Landscape genomics reveal signatures of local adaptation in barley (Hordeum vulgare L.)

Abstract

Land plants are sessile organisms that cannot escape the adverse climatic conditions of a given environment. Hence, adaptation is one of the solutions to surviving in a challenging environment. This study was aimed at detecting adaptive loci in barley landraces that are affected by selection. To that end, a diverse population of barley landraces was analyzed using the genotyping by sequencing approach. Climatic data for altitude, rainfall and temperature were collected from 61 weather sites near the origin of selected landraces across Ethiopia. Population structure analysis revealed three groups whereas spatial analysis accounted significant similarities at shorter geographic distances (< 40 Km) among barley landraces. Partitioning the variance between climate variables and geographic distances indicated that climate variables accounted for most of the explainable genetic variation. Markers by climatic variables association analysis resulted in altogether 18 and 62 putative adaptive loci using Bayenv and latent factor mixed model (LFMM), respectively. Subsequent analysis of the associated SNPs revealed putative candidate genes for plant adaptation. This study highlights the presence of putative adaptive loci among barley landraces representing original gene pool of the farming communities.

Keywords: Hordeum vulgare; adaptive loci; genotyping by sequencing; landscape genomics; local adaptation; spatial genetic structure.

Figure 1

Distribution of barley landraces and weather sites across Ethiopia on an altitudinal map.

Figure 2

Distribution of barley landraces and assignment of population membership coefficient along Ethiopian map (A). Each barley landrace was assigned to its respective inferred cluster based on the membership coefficient obtained from population structure analysis carried out using the model based software STRUCTURE. The scatter plot of principal components represented by the first and second principal components depicting the groupings of the barley individuals based on the subpopulations (B).

Figure 3

Spatial autocorrelation correlogram plots. The plot depicts results obtained from all geographic regions (A), and after accessions collected from Tigray region were removed and isolation by distance was calculated for accessions from the rest of the regions (B). The analysis considered geographic distances with even distance class of 20 km. Dashed lines encompass the 95% confidence interval of the null hypothesis, and each point represents the autocorrelation coefficient (r).

Figure 4

The partial RDA variance partitioning was computed for entire dataset (A) and after accessions from Tigray region were excluded (B). The variance explained due to all variables (Model 1), the variance explained after controlling the effect introduced due to geographic distance (Model 2), and variance explained by geographic coordinates after the variance due to climate variables controlled (Model 3).

Figure 5

Partial RDA analysis was performed to determine the relative contribution of climate and geographic variables shaping the genetic structure. The biplot depicts the eigenvalues and lengths of eigenvectors for the RDA conditioned on geographic distance.

Figure 6

Allele frequency of putative adaptive loci correlated with altitude classes and Rainfall in Kiremt as detected by LFMM. The major and minor alleles of putative adaptive loci Hv_SNP27845 showed frequency pattern along the altitude classes (A). The allele distribution depicting on the Ethiopian map (B) and the rainfall pattern along the coordinates of Ethiopia displayed in scatter plot (C).

Figure 7

A Bayesian based BayScan program were employed to scan for the presence of putative outlier loci affected by selection. This plot presents F_ST against log 10(q-value), which is the FDR analog of the p-value. The line represents the threshold FDR = 0.05 and the red dots indicate the outlier loci which are potentially affected by directional selection.