Competition and relative yield: estimation and interpretation at different densities and under various nutrient concentrations using Silybum marianum and Cirsium vulgare

(1) The role of the yield-density relationship in determining the choice of an additive or substitutive experimental design for competition experiments is examined, with particular reference to the design of multispecies experiments. A competition experiment between two thistle species Silybum marianum and Cirsium vulgaris at five density levels and six nutrient concentrations is described. (2) The design allows the influence of density and nutrient concentration on the relative yield in mixture compared with monoculture (RY) of the two species (as estimated from a substitutive design of N/2 plants in mixture) to be examined, and compared with that estimated from an additive design of N plants in mixture, where N is number of plants in monoculture. (3) Yields of the two species in monoculture show similar significant responses to both nutrient concentration and density: maximum yield occurs at a nutrient concentration four times standard Hoagland, and response to density is asymptotic. S. marianum has similar performance in mixture while C. vulgare yields are markedly reduced. Nutrient concentration has a significant influence on the RY of both species. A clear influence of density is only apparent at extremely high nutrient concentrations. (4) Acceptance of an asymptotic yield-density function implies that RY values between 0.5 and 1.0 cannot be interpreted unambiguously as being due to competition in substitutive experiments. Analysis of the substitutive design for S. marianum gives RY between 0.5 and 1.0, which cannot be interpreted as due to competition from C. vulgare. The additive design provides no evidence for such a competitive effect. RY values for C. vulgare are less than 0.5 in both designs indicating a competitive effect by S. marianum. (5) Choice of substitutive or additive design depends on knowledge of the yield-density function. Unequivocal results require a range of density combinations to be included in the design.