Deploying Genomic Selection for Improving Complex Traits in Peanut (Arachis hypogea L.)

Genomic selection (GS) has emerged as the most promising breeding approach for improving complex traits. The main strength of GS is that it can capture several small-effect genetic factors and can also improve multiple traits simultaneously. In this context, GS approach has been initiated in peanut to improve complex traits like yield under drought stress. The ‘minicore collection’ with 184 genotypes was phenotyped for three important agronomic traits (days to flowering, seed weight and pod yield) and genotyped with 15,360 diversity array technology (DArT) features. A total of six GS models (Ridge Regression-BLUP, Bayesian LASSO, Random Forest Regression, Kinship GAUSS, BayesCπ and BayesB) were tested using the phenotypic and genotypic (2,356 polymorphic DArT markers) data for estimation of correlation and cross-validation values. As a result, two best performing GS models namely Ridge Regression-BLUP and Bayesian LASSO were identified for predicting genomic estimated breeding values (GEBVs). Further to deploy GS breeding in peanut, a training population comprising of 310 elite genotypes has been constituted and genotyped with 15,360 DArT/DArTseq features. The population is being phenotyped for several agronomically important traits. The genotypic and phenotypic data will be used to define the appropriate GS model for deploying GS in peanut breeding.