Agent-Based Computational Modelling

Part of the series Contributions to Economics pp 139-152

Agent-Based Models in Ecology: Patterns and Alternative Theories of Adaptive Behaviour

  • Volker GrimmAffiliated withUFZ Centre for Environmental Research Leipzig-Halle
  • , Steven F. RailsbackAffiliated withLang, Railsback & Associates

* Final gross prices may vary according to local VAT.

Get Access


Ecologists have used agent-based models for a long time, but refer to them as “individual-based models” (IBMs). Common characteristics of IBMs are discrete representation of unique individuals; local interactions; use of adaptive, fitness-seeking behaviour; explicit representation of how individuals and their environment affect each other; and representation of full life cycles.

Ecology has contributed to agent-based modelling in general by showing how to use agent-based techniques to explain real systems. Ecologists have used IBMs to understand how dynamics of many real systems arise from traits of individuals and their environment. Two modelling strategies have proven particularly useful.

The first strategy is “pattern-oriented modelling” (POM). POM starts with identifying a variety of observed patterns, at different scales and at both individual and system levels, that characterize the system’s dynamics and mechanisms. These patterns, along with the problem being addressed and conceptual models of the system, provide the basis for designing and testing an IBM. A model’s variables and mechanisms are chosen because they are essential for reproducing these characteristic patterns. After an IBM is assembled, alternative versions (different theories for individual behaviour; different parameterizations) can be tested by how well they reproduce the patterns.

The second strategy is developing general and reusable theory for the adaptive behaviour of individuals. A “theory” is a model of some specific individual behaviour from which system-level dynamics emerge. Theory can be developed by hypothesizing alternative models for the behaviour, then using the IBM to see which alternative best reproduces a variety of patterns that characterize the system dynamics of interest. Empirical observations are used to develop both theories and the patterns used to test and falsify them.

These two strategies are demonstrated with example models of schooling behaviour in fish, spatiotemporal dynamics in forests, and dispersal of brown bears.