Prediction of oil palm (Elaeis guineensis, Jacq.) agronomic performances using the best linear unbiased predictor (BLUP)

Abstract 

Reciprocal recurrent selection (RRS) has been adopted for oil palm breeding in Indonesia.Due to a long selection cycle and the large area required, a satisfactory oil palm progeny trial is difficult to conduct.Knowledge of the parental genetic parametersis very important in achieving the expected genetic progress, but the evaluation of these parameters is constrained by highly unbalanced data sets. In this study, the unbalanced agronomic data sets and the pedigree information of an oil palm breeding programme in Indonesia were analysed by using the restricted maximum likelihood (REML) and the best linear unbiased predictor (BLUP)methods. The characters analysed were bunch and oil yields of the adult period (from 7 to 9 years after planting). The coefficients of parentage varied from 0.125 to 0.891 and from zero to 0.750 between parents in the Deli and African groups, respectively.The average coefficients of inbreeding were 0.269 and 0.166 for the parents within the Deli and African groups, respectively. The additive variances of the bunch number, industrial oil-extraction rate and oil yield characters were higher in the parents of the Deli group than those in the African ones. The coefficients of correlation between the predicted and observed hybrids performances varied from 0.55 to 0.64 for oil yield, 0.49 to 0.71 for bunch number, 0.47 to 0.58 for bunch production, 0.48 to 0.64 for industrial oil-extraction rate and 0.42 to 0.56 for plant-height increment. For selection on the basis of oil yield character, BLUPs ability to predict single-cross performance should be sufficient, and will result in a significant contribution to the oil palm seed and clone productions.