An emergent computational approach to the study of ecosystem dynamics

Abstract Despite success in theory formulation and prediction of quantities and patterns in nature, traditional modeling approaches have not proven particularly valuable as “surrogate experimental systems” in applied ecology. Theoretical models, while useful as embodiments of ecological theory, are too simplistic to be effective surrogate systems. Although simulation models can represent systems of realistic complexity, they are limited by factors which arise from the way in which they are built. We propose an alternative paradigm for modeling biotic systems which promises to enhance their usefulness as surrogate experimental systems. This paradigm is based on the premise that dynamic behavior in biotic systems emerges from the low-level interactions of independent agents. It forms the basis for the new field of artificial life (ALife), which involves the study of life-like behavior in artificial systems. In an ALife model, the target biological system is modeled as a population of independent computer programs called machines. The complete behavioral repertoire of each individual, including its interaction with others, is specified within the entity itself. A spatially-referenced “environment” is provided within which the machines interact with each other and their local environment. There is no overall controlling program or agent. Thus, the overall behavior of the system emerges from local interactions between independent agents. In this paper, we examine the premises upon which ALife is based (including the concept of emergence) and discuss several examples of ALife models at ecological scales, which we call “artifical ecosystems”. We next introduce LAGER, an environment for producing and running artificial ecosystems. Finally, we present PARE, a host/parasitoid dynamics model built in LAGER, and compare its behavior to two similar systems in the literature.