Figure 2From: A comparison of principal component regression and genomic REML for genomic prediction across populationsPattern of accuracies for principal component regression models with increasing numbers of principal components (PC). PCA was performed in the reference dataset. Selection of PC was based on eigenvalues (left panel) or on the sum of square contributions (right panel). Traits analysed were average daily milk, fat and protein yields for test populations from Ireland (IRL), the Netherlands (NLD), Sweden (SWE) and two divergent selection lines from United Kingdom (UK_1 and UK_2).Back to article page