However, such a model failed to reproduce the domain arrangement observed in Strontium ranelate actual meristems for the model��s parameter space which we explored using a stochastic parameter estimation technique, namely an evolutionary algorithm. For the evolutionary algorithm, we used an objective function described previously, extended by a simple domain recognition procedure capable of identifying circular domains. This indicates that an essential component was missing from this model. The most common outcome which we achieved was not juxtaposition, but an overlap of the SCD with the OC. To improve the spatial separation of the two domains, we considered that cells within an actual meristem differ from each other by their position. Several misexpression experiments using WUS had previously shown that the cellular response to WUS-derived signals depends on a cells�� relative position within the meristem, corresponding to its developmental trajectory. Only by adding spatial components to our model we were able to achieve a realistic sizing and arrangement of the two domains within the meristem; removal of this spatial component causes extensive spatial overlapping of the SCD and OC. Jo��nsson et al. had previously described a model for the WUS/ CLV3 interaction that concentrated on the generation of the CLV3 domain by a WUS derived signal. This model did not yet include the negative feedback regulation of CLV3 signalling upon WUS expression, and the creation and maintenance of the WUS domain was not simulated. To confine the CLV3 domain to the meristem tip, the authors proposed that an factor diffusing from the outermost meristem layer, the L1, together with the WUS-dependent Aconine signal, induced CLV3 expression. They later used a reaction-diffusion model combined with two repressive signals, derived from the L1 and stem tissue, to activate WUS expression in a deeper meristem region. Both models were successful at reproducing either the OC or the SCD, but did not incorporate the mutual interdependence between the factors that shape the two domains, and were less parsimonious with system components than the model we describe here.