Genomic selection to introgress exotic maize germplasm into elite maize in China to improve kernel dehydration rate

• Published in: Euphytica 2021-08, Vol.217 (8), Article 168
• Main Authors: Yong, Hongjun, Wang, Nan, Yang, Xiaojun, Zhang, Fengyi, Tang, Juan, Yang, Zhiyuan, Zhao, Xinzhe, Li, Yi, Li, Mingshun, Zhang, Degui, Hao, Zhuanfang, Weng, Jianfeng, Han, Jienan, Li, Huihui, Li, Xinhai
• Format: Article
• Language: English
• Subjects: Agricultural economics; Biomedical and Life Sciences; Biotechnology; Corn; Crop yield; Dehydration; Dehydration (Physiology); Gene frequency; Genetic improvement; Genomics; Germplasm; Grain; Harvesting; Inbreeding; Kernels; Life Sciences; Mathematical models; Moisture content; Plant Genetics and Genomics; Plant Pathology; Plant Physiology; Plant Sciences; Population; Population (statistical); Predictions; Single-nucleotide polymorphism; Statistical models; Water content
• ISSN: 0014-2336; 1573-5060
• DOI: 10.1007/s10681-021-02899-5

Genomic selection (GS) is an efficient way for trait improvement in crops. GS for kernel dehydration rate (KDR) has not been reported until now. The elite single-cross hybrid Zhengdan958 is the most widely planted hybrid in China, but has slow KDR and high grain moisture at harvest that seriously hamper mechanical harvesting efficiency. The present study aimed to determine whether GS is an effective strategy for improving KDR in an exotic × adapted population and to identify a lower-cost SNP panel and suitable statistical model for GS prediction. Here, the elite U.S. population BS13(S)C7 was crossed to inbred line Zheng58 to establish a training population that was then testcrossed to the inbred Chang7-2. Phenotypic traits including days to anthesis (DA), ear height (EH), water content of the ears (WC), KDR, and grain yield (GY) were measured in two locations during 2016 and 2017. We found that the rrBLUP model using 24,435 filtered SNPs with minimum call rate > 50% and minor allele frequency > 0.05 resulted in the highest prediction accuracy. Further, a subset of 5000 SNPs randomly selected from 24,435 high-quality SNPs provided a lower-cost SNP panel with sufficient prediction accuracy for GS. The breeding efficiency of GS compared with conventional selection varied from 0.28 to 0.66. Predicted genetic gains were − 0.15%, − 1.42%, − 0.64%, 1.89%, and 1.30% for DA, EH, WC, KDR, and GY, respectively, indicating that GS was adequate for improving KDR and other important traits, with advantages over pedigree breeding for both simple and complex traits in an exotic × adapted population.