多环境下水稻DH群体剑叶长度的QTL分析

种植由籼稻品种和粳稻品种杂交衍生的DH群体,连续4年测定剑叶长度,运用基于混合模型的复合区间作图法,定位其QTL及上位性互作,估算遗传主效应和环境互作效应。结果表明,全部18个QTL都参与了上位性的形成,其中3个没有自身的遗传效应,但参与了3对上位性互作,这是传统方法不能发现的。另外,一个QTL可与多个QTL发生互作,这可能预示着存在更高阶互作。QTL与上位性互作可以具有不受环境影响而稳定表达的效应,以及与环境的互作效应。有些QTL与环境的互作效应可以在多环境下被检测到,但却不具有主效应,这种QTL可能易受环境因子的影响。QTL与环境的互作效应为随机效应,一个QTL或一对上位性与环境的互作效应总和理论上应等于零,否则会影响对遗传效应的估算,因此多环境下估算的遗传效应更可靠。     

大豆籽粒硬实加性和上位性QTL定位

硬实是植物种子的普遍特性, 是影响大豆种子发芽率、生存能力及储存期的重要数量性状, 同时影响着大豆的加工品质。本实验通过对大豆籽粒硬实性状的加性和上位性互作QTL (quantitative trait locus)分析, 明确控制大豆籽粒硬实的重要位点及效应, 旨在为进一步解析硬实性状复杂的遗传机制提供理论依据。以冀豆12和地方品种黑豆(ZDD03651)杂交构建的包含186个家系的F_6:8和F_6:9重组自交系群体为材料, 采用WinQTL Cartographer V. 2.5的复合区间作图法(composite interval mapping, CIM)定位不同年份的籽粒硬实性状相关的加性QTL, 同时采用IciMapping 4.1软件中的完备区间作图法(inclusive composite interval mapping, ICIM)检测籽粒硬实性状的加性及上位性QTL。共检测到3个籽粒硬实性状相关的加性QTL, 分别位于第2、第6和第14染色体, 遗传贡献率范围为5.54%~12.94%。同时检测到4对上位性互作QTL, 分别位于第2、第6、第9、第12和第14染色体, 可解释的表型变异率为2.53%~3.47%。同时检测到籽粒硬实性状加性及上位性互作QTL, 且上位性互作多发生在主效QTL间或主效QTL与非主效QTL间, 表明上位性互作效应在大豆籽粒硬实性状的遗传基础中具有重要的作用。     

水稻幼苗耐Al3+胁迫的QTL定位分析

用珍汕97B/密阳46构建RIL群体及其遗传图谱,对其种子采用纸培法育苗和培养,并设2个Al³⁺浓度（20 mg/L和30 mg/L）胁迫处理,以处理20 d后的幼苗相对根长（%）和相对苗高（%）为耐Al3+胁迫指标,用于QTL定位分析。结果表明,以相对根长为指标,检测到2个耐Al³⁺胁迫的主效应QTL,即qRAC(r)2和qRAC(r)11,其中qRAC(r)2在2个胁迫处理下均被检测到,有效基因来自于珍汕97B,贡献率较大（12.92%和16.15%）,表现出较强的耐Al³⁺胁迫功能。以相对苗高为指标,还检测到qRAC(s)11-2（20 mg/L Al³⁺）和qRAC(s)11-1（30 mg/L Al³⁺）2个主效应QTL,它们均位于第11染色体。耐Al³⁺胁迫的QTL上位性分析还表明,总的QTL上位性互作效应比主效应QTL的作用更大,且显示出相当的复杂性,在不同胁迫浓度下,基因间可以通过不同的互作方式,表现出对高Al³⁺胁迫的耐性。以相对根长为指标,检测到8对上位性互作,涉及1、2、3、5、6、10等6条染色体的15个QTL位点;以相对苗高为指标,共检测到6对上位性互作,涉及第1、2、3、5、6、7、8、9、12等9条染色体;且几乎所有互作均发生在背景因子的QTL位点间。     

A comparison of grain protein content QTLs and flour protein content QTLs across environments in cultivated wheat

The protein content of cultivated wheat (Triticum aestivum L.) is an important determinant factor of the nutritional value of the grain and the technological properties and rheological properties of flour. In order to examine the genetic basis of protein content, we searched for grain protein content quantitative trait loci (QTLs) and flour protein content QTLs in a newly developed doubled haploid (DH) line and identified the genetic correlation between grain protein content and flour protein content in the same DH population. Both the DH population and its parental lines were evaluated for grain protein content and flour protein content in three field trials. Four additive effect QTLs, two pairs of epistatic QTLs, and two QTLs × environment (QE) interaction for grain protein content were identified. The model explained 51.52% of the phenotypic variation (PVE), with epistatic effects being better explained by the higher PVE than additive effects. Four additive effect QTLs, five pairs of epistatic QTLs, and one QE were detected for flour protein content. The model explained 45.8% of the PVE. Of the 15 QTLs identified, three additive QTLs and one pair of epistatic QTLs were determined for both grain protein content and flour protein content; of these, the QTLs for protein content were considered to be more 'stable' than those detected for only grain protein content or for only flour protein content. The data reported here may be useful for manipulating the QTLs for protein content by marker-assisted selection in future wheat breeding programs.     

大豆叶片性状和叶绿素含量QTL间的上位性和环境互作效应

以丰产性好、抗旱力强的栽培大豆晋豆23为母本,山西农家品种半野生大豆灰布支黑豆为父本杂交衍生的447个RIL作为供试群体。将亲本及447个家系分别于2011、2012和2013年采用随机试验种植,按照标准测量叶长、叶宽和叶柄长3个性状,并于2012年8月1日和8月8日和2013年8月2日和8月9日各测量1次叶绿素含量。采用QTLNETwork 2.0混合线性模型分析方法和主基因+多基因混合遗传分离分析法,对大豆叶片性状和叶绿素含量进行遗传分析和QTL间的上位性和环境互作效应研究。结果表明,叶长受2对加性-加性×加性上位性混合主基因控制,叶宽受3对等效主基因控制,叶柄长受4对加性-加性×加性上位性主基因控制,叶绿素含量受4对加性主基因控制;检测到10个与叶长、叶宽、叶柄长和叶绿素含量相关的QTL,分别位于A1、A2、C2、H_1、L和O染色体。其中2个叶长QTL分别位于C2和L染色体,是2对加性×加性上位互作效应及环境互作效应QTL;3个叶宽加性与环境互作QTL分别位于A2、C2和O染色体;2个叶柄长QTL分别位于L和O染色体;3个叶绿素含量QTL分别位于A1、C2和H_1染色体。叶片性状和叶绿素含量的遗传机制较复杂,加性效应、加性×加性上位互作效应及环境互作效应是大豆叶片性状和叶绿素含量的重要遗传基础。建议大豆分子标记辅助育种中,一方面要考虑起主要作用的QTL,另一方面要注重上位性QTL的影响,这对于性状的遗传和稳定表达具有积极的意义。     

小麦DH群体穗下节间直径、茎壁厚及茎壁面积的QTL定位

由小麦品种花培3号和豫麦57杂交获得168个DH株系,连续两年在山东泰安种植,利用324个SSR标记构建遗传连锁图谱,并基于混合线性模型对控制穗下节间直径、茎壁厚及茎壁面积的QTL遗传效应和环境互作效应进行分析。共检测到10个加性效应位点和6对上位效应位点,其中3个加性位点参与环境互作效应。检测到位于染色体2D、3D和5D(2个)控制穗下节间直径的4个加性QTLs,与控制茎壁厚的3个加性位点相同或相邻,表现出一因多效或紧密连锁效应。两个位于染色体2D和5D控制茎壁厚和茎壁面积QTL有较大遗传贡献率,分别为11.37%和10.98%,适用于分子标记辅助育种和聚合育种。6对上位性效应遗传贡献率较小、无环境互作效应。     

不同磷水平下大麦分蘖期磷效率相关性状QTL定位分析

磷素营养与大麦品质及产量密切相关,磷高效遗传机制和品种改良是近年的研究热点之一。本研究利用由大麦栽培品种Baudin和种质材料CN4079杂交构建的重组自交系(RIL)群体,低磷胁迫(0.02 mmol L~(-1) KH_2PO_4)与正常供磷(0.2 mmol L~(-1) KH_2PO_4)条件下,对地上部和地下部磷素利用效率、磷素吸收效率和干重,以及分蘖数相关的QTL定位,并预测相关位点基因。表型鉴定结果表明,各性状在RIL群体中表现连续变异,并存在超亲分离。两种磷水平下,共检测到16个QTL,分布在2H、3H和5H染色体上,表型贡献率14.1%~28.5%。3H染色体上含有3个磷素利用效率位点,其增效等位基因均来源于Baudin,其中Qspue.sau-3H.1和Qrpue.sau-3H与控制磷素吸收效率的Qspae.sau-3H和Qrpae.sau-3H处于同一区段,而Qspue.sau-3H.2与控制分蘖数的位点Qtn.sau-3H处于同一区段。5H染色体上含有3个磷素吸收效率位点,其中Qspae.sau-5H.2和Qrpae.sau-5H的增效等位基因来自CN4079,且与控制磷素利用效率的Qspue.sau-5H和Qrpue.sau-5H,以及控制干重的Qsdw.sau-5H和Qrdw.sau-5H处于同一区段。在磷效率相关的4个区段中,除Qspue.sau-3H.1所处区间仅含有磷酸代谢与磷脂代谢相关基因外,其他区间均包含磷酸盐转运蛋白基因、磷酸代谢与磷脂代谢相关基因。     

多种环境下大豆单株粒重QTL的定位与互作分析

定位大豆单株粒重QTL、分析QTL间的上位效应及QTL与环境互作效应, 有利于大豆单株粒重遗传机理的深入研究。利用147个F_2:14~F_2:18 RIL群体, 5年2点多环境下以CIM和MIM方法同时定位大豆单株粒重QTL, 检测到17个控制单株粒重的QTL, 分别位于D1a、B1、B2、C2、F、G和A1连锁群上, 贡献率为6.0%~47.9%;用2种方法同时检测到3个QTL, 即qSWPP-DIa-3、qSWPP-F-1和qSWPP-D1a-5, 贡献率为6.3%~38.3%;2年以上同时检测到4个QTL, 即qSWPP-DIa-1、qSWPP-DIa-2、qSWPP-B1-1和qSWPP-G-1, 贡献率为8.1%~47.9%;利用QTLMapper分析QE互作效应和QTL间上位效应, 7种环境下的数据联合分析得到1个QE互作QTL和4对上位效应QTL, 贡献率和加性效应都较小。在分子标记辅助育种中应该同时考虑主效QTL及各微效QTL之间的互作。     

Genetic analysis and QTL mapping of oil content and seed index using two recombinant inbred lines and two backcross populations in Upland cotton

Cottonseed is one of the main by‐products of cotton. To explore the genetic composition of oil content (OC) and seed index (SI) is helpful for utilizing the cottonseed. Under multiple environmental conditions, the genetic structures of OC and SI were explored using two recombinant inbred lines (RILs) and corresponding backcross (BC) populations in Upland cotton. Twenty‐four and 31 quantitative trait loci (QTLs) for OC and SI, respectively, were detected using composite interval mapping, in which 9 QTLs for OC and 18 QTLs controlling SI were simultaneously identified in more than two environments or two populations. Forty‐seven and 37 QTLs with main effects (M‐QTLs) for OC and SI and 114 and 74 QTLs involved in digenic interactions (E‐QTLs), respectively, identified by inclusive composite interval mapping. On average, the E‐QTLs explained a larger portion of the phenotypic variation than the M‐QTLs did. It was concluded that additive effects of single‐locus and epistasis derived from complementary loci with few detectable single‐locus effects played an important role in oil content and seed index in Upland cotton.     

Comparison and analysis of main effects,epistatic effects,and QTL × environment interactions of QTLs for agronomic traits using DH and RILs populations in rice

Two genetic linkage maps based on doubled haploid (DH) and recombinant inbred lines (RILs) populations, derived from the same indica-japonica cross ‘Samgang × Nagdong’, were constructed to analyze the quantitative trait loci (QTLs) affecting agronomic traits in rice. The segregations of agronomic traits in RILs population showed larger variations than those in DH population. A total of 10 and 12 QTLs were identified on six chromosomes using DH population and seven chromosomes using RILs population, respectively. Three stable QTLs including pl9.1, ph1.1, and gwp11.1 were detected through different years. The percentages of phenotypic variation explained by individual QTLs ranged from 8 to 18% in the DH population and 9 to 33% in the RILs population. Twenty-three epistatic QTLs were identified in the DH population, while 21 epistatic QTLs were detected in the RILs population. Epistatic interactions played an important role in controlling the agronomic traits genetically. Four significant main-effect QTLs were involved in the digenic interactions. Significant interactions between QTLs and environments (QE) were identified in two populations. The QTLs affecting grain weight per panicle (GWP) were more sensitive to the environmental changes. The comparison and QTLs analysis between two populations across different years should help rice breeders to comprehend the genetic mechanisms of quantitative traits and improve breeding programs in marker-assisted selection (MAS).     

Population-specific QTLs and their different epistatic interactions for pod dehiscence in soybean [<Emphasis Type="Italic">Glycine max</Emphasis> (L.) Merr.]

Pod dehiscence (PD) prior to harvest results in serious yield loss in soybean. Two linkage maps with simple sequence repeat (SSR) markers were independently constructed using recombinant inbred lines (RILs) developed from Keunolkong (pod-dehiscent) × Sinpaldalkong (pod-indehiscent) and Keunolkong × Iksan 10 (pod-indehiscent). These soybean RIL populations were used to identify quantitative trait loci (QTLs) conditioning resistance to PD. While a single major QTL on linkage group (LG) J explained 46% of phenotypic variation in PD in the Keunolkong × Sinpaldalkong population with four minor QTLs, three minor QTLs were identified in the Keunolkong × Iksan 10 population. Although these two populations share the pod dehiscent parent, no common QTL has been identified. In addition, epistatic interactions among three marker loci partially explained phenotypic variation in PD in both populations. The result of this study indicates that different breeding strategies will be required for PD depending on genetic background.     

大豆籽粒维生素E含量的QTL分析

维生素E(VE)具有提高人体免疫力、抗癌、预防心血管疾病等保健作用,从大豆中提取的VE安全性更高。本研究采用高效液相色谱技术(HPLC)检测大豆BIEX群体(Essex×ZDD2315)维生素E的α-生育酚、γ-生育酚和δ-生育酚含量。应用QTLNetwork 2.1软件分别检测到8个和12对控制大豆维生素E及组分含量的加性和互作QTL。α-生育酚含量加性和互作QTL累计贡献值分别为8.68%(2个)和15.57%(4对),γ-生育酚含量加性和互作QTL累计贡献值分别为8.59%(2个)和11.57%(2对),δ-生育酚含量加性和互作QTL累计贡献值分别为5.44%(1个)和17.61%(3对),维生素E总含量的加性和互作QTL累计贡献值分别为11.39%(3个)和9.48%(3对)。未检测到维生素E及组分含量和环境互作的QTL。未定位到的微效QTL累计贡献值为66.16%~75.32%,说明未定位到的微效基因的变异占2/3以上。各性状的遗传构成中,未检测出的微效QTL份额最大,加性QTL和互作QTL贡献相差不大。在育种中应考虑常规方法聚合微效QTL与标记辅助方法聚合主要QTL相结合。     

QTL identification for molecular breeding of fibre yield and fibre quality traits in jute

In jute (Corchorus olitorius), quantitative trait loci (QTL) analysis was conducted to study the genetics of eight fibre yield traits and two fibre quality traits. For this purpose, we used a mapping population consisting of 120 recombinant inbred lines (RILs) and also used a linkage map consisting of 36 SSR markers that was developed by us earlier (Das et al. 2011). The RIL population was derived from the cross JRO 524 (coarse fibre) × PPO4 (fine fibre) following single seed descent. Using single-locus analysis involving composite interval mapping, a total of 21 QTLs were identified for eight fibre yield traits whereas for fibre quality (fibre fineness), only one QTL was detected. The QTL for fibre fineness explained 8.31–10.56% of the phenotypic variation and was detected in two out of three environments. Using two-locus analysis involving QTLNetwork, as many as 11 M-QTLs were identified for seven fibre yield traits (excluding top diameter) and one M-QTL was identified for fibre fineness which accounted for 4.57% of the phenotypic variation. For six fibre yield traits, we detected 16 E-QTLs involved in nine QQ epistatic interactions. For fibre fineness, four E-QTLs involved in two QQ epistatic interactions and for fibre strength, six E-QTLs involved in three QQ epistatic interactions were identified. Eight out of the 11 M-QTLs observed for the fibre yield traits were also involved in QE interactions; for fibre fineness and fibre strength, no QE interactions were observed.     

Transgressive segregation due to linked QTLs for grain characteristics of rice

Summary It has been theoretically proposed that multiple linked quantitative trait loci (QTLs) play a role in the accumulation of hidden variation within and between populations. In this study, the genetic bases for grain characteristics were examined by comparing two accessions representing the two rice subspecies by QTL analysis. Grain dimensions are known to be quantitative traits and to be diagnostic between these two subspecies. To enhance the power to detect QTL with small effects, after transferring a segment of chromosome 6 from an Indica type into a Japonica type of rice by repeated backcrosses, the introgressed segment was dissected by making recombinant inbred lines (RILs) which were expected to have different sizes of the introgressed segment in the same genetic background. The resulting RILs showed distinct transgression of the grain characteristics examined. Multiple QTLs controlling each of the length and breadth of seeds were detected on the introgressed segment, and showed positive and negative additive effects as well as epistatic interactions. The present study confirmed that transgressive segregation resulted from a breakdown of linkage and that the detection of QTLs was highly dependent upon the genetic effects of the neighboring QTLs, indicating the need for caution in interpreting QTL effects.     

Mapping quantitative trait loci conferring blast resistance in upland indica rice (<Emphasis Type="Italic">Oryza sativa</Emphasis> L.)

A genetic analysis of blast resistance in upland rice variety is very crucial. In this study, we performed a linkage mapping of quantitative trait loci (QTLs) for blast resistance using an advanced backcross population from a cross between Way Rarem (susceptible indica variety) and Oryzica Llanos 5 (durable resistant indica variety). A transgressive segregation was observed in the advanced backcross population of Way Rarem//Oryzica Llanos 5. A total of 16 QTLs have been identified along chromosomes 1, 3, 5, 6, 7, 8, 9, and 11 against eight blast pathogen isolates. Each QTL accounted from 11.31 to 45.11% of the variation in blast resistance. Most QTLs showed race specificity, demonstrating the small effect of such QTLs. Unexpectedly, several superior blast resistance alleles were contributed by Way Rarem, the susceptible-recurrent parent. Among eight candidate defense response genes detected in several loci, a single gene (oxalate oxidase) present on chromosome 3 was found to be associated with blast resistance in upland indica rice. Ultimately, these advanced backcross lines with resistance to blast tagged by markers might be useful for pyramiding blast resistance alleles in upland rice.     

控制水稻株高的QTL定位及环境互作分析

为水稻的基因水平研究提供了重要的平台,利用由小穗小粒型品种‘密阳46’和大穗大粒型品种FJCD建立的一个包含130个家系F10的重组自交系群体,分别在武夷山和莆田环境下测定其株高,进行QTL定位及环境互作分析。武夷山环境下检测到2个加性QTL,位于1、2号染色体上,其中主效qPH-2-5解释了26.2%的表型变异;莆田环境下检测到4个加性QTL,分别位于2、2、2、6号染色体上,共解释了20.61%的表型变异。经过GE互作分析,2个背景QTL存在显著的加性×环境互作效应,共解释了17.36%的表型变异。此研究从一定程度上揭示了株高的数量遗传规律,同时为株高分子育种提供理论依据。     

利用高密度遗传图谱发掘水稻抽穗期新位点

水稻抽穗期是决定水稻种植地区及其季节适应性的关键因素,发掘控制水稻抽穗期相关的新主效QTL至关重要。利用包含527个bin标记的高密度遗传连锁图谱,通过靶向测序基因型检测技术对水稻“空育131/小白粳子”衍生的RIL群体进行抽穗期基因型分析。通过对双亲和RIL群体的基本统计分析发现,双亲抽穗期呈极显著差异,表型处于RIL群体范围内,RIL群体有明显的超亲分离现象,符合正态分布。利用IciMapping 4.2软件的完备区间作图法,在水稻第1、3和7号染色体上共检测到4个QTL,其中3个QTL区间内分别含有与抽穗期相关的已知基因OsGI、Hd6和Ghd7,而qHD-3-1是控制水稻抽穗期的新位点。     

利用BC2F2高代回交群体定位水稻籽粒大小和形状QTL

以我国优良籼稻恢复系蜀恢527为轮回亲本, 以来自菲律宾的Milagrosa为供体亲本, 培育了样本容量为199株的BC₂F₂高代回交群体。选取85个均匀分布在12条染色体上的多态性SSR标记进行基因型分析, 同时对粒长、粒宽、长宽比和千粒重4种性状进行了表型鉴定。采用性状－标记间的单向和双向方差分析对上述性状进行了QTL定位。单向方差分析(P<0.01)共检测到了10个控制粒长、粒宽、长宽比和千粒重的QTL, 其中有3个具有多效性。由于粒长和长宽比的高度相关性, 控制长宽比的2个QTL均能在粒长QTL中检测到。位于第3染色体着丝粒区域的qgl3b是一个控制粒长、长宽比和千粒重的主效QTL, 它可以分别解释粒长、长宽比和千粒重表型变异的29.37%、26.15%和17.15%。该QTL对于粒长、长宽比和千粒重均表现较大的加性效应(来自蜀恢527的等位基因为增效)和负向超显性。位于第8染色体的qgw8位点是一个控制粒宽的主效QTL, 同时也是控制千粒重的微效QTL, 能解释粒宽表型变异的21.47%和千粒重表型变异的5.16%。该QTL对粒宽和千粒重均具有较大的加性效应(来自蜀恢527的等位基因为增效)和正向部分显性。双向方差分析(P<0.005)共检测到61对显著的上位性互作, 涉及54个QTL, 其中23个是能同时影响2~4个性状的多效位点, 且有8个位点与单向方差分析检测到的相同。控制长宽比的13对上位性互作位点中, 与控制粒长的上位性互作位点完全相同的有8对。以上结果为进一步开展水稻籽粒大小和形状有利基因的精细定位、克隆和分子设计育种奠定了基础。     

水稻种子耐贮藏性QTL主效应和上位性效应分析

利用珍汕97B／密阳46构建的RIL群体及其相应分子遗传图谱,采用种子加速老化鉴定法,以处理后的相对发芽率(％)作为该材料耐贮藏性的考察指标,分别对老化处理7d和14d后测得的数据进行QTL定位和上位性分析。共检测到2个主效应QTL与控制种子耐贮藏性有关。在老化程度相对较轻时,qSS9-1起忍耐作用：而在老化程度相对较重时,则有qSS4基因在起作用。试验还检测到5对耐贮藏性的上位性互作OTL,其贡献率为8．14％～13．51％,涉及第1、2、3、4、5、6、9、12等8条染色体。其中在7d老化处理试验中,检测到2对效应值相对较小的上位性互作;而在14d老化处理试验中,检测到3对效应值相对较大的上位性互作。     

Mapping of quantitative trait loci (QTLs) for rice protein and fat content using doubled haploid lines

Rice protein content (RPC) and rice fatcontent (RFC) are two important componentsof rice nutritional quality. In order toexamine the genetic basis of these traits,a doubled haploid (DH) population and anRFLP linkage map consisting of 232 markerloci were used to search QTLs for thetraits with the computer programQTLMapper1.0. This program is based onmixed linear models and allows simultaneousmapping of both main-effect and digenicepistastic QTLs in a DH population. RPC andRFC were evaluated based on a dry weightbasis of head rice by the Kjeldahl andSoxhlet methods respectively. A total offive main-effect QTLs for RPC wereresolved. The five QTLs collectivelyexplained 74% of the phenotypic variationwith LOD=15.2. Among these QTLs, the majorQTL qRPC-5 with the largest effectwas mapped in the interval of RG435-RG172aon chromosome 5. It accounted for 35% ofthe phenotypic variation with a LOD of16.7. At this locus the allele from theparent `Gui 630' increased RPC by 1.32%.The second QTL qRPC-7 was mapped inthe interval ZG34B-G20 on chromosome 7. Itexplained 23% of the phenotypic variancewith a LOD of 6.1. Its positive alleles,also from the parent `Gui 630', increasedRPC by 1.05%. As for the remaining threeQTLs, their additive effects wererelatively small and their positive alleleswere all inherited from the parent `02428'.Three QTLs for RFC were mapped onchromosome 1, 2 and 5 respectively. Theycollectively explained 44% of thephenotypic variation. Among these loci,QTLs qRFC-2 and qRFC-5 withlarger effects individually accounted for24% and 26% of the phenotypic variancerespectively. At QTL qRFC-2 thepositive allele came from the parent `Gui630', while at QTL qRFC-5 thepositive allele from the parent `02428'.The fact that both parents possess thepositive alleles at the QTLs for the twotraits provides an appropriate explanationfor the large transgressive segregationobserved in the DH lines. Furthermore, onlyone pair of epistatic loci explaining only5.1% of the phenotypic variance wasdetected for RPC, whereas seven pairs ofepistatic loci were resolved for RFC. Thetotal absolute effects of these RFCinteractions amounted to 0.97% which ismuch larger than that (0.42%) of the threemain-effect QTLs for the trait. Alongwith the observation that RPC showed a highheritability (78%), these resultsdemonstrate that RPC in the DH populationcould be mainly controlled by relativelyfew QTLs with large main-effects. As forRFC, epistatic interactions might be aneven more important component of thegenetic basis and the segregation of the DHlines could be largely explained by a fewmain-effect QTLs and many epistatic loci.In addition, a highly negative correlation(r = –0.45) between RPC and RFC inthe DH population was observed. Thiscorrelation could be largely explained bythe linkage of qRPC-5 and qRFC-5 with the directions of effectsopposite and the co-locations of the twoepistatic loci for RFC respectively withtwo different main-effect QTLs for RPC. Theinformation reported in the present papermay be useful for improving ricenutritional quality by means ofmarker-assisted selection.