Experimental methods in plant competition research in crop-weed systems

The implications of recent studies on the inappropriatness of replacement series and additive methods in competition studies, and some possible alternatives, are discussed in the context of weed research, Replacement series are usually inadequate to assess competitive interactions and can be misleading. In particular they may be biased in favour of the larger species. Many of the criticisms of replacement series also apply to additive experiments. Response models relating yield per individual to the densities of the species in the mixture provide methodology for answering many questions about mixtures. This paper proposes a framework for using these models to: (i) measure the effect of weed species on yield per individual and yield per unit area for the crop and weed species; (ii) develop methods of biological control of weeds both within a seaons and over seasons, based on the interference between crop and weed species and the population biology of the weed species; (iii) establish a cost-benefit analysis of certain of the biological weed-control programmes. The inclusion of the effect of relative emergence time and management practices in response models is considered as well as experimental design for crop-weed experiments.