Realized niche spaces and functional types: a framework for prediction of compositional change

High resolution and comprehensive quantitative plant and environmental data bases permit an iterative screen- ing approach to provide greater freedom from a priori classification of plant functional types. Realized niche spaces are generated at various plant densities for species across a wide range of environmental conditions in South Africa. Resultant response surfaces suggest different putative func- tional types. Determination of traits in common within such types provides testable hypotheses for application to other species from the available species pool. The approach pro- vides a potentially powerful tool for grouping and regrouping species according to different functional frameworks at re- gional level. This includes deriving types that are sensitive to