Best linear unbiased prediction of maize single-cross performance given erroneous inbred relationships

Best linear unbiased prediction (BLUP) of maize (Zea mays L.) single- cross performance requires estimates of genetic relationship among parental inbreds. Error in estimates of genetic relationship may result from selection, genetic drift during inbreeding, or incomplete or unknown pedigrees. The objective of this study was to investigate the robustness of BLUP when estimates of genetic relationship among inbreds are erroneous. Grain yield, moisture, stalk lodging, and root lodging data were obtained for 2043 maize single crosses evaluated in the multilocation testing program of Limagrain Genetics, from 1990 to 1994. Three types of errors in inbred relationships were studied: (i) simulated random deviations from the expected contributions of parents to progeny; (ii) unknown parentage for 10% of the inbreds; and (iii) unknown parentage for 25% of the inbreds. Malecot's coefficients of coancestry (f) were calculated by tabular analysis of the correct and erroneous pedigrees and were subsequently used in BLUP. Average absolute deviations between erroneous f and f calculated from correct pedigrees ranged from near-zero to ≃0.30. But the correlations between predicted and observed single-cross performance were not affected by the relatively large errors in f. Most of the differences, caused by erroneous f, in the correlations between predicted and observed performance ranged from 0.001 to 0.030 and were too small to have any practical significance. These results indicated that BLUP is robust when inbred relationships are erroneously specified. Estimating f with molecular markers, prior to predicting the single-cross performance of inbreds with incomplete or unknown pedigrees, seems unnecessary.