Relations between community theory and community analysis are elucidated from a historical viewpoint. Four groups of relevant aspects are distinguished. This paper concentrates on aspects of spatial variation in species composition and species distribution relations.
Three major elements of community theory are discussed: the response of species to habitat conditions, interactions between species (competition, facilitation), and chance and random processes, both in the environment (disturbance) and in populations (dispersal). The debate island biogeography is used as a starting point for reviewing some theories about vegetation in terms of the elements they involve.
Early community theories are discussed with emphasis on Clements ('competition', 'facilitation') versus Gleason ('chance'), and Braun-Blanquet ('habitat') versus Du Rietz ('habitat', 'competition', 'chance'). Du Rietz's theory of vegetation is manysided and well ahead of its time. Then the development and significance of quantitative plant ecology for community theory is treated with emphasis on Whittaker ('habitat', 'competition') versus Goodall ('habitat', 'facilitation'). The potential use of objective and quantitative methods of analysis for hypothesis testing should be explored much further.
New theoretical ideas are summarized, including the role of the regeneration cycle of individual plants and neighbour interactions. Both interaction experiments and models for community dynamics will contribute to the general theory of plant communities. Perspectives for the development and testing of null hypotheses in vegetation science are discussed in view of the mutual relationships between community theory and community analysis, and the relations with dominance -diversity studies and multi-species pattern analysis indicated. The simplest hypothesis, i.e. that species combinations occur in sites independently and at random, is of little value. The next step would be that observed co-occurrences are determined by the overlap of species distributions over habitats. Then hypotheses on interaction and facilitation become relevant.