Resequencing 200 Flax Cultivated Accessions Identifies Candidate Genes Related to Seed Size and Weight and Reveals Signatures of Artificial Selection

Abstract

Seed size and weight are key traits determining crop yield, which often undergo strongly artificial selection during crop domestication. Although seed sizes differ significantly between oil flax and fiber flax, the genetic basis of morphological differences and artificial selection characteristics in seed size remains largely unclear. Here we re-sequenced 200 flax cultivated accessions to generate a genome variation map based on chromosome assembly reference genomes. We provide evidence that oil flax group is the ancestor of cultivated flax, and the oil-fiber dual purpose group (OF) is the evolutionary intermediate transition state between oil and fiber flax. Genome-wide association studies (GWAS) were combined with LD Heatmap to identify candidate regions related to seed size and weight, then candidate genes were screened based on detailed functional annotations and estimation of nucleotide polymorphism effects. Using this strategy, we obtained 13 candidate genes related to seed size and weight. Selective sweeps analysis indicates human-involved selection of small seeds during the oil to fiber flax transition. Our study shows the existence of elite alleles for seed size and weight in flax germplasm and provides molecular insights into approaches for further improvement.

Keywords: domestication; flax (Linum usitatissimum L.); genetic diversity; genome-wide association studies; seed size; selective sweeps.

Figure 1

Population genetic structure and Linkage disequilibrium (LD) decay among 200 flax cultivars. (A) Neighbor-joining tree of 200 flax accessions was generated using 1,179,402 SNPs. (B) Population structure analysis with different numbers of clusters (K = 2, 3 and 4). (C, D) Graphical phenotype of 7 major agronomic traits. (C) 2016DL, (D) 2017UR. (E) PCA for the 200 flax varieties. (F) The most likely group number is determined using ad-hoc statistic △K for K values ranging from 2 to 8. (G) LD decay distance in three groups. Oil flax, oil-fiber dual purpose flax (OF), and fiber flax are drawn in green, blue and red colors, respectively.

Figure 2

GWAS for seed size and 1000-seed weight traits, and candidate genes were obtained for the peak region on chromosome 11. (A) Manhattan plot for 1000-SW in 2016DL. The red arrow indicates the position of the strong peak on chromosome 11. The horizontal dashed line indicates the significance threshold (P = 1 × 10⁻⁵). (B) Local Manhattan plot (top) and LD heatmap (bottom) surrounding the peak on chromosome 11. The red arrow indicates a significant SNP in candidate gene Lus10036372. (C) Exon-intron structure of Lus10036372 and haplotypes in that gene. (D–G) Boxplots for 1000-SW based on the haplotypes (Hap.) for Lus10036372 in 2016DL (D), 2017UR (E), 2019UR (F) and 2019YL (G). The difference between haplotypes was analyzed by non-parametric t tests.

Figure 3

Analyses of the peak for seed size and 1,000-seed weight on chromosome 12. (A–C) Local Manhattan plot for SL (A), SW (B), and 1,000-SW (C) in 2017UR and LD heatmap (D) surrounding the peak on chromosome 12. Red dotted line indicates that the highest peak (SNP 16347940) on (A–C) and located in Lus10043126. (E) Exon–intron structure of Lus10043126 and DNA polymorphism in this gene. (F) A schematic map of structural variation of Lus10043126 was presented. The major and minor alleles at each polymorphic locus were represented in red and green, respectively, and the miss sites were represented in black. (G–J) Boxplots for 1000-SW based on Lus10043126 function as follows: haplotypes A–I (functional alleles) and haplotypes j-K (nonfunctional alleles) in 2016DL (G), 2017UR (H), 2019 UR (I) and 2019YL (J). The difference between haplotypes was analyzed by non-parametric t tests.

Figure 4

Distribution of nucleotide diversity (π) within candidate genes and allele frequency differences among Oil, OF and Fiber groups. (A) Boxplots for nucleotide diversity in seed size candidate genes among Oil, OF and Fiber groups. The difference was analyzed by two-tailed Student’s t tests. (B, C) Local nucleotide diversity (π) was distributed around Lus10008949 on chromosome 9 (B) and Lus10008438 on chromosome 1 (C) among Oil (green), OF (blue) and Fiber (red) groups. (D) The distribution of allele frequencies of strong SNPs are located in candidate genes were distributed in Oil, OF and Fiber groups. The large-seed and small-seed alleles are shown in purple and orange, respectively.