We discuss the development of two simulation models, a mechanistic cotton crop model and a spider mite population model. In a strategic mode, the simulation models are used to address basic hypotheses involving the interaction of cotton with its environment, and with the spider mite population, and to develop general strategies for managing the crop and its herbivores. In a tactical mode, these models are used to provide estimates of the anticipated severity of a spider mite population. Tactical modelling is made possible by using a statistically-based adaptive interface. The departure of simulated patterns of growth from observed patterns is used to 'adapt' several crop parameters including rate of photosynthesis, rate of vegetative growth, and metabolite allocation priority for fruit. For the spider mite population model, fecundity, predator-mediated mortality, and acaricide-induced mortality are adapted to specific field conditions. 1992 Elsevier Science Publishers B.V.
