Phylogenomic investigation of safflower (Carthamus tinctorius) and related species using genotyping-by-sequencing (GBS)

Abstract

Safflower (Carthamus tinctorius, Asteraceae) is a source of high-quality edible oil growing in moisture-limited environments. Despite its economic importance, the relationships to close wild species in Carthamus and the presence and relationships of ecotypes within safflower are still not fully clarified. Here we use genotyping-by-sequencing to identify the wild progenitor of C. tinctorius, infer phylogenetic relationship within the series Carthamus and identify groups of closely related lineages within cultivated safflower. Phylogenetic and population genomic analyses found C. palaestinus to be the closest relative and single progenitor of C. tinctorius, which confirms the Levant as the area of domestication of the crop. Flow cytometry showed all analyzed samples of C. oxyacantha, C. palaestinus and C. tinctorius to be diploid (2n = 2x = 24) with 2C genome sizes of 2.4-2.7 pg. Analyses of a set of 114 worldwide distributed safflower accessions arrived at two to five genetic groups, which showed, however, no correlation with the geographic origins of these accessions. From this, we conclude that the trade of safflower seeds resulted in multiple introductions of genotypes from the Levant into other areas with suitable climate conditions for the plant, as well as exchange of genotypes among these areas.

Figure 1

Unrooted strict consensus tree of 2484 MP trees derived from the analysis of GBS data. Asterisks indicate branches with support values of ≥ 98% in MP bootstrap analysis and posterior probabilities ≥ 0.99 in Bayesian phylogenetic inference for the backbone clades of the tree. The inserted tree on the left is one of the most parsimonious trees and indicates branch lengths in MP analysis.

Figure 2

Unrooted SVDquartet species tree using the multi-species coalescent as tree model on the GBS dataset of the Carthamus target species. The individuals also analyzed in Fig. 1 were partitioned according to their species affiliation, with the introgressed C. × oxyacantha individuals treated as a separate category. Ploidy levels and chromosome numbers are listed behind the species names. Numbers along branches indicate bootstrap support values (≥ 50%).

Figure 3

One out of four equally parsimonious trees of a GBS-derived MP analysis of a large and geographically diverse set of 114 safflower accessions. The tree was rooted with C. oxyacantha. Bootstrap values ≥ 75% are indicated by asterisks at the branches. Gray branches depict clades that were not recovered in all MP trees and collapse in the strict consensus tree. Colored bars provide the assignment of safflower individuals to four intraspecific groups according to a population structure analysis. To the right of the groups the proportion of members derived from certain geographical areas are given.

Figure 4

Population structure analysis based on 3720 unlinked SNP loci in LEA with K = 3 for C. oxyacantha, C. palaestinus, C. tinctorius and C. glaucus.

Figure 5

Principal component analysis (PCA) of C. oxyacantha (Co), C. glaucus (Cg), C. palaestinus (Cp) and C. tinctorius (Ct) along the first and second PCs.

Figure 6

Population structure analysis based on 7556 unlinked SNPs in LEA with K = 5 for C. tinctorius and C. palaestinus individuals.