Starling flock at sunset in Denmark
Collective animal behavior is a form of social behavior involving the coordinated behavior of large groups of similar animals as well as emergent
properties of these groups. This can include the costs and benefits of
group membership, the transfer of information across the group, the
group decision-making process, and group locomotion and synchronization.
Studying the principles of collective animal behavior has relevance to
human engineering problems through the philosophy of biomimetics.
For instance, determining the rules by which an individual animal
navigates relative to its neighbors in a group can lead to advances in
the deployment and control of groups of swimming or flying micro-robots
such as UAVs (Unmanned Aerial Vehicles).
Examples
Examples of collective animal behavior include:
- Flocking birds
- Herding ungulates
- Shoaling and schooling fish
- Schooling Antarctic krill
- Pods of dolphins
- Marching locusts
- Nest building ants
Proposed functions
Many
functions of animal aggregations have been proposed. These proposed
functions may be grouped into the four following categories: social and
genetic, anti-predator, enhanced foraging, and increased locomotion
efficiency.
Social interaction
Support
for the social and genetic function of aggregations, especially those
formed by fish, can be seen in several aspects of their behavior. For
instance, experiments have shown that individual fish removed from a
school will have a higher respiratory rate than those found in the
school. This effect has been partly attributed to stress, although
hydrodynamic factors were considered more important in this particular
study.
The calming effect of being with conspecifics may thus provide a social
motivation for remaining in an aggregation. Herring, for instance, will
become very agitated if they are isolated from conspecifics. Fish schools have also been proposed to serve a reproductive function since they provide increased access to potential mates.
Protection from predators
Several anti-predator functions of animal aggregations have been proposed. One potential method by which fish schools or bird flocks may thwart predators is the ‘predator confusion effect’ proposed and demonstrated by Milinski and Heller (1978).
This theory is based on the idea that it becomes difficult for
predators to pick out individual prey from groups because the many
moving targets create a sensory overload of the predator's visual
channel. Milinski and Heller's findings have been corroborated both in
experiment and computer simulations.
A second potential anti-predator effect of animal aggregations is
the "many eyes" hypothesis. This theory states that as the size of the
group increases, the task of scanning the environment for predators can
be spread out over many individuals. Not only does this mass collaboration presumably provide a higher level of vigilance, it could also allow more time for individual feeding.
A third hypothesis for an anti-predatory effect of animal aggregation is the "encounter dilution"
effect. Hamilton, for instance, proposed that the aggregation of
animals was due to a "selfish" avoidance of a predator and was thus a
form of cover-seeking.
Another formulation of the theory was given by Turner and Pitcher and
was viewed as a combination of detection and attack probabilities.
In the detection component of the theory, it was suggested that
potential prey might benefit by living together since a predator is less
likely to chance upon a single group than a scattered distribution. In
the attack component, it was thought that an attacking predator is less
likely to eat a particular animal when a greater number of individuals
are present. In sum, an individual has an advantage if it is in the
larger of two groups, assuming that the probability of detection and
attack does not increase disproportionately with the size of the group.
Enhanced foraging
A
third proposed benefit of animal groups is that of enhanced foraging.
This ability was demonstrated by Pitcher and others in their study of
foraging behavior in shoaling cyprinids.
In this study, the time it took for groups of minnows and goldfish to
find a patch of food was quantified. The number of fishes in the groups
was varied, and a statistically significant decrease in the amount of
time necessary for larger groups to find food was established. Further
support for an enhanced foraging capability of schools is seen in the
structure of schools of predatory fish. Partridge and others analyzed
the school structure of Atlantic bluefin tuna from aerial photographs
and found that the school assumed a parabolic shape, a fact that was
suggestive of cooperative hunting in this species (Partridge et al.,
1983).
Increased locomotion efficiency
This
theory states that groups of animals moving in a fluid environment may
save energy when swimming or flying together, much in the way that
bicyclists may draft one another in a peloton. Geese flying in a Vee formation are also thought to save energy by flying in the updraft of the wingtip vortex generated by the previous animal in the formation. Ducklings have also been shown to save energy by swimming in a line. Increased efficiencies in swimming in groups have also been proposed for schools of fish and Antarctic krill.
Group structure
The
structure of large animal groups has been difficult to study because of
the large number of animals involved. The experimental approach is
therefore often complemented by mathematical modeling of animal
aggregations.
Experimental approach
Experiments
investigating the structure of animal aggregations seek to determine
the 3D position of each animal within a volume at each point in time. It
is important to know the internal structure of the group because that
structure can be related to the proposed motivations for animal
grouping. This capability requires the use of multiple cameras trained
on the same volume in space, a technique known as stereophotogrammetry.
When hundreds or thousands of animals occupy the study volume, it
becomes difficult to identify each one. In addition, animals may block
one another in the camera views, a problem known as occlusion. Once the
location of each animal at each point in time is known, various
parameters describing the animal group can be extracted.
These parameters include:
Density: The density of an animal aggregation is
the number of animals divided by the volume (or area) occupied by the
aggregation. Density may not be a constant throughout the group. For
instance, starling flocks have been shown to maintain higher densities
on the edges than in the middle of the flock, a feature that is
presumably related to defense from predators.
Polarity: The group polarity describes if the group
animals are all pointing in the same direction or not. In order to
determine this parameter, the average orientation of all animals in the
group is determined. For each animal, the angular difference between its
orientation and the group orientation is then found. The group polarity
is then the average of these differences (Viscido 2004).
Nearest Neighbor Distance: The nearest neighbor
distance (NND) describes the distance between the centroid of one animal
(the focal animal) and the centroid of the animal nearest to the focal
animal. This parameter can be found for each animal in an aggregation
and then averaged. Care must be taken to account for the animals located
at the edge of an animal aggregation. These animals have no neighbor in
one direction.
Nearest Neighbor Position: In a polar coordinate
system, the nearest neighbor position describes the angle and distance
of the nearest neighbor to a focal animal.
Packing Fraction: Packing fraction
is a parameter borrowed from physics to define the organization (or
state i.e. solid, liquid, or gas) of 3D animal groups. It is an
alternative measure to density. In this parameter, the aggregation is
idealized as an ensemble of solid spheres, with each animal at the
center of a sphere. The packing fraction is defined as the ratio of the
total volume occupied by all individual spheres divided by the global
volume of the aggregation (Cavagna 2008). Values range from zero to one,
where a small packing fraction represents a dilute system like a gas.
Cavagna found that the packing fraction for groups of starlings was
0.012.
Integrated Conditional Density: This parameter
measures the density at various length scales and therefore describes
the homogeneity of density throughout an animal group.
Pair Distribution Function:
This parameter is usually used in physics to characterize the degree of
spatial order in a system of particles. It also describes the density,
but this measures describes the density at a distance away from a given
point. Cavagna et al. found that flocks of starlings exhibited more
structure than a gas but less than a liquid.
Modeling approach
The simplest mathematical models of animal aggregations generally instruct the individual animals to follow three rules:
- Move in the same direction as your neighbor
- Remain close to your neighbors
- Avoid collisions with your neighbors
A diagram illustrating the difference between 'metric distance' and 'topological distance' in reference to fish schools
An example of such a simulation is the Boids program created by Craig Reynolds in 1986. Another is the Self Propelled Particle
model. Many current models use variations on these rules. For instance,
many models implement these three rules through layered zones around
each animal. In the zone of repulsion very close to the animal, the
focal animal will seek to distance itself from its neighbors in order to
avoid a collision. In the slightly further away zone of alignment, a
focal animal will seek to align its direction of motion with its
neighbors. In the outmost zone of attraction, which extends as far away
from the focal animal as it is able to sense, the focal animal will
seeks to move towards a neighbor. The shape of these zones will
necessarily be affected by the sensory capabilities of the animal. For
example, the visual field of a bird does not extend behind its body.
Fish, on the other hand, rely on both vision and on hydrodynamic signals
relayed through its lateral line. Antarctic krill rely on vision and on hydrodynamic signals relayed through its antennae.
Recent studies of starling flocks have shown, however, that each
bird modifies its position relative to the six or seven animals directly
surrounding it, no matter how close or how far away those animals are.
Interactions between flocking starlings are thus based on a topological
rule rather than a metric rule. It remains to be seen whether the same
rule can be applied to other animals. Another recent study, based on an
analysis of high speed camera footage of flocks above Rome and assuming
minimal behavioural rules, has convincingly simulated a number of
aspects of flock behaviour.
Collective decision making
Aggregations
of animals are faced with decisions which they must make if they are to
remain together. For a school of fish, an example of a typical decision
might be which direction to swim when confronted by a predator. Social
insects such as ants and bees must collectively decide where to build a
new nest.
A herd of elephants must decide when and where to migrate. How are
these decisions made? Do stronger or more experienced 'leaders' exert
more influence than other group members, or does the group make a
decision by consensus? The answer probably depends on the species. While
the role of a leading matriarch in an elephant herd is well known,
studies have shown that some animal species use a consensus approach in
their collective decision-making process.
A recent investigation showed that small groups of fish used
consensus decision-making when deciding which fish model to follow. The
fish did this by a simple quorum rule such that individuals watched the
decisions of others before making their own decisions. This technique
generally resulted in the 'correct' decision but occasionally cascaded
into the 'incorrect' decision. In addition, as the group size increased,
the fish made more accurate decisions in following the more attractive
fish model. Consensus decision-making, a form of collective intelligence, thus effectively uses information from multiple sources to generally reach the correct conclusion.
Some simulations of collective decision-making use the Condorcet method to model the way groups of animals come to consensus.
