Phenomic selection in soybean breeding

Abstract

Phenomic selection is a promising complement and alternative to genomic selection for improving breeding efficiency. The major advantage of using near-infrared spectroscopy (NIRS) to indirectly capture phenotypic variants and predict complex traits is its high-throughput and low cost. Using NIR spectra to predict individual performances in the context of breeding for yield remains relatively novel. Certain wavelengths of NIR light are absorbed by specific chemical bonds that constitute the components of tissues. The relative proportion of each of these bonds within the tissue quantitatively influence the nature of the absorbance or reflection of light at different wavenumbers. Phenomic selection was tested on 206 soybean genotypes, collecting yield and NIRS data. Spectra were obtained from different tissues, grains and dried, milled leaves, measuring absorbance in range 4000 – 10000 cm−1. RR-BLUP model was used for phenomic predictions, considering NIRS data instead of molecular.