Genomic Selection for Biomass Yield of Perennial and Annual Legumes

Abstract

This study aimed to assess the predictive ability of genomic selection (GS) for biomass yield of alfalfa and pea, considering different data sets, GS models, and thresholds for genotype missing data. An additional aim was to briefly devise the incorporation of GS into breeding schemes of these crops. For alfalfa, the predictive ability of best GS models ranged from r = 0.18 to r = 0.36 in three data sets. The lowest value (observed in a data set with higher experimental error or lower genetic variation relative to the other data sets) may still be of practical interest, given the long selection cycle and the low narrow-sense heritability of biomass yield in this crop. For pea biomass yield, the predictive ability of best GS models averaged r = 0.45 across three recombinant inbred line (RIL) populations. Predictions were less accurate for this trait than for pea straw or grain yield. GS is a promising approach but its adoption implies important modifications of alfalfa and pea breeding schemes. We identified five stages of GS-based selection schemes, whose implementation depends largely on the reproductive system of the target species.