Mapping quantitative trait loci for milling
Abstract Milling properties, protein content, and flour color are important factors in rice. A marker-based ge-netic analysis of these traits was carried out in this study using recombinant inbred lines (RILs) derived from an elite hybrid cross ‘Shanyou 63’, the most-widely grown rice hybrid in production in China. Correlation analysis shows that the traits were inter-correlated, though the co-efficients were generally small. Quantitative trait locus (QTL) analysis with both interval mapping (IM) and composite interval mapping (CIM) revealed that the milling properties were controlled by the same few loci that are responsible for grain shape. The QTL located in the interval of RM42-C734b was the major locus for brown rice yield, and the QTL located in the interval of C1087-RZ403was the major locus for head rice yield.These two QTLs are the loci for grain width and length,respectively. The Wx gene plays a major role in deter-mining protein content and flour color, and is modified by several QTLs with minor effect. The implications of the results in rice breeding were discussed.
Keywords Rice quality · Milling characteristics · Protein content · Flour color · Quantitative trait locus (QTL) · Molecular marker
Introduction
Rice is one of the major staple cereal foods, feeding more than half the world population. Demand for quality rice has always been a major factor in rice marketing and becomes more important in developing countries as the economic status of the people increases (Juliano et al.1990; Unnevehr et al. 1992).
Although rice quality has many components and is re-lated to preference in different cultures, its major ele-ments include milling properties, grain size, shape and appearance, and cooking and eating characteristics.Among these, protein content, milled rice recovery (es-pecially head rice recovery) and grain color are primary concerns (Khush et al. 1979; Unnevehr et al. 1992; Juliano 1998; Siebenmorgan 1998). Rice is the major protein source in most rice-eating areas and protein can also influence the physicochemical properties of cooked rice (Hamaker and Griffin 1990, 1991; Marshall et al.1990; Juliano 1993; Hamaker 1994). Head rice yield;which affects market value, is directly related to brown rice yield and milled rice yield which together form the ”milling quality” (Webb 1980; Juliano 1985; Unnevehr et al. 1992). Color is important in consumer acceptance of grain appearance and for end-products such as noodles (Juliano 1985; Collado et al. 1997; Bhattacharya et al. 1999). Few reports on the genetic basis of such traits are available because of the complexity of their in-heritance and the effect of environmental and other con-ditions. For example, protein content can be highly af-fected by the degree of milling and by environmental conditions, e.g. nitrogen fertilizer and growth duration (Perez et al. 1996). This complexity has led to the failure of breeding efforts to improve the protein content of rice grain (IRRI 1983; Coffman and Juliano 1987). Elucida-tion of their genetic basis would greatly help to improve the above-mentioned traits.
The recent development of DNA markers and linkage maps of rice have provided new opportunities for the ge-netic improvement of rice grain quality (Causse et al 1994; Harushima et al. 1998). With linkage maps based
Communicated by F. Salamini
Y .F. Tan · Y .Z. Xing · J.P. Hua · X.L. Sun · Q.F. Zhang National Key Laboratory for Crop Genetic Improvement, Huazhong Agricultural University, 430070 Wuhan, China M. Sun
Population Genetics Laboratory, Department of Zoology, The University of Hong Kong, Pokfulam Road, Hong Kong SAR, China
Y .F. Tan · H. Corke (?)
Cereal Science Laboratory, Department of Botany, The University of Hong Kong, Pokfulam Road, Hong Kong SAR, China e-mail: hcorke@https://www.sodocs.net/doc/8d17924016.html,
Tel.: 852-********, Fax: 852-********
Theor Appl Genet (2001) 103:1037–1045?Springer-Verlag 2001
Y.F. Tan · M. Sun · Y.Z. Xing · J.P. Hua · X.L. Sun Q.F. Zhang · H. Corke
Mapping quantitative trait loci for milling quality, protein content and color characteristics of rice using a recombinant inbred line population derived from an elite rice hybrid
Received: 15 September 2000 / Accepted: 31 March 2001
on DNA markers, complex polygenic traits can be dis-sected into single Mendelian quantitative trait loci (QTLs) (Paterson et al. 1988; Tanksley 1993). Many QTLs for traits of agronomic importance, such as yield and quality attributes, have been detected (McKenzie and Rutger 1983; Anh et al. 1993; Yu et al. 1997; Tan et al. 1999, 2000). Information on QTL analysis has ac-cumulated quickly, and will eventually help the manipu-lation of the complex traits in genetic engineering and rice breeding (Tanksley 1993; Xu 1997; Yano and Sasaki 1997).
In this study we used Shanyou63, the top elite indica hybrid rice of China, as the material to investigate the genetic basis of milling characteristics, the protein con-tent and the flour color of milled rice. The objectives were to determine: (1) the relationships among the traits, and (2) the number, positions, and genetic effect of the QTLs responsible for the traits.
Materials and methods
Plant materials
The two parents, Zhenshan 97 (ZB, maternal) and Minghui 63 (MH, paternal), the F1hybrid and a population of 238 F10recom-binant inbred lines (RILs) derived from the F2plants by single-seed descendent (SSD), were planted in a randomized block de-sign with three replications (Yu et al. 1997; Xing 1999) in the summer rice growing season, 1997, at Huazhong Agricultural Uni-versity, Wuhan. Field management followed the normal agronom-ic procedures as described, and natural ripening of the grain oc-curred.
Trait measurement
Milling characteristics
The harvested paddy rice from different replications was com-bined and stored at room temperature for at least 3 months before testing. The paddy rice was de-hulled and milled as described in Tan et al. (1999). Rough rice (40 g) was de-hulled and milled in duplicate using a mill (Jiading Food and Oil Machinery Factory, Shanghai, China) according to the National Standard NY 147–88. Head rice was manually separated with a set of screens. Broken grains with two-thirds of the whole grain were included in the head rice. Brown rice percentage, milled rice percentage, and head rice percentage were calculated based on the rough rice weight.
The duplicate milled rice grain samples were combined and ground into powder using an Udy Cyclone Sample Mill (Udy, Fort Collins, Colo., USA), through a 100-mesh sieve.
Protein content
Crude protein content was measured using the Kjeldahl method (Kjeltec System 1002, Tecator, Sweden). A nitrogen conversion factor of 5.95 was used to calculate the protein content of the rice flour (AACC 1995).
Flour-color parameters
Flour color was determined with a chromometer (CR-300, Minolta Camera Co., Ltd., Tokyo, Japan) using the CIE 1976 L*a*b* color system (Pomeranz and Meloan 1987). L* is the brightness value ranging from 0 (black) to 100 (white); a* is a function of red-green (positive a* indicates redness and negative indicates green-ness); b* is a yellow-blue value (positive value for yellowness and negative for blueness) (Oliver et al. 1992).
Linkage map construction and QTL assays
The linkage map consisted of 162 RFLP (restriction fragment length polymorphism) and 48 SSR (simple sequence repeat) mark-ers covering 12 chromosomes as described in Xing (1999) and Tan et al. (2000). Pearson correlation coefficients among the traits and one-way analysis of variance with the marker genotypes as groups were conducted using the statistical package Statistica (StatSoft, Tusla Okla.). The whole genome was scanned for quantitative trait loci (QTLs) using MAPMAKER/QTL 1.0 (Lander et al. 1987; Lincoln et al. 1992) with a LOD threshold of 2.0 (Lander and Botstein 1989; van Ooijen 1999). If two or more QTLs were de-tected from the scanning results of interval mapping, the QTL with the largest effect was fixed to re-scan the whole genome. Ad-ditionally, QTL Cartographer Version 1.13 was also used for com-posite interval mapping as the threshold LOD 2.0 is somewhat low (Zeng 1994; Basten et al. 1999; van Ooijen 1999). Only the QTLs detected by both methods were listed and all the QTLs for a spe-cific trait were combined together for the calculation of the total-likelihood and variance contribution.
Results
Distribution and heritability of the traits
The distributions of protein content and the flour color values L*, a*, b* of the RIL population, as well as the parents and the hybrid (F1) (Table 1, Figure 1), showed
1038
Table1Means and standard deviations (in brackets) of traits for parents and the hybrid, and variation of the RIL population and herita-bility of traits
Source BR a MR HR PRO L*a*b*
MH b77.9 (2.11) a c70.9 (2.75) a51.0 (6.93) b7.1 (0.21) b102.8 (1.06) a?0.29 (0.03) a 6.76 (0.53) a ZB79.6 (1.05) a71.8 (1.55) a59.7 (4.58) a8.6 (0.29) a102.7 (1.10) a?0.25 (0.02) a 6.83 (0.62) a F179.4 (1.01) a72.1 (1.88) a59.1 (4.53) a 6.1 (0.28) c104.0 (0.63) a?0.28 (0.02) a 5.83 (0.54) b RIL 79.8 (1.59)71.5 (2.49)56.2 (11.10)7.1 (0.89)103.3 (1.04)?0.24 (0.04) 6.34 (0.84) Range72.2–85.861.2–77.524.4–77.5 4.7–9.3100.4–105.6?0.40–?0.14 4.52–8.84
h2 d29.8%30.1%30.8%31.5%39.9%37.5%39.6%
a BR=brown rice grain percentage (%), MR=milled rice grain per-centage (%), HR=head rice grain percentage (%), PRO=protein content of the whole milled flour (%), See Materials and methods for details of parameter calculations
b MH=Minghui 63, paternal line of the cross, ZB=Zhenshan 97, maintainer of the sterile line (i.e. maternal line) of the cross
c The same letter indicates that the character is not significantly different at P<0.05 by Duncan’s multiple range test
d Broad-sens
e heritability calculated as: h2=δg2/(δg2+δe2)×100%
that the differences in all traits between the two parents were small. The bell-shaped phenotypic distributions and the wide range of variation of the investigated parame-ters indicated transgressive segregations, suggesting the polygenic inheritance of the traits.
Though the differences of the traits between the two parents were generally small, significant differences were still observed for head rice percentage (HR), pro-tein content (PRO), and the b* value of the flour color (B) (Table 1). The values of the hybrid are close to that of ZB, indicating the possibility of a maternal effect and/or dominance in the cross.
The estimates of broad-sense heritability of the traits did not vary much from trait to trait. BR, MR, HR and
1039
Fig.1Distributions of milling characteristics,
protein content and flour color parameters in the
RIL population
PRO were estimated to have heritabilities of around 30%, whereas those of the color parameters were around 40%. The heritabilities of the investigated traits were much lower (30%–40%), compared with those of yield (67%) and yield components (62%–87%) (Yu et al. 1998), a result which is consistent with the correlation coefficients of the traits (see below).
Correlation of the traits
The phenotypic correlation coefficients among the inves-tigated parameters (Table 2) were generally small, indi-cating the complexity of the relationship among the traits. A significantly positive correlation was observed between BR and MR (r=0.273), and between MR and HR (r=0.570). This is easy to understand as the milling degree was limited to bran-removal of 8–10% of the brown rice weight. Both parameters are calculated based on the rough rice weight. More brown rice would yield more milled rice and thus give more head rice. PRO was not significantly correlated with the milling parameters,
indicating good control of the milling.
For the flour color parameters, the L* value was posi-
tively related to BR (r=0.312), and negatively correlated
to MR (r=?0.326), HR (r=?0.352) and PRO (r=?0.396),
respectively. The correlation among the milling parame-
ters and PRO is also consistent with the former result, i.e.
the milling process can influence PRO (Juliano 1985).
The a* value was negatively correlated with MR (r=
?0.172) and HR (r=?0.168), respectively. The b* value was negatively correlated to BR (r=?0.286), and positive-
ly to MR (r=0.333), HR (r=0.392) and PRO (r=0.462), re-
spectively. These results were consistent with the high
correlation between L* and b* values (r=?0.928).
QTL mapping of the traits
Milling characteristics
One QTL was detected for BR in the interval RM42-
C734b of chromosome 5 (Table 3). The QTL could ex-
1040
Table 2Correlation coeffi-cients among the parameters from 238 RILs derived from the cross Zhenshan 97×Ming-hui 63Item BR a MR HR PRO L*a*b* BR 1.000
MR0.273** a 1.000
HR?0.0610.570** 1.000
PRO?0.0640.0900.123 1.000
L*0.312**?0.326**?0.352**?0.396** 1.000
a*?0.035?0.172**?0.168**?0.1040.000 1.000
b*?0.286**0.333**0.392**0.462**–0.928**?0.207** 1.000
a Abbreviations are the same as in Table 1
b Significance at P<0.05 * or P<0.01 **
Table3QTL information for rice flour color parameters and protein contents by interval mapping with the RIL popula-tion of the cross Zhenshan
97×Minghui 63. All intervals with a QTL peak with LOD
>2.0 are listed. The positive additive value indicated a con-tribution from the allele of Zhenshan 97 (ZB) whereas the negative value is from Minghui 63 (MH)Trait Chrom Interval Position a b LOD R2
(cM)a(%)c
Brown rice5RM42-C734b8.0 1.00 4.010.0 percentage (%)Total 4.010.0
Milled rice 3C1087-RZ403 6.1 1.10 2.2 4.8 percentage (%)5C246-C144710.9?1.32 2.87.0
Total 4.911.3
Head rice3C1087-RZ403 6.2 6.00 5.210.1 percentage (%)Total 5.210.1
Protein (%)6C952-Wx 1.5?0.61 6.813.0 7R1245-RM234 1.4?0.43 3.2 6.0
Total9.217.7 L*5R3166-RG3607.50.46 2.1 4.5 6C952-Wx 4.60.848.615.7
8L363A-RZ6626.00.50 2.410.2
Total11.421.9 a*4G102-G2358.70.02 3.3 6.9 7C1023-R1440 5.60.03 5.210.5
Total8.316.6 b*1G359-RG5327.9?0.33 2.5 5.9 3C1087-RZ403 6.90.27 2.3 4.3
6C952-Wx 1.7?0.7913.725.4
8RM223-L363 A 2.8?0.33 3.0 5.6
Total20.535.8
a Distance by Haldane function (Haldane 1919) from the left marker of the interval
b Additive effect computed as: (ZB–MH)/2
c Phenotypic variance ex-plaine
d by th
e QTL(s)
plain 10.0% of the phenotypic variance with LOD=4.0.At this QTL the ZB allele increased the BR by 1.00%.We found that the QTL for grain width also maps in this region (Tan et al. 2000). This is also consistent with the significant positive correlation between BR and grain width (data not shown).
Two QTLs were detected for MR with opposite ef-fects in the intervals C1087-RZ403and C246-C1447of chromosomes 3 and 5, respectively (Table 3). The QTL on chromosome 3 explained 4.8% of the phenotypic variance with an increasing additive effect of 1.07%from the ZB allele, whereas the one on chromosome 5explained 7.0% of the phenotypic variance with an in-creasing additive effect of 1.28% from the MH allele. In total, the two QTLs could explain 11.3% of the variance with LOD=4.9.
One QTL was mapped on C1087-RZ403of chromo-some 3 to have an effect on HR, where a QTL for grain length was also detected (Tan et al. 2000). The QTL ex-plained 10.1% of the variance, and the ZB allele at this locus increased head rice percentage by 6.0%.
To avoid ‘‘false” QTLs from close linkage, re-scan-ning of the chromosome was carried out by alternately fixing one of the two QTLs (Lander and Botstein 1989;Lincoln et al. 1992). Also one-way analysis of variance (ANOV A) and composite interval mapping (CIM) (Zeng 1994) were performed to scan the chromosome region.The QTLs were also detected by ANOV A and CIM, thus confirming the presence of the QTLs in this chromo-some.
It was interesting to note that a QTL for grain length was mapped in the interval C1087-RZ403(Tan et al.2000), which is again consistent with a negative correla-
tion of the traits (data not shown). Generally, longer grains can be easier to break during abrasive milling than short grains under the same conditions.Protein content
Two QTLs were detected to have an effect on protein content. One mapped in the interval of C952-Wx on chromosome 6, with the larger effect explaining 13.0%of the phenotypic variance and LOD=6.8. In this locus the MH allele increased the protein content by 0.61%.The other, with a smaller effect, mapped on chromosome 7 in the interval R1245-RM234(Table 2). In total, the two QTLs explained 17.7% of the phenotypic variance with LOD=9.2.Color parameters
Three QTLs were detected to have an effect on L*, i.e.the brightness of the whole milled flour, on chromo-somes 5, 6 and 8. One mapped on chromosome 6 and explained 15.7% of the phenotypic variance with LOD=8.6. The other two had relatively small effects. In total, the three QTLs explained 21.9% of the phenotypic variance.
Two QTLs were detected for the flour a* value, on chromosomes 4 and 7. The QTL on chromosome 7 had a larger effect, explaining 10.5% of the variance with a LOD score of 5.2, whereas the other QTL accounted for 5.6% of the variance. In both cases alleles from ZB in-creased the a* value (i.e. decreased the greenness of the flour). In total, they explained 14.8% of the phenotypic variance.
Four QTLs on chromosomes 1, 3, 6 and 8 influenced the b* value of the flour. The QTL in the interval C952-Wx of chromosome 6 had the largest effect with LOD=13.7 and explained 25.4% of the phenotypic vari-ance. This is consistent with the negative correlation of
1041
Fig.2Locations of the QTLs for milling parameters, protein con-tent and color parameters of rice grain and flour. The numbers on the top indicate chromosome order. Dashed lines show linkage
gaps or regions with a distance >50 cM. The bars indicate the 1-LOD support intervals of the QTLs identified. Small triangles indicate peaks of the LOD contours
L* and b* (Table 2) and the mapping result of L* on chromosome 6 (Table 3 and Fig. 2). The MH allele at this locus increased b* by 0.79. The other three QTLs had smaller effects on this trait. The four QTLs ex-plained 35.8% of the phenotypic variance of b* with a LOD score of 15.8.
QTL interaction for the traits
Because of the transgression of the traits and the signifi-cantly lower-PRO and higher-B value of the hybrid as compared with the parents (Table 1 and Fig. 1), the QTLs detected were of small magnitude and accounted for a small proportion of the variation (Table 3). There-fore, we carried out a two-way analysis to detect epistat-ic interactions of all two-marker combinations across the whole genome. In total 21945 combinations were as-sayed and 136 combinations showed a significant inter-action, covering all the chromosomes (data not shown). Two kinds of interaction were detected: one was between loci that did not have significant effects on the traits (non-effect locus), and the other was between QTLs and
1042
Table4List of significant two-way interactions between different loci covering the ge-nome Trait Marker 1Marker 2F-Test a MC-Test b
p1-(1-p)n Name https://www.sodocs.net/doc/8d17924016.html, Chrom.
Brown rice G1128b1RM20030.0040.047 percentage (%)RZ5992C148100.010.114
C6245*c G35910.000d0.000
C14475C732120.00020.002
R17897C87120.0040.047
R16298C47290.0020.024
RM25810R496120.0020.024
Milled rice RG1731RZ40490.0080.092 percentage (%)R3213RG36050.002d0.024
R31665RG33380.0010.010
Head rice RG1731RM4250.0020.024 Percentage (%)C633*C96260.00010.001
C10873*C1003B110.000d0.000
RZ4674C123290.010.114
C10164L1044110.0040.047
C14475TEL3110.0020.024
Protein (%)C9221R1930.0030.035
C10164C909B120.0060.070
R16298TEL3110.0002d0.002
RM22810R496120.0070.081
L*G1128b1RM5320.0020.024
RG1731C734b50.0040.047
R25102RG65360.0010.012
R10146*RG65360.000d0.000
R27496*C48380.0000.000
RZ4717R217460.00040.005
RM707RM239100.00010.001
RG56110C1237110.00010.001
a*G3931RZ59920.0001d0.001
RZ5992RG39330.00010.001
C7463C5640.0030.035
RZ5993RZ66760.0000.000
R7123L1044110.0010.012
C7463R887120.0020.024
RG3605R26590.00040.005
RG3605C1003B110.00060.007
RM2347R543a110.0040.047
b*RG1011C11210.0010.012
G1128b1RM5320.0040.047
R25102RG65360.0010.012
C9526*C153B20.0000.000
R28696*C48380.000d0.000
C9526*C290.0000.000
C4746*Y6854L110.0000.000
RM707RM222100.00020.002
a F test for the four sub-groups of the two marker alleles
b Monte Carlo simulation using EPISTAT program (Lark et al. 1995)
c * Significant effect (QTL) was also detecte
d in th
e region
d Th
e mean value comparison within the four groups o
f the combinations were listed in Table 5
the non-effect locus (Table 4). No interaction was detect-ed between QTLs. The significant combinations with the highest likelihood [if the markers are linked (<50 cM)]are listed in Table 4, and mean-value comparisons of the four groups are listed in Table 5.Taking BR as an example, when C624(for simplicity using the marker name to denote the linked QTL) had the allele from Zhenshan 97 (BB) and G359had the allele from Minghui 63 (AA), the combination (BBAA)had a significantly higher value than the other three groups (P <0.05). The difference between the BBAA and BBBB genotypes should be due to the epistasis of the QTL linked to the two loci. Another example is the com-bination R321/RG360for MR: when both alleles were from Minghui 63 (AAAA) or Zhenshan 97 (BBBB) MR had significantly lower values (P <0.05) and vice versa. Similar situations were observed in other traits (Table 5).Figure 3 shows schematically the interactions be-tween loci. Although neither of the loci (G1128B /RM200) had a significant effect on BR, the four sub-groups had different values showing the dependence of the two alleles (Fig. 3a). When the G359alleles were from Minghui 63 (AA), C624had significantly different BR between the two alleles (Fig. 3b), i.e. the phase of G359could significantly influence the effect of C624.Discussion
Generally, the QTLs all had a low magnitude which is consistent with the low heritability (30%–40%) of the
traits. Another reason may be the limited difference be-tween the two parents, with most of the traits not being significantly different (Table 1). However, we still de-tected QTLs which could not be found using traditional methods, by using interval mapping, and confirmed their occurrence with the composite interval-mapping algo-rithm.It is not surprising that the QTLs for the milling parameters are located in the regions for grain shape (Table 2 and Tan et al. 2000). The most important mill-ing parameter is head rice percentage which shares the QTL for grain length (Tan et al. 2000). This result is easily understandable, i.e. the longer rice tends to break more easily during milling, and implies that medium-long and slender rice is preferable in breeding practice.Alternatively, the control of the milling method is also very important to obtain a higher yield of head rice for the long-grain type (van Ruiten 1985).We detected a QTL in the Wx gene region, responsi-ble for the protein content, that had a large effect. This result is consistent with the previous reports that starch synthetase, which is correlated with amylose content and is embedded in the starch granule, is one of the milled rice proteins (Sano 1984; Villareal and Juliano 1986,1989). Although it is generally considered that protein content is influenced largely by environmental condi-tions and the level of nitrogen fertilization (Nanda and
1043
Table 5Comparison of selected two-locus combinations from Table 4 within the RIL population and the hybrid indicating significant epistasis between two markers
Marker combination a BR b MR HR PRO L*a*b*
(C624/G359)(R321/RG360)(C1087/C1003B)(R1629/TEL3)(R1014/RG653)(G393/RZ599)(R2869/C483)
Marker 1Marker 2AA c AA 78.8 c B d 70.8 b A
55.0 b A 7.3 a A 102.9 b B ?0.23 a A 6.97 a A AA BB 79.8 b A 72.1 a A 48.3 c B 6.7 b B 103.3 b B ?0.26 b B 6.35 b B BB AA 80.4 a A 72.3 a A 57.5ab A 6.8 b B 103.9 a A ?0.25 b B 5.97 c B BB
BB
79.8 b A
71.2
b A
61.6
a A
7.2 a AB 103.2 b B ?0.24 a A 6.04 c B
a Marker order the same as in Table 4
b Abbreviations the same as in Table 1
c AA represents alleles from Minghui 63, an
d BB from Zhenshan 97d
The same letter indicates that the character is not significantly different at P 0.05(lower case) or P 0.01(upper case) by Duncan’s multiple range test
Fig.3Schematic representation of two-way interactions for BR indicating epistasis between loci
Coffman 1979; Perez et al. 1996), our results strengthen the recognition of a genetic component for protein con-tent.
It is worth emphasizing that as rice is unique among cereals by having a storage protein primarily made of glutelin [which has a more balanced amino-acid profile than the prolamin-rich storage proteins (Juliano 1985)], increasing the protein content will increase and balance the protein intake people whose staple food is rice. The recent report of “golden rice” promotes the potential of genetic engineering to enhance the nutritional quality of rice (Ye et al. 2000), and hence improve the nutritional state of people in poor areas where rice is the staple food. The availability of the Wx gene sequence provides the possibility of improving the protein content via Wx gene modification (Wang et al. 1995).
Also, one of the major QTLs for color parameters happens to be located in the Wx gene region, which is also the major one responsible for protein content. The two QTLs on chromosome 5 (R3136-RG360) and chro-mosome 3 (C1087-RZ403) are also responsible for grain width and length (Table 3; Tan et al. 2000). This result is consistent with the fact that milling removes the outer parts of the brown rice grain, i.e. the bran, which con-tains more of the pericarp, seed coat, aleurone layer, and the embryo and, hence, has a higher protein content. These components are all darker than the starchy endo-sperm (Juliano 1985). The wider and longer the grain, the more bran will be removed, and therefore the color of the grain is lighter. Overall a compromise for long grain, a better color of the grain/flour and high head rice yield, seems to have been reached. Genetic manipulation can be focused on the corresponding regions of chromo-somes 3, 5 and the Wx gene when marker-assisted selec-tion is carried out.
The results show strongly the importance of epistasis between different loci in accounting for transgressive segregation of the traits, consistent with previous re-ports (Lark et al. 1995; Li et al. 1997; Yu et al. 1997). First, the numbers of loci involved in the interactions are much higher than those of the QTLs detected. For example, only one QTL was detected on chromosome 5 for BR. However, seven pairs of loci covering ten chro-mosomes were detected as having significant effects on this trait by two-way interaction analysis (Table 4). If the higher levels of interactions are taken into consider-ation, the involved loci should be much higher. Second, although the variance explained by the QTLs was rela-tively small (most of which are <10%), that explained by the interaction should be very large. Effective meth-ods need to be developed to extract all of the variance. Finally, the epistasis has an important impact on breed-ing practice. Because of the interaction between differ-ent loci, QTL-linked C624 would have significant effects on BR only when the allele of G359 is from Minghui 63 (Fig. 3). This means that the offspring phenotype will be largely influenced by the genetic background of the receptor line when marker-directed selection is carried out.
With the use of DNA markers, improvement for these traits can be quickly achieved by focusing on the target QTLs, without sacrificing important agronomic traits. Meanwhile, the interaction between different loci should be carefully considered. These results should facilitate the improvement of hybrid rice quality in future breeding programs.
Acknowledgments We thank Dr. S. D. Tanksley’s group at Cor-nell University and the Japanese Rice Research Program for kind-ly providing the RFLP probes. This research was supported by the Hong Kong Research Grants Council, the National Natural Sci-ence Foundation of China, and the Rockefeller Foundation.
References
AACC (1995) Approved methods of the American Association of Cereal Chemists, 9th edn. Method 46–16 approved December 1988, reviewed October 1994. The Association, St. Paul, Min-nesota
Ahn SN, Bollich CN, McClung AM, Tanksley SD (1993) RFLP analysis of genomic regions associated with cooked-kernel elongation in rice. Theor Appl Genet 87:27–32
Batsen CJ, Weir BS, Zeng ZB (1999) QTL cartographer, version
1.13. Department of Statistics, North Carolina State Universi-
ty, Raleigh, North Carolina
Bhattacharya M, Luo Q, Corke H (1999) Time-dependent changes in dough color in hexaploid wheat landraces differing in poly-phenol oxidase activity. J Agric Food Chem 47:3579–3585 Causse MA, Fulton TM, Cho YG, Ahn SN, Chunwongse J, Wu K, Xiao J, Yu Z, Ronald PC, Harrington SE, Second G, McCouch SR, Tanksley SD (1994) Saturated molecular map of the rice genome based on an interspecific backcross population. Ge-netics 138:1251–1274
Coffman WR, Juliano BO (1987) Rice. In: Olson RA (ed) Nutri-tional quality of cereal grains. Genetic and agronomic improvement. Agronomy Society of America, Madison, Wisconsin, pp 101–129
Collado LS, Mabesa LB, Corke H (1997) Genetic variation in the color of sweetpotato flour related to its use in wheat-based composite flour products. Cereal Chem 74:681–686
Haldane JBS (1919) The combination of linkage values, and the calculation of distances between the loci of linked factors.
J Genet 8:299–309
Hamaker BR (1994) The influence of rice protein on rice quality.
In: Marshall WE, Wadsworth JI (eds) Rice science and tech-nology, Marcel Dekker, New York, New York, pp 177–194 Hamaker BR, Griffin VK (1990) Changing the viscoeleastic prop-erties of cooked rice through protein disruption. Cereal Chem 67:261–264
Hamaker BR, Griffin VK (1991) Potential influence of a starch granule-associated protein on cooked rice stickiness. J Food Sci 56:1327–1329
Hirushima Y, Yano M, Shomura A, Sato M, Shimano T, Kuboki Y, Yamamoto T, Lin SY, Antonio BA, Parco A, Kajiya H, Huang N, Yamamoto K, Nagamura Y, Kurata N, Khush GS, Sasaki T (1998) A high-density rice genetic linkage map with 2275 markers using a single F2population. Genetics 148:479–494 IRRI (1983) Annual report for 1982. International Rice Research Institute. Los Ba?os, Laguna, The Philippines
Juliano BO (1985) Rice chemistry and technology, 2nd edn.
American Association of Cereal Chemists, St. Paul, Minnesota Juliano BO (1990) Rice grain quality: problems and challenges.
Cereal Foods World 35:245–253
Juliano BO (1993) Rice in human nutrition. FAO Food Nutrition Series No. 26. International Rice Research Institute: Manila, The Philippines
Juliano BO (1998) Varietal impact on rice quality. Cereal Food World 43:207–222
1044
Juliano BO, Perez CM, Kaosa-Ard M (1990) Grain quality charac-teristics of export rices in selected markets. Cereal Chem 67:192–197
Khush GS, Paule CM, De la Cruz NM (1979) Rice grain quality evaluation and improvement at IRRI. Proc Workshop Chemi-cal Aspects of Rice Grain Quality. International Rice Research Institute, Los Ba?os, Laguna, The Philippines
Lander ES, Botstein D (1989) Mapping Mendelian factors under-lying quantitative traits using RFLP linkage maps. Genetics 121:185–198
Lander ES, Green P, Abrahamson J, Barlow A, Daly MJ, Lincoln SE, Newburg L (1987) MAPMAKER: an interactive computer package for constructing primary genetic linkage maps of ex-perimental and natural populations. Genomics 1:174–181
Lark KG, Chase K, Alder F, Mansur LM, Orf J (1995) Interactions between quantitative trait loci in which trait variation at one locus is conditional upon a specific allele at another. Proc Natl Acad Sci USA 92:4656–4660
Li ZK, Pinson SRM, Park WD, Paterson AH, Stansel JW (1997) Epistasis for three grain yield components in rice (Oryza sativa L.). Genetics 145:453–465
Lincoln S, Daly M, Lander ES (1992) Mapping genes controlling quantitative traits with MAPMAKER/QTL 1.1. Whitehead In-stitute Technical Report, 2nd edn. Whitehead Institute, Cam-bridge, Massachusetts
Marshall WE, Normand FL, Goynes WR (1990) Effects of lipid and protein removal on starch gelatinization in whole-grain milled rice. Cereal Chem 67:458–463
McKenzie KS, Rutger JN (1983) Genetic analysis of amylose con-tent, alkali spreading score, and grain dimensions in rice. Crop Sci 23:306–311
Nanda JS, Coffman WR (1979) IRRI’s efforts to improve the pro-tein content of rice. In: Proc Workshop Chemical Aspects of Rice Grain Quality. International Rice Research Institute, Manila, The Philippines, pp 33–47
Oliver JR, Blakeney AB, Allen HM (1992) Measurement of flour color in color space parameters. Cereal Chem 69:544–551 Ooijen JW van (1999) LOD significance thresholds for QTL anal-ysis in experimental populations of diploid species. Heredity 83:613–624
Paterson AH, Lander ES, Hewitt JD, Peterson S, Lincoln SE, Tanksley SD (1988) Resolution of quantitative traits into Men-delian factors using a complete linkage map of restriction frag-ment length polymorphisms. Nature 335:721–726
Perez CM, Juliano BO, Liboon SP, Alcantara JM, Cassman KG (1996) Effects of late nitrogen fertilization application on head rice yield, protein content, and grain quality of rice. Cereal Chem 73:556–560
Pomeranz Y, Meloan CE (1987) Food analysis: theory and prac-tice, 2nd edn. Van Nostrand Reinhold, New York Ruiten HTL van (1985) Rice milling: an overview. In: Juliano BO (ed) Rice chemistry and technology, 2nd edn. American Asso-ciation of Cereal Chemists, St. Paul, Minnesota, pp 349–388 Sano Y (1984) Differential regulation of waxy gene expression in rice endosperm. Theor Appl Genet 68:467–473 Siebenmorgen TJ (1998) Influence of postharvest processing on rice quality. Cereal Foods World 43:200–202
Tan YF, Li JX, Yu SB, Xing YZ, Xu CG, Zhang QF (1999) The three important traits for cooking and eating qualities of rice grain are controlled by a single locus. Theor Appl Genet 99:642–648
Tan YF, Xing YZ, Li JX, Yu SB, Xu CG, Zhang QF (2000) Gene-tic bases of appearance quality of rice grains in Shanyou 63, an elite rice hybrid. Theor Appl Genet 101:823–829
Tanksley SD (1993) Mapping polygenes. Annu Rev Genet 27: 205–233
Unnevehr LJ, Duff B, Juliano BO (1992) Consumer demand for rice grain quality. International Rice Research Institute, Ma-nila, The Philippines, and International Development Research Center, Ottawa, Canada
Villareal CP, Juliano BO (1986) Waxy gene factor and residual protein of rice starch granules. Starch/St?rke 38:118–121 Villareal CP, Juliano BO (1989) Comparative levels of waxy gene product of endosperm starch granules of different rice eco-types. Starch/St?rke 41:369–372
Wang ZY, Zhen FQ, Shen GZ, Gao JP, Snustad P, Li MG, Zhang JL, Hong MM (1995) The amylose content in rice endosperm is related to the post-transcriptional regulation of the waxy gene. Plant J 7:613–622
Xing YZ (1999) Molecular dissection of genetic bases of impor-tant agronomic traits in rice. PhD dissertation. Huazhong Agricultural University, Wuhan, China
Xu YB (1997) Quantitative trait loci: separating, pyramiding and cloning. Plant Breed Rev 15:85–39
Yano M, Sasaki T (1997) Genetic and molecular dissection of quantitative traits in rice. Plant Mol Biol 35:145–153
Ye XD, Al-Babili S, Kl?ti A, Zhang J, Lucca P, Beyer P, Potrykus
I (2000) Engineering the provitamin A (β-Carotene) biosyn-
thetic pathway into (carotene-free) rice endosperm. Science 287:303–305
Yu SB, Li JX, Xu CG, Tan YF, Gao YJ, Li XH, Zhang Q, Saghai Maroof MA (1997) Importance of epistasis as the genetic ba-sis of heterosis in an elite rice hybrid. Proc Natl Acad Sci USA 94:9226–9231
Yu SB, Li JX, Xu CG, Tan YF, Gao YJ, Li XH, Zhang Q, Saghai Maroof MA (1998) Epistasis plays an important role as the ge-netic basis of heterosis in rice. Science in China (Series C) 41:294–302
Zeng ZB (1994) Precision mapping of quantitative trait loci. Ge-netics 136:1457–1468
1045