Best Linear Unbiased Prediction (BLUP) for Genotype Performance
Mónica Balzarini, Scott B. Milligan · 2024
Whenever the examined genotypes (varieties, clones, cultivars, etc.) in an experiment can be regarded as random samples from larger sets or populations, genotype effects should also be considered random in the model. Commonly, random genotype effects are assumed to broaden inferences, i.e., to allow inference about a reference population. Moreover, pairwise comparisons between specific genotypes that have been examined will also be feasible. In addition to estimating genotype variance components, which are of intrinsic interest with random genotype effects, genotypes can be compared by calculating best linear unbiased predictors (BLUPs) of genotype effects. In this context, “best” means minimum mean squared prediction error (MSPE). The smaller the MSPE, the greater the relationship between the dependent and independent variables.