The potential of pale flax as a source of useful genetic variation for cultivated flax revealed through molecular diversity and association analyses

Abstract

Pale flax (Linum bienne Mill.) is the wild progenitor of cultivated flax (Linum usitatissimum L.) and represents the primary gene pool to broaden its genetic base. Here, a collection of 125 pale flax accessions and the Canadian flax core collection of 407 accessions were genotyped using 112 genome-wide simple sequence repeat markers and phenotyped for nine traits with the aim of conducting population structure, molecular diversity and association mapping analyses. The combined population structure analysis identified two well-supported major groups corresponding to pale and cultivated flax. The L. usitatissimum convar. crepitans accessions most closely resembled its wild progenitor, both having dehiscent capsules. The unbiased Nei's genetic distance (0.65) confirmed the strong genetic differentiation between cultivated and pale flax. Similar levels of genetic diversity were observed in both species, albeit 430 (48 %) of pale flax alleles were unique, in agreement with their high genetic differentiation. Significant associations were identified for seven and four traits in pale and cultivated flax, respectively. Favorable alleles with potentially positive effect to improve yield through yield components were identified in pale flax. The allelic frequencies of markers associated with domestication-related traits such as capsular dehiscence indicated directional selection with the most common alleles in pale flax being absent or rare in cultivated flax and vice versa. Our results demonstrated that pale flax is a potential source of novel variation to improve multiple traits in cultivated flax and that association mapping is a suitable approach to screening pale flax germplasm to identify favorable quantitative trait locus alleles.

Keywords: Association mapping; Cultivated flax; Exotic alleles; Genetic diversity; Pale flax; Wild ancestor.

Fig. 1

Population structure and genetic relationships between pale flax (L. bienne) and cultivated flax (L. usitatissimum). a structure analysis for K = 2 (upper panel) and the merged STRUCTURE (K = 2 and K = 3) populations (lower panel) of L. bienne (red) and L. usitatissimum (blue) accessions. P1, P2 and P3 represent populations within L. bienne major group and P4 and P5 represent populations within L. usitatissimum major group. b Principal coordinate analysis (PCoA) of the 532 pale and cultivated flax accessions. Cultivated flax was further divided into four convarieties, namely L. usitatissimum convar. mediterraneum, L. usitatissimum convar. usitatissimum, L. usitatissimum convar. elongatum and L. usitatissimum convar. crepitans. c Pairwise F _ST comparisons among the five populations inferred by STRUCTURE. d Phylogenetic tree created using the Neighbor-joining (NJ) algorithm (Nei 1973). Colored clusters represent pale and cultivated flax STRUCTURE (K = 2) major groups. P1–P5 correspond to the merged STRUCTURE (K = 2 and K = 3) populations, indicating their geographic distribution. The scale bar indicates the Nei (1973) minimum genetic distance

Fig. 2

Comparisons of five association mapping models in pale and cultivated flax for seeds per boll, plant height and capsular dehiscence. Probability–probability (P–P) plots of observed versus expected −Log₁₀ (P) values for a Pale flax. b Cultivated flax. Q general linear model using the Q matrix, PCA general linear model using the PCA matrix, K mixed linear model using the kinship matrix, Q + K mixed linear model using the Q and K matrices, PCA + K mixed linear model using the PCA and K matrices

Fig. 3

Comparisons of allelic effects of associated markers with agronomic and phenological traits. a Allelic effects on thousand seed weight, seeds per boll and capsular dehiscence in pale flax. b Allelic effects on thousand seed weight, plant height and 5 % flowering in cultivated flax. Same letters above the box plots indicated values that do not differ statistically according to the Kruskal–Wallis test (α = 0.01)