Non-local Evolutionary Adaptation in Gridplants
Submitted to CEC2004

David Doty

Abstract PDF eprint

A simulated model of plant growth and evolution was studied. Plants start out as seeds on a 2-dimensional grid. Plant genomes are modelled as instructions telling a plant where to grow and where to place seeds. Energy is gained by occupying grid space in analogy to collection of light by leaf surface area. At the end of a generation, cells currently occupied by plants are clered and the seeds dropped by all the plants sprout to become the new plants. Each seed produced has a probability of mutation to the genome it contains. The simulated plants evolve to play a game of competitive exclusion, in which grid space is a limited resource.

This paper tested the hypothesis that the evolved plants would display nonlocal adaptation, i.e. that the plants would not only adapt to their local environment, but would acquire general skill that would enable them to grow competitively against plants that were never a part of their environment. Statistical tests show that populations of plants that have evolved for a larger number of generations are able to occupy more grid space when played against populations of plants evolved for a shorter time. This occurs even if the two competing populations come from entirely different lineages. This improvement in competative ability continues over the course of the evolution performed in this study, without appearing to reach an equilibrium after which further evolution fails to improve the plants. This suggests that the plants are continually discovering generally useful strategies, rather than adapting only to their local environment. This is in marked contrast to the prevalent school of thought in evolutionary biology, which holds that organisms adapt only to their local environments.