Relationships between crop yield and weed time of emergence/removal: modelling and parameter stability across environments

Seventeen Italian experiments relating to maize, soyabean and durum wheat were used to analyse the variability of duration of tolerated competition (DTC) and weed-free period (WFP) curves across reasonably homogeneous areas. The data sets were analysed by regression analysis using four models relating yield loss to weed density, time of emergence and removal. These models differ in the way they account for the effect of time of weed removal on potential competitiveness. A sigmoidal relationship between these two variables appears necessary. The model with the best overall performance was then used to test the stability of the parameters that give the shape of the curve in relation to time of emergence and removal. This was done by comparing a full model with shape parameters specific to each experiment and a reduced model with a common set of parameters for all the experiments referring to a specific crop. For all three crops, the residual sum of squares of the reduced model did not increase significantly, indicating that, across tested environments, the yield loss caused by mixed weed infestations can be expressed by a single set of parameters relating weed competitivity to time of emergence and removal. For a given area, it should therefore be possible to predict yield loss on the basis of a quite limited set of experiments, thus greatly simplifying the development and use of decision support systems (DSS).