Introduction
During the domestication of crops from their wild progenitors, several traits that are useful for cultivation have been selected: reduced grain shattering and awn length, increased panicle branching and increased spikelet number per panicle. Among the domesticated traits, seed dormancy causes plants to arrest their maturation status for variable periods of time, allowing them to survive various durations of exposure to hostile environmental conditions and to germinate under favourable conditions.
Dormancy is thought to prevent pre-harvest sprouting (PHS), whereas the early breakage of dormancy causes PHS, which reduces grain yields and grain quality in cereal crops (Bewley and Black, Reference Bewley and Black1982; Li et al., Reference Li, Ni, Francki, Hunter, Zhang, Schibeci, Li, Tarr, Wang and Cakir2004). In rice, PHS frequently occurs in Southeast Asia due to the long rainy periods in early summer and autumn (Wan et al., Reference Wan, Jiang, Tang, Wang, Hou, Jing and Zhang2006), with hybrid rice in South China showing higher than average PHS ratios (Guo et al., Reference Guo, Zhu, Xu, Zeng, Wu and Qian2004). Since PHS causes severe economic losses by reducing the production and quality of grain, improving PHS resistance has long been one of the main breeding targets for cereal crops worldwide (Zhang et al., Reference Zhang, Miao, Xia and He2014).
Seed dormancy can be caused by an embryo-outside effect or an embryo-imposed effect. Among the surroundings of embryo, the seed coat acts as a physical barrier against water imbibition and radical emergence. Seed pigmentation is associated with seed dormancy in cereal crops such as rice, wheat and proso millet (Khan et al., Reference Khan, Cavers, Kane and Thompson1997; Himi et al., Reference Himi, Mares, Yanagisawa and Noda2002; Gu et al., Reference Gu, Kianian and Foley2005a; Sweeney et al., Reference Sweeney, Thomson, Pfeil and McCouch2006). Embryo-imposed dormancy is developmentally regulated; the phytohormones abscisic acid (ABA) are required for elaborate developmental regulation and play key roles in seed dormancy acquisition (Cutler et al., Reference Cutler, Rodriguez, Finkelstein and Abrams2010; Raghavendra et al., Reference Raghavendra, Gonugunta, Christmann and Grill2010). To date, the major genes found to be associated with seed dormancy and germination are related to ABA biosynthesis, catabolism and signal transduction. AtABI3 of Arabidopsis, which is orthologous to ZmVP-1 in maize, OsVP1 in rice, and TaVp-1 in wheat, functions in the global regulation of seed maturation by participating in ABA signal transduction (Hattori et al., Reference Hattori, Terada and Hamasuna1994; Nakamura and Toyama, Reference Nakamura and Toyama2001; De Laethauwer et al., Reference De Laethauwer, Reheul, De Riek and Haesaert2012).
Among cereal crops, rice, which has a relatively small genome size of approximately 300 Mb, is widely used as a model plant for genomic studies. Using newly detected genes from rice is an excellent approach for finding functional genes in grass crops such as wheat (hexaploid) and barley via comparative genomics (He et al., Reference He, He, Zhang, Sun, Morris, Fuerst and Xia2007; Liu et al., Reference Liu, He, Appels and Xia2012). Zhang et al. (Reference Zhang, Miao, Xia and He2014) isolated TaSdr, a gene associated with seed dormancy in wheat, and developed a functional marker in wheat via a comparative study based on the rice Sdr4 gene, which is positively regulated by OsVP1 (Sugimoto et al., Reference Sugimoto, Takeuchi, Ebana, Miyao, Hirochika, Hara, Ishiyama, Kobayashi, Ban and Hattori2010).
To date, most seed dormancy and germination studies have been conducted using the limited available rice genetic resources. In this study, we surveyed naturally occurring variations in seed germination ability in diverse rice genetic resources. In addition, we surveyed the changes in ABA content during seed maturation to investigate their association with seed dormancy.
Materials and methods
Plant materials
In total, 205 rice accessions (Table 1 and Supplementary Table 1) were selected based on geographical origin and ecotype among the conserved accessions in RDA GeneBank at the National Agrobiodiversity Center, National Institute of Agricultural Science (NAC), Rural Development Administration (RDA), Republic of Korea. These accessions included 109 japonica, 79 indica and 17 Tongil rice (Kim et al., Reference Kim, Kim, Lee, Seo, Choi, Choi, Yang, Kim, Lee and Chin2014) accessions and were grown using conventional cultural practices at the experiment field of Natonal Agrobiodiversity Center, Jeonju, in 2015. Their ecotypes were confirmed by genotyping using six microsatellite markers (RM235, RM242, RM267, RM23, RM211 and RM270; data not shown), which exhibit distinct allele distribution patterns in indica, Tongil and japonica rice (Qi et al., Reference Qi, Zhang, Zhang, Wang, Sun, Ding, Wang and Li2009).
Table 1. Information about the rice genetic resources investigated in this study

a PHS: less than 20% sprouting at harvesting time.
b PHS: greater than 20% sprouting at harvesting time.
Characterization of seed germination ability
Seeds and panicles were harvested from five plants at 42 d after flowering (DAF) and the moisture content was adjusted by drying at 15°C (RH 10%) during 7 d. The susceptibility for PHS value was surveyed using three freshly harvested panicles, which were incubated at 25°C (RH 100%) for 7 d, after which the number of germinated seeds on each panicle was recorded and expressed as a percentage of the total grain number per panicle. The lower the PHS value, the lower is the susceptibility and the higher is the PHS resistance. Accessions with a PHS value below 20% were classified as ‘resistant’. Seed germination at harvesting time (GHT) was determined by threshing panicles and planting three replications of 50 seeds (fruits in hulls) onto moistened Whatman filter paper (10 ml of distilled water) in Petri dishes. They were incubated at 25°C (RH 100%) based on International Seed Testing Association (ISTA) guidelines and germinated seeds (radicle and coleoptile emerged from the hull) were counted daily for a period of 10 d. The cumulated number was expressed as per cent of seeds planted. Germination after an after-ripening process (GAR) was determined by storing dry seeds at 25°C for 3 weeks before germination was performed as described above. Germination after a dormancy breakage (GDB) was determined after a dry storage at 50°C for 7 d before germination was performed as described above. Germination index (GI) was calculated as described by Basra et al. (Reference Basra, Farooq, Tabassam and Ahmad2005):

where N i is the mean number of germinated seeds on the ith day and D i is the number of days after the beginning of the germination test.
Determination of changes in ABA content
To reduce the effects of environmental factors on phenotypic variation among the tested rice accessions, including differences in PHS and hormonal contents, 14 samples (eight japonica accessions and six indica accessions; Supplementary Table 1) with similar flowering times (early August) were selected from among the 205 rice genetic resources. The rice grain parts were sampled at the following stages: +0 DAF, +3 DAF, +6 DAF, +10 DAF, +16 DAF, +20 DAF and +42 DAF.
ABA extraction and purification were performed as described by Bollmark et al. (Reference Bollmark, Kubát and Eliasson1988) with some modifications. Samples including 10 dehulled grains of the 14 rice accessions at seven ripening stages (+0, +3, +6, +10, +16, +20 and +42) were ground in liquid nitrogen and combined with 80% (v/v) methanol extraction medium containing 1 mM butylated hydroxytoluene as an antioxidant. The extracts were incubated at 4°C for 4 h, and, after centrifugation (5000 g , 15 min, 4°C), the supernatants were filtered through pre-washed Chromosep C18 columns; pre-washing was conducted using 10 ml 100% (v/v), followed by 5 ml 80% (v/v) methanol. An approximately 5 ml purified fraction containing ABA, gas and other plant hormones was dried under N2 and dissolved in 2 ml of TBS (Tris - buffered saline) medium for analysis by enzyme-linked immunosorbent assay (ELISA). ABA concentrations in extracts of dehulled rice grains were determined by ELISA using a Phytodetek ABA Test Kit (Agdia, USA). The ELISA procedures were conducted according to the manufacturer's instructions with three replicate experiments.
Data analysis
The arcsine transformation of PHS values and germination (%) was used for statistical analysis. One-way analysis of variance (ANOVA), Duncan's multiple range test (DMRT) and correlation analysis were conducted using R software (ver. 3.2.3, http://www.r-project.org/); differences were considered significant when the P value was <0.05.
Results
PHS value in diverse rice accesssions
Among the 205 accessions, 90 were classified as PHS resistant, as their PHS value was below 20% (Table 1, Fig. 1). The proportion of accessions with PHS resistance was highest for Tongil-type accessions, with a value of 76.4% [13/17 accessions (accs)], followed by indica at 54.4% (43/79 accs) and japonica at 31.2% (34/109 accs). Focusing on geographical origin, accessions from South and Southeast Asia had relatively high proportion of accessions with PHS resistance of 59.1% (13/22 accs) and 78.7% (37/47 accs), respectively; while only 29% of accessions from East Asia (28/95 accs) and other regions (12/41 accs) were PHS resistant.

Fig. 1. Investigation of PHS. (a) Representative PHS-resistant and -susceptible accessions; (b) distribution of PHS among 205 rice genetic resources.
Focusing on domestication levels, breeding lines and cultivars showed a low proportion of PHS-resistant accessions of 40.0% (24/60; East Asia, 33.3%; South and Southeast Asia, 66.7%; other regions, 53.3%). On the other hand, weed types and landraces showed a high proportion of PHS-resistant accessions of 55.2% (16/29; East Asia, 50.0%; South and Southeast Asia, 60.0%; other regions, 0%) and 54.1% (20/37; East Asia, 36.4%; South and Southeast Asia, 76.2%; other regions, 0%), respectively (Supplementary Table 1) Among the 79 non-categorized accessions 38.0% (30/79) were PHS resistant.
Focusing on flowering time within the indica and Tongil types, the accessions that flower in early August had the highest mean PHS value (35.0 ± 33.9%), followed by accessions flowering at the end of July (28.6 ± 33.8%), middle of August (25.8 ± 30.4%) and end of August (11.3 ± 19.5%). By contrast, the japonica accessions did not show significant differences in PHS value at the P < 0.01 level (Table 2).
Table 2. Average PHS values at harvesting time (42 DAF)

*P < 0.01.
a and b were ranked by Duncan's test.
PHS values vs. seed GHT
Analysis of variance revealed differences among ecotypes regarding the difference between PHS values (overall mean 31.0 ± 28.8%) and the GHT (51.9 ± 35.6%) (P < 0.001) (Fig. 2); the japonica (P < 0.001) and indica (P < 0.01) accessions exhibited higher GHTs than PHS values, whereas the Tongil group did not. Across all accessions, there was a significant correlation between PHS values and GHT, with a Pearson correlation coefficient value of 0.814 (P < 0.001). The GI at 42 DAF was highly correlated with the GHT, with a correlation coefficient of 0.947 (P < 0.001).

Fig. 2. Comparison of germination ability. (a) Changes in seed germination (%); germination of grains in the panicle and germination after separation from the panicle (b) correlation between PHS, GI, and average germination ratio at harvesting time (7th day of incubation). #Grains in the panicle, ##Grains detached from the panicle. ***P < 0.001, **P < 0.01, ND, not significant.
Effects of after-ripening and dormancy breakage
The average GI values of the PHS-resistant and non-resistant accessions were 2.8 (±3.4) and 10 (±4.1), respectively. These values increased following after-ripening treatment (10.8 ± 4.6 and 14.6 ± 3.3 for the PHS-resistant group and the PHS non-resistant group, respectively) and further increased by dormancy breakage (18.4 ± 2.9 and 18.7 ± 2.6 for the PHS-resistant group and the PHS non-resistant group, respectively) (Fig. 3). In the PHS-resistant group, the seed germination ability increased by after-ripening treatment and dormancy breakage differently in the three ecotypes. The japonica accessions showed relatively low average germination abilities following after-ripening and dormancy breakage, with values of 69.5 (±31.2)% and 91.9 (±15.0)%, respectively, compared with indica and Tongil accessions, with values of 81.5 (±21.4)% and 97.4 (±4.4)%, respectively (P < 0.05; Fig. 3.).

Fig. 3. Changes in seed germination ability after separation from the panicle and dormancy breakage. (a) Changes in average GI values during 10 d of incubation (25°C, RH 100%) in 205 rice genetic resources at harvesting time (42 DAF) and dormancy breakage treatment; (b) changes in average germination ratios on the 7th day of incubation (25°C, RH 100%) at harvesting time and following after-ripening and dormancy breakage for 90 PHS-resistant (PHS <20%) rice genetic resources.
Changes in ABA content after fertilization
The ABA content increased until 10 DAF and decreased thereafter in all 14 accessions examined (Fig. 4). The maximum ABA content was detected during the ripening period (at 10 DAF). During this period, the indica accessions had higher average ABA contents (95.9 ± 34.2 mM) than the japonica accessions (48.7 ± 10.5 mM) (P < 0.01). Among the resistant accessions, the mean ABA content of the indica accessions was 106.0 (±26.1) mM, whereas that of the japonica accessions was only 41.8 (±7.4) mM. Interestingly, the PHS-non-resistant japonica accessions had higher or similar levels of ABA compared with the PHS-resistant accessions, with differences not significant at the P value of 0.05 level, and among the indica accessions, there was no significant difference in ABA content at 10 DAF between PHS-resistant and -non-resistant accessions.

Fig. 4. Changes in ABA content after fertilization in 14 rice accessions. (a) Eight japonica accessions; (b) six indica accessions.
Discussion
Climate change is expected to cause greater fluctuations in rainfall during the ripening periods and harvesting times of various cereal crops, leading to an increase in damage caused by PHS. Various studies, for example involving QTL mapping, gene identification, and the analysis of associated hormones, have been conducted to help overcome these problems in rice, a widely used model cereal crop plant (Hattori et al., Reference Hattori, Terada and Hamasuna1994; Lin et al., Reference Lin, Sasaki and Yano1998; Dong et al., Reference Dong, Tsuzuki, Kamiunten, Terao, Lin, Matsuo and Zheng2003; Guo et al., Reference Guo, Zhu, Xu, Zeng, Wu and Qian2004; Gu et al., Reference Gu, Kianian and Foley2005b; Wan et al., Reference Wan, Jiang, Tang, Wang, Hou, Jing and Zhang2006; Gu et al., Reference Gu, Liu, Feng, Suttle and Gibbons2010; Hori et al., Reference Hori, Sugimoto, Nonoue, Ono, Matsubara, Yamanouchi, Abe, Takeuchi and Yano2010; Sugimoto et al., Reference Sugimoto, Takeuchi, Ebana, Miyao, Hirochika, Hara, Ishiyama, Kobayashi, Ban and Hattori2010; Gu et al., Reference Gu, Foley, Horvath, Anderson, Feng, Zhang, Mowry, Ye, Suttle and Kadowaki2011; Kim et al., Reference Kim, Hwang, Hong, Lee, Ahn, Yoon, Yoo, Lee, Lee and Kim2011; Lu et al., Reference Lu, Xie, Yang, Wang, Liu, Zhang, Jiang and Wan2011; Marzougui et al., Reference Marzougui, Sugimoto, Yamanouchi, Shimono, Hoshino, Hori, Kobayashi, Ishiyama and Yano2012; Challam et al., Reference Challam, Kharshing, Yumnam, Rai and Tyagi2013). However, these investigations of valuable alleles were limited in that only a small number of resistant and susceptible accessions were used to uncover various aspects of seed germination and dormancy. Seed germination and dormancy are quantitative traits that retain broadly observed phenotypic variation, the degrees of which vary according to allele type. Therefore, these diverse alleles might serve as valuable resources for regulating seed germination levels, including PHS (Haussmann et al., Reference Haussmann, Parzies, Presterl, Susic and Miedaner2003). The aim of the present study was to survey naturally occurring variations in natural rice populations, so that the resources together with the data can be used in further studies for investigating seed dormancy and germination.
Among the 205 rice accessions examined (109 japonica, 79 indica and 17 Tongil), more PHS-resistant accessions (54.4%) were presented in the indica ecotype than the japonica type (31.2%). This finding was not unexpected, as most indica rice accessions are domesticated and cultivated in tropical and subtropical regions, which have a wet period during ripening and harvesting time. By contrast, japonica rice is dispersed throughout south China and has been domesticated in temperate regions, which have relatively dry climates (Khush, Reference Khush1997; Huang et al., Reference Huang, Kurata, Wei, Wang, Wang, Zhao, Zhao, Liu, Lu and Li2012). The selective pressure for PHS resistance is lower in East Asia, which has relatively few environmental constraints, compared with South and Southeast Asian regions. We found that breeding lines and cultivars from East Asia were more susceptible to PHS than those from other regions, with an average proportion of resistant accessions of 33.3%. This finding reflects the notion that selection by humans has placed a relatively low value on PHS resistance in this region compared with tropical and subtropical regions. As a result, unexpected rainfall during ripening and harvesting due to rapid environmental changes can severely damage grain quantity and quality.
Despite the correlation between the germination of grains in the panicle and of detached grains, we detected an increase in germination (%) after separation from the panicle at 42 DAF in both the japonica (P < 0.001) and indica (P < 0.01) accessions. Physical dormancy due to the presence of an impermeable seed coat is one cause of seed dormancy in some orthodox seeds, and seed pigmentation is associated with seed dormancy in cereal crops (Werker, Reference Werker1980; Khan et al., Reference Khan, Cavers, Kane and Thompson1997; Himi et al., Reference Himi, Mares, Yanagisawa and Noda2002; Gu et al., Reference Gu, Kianian and Foley2005a; Sweeney et al., Reference Sweeney, Thomson, Pfeil and McCouch2006). Kermode and Bewley (Reference Kermode and Bewley1989) reported that the seed detachment from the mother plant in combination with some water loss could elicit germination. In this regard, the increased seed germination after the separation from panicles might be due to the weakening of the mechanical barrier correlated with dehydration in the detachment site of grains. The germination ability of the PHS-resistant group (90 accessions, PHS < 20%) increased following after-ripening treatment, and remained dormancy was nearly absent after the high-temperature dormancy-breaking treatment (Fig. 3). In particular, there was a relatively high increase in germination ability in indica-type rice following after-ripening and dormancy breakage compared with japonica-type rice. This result indicates that the dormancy of indica rice is easily broken, which might be due to its geographically distinct domestication process (Cheng et al., Reference Cheng, Motohashi, Tsuchimoto, Fukuta, Ohtsubo and Ohtsubo2003; Vitte et al., Reference Vitte, Ishii, Lamy, Brar and Panaud2004; Zhu and Ge, Reference Zhu and Ge2005; Huang et al., Reference Huang, Kurata, Wei, Wang, Wang, Zhao, Zhao, Liu, Lu and Li2012).
ABA plays a central role in maintaining primary dormancy and germination throughout the plant's lifecycle and affects its seed germination ability and dormancy through active or passive transport (Finch-Savage and Leubner-Metzger, Reference Finch-Savage and Leubner-Metzger2006; Finkelstein et al., Reference Finkelstein, Reeves, Ariizumi and Steber2008; Seo and Koshiba, Reference Seo and Koshiba2011). Similar to previous results (Gu et al., Reference Gu, Liu, Feng, Suttle and Gibbons2010; Kanno et al., Reference Kanno, Jikumaru, Hanada, Nambara, Abrams, Kamiya and Seo2010), we found that the ABA content was maximal during the middle stage of development (10 DAF) and decreased thereafter in all 14 accessions examined. The indica accessions had higher ABA contents than the japonica accessions, which is thought to result from natural or human-driven selection to help the plants cope with a long wet period (Khush, Reference Khush1997). Interestingly, PHS-non-resistant accessions of both the japonica and indica type maintained high or similar levels of ABA compared with PHS-resistant accessions. In higher plants, ABA concentration and sensitivity to ABA determine the physiological responses mediated by this phytohormone (Seo and Koshiba, Reference Seo and Koshiba2011). ABA levels are regulated by its biosynthesis and catabolism, and ABA transport, perception and signal transduction are key factors affecting the sensitivity to this hormone. To date, several genes in the ABA biosynthesis and catabolism pathways have been discovered, including NCEDs for ABA biosynthesis during water stress responses (Iuchi et al., Reference Iuchi, Kobayashi, Taji, Naramoto, Seki, Kato, Tabata, Kakubari, Yamaguchi-Shinozaki and Shinozaki2001; Urano et al., Reference Urano, Maruyama, Ogata, Morishita, Takeda, Sakurai, Suzuki, Saito, Shibata and Kobayashi2009) and CYP707As for ABA catabolism (Kushiro et al., Reference Kushiro, Okamoto, Nakabayashi, Yamagishi, Kitamura, Asami, Hirai, Koshiba, Kamiya and Nambara2004; Umezawa et al., Reference Umezawa, Okamoto, Kushiro, Nambara, Oono, Seki, Kobayashi, Koshiba, Kamiya and Shinozaki2006). Furthermore, the discovery of ABA receptors such as OsPYL/RCAR5, the rice orthologue of an intracellular ABA receptor of Arabidopsis (Kim et al., Reference Kim, Hwang, Hong, Lee, Ahn, Yoon, Yoo, Lee, Lee and Kim2011), revealed that ABA is not simply transferred by passive transport (Seo and Koshiba, Reference Seo and Koshiba2011; Ye et al., Reference Ye, Jia and Zhang2012). While PHS of less matured seeds could be affected by ABA content in that we measured the PHS level at harvesting time (42 DAF), these findings suggest that the key factors for seed dormancy are the perception and signal transduction of ABA rather than the total content of ABA in a natural population. Indeed, Gianinetti and Vernieri (Reference Gianinetti and Vernieri2007) also reported that there is little correlation between ABA content and seed dormancy.
Based on the results of the present study, we are currently searching for genes associated with seed dormancy affecting PHS at the molecular level. The diversity of seed germination ability in rice genetic resources detected in this study could be sustainably used for crop improvement programmes and for detecting the genetic and physiological basis of this quantitative trait.
Supplementary Material
The supplementary material for this article can be found at http://dx.doi.org/10.1017/S1479262116000319
Acknowledgements
This work was supported by a grant from Next-Generation BioGreen 21 Program (Plant Molecular Breeding Center, No. PJ01102401) and was carried out with the support of ‘Research Program for Agricultural Science & Technology Development (Project No. PJ010883)’, National Institute of Agricultural Sciences, Rural Development Administration, Republic of Korea.