Aggregation of an individual-based space-dependent model of forest dynamics into distribution-based and space-independent models

A question addressed in forest modelling is: which level of detail is required to reproduce the ecosystem function? In this paper four versions of the same model of a natural rain-forest are compared. The most detailed is an individual-based distance-dependent model, where every individual is described and interacts with other trees through a spatially explicit neighbourhood. Performing aggregation on the stand description yields a distribution-based distance-dependent model. Applying the mean field approximation brings two distance-dependent models. Three of the four models appear to predict the same stationary state in the limit where the number of trees is infinite (individual-based distance-independent+distribution-based models). The individual-based distance-dependent model differs because of the occurrence of space-dependent effects. An alternative aggregation method based on moment equations is then suggested to render the space-dependent effects.