Using Best Linear Unbiased Predictions to Enhance Breeding for Yield in Soybean: I. Choosing Parents

Abstract

In serf‐pollinated crops, choosing parents typically is accomplished by calculating parental performance from historical data and then calculating the midparent value (MPV) for potential crosses. When limited or no data exist for parents of interest, precise predictions are difficult or impossible to obtain. Best linear unbiased prediction (BLUP), has been used to determine paired matings in dairy cattle under conditions described above. The objectives of this study were to compare the elficiencies of two methods of parental selection, MPV and BLUP, for identifying superior soybean [Glycine max (L.) Merr.] cross combinations when (i) equal and unequal amounts of yield data on all potential parents were available, and (ii) unequal amounts of yield data were available for some parents and no data were available for others. F4–F6 bulks and F5:6 lines from 24 soybean crosses were evaluated to estimate the mean yield performance of each cross. Historical yield records on the parents of each cross were used to predict the performance of the 24 crosses. Numbers of records on the parents were restricted to provide simulated situations of balanced and unbalanced parent data availability. The performance of each cross was predicted with MPV and BLUP for each situation. Standard errors of the predicted differences (SE) and rank correlations between the actual and the predicted performances were calculated to determine the relative efficiencies of MPV and BLUP. In every case, predictions from BLUP provided higher rank correlations, lower SE, and identified higher percentages of the superior crosses than MPV.