Landscape genomics reveals adaptive genetic differentiation driven by multiple environmental variables in naked barley on the Qinghai-Tibetan Plateau

Abstract

Understanding the local adaptation of crops has long been a concern of evolutionary biologists and molecular ecologists. Identifying the adaptive genetic variability in the genome is crucial not only to provide insights into the genetic mechanism of local adaptation but also to explore the adaptation potential of crops. This study aimed to identify the climatic drivers of naked barley landraces and putative adaptive loci driving local adaptation on the Qinghai-Tibetan Plateau (QTP). To this end, a total of 157 diverse naked barley accessions were genotyped using the genotyping-by-sequencing approach, which yielded 3123 high-quality SNPs for population structure analysis and partial redundancy analysis, and 37,636 SNPs for outlier analysis. The population structure analysis indicated that naked barley landraces could be divided into four groups. We found that the genomic diversity of naked barley landraces could be partly traced back to the geographical and environmental diversity of the landscape. In total, 136 signatures associated with temperature, precipitation, and ultraviolet radiation were identified, of which 13 had pleiotropic effects. We mapped 447 genes, including a known gene HvSs1. Some genes involved in cold stress and regulation of flowering time were detected near eight signatures. Taken together, these results highlight the existence of putative adaptive loci in naked barley on QTP and thus improve our current understanding of the genetic basis of local adaptation.

Fig. 1. Inference of population genetic structure.

A Distribution of 157 naked barley landraces. Blue, yellow, orange, and green dots correspond to Group 1, Group 2, Group 3, and Group 4, respectively. B The first two axes of the principal component analysis on SNP markers. Individual genotypes were colored based on the cluster allocation of the discriminant analysis of principal components (DAPC). Colors are explained in the legend to the right. The axes denote the relative proportions of explained genetic variance. C Individual ancestry coefficients of 157 naked barley landraces estimated using Admixture with K = 2–6. With K = 4, blue, yellow, orange, and green bars correspond to Group 1, Group 2, Group 3, and Group 4, respectively. D Principal component analysis. E Neighbor-joining tree.

Fig. 2. Results of principal component analysis of climate variables and RDA biplot.

A Principal component analysis of the climatic diversity of 145 georeferenced barley landraces. The axes denote the proportion of explained climatic variance; the dots’ colors represent the groups corresponding to the results of population structure analysis, and the vectors represent the scale, diversity, and direction of drivers of differentiation. B RDA showed that the six nonredundant climate variables were correlated with the observed population structure and might have had selective impacts in the past. The biplot depicted the eigenvalues and length of eigenvectors for the RDA. Blue, yellow, orange, and green dots correspond to Group 1, Group 2, Group 3, and Group 4, respectively.

Fig. 3. Identification of outliers.

A The signatures identified using EnvGWAS on the 37,636 SNPs with MAF ≥ 0.05 for precipitation in July. The dark red line refers to Bonferroni correction based on α = 0.05 (− log10 (p) = 5.877), while the green line is based on a false discovery rate with a q-value > 0.05. The red dots indicate the significant loci. B, C The candidate SNPs identified using LFMM on the 37,636 SNP markers with MAF ≥ 0.05 for precipitation in September and May, respectively. The dark green line is based on a false discovery rate with a q-value > 0.05, specific for each trait. The red dots indicate the significant loci.

Fig. 4. Patterns of the seven chromosomes’ diversity.

A Patterns of the seven chromosomes’ diversity in 145 naked barley landraces based on 37,636 SNP markers. A sliding window analysis (window size = 20,000 bp, step size = 10,000 bp) was used to compute the SNP density, nucleotide diversity (pi), Watterson’s theta estimator (θ), and Tajima’s D (from top to bottom). The red lines represented the selected intervals. Gray dashed horizontal lines indicated genome averages. B Comparison of seven chromosomes’ diversity between associated and nonassociated intervals. The associated intervals contained at least one marker associated with climate variables. SNP density, nucleotide diversity (pi), Watterson’s theta estimator (θ), and Tajima’s D (from left to right) were calculated in a sliding window (window size = 20,000 bp, step size = 10,000 bp).