Modelling Interactions between Nitrogen Fertilizer and Herbicide in Jimsonweed (Datura stramonium L.) and Redroot Pigweed (Amaranthus retroflexus L.)

In order to select appropriate combined model to describe the relationship between herbicide dose–response and nitrogen fertilizer levels in predicting jimsonweed and redroot pigweed control a experiment was laid out in a factorial design based on complete randomized design (CRD) with three replicates in research greenhouses of Agriculture Campus of Tehran University, Karaj in 2009. The four levels of nitrogen (0, 90, 180 and 360 kg ha-1 urea fertilizer) and five levels of Nicosulfuron (0, 0.5, 1, 1.5 and 2 L.ha-1) were applied. Analysis of variance demonstrated the there were significant effects of fertilizer and herbicide treatment and interaction between herbicide and nitrogen on both of weed biomass.‌ The standard dose–response curve was considered for base model. Then efficiency of this model assesses to weed biomass determination at each nitrogen level that accuracy of this model confirms. For determination of relation each of parameters of base model with fertilizer levels, possible models were considered and appropriate models as sub-model were settled at base model, ultimately final combined model presented. The biomass of both weeds at no-herbicide treatment (Wo) showed a different behavior with increasing nitrogen. Trend of these changes at nitrogen levels was well described by the linear and quadratic models for redroot pigweed and jimsonweed, respectively. Increasing nitrogen didn’t change herbicide dose-response of jimsonweed, because steepness of the curve (β) and the effective dose required to reduce weed biomass by 50% (ED50) parameters of this weed not showed a clear behavior with increasing nitrogen. Therefore standard dose–response curve with constant β and ED50 parameters was best describing of jimsonweed biomass as affected by both the herbicide dose and nitrogen level. But standard dose–response curve of redroot pigweed was modified by replacing the parameter β and ED50 with the exponential and linear curves, respectively. The final presented models can be used to predict weed control by an herbicide as affected by both the herbicide dose and nitrogen level.