Resource capture by crops.

Interplant competition for capture of the essential resources for plant growth i.e. light, water and nutrients, strongly affects the performance of natural, semi-natural and agricultural ecosystems. Ecologists have studied competition to understand the diversity and stability of plant communities, succession patterns of vegetation, and to help define management strategies for semi-natural ecosystems. Agroecologists have studied competitive phenomena to optimize plant densities of crops, to optimize intercropping systems and to quantify crop-weed interactions to improve weed management systems with minimum herbicide inputs. Similar approaches have been used to study interplant competition in natural and disturbed systems. However, because of the complex nature of interplant competition, it has taken a long time to develop generalized concepts and theories. The first systematic approaches for studying competitive phenomena were developed in the 1960s. For monocultures, much of our current understanding is based on the work of a Japanese group (e.g. Shinozaki and Kira, 1956) whereas de Wit (1960) developed the first systematic approach to study competition in mixtures. He introduced an experimental design (the replacement series in which the mixing ratio varies, but total density remains constant) with a model to analyse the results. These approaches were based on a hyperbolic relationship between plant density and yield, and have been used extensively in agricultural and ecological sciences to study competition between plants, plant population dynamics, and component contributions of intercropping systems (see reviews by Trenbath, 1976; Harper, 1977; Radosevich and Holt, 1984; Grace and Tilman, 1990). Recently, several papers discussed the disadvantages and pitfalls of the replacement series approach, such as its total density dependence ( cf. Connolly, 1986; Taylor and Aarssen, 1989). Only in the early 1980s, approaches were developed to describe competition over a range of population densities with varying mixing ratios and at a range of total densities, generally also based on the hyperbolic yield-density relationship (Suehiro and Ogawa, 1980; Wright, 1981; Spitters, 1983a, b; Cousens, 1985; Spitters, Kropff and de Groot, 1989). Similar approaches have been developed using the neighbourhood approach, in which the number of neighbours of an individual plant in a predefined area is related to the weight of the central plant (Silander and Pacala, 1985; Firbank and Watkinson,