Global genomic population structure of wild and cultivated oat reveals signatures of chromosome rearrangements

Abstract

The genus Avena consists of approximately 30 wild and cultivated oat species. Cultivated oat is an important food crop, yet the broader genetic diversity within the Avena gene pool remains underexplored and underexploited. Here, we characterize over 9000 wild and cultivated hexaploid oat accessions of global origin using genotyping-by-sequencing and explore population structure using multidimensional scaling and population-based clustering methods. We also conduct analyses to reveal chromosome regions associated with local adaptation, sometimes resulting from large-scale chromosome rearrangements. We report four distinct genetic populations within the wild species A. sterilis, a distinct population of cultivated A. byzantina, and multiple populations within cultivated A. sativa. Some chromosome regions associated with local adaptation are also associated with confirmed structural rearrangements on chromosomes 1A, 1C, 3C, 4C, and 7D. This work provides evidence suggesting multiple polyploid origins, multiple domestications, and/or reproductive barriers amongst Avena populations caused by differential chromosome structure.

Fig. 1. Populations of collected wild oat.

a Heatmap of taxa percentages by species in each of K = 21 populations from sparse nonnegative matrix factorization (sNMF) analysis. Light blue indicates a low percentage (2–10%) and dark blue indicates a high percentage (over 50%) while no color indicates 0–1%. The sum is the total number of taxa. Percentages may not sum to 100 due to rounding differences. A. hausknechtii, a contested species with a single duplicated accession in P04 is omitted. b Multi-dimensional scaling of n = 9112 taxa, colored by population membership. Four populations composed primarily of Avena sterilis (P02, P03, P04, P05) are indicated, as well as the population of A. byzantina (P01) and a population of A. sativa landraces from Spain (P06). The remaining non-labeled populations are primarily A. sativa. An interactive, rotatable version of this plot is available online (http://graingenes.shinyapps.io/Avena_diversity/). c Map of collection sites for taxa from four A. sterilis populations. An interactive version of this map is available at https://www.google.com/maps/d/edit?mid=1kYjxF-c5K-VHeIth0oSCwz5zt6-P5Yo&usp=sharing. Source data are provided as a Source Data file.

Fig. 2. Populations of cultivated oat.

Multi-dimensional scaling of n = 6928 taxa from populations containing primarily cultivated A. sativa (a) shown on axes 1 and 2 and (b) shown on a projection of the first three axes. Membership in applicable populations from K = 21 is colored as in Fig. 1. Populations P07 (Australian oat), P08 (southern USA), P21 (winter oat), and P09 (North Dakota type) are indicated, as discussed in the text. Online interactive versions of these plots can be used to visualize or highlight additional populations or features (https://graingenes.shinyapps.io/Oat_diversity/). Source data are provided as a Source Data file.

Fig. 3. Heatmaps of K = 21 sparse nonnegative matrix factorization (sNMF) populations of A. sativa taxa.

a Population membership in each of K = 21 sNMF populations summarized by country. b Population membership summarized by North American state and province. Numbers indicate percentage membership of a country, state, or province in a given population. Light blue indicates a low percentage (2–10%) and dark blue indicates a high percentage (over 50%) while no color indicates 0–1%. Populations without taxa are not shown. Countries or states with fewer than ten accessions are omitted. The sum is the total number of taxa in each row. Percentages do not always sum to 100 due to rounding. Source data are provided as a Source Data. file.

Fig. 4. Impact of chromosome rearrangements on hexaploid Avena population structure and local adaptation revealed by complementary population genomic approaches.

Manhattan plots show the PCA-based genome-wide scans of local adaptation across the 21 chromosomes, with chromosome names shown on top of each panel, analyzed using PCAdapt for the full set of taxa. The y-axis represents the −log10(p) values of marker associations with population structure. This y-axis is truncated, such that highly significant points are crowded at the top. The red line at y = −log10(p) = 2.12 marks the false discovery rate threshold lower than 5% (adjusted for multiple comparisons) for detecting outlier loci (Supplementary Fig. 9). Shaded areas below zero and above the x-axis labels indicate putative inversions identified by Lostruct analysis. The colored bars correspond to the outlier genomic regions (green for corner 1, orange for corner 2, and purple for corner 3) in Supplementary Fig. 10. Source data are provided as a Source Data file.

Fig. 5. Chromosome 7D inversion karyotypes in hexaploid Avena reveal population structure and patterns of agroecological adaptation.

a The proportions of four putative haplotypes (as revealed by K-Means clustering) in each of six hexaploid species. The four haplotypes are haplotype 1 (7D-H1) (n = 1638), haplotype 2 (7D-H2) (n = 1983), haplotype 3 (7D-H3) (n = 4868), and haplotype 4 (7D-H4) (n = 133). b The proportions of the four haplotypes in each of the 21 Avena populations (P01 to P21). Source data are provided as a Source Data. file.