Evolution of Digital Ecosystems

Most Avida experiments to date use only single-niche (that is, single species) populations, but the evolutionary process becomes far more interesting (and more powerful) when we consider ecological communities. The selective pressures that cause the formation and diversity of ecosystems are still poorly understood [Schluter01, Tilman00]. In part, the lack of progress is due to the difficulty of performing precise, replicated and controlled experiments on whole ecosystems [Morin02]. To study simple ecosystems in a laboratory microcosm (reviewed in [Travisano00]), biologists often use a chemostat, which slowly pumps resource rich media into a flask containing bacteria, while simultaneously draining its contents to keep the volume constant. Unfortunately, even in these model systems, ecosystems can evolve to be more complex than is experimentally tractable and understanding their formation remains difficult [Notley99a, Notley99b, Rainey98]. We set up Avida experiments based on this chemostat model [Cooper02] wherein 9 resources flow into the population, and 1% of unused resources flow out. We used populations with 2500 organisms, each of which absorbed a small portion of an available resource whenever they performed the corresponding task. If too many organisms focus on the same resource, it will no longer be plentiful enough to encourage additional use.

Theory predicts that an environment with either a single resource or with resources in unlimited quantities is capable of supporting only one species [Tilman82], and this is exactly what we see in the standard Avida experiments. It is the competition over multiple, limited resources that is believed to play a key role in the structuring of communities [Schluter96, Travisano00]. In 30 trials under the chemostat regime in Avida a variety of distinct community structures developed [Cooper02]. Some evolved 9 stably coexisting specialists, one per resource, while others had just a couple of generalists that divided the resources between them. Others still mixed both generalists and specialists. In all cases, the ecosystems proved to be stable because they persisted after all mutations were shut off in the system, and if any abundant phenotype were removed, it would consistently reinvade.

Phylogeny visualizations provide a striking demonstration of the differences between populations that evolved in a single niche and those from ecosystems, as displayed in the figures above. Single niche populations can have branching events that persist for a short time, but in the long term one species will out compete the others, or simply drift to dominance if the fitness values are truly identical. By contrast, in ecosystems with multiple resources branches corresponding to speciation events persist.