Genome-wide association study revealed significant SNPs for anthracnose resistance, seed alkaloids and protein content in white lupin

Abstract

White lupin (Lupinus albus L.) is a high-protein grain legume alternative to soybean in Central Europe, but its cultivation is risky due to the fungal disease anthracnose that can cause severe yield damage. In addition, management of seed alkaloids is critical for human nutrition and animal feed. We report on a white lupin collection of genebank accessions, advanced breeding lines and cultivars that was genotyped and phenotypically characterized for anthracnose resistance and seed alkaloids and protein levels. Using genotyping by sequencing (GBS), SeqSNP-targeted GBS, BiomarkX genotyping and Sanger sequencing, a genetic resource of genome-wide SNPs for white lupin was established. We determined anthracnose resistance in two years field trials at four locations with infection rows and measured seed alkaloids and protein levels by near-infrared spectroscopy (NIRS). Few white lupin breeding lines showed anthracnose resistance comparable or better than Celina and Frieda, currently the best commercial cultivars in Germany. NIRS estimates for seed alkaloids and protein levels revealed variation in the white lupin collection. Using genome-wide association studies (GWAS), we identified SNPs significantly associated with anthracnose resistance in the field representing known and new genomic regions. We confirmed the pauper locus and detected new SNP markers significantly associated with seed alkaloids. For the first time, we present loci associated with total grain protein content. Finally, we tested the potential of genomic prediction (GP) in predicting the phenotype of these three quantitative traits. Application of results and resources are discussed in the context of fostering breeding programs for white lupin.

Fig. 1

Genetic characterization of the white lupin LUW collection. (A) Kinship matrix of the LUW panel with cluster tree (left, bottom) and corresponding phenotypes for anthracnose (right) and seed quality (top). The color code for kinship, anthracnose infection and seed quality traits is given in the legend. The kinship is clustered in three groups of related genotypes (B) principal component (PCA) analysis based on genotyping data. Boxes, stars and circles illustrate the same clusters as presented in sub figure A. Red and green color of boxes, circles and stars indicate high and low anthracnose susceptibility, respectively. The explained variation in the first 20 PCs (blue bars) and the cumulated explained variation (red curve) are shown in the small graph

Fig. 2

Results of genome-wide association study for seed protein and alkaloid content. (A) Manhattan plot for protein and (B) alkaloid content with (C) corresponding Q–Q plots by different statistical approaches (see text) showing SNP association with seed quality traits. Statistical approaches are shown in different symbols and chromosomes are shown in different colors. The dotted horizontal line indicates the -logP significance threshold of 5.7. Vertical dashed lines indicate QTLs. (D.1) Section of pauper locus on chromosome Lalb_Chr18 with significantly associated SNPs, along with (D.2) a SNP-wise linkage disequilibrium plot among all genotyped SNPs along the pauper locus. (E) Boxplot for the alkaloid seed content for every haplotype derived from an allele combination of both QTLs detected in the pauper region. Gray numbers above the boxplots illustrate the sample size for each haplotype. The NIRS seed alkaloid measurements were adjusted by subtracting the measured alkaloid content by the lowest (negative) measured value and adding an offset value of 0.005

Fig. 3

Manhattan (left) and corresponding Q–Q plot (right) by different statistical approaches (see text) showing SNP association with anthracnose resistance at three different developmental stages. Statistical approaches are shown in different symbols and chromosomes are shown in different colors. The dotted horizontal line indicates the -logP significance threshold of 5.7

Fig. 4

Genomic prediction for anthracnose and seed quality in the white lupin LUW collection. (A) Measured versus predicted values and corresponding regression lines for anthracnose infection at two developmental stages as well as seed alkaloid level and protein content using all SNP markers (purple) and only significantly associated SNP markers (QTLs, yellow). (B) Pearson correlation prediction for all traits using the two marker sets. The dots give the average prediction value from the cross-validation, while the gray bars show the confidence interval; the arrows indicate the dot when the confidence interval is zero