**1. Introduction**

Seed dormancy is an important evolutionary trait. It can optimize the distribution and timing of germination over time in nature [1]. Seed dormancy also plays an important role in agricultural production. Extremely strong dormancy leads to a low germination rate in the field, the irregular emergence of seedlings and an impact on sowing time, and can even affect the final yield. On the contrary, too weak dormancy leads to pre-harvest sprouting (PHS), especially in the high temperature and rainy environment during seed maturity. The economic loss caused by pre-harvest sprouting has become an important factor restricting the yield of cereal crops (such as rice, wheat, maize, etc.) [2], and seriously affects the sowing quality and processing quality of crops, and even causes the change of storage quality [3]. Therefore, crop seeds require a well-balanced level of dormancy to ensure a high rate of germination and to control pre-harvest sprouting in the field.

Seed dormancy is a highly complex trait and largely influenced by genetic and environmental factors [4]. Recent progress in plant genomics and various genetic populations has facilitated the identification of quantitative trait loci (QTLs) for seed dormancy in many species, for example, in *Arabidopsis* [5], *Lepidium sativum* [6], oilseed rape [7], sorghum [8], barley [9], and wheat [10–12].

By using different genetic populations constructed from cultivated rice, wild rice, and weedy rice, more than 160 QTLs have been identified that affect the germination or dormancy of rice (available online: https://archive.gramene.org/qtl/). For example, five seed dormancy QTLs were detected by BC1 and F2 populations constructed from rice variety N22 with strong dormancy and two weak dormant varieties [13]. Three seed dormancy QTLs were mapped using the chromosomal segment substitution line (CSSL) population and their derived F2 populations constructed by the strong dormant variety Nona Bokra and the weak dormant variety Koshihikari [14]. By using a population of recombinant inbred lines (RILs), nine seed dormancy QTLs were identified in three developmental stages [15]. In addition, besides biparental genetic populations, genome-wide association analysis across natural accession also revealed genetic variation of seed dormancy among rice natural populations, and the analysis can be used to identify new candidate genes related to seed dormancy [16,17].

In 2010, Japanese scientists isolated the first rice seed dormancy gene *Seed dormancy 4* (*Sdr4*) by map-based cloning [18]. *Sdr4* can be positively regulated by the seed maturation-related gene *OsVP1* and two *Arabidopsis* dormancy gene *Delay of Germination 1 (DOG1)* [19] homologous genes, thereby enhancing seed dormancy. After that, several rice seed dormancy genes were identified and proved to be involved in the hormonal regulation of seed dormancy. By rice mutant screening, *PHS8* was isolated to be a starch debranching enzyme named isoamylase1, and it determined seed dormancy and germination by affecting abscisic acid (ABA) signaling [20]. The rice *GERMINATION DEFECTIVE 1* regulated seed germination by integrating gibberellin acid (GA) and carbohydrate metabolism [21]. A weedy red rice dormancy QTL (*SD7-1*/*Rc*) was identified as a basic helix-loop-helix transcription factor that controls ABA synthesis, influencing red pericarp color and seed dormancy [22]. By map-based cloning, a gibberellin synthesis gene *OsGA20ox2* was identified within QTL *Seed Dormancy1-2* (*qSD1-2*) [23]. *OsGA20ox2* (the green revolutionary gene *SD1*) is involved in the biosynthesis of GA, regulating the development of endosperm-imposed dormancy in rice.

Hormonal regulation may be a highly conserved mechanism of seed dormancy among many species. The balance of ABA and GA or other hormones plays crucial roles in the regulation of seed dormancy and germination [24,25]. ABA is an essential positive regulator of both dormancy induction during seed maturation and maintenance of the dormant state after imbibition [26–28]. In *Arabidopsis*, *DOG1* was the first cloned dormancy QTL, and encoded a protein with unknown functional domain [19] and had conserved function throughout many species. In recent years, *DOG*1 was reported to play a regulatory role in ABA signaling. It encodes for a plant-specific protein that enhances ABA signaling through its binding to protein phosphatase 2C (PP2C) ABA HYPERSENSITIVE GERMINATION1 (AHG1) and AHG3 [29,30]. In addition, *DOG1* may mediate a conserved seed coat dormancy mechanism in the temperature- and GA-dependent pathways [31]. Besides *DOG1*, the previously mentioned seed dormancy genes, such as *PHS8* and *SD7-1*/*Rc*, were involved in either ABA metabolism or signaling pathway. Thus, it is worthwhile to investigate whether there are more genes involved in the hormonal regulation of seed dormancy.

Here, we presented the identification of QTLs for seed dormancy in a set of genome-wide single nucleotide polymorphism (SNP) genotyped chromosomal segment substitution lines (CSSLs) by backcrossing and marker-assisted selection, in which *japonica* Nipponbare (NIP) was the donor parent and the recurrent parent was *indica* Zhenshan 97 (ZS97). The CSSL population, which comprised 146 lines, was developed and genotyped in a previous study [32,33] and has not been used for a seed dormancy study.

Therefore, the objectives of the present study were to dissect the genetic base of seed dormancy and germination performances in the CSSL population, and to fine map the major QTLs using the CSSL-derived population. Moreover, we investigated how the candidate gene was involved in the ABA regulation of seed dormancy.
