Skip to main content
. 2024 Nov 14;19(11):e0310886. doi: 10.1371/journal.pone.0310886

Fig 1. Cross-validation scheme: Same trait in two field conditions.

Fig 1

The training population (TP) is represented in green, and the validation population (VP) in red. (A) Univariate cross-validation 1 (UV-CV1): this scheme was applied for all traits, with 4/5 of the population used for training the model and 1/5 for validation. (B) Multivariate cross-validation 1 (MV-CV1): the phenotypical information for the same trait in both high nitrogen and well-watered (HNW) and low nitrogen and rainfed (LNR) conditions were included in the model for the individuals involved in the TP. The correlation between the genomic estimated breeding values (GEBVs) and the BLUEs was performed separately for the target trait in its HNW and LNR conditions. (C) Multivariate cross-validation 2 (MV-CV2): phenotypic information from the LNR condition of the individuals in the VP was included in the model to predict the HNW condition. (D) Multivariate cross-validation 2 (MV-CV2): the same scenario as in (C), but HNW and LNR conditions reversed.