Euphytica Генетика

Integrating GWAS and NIR-based phenotyping reveals candidate loci associated with grain quality traits in sorghum

Zhou, Ying, Liu, Yansheng, Qin, Xiner, Huang, Yuanxu, Li, Cong, Deng, Zhuying, Zeng, Gongjian, Mao, Hailiang, Hu, Yongfeng, Li, Yin, Shen, Xiangling
Springer Nature · 2026-11-16 · DOI 10.1007/s10681-026-03837-z

Аннотация

Sorghum ( Sorghum bicolor (L.) Moench) is a crucial cereal crop for food, feed, brewing and bioenergy, and grain traits and quality breeding underpin its diversified applications, yet the correlations between its grain morphology and quality remain underexplored, hindering synergistic breeding. To fill this gap, we evaluated a diverse panel comprising 774 sorghum accessions to dissect the genetic basis of grain-related traits. Population structure analysis revealed extensive admixture. Field evaluations were conducted in 2023 at Lingshui, Hainan Province, and Yichang, Hubei Province, China. A core collection of 280 accessions selected on the basis of SSR clustering was subjected to whole-genome resequencing, generating over 13 million high-quality SNPs. Genome-wide association study (GWAS) identified key loci for grain morphology, including SbSLP ( Sobic.003G269900 ) and SbSEP ( Sobic.009G256700 ). For grain physicochemical traits, SbSGPP1 ( Sobic.010G007300 ) was identified as a candidate gene associated with tannin content. A near-infrared spectroscopy (NIR) model was constructed using 140 samples, and 29 independent sample sets were used for validation. The root mean square errors for the validation of starch and amylose were 9.55% and 22.13% respectively, indicating significant prediction errors. GWAS based on sequencing and NIR phenotypic prediction identified significant associations with SbKAO ( Sobic.010G007700 ). Haplotype differences supported the candidate association, but failed to determine the gene function or verify the accuracy of the NIR prediction. These findings prioritize the candidate loci that require further study, and also highlight the necessity of improving the NIR prediction ability and functional validation.

Читать у издателя →