Chapter 1: The Ant Colony as a Complex System

(1) Our ideas about ants over history have mirrored political fashions and fixations.

(2) The behavior of ant colonies is guided by a dynamic network of interactions.

In 19th C colonial expansion put Europeans in contact with stunning biological diversity of the tropics
Choice of experimental systems (sea urchins, frogs) affects theoretical stance
Our thinking about ants and how they do what they do has tended to mirror current political fashions and fixations
An ants moment to moment response depends on what it is doing and its environment; a colony’s response depends on what happened last week.
Colonies behaviors change over years…
There are always dense webs of contingency in systems of interacting parts
This book presents a single idea about ants: the behavior of ant colonies arises from dynamical networks of interaction . … “A colony’s behavior is guided by a pulsing, shifting web of interactions, in which the pattern of the interactions is more important than the content.”

Chapter 2: Colony Organization

(1) Task allocation: ants can switch tasks, although some roles are sinks. Young workers tend to stay inside the nest, and the transition to roles that take them outside.
(2) Ants make moment to moment decisions about whether to engage in a task or do nothing. And ants don’t have to do things perfectly… just enough to (collectively) get the job done.

We know little about ants and their behavior. They exhibit an astonishing diversity of behavior, and individual exceptions to how even well-studied ants are expected to behave is not uncommon.

“Task Allocation” rather than “Division of Labor,” to emphasize that ant colonies shift their behavior in response to a changing world.
“Even the most elegant argument about what ought to happen does not demonstrate that it does happen.” p. 26
Ants switch tasks if more ants are needed to perform a particular task. However, not all task-to-task transitions are possible. In particular, once a worker has become a forager, it does not switch back to other tasks.
We don’t know much about how ants manage to perform tasks. “It’s important to remember that whatever the ant is doing, it’s not rocket science.” An ant’s performance doesn’t have to be perfect. Instead, enough ants have to perform well enough to get the job done.
How an ant’s task changes with age is extremely variable. However, a general pattern is that younger workers stay inside the nest doing brood care and nest maintenance and move outside of the next when they are older.
Some researchers suggests that as an ant gets older it responds to more stimuli and becomes involved in a greater variety of tasks.
Factors that result in task shifts: (1) the spatial dynamic (what is near to hand); (ii) recent history of the colony’s needs; (iii) current demands of the colony and its environment; (iv) individual differences — some ants are consistently more active than others.
As well as switching tasks, ants make moment to moment decisions about whether to actively engage in a task. When the numbers of workers doing nest maintenance increase, the number of workers foraging decrease — even though foragers themselves are not shifting to maintenance (since foragers don’t shift).
? Why is foraging a sink? Why don’t foragers shift?

Chapter 3: Interaction Networks

(1) Colonies perform a standard sequence of tasks each day. Colonies behavior and responses to disturbances change over time.
(2) Ants interact via antennae touching and sensing cuticular hydrocarbons, and the rate at which they have interactions is crucial. (e.g., foraging is triggered by foragers returning with food at a rate >= 1/10 seconds)
(3) “Differences among species in terms of the speed and intensity of the colony’s reaction come from differences in the rate at which the network is ticking, how often the ants interact, and how quickly and how much they respond. All of the variation among species in interaction networks begins with differences in the shape of the paths that ants use to move around.”
(4) How quickly a group of ants can find something and how quickly information about it spreads through the colony are both enhanced by increased colony size and by straighter paths.… “Adjusting path shape to density [increasing convolutions with density] makes Argentine ants more effective searchers.”
(5) Some ants cluster together when their density is low, but actively avoid one another when density is higher. … As density decreased, ants tended to stick to the boundaries of an area, which is a good strategy for increasing interaction since the boundaries of an area increase linearly whereas the area increases geometrically

Older larger colonies are more stable than young small ones
In a particular ant species, colonies perform a standard sequence of tasks each day. Older colonies respond to disturbances in much the same way each time. A colony’s behavior transforms in predictable ways as it grows older and larger. One colony’s relations with its neighbors look much like another’s.
Task partitioning is the name for a series of tasks that accomplish a goal. For example, foraging, processing food, and transporting it to larva produce the flow of food into the colony.
When ants interact by touching antennae, one ant perceives the cuticular hydrocarbons of another. These are greasy fatty acids that are spread over the hard outer surfaces of an ants body by grooming. …Changes in the cuticular hydrocarbons of an ant triggered by chance events — e.g., when some ants are fed a particular species of cockroaches — can lead to interaction changes.
Foraging and its control by interaction networks. Colony activity begins in the early morning — probably stimulated by the sun striking the earth and warming the ground — when patrollers leave the nest. They lay short trails that show foragers which direction in which to forage, and foragers only go out when patrollers return. (This is ‘wise’ as the failure of patrollers to return may signal a problem.) Once foraging begins, the number of ants out foraging at any given time is regulated by rate of interactions with foragers returning to the nest with food. A change in the rate of forager return alters the rate at which foragers go out very quickly — within about two minutes. This response rate may be determined by the length of an ant’s memory, which is about 10 seconds.
Other cases in which interaction networks control activities are nest choice, midden work, and trail laying.
Interaction rhythms produce colony behavior as a result of the relation of two rates: the rate at which interactions occur and the rate at which ants respond to them, the latter which depends on the ant’s ‘memory.’
Harvester ants appear to be able to remember location from one day to another, in that they will return to the same area to forage if not provided with other cues by patrollers. Species seem to differ in how long they can remember things, with some ants showing memories that last for days (of the colony’s smell) and others showing an apparent memory in ‘old’ ants that can last over the winter.
There are some ants that appear to be much more active than others — about 10% of ants make many trips, while the rest make only a few. But if you take away that most active set of ants, the next day there will be another 10% of highly active ants. (???If you don’t take them away, will the same or different ants be most active the next day???)
“Differences among species in terms of the speed and intensity of the colony’s reaction come from differences in the rate at which the network is ticking, how often the ants interact, and how quickly and how much they respond. All of the variation among species in interaction networks begins with differences in the shape of the paths that ants use to move around.”
In general, if an ant reacts to its rate of encounter by changing the way that it moves, then each encounter will change the probability of future encounters.
How quickly a group of ants can find something and how quickly information about it spreads through the colony are both enhanced by increased colony size and by straighter paths. These two effects interact. Smaller groups must use straighter paths to get the same result as larger groups. … “Adjusting path shape to density [increasing convolutions with density] makes Argentine ants more effective searchers.” [[Another issue is the distribution of what is being searched for — if the distribution is not random (e.g., seeds scattered in a circle around a plant) — then different search algorithms may be called for.]]
Some ants cluster together when their density is low, but actively avoid one another when density is higher. These particular ants can see about an ant-length in front of them, and so can take evasive action. As density decreased, ants tended to stick to the boundaries of an area, which is a good strategy for increasing interaction since the boundaries of an area increase linearly as the sum of the lengths, but its area increase geometrically. In general, when contract rate is random, and each ant may contact any of the others, interaction rate changes as the square of the number of ants, so small changes in density can have large effects on contact rates. So it makes sense to be able to modulate contact rate as a function of density.

Chapter 4: Colony Size

(1) 90% of colonies die before they are 1; if the reach 2 years old, it is almost certain they will live 20-25 years.
(2) Colony behavior changes with the age of the colony. Older colonies
(i) devote fewer ants to foraging,
(ii) avoid foraging overlap with neighboring colonies
(iii) respond to disturbances consistently,
(iv) more readily return to homeostasis,
(v) are more likely to detect events, and
(vi) develop larger and more complex nest structures.
(3) Ants not foraging (or doing other tasks) just wait

In Harvester ants, about 90% of colonies die before they are a year old. If they reach two years old it is almost certain that they will live 20 to 25 years. One reason that young colonies may be prone to death is that there may not be enough ants to maintain functioning interaction networks. An ant might get stranded in a location where the is no other ant to meet, and so it will not do anything.
Colony behavior changes with the age of the colony. Younger colonies devote a higher proportion of ants to foraging than older colonies.
As a colony grows, if ants are not foraging, what are they doing? Surprisingly, the answer seems to be that they are doing nothing. This might be useful for any of a number of reasons. Perhaps they are a reserve (though this has not been observed in 25 years of observations. Perhaps they are a mechanism for food storage. Perhaps they provide a buffer that dampens increases in the interaction rate as the colony grows.
Just as younger colonies devote a higher proportion of ants to foraging, it is also the case that in younger colonies ants switch tasks more readily (is foraging still a sink?); in older colonies, an ant tends to do the same task from day to day unless there is a drastic change in the environment. Not only is task allocation more consistent in older colonies, but when perturbed older colonies tend to respond the same way time after time (within a colony, not between colonies) where younger colonies respond differently each time. Older colonies are also more homeostatic, and tend to return to base conditions after each perturbation
The larger a colony, the more likely it is that if something changes in the environment an ant will detect it
As Harvester ant colonies age, the structure of the nests become not just larger but more complex.
It is interesting to speculate about how very large colonies — of millions of ants — are organized to support interaction among groups and subgroups. One way to deal with this is if the response of an individual in a smaller group requires multiple interactions with ants in a larger group to trigger it.

Chapter 5: Relations with Neighbors

How colonies interact with their neighbors

competition for food source / resource
avoiding overlap while foraging
sharing nests
inhabiting abandoned nests of neighbors
using pherome trails
Avoiding overlap with neighbors seems to be an important function of the patrolling system. It appears that an encounter with a neighbor makes a patroller less likely to reinforce a foraging direction. … Each patroller that meets a neighbor’s patroller avoids returning directly to the nest. Coming back to the nest from another side, it doesn’t put down its chemical secretion in the direction that led to the encounter with the neighbor. Meanwhile, other patrollers that went in other directions and did not meet the neibhbor’s patrollers retun to the nest and do put down the secretion that leads foragers back the way they came. This is sufficient, most of the time, to lead foragers away from the site of the encounter…
However, it is only mature colonies, older than 5 and at their stable size, that avoid foraging towards the place they met a neighbor the previous day.
There is less hostility between colonies founded by closely related queens than those founded by unrelated queens. [cuticular hydrocarbons]
Just as ants within a colony use interaction rate as a cue to the numbers of nestmates performing a task, or finding a nest or food source, so in interaction sbetween colonies, and can use interaction rate o assess the numbers of ants from another colony they are dealing with. The rate at which ants of one colony meet ants of the other depends on the relative sizes fo the two colonies — that is, the ration of numbers of workers in one colony to numbers in another.
colonies sometimes move into the nests of dead neighbors
— ??? I wonder if this changes the behavior of the colony — to what extent does the architecture of the nest affect the ‘processing’? Could or have researchers studies nest structure? It would be interested to apply computer aided tomography to deciphering and studying nest architecture. ??? —
In conflicts between colonies, numbers matter much more than size. In fact, smallness may be an advantage because a colony with access to the same amount of resource can produce larger numbers of smaller ants.
The argentine ants … were less aggressive, and moree likely to avoid interaction with the other species when their colonies were small, but plunged into all-out conflict when there were 1,0000 workers or more.
These conflicts can go on for years…
A colony o fone speices may use the interaction network of the other — e.g., the pheromone trails.
Dominance hierarchies vary according to conditions…
There is no single characteristic that makes invasive ants successful. Instead, their success arises from the local details of their network of ecological relations with other species.
The argentine ant forms supercolonies of genetically related colonies. …In the winter a colony aggregates in one or a few large nests, often at the base of shrubs…. in the spring the ants move out into distinct nests connected by trails… by the end of the summer the colony is at its most dispersed, spanning about 200 meters, its many small nests connected by trails.
How neighboring ants interact with each other determines how they get the resources they need to survive. Over time, these interactions will produce the spacing of colonies on the landscape — for example, if two colonies can’t share a tree, eventually there will only be one in each tree. Behavioral interactions create the spacing of colonies, which in turn influences the availability of resources and thus colony growth, which both feed back to influence the frequency and outcome of behavioral interactions.

Chapter 6: Ant Evolution

Ants evolved from wasps about 130 million years ago. About 90 million years ago they began to diversify, and it is clear that diversity in ants is closely related to the evolution of diversity in plants. A burst of diversity in ant species occurred at the same time as the origin and radiation of the flowering plants. After the K-T boundary, flowering plants increased in diversity, scale insects appeared, and ants evolved to take advantage of scale insects.
Plant-ant mutualisms. What is most remarkable about these examples is that the ants behavior draws on very precise and intimate use of the plant’s physiology.
Discussion of reproduction and the evolution of worker sterility.

Chapter 7: Modeling Ant Interaction

Any model of ant interaction has to have at least two levels. The first specifies how the workers interact within the colony to regulate the acquisition, processing, and distribution of resources [behavior]. The second specifies how the internal processes of the colony connect to the colony’s environment [ecology]. That is, how the colony’s interactions with the rest of the world determine its grownth and reproduction, and how this changes the colony’s environment, which in turn changes its interactions and feeds back on its growth and reproduction.
How do decentralized systems combine variable input and imprecise response, yet manage to have the system respond correctly enough of the time that it can function?
When you watch real ants, they don’t look like they do on TV. You see a lot of bmbling around, a few ants going the wrong way, ants pulling an object in different directions. … The achievements of colonies do not arise from the skill and determination of individual ants. … The colony isn’t like clockwork, but it is ticking.
What is most amazing about ants is that such variable, noisy processes create a system that can accomplish so much. The system is turbulent in every way. The experience of each ant is variable, only loosely tuned to the state of the world, because the rate at which each ant meets others depends on so many small contingencies in how the nats happen to move around. The reaction of each ant to the pattern of encounters it experiences is imprecise, because ants don’t count very carefully and because an ant’s respond to the same experience will be different from one time to the next.

# # #