Table 4 Predictive abilities from BayesB and RR-BLUP for simulations with different numbers of QTL and heritabilities
Rice (n = 413)Wheat (n = 254)Arabidopsis (n = 199)
h2No. QTLBayesBRR-BLUPBayesBRR-BLUPBayesBRR-BLUP
0.110.16 ± 0.0350.17 ± 0.0330.17 ± 0.0210.18 ± 0.0200.08 ± 0.0380.06 ± 0.030
100.26 ± 0.0250.26 ± 0.0210.14 ± 0.0230.13 ± 0.0180.07 ± 0.0250.08 ± 0.032
1000.30 ± 0.0120.30 ± 0.0140.13 ± 0.0340.22 ± 0.0220.09 ± 0.0310.09 ± 0.038
0.510.68 ± 0.0120.56 ± 0.0320.68 ± 0.0080.50 ± 0.0210.64 ± 0.0270.27 ± 0.035
100.65 ± 0.0060.65 ± 0.0080.50 ± 0.0120.47 ± 0.0140.33 ± 0.0240.27 ± 0.015
1000.71 ± 0.0040.71 ± 0.0040.48 ± 0.0270.55 ± 0.0230.29 ± 0.0300.27 ± 0.034
0.910.94 ± 0.0030.81 ± 0.0180.94 ± 0.0020.75 ± 0.0200.94 ± 0.0030.43 ± 0.039
100.94 ± 0.0010.88 ± 0.0140.93 ± 0.0030.74 ± 0.0080.92 ± 0.0090.51 ± 0.029
1000.94 ± 0.0020.94 ± 0.0020.78 ± 0.0190.83 ± 0.0150.47 ± 0.0300.50 ± 0.028
  • The simulations were conducted according to procedure 2 (Figure S6). Predictive abilities were estimated with fivefold cross-validation for BayesB and RR-BLUP based on marker genotypes for 2000 randomly selected markers and 1, 10, and 100 simulated causal mutations. We report the average predictive ability ± SE of 10 replications for each scenario.