Table 6 Pearson correlation of predictions of a test set consisting of 60 lines when predicted by additive and interaction effect estimates based on a model including 50 randomly chosen markers and their pairwise interaction effects. The numbers represent the means of 200 repetitions with independently, uniformly drawn marker and test and training sets (The latter of size 539, 300 or 200). The standard error was in all cases smaller than 0.022. Colors indicate the underlying data: Environment 1, 2, 3 or 4. The case of a training set size of 539 is the case of 50 markers in Table 5.