Appendix 1 - Simulations of geographical trends with an agent based ecological model

Among important recent approaches to solving problems of evolutionary ecology are applications of agent-based models. We have used an agent-based ecosystem model, i.e., the Chowdhury model, to examine latitudinal gradients in species diversity (Rohde and Stauffer 2005), Rapoport's rule (Stauffer and Rohde 2006) and the latitude-niche breadth hypothesis (Stauffer et al. 2007). Application of one particular kind of model cannot be expected to give a definitive "explanation" of any ecological or zoogeographical pattern, but it may significantly contribute to such an explanation, if supported by other evidence. The Chowdhury ecosystem model (Chowdhury and Stauffer, 2005; Stauffer et al. 2005) is one of the most complex agent- or individual-based (Billari et al. 2006) ecological models (Pekalski, 2004; Grimm and Railsback, 2005). It was recently reviewed in Stauffer et al. (2006). It permits simulations over evolutionary (millions of years) time scales at the species level and above, and over much shorter ecological time scales at the population level. In agent based models, each individual is treated separately with its own random birth and death, instead of by a differential equation describing how the total number of individuals changes with time. Sometimes, differential equations can give qualitatively wrong results compared to the more realistic individual treatment.