A Darwinian evolutionary system. 3. Experiments on the evolution of feeding patterns

Another application of the Darwinian evolutionary system introduced earlier is presented. The evolution of spiraling and meandering feeding patterns, as exemplified by the marine polychaete Paraonis fulgens, was simulated by constructing an artificial “worm”, called Rectangulus, because it was capable of turning around 90 ° only. The behavioural program of Rectangulus contained six parameters, which controlled the following properties: (i) turning spontaneously after a given number of steps, (ii) turning before a former track, (iii) avoiding entering a “channel”, i.e. a path flanked by former tracks, (iv) keeping contact with former tracks, (v) performing a turn at the beginning (which in combination with contact-keeping leads to a spiral), and (vi) switching from spiraling to meandering behaviour after a given number of steps. These parameters were binary-coded in the “genomes” of Rectangulus, and were capable of mutating and recombining.Evolutionary experiments were performed with populations consisting of 50 individuals, inhabiting an area of 100 × 100 “unit steps”. Each individual had to “forage” for 140 steps, and the number of successful (uncovered area) and unsuccessful (area covered previously) steps were recorded. Individuals with greater “yield” (successful minus unsuccessful steps, divided by the total number of steps) had a greater probability of being reproduced (selection). It was found that Rectangulus soon “learned” to turn in front of a former track and a channel, and to keep contact, both if one starts with individuals who at the beginning could only go straight ahead, or with ones turning after every step. Evolution from this stage onward proceeded more variably. Different types of meandering, spiraling, and a combination of both emerged; however, in two out of six cases, evolution did not proceed beyond the turning and contact-keeping stage. It is suggested that the different outcomes represent separate adaptive peaks. The highest yield was obtained by the combination of spiraling and meandering, as used by Paraonis fulgens, followed by pure spiraling and meandering. This was confirmed by calculations made for populations with fixed parameters. The results further show that selection for non-crossing could only have been effective in producing these patterns for population densities much higher than the ones found for P. fulgens at the present time.