The genome wide selection (GWS) was proposed as a way to increase efficiency and accelerate the genetic improvement, emphasizing the simultaneous prediction of the genetic effects of a large number of DNA genetic markers scattered throughout the genome of an organism, in order to capture the effects of all loci and explain the genetic variation of a quantitative character. The objective of this study was to apply the principle of GWS in the cashew breeding, simultaneously estimating the effects of 238 markers assessed in 74 individuals from a full-sib family to explain the large percentage of total genotypic variation of the character kernel weight and increasing the efficiency of cashew breeding. It was found that the predictive power and accuracy are nearly maximized in the analysis with 70 markers to the greatest effect. The increase in the number of markers does not increase linearly the accuracy of the method GWS RR-BLUP. The 70 markers with the greatest effects capture 74% of the total genotypic variation and provide a selective high accuracy (86%) of the selection for kernel weight, while the five markers with the greatest effects capture only 19% of the total genotypic variation and provide a selective accuracy of only 44%. Thus assisted selection (MAS) based on a few (five) markers provide significant effects efficiency much lower than GWS. The predicted genomic genetic values in the population of cross-validation had a good approximation of the observed phenotype values, with a correlation of 0.79. The simultaneous estimation of the effects of the markers according to the concept of GWS is an interesting alternative to enhance the efficiency of cashew breeding.
assisted selection; breeding; random regression; BLUP
