Fig. 2: The duration of the vegetative phase, a trait measured in most breeding schemes, was not better related between different fields than other traits compared indoor and in the field in Figs. 1–4.
From: Robotized indoor phenotyping allows genomic prediction of adaptive traits in the field
a, b Comparison of observed values in 3 field experiments. c Comparison of observed mean genotypic values and mean predicted values (G-BLUPs) in a 5-fold cross-validation scheme with 10 iterations. d Comparison of observed mean genotypic values and predicted values, in the independent dataset. Observed values originated from data of a, b (BLUEs) and G-BLUP model calibration was performed using dataset of c. In a, b and d, light blue circles, mid-early hybrids (G2), dark blue squares, intermediate hybrids (G3), red triangles, mid-late hybrids (G4). In c, purple empty circles, diversity panel, red and yellow empty circles, genetic progress panel, hybrids released before 1980 and 2000, respectively; green empty circles, hybrids released after 2000. In a, r = 0.69 (95% CI = 0.52–0.81), n = 53, df = 51, p-value = 9.4E-09, CVRMSE = 4.4%. In b, r = 0.47 (95% CI = 0.23–0.65), n = 55, df = 53, p-value = 0.0003, CVRMSE = 7%, In c, r = 0.84 (95% CI = 0.81–0.89), n = 302, df = 300, p-value < 2.2E-16, CVRMSE = 2.7%. In d, r = 0.71 (95% CI = 0.56–0.82), n = 60, df = 58, p-value = 1.5E-10, CVRMSE = 2.5%. Significance of the correlation coefficients was tested using two-sided t-test. For rho and other statistics, see Supplementary Tables 4 and 5. Source data are provided as a source Data file.
