Figure 7.

Evaluation of the multi-trait genomic prediction framework where missing values included within the training partition, with an increasing missing rate ranging from 0 to 90% of focal traits (Ethylene, Sucrose, Glucose, Fructose, Citric.A. and Malic.A.) predicted using proxy traits (F.Weight, Hue.G, RI and TA). Multi-trait strategy was modeled using either phenotypic values (multi-trait model) or genomic estimated breeding values (model-based index) in comparison with single-trait GBLUP model.