*2.3. Seed Dormancy QTL Detection in CSSL Population*

To determine the genetic regions controlling seed dormancy, the CSSL population was genotyped by the RICE6K SNP array. A total of 518 bins (defined as Bin1 to Bin518) across the whole genome were obtained [33]. QTL mapping was carried out by using ridge regression analysis with the 518 bins, and it identified 9, 19, 25, 23, 17, and 21 QTLs for Gmax, G3d, T50, U8416, AUC, and GI, respectively (Figure 2). Detailed information about the *p*-value, phenotypic variation explained, and the effect of the identified QTLs are shown in Table 2.

**Figure 2.** Circos plot illustrating the quantitative trait loci (QTLs) of six germination parameters. Chr: Size of the 12 chromosomes of *Oryza sativa*; Gmax, maximum germination percentage of seven days germination; G3d, germination percentage at three days; T50, time to reach 50% germination of the total number of germinated seeds; U8416, germination uniformity, which is time interval between 84% and 16% of viable seed to germinate; AUC; area under the germination curve until 168 h; GI, germination index. Blue indicates significant QTLs identified by all six germination parameters. Orange indicates significant QTLs identified in the corresponding germination parameter. Gray indicates non-significant QTLs.


U8416 the positive E value represents the ZS97 alleles that increased the effect; c. The genes near 500 kb of the most significantly associated SNP; d. - Indicates no QTLs detected.

In total, 36 QTLs were detected for seed dormancy using the six parameters in the CSSL population (Table 2). Among the six parameters, U8416 explained the highest phenotypic variance (77.9%) with 23 QTLs detected, while T50 explained 71.5% of phenotypic variance with the largest number of QTLs detected (25 QTLs). G3d, AUC, and GI explained 77%, 66.3%, and 52.7% of phenotypic variance with 19, 17, and 21 QTLs detected, respectively. Gmax only detected 9 QTLs and explained 50.5% of phenotypic variance. The 36 QTLs were distributed on each of the chromosome in which both chromosome 6 and 10 had the highest -log10(*p*) value (15.7) for U8416, and the phenotypic variance were 6.5% and 10.4%, respectively. Among all 36 QTLs, 10 QTLs were identified or cloned previously for seed dormancy, suggesting the consistency of our QTL analysis with others. The other remaining 26 QTLs may be new ones for seed dormancy, as they do not contain any QTLs for dormancy in rice that have been described before (https://archive.gramene.org/qtl/).

Five out of thirty-six QTLs were common QTLs detected in all six parameters and distributed on chromosomes 3, 6, and 10. One common QTL (*qDOM3.3*) on chromosome 3 was identified as a seed dormancy QTL in the "Asominori×IR24" CSSL population [35], and *qDOM10.3*, which was detected in all six parameters, contained a gene, namely *OsFbx352* [36], that plays a regulatory role in the regulation of glucose-induced suppression of seed germination by targeting ABA metabolism. The other three common QTLs (*qDOM3.1*, *qDOM6.2*, and *qDOM10.2*) have not been reported before. *qDOM3.1* was detected in almost the same region on the upper end of chromosome 3 by the six germination parameters, suggesting the robustness of this QTL in the present CSSL population. U8416 of *qDOM3.1* had the highest -log10(P) value (9.2) among the six parameters and explained 8.3% of phenotypic variance.

#### *2.4. Verification of qDOM3.1 for Seed Dormancy*

To validate and fine map *qDOM3.1*, one line, namely NQ96, in the CSSL population was selected (Figure S2, Supplementary Materials). It carries a NIP substitution segment encompassing *qDOM3.1* on top of chromosome 3 in the ZS97 genetic background, with another NIP substitution segment on chromosome 9. The germination behavior of NQ96 was significantly lower and slower than ZS97 (Table 3). This indicated that the introduced NIP segment contained the QTL of seed dormancy. To confirm the genetic effect of the *qDOM3.1* on seed dormancy, we generated an F2 segregating population comprising 338 individuals by crossing NQ96 with ZS97. The F2 population was genotyped using ten polymorphic markers in the *qDOM3.1* region and one polymorphic marker on the other introgressed segment on chromosome 9. There was no significant difference on seed dormancy for NIP and ZS97 allele on chromosome 9 with marker RM410, denoting that the introgressed segment on chromosome 9 had no effect on seed dormancy. Thus, NQ96 only contained *qDOM3.1*, which had a genetic effect on seed dormancy.


Germination behaviors that are significantly different from that of ZS97 are indicated by asterisks (\* *p* < 0.05, \*\* *p* < 0.01). - no data available.

To determine the genetic effect of *qDOM3.1*, we performed a genetic segregation analysis of seed dormancy using the *qDOM3.1*-derived F2 population. Frequency distribution of Gmax in the population indicated that *qDOM3.1* from ZS97 was dominant (Figure 3).

**Figure 3.** Frequency distribution of Gmax in F2 segregation population.

Afterward, based on the genotyping results of ten polymorphic markers distributed within the target region RM14238-RM14317 for the F2 individuals, *qDOM3.1* was detected in the interval RM14238-MP030012 (approximately 252 kb) with a logarithm of the odds (LOD) score peaked around MP03008 (Figure 4), which explained 69.9%, 75.4%, 73.2%, 71.4%, 38.3%, and 46.5% of the phenotypic variance in Gmax, G3d, AUC, GI, U8416, and T50, respectively.

**Figure 4.** Verification of the QTL effect in the F2 primary segregation population. QTL scans along chromosome 3 for the six indexes in the CSSL-derived F2 population from the cross of NQ96 and ZS97. Logarithm of odds profile of QTL region on chromosome 3 in the F2 population, showing a QTL (*qDOM3.1*) for Gmax, G3d, T50, U8416, AUC, and GI.
