1 L. M. Hilty, P. W. Gilgen (Eds.).: ”Sustainability in the Information Society”, 15th International Symposium Informatics for Environmental Protection. Metropolis-Verlag, pp. 631-636 Asymmetric cellular automata for the modelling of ecological systems Michael Sonnenschein and Ute Vogel 1 1. Modelling with cellular automata in ecology Cellular automata (Wolfram 1986, Toffoli/Margolus 1987) in combination with individual based or individual oriented approaches (Caswell 1989, DeAnge- lis/Gross 1992) have successfully been used to build spatially explicit models of population dynamics. Current examples are a fox-rabies-model (Thulke et al. 1999), or a grasshopper model (Schr¨oder 2000). Combination of such models with geographical information systems (GIS) allows to integrate real landscape data quite easily. The software tool RAMAS c  GIS for example combines a generic metapopulation model (Hanski, 1999) with a raster-based GIS provid- ing habitat data. As a result it can be used for population viability analysis (PVA) based on concrete data (Ak¸cakaya 1998). Traditionally, cellular automata cover the investigated area with a regular pattern of cells which are identical in shape (usually squares) and size. These cells are related by a simple, regular neighborhood relation. In population ecology, each cell models the local population dynamics by a life-cycle model. The simple structure of such models is undoubtedly an advantage for formal analysis. But – as also discussed in (Wittmann 2000) – we can consider the uniform structure of cells to be an essential disadvantages of this approach: If e.g. concrete spatial situations (for example barriers like streets or rift valleys between neighbored cells) can be modelled at all, then only in simplified man- ner. To get more realistic results, one can use a generalized definition of cells and their neighborhood, as e.g. in cellular systems (Wunsch 1977). A second disadvantage of traditional cellular automata consists in the lack of of a global layer to model common, dynamic conditions (e.g. weather) for all cells. Such ‘global variables’ are important to model correlation between local dynamics of the cells. 1 Carl von Ossietzky Universit¨at Oldenburg, Faculty of Computing Science, D-26111 Oldenburg (Germany) Email: {michael.sonnenschein, ute.vogel}@informatik.uni-oldenburg.de Internet: http://www-ui.informatik.uni-oldenburg.de/ai