Wheat (Triticum aestivum L.) is one of the most impor-tant staple crops, providing approximately 20% of the dietary calorie intake for the world population (FAO 2023). While the wheat production in many developing countries almost doubled during the Green Revolution in the 1960s, productivity has been stagnated over the recent few decades (Hazell 2009; Ray et al. 2012). To meet the demand of a rapidly growing world population, at least a 15% increase in wheat production is required by 2050 (Gupta et al. 2020). Therefore, improving yield is an essential goal of wheat breeding programs to meet the rising food demand and overcome yield stagnation. Wheat yield is a complex trait determined by the major yield component traits (i.e., tillers per unit area, kernels per spike, and kernel weight) and affected by other physiolo-gical traits including heading date and culm length as well as various environmental factors such as day length and temperature (Chen et al. 2012). As the final crop yield is a product of numerous genetic and environmental factors inter-related to each other, genetic studies on yield are generally conducted on individual yield components such as tillers per unit area, spikelets per spike, kernels per spikelet, and kernel weight that are less affected by the environment and have higher heritability than yield itself (Cui et al. 2014; Wu et al. 2014; Assanga et al. 2017; Zhang et al. 2018; Deng et al. 2019; Liu et al. 2019).

Dissecting the genetic loci underlying culm length and days to heading date (DHD) is also crucial to maximize yield, as the major genes controlling these traits affect yield components (Tshikunde et al. 2019; Cao et al. 2020). During the Green Revolution, adopting the semi-dwarfing alleles (i.e., Rht-B1b and Rht-D1b) provided substantial benefits on improving yield by conferring lodging tole-rance under high fertilization and increasing the harvest index (Peng et al. 1999). However, as these alleles reduce seedling vigor and coleoptile length as well as the final plant height, their applicability is limited in warm and dry conditions where deep sowing is required (Rebetzke et al. 1999). Therefore, researchers have been searching for alternative genes that can better adapt to suboptimal areas and maintain grain yield under low input conditions. (Thomas 2017; Pearce 2021; Xiong et al. 2022). Opti-mizing heading date is also essential for adaption across various environments, as the yield performance is influ-enced by the duration of the vegetative phase, floret development, and maturity (Zhang et al. 2021). For exam-ple, early heading may help avoid heat and drought stress during anthesis and grain filling, thereby enabling high yield in hot and dry regions (Bentley et al. 2013; Mondal et al. 2016). Therefore, understanding the genetic architecture underlying heading date is crucial for breeding cultivars with proper heading dates in a given environment. The main genetic factors controlling wheat heading date include genes related to vernalization (Vrn), photoperiod (Ppd), and earliness per se (Eps) (Herndl et al. 2008). Vrn genes such as VRN1, VRN2, and VRN3 regulate vernaliza-tion sensitivity (Herndl et al. 2008, 2006), and Ppd genes including PPD1 and PPD2 mainly control the photoperiod response (Law et al. 1978; Scarth and Law 1983; Khlestkina et al. 2009). Eps is known to be responsible for the flowering time differences after the vernalization and photoperiod requirements are met (Kato et al. 1999a, 1999b; Kamran et al. 2014). EARLY FLOWERING 3 (ELF3), a circadian clock regulator gene, was identified as the candidate gene for Eps (Alvarez et al. 2016).

In Korea, wheat is mostly cultivated as a secondary crop in rice-wheat double cropping practices. Therefore, late heading causes wheat harvesting without sufficient grain filling in order to sow the main crop such as rice in an optimal planting season, resulting in a decrease in wheat yield and quality (Cho et al. 2015a). Also, the maturation stage of late-heading wheat cultivars often overlaps with the rainy season, which increases the risk of pre-harvest sprouting and deteriorates end-use quality (Shin et al. 2013). Therefore, developing early-heading cultivars re-mains one of the most important goals for Korean wheat breeders since the 1970s (Cho et al. 2015b). However, previous studies showed that the major heading date genes such as VRN1 and PPD1 exhibit highly limited allelic variation in Korean wheat cultivars and landraces, neces-sitating research on other genes affecting heading date (Lee et al. 2014; Cho et al. 2015a, 2015b, 2018). As early heading often accompanies a decrease in yield due to reduced vegetative growth, it is also important to evaluate the effects of heading date genes on yield and its com-ponents. In this study, we aimed to identify quantitative trait loci (QTLs) for yield-related traits including DHD, using doubled haploid lines derived from the cross between two Korean cultivars, Keumkang and Olgeuru. We also evaluated the impact of the DHD QTLs on major yield- related traits and discussed the potential application of them in breeding programs.

Plant materials and field trials

A population of 94 doubled haploid lines was produced from the cross between Korean wheat cultivars, Keumkang and Olgeuru, using the wheat × maize system described by Inagaki and Mujeeb-Kazi (1995). Keumkang is a popular cultivar with good milling, end-use quality and early matu-rity (Song 1998). Olgeuru exhibits a higher yield than Keumkang (Nam 1994). The doubled haploid lines were grown at the Upland Crop Experimental Farm of the National Institute of Crop Science, Rural Development Administration (Iksan, Korea) for seed multiplication in 2009. Field trials were performed in upland condition at Jeonbuk National University (35°85΄N 127°13΄E) in Jeonju, South Korea, during three growing seasons (2017, 2018, and 2019). The seeds were sown in late October, and each plot consisted of three 2 m rows spaced 25 cm apart. The plots were combine-harvested in mid-June. The field management was conducted according to the standard wheat cultivation manual of Rural Development Admini-stration (RDA), South Korea (RDA 2012).

Evaluation of yield-related traits

A total of eight yield-related traits were evaluated: DHD, culm length (CL), spike length (SL), kernels per spike (KPS), tiller number per m² (TN), test weight (TW), thousand kernel weight (TKW), and yield (YD). DHD was recorded as the number of days from sowing date to heading date when 50% of the plants in each plot showed the emergence of spikes from the stem. CL was determined by measuring the length from the ground to the base of the spike and SL was measured as the length of the spike excluding awns. KPS was measured as the number of kernels per main spike. CL, SL, and KPS were measured from the main culms of ten plants randomly selected in each plot at maturity. TN was counted as the number of stems per m². TW, TKW, and YD were measured after harvesting and drying the grains at 14% moisture content. TW and TKW were evaluated by weighing one liter of grain and 1,000 kernels, respectively. YD was recorded by weighing the grains from each plot and converting the value in kg/10a.

QTL mapping and statistical analysis

Genomic DNA was extracted from young leaf tissue using a DNA extraction kit (Solgent, South Korea) according to the manufacturer’s instruction. The 94 dou-bled haploid lines and the parental cultivars were geno-typed using Axiom^TM 35K Wheat Breeder’s Genotyping Array (ThermoFisherScientific, Applied Biosystems^TM, US). QTL mapping was performed using IciMapping program version 4.2 (Meng et al. 2015). Out of 35,042 SNPs, 3,325 were polymorphic between Keumkang and Olgeuru. After filtering the SNPs with more than 10% missing data and those with overlapping genetic positions using the ‘BIN’ function, 641 SNPs were finally used to construct a linkage map by the ‘MAP’ function. QTL analy-sis was performed by the inclusive composite interval mapping with the LOD threshold of 3.0, and those identi-fied in at least two of the three years were declared as stable QTLs. Statistical analyses including ANOVA and correlation analysis were performed using R (The R Project for Statistical Com-puting version 4.2.0, https://www.r-project.org). Three-way ANOVAs and Tukey’s Honestly Significant Difference (HSD) test were conducted using the “agricolae” package (de Mendiburu 2023).

Phenotypic variation and correlations among yield-related traits

Distribution of the eight yield-related traits (DHD, CL, SL, KPS, TN, TW, TKW, and YD) in the 94 doubled haploid lines and their parents evaluated in three years are illustrated in Fig. 1 and Supplementary Table S1. The average KPS (33.6) and YD (437.3 kg/10a) of the doubled haploid lines were between the values of Keumkang (33.4 and 427.1 kg/10a) and Olgeuru (35.1 and 485.7 kg/10a) in three years, while the average SL (8.1 cm), TN (851.6/m²), TW (806.0 g/l), and TKW (39.7 g) were lower than the values of both Keumkang (8.4 cm, 940.3/m², 815.7 g/l and 42.4 g) and Olgeuru (8.6 cm, 1008.4/m², 827.0 g/l, 41.4 g) in three years. The parental values for DHD and CL showed inconsistent patterns (Keumkang headed earlier than Olgeuru in 2017 and 2019, while Olgeuru headed earlier in 2018; Keumkang was taller than Olgeuru in 2017 and 2018, while Olgeuru was taller in 2019). Transgressive segregation was observed for all investigated traits.

Figure 1. Distribution of the yield-related traits in the Keumkang/Olgeuru doubled haploid population. Black vertical line indicates the mean value. Red and blue arrows indicate the value of Keumkang and Olgeuru, respectively.

The correlations among the yield-related traits are sum-marized in Supplementary Figs. S1-S3. Thirteen pairs of traits showed sig-nificant correlations in all three years. While the DHD-CL (r = 0.33-0.45), DHD-SL (r = 0.36-0.46), DHD-KPS (r = 0.55-0.74), CL-SL (r = 0.22-0.45), CL-TKW (r = 0.31-0.46), SL-KPS (r = 0.44-0.48) correlations were signifi-cantly positive in all three years, the DHD-TN (r = ‒0.38 to ‒0.21), DHD-TW (r = ‒0.45 to ‒0.29), SL-TN (r = ‒0.31 to ‒0.25), SL-TW (r = ‒0.47 to ‒0.34), KPS-TN (r = ‒0.41 to ‒0.25), KPS-TW (r = ‒0.50 to ‒0.35), and TN-TKW (r = ‒0.40 to ‒0.27) correlations were significantly negative in all three years. However, YD did not exhibit such consis-tent correlation with any other trait across three years, reflecting its complex nature affected by different yield components and environmental factors.

QTLs for yield-related traits

A linkage map covering 2639.0 cM was constructed using 641 SNPs with unique genetic positions out of the 3,325 polymorphic SNPs. The average chromosome length was 125.7 cM, ranging from 7.9 cM (6D) to 288.7 cM (5A). Chromosome 5B had the greatest number of markers (76), while chromosome 5D had the smallest number of markers (3). The overall SNP density was 4.39 cM, with the highest density on 6D (1.3 cM) and the lowest density on 7D (9.6 cM).

A total of seven QTLs, three for DHD and one each for CL, KPS, TKW, and TW, were detected in at least two of the three years (Table 1 and Fig. 2). Three QTLs for DHD, QDhd.jbnu-3B, QDhd.jbnu-6B, and QDhd.jbnu-7D, were stably detected with the LOD scores of 6.3-18.9, 4.7-27.2, and 4.6-8.1, respectively. QDhd.jbnu-6B explained 12.2- 33.1% of the phenotypic variance, and Keumkang contri-buted the allele for late heading with the additive effects of 0.9-2.2 days. For QDhd.jbnu-3B and QDhd.jbnu-7D, Olgeuru contributed the allele for late heading with the additive effects of 1.1-2.2 and 1.0 days, explaining 17.4-40.9% and 6.0-8.7% of the phenotypic variance, respectively. For CL, QCl.jbnu-6A explained 13.8-27.5% of the phenotypic variance with the LOD scores of 4.7-10.3. Keumkang contributed the allele for a long culm length with the additive effects of 2.8-4.0 cm. For KPS, QKps.jbnu-3B explained 15.6-30.6% of the phenotypic variance with the LOD scores of 4.6-8.1. Olgeuru con-tributed the allele for high kernels per spike with the addi-tive effects of 1.1-2.6. For TKW, QTkw.jbnu-6A explained 15.2-22.7% of the phenotypic variance with the LOD scores of 3.3-5.3. The positive allele for QTkw.jbnu-6A was contributed by Keumkang with the additive effects of 1.3-1.6 g. For TW, QTw.jbnu-1A explained 13.2-17.9% of the phenotypic variance with the LOD score 3.7-3.9. Keumkang contributed the allele for high test weight with the additive effects of 5.7-6.1 g/l. QDhd.jbnu-3B and QKps.jbnu-3B were detected on the same position (117.0 cM), and QCl.jbnu-6A and QTkw.jbnu-6A were located within a 3.0 cM interval, indicating that these QTLs may have pleiotropic effects on two different traits (Supplementary Table S2). No stable QTL was detected for SL, TN, and YD.

Table 1 . QTLs for yield-related traits identified from the Keumkang/Olgeuru doubled haploid population

QTLz)	Year	Position (cM)	Left marker	Right marker	Interval (cM)	LODy)	PVEx) (%)	ADDw)
QDhd.jbnu-3B	2017	117.0	AX-95081398	AX-95147328	6.0	16.1	40.9	‒2.2
	2018	106.0	AX-94527403	AX-95081398	9.0	6.3	17.4	‒1.1
	2019	117.0	AX-95081398	AX-95147328	5.0	18.9	18.0	‒1.7
QDhd.jbnu-6B	2017	124.0	AX-94511377	AX-94916148	5.0	6.5	12.8	1.2
	2018	124.0	AX-94511377	AX-94916148	5.0	4.7	12.2	0.9
	2019	119.0	AX-94649598	AX-94511377	3.0	27.2	33.1	2.2
QDhd.jbnu-7D	2017	89.0	AX-94523269	AX-94944119	9.0	4.6	8.7	‒1.0
	2019	91.0	AX-94523269	AX-94944119	10.0	8.1	6.0	‒1.0
QCl.jbnu-6A	2017	75.0	AX-94718756	AX-94415763	8.0	10.3	27.5	4.0
	2018	75.0	AX-94718756	AX-94415763	4.0	6.1	25.6	3.2
	2019	72.0	AX-94722330	AX-94718756	10.0	4.7	13.8	2.8
QKps.jbnu-3B	2017	117.0	AX-95081398	AX-95147328	6.0	8.6	30.6	‒2.6
	2018	117.0	AX-95081398	AX-95147328	7.0	5.5	20.8	‒1.6
	2019	117.0	AX-95081398	AX-95147328	8.0	3.9	15.6	‒1.1
QTkw.jbnu-6A	2017	73.0	AX-94718756	AX-94415763	8.0	5.0	21.9	1.6
	2018	73.0	AX-94718756	AX-94415763	9.0	3.3	15.2	1.3
	2019	73.0	AX-94718756	AX-94415763	7.0	5.3	22.7	1.6
QTw.jbnu-1A	2018	5.0	AX-94895904	AX-94739433	7.5	3.9	17.9	6.1
	2019	4.0	AX-94448644	AX-94895904	7.5	3.7	13.2	5.7

^z)QTLs stably identified at least two of the three years are listed in the table. QTLs were named according to Mcintosh et al. (2017). DHD: days to heading date, CL: culm length, KPS: kernels per spike, TKW: thousand kernel weight, TW:

test weight.

^y)Logarithm of the odds.

^x)Phenotypic variance explained.

^w)Additive effect of allele substitution. The units are those of the corresponding traits. A positive sign indicates that the

Keumkang allele increased the trait value.

Figure 2. QTL mapping for yield-related traits in the Keumkang/Olgeuru doubled haploid population. TW: test weight, DHD: days to heading date, KPS: kernels per spike, CL: culm length, TKW: thousand kernel weight.

Effects of QDhd.jbnu-3B, QDhd.jbnu-6B, and QDhd.jbnu-7D on days to heading date and other traits

As shortening DHD with maintaining productivity is one of the most important goals of Korean wheat breeding programs, we focused on the three DHD QTLs for further analyses. To characterize the effects of the three DHD QTLs and their interactions, we conducted three-way ANOVAs of QDhd.jbnu-3B, QDhd.jbnu-6B, and QDhd.jbnu-7D for the yield-related traits (Table 2). The main effects of QDhd.jbnu-3B were highly significant for CL, SL, KPS, TN, TW, YD as well as DHD. Lines carrying the Keumkang allele at QDhd.jbnu-3B headed 3.1 days earlier than those carrying the Olgeuru allele. The Keumkang allele was also associated with a decrease in culm length (‒3.9 cm), spike length (‒0.4 cm), kernels per spike (‒3.5), and the final yield (‒27.9 kg/10a), while an increase in the number of tillers (+15/m²), test weight (+7.7 g/l), and thousand kernel weight (+0.7 g). The main effects of QDhd.jbnu-6B were significant for all traits except yield. For QDhd.jbnu-6B, lines carrying the Olgeuru allele headed 2.0 days earlier than those carrying the Keumkang allele. The Olgeuru allele at QDhd.jbnu-6B was associated with decreased culm length (‒2.6 cm), spike length (‒0.5 cm), kernels per spike (‒1.3), and thousand kernel weight (‒1.7 g), while increased number of tillers (+25.6/m²) and test weight (+8.9 g/l). The main effects of QDhd.jbnu-7D were sig-nificant for DHD, KPS, and TW. Lines carrying the Keumkang allele headed 1.7 days earlier than those with the Olgeuru allele. The Keumkang allele was associated with fewer kernels per spike (‒1.4) and higher test weight (+3.7 g/l). The alleles for early heading of the three QTLs were consistently associated with a decrease in kernels per spike and an increase in test weight.

Table 2 . Three-way ANOVAs of QDhd.jbnu-3B (Q3), QDhd.jbnu-6B (Q6), and QDhd.jbnu-7D (Q7) for different yield-related traits.

Traitz)	Allele	Main effecty)				Interaction
		Q3	Q6	Q7		Q3 × Q6	Q3 × Q7	Q6 × Q7	Q3 × Q6 × Q7
DHD (d)	Keumkang	181.0	183.8	182.0
	Olgeuru	184.1	181.8	183.7
	P-value	***	***	***		ns	ns	ns	**
	PVE (%)	27.7	13.1	9.5					1.2
CL (cm)	Keumkang	74.8	78.4	76.4
	Olgeuru	78.7	75.8	78.2
	P-value	***	*	ns		ns	ns	ns	ns
	PVE (%)	4.2	1.9
SL (cm)	Keumkang	7.8	8.3	8.1
	Olgeuru	8.2	7.8	8.1
	P-value	***	***	ns		ns	***	ns	ns
	PVE (%)	7.4	7.4				3.9
KPS (no.)	Keumkang	31.7	34.3	33.0
	Olgeuru	35.2	33.0	34.4
	P-value	***	***	***		**	ns	*	*
	PVE (%)	22.2	4.4	3.6		2.2		1.3	1.3
TN (no./m2)	Keumkang	873.0	819.8	854.8
	Olgeuru	835.5	883.8	847.0
	P-value	**	***	ns		ns	ns	ns	ns
	PVE (%)	2.9	9.6
TW (g/l)	Keumkang	811.1	801.9	807.5
	Olgeuru	803.4	810.8	803.8
	P-value	***	***	**		**	ns	ns	*
	PVE (%)	5.6	9.2	2.4		2.0			1.5
TKW (g)	Keumkang	39.9	40.4	39.7
	Olgeuru	39.2	38.7	39.6
	P-value	ns	***	ns		ns	ns	*	ns
	PVE (%)		5.7					1.4
YD (kg/10a)	Keumkang	421.5	440.8	436.3
	Olgeuru	449.4	434.6	438.7
	P-value	***	ns	ns		ns	ns	ns	ns
	PVE (%)	8.7

^z)DHD: days to heading date, CL: culm length, SL: spike length, KPS: kernel per spike, TN: tiller number per m², TW: test weight, TKW: thousand kernel weight, YD: yield.

^y)Q3, Q6, and Q7 indicate QDhd.jbnu-3B, QDhd.jbnu-6B, and QDhd.jbnu-7D, respectively. Mean values of the doubled haploid lines carrying the Keumkang allele and the Olgeuru allele are indicated for each QTL. Asterisks indicate significance (*P < 0.05, **P < 0.01 and ***P < 0.001; ns: not significant) from the three-way factorial ANOVA with the year (2017, 2018, and 2019) as a random factor and the three QTLs as fixed factors. Phenotype variance explained (PVE) is indicated only for significant effects. QDhd.jbnu-3B, QDhd.jbnu-6B, and QDhd.jbnu-7D were represented by the markers AX-95081398, AX-94511377, and AX-94523269, respectively.

Although the two-way and three-way interactions bet-ween the three DHD QTLs were significant for a few traits (i.e., QDhd.jbnu-3B × QDhd.jbnu-6B interaction for KPS and TW, QDhd.jbnu-3B × QDhd.jbnu-7D interaction for SL, QDhd.jbnu-6B × QDhd.jbnu-7D interaction for KPS and TKW, and the three way interaction for DHD, KPS, and TW), they explained limited proportion (< 4%) of the phenotypic variances. Notably, pyramiding two or more alleles for early heading at QDhd.jbnu-3B, QDhd.jbnu-6B, and QDhd.jbnu-7D was highly effective in accelerating heading, ranging from 179.7 days in the lines with the three alleles for early heading to 186.2 days in the lines with the three alleles for late heading (Fig. 3).

Figure 3. Days to heading date of the doubled haploid lines carrying different allele combinations of QDhd. jbnu-3B, QDhd.jbnu-6B, and QDhd.jbnu-7D. ‘+’ indicates the allele for early heading while ‘‒’ in-dicates the allele for late heading. The Keumkang allele is associated with early heading for QDhd. jbnu-3B and QDhd.jbnu-7D, while the Olgeuru allele is associated with early heading for QDhd.jbnu-6B. Different letters above the bars indicate significant difference from the Tukey’s HSD test at P < 0.05.

This study identified seven stable QTLs for yield-related traits using the doubled haploid lines derived from the two Korean wheat cultivars, Keumkang and Olgeuru. As early heading is especially important in Korean wheat breeding programs, we focused on the three DHD QTLs (i.e., QDhd.jbnu-3B, QDhd.jbnu-6B, and QDhd.jbnu-7D) to cha-racterize their effects and interactions on DHD and other yield-related traits. In this section, we compare the three DHD QTLs with those identified in previous studies and discuss their potential uses in wheat breeding programs.

Comparison of the three DHD QTLs with previous studies

QDhd.jbnu-3B was detected in three consecutive years by the flanking marker AX-95081398 (810.6 Mb; IWGSC RefSeq v1.0) on chromosome 3B (Table 1 and Fig. 2). Com-parative analysis revealed that it was co-located with QKNS.caas-3B, a QTL for KPS flanked by the markers BobWhite_c22016_155 (IWB1543; 789.3 Mb) and RAC875_ c10909_1180 (IWB53203; 805.6 Mb) (Gao et al. 2015). While the effects of QKNS.caas-3B on DHD are unknown, the co-location of QDhd.jbnu-3B and QKps.jbnu-3B by the same flanking marker (AX-95081398) in our study (Table 1) indicate that the gene underlying this locus may have a pleiotropic effect on DHD and kernels per spike. QDhd.jbnu-6B identified in three consecutive years was flanked by AX-94511377 (663.5 Mb; IWGSC RefSeq v1.0) on chromosome 6B (Table 1 and Fig. 2). While no known QTL for DHD was co-located with this region, it was in a similar chromosomal region with two loci identified in genome-wide association mapping studies (GWAS), i.e., wsnp_Ex_c1276_2445537 (642.3 Mb) associ-ated with kernel weight per spike (Lozada et al. 2017) and AX_109917592 (677.4 Mb) associated with TKW (Li et al. 2019). As QDhd.jbnu-6B exhibited pleiotropic effects on DHD and TKW in the present study (Table 2), further study is required to test if the gene underlying QDhd.jbnu-6B is the same gene with those underlying the two kernel weight loci identified from GWAS. As the pseudomolecule locations (IWGSC RefSeq v1.0) of the flanking markers for QDhd.jbnu-7D were unclear (e.g., the location of AX-94523269 is unknown while AX-94944119 is on chromosome 7B; but both are located on chromosome 7D according to the consensus map), we did not compare QDhd.jbnu-7D with previously reported QTLs.

Potential applications of the DHD QTLs in wheat breeding programs

Heading date is an important trait to secure stable yield under target environments and cropping system. In Korea, developing early heading wheat cultivars is especially important to allow the optimum transplanting period of rice in the wheat-rice double cropping system while maintain-ing stable yield and quality of wheat. As Korean wheat breeding programs have focused on developing early-heading cultivars since the 1970s, most recent commercial cultivars already carry the alleles for early heading at the major heading date genes such as VRN1 and PPD1 (Lee et al. 2014; Cho et al. 2015a; Cho et al. 2015b; Cho et al. 2018). Therefore, the three DHD QTLs identified in this study provide useful genetic resources to develop early heading cultivars in Korean wheat breeding programs. When the alleles for early heading at QDhd.jbnu-3B, QDhd.jbnu-6B, and QDhd.jbnu-7D were pyramided, the heading date could be advanced by approximately one week (Fig. 3).

However, caution is required because early heading is often associated with decreased spikelets per spike and kernels per spike, thus reduces the yield potential (Muterko et al. 2015; Hu et al. 2023). For example, the spring alleles of Vrn-A1 significantly shortened the vegetative phase and early reproductive phase, resulting in the reduced spikelet number per spike (Stelmakh 1992, 1998; Dreisigacker et al. 2021). Similarly, the photoperiod-insensitive alleles of Ppd-D1 are associated with the shortened period of terminal spikelet formation, which resulted in accelerated heading and a decreased spikelet number per spike (Ransom et al. 2007; Dreisigacker et al. 2021). Similar to these, the three DHD QTLs detected in this study affected heading date and kernels per spike at the same time, where the alleles for early heading were associated with reduced kernels per spike (Table 2). As the yield reduction associated with the allele for early heading was more severe for QDhd.jbnu-3B (‒6.21%) compared with QDhd.jbnu-6B (‒1.40%) and QDhd.jbnu-7D (‒0.55%), pyramiding the alleles for early heading at QDhd.jbnu-6B and QDhd.jbnu-7B might be beneficial for accelerating heading while minimizing the negative impact on yield. Further research is required to evaluate the effects of these QTLs on heading date and other important agronomic traits under different genetic backgrounds and environments.

The datasets generated from this research are available from the corresponding author upon reasonable request. 

This work was carried out with the support of the “Cooperative Research Program for Agriculture Science & Technology Development (RDA PJ0159652023)”, Rural Development Administration, Republic of Korea.

Sumin Hong performed the experiments, analyzed the data and wrote the first draft. Kyeong-Min Kim, Changhyun Choi, and Seong-Woo Cho conducted field trials and analyzed the data. Chul Soo Park designed the experiments and supervised the project. Youngjun Mo analyzed the data, supervised the project, and wrote the final draft.

The authors declare that they have no conflicts of interest.