Multi-locus genome-wide association studies for five yield-related traits in rice

Abstract

Background: Improving the overall production of rice with high quality is a major target of breeders. Mining potential yield-related loci have been geared towards developing efficient rice breeding strategies. In this study, one single-locus genome-wide association studies (SL-GWAS) method (MLM) in conjunction with five multi-locus genome-wide association studies (ML-GWAS) approaches (mrMLM, FASTmrMLM, pLARmEB, pKWmEB, and ISIS EM-BLASSO) were conducted in a panel consisting of 529 rice core varieties with 607,201 SNPs.

Results: A total of 152, 106, 12, 111, and 64 SNPs were detected by the MLM model associated with the five yield-related traits, namely grain length (GL), grain width (GW), grain thickness (GT), thousand-grain weight (TGW), and yield per plant (YPP), respectively. Furthermore, 74 significant quantitative trait nucleotides (QTNs) were presented across at least two ML-GWAS methods to be associated with the above five traits successively. Finally, 20 common QTNs were simultaneously discovered by both SL-GWAS and ML-GWAS methods. Based on genome annotation, gene expression analysis, and previous studies, two candidate key genes (LOC_Os09g02830 and LOC_Os07g31450) were characterized to affect GW and TGW, separately.

Conclusions: These outcomes will provide an indication for breeding high-yielding rice varieties in the immediate future.

Keywords: ML-GWAS; MLM; Oryza sativa; QTNs; Yield.

Fig. 1

Distribution of five yield-related traits in rice and Pearson coefficient analysis. The lower left represents the linear regression statistics between each two traits, the diagonal histogram represents the distribution of each trait, and the upper right number represents the correlation coefficient (positive numbers represent positive correlation, negative numbers represent negative correlation); asterisks represent significance (* stands for p-value less than 0.05; *** stands for p-value less than 0.001); yellow circles indicate that the absolute value of the correlation is greater than 0.5, and blue represents less than 0.5

Fig. 2

Genetic structure of the 529 rice panel. (a - b) PCA plots of the 529 rice core varieties. PCA plots present the genetic variation in the rice accessions with PC1 and PC2, PC2 and PC3, separately. (c) Phylogenetic tree clustering of 529 rice core germplasm accessions. (d) Genome-wide LD decay is estimated from all population and subpopulations. The x-axis represents the physical distance and the y-axis represents the average pairwise correlation coefficient (r²) of SNPs. The black, grey, purple, blue, green, orange, and red colors represent All, Adm, Aus, IndI, IndII, Tej, and Trj, successively

Fig. 3

Comparison of location of candidate QTNs and cloned genes on chromosomes. The common QTNs mapped in this study are labeled on the right side of chromosomes, and different colors display different traits: black, GL; red, TGW; green, YPP; brown, GW; olive, GT. Location of known genes is marked on chromosomes, and pink highlights genes covered by QTNs

Fig. 4

(a) Local linkage disequilibrium for qTGW-7-1. (b) Box plot of TGW traits about four haplotypes of LOC_Os07g31450. The x-axis represents four haplotypes of LOC_Os07g31450 and the y-axis represents 1000-grain weight. The table below is the detailed information of four haplotypes. (c) Heatmap of the expression pattern of LOC_Os07g31450 in various tissues among three local rice species. The y-coordinate indicates three species and relative expression, and x-coordinate indicates 39 different parts and development stages of rice tissue. Red represents higher gene expression and green indicates lower gene expression level, the gene expression levels are log₂ transformed