A cellular automata model for species competition and evolution

Abstract

We propose a inhomogeneous cellular automata (CA) model in which several species compete for their territory and can co-evolved in regions where several of them coexist. Our model has as few parameters as possible. Each cell represent an individual and the associated CA rule represents its genome. The state evolution of each cell is interpreted as a phenotype. The fitness is defined as the cell activity, i.e. the variability of the state over time. Individuals of low fitness evolves by copying part of the genomes of neighboring high fitness individuals. We then consider a computer experiment implementing the competition-evolution of two species (two rules) each populating initially half of cellular space. © Springer-Verlag Berlin Heidelberg 2006.