**3. Discussion**

Cross-validation remains the most popular method to evaluate GP accuracy [14,28]. Our RR-BLUP model prediction accuracy of 0.92 for PR is the highest of all published GP models for plant disease resistance traits [14]. This model is especially valuable because PR has low heritability and high inheritance complexity [3,4]. The QTL markers, multi-year phenotypic data and the genetic diversity and size of the population likely contributed positively to this high prediction accuracy [29].
