EXPERIMENTAL DISCRIMINATION BETWEEN CONTRASTING MODELS OF NEIGHBORHOOD COMPETITION

1. Previously published competition models have been categorized into four types, based on the territorial extent over which individual plan ts are assumed to draw resources. 2. The different types of models giv e different expectations for the growth responses of focal plants when the spacing of neighbours is systematically varied, thus allowing the different types of model to be discriminated. 3. These systematic spa cings were imposed in three experiments using carrots (Daucus carota L ). The absolute growth rate (AGR) of single plants located at the cent re of a ring of four plants was less than that of such plants encompas sed by three rings of plants. The difference in AGR between these two treatments initially increased but then became constant. 4. These find ings support the 'overlapping domain' type of model and are shown to f alsify the assumptions made by other types of competition model. 5. Ex periments in which the spacing of individual plants were systematicall y varied in contrasting light and nutrient regimes did not support the expectation that a 'non-overlapping domain' type of model would be mo st appropriate to simulate the competitive interactions between plants when light is the resource subject to competition.