Fig. 2

Predicted accuracy of estimated breeding values of the whole candidate population for different reference population types (PB: purebreds in GP2; CB_extreme: two_tailed crossbreds in DLY; CB_random: random crossbreds in DLY) and sizes (500, 1000, 2000, 3000, 4000, 5000, 6000, 6500). a GBLUP method, \({h}^{2}\) = 0.5, b BSLMM method, \({h}^{2}\) = 0.5, c GBLUP method, \({h}^{2}\) = 0.3, d BSLMM method, \({h}^{2}\) = 0.3, e GBLUP method, \({h}^{2}\) = 0.1, f BSLMM method, \({h}^{2}\) = 0.1. Predicted accuracy was averaged across 50 replications for scenarios involving randomization