Population structure of the wild einkorn wheat accessions
First, we characterized the genetic diversity and traits of 43 wild einkorn wheat (T. monococcum ssp. aegilopoides) accessions. To clarify the intraspecific genetic diversity and population structure of wild einkorn, simple sequence repeat (SSR) marker-based PCR analysis was performed using the 43 wild einkorn accessions that were used for creating the synthetic hexaploids (Additional file 1: Table S1). One T. urartu accession was used as an outgroup species. Forty-two SSR markers covering all the chromosomes were selected based on the linkage maps of the A genome [32, 33] (Additional file 1: Table S2). The genotype accumulation curve plateaued at 15 SSR markers, indicating that these markers were enough to discriminate between the genotypes of the tested accessions (Additional file 2: Fig. S1). All the tested markers were polymorphic and had two to seven alleles (Additional file 1: Table S3). Mean Simpson’s index, expected heterozygosity, and evenness for the SSR markers were 0.4639, 0.4747, and 0.6993, respectively.
To estimate the population structure of the 43 wild einkorn accessions, we conducted Bayesian clustering analysis. The delta K for differing number of subpopulations presumed two clusters (k = 2) to be optimal (Fig. 1b, c). These two clusters corresponded to the two distinct lineages in the previous RNA-seq-based polymorphism analysis [34]. The lineage mainly distributed in southern Turkey, Iraq, and Iran was called Lineage 1 (L1), and the other lineage distributed in Turkey, Greece, and Armenia was called Lineage 2 (L2) (Fig. 1d). These two lineages were also detected in the UPGMA phylogenetic tree (Additional file 2: Fig. S2). T. monococcum ssp. aegilopoides KU-8143 was classified into L1 in Bayesian clustering analysis (Fig. 1c), but it belonged to L2 in the tree (Additional file 2: Fig. S2). KU-8143 could be a recombinant between L1 and L2. We designated KU-8143 as L1 for subsequent analyses.
To evaluate the genetic diversity of L1 (29 accessions) and L2 (14 accessions), we estimated the Simpson’s index and expected heterozygosity for these lineages. The Simpson’s index for L1 and L2 were 0.9655 and 0.9286, respectively. Expected heterozygosity for L1 and L2 was 0.3771 and 0.4571, respectively. These results indicate that the genetic diversity for L2 was higher than that for L1.
Phenotypic differences between the two lineages of wild einkorn wheat
Plant architecture was distinct between the two wild diploid wheat species, T. monococcum ssp. aegilopoides and T. urartu (Fig. 1a, Table 1, Additional file 1: Table S4, Table S5). Heading date (HD) and flowering date (FD) were earlier, plant height (PH) was higher, and spike length (SL) was longer in wild einkorn than in T. urartu. The number of spikelets per spike (NSp) was higher in wild einkorn than in T. urartu. To estimate phenotypic variations in the 43 wild einkorn accessions, we evaluated 26 traits in two seasons. Sixteen traits, including flowering date and heading date; awn-related traits including top awn length (TAL), middle awn length (MAL), and bottom awn length (BAL); 5th internode length (5InL), leaf width (LW), and flag leaf width (FLW); and spikelet-related traits including spikelet length (SpL), spikelet width (SpW), number of spikelets (NSp), spikelet density (SpD), and length-width ratio (SpLWr), showed significant correlations between the two seasons, implying that the variations of these traits among the accessions were stable. By contrast, leaf length, flag leaf length, spike width (SW), and spike length and stem-related traits including plant height, 2nd internode length (2InL), 3rd internode length (3InL), 4th internode length (4InL), and stem length (StL) showed no significant correlations between the seasons. Variations of these traits were not stable and could be influenced by environmental factors (Additional file 2: Fig. S3). To compare the extent of variation among the traits, we calculated the coefficient of variation (CV). Top awn length, 5th internode length, and self-seed fertility (SSF) were highly variable among the accessions (CV > 0.46), whereas heading date, flowering date, spikelet length, and spikelet width were less variable (CV < 0.10) (Table 1, Additional file 1: Table S4).
To evaluate phenotypic differences between L1 and L2 of wild einkorn, we compared the 16 stable traits in addition to five grain-related traits, including grain area size (GAS), grain length (GL), grain width (GW), grain circularity (GC), and perimeter length of grain (GPL), between these lineages. Of the 21 traits, 19 showed significant differences between the two lineages (Fig. 2a). Flowering date of L1 was earlier than that of L2. L1 had longer awns than L2. Number of spikes in L1 was smaller than that in L2, implying that L1 had a smaller number of tillers. Spikes of L1 were characterized by a small number of spikelets, low spikelet densities, and large spikelets and grains compared to the spikes of L2.
Given that most L1 and L2 accessions were geographically separated, these genetically and morphologically divergent lineages were assumed to have adapted to the local temperature and precipitation of their habitats. Average temperature (°C) per month and average precipitation (mm) per month were estimated in the habitats of each linage based on WorldClim global climate datasets (Fig. 2b). The average temperature per month in the L1 habitats from March to November was significantly higher than that in the L2 habitats. The average precipitation per month in the L1 habitats from January to April was greater than that in the L2 habitats, while the average precipitation per month in the L1 habitats from June to September was more than that in the L2 habitats. Overall, during the growing season of wild einkorn, the habitats of L1 accessions were characterized by relatively high temperatures, more precipitation in winter, and rapid drying after May. The habitats of L2 accessions were characterized by relatively low temperature, little precipitation in winter, and gradual drying after May.
Generation of synthetic wheat hexaploids with the Am genome
In total, 42 synthetic lines were generated through interspecific crosses between Ldn and 42 accessions of T. monococcum ssp. aegilopoides (Fig 3a). These synthetic hexaploid lines produced self-pollinated seeds. We conducted genomic in situ hybridization (GISH) analysis to evaluate the somatic chromosomes of the ABAm synthetic lines (Fig 3b, c, d). Forty-two somatic chromosomes were observed in the root cells of the F3 plants as expected. GISH analysis using the wild einkorn DNA as probes detected 28 chromosomes of wild einkorn descent in the ABAm synthetic line (Fig 3c, d). Fourteen chromosomes were stained with relatively stronger probe signals among the 42 chromosomes of the ABAm polyploids, and the weaker-stained chromosomes were presumed to be the B-genome chromosomes (Fig 3c, d). This observation indicated that the Am-genome chromosomes could not be clearly distinguished from the A-genome chromosomes by the GISH analysis using the wild einkorn DNA.
The position of each marker was deduced according to information for the chromosomal position of the SNP site distinguished between the A and Am genomes (Michikawa et al. 2019).
In nascent allohexaploid wheat, whole-chromosome aneuploidy has been reported [5]. To confirm the Am-genome chromosomes in the ABAm synthetics, we conducted PCR analysis with Am-chromosome-specific SSR markers (Fig 4, Table 2). The SSR markers used for this confirmation showed clear polymorphisms between Ldn and the wild einkorn accessions. For each of the Am-genome chromosomes, the ABAm synthetics contained both Ldn- and wild einkorn-derived PCR bands. Thus, all tested ABAm synthetic lines contained a set of the Am-genome chromosomes. Two of the 42 ABAm synthetic lines exhibited a hybrid dwarf phenotype (Fig 5a, b). The two lines showing hybrid dwarfism contained the expected 42 chromosomes including a set of the Am-genome chromosomes in the root tip cells (Fig 5c).
Phenotypic variation in the AABBAmAm synthetic lines
To examine phenotypic variation in the AABBAmAm synthetic lines, 26 traits were evaluated over two growing seasons. Significantly positive correlations between years were detected in 14 traits (Additional file 2: Fig. S4). Plant height, stem length, and 1st internode length showed high correlation coefficients. The correlation coefficients for traits related to flowering, awn, and spikelet morphology were significantly positive. To compare the extent of phenotypic variation between the tested traits, we calculated the coefficient of variation (CV) for each trait (Table 1). The CV was lower in the AABBAmAm synthetic hexaploids than in their wild einkorn parents, indicating that the degree of phenotypic trait variation within the synthetic hexaploids was less than that within their wild einkorn parents. The correlation coefficient of CV between the synthetic hexaploids and wild einkorn was 0.77 (p = 1.54e-07). When excluding SSF, which was an outlier trait, the correlation coefficient of CV was 0.89 (p = 4.92e-12). The significantly positive correlation coefficient of CV indicates that the diversity of each trait in wild einkorn was transmitted into the synthetic hexaploids.
Two of the 42 AABBAmAm synthetic lines (Ldn/ssp. aegilopoides KU-8267 and Ldn/ssp. aegilopoides KU-8276) showed hybrid dwarfism. The stem length and plant height of these two lines were significantly shorter than those of the other synthetic lines (Additional file 2: Fig. S5). These two lines are hereafter called HDW for hybrid dwarf. The heading and flowering dates of the HDW lines were delayed. The number of spikelets for the HDW lines was also smaller than those of the other AABBAmAm synthetic lines, but awn length, leaf-related traits, and selfed seed fertility were not different between them.
Transmission of the Am-genome variations to the AABBAmAm synthetic lines
To evaluate transmission of trait variation within the wild einkorn T. monococcum ssp. aegilopoides accessions to the AABBAmAm synthetic lines, the correlation coefficient for the 26 traits between the pollen parents and the synthetic lines was calculated. A significantly positive correlation was detected for nine traits in both seasons (Additional file 2: Fig. S6, Fig. S7). The traits with significant correlations were related to flowering, awn, and spikelets, but the traits involved in plant height and leaf morphology were not significant.
We conducted clustering analysis of correlation coefficients of the traits between the two seasons in each of the synthetic lines and the pollen parents in addition to correlation coefficients of the traits between the synthetic lines and the pollen parents. The traits were grouped into three major clusters: A, B, and C. In some traits of clusters A and B, the reproducibility of the traits between the seasons in the AABBAmAm synthetic lines was different from that in the pollen parents (Fig. 6). Plant heights, stem length, and 1st internode length showed high correlation coefficients in the AABBAmAm synthetic lines, but not in the wild einkorn accessions. In contrast, 5th internode length showed a significant correlation in the wild einkorn accessions, but not in the AABBAmAm synthetic lines.
In the traits of cluster C, both of the comparisons between the seasons in the AABBAmAm synthetic lines and the wild einkorn accessions showed high correlation coefficients. High correlation coefficients were also observed in the comparisons between the AABBAmAm synthetic lines and the wild einkorn accessions. Cluster C included flowering, awn, and spikelet-related traits. These results indicated that phenotypic variations in these traits within the wild einkorn accessions were transmitted into the AABBAmAm synthetic lines.
We examined whether phenotypic divergence between the wild einkorn accessions in L1 and L2 reflected phenotypic variations in the AABBAmAm synthetic lines (Fig. 7). Of the 16 traits showing significant divergence in the wild einkorn accessions between L1 and L2, 14 traits were also significantly different in the synthetic lines. In addition, significant differences in flag leaf width, 5th internode length, and stem length with awn between the L1 and L2 accessions were uniquely detected in the synthetic lines. The synthetic lines derived from L1 accessions had earlier flowering, longer awns, bigger spikelets and grains, and lower density spikelets than those from L2 accessions. These trait characteristics in the synthetic lines were clearly inherited from in their wild einkorn pollen parents.
Characterization of grain morphology of the AABBAmAm synthetic lines
Five grain-related traits, grain area size, grain length, grain width, grain circularity, grain perimeter length, and hundred-grain weight, were evaluated in the AABBAmAm synthetic lines. Grains of the AABBAmAm synthetic lines were longer than those of their Ldn and wild einkorn parents (Fig. 8a). Compared with the AABBDD synthetic wheat, the shape of grains of the AABBAmAm synthetic lines was slender, suggesting that addition of AmAm into AABB genomes changed grain shape from round to slender. Grain area size, perimeter length of grain, and grain length of the AABBAmAm synthetic lines were highly correlated with those of their einkorn pollen parents (Additional file 2: Fig. S8). All the grain-related traits showed significant differences between the synthetic lines derived from L1 and L2 einkorn accessions (Fig. 7). The variations of grain shapes within the synthetic lines were transmitted from the wild einkorn parents.
Grain hardness of wheat is an essential parameter for characterizing flour. Hard grain is used for Italian-style pasta, and soft grain is used for other types of noodles, such as ramen and udon. Grain hardness of the 40 AABBAmAm synthetic lines, excluding the HDW lines, was measured using a single-kernel characterization system (SKCS), which is algorithmically forced to have a value of 75 for hard wheat and 25 for soft wheat [35]. Grain hardness of the AABBAmAm synthetic lines ranged from 18.55 to 42.31 (Fig. 8b). Average grain hardness was 29.65 ± 6.00, indicating that grains of the AABBAmAm synthetic lines were soft grains.
A scanning electron microscope was used to observe cross sections of seeds of three AABBAmAm lines (Ldn/ssp. aegilopoides KU-101-3, Ldn/ssp. aegilopoides KU-8001, and Ldn/ssp. aegilopides KU-8201) (Fig. 8c). The surface of starch granules of all the lines was smooth. We also detected holes formed by starch granules that had fallen out of the endosperm during sample preparation. These observations were consistent with the characteristics of soft grains [30].
Phenotypic comparisons among AABBAmAm, AABBAA, and AABBDD synthetic lines.
Phenotypic comparisons between AABBDD synthetic wheat and AABBUU synthetic hexaploids indicated that differences in phenotypic traits between these nascent synthetic hexaploids reflect the genomes of their pollen parents [31]. To test whether this observation could extend to the other nascent synthetic hexaploids, we compared phenotypic traits of the AABBDD synthetic wheat with those of the AABBAmAm and AABBAA synthetic hexaploids (Fig. 9, Additional file 2: Fig. S5). The two HDW lines and the other 40 AABBAmAm synthetic lines were separately analyzed in the comparisons among the synthetic hexaploids. Heading and flowering dates of the AABBAmAm synthetic lines were later than those of the AABBDD synthetic wheat and almost the same as those of the AABBAA synthetic hexaploids. Heading and flowering dates of the HDW lines were significantly delayed. The plant height and stem length of the AABBAmAm synthetic lines were longer than those of the AABBDD and AABBAA synthetic hexaploids. The awn length of the AABBAmAm synthetic lines was also longer than those of the AABBDD and AABBAA synthetic lines. Compared with the AABBDD synthetic wheat, spike length in the AABBAmAm synthetic lines was shorter and the number of spikelets was larger, resulting in the high density of spikelets in the AABBAmAm synthetic lines. The grain length of the AABBAmAm synthetic lines was longer and grain width was shorter, forming the more slender grain of the AABBAmAm synthetic lines compared with the other synthetic hexaploids.