Table 3 MAIZE2 data set. Average Pearson’s correlations between observed and predictive values (and their standard deviation in parentheses) for 3 methods for single-environment and GE multi-environment models for 50 random partitions with 70% of the lines in the training set and 30% of the lines in the testing set. Methods GB, GK, and AK are the GBLUP, Gaussian Kernel and Arc-Cosine Kernel, respectively. The best predictive model for each environment (E1-E5) is in boldface
Single-environment model
EnvironmentGBAKGK
E10.647 (0.07)0.755+* (0.05)0.733 (0.05)
E20.384 (0.07)0.533+* (0.06)0.488 (0.07)
E30.678 (0.03)0.759+* (0.02)0.718 (0.04)
E40.368 (0.03)0.502+* (0.06)0.469 (0.05
E50.393 (0.07)0.564+* (0.06)0.522 (0.06)
GE multi-environment model
E10.679 (0.08)0.811+ (0.03)0.810 (0.05)
E20.476 (0.09)0.600+ (0.05)0.600 (0.06)
E30.694 (0.05)0.784+ (0.03)0.784 (0.03)
E40.409 (0.09)0.547+ (0.05)0.550 (0.06)
E50.478 (0.09)0.643+ (0.04)0.644 (0.04)
  • * Significant at the 0.05 probability level of the t-test for the hypothesis that the average of the correlation of kernel AK is superior to the mean of the correlation of kernel GK.

  • +Significant at the 0.05 probability level of the t-test for the hypothesis that the average of the correlation of kernel AK is superior to the mean of the correlation of kernel GB.