is an army of 20 ants compared with one of 10, of equal quality?
4 times (20² / 10²).
Why do the game's red and black ants fight at sight, while real neighbouring colonies often tolerate each other?
The game uses colour as the only signal. Real ants use smell and learn the smell of neighbours, which can lead to a "dear enemy" truce.
A soldier is four times as strong in a bite as a worker. Is one soldier worth four workers?
Four workers in contact deal 4 × 0.019 = 0.076 per step, as much as one soldier (0.077). But the soldier has twice the hit points (12 against 6), and each worker has to be reached to be hurt, so in a mass fight the soldier is worth a bit more than four workers.
Franks, N. R., & Partridge, L. W. (1993). Lanchester battles and the evolution of combat in ants. Animal Behaviour, 45, 197-199.
Plowes, N. J. R., & Adams, E. S. (2005). An empirical test of Lanchester's square law: mortality during battles of the fire ant Solenopsis invicta. Proceedings of the Royal Society B, 272, 1809-1814.
Giraud, T., Pedersen, J. S., & Keller, L. (2002). Evolution of supercolonies: the Argentine ants of southern Europe. PNAS, 99, 6075-6079.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
An ant that found something cannot give an order. It can only convince the others to follow.
The real mechanism
When a scout finds food or a new nest site, it must bring other ants to it. Species do it in different ways. In tandem running, as in Temnothorax, a leader walks to the target with one follower that taps its abdomen; the follower learns the way, and the pair slows down when the follower lags: Franks and Richardson (2006) showed this is a form of teaching. In group recruitment a small band follows; in mass recruitment, as in black garden ants, the scout lays a trail (sheet 1) and the crowd comes. For the choice of a new nest, Temnothorax scouts use a quorum: they switch from slow tandem runs to quick carrying when enough of them are at the same site.
To find the way back, many ants do not rely on trails alone. Desert ants (Cataglyphis) keep a running total of their path with a sun compass and a count of their steps (their path integration), which brings them back in a straight line after a long and winding search; Wittlinger et al. (2006) showed this by changing the length of their legs with stilts and stumps. Finally, no ant is in command. Experiments on harvester ants show that foragers go out depending on how often returning foragers come back with food: the rate of encounters is the information (Greene and Gordon, 2007).
In Ants Life
The yellow ant is a worker (hit points 6, speed 1.1) that never ages. A click sends her to the target: once she is within 16 pixels, the report is opened (and it times out after 3,600 steps, a minute). The orders in the report act on the colony's free workers (not carrying anything, not soldiers): Call the others takes up to 10 within 110 pixels, Bring the colony here up to 14 within 260 pixels. They follow her for 1,125 steps (18.75 seconds): they turn towards her by 0.19 radian per step, stop closing in at 18 pixels, and wander a little (noise 0.24). A follower drops the order as soon as it finds a crumb. Mark a trail home lays a trail of strength markTrail = 12 (the maximum is 14) in a straight line from the spot to the nest, and calls the workers: up to trailCallMax = 12 free workers within trailCallRadius = 320 pixels of its start turn to it at once; Clear the leaves removes the leaves that landed within 36 pixels and pushes 0.6 of a plug. On the way home an ant keeps its direction to the nest at 0.077 radian per step and follows its trail by the same amount.
What Ants Life simplifies
The yellow ant can do things that no real ant can: she sees her colony's workers and addresses them, and she knows exactly where her nest is. The orders are the player's wishes, shaped as the abilities of one ant.
The recruitment is not an interaction with a body contact and a smell; it is a flag on each ant. The order itself is what a scout would do over a longer time.
The trail marked by the order has the same strength all along: it shows where the path is but gives no direction. A real return trail is stronger near the food.
There is no tandem running, no quorum and no teaching: ants do not learn the way, they follow a slope.
Soldiers cannot be called; they follow only the rules of their caste.
Experiment: does calling the others pay?
Action
In the yard, find a pile of crumbs far from your nest with no trail leading to it (turn the trails on with T). Click the pile: the yellow ant goes, looks, and reports. Choose Call the others and carry on beside the pile. Then R and do the same without calling, or with Mark a trail home. Follow one of the ants that came (click it).
Explanation
Followers stay with the yellow ant for about 19 seconds and leave the order when they find a crumb. They then carry it home along a trail they lay.
Parameters
The three orders, the distance of the pile, and the speed of time. The goals Send the yellow ant to look around 3 times and Give 3 orders to the colony count your use.
Check
With the call, up to 10 workers reach the pile at once and a trail forms sooner. Without it, the first trail comes from chance alone. A trail marked by the order starts at strength 12, which disappears in about 24 seconds if no ant renews it (0.9962 per step), so it is a start, not a road.
Common pitfall
If she is taken in hand with the arrows before arriving, or if the game is paused, there is no report. Free workers can be few if most are already carrying.
Food for thought
Why is it exact to say that the player does not command the colony, but gives suggestions?
The followers drop the order after 19 seconds, or at the first crumb, and the colony's real behavior comes from the trails; like a scout, the yellow ant can only attract.
A mark of strength 12 evaporates at 0.9962 per step. How long before it is under the level of 0.05?
ln(0.05 / 12) / ln(0.9962) is about 1,440 steps, 24 seconds.
Which is better for finding food: a colony where an ant gives orders, or one where ants follow trails?
No ant has the whole map in a real colony, so a leader would decide with less information than the sum of the ants' local knowledge; trails collect it for free and adapt as it changes.
Franks, N. R., & Richardson, T. (2006). Teaching in tandem-running ants. Nature, 439, 153.
Wittlinger, M., Wehner, R., & Wolf, H. (2006). The ant odometer: stepping on stilts and stumps. Science, 312, 1965-1967.
Greene, M. J., & Gordon, D. M. (2007). Interaction rate informs harvester ant task decisions. Behavioral Ecology, 18, 451-455.
Pratt, S. C., Mallon, E. B., Sumpter, D. J. T., & Franks, N. R. (2002). Quorum sensing, recruitment, and collective decision-making during colony emigration by the ant Leptothorax albipennis. Behavioral Ecology and Sociobiology, 52, 117-127.
12 Scale: real numbers of ants, lifespan of queens, biomass
The whole gameOrders of magnitude
A yard of 130 ants is a drawing of a society. The real ones count in thousands, millions, and sometimes more.
The real mechanism
There are about 15,000 described species of ants, on every continent except Antarctica. The size of a colony runs over several orders of magnitude: a few dozen workers in a Temnothorax nest in an acorn; a few thousand in the black garden ant; hundreds of thousands in a wood ant mound; one to several million in a mature leaf-cutter ant colony; tens of millions in the driver ants of Africa. The record is held by supercolonies: on the coast of Ishikari in Japan, a colony of Formica yessensis was estimated at 306 million workers and 1 million queens in about 45,000 nests (Higashi and Yamauchi, 1979).
Queens are among the longest-lived insects: a queen of the black garden ant has lived 28 years in a laboratory (Keller and Genoud, 1997), while workers of most species live between a few weeks and a few years. Added up, ants are everywhere: a 2022 estimate puts the number of ants on Earth at about 20 quadrillion (2 × 10^16), with a dry carbon biomass of about 12 megatonnes, more than all the wild birds and wild mammals together, and about a fifth of the biomass of humans (Schultheiss et al., 2022). An ant brain has a few hundred thousand neurons. The colony, not the ant, is what is intelligent.
In Ants Life
A colony is capped at maxAnts = 130 (ants and eggs together). The yard holds at most maxColonies = 4 colonies, so at most about 520 ants. A game ant lives 14,000 to 25,000 steps, that is 3.9 to 6.9 minutes, against a year of 13 minutes 20 seconds: about 0.3 to 0.5 of a game year. An egg hatches in 560 steps (9.3 seconds). The queen does not age: she dies only in a flood (above 90% for about 33 seconds), under an earthworm, or when her colony is destroyed. A colony can send a princess from 36 ants. The yard begins with 15 ants in each colony; the double bridge has 30 immortal ants.
What Ants Life simplifies
Population: 130 at most, against thousands. Crowding effects (traffic jams, trail saturation) do not appear at that size.
Time: a year of 13 minutes is a compression of about 39,000 times. An egg that hatches in 9 seconds corresponds, at the same ratio, to about 4 days, against several weeks in reality. A worker lives about 5 minutes, which is about 4 months of the compressed year, within the real range.
Queen: she has no age. A real queen is a long-term investment, and her death means the end of the colony after some months, as the workers die one by one.
A colony takes years to mature in nature. In the game, a mother can send a princess after 100 seconds.
One species, two colours: ants are shown as black and red, without the diversity of the 15,000 species.
Experiment: how long to reach the cap?
Action
Start a game and run it at 8×. Read the row of the black colony in the panel at intervals of 30 game seconds: note the sum of workers and soldiers, and the eggs. Stop when the sum with the eggs reaches 130. Then check the row of the other colonies and the number of colonies (four at most).
Explanation
The queen lays one egg per 140 steps while her store allows (sheet 5), and an egg hatches after 560 steps, so growth is limited by the laying rate and by the food.
Parameters
The speed of time, and the weather (a cold snap slows laying; winter stops it).
Check
At best one egg per 2.3 seconds is laid, so the 115 more ants of a full colony take at least about 4.5 minutes of game time if the store is never short; ants that die (after 4 to 7 minutes) slow the growth further, so a colony often stalls below the cap. To lay with 100 ants the store needs 60 crumbs.
Common pitfall
Eggs count towards the cap, so the ant count can read lower than 130 when the cap is reached. The first flight of a princess (36 ants and 40 crumbs) comes long before the cap.
Food for thought
At 130 ants per colony, what is the largest number of ants the yard can hold?
4 colonies × 130 = 520 (eggs included).
The game's year lasts 800 seconds. How many times faster than the real one is it?
A real year is about 31.6 million seconds; divided by 800 that is about 39,000 times faster.
Together, ants weigh about a fifth of what humans weigh (in carbon). Why can they still outweigh all the wild birds and mammals?
Because the total counts, not the size of one animal: with about 2 × 10^16 ants, the sum of their small bodies (about 12 megatonnes of carbon) is bigger than that of the wild birds and wild mammals together.
Schultheiss, P., et al. (2022). The abundance, biomass, and distribution of ants on Earth. PNAS, 119(40), e2201550119.
Keller, L., & Genoud, M. (1997). Extraordinary lifespans in ants: a test of evolutionary theories of ageing. Nature, 389, 958-960.
Higashi, S., & Yamauchi, K. (1979). Influence of a supercolonial ant Formica (Formica) yessensis Forel on the distribution of other ants in Ishikari Coast. Japanese Journal of Ecology, 29, 257-264.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
13 Ants, aphids and plants: honeydew, leaf-cutters and ladybirds
The yardMutualism
Ants do not only eat what falls on the ground: some tend herds of aphids for their sugar, and some cut leaves to grow a fungus.
The real mechanism
Plants make sugar, and aphids tap it. An aphid pierces a leaf or a stem and drinks the sap of the phloem, which is rich in sugar and poor in amino acids. To get enough nitrogen it has to process a lot of sap, and it gets rid of the excess sugar as droplets of honeydew. Many ants collect it: they stroke the aphid with their antennae, which prompts it to release a drop, and drink it. For many species, honeydew is a main source of carbohydrates.
This is a mutualism, an exchange that suits both sides. The ants protect the aphids from predators such as ladybirds and lacewing larvae and from parasitic wasps, and they often keep the herd clean. Tended aphids frequently do better than aphids left alone. The relation is not always so sweet: ants also eat some aphids, and the benefit depends on the species and on the conditions (Stadler and Dixon, 2005). Some ants go very far in this direction: the yellow meadow ant, Lasius flavus, lives largely on root aphids that it keeps inside its own nest.
Other ants turn plants into food in a quite different way. The leaf-cutter ants of the Americas (the genera Atta and Acromyrmex, a few dozen species) cut pieces from fresh leaves and flowers and carry them to the nest. They do not eat the leaves: the pieces are chewed into a compost on which the ants grow a fungus, and it is the fungus that feeds the colony. Ants and fungus depend on each other, an agriculture that is tens of millions of years old (Hölldobler and Wilson, 2011). A mature colony counts millions of workers of several sizes: the big ones cut and carry, the small ones tend the fungus, and small workers often ride on the pieces, which is thought to protect the carriers from parasitic flies.
A herd of aphids draws predators. Ladybirds, as larvae and as adults, eat aphids by the dozen every day. Ants attack them, but an adult ladybird has a hard shell and bitter chemicals for defense, so guarding ants reduce the losses without always preventing them.
In Ants Life
Plants. The yard holds plants = 9 plants, half flowers (strength plantHpFlower = 18) and half bushes (plantHpBush = 30), at least 100 pixels from a nest and 90 pixels from each other. A plant grows back by plantRegrow = 0.0035 point per step, 0.21 per second: a flower is whole again in about 86 seconds, a bush in about 143. Under 1 point of strength only a stump is left.
Big ants. Each new worker of the yard has a chance majorShare = 0.08 to be a big ant: strength majorHp = 14 (a worker has 6), speed 0.62 (against 0.8). It looks for the nearest plant within majorSight = 240 pixels, cuts one point of strength when within majorReach = 9 pixels, and carries the piece home, where it counts majorValue = 2 crumbs. It leaves alone a plant that carries 3 aphids or more. When no plant is left to cut it forages like a worker.
Aphids and honeydew. About a third of the plants (34%) start with a herd of 8 to 16 aphids. A herd grows by 0.0007 times its size per step, slowing down as it nears herdCap = 40 (a logistic curve: from 12 to 35 aphids takes about 66 seconds). Each aphid gives honeyRate = 0.00022 drop per step, and a drop is a crumb of food within honeyRadius = 10 pixels of the plant: a herd of 10 gives a crumb every 7.6 seconds, a herd of 40 one every 1.9 seconds. Nothing drips while the plant has half a point of strength or less, or when the cold is deeper than 0.3 on the hibernation scale (sheet 8). Every herdSpread = 3,000 steps (50 seconds) a herd of more than 30 sends 6 aphids to a bare plant within 260 pixels, and when every herd is gone a winged aphid lands on a plant and starts a herd of 6.
Ladybird. The first flies in 16,000 to 26,000 steps into the game (4.4 to 7.2 minutes), the next ones 30,000 to 50,000 steps after the last has gone, and only if no rain falls, the cold is not deep and some herd has more than 3 aphids. It walks at 0.9, eats ladyEat = 0.04 aphid per step (2.4 per second) at the nearest herd, stays 2,500 to 4,500 steps, and has ladyHp = 10 hit points. A bite costs it 0.08 from a soldier, 0.03 from a worker, 0.06 from the yellow ant. Brought down, it leaves 8 crumbs.
What Ants Life simplifies
One kind of ant does everything. In nature the aphid tenders (Lasius, Formica, mostly in temperate regions) and the leaf-cutters (tropical America) are different ants. And a leaf-cutter's piece is compost for a fungus; in the game it is simply food for the store.
Honeydew falls on the ground as ordinary crumbs. Real ants drink it straight from the aphid, carry it in their crop and share it by trophallaxis (sheet 5). There is no stroking, and no ant standing guard on the plant.
The aphids do not need the ants: the herd grows by itself, and the only protection is the bite that the ants give the ladybird. Real ants also build shelters, move aphids and, for some, keep their eggs over the winter. And real aphids weaken the plant they feed on; in the game they cost it nothing.
The ants beat the ladybird, and eat it. A real ladybird is protected by its shell and its bitter chemicals, and ants rarely get the better of it.
A real herd multiplies much faster (aphid females give birth to live young, without mating, so a herd can double in days) and produces winged forms when it is crowded, which fly off to other plants. The game keeps only the slow curve and the winged aphid that lands.
Experiment: a herd, its honeydew and the ladybird
Action
In the yard, hover the plants until the card says "a herd of N aphids", and note N and the position of the plant. Run the game at 4× for 60 seconds of game time and hover again. Watch the crumbs appear near the plant and the foragers come for them. Then wait for the toast "A ladybird flies in" (between 4.4 and 7.2 minutes), pause with Space and follow it with the pointer: how fast does the herd shrink? Call your soldiers with E or the order Bring the colony here.
Explanation
The herd grows slowly towards 40 and drips honeydew in proportion to its size; the ladybird eats 0.04 aphid per step, however big the herd is.
Parameters
The speed of time, the soldiers (the slider, before the ladybird comes), the order Bring the colony here (14 ants at most).
Check
A herd of 20 gives 20 × 0.00022 = 0.0044 drop per step, a crumb every 3.8 seconds, about 16 crumbs a minute. A ladybird that stays its 42 to 75 seconds eats 100 to 180 aphids, as many as the three herds that the yard starts with: it can empty the yard of aphids. A soldier needs 10 / 0.08 = 125 steps, about 2 seconds, to bring it down alone. The next herd comes from a winged aphid at the next multiple of 50 seconds of game time.
Common pitfall
The big ants never cut a plant that carries 3 aphids or more: to see them at work, look at the plants without a herd. And the honeydew crumbs are food for the rival kind too.
Food for thought
A herd of 30 aphids stays for 5 minutes. How many crumbs of honeydew does it give?
30 × 0.00022 = 0.0066 drop per step; 5 minutes are 18,000 steps: about 119 crumbs (a little more, if the herd grows, a little less when the cold comes).
Why does a big ant in the game spare a plant with a herd, and what does the real relation say about this?
A stripped plant would dry up and the herd would starve, so a colony that eats the plant loses its sugar. Real tending ants protect the source of the honeydew, not only the aphids.
Leaf-cutter ants cut leaves they do not eat. Why?
The pieces are the compost of a fungus that the ants grow in the nest and eat. They are farmers, not grazers.
Stadler, B., & Dixon, A. F. G. (2005). Ecology and evolution of aphid-ant interactions. Annual Review of Ecology, Evolution, and Systematics, 36, 345-372.
Hölldobler, B., & Wilson, E. O. (2011). The Leafcutter Ants: Civilization by Instinct. W. W. Norton.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
14 The ant in the house: sugar trails, walls and the dangers of a home
The houseAnts and people
The trail along a wall is the part of an ant colony that we see; the nest is somewhere behind it.
The real mechanism
A few ant species have become household guests. In temperate homes the usual visitors are the black garden ant Lasius niger, which nests outside and sends scouts to the sweet things of the kitchen, the Argentine ant Linepithema humile and, in North America, the odorous house ant Tapinoma sessile; carpenter ants (Camponotus) nest in damp wood. A tropical species, the Pharaoh ant Monomorium pharaonis, lives only in heated buildings in cooler climates, in kitchens, hospitals and blocks of flats. They come for water and for food, and the food they want changes with the needs of the colony: sugar for the energy of the workers, proteins and fats when there is brood to feed.
The way in is the trail logic of sheet 1. A scout that finds a spill goes home laying a trail, and within minutes dozens of ants follow the same line, along a wall, a cable or a pipe, because ants like to follow edges and lines. Pharaoh ants add a signal that says "do not come" at the entrance of a branch that led to nothing (Robinson et al., 2005).
The nests are seldom seen. They hide in the walls, behind baseboards, under floors, close to warmth and humidity. Pharaoh ant colonies have many queens and spread by budding: part of the colony, with queens, brood and workers, walks off and starts a new nest, with no nuptial flight (sheet 3). A building can thus hold several linked nests. Pest controllers know that a repellent spray can scatter such a colony into more nests, while a slow poison bait, which the workers carry home and share by trophallaxis (sheet 5), can reach the queens.
The people of the house are, for the colony, a source of accidents rather than a predator: a foot, a vacuum cleaner, a cloth, a spray. They kill foragers, and only foragers. The colony is spread out, with its queens and its brood in the walls, and losing a few dozen workers on a kitchen floor is a small cost.
In Ants Life
The place. The window is a garden, and the house fills the middle of it, with a margin of 7% of the width on each side and 9% of the height at top and bottom. A cross of walls (8 pixels thick) cuts the house into four rooms: kitchen, living room, bedroom and bathroom. The plan is drawn at random at the start of every game, from HOUSE_DRAWS = 48 random numbers that both engines draw first and in the same way: the cross falls between 42% and 58% of the width and of the height; the four doorways, 12% of the height or of the width of the house, sit at random places along their arm; the two cracks in the outer wall are at random heights; the four rooms are shuffled between the four corners; the furniture is flipped and moved a little inside its room, and so are the spills, the web of the spider, the rugs and the objects lying on the floor. Every room, both nests and every spill can always be reached (checked on 500 plans). The two anthills stand outside, in the garden (the black one on the left, the red one on the right), and the ants come in by a crack in the outer wall. Each kind starts with 2 scouts (workers); at houseArrive = 1,200 steps (20 seconds) the colonies follow: 10 workers and 3 soldiers for the black, 6 workers and 2 soldiers for the red, that is 12 + 3 and 8 + 2 in all, each with a queen. The furniture no longer blocks, the ants climb it, except the bathtub, full of water, which they go round. Eleven spills of food lie on the floor (140, 100, 70, 90, 50, 20, 120, 60 and 30 crumbs, and two of seeds of 25 in the garden); a spill with fewer than houseLow = 25 crumbs left is filled again, checked every houseFoodEvery = 470 steps (7.8 seconds). The trails, the store, the eggs and the fights are those of the yard (sheets 1, 2, 5 and 10). There is no weather, no season and no plant.
The doors. Each doorway has a door that opens and shuts by itself: it stays open for doorOpenFor = 3,000 to 9,000 steps (50 to 150 seconds), then shut for doorClosedFor = 1,500 to 4,000 steps (25 to 67 seconds), and so on, and it never shuts while an ant is in the doorway. A shut door is a wall: the trails through it are cut, and the vacuum cleaner and the spider cannot cross it. An ant that dies in the house (under the foot, in the cloud, in the vacuum cleaner or by a tool of the player) stays on the floor for corpseLife = 2,400 steps (40 seconds), at most corpseMax = 80 at a time.
Holding a room. Every 30 steps (half a second) the game counts the ants of each kind in each room. A room is yours if you have at least HOUSE_HOLD = 6 ants in it and more than the rival, and the rival's by the reverse rule; otherwise nobody holds it. Holding the four rooms wins the house.
The dangers. One at a time, of a kind drawn at random and placed within 30 pixels of an ant drawn at random. The first comes hazardFirst = 9,000 to 16,000 steps into the game (2.5 to 4.4 minutes), then every hazardGap = 14,000 to 26,000 steps after the last. Giant foot: a shadow for footWarn = 120 steps, then footSteps = 3 footfalls, 70 steps and footStride = 150 pixels apart, each an ellipse of 46 by 22 pixels (half-axes) where every ant dies; the trails under it keep a tenth. Insecticide: a hiss for sprayWarn = 90 steps, then a cloud that grows to a radius of sprayRadius = 80 pixels in 60 steps and stays sprayLife = 420 steps (7 seconds); each ant inside loses sprayDamage = 0.06 hit point per step (a worker lasts 100 steps, a soldier 200), and the trails inside keep 96% per step. Vacuum cleaner: a hum for vacuumWarn = 150 steps, then a nozzle that wanders for vacuumLife = 600 steps (10 seconds) at 1.1 pixel per step; it sucks up the ants within vacuumKill = 16 pixels, draws in those within vacuumPull = 38 pixels at 1.5 pixel per step, and the trails under it keep 70% per step. It stays on the floor and acts only in the open: it neither sucks up nor kills an ant behind a wall, and its victims lie on the floor. The foot, a giant one, ignores the walls.
The house spider. It lives in a web in one room and is there houseSpiderFirst = 600 steps into the game. It bites the nearest ant within spiderSight = 90 pixels that is also within spiderLeash = 170 pixels of its web, from spiderReach = 11 pixels, one bite every spiderBite = 45 steps; it has 80 hit points, walks at 0.7 and is drawn a quarter smaller than the yard spider. It hunts only an ant it can reach in a straight line: with a wall or a shut door between them it ignores the ant. The ants can bring it down (a feast of 150 crumbs), and soldiers and workers gang up on it as in the yard (sheet 9); another spider settles after houseSpiderGap = 30,000 to 50,000 steps (8 to 14 minutes).
The tools of the player. In the house the player has toolCharges = 5 puffs of insecticide and 3 ant traps, armed by a button of the dock or by the keys 6 and 7 and put down by a click on the floor (a click in a wall is refused). A puff makes a cloud that grows to sprayRadius2 = 60 pixels in spraySwell = 40 steps and stays sprayLife2 = 300 steps (5 seconds); each ant inside loses sprayDamage2 = 0.06 hit point per step, whatever its kind, and the trails inside keep 96% per step. A trap holds trapBait = 600 units of bait and a pile of trapFood = 18 crumbs; every ant within trapRadius = 10 pixels of it loses trapDamage = 0.05 per step and uses up one unit of bait per step, until the bait is gone. The tools come back with a new game.
What Ants Life simplifies
A real colony does not "hold" a room. It holds nests and a network of trails, often several nests linked together, and a room is crossed by the foragers of a colony without being owned. The game turns the territory of a colony into a head count.
The nests of the game are anthills in the open, in the garden, and the ants come in through a crack. Real nests of garden ants are often outside too, but the ants of a house more often come from nests in the walls, behind baseboards and under floors, and the tunnels of the nest view are the underground ones of the yard.
The house is flat. Real ants climb walls, run along pipes and cables, and go from floor to floor; there is no ceiling and no wall to climb (the ants climb the furniture, not the walls). The doors open and shut on a random timer, not because someone goes through, and the ants cross only at the doorways, where a real ant would slip under a door, along a pipe or through a crack.
The plan is a variation of one scheme, a cross of four rooms, not a free architecture: every game has the same four rooms and four doorways, in other proportions and other places. A real home has corridors, stairs, cupboards and many more places to hide in or to be missed.
The dangers are aimed at an ant at random and hit both colonies alike. A real accident is rare for a given colony, and the people who step on ants do not choose them. A real insecticide acts long after the spray (residues, baits carried home, repellents that scatter the colony), whereas the cloud of the game is a short poison that goes away. The house spider stays near its web, as a real one does, but it bites at sight, where a real spider in a web waits for the vibrations of what blunders into it. The puffs and traps of the player kill both kinds alike, whereas a real bait is chosen to be carried home and shared slowly (sheet 5).
Food reappears every few seconds, without limit. In a clean kitchen, food is the scarce thing, and it is a spilled crumb that brings the trail in.
Experiment: surviving a danger
Action
Choose The house in the menu. Run it at 4× until the red toast of the first danger (2.5 to 4.4 minutes), then pause with Space. Hover the shadow, the can or the nozzle to read its name, and look where your ants are. Open the charts (G) and the heat map (K) to see which places are busy. Resume and count: the toast at the end gives the number of ants lost. Start again with R and play it the other way: keep the ants far from the spot, or out in the garden, when the warning comes. Then try the tools: in a new game, press 6 and click on the floor where ants gather to put down a puff of insecticide, or 7 and click next to a spill of food to lay a trap, and count the ants lost.
Explanation
The damage is proportional to the number of ants in the area, and the warning (2 seconds for the foot, 1.5 for the spray, 2.5 for the vacuum cleaner) is the only time you have to act.
Parameters
The speed of time, the orders of the yellow ant (Call the others, Bring the colony here), where the trails lead.
Check
A footprint covers about π × 46 × 22 = 3,200 square pixels, the cloud π × 80² = 20,100. A worker crossing the cloud through its middle needs 160 / 0.8 = 200 steps (3.3 seconds), but it dies after 6 / 0.06 = 100 steps (1.7 seconds): it does not come out, and a soldier (12 / 0.06 = 200 steps against 160 / 0.67 = 239 steps) does not either. Only the ants that stay out of the cloud, or that leave it at once, survive. The nozzle sweeps about 32 × 660 = 21,000 square pixels in its 10 seconds. A puff of the player covers π × 60² = 11,300 square pixels: a worker needs 120 / 0.8 = 150 steps to cross it through the middle but dies after 100, about two thirds of the way. A trap kills a worker in 6 / 0.05 = 120 steps (2 seconds), and its 600 units of bait last for 600 ant-steps: about 5 workers, or 2.5 soldiers.
Common pitfall
The danger also strikes the rival kind: do not count on it to lose the room you want. And the foot wipes the trails under it, so the foragers that survive may lose the way to the food until a new trail is laid. The tools kill your own ants too.
Food for thought
How long can a worker stay in the insecticide cloud before it dies, and how long would it need to cross the cloud through its middle?
6 / 0.06 = 100 steps, 1.7 seconds, against 160 pixels at 0.8 pixel per step, 200 steps, 3.3 seconds: it dies about halfway.
Why can a real colony in a wall lose dozens of foragers and carry on?
Its queens and brood are in the nest, out of reach, and the foragers are the part of the colony that is replaced; a new trail is laid within minutes.
The rival has 10 ants in a room. How many of yours does the room need to be yours?
11 are enough (more than 10), as long as there are at least 6. With 10 against 10, nobody holds the room.
A door stays shut for 1,500 to 4,000 steps. How many seconds is that, and what share of the time is a door open on average?
1,500 / 60 = 25 seconds and 4,000 / 60 = about 67 seconds. A door is open 3,000 to 9,000 steps (6,000 on average) and shut 1,500 to 4,000 (2,750 on average), so it is open 6,000 / (6,000 + 2,750) = about 69% of the time.
A trap holds 600 units of bait, and every ant within 10 pixels uses up one unit per step while it loses 0.05 hit point per step. How many workers (6 hit points) can it kill, if each one stays until it dies?
A worker dies after 6 / 0.05 = 120 steps and uses 120 units, so 600 / 120 = 5 workers. A soldier (12 hit points) needs 240 steps, so the same trap kills only 2.5 of them.
Hölldobler, B., & Wilson, E. O. (1990). The Ants. Harvard University Press.
Robinson, E. J. H., Jackson, D. E., Holcombe, M., & Ratnieks, F. L. W. (2005). "No entry" signal in ant foraging. Nature, 438, 442.
The biological statements of the sheets were written with the cited sources in mind, and are kept deliberately cautious where the literature varies from species to species. The Ants Life values are those of the code (js/formicarium.js and engine/src/formicarium.hpp) at the time of writing: if the game changes, they may change too.
Ants Life is my modest contribution to education, born from the wish to explore science.