008 Evolution experiments with an Artificial Ecosystem

Each organism is attached to some place in the world. It should be emphasized that the organism is not located only at this place, and the place does not correspond to an ecological niche. (Certainly such niches must not be artificially imposed by the programming.) Rather, the organism is allowed to operate over a certain number of contiguous places (its territory), but is associated with the place of attachment for certain of its activities and at the end of every period or cycle of the program.

The temporal organization of the model is not in real physical time since events occur at discrete periods. The periods are not generation times, and the lifetime of an organism can extend over any number of these periods. However, the consequences of an organism’s behavior must depend on the behavior of other organisms — that is, it must appear as if organisms are functioning simultaneously. This is achieved, inside the computer, by using the device of a two-pass system, allowing interaction among processes arbitrarily separated by the sequential operations of the computer. In the first pass organisms interact with the environment and with other organisms in local sequence time. The consequences of an organism’s behavior in what corresponds to global, physical time are determined during the second pass. Here chips are collected, organisms reproduce, and chips from decaying organisms (detritus chips) are returned to the matter pool. The net effect of birth and decay determine the composition of the new biota, and the process is repeated. In the actual program the second pass consists of a number of subpasses. The overall flow of information in the program is illustrated in Fig. 1.

The organisms have a genome and a phenome. The genome is mapped into the phenome according to a doublet code. This mapping is not designed to represent the known processes of protein synthesis, or embody any particular logic of self-reproduction. However the genotype-phenotype distinction serves as a basis for efficiently describing various representative strategies of construction and interaction, which is the same function it performs in real cells.