Fig. 6

Scheme illustrating how the optimized training set populations were formed. The algorithms APY (to define the number of information) and LA-GA-T (to select information optimally) were applied to the kernels GWT (genotype x environmental covariates x trait) and GET (genotype x environment x trait) to obtain three replications for each kernel, which formed our OTS 1 (optimized training set 1). From OTS 1, two other scenarios were created: OTS 2 and OTS 3. The whole process was done for the two datasets, HEL and USP