QTLs for agronomic traits detected in recombinant inbred lines derived from a bread wheat × spelt cross.

Spelt wheat (Triticum aestivum subsp. spelta), a subspecies of common wheat, is a genetic resource for the breeding of bread wheat (T. aestivum subsp. aestivum); however, genetic analyses of agronomic traits in bread wheat × spelt crosses are insufficient. Here, we conducted QTL analysis in the recombinant inbred lines from a bread wheat × spelt cross. In addition to the major Q locus, QSpd.obu-4D was detected with the spelt allele conferring a higher spikelet density than the bread wheat allele. The effect of QSpd.obu-4D was evident in the presence of the Q allele of bread wheat, suggesting that this variation might be cryptic in spelt wheat with the q allele. Two QTLs with stable effects were identified for grain length, one of which (QGl.obu-1A) has never been detected in a bread wheat × spelt cross. The spelt wheat allele at QHt.obu-7B conferring later heading was identified in the Vrn-B3 region and could be a novel gene source for modifying heading time. Furthermore, QGi.obu-2B, responsible for low grain dormancy of spelt wheat, was detected. Further exploration and identification of useful QTLs could accelerate the utilization of spelt wheat as a genetic resource for bread wheat breeding programs.

Introduction

Wheat has been an important cereal crop since the ancient times and is cultivated under various climatic conditions around the world. Approximately 95% of wheat cultivated today is bread wheat (Triticum aestivum subsp. aestivum), although wheat comprises a variety of species and subspecies (Matsuoka 2011). Bread wheat is a hexaploid wheat species, with AABBDD genomes, and shows the free threshing trait. In contrast, spelt wheat (T. aestivum subsp. spelta) shows hulled or non-free threshing traits, although it is considered as one of the subspecies of common wheat and carries AABBDD genomes. Spelt and bread wheat are closely related and can produce fertile descendants via hybridization, although spelt wheat harbors distinct genetic variations in its genome (Kippes ). Even though spelt wheat has only minor economic importance at present, it has potential as a genetic resource for the breeding of bread wheat, because of its superior genetic variation for traits such as yield and yield components, disease resistance, and abiotic stress tolerance (Campbell 1997, Raman , Xie ). However, spelt wheat also has many undesirable traits, including lower yield, higher plant height, and increased lodging, compared to the modern bread wheat varieties, which are disadvantageous in the modern agricultural system (Miedaner and Longin 2016).

Spelt wheat is best distinguished from bread wheat by its non-free threshing trait and the long and lax spike morphology. In spelt wheat, the glumes tenaciously enclose grains and a strong mechanical force is required to separate the grains from the glumes. The differences in spike morphology and threshability are largely controlled by the Q locus on chromosome 5A, with spelt and bread wheat carrying the q and Q alleles, respectively (Simons ). The q allele of the spelt wheat confers the non-free threshing trait and lax spike morphology (speltoid type), whereas the Q allele of the bread wheat confers the free-threshing trait and spike morphology of the square head type (Simons ). The non-free threshing trait and the speltoid type spike prevent efficient harvesting and could impact the yield; therefore, the q allele should preferably be eliminated from current bread wheat breeding programs. In addition to the effects of the Q locus, a wide range of variation in spike morphology, such as spike length and spikelet density has been reported in hybrid populations between spelt and bread wheat (Dvorak , Manickavelu , Onishi , Xie ). Therefore, understanding the genetic control of spike morphology, which is closely related to yield and other yield components, is necessary for the utilization of spelt wheat in bread wheat breeding programs.

Grain morphology is associated with yield, processing properties, and end-use quality; therefore, it is an important agronomic trait for wheat. Spelt wheat tends to have longer and larger grains than bread wheat (Gegas ), although bread wheat displays a lot of variations. Heading time and pre-harvest sprouting (PHS) tolerance are also important agronomic traits for adaptation to local environments in diverse regions. The degree of grain dormancy is considered to be the most critical factor for PHS tolerance and could be important especially in wet and humid climatic regions, such as Japan, because PHS is a major problem affecting grain quality. A number of QTLs for grain morphology, heading time and grain dormancy have been reported in bread wheat (Kamran , Mares and Mrva 2014), while there have been only a few studies involving the genetic analysis of spelt wheat (Liu , Manickavelu , Xie , Zanetti ).

To use spelt wheat as a breeding material, it is necessary to compare the QTLs responsible for agronomic traits detected in different populations derived from the crosses using diverse spelt germplasm as wide genetic variation in spelt wheat may exist. However, to date, only a few populations have been used in genome-wide QTL analyses (Liu , Manickavelu , Xie ). In the present study, we developed new recombinant inbred lines from a cross between spelt and bread wheat and conducted QTL analysis for agronomic traits, including spike and grain morphology, heading time, and grain dormancy.

Materials and Methods

Plant materials

The plant materials used in this study were recombinant inbred lines (RILs) developed from the F2 population of a cross between ‘Chinese Spring’ (CS) and a spelt wheat strain, ‘st. Rumania’ (RU), by the single-seed descent method. RU was the strain collected in Romania and provided by the National BioResource Project-Wheat (Strain ID: KT019-002), with support, in part, from the National BioResource Project of the Ministry of Education, Culture, Sports, Science & Technology (MEXT), Japan. Ninety-six RILs from F9 or later generations were used for the phenotypic evaluation and linkage map construction.

Cultivation

The RILs and their parental strains (CS and RU) were cultivated under three environmental conditions at the Obihiro University of Agriculture and Veterinary Medicine, in Obihiro, Japan.

In the experiments conducted in 2014 (designated as 2014), RILs and their parental strains were grown in the field. Seeds were sown in cell flats in late April. After approximately three weeks, the seedlings were transplanted in rows of 40 cm length (10 cm between plants and 30 cm between rows) in the experimental field covered with a transparent plastic roof to prevent damages from rain. Four plants per row were grown with one replication (row) for each RIL and with nine replications (rows) of parental strains. Eighteen of 36 plants were used for phenotypic evaluation of the parental strains. Fertilizer (N, 10 g; P2O5, 15 g; K2O, 9 g m−2) was applied before transplanting.

In the experiments conducted in 2015 (designated as 2015), seeds were sown in plastic boxes (60 × 15 cm, 8 l volume) in late April. The seedlings were grown for approximately three weeks in a greenhouse and then the plastic boxes were placed outside. Eight plants per box were grown for each RIL with one replication (box) and with five replications (boxes) for parental strains. After anthesis, the plastic boxes were moved under a plastic roof to prevent damages from rain. Four of eight plants for each RIL and 20 of 40 plants of the parental strains were used in the experiments. To eliminate variability due to soil factors, soil from the field was mixed well and apportioned to all plastic boxes. Fertilizer (N, 0.5 g; P2O5, 1.0 g; K2O, 0.5 g per box) was applied for each plastic box before seed sowing.

In the experiment conducted in 2016 (designated as 2016), the RILs were grown in a controlled environment to maintain constant temperature during the growth period, because variations in heading time triggers the RILs to develop spikes and grains under different temperature conditions. In addition, RILs were grown after vernalization because the range of heading time in the RILs was narrower post-vernalization than without. Seeds were sown in cell flats and grown for 10 days in an incubator (LH-350SP, Nippon Medical & Chemical Instruments Co., Ltd., Osaka, Japan) under an 8 h day-length and at 23°C day/18°C night temperature. Then, the seedlings were vernalized (8 h daylength and 5°C day and night temperature) for 30 days in the incubator. After vernalization, the seedlings were transplanted into pots (13 cm diameter, 1 l volume) and grown under controlled-temperature conditions with natural sunlight (22°C, 12 h day/16°C, 12 h night). Two plants for each RIL and eight plants for each of the parental strains were grown. Fertilizer (N, 0.05 g; P2O5, 0.1 g; K2O, 0.05 g per pot) was applied to each pot before transplanting.

Phenotypic evaluation

Spike and grain morphology and heading time (HT) were measured in all (2014, 2015, and 2016) the experiments. Spikes were harvested after maturity and were stored at room temperature (approximately 25°C) until measurement. Spike length (SPL) was measured from the base of the spike to the tip of the top-most spikelet excluding the awn. Number of spikelets (SPN) was counted and spikelet density (SPD) was calculated as SPN/SPL. After hand-threshing, grain length (GL) and width (GW) were measured using SmartGrain (Tanabata ). Grain thickness was measured using Vernier calipers. Two to four spikes and 10 to 40 grains were measured for each RIL. Eight to 20 spikes and 60 to 180 grains were measured for each parental strain.

Germination tests were conducted to evaluate the degree of grain dormancy in the 2014 and 2015 experiments. Germination tests could not be conducted in 2016 due to limited harvest material. Each spike was checked daily and harvested at physiological maturity (when the green color of the spike disappeared completely). Spikes were air-dried at room temperature (approximately 25°C) for approximately three days and were preserved at −20°C in a freezer until use. Two to six spikes of RILs and six to ten spikes for the parental strains were harvested. Grains were carefully separated from the glumes by hand-threshing and were placed on a filter paper wetted with water in petri dishes (90 mm diameter). The dishes were incubated at 20°C for 10 days in the dark. The number of germinated grains was counted daily. When the seed coat ruptured, the grain was considered to have germinated. The degree of grain dormancy was calculated as the germination index (GI), which was: GI = (10 × s + 9 × s + ... + 1 × s)/(total number of grains examined × 10), where s is the number of grains that germinated on the nth day (Mohan ).

Marker analysis and linkage map construction

Total DNA from each RIL was extracted as described by Monna and with slight modifications as per Cao . Briefly, small pieces of young leaves were sampled in a 2-ml tube. Leaves were frozen using liquid nitrogen and crushed using stainless steel beads. Approximately 500 μl of DNA extraction buffer, as per Monna , was added. After 15 min, 400 μl of chloroform:isoamyl alcohol (24:1) was added and mixed with the solution by inversion. After mixing, the tube was spun at 13,000 rpm for 15 min and the supernatant was collected in a new tube. Isopropanol was added to the tube and centrifuged again at 13,000 rpm for 15 min to precipitate the crude DNA. The DNA pellet was rinsed with 70% ethanol and dissolved in Tris-EDTA Buffer.

Amplification of genetic markers was performed by polymerase chain reaction (PCR) and using GoTaq® DNA polymerase (Promega Corporation, Madison, WI, USA) or KOD FX and KOD FX Neo DNA polymerase (TOYOBO CO., LTD., Osaka, Japan) or PrimeSTAR® GXL DNA polymerase (Takara Bio Inc., Shiga, Japan), according to the manufacturer’s protocols. The PCR products were electrophoresed in 3–4% agarose gels or 10% acrylamide gels (acrylamide: N,N′-methylenebisacrylamide = 29:1). The genotype bands were evaluated after staining with ethidium bromide.

For the linkage map construction, 160 simple sequence repeat (SSR) markers and four gene-specific markers were used. The SSR markers were selected based on the information from Somers and GrainGenes (https://wheat.pw.usda.gov/GG3/) (Supplemental Table 1). The four gene-specific markers of the Q (Asakura ), Vrn-A1 (Yan ), Vrn-D1 (Fu ), and ALMT-1 loci (Sasaki ), which were polymorphic between the parental strains were used for map construction. Based on the genotypes of 96 RILs, the linkage map was constructed using JoinMap® 4 software (van Ooijen 2006).

Data analysis

The MQM (multiple QTL model) mapping method was used for QTL analysis by using MapQTL® 6 (van Ooijen 2002). In the first step, interval mapping (IM) was performed to identify the putative QTLs. one marker at each putative QTL (with LOD greater than two) detected by IM was selected as a cofactor and multiple-quantitative trait loci model (MQM) mapping was performed. The markers closest to the significant QTL were finally selected as a set of cofactors. The genome-wide LOD threshold at the 5% level were determined using a permutation test for each trait (Churchill and Doerge 1994).

To analyze the interactions between the two QTLs for spikelet density, the differences between the genotypes were tested using the Tukey-Kramer test for multiple comparisons using PASW® statistics 18 (SPSS Inc., Chicago, IL, USA).

Results

Genetic variation in agronomic traits in RILs

The spike and grain morphology in RILs and their parental strains was examined under three environmental conditions (Table 1). The spelt wheat strain (RU) consistently showed longer SPL (13.1–13.9 cm) and lower SPD (1.3 spikelets/cm) than the bread wheat strain (CS) (6.9–8.0 cm in SPL, 2.5–2.6 spikelets/cm in SPD). Differences in SPN between CS and RU were small (0.5–2.1), whereas they were significant only in 2015 (p < 0.001) and 2016 (p = 0.015). In the RILs, there was large transgressive segregation in SPL, SPN, and SPD, and the lines with shorter SPL and greater SPD than CS were appeared in RILs (Table 1, Fig. 1A). The GL of RU (7.6–7.8 mm) was greater than that of CS (5.7–6.1 mm) and there were only small differences (within 0.2 mm) in GW between the two strains (Table 1, Fig. 1B). No significant differences were found with respect to GT. RILs showed continuous variation of GL between the values for parental strains. Transgressive segregation was observed for GW and GT.

Table 1

Comparison of spike and grain related traits in parental strains and RILs

Trait	Trial	CS	RU	t-test	RILs
					Mean	Range
SPL (cm)	2014	7.3	13.8	***	9.3	4.1–14.8
	2015	8.0	13.9	***	9.4	4.8–14.1
	2016	6.9	13.1	***	9.5	4.7–14.5
SPN	2014	18.1	18.6	ns	17.1	12.3–22.3
	2015	20.5	18.4	***	19.0	15.0–24.8
	2016	18.3	17.0	*	17.5	13.5–21.0
SPD (spikelets/cm)	2014	2.5	1.3	***	2.0	1.1–3.5
	2015	2.6	1.3	***	2.2	1.3–4.2
	2016	2.6	1.3	***	2.0	1.0–3.6
GL (mm)	2014	5.7	7.7	***	6.6	5.6–7.6
	2015	5.7	7.6	***	6.7	5.4–7.8
	2016	6.1	7.8	***	7.1	5.9–8.3
GW (mm)	2014	3.0	3.0	ns	3.0	2.4–3.2
	2015	2.7	2.9	**	2.9	2.6–3.3
	2016	3.1	3.0	*	3.2	2.7–3.4
GT (mm)	2014	2.6	2.6	ns	2.7	2.3–3.0
	2015	2.5	2.7	ns	2.7	2.3–3.2
	2016	2.9	2.9	ns	3.0	2.5–3.5
HT (days)	2014	63.7	65.5	*	68.8	61.3–86.0
	2015	61.8	60.5	ns	65.7	55.5–82.5
	2016	73.4	82.3	***	81.7	74.0–92.0
GI	2014	0.52	1.00	***	0.68	0.04–1.00
	2015	0.37	0.94	***	0.51	0.00–1.00

GL: Grain length, GW: Grain width, GT: Grain thickness, SPL: Spike length, SPN: Spikelet numbers per spike, SPD: Spikelet density (SPN/SPL), HT: Heading time, GI: Germination index.

indicate that the significance at the 5, 1 and 0.1% levels respectively. ns indicates non-significance.

Fig. 1

Variation in spike (A) and grain (B) morphologies observed in the parental strains and RILs.

The difference of HT between CS and RU in 2014 and 2015 without vernalization was small (1.8 and 1.3 days, respectively) but it increased (8.9 days) after vernalization in 2016 (Table 1). Transgressive segregation towards later HT was observed in all the experiments. The GI of RU was significantly higher than that of CS (Table 1). The GI of RU was nearly 1.0, indicating that most grains germinated on the first and second day because of a very low level of grain dormancy. Continuous distributions of GI were observed in the RILs.

QTL analysis

A genetic linkage map was constructed using 164 molecular markers (Fig. 2). The total map length was 1,902 cM and it covered the majority of the wheat genome compared to a consensus map of 2,569 cM as reported in Somers , although markers were scarce in some chromosomal regions.

Fig. 2

Linkage map indicating the positions of QTLs detected in the RILs. Markers used for map construction are shown on the left side of the chromosomes. Locations of QTLs detected by MQM mapping are represented as boxes with two-LOD support intervals. Filled and white boxes indicate the effects of the QTLs that increase the trait values by the RU- and CS-derived alleles, respectively. a, b, and c indicate QTLs detected in 2014, 2015, and 2016, respectively.

QTL analysis was conducted for eight examined traits and significant QTLs were detected for all traits under at least one environmental condition (Fig. 2, Table 2). For spike morphology, QTLs were detected on chromosomes 4D and 5A. The QTLs on chromosome 5A (QSpl.obu-5A and QSpd.obu-5A) were detected in the region including the Q locus and had strong effects that increased SPL and decreased SPD due to the RU-derived allele. The effects of QSpl.obu-5A and QSpd.obu-5A were consistent with the expected genetic effects of the Q locus, a major genetic locus that controls spike morphology and threshability (Simons ). In contrast, QTLs on chromosome 4D (QSpl.obu-4D and QSpd.obu-4D) had effects that decreased SPL and increased SPD due to the RU-derived allele, although the magnitude of these effects was smaller than QSpl.obu-5A and QSpd.obu-5A. QSpd.obu-4D was detected in all three environmental conditions.

Table 2

Results of MQM mapping for spike and grain related traits in the RILs

Trait	Trial	Chromosome	QTL	LOD	PVE (%)	Additive effect
SPL	2015	4D	QSpl.obu-4D	3.5	6.0	7.0
	2014	5A	QSpl.obu-5A	20.5	63.8	−2.3
	2015	5A	QSpl.obu-5A	21.9	63.8	−2.1
	2016	5A	QSpl.obu-5A	29.7	75.9	−2.6
SPN	2014	5A	QSpn.obu-5A	5.3	23.0	0.9
SPD	2014	4D	QSpd.obu-4D	4.6	7.6	−0.2
	2015	4D	QSpd.obu-4D	4.9	9.2	−0.2
	2016	4D	QSpd.obu-4D	3.3	5.3	−0.2
	2014	5A	QSpd.obu-5A	22.0	56.0	0.6
	2015	5A	QSpd.obu-5A	18.0	50.1	0.5
	2016	5A	QSpd.obu-5A	21.4	56.0	0.5
GL	2014	1A	QGl.obu-1A	2.9	9.9	−0.1
	2015	1A	QGl.obu-1A	3.1	6.7	−0.1
	2016	1A	QGl.obu-1A	4.4	16.2	−0.2
	2014	2D	QGl.obu-2D	5.4	19.3	−0.2
	2015	2D	QGl.obu-2D	5.0	11.3	−0.2
	2016	2D	QGl.obu-2D	2.9	10.3	−0.2
	2015	3A	QGl.obu.-3A	3.5	7.7	0.1
	2015	4D	QGl.obu-4D	3.9	8.5	0.1
GW	2016	2B	QGw.obu-2B	5.9	24.7	0.1
	2014	5A	QGw.obu-5A	3.0	13.7	−0.1
	2015	5A	QGw.obu-5A	4.6	20.1	−0.1
GT	2014	5A	QGt.obu-5A	5.4	23.4	0.1
	2015	5A	QGt.obu-5A	8.4	33.3	0.1
	2016	5A	QGt.obu-5A	7.8	31.0	0.1
HT	2016	2B	QHt.obu-2B	6.7	25.3	−2.0
	2014	5A	QHt.obu-5A	10.5	37.9	3.5
	2015	5A	QHt.obu-5A	17.3	54.6	5.6
	2014	7B	QHt.obu-7B	4.0	12.7	−2.2
	2015	7B	QHt.obu-7B	4.5	9.9	−2.7
	2016	7B	QHt.obu-7B	3.8	12.6	−1.8
GI	2014	2B	QGi.obu-2B	3.1	14.6	−0.10
	2015	2B	QGi.obu-2B	3.8	19.4	−0.12

LOD values, PVE (phenotypic variance explained by each QTL), and additive effect are listed. Positive and negative values of additive effect indicate the effect increasing trait values by the CS and RU alleles, respectively.

QTLs with stable effects on grain morphology under all environmental conditions were detected on chromosomes 1A (QGl.obu-1A) and 2D (QGl.obu-2D) for GL, and near the Q locus on chromosome 5A for GT (QGt.obu-5A) (Fig. 2, Table 2). The RU-derived alleles at QGl.obu-1A and QGl.obu-2D increased GL, while the CS-derived alleles at QGt.obu-5A increased GT. QTLs for GL on chromosomes 3A and 4D and those for GW on chromosomes 2B and 5A were significant under only one or two environmental conditions.

For HT, three QTLs were found on chromosomes 2B (QHt.obu-2B), 5A (QHt.obu-5A) and 7B (QHt.obu-7B) (Fig. 2, Table 2). The QTL QHt.obu-5A had the largest effect (37.9 and 54.6% in PVE) in 2014 and 2015 without vernalization, while it could not be detected in 2016 with vernalization. In contrast, the QTL QHt.obu-2B was only visible in the 2016 experiments and required vernalization. The QTL QHt.obu-7B had stable effects in conferring later heading due to the RU-derived allele in all environments with or without vernalization. Most QTLs for spike and grain morphology were independent to QTLs for HT, except that a number of QTLs were tightly linked in the Q locus—QHt.obu-5A region.

For GI, a significant QTL (QGi.obu-2B) was detected on chromosome 2B in two environmental conditions examined (Fig. 2, Table 2). The RU-derived allele increased the GI (decreased the level of grain dormancy). Although QGi.obu- 2B and QHt.obu-2B were co-located on chromosome 2B, these two QTLs were detected independently in different environments.

Interaction of QTLs for spikelet density

Conspicuous transgressive segregation was observed for SPD, in which a compact spike appeared in the RILs (Fig. 1A). A QTL for SPD on chromosome 4D (QSpd.obu-4D) led to an increase in SPD through the RU-derived allele, although RU normally showed a lax spike morphology. To understand the genetic mechanism of transgressive segregation, the interaction between QTLs on chromosomes 5A (Q locus) and 4D (QSpd.obu-4D) was examined (Fig. 3). The RILs were classified into four genotype classes based on the allele combinations of the two loci. Under the presence of the q allele derived from RU, QSpd.obu-4D had no effect on SPD; however, the RU-derived alleles at QSpd.obu-4D caused significant increases in SPD in combination with the Q allele derived from CS, indicating that the q allele suppressed the effect of the RU-derived alleles at QSpd.obu-4D.

Fig. 3

Interaction of two QTLs for spikelet density on chromosomes 5A (the Q locus) and 4D (QSpd.obu-4D). Two markers (wmc720 and ALMT-1) or three markers (barc319, Q, and Vrn-A1) were used to determine the genotype of QTL on chromosomes 4D and 5A, respectively. Different letters on the bars indicate significant differences between genotypes in the same experiment. *: Six lines were used in 2015.

Comparison to known QTLs or genes

The QTLs with stable effects for spike and grain morphology detected in the present study (QGl.obu-1A, QGl.obu-2D and QSpd.obu-4D) were compared with the previously reported QTLs in populations derived from crosses between spelt and bread wheat (Fig. 4) (Manickavelu , Xie , 2015b). The QTL QGl.obu-2D was consistent with the region detected in Manickavelu , in which the same parent strain of bread wheat (CS) was crossed with a different spelt wheat strain. The Tg-D1 gene for non-free threshing on chromosome 2D has pleiotropic effects on increasing grain length (Okamoto ); the location of Tg-D1 is different from that of QGl.obu-2D, as this QTL was mapped at the terminal region of the long arm of the chromosome. Manickavelu also identified a QTL for SPD (QCpt) near QSpd.obu-4D on chromosome 4D, with both the Q locus and QCpt on chromosome 4D increasing SPD due to the spelt wheat allele in this population (Manickavelu ). It was still unclear whether QSpd.obu-4D and QCpt are identical QTL or not because of lack of common molecular markers used in Manickavelu and this study. The QTL QGl.obu- 1A was detected only in the present study.

Fig. 4

Comparative maps of QTL locations on chromosomes 2D (A) and 4D (B). The linkage map in this study and the framework map based on Somers are shown on the left and right, respectively. QTLs detected in this study are shown on the left linkage map, as boxes with two-LOD support intervals. The locations of reported QTLs or genes are shown on the right linkage map. a) Bars represent the regions including the LOD peaks of QTLs reported by Manickavelu . b) The location of Tg-D1 was reported by Dvorak . The location of ALMT-1 was reported by Ma .

Regarding heading time, the LOD peak of QHt.obu-5A conferring later heading without vernalization due to the allele from CS was located on the gene specific marker from the Vrn-A1 locus. This result was consistent with the fact that CS has the vrn-A1 allele (winter allele) on the Vrn-A1 locus (Yan ). The location of QHt.obu-2B and QHt.obu-7B were similar to the location of Ppd-B1 (Fig. 5, Nguyen ) and Vrn-B3 (Yan ), respectively. Although a QTL for heading time was also identified on chromosome 7B in a bread wheat × spelt cross (Manickavelu ), its location and effect were different from that of the QTL QHt.obu-7B (CS-derived allele conferred later heading). In bread wheat, a number of QTLs for grain dormancy or PHS tolerance exists near the QGi.obu-2B region (Fig. 5, Chao , Kulwal , Kumar , 2015, Liu , Mohan , Singh , Somyong , Zhang ). Fine mapping by Somyong revealed at least two QTLs in this region. In addition, the wheat TaSdr gene, the ortholog of OsSdr4 conferring grain dormancy in rice, was also present in this region (Zhang ).

Fig. 5

Comparative maps of QTL locations for grain dormancy or PHS tolerance on chromosome 2B. The linkage map in this study and the framework map based on Somers are shown on the left and right, respectively. QTLs detected in this study are shown on the left linkage map as boxes with two-LOD support intervals. The locations of the reported QTLs are shown on the right linkage map. Bars represent the regions including the LOD peaks of QTLs reported by a) and b) Chao ; c) Kumar ; d) Kulwal ; e) Liu ; f) Mohan ; g) Singh ; h) Somyong ; i) Kumar ; and j) Zhang . The location of Ppd-B1 was reported by Nguyen .

Discussion

The potential of spelt wheat as a genetic resource for the breeding of bread wheat has long been recognized (Campbell 1997). A wide range of phenotypic variation in spelt wheat for traits such as protein and mineral concentration of grains (Gomez-Becerra ), aluminum tolerance (Raman ), and various agronomic traits (Miedaner and Longin 2016, Würschum ) have been evaluated. However, studies identifying genes or QTLs associated with these traits in populations derived from crosses between spelt and bread wheat remain scarce (Liu , Manickavelu , Xie , Zanetti ). Identification of QTLs for valuable agronomic traits in spelt wheat is necessary to exploit its full potential for use in bread wheat breeding programs.

Spike morphology is a critical trait that affects the number of grains per spike and the grain morphology. In addition, spike morphology, including threshability, is taxonomically important because it is the most distinguishing trait between the bread and spelt wheat types. Therefore, understanding the genetic control of spike morphology is important not only for breeding, but also to clarify the origin of spelt wheat (Dvorak , Mac Key 1966). Transgressive segregations showing segregants with higher SPD than bread wheat were reported in several cross combinations between spelt and bread wheat (Dvorak , Manickavelu , Onishi , Xie ). Dvorak indicated that spelt wheat had the C allele for compact spikes on chromosome 2D despite the fact that spelt wheat had non-compact spikes, which supported the hypothesis that some spelt wheat, especially European spelt wheat, may have originated from hybridization between T. aestivum subsp. compactum carrying the C allele and emmer wheat, which is a hulled or nonfree-threshing tetraploid wheat (Dvorak , Mac Key 1966). In contrast, QTLs increasing SPD, instead of the C locus on chromosome 2D, due to alleles from spelt wheat were reported (Xie ). In the present study, a QTL for SPD showing stable effects under all the tested environmental conditions was detected on chromosome 4D (QSpd.obu-4D). In addition, the effect of QSpd.obu-4D was suppressed in the presence of the q allele, implying that a cryptic variation, which can be revealed by crossing with bread wheat, has accumulated in spelt wheat and may reflect the different origins of spelt wheat.

The longer grain length observed in spelt wheat is one of the distinguishing characteristics in comparison to bread wheat (Campbell 1997, Gegas ). A number of QTLs for grain length were detected in bread wheat × spelt crosses (Manickavelu , Xie ). Although the Tg-D1 gene on chromosome 2D was reported to have pleiotropic effects in increasing grain length (Okamoto ), its utilization in bread wheat breeding programs has been difficult because Tg-D1 also confers the non-free-threshing trait. The QTLs QGl.obu-1A and QGl.obu-2D were independent of the QTLs for spike-related traits, thus highlighting their potential use in controlling grain length in bread wheat. The QTL QGl.obu-1A is a novel QTL obtained from the bread wheat × spelt cross although a few QTLs were previously reported at similar regions in bread wheat (Gegas ). Further mapping of QGl.obu-1A is required for comparison with other QTLs found in bread wheat.

Regarding the heading time, the QTL QHt.obu-7B, which had stable effects in environments with and without vernalization, was detected near the Vrn-B3 region on the chromosome 7B. The Vrn-B3 locus was cloned as the TaFT1 gene and a retrotransposon insertion was detected in the promoter region of the Vrn-B3 allele from the cultivar HOPE, which conferred earlier heading time compared to the CS-derived allele (vrn-B3) (Yan ). Previously, we identified a QTL in the Vrn-B3 region in RILs between Zenkouji-komugi × CS and the Zenkouji-komugi-derived allele conferred earlier heading time compared to the CS-derived allele, despite the lack of retrotransposon insertions in Zenkouji-komugi (Cao ). These results provided evidences that at least four alleles in a single or multiple QTL(s) could be located in the Vrn-B3 region, and that spelt wheat could serve as a novel genetic resource for QTL to modify heading time through breeding.

Grain dormancy of RU was lower than that of CS, although CS was revealed to be a weaker strain for grain dormancy in comparison to most Japanese wheat varieties (Chono ). This result suggests that weak PHS tolerance may be a major obstacle in the utilization of spelt wheat as a genetic material for breeding in wet and humid regions such as Japan. To date, only a few studies have reported the evaluation of grain dormancy in spelt wheat (Liu , Ruzhitskaya and Borysova 2014) or PHS tolerance (Zanetti ). In the present study, a QTL with stable effects (QGi.obu-2B) conferring weak dormancy due to the spelt wheat-derived allele was detected on chromosome 2B. In this region, as shown in Fig. 5, a number of QTLs, as well as some candidate genes affecting grain dormancy or PHS tolerance were reported. This indicates that the QGi.obu-2B region might be important for diversity of grain dormancy, not only among bread wheat varieties, but also between spelt and bread wheat. Fine mapping of QGi.obu-2B is needed for marker-assisted selection against the weak dormancy allele of spelt wheat. In addition, further elucidation of the genetic variation associated with grain dormancy in spelt wheat is important to determine its utility in breeding for diverse environmental conditions.

Spelt wheat was genetically differentiated from bread wheat based on DNA polymorphisms, such as single nucleotide polymorphisms (SNPs) in their genomes (Kippes ). In addition, spelt wheat has specific taste and quality characteristics (Longin and Wurschum 2016). These results imply that there might be spelt wheat-specific genes or QTLs associated with phenotypes important for breeding. Further exploration and identification of useful genes or QTLs could accelerate the utilization of spelt wheat as a genetic resource in bread wheat breeding programs.