Genetic architecture of ear traits based on association mapping and co-expression networks in maize inbred lines and hybrids

Abstract

Ear traits are key contributors to grain yield in maize; therefore, exploring their genetic basis facilitates the improvement of grain yield. However, the underlying molecular mechanisms of ear traits remain obscure in both inbred lines and hybrids. Here, two association panels, respectively, comprising 203 inbred lines (IP) and 246 F₁ hybrids (HP) were employed to identify candidate genes for six ear traits. The IP showed higher phenotypic variation and lower phenotypic mean than the HP for all traits, except ear tip-barrenness length. By conducting a genome-wide association study (GWAS) across multiple environments, 101 and 228 significant single-nucleotide polymorphisms (SNPs) associated with six ear traits were identified in the IP and HP, respectively. Of these significant SNPs identified in the HP, most showed complete-incomplete dominance and over-dominance effects for each ear trait. Combining a gene co-expression network with GWAS results, 186 and 440 candidate genes were predicted in the IP and HP, respectively, including known ear development genes ids1 and sid1. Of these, nine candidate genes were detected in both populations and expressed in maize ear and spikelet tissues. Furthermore, two key shared genes (GRMZM2G143330 and GRMZM2G171139) in both populations were found to be significantly associated with ear traits in the maize Goodman diversity panel with high-density variations. These findings advance our knowledge of the genetic architecture of ear traits between inbred lines and hybrids and provide a valuable resource for the genetic improvement of ear traits in maize.

Supplementary information: The online version contains supplementary material available at 10.1007/s11032-023-01426-9.

Keywords: Association mapping; Co-expression network; Ear traits; Hybrid; Inbred line; Maize.

Fig. 1

Phenotypic analysis of six ear traits based on best linear unbiased estimations (BLUEs). A Violin plot of phenotype performance of six ear traits including ear diameter (ED), ear length (EL), ear row number (ERN), kernel number per row (KNR), ear tip-barrenness length (ETB), and kernel number per ear (KNE). B Correlation coefficients between six ear traits. The correlations of traits are shown above the diagonal for the IP and below the diagonal for the HP. *, **, and *** indicate significant correlation at the 0.05, 0.01, and 0.001 probability levels, respectively

Fig. 2

Comparison analysis of kernel number per ear (KNE) for tester lines and their F₁ progeny. A Comparison analysis for KNE in Yulin in 2018 (18Yu). B Comparison analysis using best linear unbiased estimations (BLUEs) for KNE. Red point refers to the KNE of tester lines and the box plot indicates the KNE of their F₁ progeny

Fig. 3

Distribution of genome-wide significant single-nucleotide polymorphisms (SNPs). A Significant SNPs for each ear trait identified in the inbred-based population (IP). B Significant SNPs for each ear trait identified in the hybrid-based population (HP). a, ear diameter (ED); b, ear length (EL); c, ear row number (ERN); d, kernel number per row (KNR); e, ear tip-barrenness length (ETB); f, kernel number of per ear (KNE); g, significant SNP number (window size of 1 Mb); Chr, chromosome. The outer gray circle represents the chromosome length (Mb). Significant SNPs are marked by scatter plots and displayed by the lowest P value in all environments or models. Circle represents the SNP identified in a single environment or best linear unbiased estimations (BLUEs). Triangle represents the SNP identified in both at least one environment and BLUEs. Green, blue, orange, white, and purple, respectively, indicate SNP with additive, complete–incomplete dominance, over-dominance, missing dominance effect index, and varied dominance effect index in different environments

Fig. 4

Gene ontology enrichment (GO) analysis of candidate genes. A Significant GO terms in the IP. B Significant GO terms in the HP

Fig. 5

High-density association mapping and expression levels of candidate genes. Candidate gene-based association analysis for GRMZM2G143330 (A) and GRMZM2G171139 (B) in the maize Goodman diversity panel. Each set of plots includes a partial Manhattan plot (top left), the linkage disequilibrium heatmap (bottom left), and the phenotypic values of different haplotypes (box plot). Chromosome position is marked with B73 (AGPv4) genome. The significant of differences was evaluated by two-tailed Student’s t-test. EL, ear length; ERN, ear row number. Expression pattern of GRMZM2G143330 (C) and GRMZM2G171139 (D) in different tissues from different development stages of B73, Mo17, and their offspring (B73×Mo17). DAP, days after pollination; DAS, days after seeding; TPM, transcripts per million. Gene expression is shown by log₂ (TPM + 1)