Rice is one of the most important cereal crops and staple foods in Asia. According to its grain shape, rice is primarily classified into long, medium, and short categories, by which the ratios of grain length (GL) to grain width (GW) are more than 3.0, between 2.1 and 2.9, and smaller than 2.0, respectively http://www.ams.usda.gov/AMSv1.0/getfile?dDocName=STELDEV3003761. People from North and South America, Southern China, and Europe usually prefer rice with long and slender grains, whereas those from Japan, Northern China, and North and South Korea prefer rice with short and round grains [1, 2]. Juliano and Villareal reviewed grain quality of world rice types and reported that the correlations between GW and cooked rice hardness were significant in 18 countries/locations . In addition, Yang et al. and Luo et al. also mentioned that cooking quality is associated with grain shape [4, 5]. However, difference in cooking quality among the three shape categories is mainly determined by the chemical components and texture of their grains. In most cases, long grain rice has a high grain amylose content and after cooking, it is often firm and fluffy (not sticky); medium grain rice has a low amylose content and after cooking, it is often soft, moist, and sticky in texture. The cooking quality and amylose content in short grain rice are similar to those of rice in the medium grain category.
Grain size is usually evaluated by 1000-grain weight (TGW), which is one of the three key components, along with grain number per panicle (GPP) and panicle number per plant, contributing to rice grain yield and is positively correlated with grain shape traits including GL, GW, and grain thickness (GT) [6, 7]. Grain shape has attracted significant attention in rice breeding programs due to its contributions to rice yield and quality .
Grain shape has been widely accepted as a complex trait controlled by multiple genes with small effects. Extensive efforts to determine the genetic basis of grain shape have used forward and reverse genetic strategies. Initially, studies focused on characterizing mutants and the expression of major genes associated with grain size, for example, the Lk-f gene, which confers long kernel size , or Mi, which confers short kernel size . However, finding mutants of most grain shape genes in nature is not easy. Alternatively, quantitative trait locus (QTL) analysis based on genome wide mapping is a good strategy that has been widely used for mapping rice grain shape genes during the last 20 years. Previous studies used different populations, such as F2, F3, backcross, double-haploids, and recombinant inbred lines (RILs) http://www.gramene.org/qtl. Among these, the RIL population has particular advantages including repeatability, which favors the genetic analysis of quantitative traits because experiments can be replicated over years and locations. Combined with several fine genetic maps developed in recent years, many QTLs for grain shape have been identified. Tan et al., using F2:3 and RIL populations, detected two major QTLs for GL and GW on chromosomes 3 and 5, respectively. Several minor QTLs were also detected . The two major QTLs for grain shape were also commonly detected in the same intervals on chromosomes 3 and 5, respectively, across different populations [7, 11–15].
To date, thousands of QTLs have been detected in primary rice mapping populations. For fine mapping and cloning these QTLs, it is vital to make a population in which the targeted QTLs behave as the characters of a single mendelian factor. Therefore, an advanced population such as a near isogenic line population (NIL), which minimizes the noise of genetic background, is a prerequisite for QTL map-based cloning. Recently, increasing numbers of NILs have been used for QTL fine mapping and cloning. In terms of grain shape, Li et al. reported fine mapping of gw3.1 in the pericentromeric region of chromosome 3 using NILs, which is derived from a tropical japonica cultivar as the recurrent parent backcrossed to a wild rice . Wan et al. found QTL qGL-3 for GL and qGW-5 for GW using chromosomal segment substitution lines (CSSLs) derived from the cross Asominori × IR24 . Interestingly, these two QTLs were repeatedly identified in eight different environments. Fan et al. finely mapped the GS3 locus, a major QTL for GL and GW, and determined its candidate gene using advanced populations . Based on these related reports, Zhang suggested that GL is mostly controlled by the GS3 locus on chromosome 3 and GW is largely conditioned by GS5 on chromosome 5 . So far, one GL QTL, GS3 (encoding a transmembrane protein), two GW QTLs, GW2 (encoding RING-type E3 ubiquitin ligase) and qSW5/GW5 (whose biochemical function remains unclear), have been cloned [17, 19–21]. In addition, Xie et al. finely mapped two TGW QTLs, gw8.1 and gw9.1, and concluded that there were significant correlations between TGW and grain shape [22, 23].
Consecutive backcrossing is the conventional strategy for developing NILs and has been extensively used in mapping QTLs. Although molecular marker-assisted selection (MAS) can aim to any genome region, advanced backcross QTL-NIL development is both laborious and time consuming. An efficient method for a rapid NIL development is to seek the heterogeneous inbred family (HIF) at a QTL region . The basic work flow for HIF-based NIL (HIF-NIL) development is first to screen the RIL heterozygous in the target QTL (high generation, eg, F5 or F6), second to evaluate the variation of target trait phenotype, and finally, to get the NILs (seeds harvested from the heterozygotes at the targeted genome region). Therefore, HIF-NIL development is much more efficient and more rapid for either major or minor QTLs as compared to the conventional strategies.
In this study, an RIL population derived from a cross between two rice cultivars, Nanyangzhan and Chuan7, was used for QTL mapping for grain size and grain shape. An NIL-F2 population of minor QTLs was constructed. The objectives of this study were (1) to detect QTLs for GL, GW, and GT, (2) to evaluate the power of QTL detection for grain shape as compared with other reports, and (3) to fine map a minor QTL, qGL7, with pleiotropic effects on GL, TGW, and spikelets per panicle (SPP).