Comparison of BLUE and BLUP/REML in the selection of clones and families of potato (Solanum tuberosum).

Abstract

The use of best linear unbiased prediction/restricted maximum likelihood (BLUP/REML) in perennial crops and animal breeding enhances selection gain. However, its advantage with respect to annual crops is not clear. We compared the BLUP and best linear unbiased estimator selection efficiency in the breeding of various potato generations. This was done by simulating various selection intensities on clonal families (full sibs), and clones. The characters evaluated were tuber yield and tuber specific gravity. Two criteria were adopted for comparison: a) incidence of families or clones and b) selection gain. For tuber yield, BLUP/REML method was slightly more efficient for selecting families in the first clonal generation, if it were above 50%. Below this value, both methods were equivalent. However, they both presented equal behavior for family selection of tuber specific gravity. For clonal selection, BLUP/REML showed robust superiority from 10 to 90% selection intensities in both characters. Therefore, the adequate use of BLUP/REML in potato breeding can enhance the selection gain on the yield and specific gravity of tubers.