From the CART model, the relationship between the input and output and the interaction between predictor variables can be perceived by starting at the top of the tree and moving down along the branches until reaching the terminal/output node. The relative position of the predictors in the tree was indicative of the fact that LON, SR, GLSR, and PD were again the most important variables in predicting disease severity classes using the CART modeling approach (Figure 2.2). PC was unimportant. The highest disease severity (>80%) occurred when the most susceptible hybrids (with resistance ratings less that 3) were planted at sites located between longitudes 91.29W and 91.41W, or at sites located east of longitude 91.29W and had surface residue cover greater than 60%. For all locations west of longitude 91.41W, disease severity was between 0 and 20%.