The relationships coupling ecologically important processes to pattern formation at different spatial and temporal scales can be explored through a multidisciplinary approach that involves observation, experimentation, and theory. We are using this approach in the study of a serpentine grassland located in the Jasper Ridge Biological Preserve of Stanford University [California USA], and we present preliminary results for that system to illustrate the approach. We argue that the development and analysis of simplified null models is an essential precursor to the successful development of more detailed models. Field and experimental studies can be used to examine the mechanisms producing pattern at small spatial and temporal scales; these then form the basis for spatially explicit, landscape level models. Simplified version of the models can be used as computational experimental tools to explore the consequences of varying disturbance regimes and other critical influences on pattern formation; more detailed versions can be used as limited predictive tools and to guide experimental work.