By analyzing the yield of the hybrids from the inbred lines bred from the Shaan A and Shaan B group, the combining ability of the inbred lines were evaluated, genome-wide association analysis, and mining associated loci for yield and its combining ability conducted. It will provide references for improving maize inbred lines selected from Shaan A group and Shaan B group and applying them in varieties breeding.

Based on NCⅡ genetic design, 85 excellent inbred lines from Shaan A group and Shaan B group were used to construct a hybrid population containing 246 F₁. Then, the yield of the hybrid population was tested in three environments to evaluate their general combining ability (GCA) and special combining ability (SCA). Using the 6H90K maize array to detect the parental genotypes, 63 879 high-quality SNPs were obtained, which were used to analyze the genetic characteristics of parental lines. According to the parental genotypes, 55 951 high-quality SNPs were inferred in the hybrid population for genome-wide association analysis of hybrid yield, GCA, and SCA using additive model and non-additive model. Meanwhile, candidate genes around the significant SNPs were screened and annotated based on the maize B73 reference genome.

The yield in the three environments accorded to the normal distribution with wide variation, the broad-sense heritability of yield was 59.04%, and the environmental effect was significant. There was significant positive correlation between hybrid yield and combining ability, and the correlation between hybrid yield and SCA (r=0.95) was higher than that between hybrid yield and GCA (r=0.62). The genetic characteristic of Shaan A group and Shaan B group was different, and inbred lines from Shaan A group have higher general combining ability. Totally, five, seven and nine significant SNPs were detected (-log₁₀(P)>3.86) for GCA, hybrid yield and SCA, respectively. Among them, four SNPs were co-located in hybrid yield and SCA. Ultimately, 17 associated SNPs were anchored. Dominant allele analysis of different trait-associated loci showed that four GCA-associated SNPs were controlled by additive effects, and the F₁ BLUE-associated loci could be divided into 4 types mainly by the dominant effect, and the heterozygous genotype is the favorite allele or sub-optimal allele for yield in F₁. Through functional annotation, the candidate genes were specifically expressed in maize growth and kernel establishment, for example, GRMZM2G165828 and GRMZM2G057557 were related to maize kernel development.

Based on this study, we consider that GCA and SCA jointly affect the yield of hybrids, and the effect of SCA is greater. Moreover, GCA and SCA may have different genetic basis, and GCA can be increased with the accumulation of favorable alleles. Using the genome-wide association analysis in the F₁ hybrid population can carry out genetic analysis related to combining ability, mine the genetic loci related to yield and combining ability, and accelerated the application of the associated loci in molecular breeding.

Ear traits are important components of grain yield in maize. Dissecting their genetic basis and mining significant SNPs using genome-wide association study (GWAS) can provide references for cloning functional genes and breeding high-yield maize varieties.

A total of 115 superior inbred lines from Shaan A group and Shaan B group, as well as four domestic backbone lines were selected as parents. Based on NCⅡ genetic design, an association population consisting of 442 hybrids was constructed, which was planted in two different environments to collect phenotype data of ear traits. Meanwhile, all parental lines were sequenced by the tunable genotyping by sequencing (tGBS) protocols. According to the genotype of inbred lines, altogether 19 461 high-quality SNPs were inferred in the association population. Then, GWAS was performed using 19 461 SNPs and phenotype data by three models including additive, dominance and epistasis, respectively. Combining with the transcriptome data of maize ear related tissues in the public database and the annotation information of genes, candidate genes were predicted.

Phenotypic data analysis showed that eight ear traits followed a continuous distribution, and there were 3.78%-45.25% of phenotypic variation. Analysis of variance indicated that environment and genotype effects reached an extremely significant level (P<0.001), and the range of broad-sense heritability was from 54.15% to 68.89%. And there were significantly positive or negative correlations among ear traits of hybrids. In total, 16, 3, 79 significant SNPs/pairs were identified under additive, dominant, and epistatic models, respectively. The significant loci detected by the three models cumulatively explained 38.21%-60.69% of the phenotypic variation of each trait. The cumulative phenotypic variation of significant SNP detected by additive model and epistatic model was 0.00-41.26% and 15.18%-45.36%, respectively. Effect analysis of significant SNPs identified by additive and dominant models showed most SNPs with additive or partial dominance effects, and only two with over-dominance effects. Further, only seven single-SNPs and five interaction pairs explained more than 5% of the phenotypic variation, and 17 candidate genes were predicted based on the SNP locations and gene expression information.

Ear traits of maize hybrids were mainly affected by additive and epistasis effects, but less by dominance effects. Multiple SNPs identified by additive and dominant models showed additive and partially dominance effects, and aggregating favorable alleles of these SNPs could improve the target traits.