Use of foliar symptoms and plant height to predict yield loss in potatoes due to metribuzin injury

A model to predict yield loss due to metribuzin injury was developed using data from four experiments involving postemergence applications of metribuzin to numerous cultivars and breeding selections. Two explanatory variables, plant height proportion (PHR), defined as injured/uninjured, and percent foliar injury (FI), were used to predict yield loss. Three multiple regression models, including linear, quadratic, and a linear-log were considered and evaluated. Acceptance of the best model was based on comparison of fit as indicated by Residual Mean Squares, Adjusted R², and Predicted Sums of Squares. The best fit was for a linear-log model utilizing percent foliar injury and a natural log transformation of the plant height proportion (injured/uninjured). The predictive model was statistically validated by utilizing residual analysis on an independent data set. The model can be used to simply and quickly predict yield losses due to metribuzin injury