Fig. 4

Cross-validation schemes employed. ST-CV1: single-trait cross-validation scheme where a training set of 70% of random genotypes are phenotyped and genotyped and a testing set of remaining 30% of genotypes are genotyped, not phenotyped; MT-CV1: multi-trait cross-validation scheme where a training set of 70% of random genotypes are phenotyped (primary + secondary traits) and genotyped and remaining 30% of genotypes are genotyped only, not phenotyped; MT-CV2: multi-trait cross-validation scheme where 100% information from secondary traits, a training set of 70% of random genotypes are phenotyped for primary traits and remaining 30% of genotypes as testing set (phenotyped for correlated traits but not primary traits and genotyped)