The Species Problem in Artificial Life

This paper deals with the species problem in an ALife approach. We present the problem and identify three paradigmatic species concepts that permit different operationalizations. We use the FATINT system to test and compare them in a fully embedded, agent based environment with complete information about lineages, individual phenotypes and reproduction capabilities. We find that under mild conditions (we propose to call them the Darwin condition) the different species concepts tend to yeld similar or identical categorizations. We outline further studies using different models, various species concepts, and new testable hypotheses.