Genetic diversity analysis of a flax (Linum usitatissimum L.) global collection

BACKGROUND: A sustainable breeding program requires a minimum level of germplasm diversity to provide varied options for the selection of new breeding lines. To maximize genetic gain of the North Dakota State University (NDSU) flax breeding program, we aimed to increase the genetic diversity of its parental stocks by incorporating diverse genotypes. For this purpose, we analyzed the genetic diversity, linkage disequilibrium, and population sub-structure of 350 globally-distributed flax genotypes with 6200 SNP markers. RESULTS: All the genotypes tested clustered into seven sub-populations (P1 to P7) based on the admixture model and the output of neighbor-joining (NJ) tree analysis and principal coordinate analysis were in line with that of structure analysis. The largest sub-population separation arose from a cluster of NDSU/American genotypes with Turkish and Asian genotypes. All sub-populations showed moderate genetic diversity (average H = 0.22 and I = 0.34). The pairwise F(st) comparison revealed a great degree of divergence (F(st) > 0.25) between most of the combinations. A whole collection mantel test showed significant positive correlation (r = 0.30 and p < 0.01) between genetic and geographic distances, whereas it was non-significant for all sub-populations except P4 and P5 (r = 0.251, 0.349 respectively and p < 0.05). In the entire collection, the mean linkage disequilibrium was 0.03 and it decayed to its half maximum within < 21 kb distance. CONCLUSIONS: To maximize genetic gain, hybridization between NDSU stock (P5) and Asian individuals (P6) are potentially the best option as genetic differentiation between them is highest (F(st) > 0.50). In contrast, low genetic differentiation between P5 and P2 may enhance the accumulation of favorable alleles for oil and fiber upon crossing to develop dual purpose varieties. As each sub-population consists of many genotypes, a Neighbor-Joining tree and kinship matrix assist to identify distantly related genotypes. These results also inform genotyping decisions for future association mapping studies to ensure the identification of a sufficient number of molecular markers to tag all linkage blocks.