2006 genetic basis of drought resistance at reproductive stage in rice
DESCRIPTION
TRANSCRIPT
Copyright � 2006 by the Genetics Society of AmericaDOI: 10.1534/genetics.105.045062
Genetic Basis of Drought Resistance at Reproductive Stage in Rice:Separation of Drought Tolerance From Drought Avoidance
Bing Yue,* Weiya Xue,* Lizhong Xiong,* Xinqiao Yu,† Lijun Luo,† Kehui Cui,*Deming Jin,* Yongzhong Xing* and Qifa Zhang*,1
*National Key Laboratory of Crop Genetic Improvement and National Center of Plant Gene Research (Wuhan), Huazhong AgriculturalUniversity, Wuhan 430070, China and †Shanghai Agrobiological Gene Center, Shanghai 201106, China
Manuscript received May 1, 2005Accepted for publication October 19, 2005
ABSTRACT
Drought tolerance (DT) and drought avoidance (DA) are two major mechanisms in drought resistance ofhigher plants. In this study, the genetic bases of DTand DA at reproductive stage in rice were analyzed using arecombinant inbred line population from a cross between an indica lowland and a tropical japonica uplandcultivar. The plants were grown individually in PVC pipes and two cycles of drought stress were applied toindividual plants with unstressed plants as the control. A total of 21 traits measuring fitness, yield, and theroot system were investigated. Little correlation of relative yield traits with potential yield, plant size, and roottraits was detected, suggesting that DTand DA were well separated in the experiment. A genetic linkage mapconsisting of 245 SSR markers was constructed for mapping QTL for these traits. A total of 27 QTL wereresolved for 7 traits of relative performance of fitness and yield, 36 QTL for 5 root traits under control, and 38for 7 root traits under drought stress conditions, suggesting the complexity of the genetic bases of both DTand DA. Only a small portion of QTL for fitness- and yield-related traits overlapped with QTL for root traits,indicating that DT and DA had distinct genetic bases.
DROUGHT is one of the major abiotic stresseslimiting plant production. The worldwide water
shortage and uneven distribution of rainfall makes theimprovement of drought resistance especially impor-tant (Luo and Zhang 2001). Fulfillment of this goalwould be enhanced by an understanding of the geneticand molecular basis of drought resistance.
However, little progress has been made in character-izing the genetic determinants of drought resistance,because it is a complex phenomenon comprising a num-ber of physio-biochemical processes at both cellularand organismic levels at different stages of plant de-velopment (Tripathy et al. 2000). Drought resistanceincludes drought escape (DE) via a short life cycle ordevelopmental plasticity, drought avoidance (DA) viaenhanced water uptake and reduced water loss, droughttolerance (DT) via osmotic adjustment (OA), antioxi-dant capacity, and desiccation tolerance. The recentdevelopment of high-density linkage maps has providedthe tools for dissecting the genetic basis underlyingcomplex traits, such as drought resistance, into individ-ual components. Quantitative trait locus (QTL) map-ping has been carried out in an attempt to determinethe genetic basis of several traits that may be related todrought resistance, including OA (Lilley et al. 1996;Zhang et al. 1999, 2001; Robin et al. 2003), cell-
membrane stability (Tripathy et al. 2000), abscisic acid(ABA) content (Quarrie et al. 1994, 1997), stomatalregulation (Price et al. 1997), leaf water status, and rootmorphology (Champoux et al. 1995; Ray et al. 1996;Price and Tomos 1997; Yadav et al. 1997; Ali et al. 2000;Courtois et al. 2000; Zheng et al. 2000; Zhang et al.2001; Kamoshita et al. 2002; Price et al. 2002). However,it is not clear how these attributes are related to theperformance of the genotypes at the whole-plant level,and how they function to reduce the drought damage tofitness- and productivity-related traits.
Plants are most susceptible to water stress at thereproductive stage. Dramatic reduction of grain yieldoccurs when stress coincides with the irreversible re-productive processes, making the genetic analysis ofdrought resistance at the reproductive stage crucially im-portant (Cruz and O’Toole 1984; Price and Courtois
1999; Boonjung and Fukai 2000; Pantuwan et al.2002). However, variation of flowering time in segre-gating populations often made the phenotyping ofdrought resistance rather inaccurate. Staggering theseed-sowing time has been suggested to synchronize theflowering time of a population in QTL mapping (Price
and Courtois 1999). Lanceras et al. (2004) also re-ported QTL mapping of yield and yield componentsunder different water regimes in the field by synchro-nizing flowering time of the mapping population. How-ever, the success has been limited because of thedifficulty in achieving a real synchronization of theflowering time in a segregating population. In addition
1Corresponding author: National Key Laboratory of Crop GeneticImprovement, Huazhong Agricultural University, Hongshang District,Wuhan 430070, China. E-mail: [email protected]
Genetics 172: 1213–1228 (February 2006)
to flowering time, segregation for plant size and rootvolumes also confounds the accuracy of QTL mapping.It is almost impossible to distinguish the genetic basis ofDT from other contributing factors (such as DA andDE) in drought resistance under field conditions inwhich drought stress is applied to and withdrawn fromall plants simultaneously.
In this study, we adopted a protocol for droughttreatment by planting and stressing rice plants of arecombinant inbred line (RIL) population in individualpolyvinyl chloride (PVC) pipes in which the variousgenotypes were stressed to the same extent at the samedevelopmental stage. We showed that such an experi-mental design cleanly separated DT from DA, thusallowing relatively independent analyses of the geneticbases of DT and DA.
MATERIALS AND METHODS
Plant materials and drought stress treatment: A populationconsisting of 180 RILs at F9/F10 generation was developedfrom a cross between the lowland rice cultivar Zhenshan 97(Oryza sativa L. ssp. indica) and the upland rice cultivarIRAT109 (O. sativa L. ssp. japonica). Zhenshan 97 is the main-tainer line for a number of elite hybrids widely cultivated inChina, and IRAT109 was developed in Cote d’Ivoire.
For phenotyping, rice plants were grown in PVC pipes, oneplant per pipe, under a rain-out shelter with movable roofs.The pipe was 20 cm in diameter and 1 m in length with holeson two sides at 25, 50, and 75 cm from the top. Each pipe wasloaded with a plastic bag filled with 38 kg of thoroughly mixedsoil composed of two parts of clay and one part of river sand, towhich 25 g of fertilizers (including 4 g each of N, P2O5, andK2O) was added.
Sowing time was staggered among the lines to synchronizeflowering on the basis of the heading dates of the linesobserved in 2002. Three to five germinated seeds were directlysown in each pipe and only one healthy plant was kept at30 days after sowing. At the beginning of the tillering stage, 1 gof urea (dissolved in water) was applied to each pipe. Theplants were fully irrigated by watering every day until thedrought treatment. Drought stress was individually applied toeach plant at the booting stage. To apply drought stress, waterwas added to the full capacity of the pipe, the plugs on the pipewere then removed, and small holes were punched on theplastic bag to drain the water slowly. Rain was kept off byclosing the roof during periods of rain. When all leaves of astressed plant became fully rolled, as visualized at noon—apoint corresponding to the relative water content in the rangeof 72–75%, as checked in this study—watering was applied tothe full capacity of the pipe. With the full water level main-tained for 1 day, the second cycle of drought stress was appliedto the plant until all leaves became fully rolled again. After thesecond round of stress, watering was resumed for the rest ofthe life cycle.
The pipes were laid out in six blocks following a randomizedcomplete block design. Drought stress was applied to three ofthe blocks with the other three blocks used as control. In 2003,150 RILs and the parents were phenotyped with two pipes perblock for each genotype. In 2004, 75 RILs and the parents weretested to represent the resistant and susceptible lines on thebasis of relative yield in 2003, with only one pipe per block foreach genotype.
Traits and measurements: A total of 21 traits were scored inthis study; 9 of them were traits collected from the above-ground part of the plants and the other 12 were root traits(Table 1).
The traits collected from above-ground parts were related tofitness and productivity, including yield and yield componenttraits, biomass, and fertility. Yield and yield-related traits wereexamined for all plants under stress and the control con-ditions, including grain yield per plant (in grams), number ofspikelets per panicle, 1000-grain weight (in grams), fertilepanicle rate (%), spikelet fertility (%), biomass (in grams) andharvest index (%). Fertile panicle rate was the proportion ofthe number of fertile panicles (with 5 grains or more on eachpanicle) in all the panicles of a plant. Spikelet fertility wasmeasured as the number of grains divided by the total numberof spikelets of a plant. Harvest index was scored as grain yielddivided by the total dry matter of the above-ground part. Therelative performance of the phenotypes for each trait was mea-sured simply as the ratios of the measurements taken underdrought stress and control conditions.
In addition, two traits related to the water status of theplants, leaf-drying score and number of days to leaf rolling,were also recorded. Leaf-drying score was recorded on thebasis of the degrees of leaf drying immediately after rewateringas 0 (no leaf drying) to 4 (.20% of the leaf area was drying).Number of days to leaf rolling of each plant was recorded asthe number of days from the application of drought stress tothe day when all leaves became rolled at noon.
The root traits were scored at seed maturity of the plants. Tomeasure these traits, the plastic bag containing the soil androots was pulled out from the PVC pipe and laid out on a 2-mmsieve screen frame. The lowest visible root in the soil afterremoving the plastic bag was scored as the maximum rootdepth (in centimeters). The body of soil and roots was cut intotwo parts at 30 cm from the basal node of the plant and the soilwas washed away carefully to collect roots. The volumes (inmilliliters) of roots from the two parts were measured in acylinder using the water-replacing method (Price and Tomos
1997). The root mass below 30 cm was considered to be deeproot, from which a number of measurements were derived.Root growth rate in depth and root growth rate in volume werecalculated by dividing the maximum root depth and the totalroot volume, respectively, by the root growth period (numberof days from sowing to heading of the plant). Drought-induced root growth was evaluated by two traits: drought-induced root growth in depth and drought-induced deep-rootrate in volume, which were calculated as the differences ofmaximum root depth and deep-root rate in volume betweenthe measurements obtained under drought stress and controlconditions.
The abbreviations for and descriptions of these traits arelisted in Table 1 and used hereafter.
DNA markers, map construction, and QTL analysis: A totalof 245 nuclear simple sequence repeat (SSR) markers wereused for constructing the linkage map. The SSR primers andmarker assays essentially followed Temnykh et al. (2000, 2001)and McCouch et al. (2002). The program of Mapmaker/EXP3.0 (Lincoln et al. 1992) was used to construct the geneticlinkage map. The means of the traits were used to identify QTLby Windows QTL Cartographer 2.0 (Zeng 1994). The LODthresholds were determined by 500 random permutations,which resolved that, at a false positive rate of ,0.05 for eachtrait, the LOD thresholds ranged from 1.9 to 2.4 for 20 ofthe 21 traits. The only exception was relative fertile panicles(RFP), in which the LOD threshold was 2.6 for the data of2003 and 4.1 for 2004. For ease of presentation, a uniformthreshold of 2.4 was adopted for the 20 traits, and 2.6 and 4.1were used for RFP for the 2 years, respectively. The results of
1214 B. Yue et al.
both years were presented for QTL with a LOD score .2.4 in1 year but in the range of 2.0–2.4 in the other year for the 20traits.
RESULTS
Phenotypic variation of the parents and RILs: Thephenotypic differences between parents as well as thevariation in the RIL population are summarized inTable 2. Transgressive segregation was observed in theRIL population for all the traits investigated. ANOVA ofthe data collected in 2003 indicated that variation dueto genotype differences was highly significant for all
the traits, although the relative proportions of variancevaried from one trait to another (Table 3).
IRAT109 showed more drought resistance thanZhenshan 97 in both years by having higher values inrelative performance of the traits related to fitness andproductivity (Table 2). The differences between the twoparents for relative yield, relative biomass, relative spike-let fertility, and relative grain weight were significant atthe 0.01 probability level in 2003. Thus Zhenshan 97suffered much more drought damage than IRAT109.
The reverse performance was observed between theparents for the two traits related to water status (Table2). The leaf-drying score of IRAT109 was significantly
TABLE 1
Abbreviations, full names, and descriptions of the traits investigated in this study
Abbreviation Trait Description
RY Relative yield per plant (%) Yield per plant under drought stress/Yield per plantunder control conditions
RSF Relative spikelet fertility (%) Spikelet fertility under drought/Spikelet fertilityunder control conditions
RBM Relative biomass (%) Biomass per plant under drought/Biomass per plantunder control conditions
RFP Relative rate of fertile panicles (%) Rate of fertile panicles (with five seeds or more) perplant under drought/Rate of fertile panicles perplant under control conditions
RHI Relative harvest index (grain yield/biomass) (%) Harvest index under drought/Harvest index undercontrol conditions
RGW Relative grain weight (%) Weight of 1000 seeds under drought/Weight of1000 seeds under control conditions
RSN Relative number of spikelets per panicle (%) No. of spikelets per panicle under drought/no. ofspikelets per panicle under control conditions
LDS Leaf-drying score Degrees of leaf drying immediately after rewatering,scored 1 (no drying) to 5 (.20% area dried)
DLR No. of days to leaf rolling No. of days to leaf rolling starting from day of droughttreatment
MRDC Maximum root depth under control (cm) The lowest visible root at the soil surface after removingthe plastic bag under control conditions
MRDD Maximum root depth under drought (cm) The lowest visible root at the soil surface after removingthe plastic bag under drought conditions
DIRD Drought-induced root growth in depth (cm) The difference of maximum root depth under droughtand control conditions
RGDC Root growth rate in depth under controlconditions (cm/day)
Maximum root depth divided by root growth periodunder control conditions
RGDD Root growth rate in depth under droughtconditions (cm/day)
Maximum root depth divided by root growth periodunder drought conditions
RVC Root volume under control conditions (ml) The volume of roots under control conditions measuredusing the water-replacing method
RVD Root volume under drought conditions (ml) The volume of roots under drought conditions measuredusing the water-replacing method
DRVC Deep root rate in volume under controlconditions (%)
Percentage of root volume ,30 cm in the total rootvolume under control conditions
DRVD Deep root rate in volume under droughtconditions (%)
Percentage of root volume ,30 cm in the total rootvolume under drought conditions
RGVC Root growth rate in volume under controlconditions (ml/day)
Total root volume divided by root growth period undercontrol conditions
RGVD Root growth rate in volume under droughtconditions (ml/day)
Total root volume divided by root growth period underdrought conditions
DIDRV Deep root rate in volume induced by droughtconditions (%)
The difference in deep-root rate in volume underdrought and control conditions
Genetic Basis of Drought Resistance in Rice 1215
less than that of Zhenshan 97 in both years, whileZhenshan 97 could sustain longer time than IRAT109before leaf rolling as reflected by the DLR scores.
For most of the root traits (Table 2), IRAT109 hadhigher values than Zhenshan 97 under both control anddrought stress conditions in both years. In at least oneyear, the differences between parents for maximum rootdepth under control, root volume and deep-root rateunder both drought stress and control conditions, anddrought-induced deep-root rate in volume were signif-icant. Zhenshan 97, however, showed more drought-induced root growth in depth than IRAT109 did, andthe difference was significant in 2003. Again, trans-gressive segregation was observed in all the root traits.
When the data collected from the 2 years werecompared, DLR was substantially higher in 2004 thanin 2003 for both parents (Table 2), indicating that thestress developed more slowly in 2004 due to the milderweather conditions during drought stress (the tempera-ture and evaporation was higher in 2003). Consequently, anumber of other traits also showed significant differencesbetween the 2 years in one or both parents, includingrelative yield, relative number of fertile panicles, relativegrain weight, and relative spikelet number. Significantdifferences between the 2 years were also observed inseveral root traits in one or both parents.
Correlations of the traits: The traits related to fitnessand productivity, e.g., relative yield, relative spikelet fertil-ity, relative rate of fertile panicle, relative biomass, relative
grain weight, and relative harvest index were highlycorrelated with each other (Table 4). This suggested thatthe yield loss and harvest index reduction under droughtstress in late season were associated with the reduction ofspikelet fertility, fertile panicle rate, biomass and grainweight. In particular, a very high correlation (0.85–0.95)was observed between relative yield, relative spikeletfertility, and relative harvest index in both years.
Figure 1 illustrates the relationships of relative yieldand relative biomass with yield and biomass under con-trol conditions. It was clear from Figure 1 that relativeyield was not correlated with yield under control con-ditions, and thus genotypes with high and low yieldpotential were equally stressed. Similarly, there was littlecorrelation between relative biomass and biomass un-der control conditions, and thus genotypes with largeand small plant sizes were equally stressed. Moreover,relative yield was not significantly correlated with bio-mass under control, and neither was relative biomasssignificantly correlated with yield under control.
There was no correlation between the two traits re-lated to water status of the plants (Table 4). There wereno consistent correlations between these two traits withthe relative performance of the traits related to fitnessand productivity in 2 years, except the negative corre-lation detected in both years between relative biomassand number of days to leaf rolling.
The root traits investigated were also highly corre-lated with each other (Table 5). In general, constitutive
TABLE 2
The measurements of the traits in the RIL population and the parents in 2003 and 2004
Trait Zhenshan 97 IRAT109 Mean of RILs Range of RILs
RY 43.9/65.7*** 80.6**/81.9 58.2/52.6 (19.6–90.8)/(17.9–90.5)RSF 54.2/69.1 74.3**/88.6 63.9/63.7 (24.2–94.5)/(22.4–95.6)RBM 79.0/81.8 94.9**/89.6 90.4/81.0 (70.3–100.0)/(57.1–99.2)RFP 88.3/92.5**** 93.5/100.0**** 80.0/94.0 (28.1–100.0)/(68.6–100.0)RHI 52.1/66.9 65.6/74.8 59.2/58.6 (20.3–100.0)/(18.3–96.9)RGW 73.5/76.2 88.0**/97.8*,**** 87.6/82.0 (58.0–104.1)/(63.2–104.1)RSN 89.6/98.3**** 91.9/94.8*** 84.8/94.3 (52.1–100.5)/(68.6–100.2)LDS 3.0*/2.67* 1.7/1.3 2.4/1.8 (1.0–3.8)/(0.3–3.3)DLR 18.5**/22.0*,*** 10.3/16.7**** 12.1/19.4 (7.0–17.5)/(8.0–26.7)MRDC 53.6/53.3 61.1**/67.0* 61.8/57.9 (47.2–79.8)/(39.0–75.5)MRDD 76.7/82.7 79.5/92.3*** 81.9/87.1 (64.8–94.5)/(69.0–95.7)DIRD 23.1*/29.4 18.4/25.3*** 20.1/29.2 (7.0–33.8)/(14.7–48.0)RGDC 0.8/0.8 0.8/0.9 0.8/0.9 (0.6–1.0)/(0.5–1.0)RGDD 1.2/1.3 1.0/1.3*** 1.0/1.1 (0.7–1.4)/(0.8–1.6)RVC 84.0***/51.0 84.3***/70.0* 112.3/82.6 (46.3–231.4)/(43.9–146.9)RVD 73.0***/45.2 102.5**,***/75.7 107.8/89.7 (43.0–234.6)/(29.8–175.1)DRVC 8.7/8.9 22.4**,***/12.8* 13.3/9.2 (2.5–28.8)/(0.8–22.4)DRVD 17.6/16.4 25.6/33.0*,*** 19.0/24.8 (3.7–36.3)/(10.6–44.1)RGVC 1.3***/0.8 1.1/0.8 1.4/1.0 (0.8–2.3)/(0.7–1.7)RGVD 1.1/0.7 1.3/1.1 1.3/1.1 (0.6–2.3)/(0.4–1.8)DIDRV 8.9/7.5 3.2/20.2**,**** 5.7/15.6 (�4.2–18.9)/(1.6–29.1)
The number at the left of the ‘‘/’’ is the result of 2003, and the number at the right is the result of 2004. *,**Significantly higherthan the other parent at the 0.05 and 0.01 probability levels based on t-test. ***,****Significantly higher than the other year of thesame parent at the 0.05 and 0.01 probability levels based on t-test.
1216 B. Yue et al.
root growth (maximum root depth and root volumeunder control) had high and consistent correlationswith other root traits. For example, maximum rootdepth was highly significantly correlated in both yearswith all the root traits, except drought-induced rootgrowth in volume. A similar situation was also obviousfor root volume under control that was also highly cor-related with most root traits. The highest correlation(.0.90) detected was between root volume and rootgrowth rate under both control and drought conditions.
Correlations between traits in different groups areshown in Table 6. In general, there was not much cor-relation between the relative performance of fitness-and productivity-related traits and the root traits, withexceptions of only a few marginal cases in 2004, all ofwhich suggested root growth under drought had smallnegative effects on yield and biomass. Thus, variation inroot traits contributed very little toward reducing thedrought stress of the upground parts in this experiment.In addition, relative yield, relative biomass, and relativefertility were not significantly correlated with floweringtime (data not shown), as expected on the basis of theexperimental design. All this demonstrated that thepipe planting effectively minimized the effects of DA orDE on relative yield and yield-related traits. Therefore,the relative yield, relative spikelet fertility, and relativebiomass examined in this study were indeed regulatedalmost exclusively by DT mechanisms under the ex-perimental conditions and thus can be viewed as DTtraits although the underlying mechanisms remain tobe investigated.
Table 6 also showed no correlation between leaf-drying score and the root traits. Number of days to leafrolling was negatively correlated with a number of traitsmeasuring root volumes under both drought stress andcontrol conditions; thus leaf rolling occurred sooner inplants with larger root volumes. However, there was onehighly significant positive correlation between numberof days to leaf rolling and root growth in depth underdrought, indicating drought-induced root growth indepth may have a positive effect on delaying leaf rolling.
The linkage map: A total of 410 SSR markers weresurveyed and 245 (59.8%) of them showed polymor-phism between the two parents. A linkage map was
TABLE 3
ANOVA of the traits based on the data of 2003
Trait Variation d.f. MS F P
RY Genotype 151 1262.89 7.23 0.0000Block 2 1550.86 8.88 0.0002Error 302 174.56
RSF Genotype 150 1222.68 3.83 0.0000Block 2 946.45 2.97 0.0521Error 300 319.06
RBM Genotype 150 289.24 1.38 0.0120Block 2 623.25 2.97 0.0518Error 300 209.63
RFP Genotype 150 589.01 2.90 0.0000Block 2 1738.76 8.54 0.0003Error 300 203.45
RHI Genotype 149 1560.11 3.14 0.0000Block 2 978.40 1.97 0.1391Error 298 497.14
RGW Genotype 150 138.38 2.77 0.0000Block 2 28.76 0.58 0.5683Error 300 49.93
RSN Genotype 150 323.34 2.83 0.0000Block 2 31.27 0.27 0.7655Error 300 114.38
LDS Genotype 149 2.32 4.83 0.0000Block 2 5.89 12.27 0.0000Error 298 0.48
DLR Genotype 151 16.00 7.07 0.0000Block 2 6.67 2.95 0.0525Error 302 2.26
MRDC Genotype 151 109.84 3.73 0.0000Block 2 1397.86 47.42 0.0000Error 302 29.48
MRDD Genotype 150 126.42 2.70 0.0000Block 2 3330.98 71.11 0.0000Error 300 46.84
DIRD Genotype 149 123.42 2.01 0.0000Block 2 875.99 14.24 0.0000Error 298 61.53
RGDC Genotype 151 0.02 2.59 0.0000Block 2 0.22 25.52 0.0000Error 302 0.01
RGDD Genotype 150 0.06 5.07 0.0000Block 2 0.17 15.4 0.0000Error 300 0.01
RVC Genotype 151 4398.12 10.35 0.0000Block 2 411.89 0.97 0.3824Error 302 424.96
RVD Genotype 151 5195.99 12.85 0.0000Block 2 1578.62 3.90 0.0211Error 302 404.31
DRVC Genotype 151 0.02 5.45 0.0000Block 2 0.07 19.09 0.0000Error 302 0.004
DRVD Genotype 151 0.04 4.32 0.0000Block 2 0.04 4.87 0.0083Error 302 0.01
RGVC Genotype 151 0.42 5.85 0.0000Block 2Error 302 0.07
(continued )
TABLE 3
(Continued)
Trait Variation d.f. MS F P
RGVD Genotype 151 0.43 6.9 0.0000Block 2 0.74 11.92 0.0000Error 302 0.06
DIDRV Genotype 151 209.78 2.73 0.0000Block 2 204.93 2.67 0.0702Error 302 76.72
MS, mean square; F, F-statistic.
Genetic Basis of Drought Resistance in Rice 1217
constructed using Mapmaker analysis based on data fromthe 245 SSR markers assayed on the 180 RILs (Figure 2).The map covered a total length of 1530 cM with an av-erage interval of 6.2 cM between adjacent markers.
QTL for relative performance of the traits relatedto fitness and productivity: QTL detected for relativeperformance of seven traits related to fitness and pro-ductivity are listed in Table 7(see also Figure 2). A totalof 27 QTL were resolved for the seven traits, including8 QTL detected in both years and 19 QTL observed inonly 1 year. The detection is quite consistent, consider-ing the large scale of the experiment, the nature of thetraits, and the secondary statistics of ratios as input data.All the QTL that were detected in both years appearedto have larger effects in 2004 than in 2003, as indicatedby the LOD scores and the amounts of variation ex-plained. This is expected since the lines planted in 2004were selected on the basis of the extreme phenotypesfrom the previous year.
Alleles from IRAT109 at 14 of the QTL had positiveeffects on the relative performance of these traits, whilealleles from Zhenshan 97 at the other 13 loci contrib-uted positively to the relative performance (Table 7). Ofthe 8 QTL that were consistently detected in both years,the IRAT109 alleles at 7 QTL had positive effects on therelative performance of these traits. Interestingly, oneregion on chromosome 9, RM316–RM219, was partic-ularly active by exhibiting significant effects simulta-neously on relative yield (QRy9), relative spikelet fertility(QRsf9), relative biomass (QRbm9), and relative harvestindex (QRhi9). Another region on chromosome 8,RM284–RM556, was detected to have a significant ef-fect on relative yield (QRy8), relative spikelet fertility(QRsf8), and relative number of fertile panicles (QRfp8).It is also worth noting that almost all the QTL detectedin both years had large effects on the traits as reflectedby the large proportions of the phenotypic variationexplained (10% or more).
QTL for the two plant water status traits: Six QTLwere resolved for leaf-drying score and four QTL fornumber of days to leaf rolling (Table 8; Figure 2). In
both cases, one QTL was detected in both years andthe others were detected in only 1 year. As in the traitsfor relative performance described above, the regionRM219–RM296 on chromosome 9 showed a large effecton number of days to leaf rolling (QDlr9). Also a QTLfor leaf-drying score (QLds3b) had a large effect on thetrait in both years.
QTL for root traits under control conditions: A totalof 36 QTL were resolved for the five root traits undercontrol conditions (Table 9; Figure 2), of which 7 were de-tected in both years and the remaining 29 in only 1 year.Again, the effects observed in 2004 were larger thanthose in 2003 for all the QTL detected in both years,except for one QTL, QRgvc3, for root growth rate involume under control conditions. While the IRAT109alleles at 22 of the 36 QTL contributed positively to theroot traits, alleles from Zhenshan 97 at 5 of the 7 QTLthat were observed in both years had positive effects onthe root traits. Of the 19 QTL each explaining .10% ofphenotypic variation, the IRAT109 alleles at 12 QTLcontributed to the increase of the trait measurements.Again, there were a number of regions where QTL fortwo or more traits were detected, including the intervalsRM472–RM104 on chromosome 1, RM231–RM489on chromosome 3, both RM471–RM142 and RM349–RM131 on chromosome 4, both RM125–MRG4499 andRM429–RM248 on chromosome 7, RM316-RM219 onchromosome 9, and RM287–RM229 on chromosome 11.In all the QTL having effects on multiple traits, exceptone, alleles from the same parents contributed in thesame direction to different traits, suggesting the likeli-hood that different QTL are due to the effects of thesame genes.
QTL for root traits under drought stress: A total of38 QTL were observed for the seven root traits underdrought stress conditions (Table 10; Figure 2), including6 detected in both years and 32 detected in only 1 year.Unlike other traits described above, the effects of QTLdetected in 2004 were not necessarily larger than thoseresolved in 2003 for the QTL detected simultaneously inboth years. Alleles from IRAT109 at 23 of the 38 QTL
TABLE 4
Coefficients of pairwise correlations of the above-ground traits investigated in 2003 and 2004
RY RSF RBM RFP RHI RGW RSN LDS
RSF 0.88/0.85RBM 0.35/0.40 0.15/0.03RFP 0.58/0.46 0.64/0.51 0.26/0.14RHI 0.95/0.85 0.89/0.94 0.15/�0.07 0.46/0.44RGW 0.44/0.61 0.36/0.47 0.10/0.27 0.30/0.38 0.44/0.48RSN 0.37/0.03 0.21/�0.07 0.23/�0.04 0.27/0.01 0.33/0.08 0.32/0.04LDS �0.31/0.03 �0.26/0.05 �0.23/0.13 �0.34/0.14 �0.24/0.05 �0.15/0.04 �0.21/0.09DLR �0.36/�0.21 �0.23/�0.11 �0.29/�0.37 �0.12/0.00 �0.33/�0.03 �0.39/0.05 �0.27/0.12 0.09/�0.21
Critical values at the 0.01 probability level are 0.21 and 0.30 for 2003 and 2004, respectively. The number at the left of the ‘‘/’’ isthe result of 2003, and the number at the right is the result of 2004.
1218 B. Yue et al.
contributed to the increase of the trait measurements,whereas at the other 15 QTL, alleles from Zhenshan 97were in the direction of increasing the trait measure-ments. Of the 22 QTL each explaining .10% of phe-
notypic variation, alleles from IRAT109 at 17 loci hadpositive effects on these root traits.
The QTL were distributed very unevenly among thechromosomes, with 11 QTL on chromosome 4, 5 QTL
Figure 1.—Scatter plots of relative performance of yield and biomass against yield and biomass under control conditions in2003 (left) and 2004 (right). (A) Relative yield against yield under control; (B) relative biomass against biomass under control; (C)relative yield against biomass under control; (D) relative biomass against yield under control.
Genetic Basis of Drought Resistance in Rice 1219
on chromosome 7, 4 QTL on each of chromosomes 2and 3, 3 QTL on each of chromosomes 1, 8, 9, and 11,1 QTL on each of chromosomes 6 and 10, but noneon chromosomes 5 and 12. There were also obvioushotspots where QTL for two or more of the root traitsunder drought stress were detected, including regionsmostly on chromosome 4, as well as chromosomes 3, 7,9, and 11 (Figure 2).
Comparison of chromosomal locations of QTL fordifferent types of traits: Of the 21 chromosomal re-gions resolved with QTL for relative performance offitness- and productivity-related traits, 9 overlapped withthe QTL intervals for root traits (Figure 2). One regionon chromosome 9, RM316–RM219, in which multipleQTL were detected, showed relatively large effects onboth root traits and relative performance of fitness andproductivity; the other 9 regions had only 1 QTL, eachwith relatively small effects on the respective traits (Figure2; Tables 7, 9, and 10). In addition, positive alleles forthe two types of traits were contributed by differentparents in 4 of the 9 overlapping regions, including theregion RM316–RM219 on chromosome 9. The distinctchromosomal locations between QTL for fitness- andproductivity-related traits and root traits, and the dif-ferent directions of the allelic contributions for mostoverlapping QTL, were in good agreement with theresults of correlation analysis, further suggesting thatroot traits and relative performance of the fitness andproductivity traits had different genetic determinants.
Number of days to leaf-rolling and leaf-drying scoreare two traits reflecting plant water status. All four QTLfor number of days to leaf rolling overlapped with oneor more QTL for root traits, but none of them over-lapped with QTL for the relative performance of fitness-and productivity-related traits (Figure 2). Of the sixQTL for leaf-drying score, only one with small effectoverlapped with a QTL for relative spikelet number thatalso seemed to have impact on deep-root rate in volumeinduced by drought. Again, these results agreed wellwith the correlation results above, in which number ofdays to leaf rolling was significantly correlated with someof the root traits, while the leaf-drying score had littlecorrelation with either root traits or above-ground traits(Tables 4 and 6).
DISCUSSION
The PVC pipe protocol successfully separateddrought tolerance and drought avoidance: A major dif-ficulty in genetic analysis of drought resistance by apply-ing and relieving drought treatment at the same timefor all plants, as adopted by many previous studies, isthe inability to resolve the whole-plant resistance intoindividual components, such as DE, DA, and DT. Pre-vious studies showed that the drought resistance in-dex (relative yield) was often negatively correlated with
TA
BL
E5
Co
effi
cien
tso
fp
airw
ise
corr
elat
ion
so
fth
ero
ot
trai
tsin
vest
igat
edin
this
stu
dy
in2
00
3an
d2
00
4
MR
DC
MR
DD
DIR
DR
GD
CR
GD
DR
VC
RV
DD
RV
CD
RV
DR
GV
CR
GV
D
MR
DD
0.48
/0.
42D
IRD
�0.
49/�
0.80
0.54
/0.
19R
GD
C0.
38/
0.48
0.24
/0.
15�
0.13
/�
0.42
RG
DD
�0.
20/�
0.30
0.41
/0.
190.
60/
0.44
0.67
/0.
55R
VC
0.62
/0.
580.
33/
0.31
�0.
26/�
0.42
�0.
21/�
0.25
�0.
49/�
0.65
RV
D0.
56/
0.50
0.29
/0.
29�
0.26
/�
0.35
�0.
29/�
0.27
�0.
52/�
0.62
0.89
/0.
87D
RV
C0.
79/
0.76
0.43
/0.
46�
0.33
/�
0.53
0.32
/0.
36�
0.10
/�
0.19
0.53
/0.
470.
46/
0.51
DR
VD
0.57
/0.
480.
64/
0.59
0.09
/�
0.13
0.43
/0.
220.
28/
0.03
0.22
/0.
320.
19/
0.45
0.62
/0.
75R
GV
C0.
57/
0.53
0.35
/0.
29�
0.20
/�
0.39
�0.
02/�
0.04
�0.
29/�
0.43
0.95
/0.
930.
80/
0.77
0.50
/0.
460.
26/
0.33
RG
VD
0.51
/0.
430.
29/
0.31
�0.
20/�
0.27
�0.
15/�
0.10
�0.
34/�
0.38
0.83
/0.
730.
96/
0.94
0.42
/0.
500.
21/
0.51
0.81
/0.
73D
IDR
V�
0.13
/0.
010.
34/
0.46
0.46
/0.
280.
19/�
0.01
0.46
/0.
21�
0.27
/0.
04�
0.23
/0.
21�
0.26
/0.
190.
58/
0.80
�0.
21/
0.06
�0.
18/
0.30
See
Tab
le4
lege
nd
for
exp
lan
atio
ns.
1220 B. Yue et al.
potential yield and was also dependent on the actualdevelopmental stage of the plants when stress treatmentwas applied (Price and Courtois 1999; Venuprasad
et al. 2002; Toorchi et al. 2003). In an attempt toseparate the components in the field experiments,several approaches have been adopted, including stag-gering the sowing date, installing a drip irrigation sys-tem in the plots, normalizing the data by statisticalmethod, and utilizing advanced backcross lines (Blum
1988; Price and Courtois 1999; Robin et al. 2003).Although such measures were useful for improving theaccuracy of QTL mapping, it is nonetheless impossibleto assess the relative contributions of DE, DT, and DAto overall drought resistance at the whole-plant level.
In this study, the effect of DE was completely elim-inated because stress treatment was individually appliedto pipes on the basis of the developmental stage of theplants. A plant-wise drought treatment protocol wasused to ensure that all the plants received a similar levelof stress treatment such that genotypes with a deep-rootsystem or small size did not have an advantage inavoiding drought damage. This was confirmed by thevery low correlation of the relative performance of thefitness- and yield-related traits with the root traits, as wellas with potential yield and plant size as defined by yieldand biomass under the control conditions. All this indi-cates that the effects of DA and DT were well separatedin this experimental design. Thus, the relative perfor-mance of fitness- and yield-related traits under droughtstress and control conditions unambiguously providedmeasurements for DT. The root traits, however, pro-vided the measurements for DA, although the contri-bution of this component to drought resistance at thewhole-plant level was eliminated by the experimentaldesign. Therefore the genetic bases of DA and DT canbe separately analyzed using this data set.
The genetic bases of DT and DA are different: Thegenetic bases of DT and DA were rarely addressedseparately in previous studies. Zhang et al. (2001)studied QTL for OA and root traits and found that noQTL for OA overlapped with any of the QTL for roottraits. Lilley et al. (1996) reported tight linkage be-tween QTL for root traits and OA, with alleles forincreasing OA and root traits derived from differentparents.
In this study, a large number of QTL for DT and DAwere detected. The results indicated that most of theQTL for putative DT-related traits did not overlap withQTL for DA-related traits. In regions where QTL for DT-and DA-related traits were clustered, nearly a half of thepositive alleles for DT- and DA-related traits were fromdifferent parents. For example, in the QTL hotspot ofRM316–RM219 on chromosome 9, the positive allelesof QTL for deep-root traits (MRDC and DRVC) werefrom Zhenshan 97, while the IRAT109 alleles contrib-uted positively to relative yield, relative spikelet fertility,relative biomass, and relative harvest index. The distinct
TA
BL
E6
Co
effi
cien
tso
fp
airw
ise
corr
elat
ion
sb
etw
een
abo
ve-g
rou
nd
trai
tsan
dro
ot
trai
tsin
vest
igat
edin
this
stu
dy
in2
00
3an
d2
00
4
RY
RSF
RB
MR
FP
RH
IR
GW
RSN
LD
SD
LR
MR
DC
0.01
/�
0.05
0.05
/0.
020.
01/�
0.16
�0.
03/�
0.16
0.09
/�
0.01
�0.
02/�
0.25
0.09
/�
0.30
�0.
02/
0.12
�0.
12/�
0.24
MR
DD
�0.
07/�
0.14
�0.
02/�
0.15
0.11
/�
0.17
0.03
/0.
01�
0.05
/�
0.15
�0.
10/�
0.02
0.09
/�
0.12
�0.
11/�
0.12
0.20
/�
0.04
DIR
D�
0.07
/�
0.02
�0.
06/�
0.12
0.10
/0.
060.
05/
0.18
�0.
14/�
0.08
�0.
08/
0.26
0.01
/0.
25�
0.09
/�
0.20
0.32
/0.
24R
GD
C�
0.08
/0.
05�
0.08
/0.
08�
0.07
/�
0.10
�0.
09/
0.01
�0.
05/
0.11
�0.
20/
0.12
�0.
13/�
0.11
�0.
10/
0.11
0.21
/0.
22R
GD
D�
0.10
/0.
07�
0.11
/0.
01�
0.02
/�
0.02
�0.
03/
0.17
�0.
14/
0.06
�0.
18/
0.34
�0.
13/
0.13
�0.
12/�
0.07
0.36
/0.
45R
VC
0.04
/�
0.22
�0.
04/�
0.11
0.05
/�
0.21
�0.
09/�
0.1
0.08
/�
0.15
0.10
/�
0.34
0.13
/�
0.18
0.04
/0.
01�
0.34
/�
0.46
RV
D�
0.03
/�
0.29
�0.
08/�
0.31
0.11
/0.
04�
0.07
/�
0.18
�0.
04/�
0.40
0.10
/�
0.40
0.16
/�
0.18
0.07
/0.
07�
0.36
/�
0.55
DR
VC
0.06
/�
0.16
0.04
/�
0.04
0.00
/�
0.15
�0.
06/�
0.07
0.12
/�
0.10
0.06
/�
0.32
0.04
/�
0.25
0.05
/0.
12�
0.10
/�
0.34
DR
VD
�0.
10/�
0.29
�0.
03/�
0.24
�0.
03/�
0.11
0.05
/0.
01�
0.03
/�
0.31
�0.
11/�
0.29
0.01
/�
0.13
�0.
05/
0.09
0.30
/�
0.19
RG
VC
0.01
/�
0.23
�0.
09/�
0.11
0.04
/�
0.27
�0.
18/�
0.06
0.04
/�
0.12
0.04
/�
0.24
0.08
/�
0.15
�0.
02/
0.04
�0.
31/�
0.40
RG
VD
�0.
07/�
0.29
�0.
14/�
0.38
0.13
/0.
09�
0.11
/�
0.15
�0.
10/�
0.45
0.08
/�
0.30
0.13
/�
0.14
0.00
/0.
09�
0.37
/�
0.51
DID
RV
�0.
18/�
0.27
�0.
08/�
0.32
�0.
03/�
0.02
0.13
/0.
08�
0.16
/�
0.36
�0.
20/�
0.13
�0.
05/
0.04
�0.
11/�
0.02
0.46
/0.
04
See
Tab
le4
lege
nd
for
exp
lan
atio
ns.
Genetic Basis of Drought Resistance in Rice 1221
locations of the QTL and different directions of alleliccontributions of the parents in QTL in overlappingregions for DT- and DA-related traits suggested that DTand DA had different genetic bases. This also explainedthe lack of correlation between these two sets of traitsunder the experimental conditions.
The genetic complexity of the root traits: Rapiddevelopment of a deep-root system is considered a DAstrategy for plants as it enables absorption of water indeep soil layers (Fukai and Cooper 1995; Price andCourtois 1999). Although the putative contribution ofroot traits to DA or drought resistance could not beestimated in this experimental design, the QTL map-ping of various root traits under control and droughtstress conditions in this study provided a comprehensivescenario of the genetic controls of root morphologyunder normal conditions and root restructuring underdrought stress.
A total of 74 QTL were resolved for the 12 root traitsthat could be assigned to 36 genomic regions according
to the flanking markers. Comparisons with previousresults indicated that 4 of the 36 QTL regions had posi-tional correspondence with QTL for root or other DA-related traits reported in previous studies of rice. Forexample, in the region RM472–RM104 on chromosome1 where QTL for root volume, root growth in volume,and number of days to leaf rolling were detected in thisstudy, QTL were also identified for root thickness androot weight (Zheng et al. 2003), as well as for relativewater content, leaf rolling, and leaf-drying score (Babu
et al. 2003). The region RM160–RM215 on chromosome9, contributing to maximum root depth under both con-trol and drought stress in this study, was also identifiedas harboring QTL for upland seminal root length andrelative seminal root length in a previous study (Zheng
et al. 2003). The region RM470–RM303 on chromosome4, controlling deep-root rate, maximum root length,and root volume under drought stress in this study,corresponded to a region controlling root thickness,root penetration index, and penetrated root dry weight
Figure 2.—The molecular marker linkage map based on the RIL population from a cross between Zhenshan 97 and IRAT109.Genetic distance is given in Kosambi centimorgans. The QTL for above-ground traits and root traits are placed on the left andright sides of the chromosomes, respectively. QTL detected in both years are shown in boldface type. QTL in italics indicate thatthe alleles for increasing trait values are from Zhenshan 97. The full names of the traits are listed in Table 1.
1222 B. Yue et al.
TA
BL
E7
QT
Lfo
rre
lati
vep
erfo
rman
ceo
ftr
aits
rela
ted
tofi
tnes
san
dp
rod
uct
ivit
yre
solv
edu
sin
gco
mp
osi
tein
terv
alm
app
ing
inth
eR
ILp
op
ula
tio
no
fZ
hen
shan
97
/IR
AT
10
9
Tra
its
2003
2004
Ch
rom
oso
me
QT
LIn
terv
ala
LO
DA
dd
itiv
eef
fect
b
Ph
eno
typ
eva
riat
ion
cC
hro
mo
som
eQ
TL
Inte
rval
aL
OD
Ad
dit
ive
effe
ctb
Ph
eno
typ
eva
riat
ion
c
RY
2QRy2
RM
240–
RM
166
2.0
4.9
5.81
2QRy2
RM
240–
RM
166
3.8
7.4
13.6
48
QRy8
RM
284–
RM
556
3.4
�6.
29.
423
QRy3
RM
203–
RM
520
2.6
�6.
59.
849
QRy9
RM
316–
RM
219
5.0
7.7
14.1
69
QRy9
RM
316–
RM
219
6.8
10.1
25.7
010
QRy10
RM
496–
RM
228
2.5
�5.
88.
29
RSF
5QRsf5
RM
421–
RM
274
3.1
5.3
7.64
3QRsf3
RM
293–
RM
571
3.2
�7.
111
.51
8QRsf8
RM
284–
RM
556
2.4
�4.
86.
209
QRsf9
RM
219–
RM
296
5.4
10.9
30.4
69
QRsf9
RM
316–
RM
219
3.0
5.2
7.18
RB
M2
QRbm
2R
M57
3–R
M31
82.
31.
56.
172
QRbm
2R
M57
3–R
M31
83.
35.
114
.56
9QRbm
9R
M31
6–R
M21
93.
21.
67.
455
QRbm
5R
M50
7–R
M13
3.7
5.4
18.7
910
QRbm
10
RM
596–
RM
271
2.3
1.8
9.17
10QRbm
10
RM
596–
RM
271
5.6
8.5
28.6
0
RF
P8
QRfp8
RM
284–
RM
556
3.1
�4.
27.
5212
QRfp12
RM
235–
MR
G54
544.
2�
14.1
39.1
712
QRfp12
RM
235–
MR
G54
544.
7�
14.2
37.9
6
RSN
1QRsn1
RM
237–
RM
403
3.5
3.2
9.55
1QRsn1
RM
237–
RM
403
2.7
2.8
9.98
2QRsn2
RM
324–
RM
292.
6�
2.6
6.06
6QRsn6
RM
454–
MR
G43
713.
4�
3.3
14.0
412
QRsn12
RM
19–R
M45
35.
3�
4.0
20.8
6
RG
W2
QRgw
2R
M14
5–R
M32
42.
5�
1.8
5.97
3QRgw
3R
M52
3–R
M23
12.
92.
511
.78
5QRgw
5R
M50
9–R
M43
03.
22.
29.
349
QRgw
9R
M44
4–R
M31
62.
52.
39.
997
QRgw
7R
M12
5–M
RG
4449
2.6
2.1
8.36
RH
I2
QRhi2
RM
221–
RM
573
3.2
6.5
8.79
9QRhi9
RM
316–
RM
219
3.4
7.4
13.5
79
QRhi9
RM
316–
RM
219
4.4
7.4
11.4
510
QRhi10
RM
496–
RM
228
3.5
�7.
513
.90
aU
nd
erli
ned
chro
mo
som
en
um
ber
and
mar
ker
inte
rval
sin
dic
ate
QT
Ld
etec
ted
inb
oth
year
s.bT
he
po
siti
veva
lues
ind
icat
eth
eal
lele
sfr
om
IRA
T10
9w
ith
incr
easi
ng
effe
cts.
cA
mo
un
to
fp
hen
oty
pe
vari
atio
n(%
)ex
pla
ined
by
the
QT
L.
Genetic Basis of Drought Resistance in Rice 1223
as reported previously (Zhang et al. 2001). The regionRM231–RM489 on chromosome 3 controlling rootvolume and root growth rate in volume in this studycorresponded with the QTL for total root volume androot weight (Venuprasad et al. 2002). In the remaining32 chromosomal regions, 7 regions harbored QTL forroot traits detected under both control and droughtstress conditions, 7 regions included root QTL repeat-edly detected in 2 years, and 15 were regions in whichmultiple QTL were resolved.
At the 74 QTL for root traits resolved, alleles from theupland parent IRAT109 at 45 QTL had positive effectsfor increasing the trait values, while positive alleles werecontributed by the lowland parent Zhenshan 97 at theother 29 QTL. Among the 41 QTL with relatively largeeffects (explaining .10% variation), alleles from theupland parent at 29 loci had positive effects on theseroot traits. Thus, both upland and paddy rice couldmake positive contributions to DA, given the variousattributes of the root traits, although IRAT109 maycontribute more to DA than Zhenshan 97.
The likely mechanisms of DT in this population: Thestrategy for establishment of DT involves OA and main-tenance of cell-membrane stability, as well as detoxifi-cation (Tripathy et al. 2000; Chaves and Oliveira
2004). Although we did not measure these physiologicaltraits directly in this study, possible mechanisms of theDT may be deduced on the basis of collocations of QTLdetected in this and previous studies.
Three chromosomal regions with major QTL forrelative yield and yield-component traits in this studymatched very well with DT-related physiological traitsreported previously. For example, in the region RM284–RM556 on chromosome 8 harboring QTL for relativeyield, relative grain fertility and relative fertile paniclerate resolved in this study, and a QTL for OA withflanking markers RM284–RM210 was identified by Robin
et al. (2003). In the same region, QTL for OA and cell-membrane stability were also detected in other studies(Tripathy et al. 2000; Zhang et al. 2001). Compara-tive mapping indicated that this genomic region corre-sponded to a segment on wheat chromosome 7S where alocus associated with OA was identified (Tripathy et al.2000).
In the genomic region RM316–RM219 on chromo-some 9 where major QTL for relative yield, relative grainfertility, relative biomass, and relative harvest index wereidentified in this study, a QTL for cell-membrane stabil-ity (marked by RZ698–RM219) was also reported pre-viously (Tripathy et al. 2000). In another study, a QTLfor OA was identified in this region, and it was alsoshown that this region corresponded to a region inwheat where a QTL for ABA content was detected(Zhang et al. 2001).
In the region RM240–RM166 on chromosome 2where a QTL for relative yield was detected in this study,a QTL for relative yield (Babu et al. 2003) and two
TA
BL
E8
QT
Lfo
rp
lan
tw
ater
stat
us
trai
tsre
solv
edu
sin
gco
mp
osi
tein
terv
alm
app
ing
inth
eR
ILp
op
ula
tio
no
fZ
hen
shan
97
/IR
AT
10
9
Tra
its
2003
2004
Ch
rom
oso
me
QT
LIn
terv
ala
LO
DA
dd
itiv
eef
fect
b
Ph
eno
typ
eva
riat
ion
cC
hro
mo
som
eQ
TL
Inte
rval
aL
OD
Ad
dit
ive
effe
ctb
Ph
eno
typ
eva
riat
ion
c
LD
S1
QLds1
RM
237–
RM
403
2.4
�0.
25.
132
QLds2
RM
279–
RM
555
2.9
0.2
10.9
53
QLds3b
RM
520–
RM
293
4.7
�0.
314
.98
3QLds3a
RM
489–
RM
517
2.9
0.2
11.3
38
QLds8
RM
502–
RM
264
2.9
�0.
211
.05
3QLds3b
RM
520–
RM
293
4.2
�0.
316
.10
9QLds9
RM
434–
RM
257
2.8
0.2
6.98
DL
R1
QDlr1
RM
472–
RM
104
3.0
�0.
79.
058
QDlr8
RM
544–
RM
722.
4�
0.9
6.55
4QDlr4
RM
335–
RM
307
3.3
�0.
67.
119
QDlr9
RM
219–
RM
296
7.7
�1.
826
.36
9QDlr9
RM
219–
RM
296
3.6
�0.
79.
30
aU
nd
erli
ned
chro
mo
som
en
um
ber
and
mar
ker
inte
rval
sin
dic
ate
QT
Ld
etec
ted
inb
oth
year
s.bT
he
po
siti
veva
lues
ind
icat
eth
eal
lele
sfr
om
IRA
T10
9w
ith
incr
easi
ng
effe
cts.
cA
mo
un
to
fp
hen
oty
pe
vari
atio
n(%
)ex
pla
ined
by
the
QT
L.
1224 B. Yue et al.
TA
BL
E9
QT
Lfo
rro
ot
trai
tsu
nd
erco
ntr
oll
edco
nd
itio
ns
reso
lved
usi
ng
com
po
site
inte
rval
map
pin
gin
the
RIL
po
pu
lati
on
of
Zh
ensh
an9
7/
IRA
T1
09
Tra
its
2003
2004
Ch
rom
oso
me
QT
LIn
terv
ala
LO
DA
dd
itiv
eef
fect
b
Ph
eno
typ
eva
riat
ion
cC
hro
mo
som
eQ
TL
Inte
rval
aL
OD
Ad
dit
ive
effe
ctb
Ph
eno
typ
eva
riat
ion
c
MR
DC
2QMrdc2
MR
G27
62–R
M52
63.
02.
0511
.07
4QMrdc4b
RM
255–
RM
349
2.8
2.84
12.9
44
QMrdc4a
RM
307–
RM
471
4.7
�2.
0711
.16
9QMrdc9b
RM
160–
RM
215
2.6
2.53
10.9
65
QMrdc5
RM
421–
RM
274
3.0
1.74
7.70
9QMrdc9a
RM
316–
RM
219
4.5
�1.
889.
2011
QMrdc11
RM
287–
RM
229
5.6
�2.
3314
.04
RG
DC
7QRgdc7
RM
125–
MR
G44
497.
5�
0.04
17.4
93
QRgdc3
RM
473–
RM
487
2.9
�0.
0414
.80
11QRgdc11b
RM
287–
RM
229
3.6
�0.
038.
634
QRgdc4
RM
471–
RM
142
2.4
�0.
049.
7012
QRgdc12
RM
511–
MR
G43
412.
80.
026.
2411
QRgdc11a
RM
332–
RM
167
3.7
0.05
16.8
511
QRgdc11b
RM
287–
RM
229
3.3
�0.
0516
.40
DR
VC
1QDrvc1b
RM
23–R
M49
35.
6�
1.80
10.6
01
QDrvc1a
RM
428–
RM
490
2.5
1.74
16.3
52
QDrvc2a
RM
526–
RM
221
6.0
2.51
20.2
92
QDrvc2b
M26
2–M
RG
2762
2.7
1.54
12.1
34
QDrvc4a
RM
471–
RM
142
2.5
�1.
305.
434
QDrvc4a
RM
471–
RM
142
4.5
�1.
8818
.19
4QDrvc4b
RM
470–
RM
317
4.8
1.63
8.56
11QDrvc1a
RM
286–
RM
332
2.8
1.34
9.29
7QDrvc7
RM
134–
RM
248
2.8
1.23
4.93
9QDrvc9
RM
316–
RM
219
3.2
�1.
305.
5211
QDrvc11b
RM
287–
RM
229
2.4
�1.
204.
60
RV
C1
QRvc1
RM
472–
RM
104
2.6
8.83
5.30
3QRvc3
RM
231–
RM
489
8.6
�18
.28
31.4
73
QRvc3
RM
231–
RM
489
8.9
�17
.44
19.8
34
QRvc4
RM
349–
RM
131
5.0
14.9
622
.34
4QRvc4
RM
349–
RM
131
6.4
16.1
117
.70
6QRvc6
RM
527–
RM
564
2.7
9.72
9.07
7QRvc7
RM
125–
MR
G44
495.
614
.42
13.3
17
QRvc7
RM
125–
MR
G44
494.
012
.26
15.0
38
QRvc8
RM
404–
RM
339
4.9
12.3
19.
75
RG
VC
1QRgvc1
RM
472–
RM
104
2.5
0.09
5.99
3QRgvc3
RM
231–
RM
489
2.7
�0.
099.
703
QRgvc3
RM
231–
RM
489
6.6
�0.
1414
.13
7QRgvc7b
RM
134–
RM
248
2.9
0.09
10.5
34
QRgvc4
RM
349–
RM
131
5.9
0.16
17.1
711
QRgvc11
RM
202–
RM
287
3.5
�0.
1013
.71
7QRgvc7a
RM
125–
MR
G44
493.
90.
1210
.64
8QRgvc8
RM
72–R
M33
13.
60.
107.
0711
QRgvc11
RM
202–
RM
287
2.1
�0.
084.
87
aU
nd
erli
ned
chro
mo
som
en
um
ber
and
mar
ker
inte
rval
sin
dic
ate
QT
Ld
etec
ted
inb
oth
year
s.bT
he
po
siti
veva
lues
ind
icat
eth
eal
lele
sfr
om
IRA
T10
9w
ith
incr
easi
ng
effe
cts.
cA
mo
un
to
fp
hen
oty
pe
vari
atio
n(%
)ex
pla
ined
by
the
QT
L.
Genetic Basis of Drought Resistance in Rice 1225
TA
BL
E1
0
Th
em
ain
-eff
ect
QT
Lfo
rro
ot
trai
tsu
nd
erd
rou
ght
stre
ssan
dd
rou
ght-
ind
uce
dro
ot
gro
wth
reso
lved
usi
ng
com
po
site
inte
rval
map
pin
gin
the
RIL
po
pu
lati
on
of
Zh
ensh
an9
7/
IRA
T1
09
Tra
its
2003
2004
Ch
rom
oso
me
QT
LIn
terv
ala
LO
DA
dd
itiv
eef
fect
b
Ph
eno
typ
eva
riat
ion
cC
hro
mo
som
eQ
TL
Inte
rval
aL
OD
Ad
dit
ive
effe
ctb
Ph
eno
typ
eva
riat
ion
c
MR
DD
4QMrdd4
aR
M33
5–R
M30
73.
2�
1.77
7.01
4QMrdd4
cR
M34
9–R
M13
12.
61.
6211
.86
4QMrdd4
bR
M47
0–R
M30
36.
32.
5114
.63
9QMrdd9
RM
160–
RM
215
4.4
4.04
19.5
911
QMrdd1
1R
M28
6–R
M33
22.
81.
676.
19
RG
DD
3QRgdd3
RM
523–
RM
231
5.2
0.05
11.1
83
QRgdd3
RM
523–
RM
231
2.4
0.05
8.52
6QRgdd6
RM
454–
MR
G43
713.
40.
049.
237
QRgdd7
RM
125–
RM
4449
2.8
�0.
0612
.17
7QRgdd7
RM
125–
RM
4449
7.6
�0.
0616
.81
9QRgdd9
RM
219–
RM
296
2.4
�0.
059.
188
QRgdd8
RM
339–
RM
342
2.4
�0.
035.
69
DR
VD
2QDrvd2
RM
526–
RM
221
5.5
2.23
11.0
53
QDrvd3
RM
473–
RM
487
3.6
�3.
1519
.77
4QDrvd4
aR
M33
5–R
M30
73.
5�
1.77
7.28
11QDrvd1
1R
M28
6–R
M33
23.
22.
5313
.18
4QDrvd4
bR
M47
0–R
M30
36.
02.
2111
.26
9QDrvd9
RM
219–
RM
296
3.6
�1.
867.
9910
QDrvd1
0R
M59
6–R
M27
12.
4�
1.5
5.06
DIR
D2
QDird2
RM
29–R
M34
13.
0�
1.81
7.09
2QDird2
RM
29–R
M34
13.
5�
0.21
17.2
54
QDird4
bR
M47
1–R
M14
23.
61.
948.
894
QDird4
aR
M30
7–R
M47
13.
00.
1611
.92
DID
RV
1QDidrv1
RM
237–
RM
403
3.4
1.81
11.8
82
QDidrv2
RM
573–
RM
497
2.9
2.30
12.5
37
QDidrv7a
RM
125–
RM
4449
3.0
�1.
467.
227
QDidrv7b
RM
134–
RM
248
2.8
�1.
376.
66
RV
D1
QRvd1
RM
472–
RM
104
2.7
11.3
77.
433
QRvd3
RM
231–
RM
489
4.1
�15
.21
15.2
33
QRvd3
RM
231–
RM
489
8.3
�19
.00
19.7
24
QRvd4a
RM
470–
RM
303
3.5
13.2
812
.94
4QRvd4b
RM
349–
RM
131
6.1
16.4
015
.27
7QRvd7
RM
125–
RM
4449
3.8
15.9
315
.22
7QRvd7
RM
125–
RM
4449
5.9
15.6
112
.23
8QRvd8
RM
404–
RM
339
5.2
14.1
511
.13
RG
VD
1QRgvd1
RM
472–
RM
104
2.9
0.11
8.33
2QRgvd2
RM
145–
RM
324
4.1
�0.
1314
.09
3QRgvd3
RM
231–
RM
489
6.0
�0.
1514
.29
4QRgvd4
aR
M47
0–R
M30
34.
40.
1415
.97
4QRgvd4
bR
M34
9–R
M13
14.
40.
1312
.28
11QRgvd1
1R
M20
2–R
M28
72.
3�
0.11
9.89
7QRgvd7
RM
125–
RM
4449
2.7
0.10
6.36
8QRgvd8
RM
404–
RM
339
5.1
0.13
10.7
311
QRgvd1
1R
M20
2–R
M28
73.
5�
0.11
8.78
aU
nd
erli
ned
chro
mo
som
en
um
ber
and
mar
ker
inte
rval
sin
dic
ate
QT
Ld
etec
ted
inb
oth
year
s.bT
he
po
siti
veva
lues
ind
icat
eth
eal
lele
sfr
om
IRA
T10
9w
ith
incr
easi
ng
effe
cts.
cA
mo
un
to
fp
hen
oty
pe
vari
atio
n(%
)ex
pla
ined
by
the
QT
L.
1226 B. Yue et al.
OA-related candidate genes, BADH1 and BADH2, werelocated (Robin et al. 2003).
These positional correspondences were suggestive ofthe possible mechanisms underlying the QTL for DT-related traits identified in this study, including OA and/or cell membrane stability, as well as ABA response. Suchcorrespondence also enhanced the assertion that therelative performance of the fitness and productivitytraits used in this study largely resulted from the effectsof DT and provided further support for the plant-wisedrought stress protocol for investigating the geneticbasis of DT at the reproductive stage in rice with theeffect of DA eliminated.
Implications of the results in improvement of plantdrought resistance programs: The results may have im-portant implications in improvement of drought re-sistance in rice-breeding programs. The upland cultivarIRAT109 has higher values in all the important traitsof relative performance such as relative yield, relativespikelet fertility, relative biomass, relative grain weight,and relative harvest index. The QTL analysis alsoshowed that it contributed positively in most of theQTL that were consistently resolved for these traits. Inaddition, this cultivar also has high potential yield andbiomass comparable with many high-yielding cultivars(data not shown). Thus, IRAT109 may be a very goodgerm plasm resource for drought resistance breedingin rice and also for discovering genes for drought re-sistance. In this study, more than a half of the QTL eachexplained .10% of phenotypic variation for the DTandDA traits, which should be potentially useful for geneticimprovement of late season drought resistance in rice.In particular, QTL in the four regions related to DTand in the two regions related to DA that matched withQTL in previous reports may be exploited in breedingprograms.
Indicator traits for drought resistance can provideconvenient selection criteria for breeding programs. Inmaize the silk-tassel interval was identified as a highlyindicative secondary trait for drought-resistant breeding(Bolanos and Edmeades 1996). Spikelet fertility can bevisually estimated under field conditions and has beenused as an indirect index for drought screening in rice(Garrity and O’Toole 1994; Fukai et al. 1999). Theresults from this study also clearly showed that relativespikelet fertility or fertility under drought stress con-ditions is not only a highly informative indicator forseverity of drought stress, but also the most importantdeterminant of yield under drought stress conditions.In addition, a major QTL (QRsf9) governing this traitwas identified in this study together with closely linkedflanking markers on both sides. Thus, spikelet fertilityunder drought stress can be used as both an indicatorand a target in the selection processes using molecular-marker-assisted selection.
In summary, the results clearly indicated that thegenetic dissection of complex traits like drought re-
sistance could be achieved with special experimentaldesigns. In particular, the plant-wise drought treatmentprotocol as adopted in this study may provide a gen-erally useful method for independent evaluation ofthe individual components (such as DA, DE, and DT)contributing to drought resistance in rice as well as inother species. The genetic basis of each componentcould be characterized by further resolving the compo-nent into individual QTL that could be used either inbreeding programs by marker-assisted selection or asthe starting point for gene identification using variousapproaches.
We thank Abraham Blum and John O’Toole for their technicaladvice at various stages of this work. This research was supported bygrants from the National Program on the Development of BasicResearch, the National Special Key Project on Functional Genomicsand Biochips, the National Natural Science Foundation of China, andthe Rockefeller Foundation.
LITERATURE CITED
Ali, M. L., M. S. Pathan, J. Zhang, G. Bai, S. Sarkarung et al.,2000 Mapping QTLs for root traits in a recombinant inbredpopulation from two indica ecotypes in rice. Theor. Appl. Genet.101: 756–766.
Babu, R. C., B. D. Nguyen, V. Chamarerk, P. Shanmugasundaram,P. Chezhian et al., 2003 Genetic analysis of drought resistancein rice by molecular markers: association between secondarytraits and field performance. Crop Sci. 43: 1457–1469.
Blum, A., 1988 Drought Resistance: Plant Breeding for Stress Environ-ments. CRC Press, Boca Raton, FL.
Bolanos, J., and G. O. Edmeades, 1996 The importance of theanthesis-silking interval in breeding for drought tolerance intropical maize. Field Crops Res. 48: 65–80.
Boonjung, H., and S. Fukai, 2000 Effects of soil water deficit at dif-ferent growth stages on rice growth and yield under upland con-ditions. 2. Phenology, biomass production and yield. Field CropsRes. 43: 47–55.
Champoux, M. C., G. Wang, S. Sarkarung, D. J. Mackill, J. C.O’Tool et al., 1995 Locating genes associated with root mor-phology and drought avoidance in rice via linkage to molecularmarkers. Theor. Appl. Genet. 90: 756–766.
Chaves, M. M., and M. M. Oliveira, 2004 Mechanisms underlyingplant resilience to water deficits: prospects for water-saving agri-culture. J. Exp. Bot. 55: 2365–2384.
Courtois, B., G. McLaren, P. K. Sinha, K. Prasad, R. Yadav et al.,2000 Mapping QTLs associated with drought avoidance in up-land rice. Mol. Breed. 6: 55–66.
Cruz, R. T., and J. C. O’Toole, 1984 Dry land rice response to anirrigation gradient at flowering stage. Agron. J. 76: 178–183.
Fukai, S., and M. Cooper, 1995 Development of drought-resistantcultivars using physio-morphological traits in rice. Field CropsRes. 40: 67–87.
Fukai, S., G. Pantuwan, B. Jongdee and M. Cooper, 1999 Screen-ing for drought resistance in rainfed lowland rice. Field CropsRes. 64: 61–74.
Garrity, D. P., and J. C. O’Toole, 1994 Screening rice for droughtresistance at the reproductive phase. Field Crops Res. 39: 99–110.
Kamoshita, A., J. Wade, L. Ali, S. Pathan, J. Zhang et al., 2002 Map-ping QTLs for root morphology of a rice population adaptedto rainfed lowland conditions. Theor. Appl. Genet. 104: 880–893.
Lanceras, J. C., G. Pantuwan, B. Jongdee and T. Toojinda,2004 Quantitative trait loci associated with drought toleranceat reproductive stage in rice. Plant Physiol. 135: 384–399.
Lilley, J. M., M. Ludlow, S. McCouch and J. C. O’Toole, 1996 Lo-cating QTL for osmotic adjustment and dehydration tolerancein rice. J. Exp. Bot. 47: 1427–1436.
Genetic Basis of Drought Resistance in Rice 1227
Lincoln, S., M. Daly and E. Lander, 1992 Constructing geneticsmaps with MAPMAKER/EXP 3.0, Ed. 3. Whitehead InstituteTechnical Report, Whitehead Institute Cambridge, MA.
Luo, L. J., and Q. F. Zhang, 2001 The status and strategy on droughtresistance of rice (Oryza sativaL.). Chinese J. Rice Sci. 15: 209–214.
McCouch, S. R., L. Teytelman, Y. Xu, K. B. Lobos, K. Clare et al.,2002 Development and mapping of 2240 new SSR markers forrice (Oryza sativa L.). DNA Res. 9: 199–207.
Pantuwan, G., S. Fukai, M. Cooper, S. Rajatasereekul and J. C.O’Toole, 2002 Yield response of rice (Oryza sativa L.) geno-types to different types of drought under rainfed lowlands. 3.Plant factors contributing to drought resistance. Field CropsRes. 73: 181–200.
Price, A., and B. Courtois, 1999 Mapping QTLs associated withdrought resistance in rice: progress, problems and prospects.Plant Growth Regul. 29: 123–133.
Price, A., and A. D. Tomos, 1997 Genetic dissection of root growthin rice (Oryza sativa L.). II. Mapping quantitative trait loci usingmolecular markers. Theor. Appl. Genet. 95: 143–152.
Price, A., E. M. Young and A. D. Tomos, 1997 Quantitative trait lociassociated with stomatal conductance, leaf rolling and headingdate mapped in upland rice (Oryza sativa L.). New Phytol. 137:83–91.
Price, A. H., K. A. Steele, B. J. Moore and R. G. W. Jones,2002 Upland rice grown in soil-filled chambers and exposedto contrasting water deficit regimes. II. Mapping quantitative traitloci for root morphology and distribution. Field Crops Res. 76:25–43.
Quarrie, S. A., M. Gulli, C. Calestani, A. Steed and N. Marmiroli,1994 Location of a gene regulating drought-induced abscisicacid production on the long arm of chromosome 5 A of wheat.Theor. Appl. Genet. 89: 794–800.
Quarrie, S. A., D. A. Laurie, J. Zhu, C. Lebreton, A. Semikhodskii
et al., 1997 QTL analysis to study the association between leafsize and abscisic acid accumulation in droughted rice leavesand comparisons across cereals. Plant Mol. Biol. 35: 155–165.
Ray, J. D., L. Yu, S. R. McCouch, M. C. Champoux, G. Wang et al.,1996 Mapping quantitative trait loci associated with root pene-tration ability in rice (Oryza sativa L.). Theor. Appl. Genet. 92:627–636.
Robin, S., M. S. Pathan, B. Courtois, R. Lafitte, S. Carandang et al.,2003 Mapping osmotic adjustment in an advanced back-crossinbred population of rice. Theor. Appl. Genet. 107: 1288–1296.
Temnykh, S., W. D. Park, N. Ayres, S. Cartihour, N. Hauck et al.,2000 Mapping and genome organization of microsatellite se-quences in rice(Oryza sativaL.). Theor. Appl. Genet.100:697–712.
Temnykh, S., G. Declerck, A. Luashova, L. Lipovich, S. Cartinhour
et al.,2001 Computationalandexperimentalanalysisofmicrosat-ellites in rice (Oryza sativa L.): frequency, length variation, trans-poson associations, and genetic marker potential. Genome Res.11: 1441–1452.
Toorchi, M., H. E. Shashidhar, T. M. Gireesha and S. Hittalmani,2003 Performance of backcrosses involving transgressant dou-bled haploid lines in rice under contrasting moisture regimes:yield components and marker heterozygosity. Crop Sci. 43:1448–1456.
Tripathy, J. N., J. Zhang, S. Robin and H. T. Nguyen, 2000 QTLsfor cell-membrane stability mapped in rice (Oryza sativa L.) un-der drought stress. Theor. Appl. Genet. 100: 1197–1202.
Venuprasad, R., H. E. Shashidhar, S. Hittalmani and G. S.Hemamalini, 2002 Tagging quantitative trait loci associatedwith grain yield and root morphological traits in rice (Oryza sativaL.) under contrasting moisture regimes. Euphytica 128: 293–300.
Yadav, R., B. Courtois, N. Huang and G. McLaren, 1997 Mappinggenes controlling root morphology and root distribution in adoubled-haploid population of rice. Theor. Appl. Genet. 94:619–632.
Zeng, Z. B., 1994 Precision mapping of quantitative trait loci.Genetics 136: 1457–1468.
Zhang, J., H. T. Nguyen and A. Blum, 1999 Genetic analysis ofosmotic adjustment in crop plants. J. Exp. Bot. 50: 291–302.
Zhang, J., H. G. Zheng, A. Aarti, G. Pantuwan, T. T. Nguyen et al.,2001 Locating genomic regions associated with componentsof drought resistance in rice: comparative mapping within andacross species. Theor. Appl. Genet. 103: 19–29.
Zheng, B. S., L. Yang, W. P. Zhang, C. Z. Mao, Y. R. Wu et al.,2003 Mapping QTLs and candidate genes for rice root traitsunder different water-supply conditions and comparative anal-ysis across three populations. Theor. Appl. Genet. 107: 1505–1515.
Zheng, H. G., R. C. Babu, M. S. Pathan, L. Ali, N. Huang et al.,2000 Quantitative trait loci for root-penetration ability and rootthickness in rice: comparison of genetic backgrounds. Genome43: 53–61.
Communicating editor: A. H. Paterson
1228 B. Yue et al.