Integrating phenomic selection using single-kernel near-infrared spectroscopy and genomic selection for corn breeding...

Area of Science:

• Plant breeding
• Quantitative genetics
• Agricultural science

Background:

• Phenomic selection (PS) predicts complex traits using phenomic data, complementing genomic selection (GS).
• Traditional PS with near-infrared spectroscopy (NIRS) often requires destructive sampling or post-selection data collection.
• Nondestructive, single-kernel NIRS offers a more efficient approach for PS in breeding programs.

Purpose of the Study:

• To explore the application of nondestructive, single-kernel NIRS for PS in sweet corn breeding.
• To predict unobserved, field-based traits using phenomic data from intact seeds.
• To evaluate the predictive ability of PS models and compare them with GS models.

Main Methods:

• Employed genomic best linear unbiased prediction (GBLUP), phenomic best linear unbiased prediction (PBLUP) using NIRS data, and a combined model.
• Utilized SNP and NIRS data to construct relationship matrices for prediction models.
• Applied PS to select doubled haploid (DH) lines for germination, validating predictions with observed data.

Main Results:

• PS demonstrated good predictive ability for traits like plant height (e.g., 0.46).
• PS successfully distinguished between high and low germination rates in untested DH lines.
• Combined genomic and phenomic models achieved the highest predictive ability, outperforming GS at low marker densities.

Conclusions:

• Nondestructive, single-kernel NIRS is a viable tool for PS in corn breeding.
• Combining genomic and phenomic data maximizes predictive accuracy and genetic gain.
• PS offers a cost-effective alternative or complement to GS, especially when genotyping is limited.