Fig. 4.

Reliability of five genomic prediction algorithms (G-BLUP, RR-BLUP, BayesB, RKHS, LASSO) and of WhoGEM method to predict quantitative traits in M. truncatula. Reliability is estimated using 50 rounds of fivefold cross-validation (repeated k-fold cross-validation). a Maximum symptom score for the response to Verticillium alfalfae. b Plant height. c Number of leaves. d Total number of nodules. e Number of nodules below 5 cm of the root. f Number of nodules above 5 cm of the root. g Root rot index for the response to Aphanomyces euteiches. If the notches of two boxes do not overlap, this suggests that the medians are significantly different. Letters identify significantly different groups, with Kruskal-Wallis rank test at α = 0.01