A precise technique is needed to determine the lowest possible rates of herbicide applications to opti-mize weed control and minimize damage. An Artificial Neural Network (ANN) was applied to test the effects of different doses of metribuzin on winter wheat. Tests were conducted at two locations of Iran, Shahrood and Mashhad. The metribuzin herbicide was applied to winter wheat pre (PRE) and post-emergence (POST) in both experiments. PRE and POST emergence treatments studied in both lo-cations each consisted of twenty doses of herbicide. In training Multilayer Perceptron ANN, test and total phase P-value indicating that there was no significant (p<0.05) difference between observed and estimated statistical parameters such as average, variance and statistical distribution. Dry matter re-duction was as much as 70% and 85% in the PRE emergence applications of metribuzin at doses of up to 112 and 455 g a.i. ha-1 at two locations respectively. Reduction in dry matter was about 30% in doses of 455 and 280 g a.i. ha-1 of metribuzin when applied POST emergence at Shahrood and Mash-had, respectively. Based on our results, The Multilayer Perceptron ANN should be considered a standard technique for the analysis of dose-response relationship involving time of herbicide activity.

Keywords