Complexity and Organized Behaviour within Environmental Bounds (COBWEB): An Agent-Based Approach to Simulating Ecological Adaptation

Abstract

An agent-based simulation model, Complex Organization and Behaviour within Environmental Bounds (COBWEB), has been developed to simulate how a group of agents adapts to environmental variability and change. The environment is an abstraction with properties that are common to many complex systems. The environment is two-dimensional and is characterized by a cellular automaton with multiple resources, variable rates of growth, famines and the presence of immovable objects. The agents are genetic algorithms with modifiable behavioural traits and resource preferences. The agents' strategies are defined by movement, consumption, asexual and sexual reproduction, communication and sharing or stealing resources from other agents. The strategies are determined genetically and through new information acquired from the environment or from other agents. COBWEB has been used to examine the linkages between population growth, resources, the availability of energy and surprise suggesting a role for this software in studying ecological processes, specifically the adaptation of different species to environmental change. The current research results illustrate how COBWEB agents can adapt to variability and change in response to changing resource availability, available energy and crossing threshold values. The second set of experiments indicates the importance of genetic diversity and the ability of an invader to extract more energy from the landscape than the indigenous species. The results illustrate that COBWEB is a useful tool for this sort of research and is capable of generating new insights.

Keywords: Complexity, ecological simulation, environmental change, genetic algorithms, invasive species