Self-Organization in Social InsectsKazuo Uyehara5/04/2008Self-organization is a process in which complex patterns or behaviorsare produced without direction from an outside or supervising source. Thismechanism is observed in a multitude of systems including phase-transitions,oscillating chemical reactions, and the development of fish schools. Self-organization in biological contexts is particularly interesting, as it may bea necessary component of the observed complexity found in organismal be-havior and play a vital part in natural selection.Self-organizing systems in biology are generally composed of many indi-viduals that work collectively to create complex patterns through positiveand negative feedback interactions. These patterns are not encoded on theindividual level nor are they being built with any knowledge of the globalstructure, they are instead products of interactions between individuals orbetween individuals and their local environment. The behavior of manysocial insects are fitting examples of this process, as the individual insectsgenerally do not have the cognitive power to generate their observed complexstructures and activities. For example, the construction of termite mounds,ant walls, honey bee comb patterns and wasp nests, honey bee thermoreg-ulation, army ant swarms, ant corpse-aggregation, and foraging in ants andbees can all be seen as a combination of simple rules on the individual levelthat produce a global response without any external direction.The emergent properties of social insects interactions are generally char-acterized by three properties: complex spatio-temporal patterns from ho-mogenous initial conditions, the existence of multiple stable states that arethe consequence of the amplification of random initial heterogeneity, andthe presence of bifurcation points, at which a patterns drastically changeas a result of small parameter changes (Camazine et al.). Self-organizingsystems are strongly dependent on their environment because of their sensi-tivity to initial conditions and they can simultaneously offer stable patternsthat resist perturbations given appropriate parameters. However, they arealso capable of variability and drastic change when near a bifurcation point.1This has major implications for the role of self-organization in the evolutionof complex patterns. The limited amount of information that can be storedin the genome may be used to produce a plethora of sophisticated behaviorsby devoting a relatively small portion of the sequence to simple rules thatcan yield emergent complexity when aggregated. Thus self-orgnization maybe a powerful tool in the optimization of genetic information and in thecreation of simple parameters that can be selected for through evolution.Although self-organization has been proven experimentally in non-biologicalsystems, it is much more difficult to perform similarly conclusive experimentswhen dealing with organisms such as social insects. It is most likely thatthe examples presented here are not fully explained by self-organizationalone and are instead the product of a synthesis of different mechanisms.However, this does not make self-organization any less attractive as an al-ternative to other mechanisms for complex patterns such as an organizingleader, blueprints, recipes, and templates.These alternative hypotheses would often require an unlikely amountof individual complexity. While social insect colonies often have a centralleader, such as a single queen, the leader would be unable to efficientlycommunicate detailed and individual commands to a large group of work-ers. This problem in communication could be circumvented by individualsalready possessing the necessary information to produce complex spatio-temporal patterns in the form of a blueprint that all individuals possess orresponse or a recipe of step-by-step instructions that direct each worker.Both of these strategies require a large amount of encoded information andrecipes make it particularly difficult for an entire community to simulta-neously work towards the same global outcome. Another alternative, thetemplate, would provide individuals with the necessary information to pro-duce the pattern. The major drawback to template-guided building is thatthere is not always a mechanism for template construction. These alternativemechanisms to self-organization are certainly used by other organisms, butit is unlikely that they are the main method of complex pattern formation insocial insects. Nevertheless, we will see an example of a possible template-driven behavior in the construction of termite mounds, and it would not besurprising if other patterns were best explained by a combination of mech-anisms.We will now examine biological examples to evaluate models of self-assembly in social insects. A very basic and well supported model of self-assembly explains the formation of ant trails through the use of pheromones.As ants travel between the nest and food sources, they lay a chemicalpheromone that attracts other ants. This positive feedback eventually pro-2Figure 1: Double-bridge experiment (Deneubourg and Goss 1989) The per-centage of ants that chose each of the branches as a function of time. Inter-estingly, the branch initially favored by random fluctuations was not chosenin the end, as the additional pheromones were not sufficient to produce arun-away positive feedback. The inset is a drawing of the experimentalsetup.duces an ordered path of ants going to and from the food. The pheromonedegrades over time, so once one path has been established, it usually per-sists, as less traveled paths disappear. This is a simple example that doesnot reflect the more subtle complexities of the trails used by some species,but clearly illustrates the positive feedback mechanism. The double-bridgeexperiment done by Deneubourg and Goss in 1989 supports this mechanismwith data. They separated a nest of ants from a food source by a bridgewith two equally long branches and observed the preferential selection ofone bridge over the other. A random heterogeneity in the path selection isamplified and one path is eventually chosen (Figure 1). Negative feedbackis manifested through the depletion of food or satiation, which prevent thepersistence of trails for unreasonable amounts of time. Overall this exampleexhibits what appears to be a community decision to favor one path over theother through individual decisions based on local pheromone concentrationsand also shows