Correlation and QTL analyses of total chlorophyll content and photosynthetic

Correlation and QTL analyses of total chlorophyll content and photosynthetic rate of rice (Oryza sativa L.) under water stress and well-watered conditions Song-Ping Hu 1, 2*, Ying Zhou 1*, Lin Zhang 1, Li-Jun Luo 2, Lin Li 3, Xiu-Dong Zhu 3, Guo-Lan Liu 2, Qing-Ming Zhou 3** ( 1 College of Resource and Environmental Science, Jishou university, Hunan 416000, China; 2 Shanghai Agrobiological Gene Center, Shanghai 201106, China; 3 College of Agriculture, Agricultural University of Hunan, Changsha, 410128, China) Abstract To explore the relevant molecular genetic mechanisms of photosynthetic rate (PR) and chlorophyll content (CC) on rice, we conducted a series of relational experiments using a population of recombinant inbred lines (RILs) (Zhenshan97B IRAT109). We found a significant correlation between CC and PR (R=0.19**) in well-watered condition, but no significant correlation during water stress (r=0.08). We detected 13 main quantitative trait loci (QTLs) located on chromosomes 1, 2, 3, 4, 5, 6, and 10 that were associated with CC, including 6 QTLs located on chromosomes 1, 2, 3, 4, and 5 in water stress condition, and 7 QTLs located on chromosomes 2, 3, 4, 6, and 10 in well-watered condition. These QTLs explained 47.39% of phenotypic variation in water stress condition and 56.19% in well-watered condition. We detected four main QTLs associated with PR; three of them (qpr2, qpr10, qpr11) were located on chromosomes 2, 10, and 11 during water stress condition, and one (qpr10) was located on chromosome 10 in well-watered condition. These QTLs explained 34.37% and 18.41% of the phenotypic variation in water stress and well-watered conditions, respectively. In total, CC was largely controlled by main QTLs, and PR was mainly controlled by epistatic QTL pairs. Key words: photosynthetic rate; chlorophyll content; water stress; Quantitative trait loci; rice (Oryza sativa L.) *: Songping Hu and Ying Zhou contribute equally to this work. **: Corresponding author: sp6974@163.com 1

Introduction Chlorophyll is one of the most important molecules associated with photosynthesis in plant leaves. Chlorophyll content (CC) is used in rice breeding programs as an effective index of high photosynthetic efficiency (Kannangara 1991). However, it is difficult to select for CC in breeding programs because it is a quantitative trait (Wu and Luo 1996). In recent years, the rapid development of molecular technology has resulted in the identification of many QTLs, some of which are associated with CC in rice leaves. Using a F2 population of the rice varieties Palawan and IR42, Wu and Luo (1996) detected three main effect QTLs and one epistatic locus associated with CC. Yang et al. (2003) found five QTLs associated with CC, and Wang et al. (2003) reported six QTLs that influenced CC. All of these studies were carried out under well-watered condition, and their results were varying. Rice production is greatly influenced by drought stress, especially in the last phase of growth. Thus drought stress is one of the key factors that restricts rice yield (Dey and Upadhyaya 1996). Photosynthetic rate (PR) directly affects the biomass and yield of crops. In rice, higher leaf PR was reported to be correlated with greater biomass (Shen 1980; Ohno 1976; Chen et al. 1995) and grain yield (Xu and Shen, 1994). Ohno (1976) and Cao et al. (2001) also reported significant differences among PR in different rice varieties, and suggested that high yielding varieties could be developed by selecting varieties with high PR. Genetic studies in rice have shown that leaf PR shows a bimodal distribution in the F2 population, with a segregation ratio close to 3:1 (Hayashi et al. 1977; Nagamine 1991). The results of those studies suggested the existence of major genes controlling leaf PR. Furthermore, only a few studies on quantitative trait loci (QTLs) of PR have been reported, probably because strong influence of environmental factors jeopardise accurate measurement of this trait. Teng et al. (2004) carried out QTL analysis of photosynthetic traits such as PR and stomatal resistance using a double haploid (DH) population of rice, and found only two QTLs for PR. This low number of QTLs may have resulted from the small research population, or the complexity of the photosynthetic mechanism. To study some of the effects of drought stress on rice growth, we constructed special drought tolerance screening facility in Shanghai, China, and evaluated a population of recombinant inbred lines (RILs, F10) from a cross between upland japonica variety (IRAT109 with strong 2

drought tolerant ability) and lowland indica variety (Zhenshan97B with weak drought tolerant capacity). We investigated leaf CC and PR and carried out QTL analyses. These results provide valuable reference for breeding drought-tolerant or water-saving rice varieties with high photosynthetic efficiency. Results CC and PR in RIL population The CC of IRAT109 was higher than that of Zhenshan97B in both water stress and well-watered conditions (0.28 mg/dm 2 higher under water stress condition, and 0.24 mg/dm 2 higher in well-watered condition Table 1).The CC of IRAT109 and Zhenshan97B decreased with 5.44% and 6.57%, respectively, in water stress condition. The ANOVA showed that the mean CC of the RIL population did not vary significantly between both the treatments (Table 1). The frequency distribution of CC in the RIL population was a standard normal distribution both in well-watered and water stress conditions (Fig. 1). This distribution pattern indicated that CC is a typical quantitative trait Wu and Luo 1996. The CC values of most lines stood between the CC values of the parents, so CC was a suitable trait for QTL analysis. The PR of IRAT109 was higher than that of Zhenshan97B in both well-watered and water stress conditions. Compared with well-watered condition, the PR of Zhenshan97B in water stress condition was lower while the PR of IRAT109 was higher. Table 1. Phenotypic values of chlorophyll content and photosynthetic rate of parents and RILs population (CC: mg/dm 2, PR: mg CO 2 /dm 2 h) Traits IRAT109 Zhenshan97B RILs population Mean ± sd b) Mean ± sd Range Mean ± sd Peak Skew F value -ness WCC a) 4.96±0.10 4.72±0.18 3.83-5.52 4.61±0.32 0.05 0.15 0.45 SCC 4.69±0.13 4.41±0.10 3.90-5.69 4.62±0.32 0.13 0.33 WPR 22.20±1.72 20.77±3.87 11.50-33.13 20.43±.27 0.22 0.19 79.23** SPR 23.10±5.70 18.13±1.18 9.27-28.93 17.84±0.23-0.12 0.11 a) WCC, chlorophyll content in well-watered treatment; SCC, chlorophyll content in stress treatment; WPR, photosynthetic rate under well-watered treatment; SPR, photosynthetic rate under stress treatment. There are the same meanings while these letters or abbreviations are appearing on the subsequent tables or figures of this paper. b) sd, standard deviation. ** indicates 1% levels of significance. 3

The PR of population differed significantly between well-watered and water stress conditions. The RILs average PR was lower than that of the parents (Table 1), and its frequency distribution was a typical normal distribution, similar to that of CC (Fig. 1). This result indicated that PR was also a suitable trait for QTL analysis. No. of lines 50 Zhenshan97B IRAT109 50 No. of lines Zhenshan97B 40 40 30 30 IRAT109 20 20 10 10 0 3.70 3.95 4.20 4.45 4.70 4.95 5.20 5.45 5.70 5.95 SCC(mg/dm 2 ) 0 3.70 3.95 4.20 4.45 4.70 4.95 5.20 5.45 5.70 5.95 NCC (mg/dm 2 ) No. of lines 30 Zhenshan97B 50 No. of lines Zhenshan97B IRAT109 40 20 30 IRAT109 20 10 10 0 5 8 11 14 17 20 23 26 29 32 SPR (mgco 2 /dm 2.h) 0 9.0 11.5 14.0 16.5 19.0 21.5 24.0 26.5 29.0 31.5 34.0 NPR (mgco /dm2 2.h) Fig. 1. Frequency distribution of chlorophyll content and photosynthetic rate in RILs derived from a cross between Zhenshan97B and IRAT109 in different treatments (arrow indicates parent s value). Relationships between CC and PR 4

CC was more stable than PR in the two different conditions (water stress vs. well-watered). The correlation coefficient of leaf CC (r=0.77**) in water stress vs. well-watered conditions was larger than that of PR (r=0.35*), suggesting that PR suffered more than CC under water stress. There was a significant positive correlation (r=0.19**) between CC and PR of rice RILs leaves in well-watered condition, but no significant correlation under water stress (r=0.08) (Table 2). Table 2. Correlations between chlorophyll content and photosynthetic rate in rice leaves Traits WPR SCC WCC SPR 0.35** 0.08 0.11 WPR 0.29** 0.19** SCC 0.77** ** P<0.01. Main QTLs for CC and PR We detected a total of 13 QTLs related to CC. These QTLs were located on chromosomes 1, 2, 3, 4, 5, 6, and 10 (Table 3). The phenotypic variations explained by a single QTL varied from 3.54% to 14.53%. Seven QTLs were found in well-watered condition and accounted for 56.19% of total phenotypic variation. Alleles of four QTLs from Zhenshan97B (qcc3c, qcc3d, qcc6b, qcc10) showed an additive effect that varied from -0.08 to -0.11 mg/dm 2, while the alleles of another three QTLs from IRAT109 (qcc2b, qcc4b, qcc6a) had additive effects ranging from 0.05 to 0.07 mg/dm 2. Six QTLs were detected in water stress condition and explained 47.39% of the total phenotypic variation. Alleles of four QTLs from IRAT109 (qcc1, qcc2a, qcc4a, qcc5) had additive effects varying from 0.07 to 0.13 mg/dm 2, and alleles of another two QTLs from Zhenshan97B (qcc3a, qcc3b) had additive effects varying from -0.09 to -0.10 mg/dm 2. No additive environment interaction effects (AEi) of CC were detected in this study under two water conditions. 5

Table 3. QTLs for chlorophyll content and photosynthetic rate detected in RILs population (Zhenshan97B IRAT109) QTLs Treat- Chromo Marker intervals LOD h 2 (%) a) Additive b) P-value ments -somes value qcc1 S 1 RM443-RM297 4.56 6.46 0.08 <0.0001 qcc2a S 2 RM438-RM424 2.69 4.81 0.07 0.0004 qcc3a S 3 RM132-RM22 5.02 8.26-0.09 <0.0001 qcc3b S 3 RM520-RM571 5.43 9.35-0.10 <0.0001 qcc4a S 4 RM349-RM127 2.13 3.98 0.07 0.0017 qcc5 S 5 RM153-RM507 5.94 14.53 0.12 <0.0001 qcc2b W 2 RM424-RM561 2.31 3.54 0.05 0.0011 qcc3c W 3 RM22-RM231 7.92 13.95-0.10 <0.0001 qcc3d W 3 RM426-RM203 5.65 10.51-0.09 <0.0001 qcc4b W 4 RM255-RM349 2.85 5.31 0.06 0.0003 qcc6a W 6 RM549-RM539 4.00 6.39 0.07 <0.0001 qcc6b W 6 RM3-RM162 4.51 9.26-0.09 <0.0001 qcc10 W 10 RM271-RM269 3.02 7.23-0.08 0.0002 qpr2 S 2 RM262-RM263 2.84 6.20 0.66 0.0003 qpr10 S 10 RM596-RM271 4.39 13.18-0.96 <0.0001 qpr11 S 11 RM206-RM144 6.37 14.99 1.03 <0.0001 qpr10 W 10 RM596-RM271 5.38 18.41-1.17 <0.0001 a) h 2, contribution rates of single locus. b) Additive, additive effects were estimated as the substitution of IRAT109 allele by Zhenshan97B allele. There are the same meanings while these words or abbreviations are appearing on the subsequent tables of this paper. We detected four QTLs related to PR; three were detected in water stress condition (qpr2, qpr10, qpr11), and one in well-watered condition (qpr10). These four QTLs were located on chromosome 2, 10, and 11 (Table 3 and Fig. 2). qpr10 was detected in both conditions. Its additive effect in Zhenshan97B was -0.96 mg CO 2 /dm 2 h during water stress and -1.17mg CO 2 /dm 2 h in well-watered condition. In water stress condition, the additive effects of the other 6

CHROM.1 CHROM.2 CHROM.3 CHROM.4 Marker Marker Marker RM499 RM495 RM476A RM220 RM490 RM259 RM243 RM572 RM23 RM493 RM157B RM9 RM294B RM486 RM237 RM443 RM297 RM302 RM476B RM315 RM472 RM104 RM414 RM110 RM211 RM279 RM555 RM492 RM145 RM438 RM424 RM561 RM341 RM475 RM262 RM263 RM526 RM525 RM318 RM6 RM240 RM250 RM166 RM213 RM535 RM132 RM22 RM231 RM489 RM545 RM517 RM157A RM7 RM282 RM411 RM16 RM426 RM203 RM520 RM571 RM143 RM130 RM514 RM442 RM85 RM335 RM551 RM307 RM261 RM471 RM119 RM273 RM252 RM241 RM470 RM451 RM317 RM255 RM349 RM127 RM131 RM559 CHROM.5 CHROM.6 CHROM.10 CHROM.11 Marker RM122 RM153 RM507 RM13 RM548 RM593 RM592 RM574 RM169 RM289 RM509 RM430 RM164 RM163 RM459 RM161 RM421 RM274 RM480 Marker RM508 RM435 RM587 RM510 RM225 RM204 RM585 RM111 RM253 RM276 RM549 RM539 RM136 RM527 RM3 RM162 RM275 RM528 RM30 RM340 RM176 Marker RM222 RM216 RM311 RM467 RM596 RM271 RM269 RM258 RM294A RM228 RM591 RM333 RM4B RM20B RM167 RM441 RM120 RM536 RM287 RM209 RM21 RM206 RM144 RM224 Marker Fig.2. Chromosome locations of QTLs for chlorophyll content and photosynthetic rate detected in RILs (Zhenshan97B IRAT109) population in water stress or well-watered conditions. Note: QTLs of CC in water stress condition; QTLs of CC in well-watered condition QTLs of PR in water stress condition; QTLs of PR in well-watered condition two QTLs (qpr2 and qpr11) were contributed by the IRAT109 allele (PR increased by 0.66 and 1.03 mg CO 2 /dm 2 h, respectively). Three QTLs accounted for 34.37% of phenotypic variation of PR in water stress condition, but only 18.41% in well-watered condition. There was no AEi of PR detected in this study under two water conditions. 7

Digenic interaction We detected 16 pairs of epistatic loci associated with CC, which were located on all chromosomes except for chromosome 10 (Table 4). Of these, 12 pairs were detected in well-watered condition and 4 pairs in water stress condition. In well-watered condition, the total digenic interactions explained 38.49% of total phenotypic variation, and ranged from 1.87% to 4.26% for a single pair. In water stress condition, digenic interactions explained 18.57% of total phenotypic variation, and ranged from 4.11% to 5.04% for a single pair. Two main effect QTLs, qcc2a (RM438-RM424) and qcc4a (RM349-RM127), exhibited epistatic interactions in well-watered condition, while in water stress condition only one main QTL, qcc4b (RM255-RM349), was involved in epistasis. No epistasis (between QTLs i and j) environment interaction effects (AAEij) of CC were detected in this study under two water conditions. Table 4. Epistatic loci associated with rice leaves chlorophyll content detected in RILs population (Zhenshan97B IRAT109) Treat- Interval i a) Chromo- Interval j a) Chromo- LOD h 2 (%) b) AAij c) P-value ments somes somes value S RM476A-RM220 1 RM134-RM420 7 3.64 5.04-0.09 0.0001 S RM492-RM145 2 RM255-RM349 4 4.78 4.53-0.08 0.0003 S RM122-RM153 5 RM337-RM407 8 9.13 4.11-0.09 <0.0001 S RM585-RM111 6 RM179-RM277 12 3.83 4.89 0.09 0.0001 W RM486-RM237 1 RM349-RM127 4 5.32 2.77-0.06 0.0005 W RM297-RM302 1 RM442-RM85 3 4.78 2.46 0.07 <0.0001 W RM476B-RM315 1 RM477-RM264 8 3.70 3.18 0.07 0.0001 W RM438-RM424 2 RM13-RM548 5 2.45 2.31 0.06 0.0013 W RM525-RM318 2 RM261-RM471 4 4.77 2.96-0.06 0.0002 W RM514-RM442 3 RM507-RM13 5 2.30 4.26-0.06 0.0024 W RM442-RM85 3 RM459-RM161 5 4.21 4.26 0.08 <0.0001 W RM261-RM471 4 RM536-RM287 11 4.44 3.29 0.08 <0.0001 W RM471-RM119 4 RM19-RM453 12 4.29 3.42 0.07 <0.0001 W RM274-RM480 5 RM245-RM205 9 5.37 1.87-0.06 0.0003 W RM455-RM351 7 RM463-RM235 12 4.17 3.59 0.07 0.0010 W RM515-RM284 8 RN287-RM209 11 5.37 4.12-0.09 <0.0001 a) Interval i or j refers to the interval of chromosomes lodged locus i or j. b) h 2 represents the contribution rate of epistatic loci. c) AAij is the effects of epistasis (between loci i and j). There are the same meanings while these letters or abbreviations are appearing on the subsequent tables of this paper. 8

The 16 pairs of epistatic loci formed a complex genetic network. The contribution rate and effects of these epistatic loci were small, indicating that CC has a complicated molecular genetic basis. Table5 Epistatic loci associated with rice leaves' photosynthetic rate detected in RILs population (Zhenshan97B IRAT109) Treat- Interval i Chromo- Interval j Chromo- LOD ments somes somes value h 2 (%) AAij P-value 6 50$50 50%50 $ $ : 5050 5050 : 5050 5050 9

: 5050 5050 : 5050 5050 : 5050 5050$ : 5050 5050 : 5050 5050 : 5050 5050 : 5050 5050 : 5050 50%50% : 5050 5050 : 5050 5050 : 5050 5050 Twenty-eight pairs of epistatic loci associated with PR were mapped onto 12 chromosomes (Table 5). Of these, 13 pairs were found in well-watered condition and 15 pairs in water stress condition. In well-watered condition, digenic interactions explained 54.71% of phenotypic variation, ranging from 1.29% to 7.54% for a single pair. In water stress condition, digenic interactions explained 46.8% of phenotypic variation, ranging from 0.98% to 4.74% for a single pair. However, none of the main effect QTLs for PR was associated with pairs of epistatic loci in either water stress or well-watered condition; videlicet, the epistatic pairs of QTLs for PR in this study occurred among those intervals of chromosomes without main effect. The 28 pairs of epistatic loci formed a complex genetic network, in which chromosomes 1, 6, and 7 had particularly important roles as they contained 6 8 epistatic pairs. We detected no AAEij of PR in this study, regardless of the water conditions. Discussion There have been many studies on chlorophyll in rice, which focused mainly on the components of 10

chlorophyll, anabolic approach, and its location and functions within the cell (Jiang et al. 1994, Alberte et al. 1977, Xu et al. 2000) etc. The rapid development of molecular marker technology in recent years has established some molecular genetic basis underlying CC and PR (Wu and Luo, 1996; Yang et al. 2003; Wang et al. 2003; Teng et al. 2004; Xu et al. 2000). However, little is known about the relationships between CC and PR and their QTLs during water stress. In this study, we analyzed QTLs affecting CC and PR in a rice RILs population grown under water stress or well-watered conditions. CC and PR under well-watered and water stress conditions Jiang et al. (1994) reported that chlorophyll in rice seedlings degraded rapidly with increasing duration and intensity of water stress. This was attributed to the increasing levels of active oxygen, hydrogen peroxide (H 2 O 2 ) and malondialdehyde (MDA), and the decreasing levels of antioxidants including ascorbic acid (AsA), glutathione (GSH), and carotenoid (CAR). These changes in cellular make-up can damage chlorophyll-protein complexes. Alberte et al. (1977) also suggested that decreasing CC during water stress may result from inhibition of the light-capturing chlorophyll a/b-proline complex within the chloroplast. Also, reduction in the levels of reactive oxygen scavengers such as deoxidized GSH, а-tocopherol, and mannitol could prevent recycling of chlorophyll, while the increased levels of active oxygen and MDA could accelerate chlorophyll degradation. In this study, the CC of two parents (IRAT109 and Zhenshan97B) decreased 5.44% and 6.57%, respectively, in water stress condition. The decrease in CC of upland rice IRAT109 was about 1.23% less than that of paddy rice Zhenshan97B. Upland rice showed higher drought tolerance, higher yield, and higher photosynthetic rate under water stress condition. However, the changes of CC differed within the RILs population. Among the offspring of RILs population (Zhenshan97B IRAT109), 45.00% of them by statistic calculating retained a stable CC in water stress condition, or even showed increased CC (See the SCC of Fig.1 and Table 1). In addition, some lines grew more slowly, had smaller leaves, or developed thicker color on leaves during water stress. These growth characteristics may be another reason underlying the high CC in water stress condition. In general, the PR of crops decreases in stress conditions such as drought, high temperature and heat harm etc. (Xu 1990; Wang 1995). Cao et al. (2001) reported that the PR of japonica rice was 11

higher than that of indica rice. The results of our study agreed with those findings (Table 1). In well-watered condition, the PR of japonica IRAT109 was 6.88% higher than that of indica Zhenshan97B, while in water stress condition the PR of japonica IRAT109 was 27.41% higher than that of indica Zhenshan97B. The PR of the different varieties differed significantly between the two water conditions. In water stress condition, the PR of the drought-tolerant variety IRAT109 was 4.10% higher than that in well-watered condition, whereas the PR of the drought-sensitive variety Zhenshan97B was 12.71% lower in water stress condition than that in well-watered condition. On average, the PR of RIL populations decreased 12.68% in water stress condition. Why did the PR of IRAT109 increase during water stress condition? It may be that the upland rice variety IRAT109 is very well adapted to drought-prone environments, and thus shows increased PR in dry condition. In addition, we have known from the former results that there was a significant positive correlation (r=0.19**) between CC and PR of RILs under well-watered condition, but no correlation under water stress (r=0.08) (Table 2).This result suggested that high CC enhanced PR in well-watered condition. PR was more sensitive to water stress than CC. Relationships and applications of main effect QTLs and epistatic QTL pairs for CC and PR Recently, various rice genetic populations have been used to map CC genes (QTLs) under various conditions. Wu and Luo (1996) found several QTLs associated with leaf CC during nitrogen stress. Yang et al. (2003) reported five QTLs for CC on chromosomes 1, 2, 3 and 10, and two epistatic pairs on chromosomes 2, 6, and 12. Wang et al. (2003) detected six QTLs and five epistatic pairs affecting chlorophyll a and b content. All these QTLs were different. In the present study, we identified 13 main effect QTLs and 16 pairs of epistatic loci located on all 12 chromosomes. Comparing our results to those previously reported, the QTL qcc1 (RM443-RM297) located on chromosome 1 corresponds to qchla1d (Xpsrpsr72a-Xpsr754) and qchlb1c (Xpsrpsr72a-Xpsr754) reported by Wang et al. (2003); qcc2a (RM438-RM424) and qcc2b (RM424-RM561) located on chromosome 2 correspond to R258/RG102 reported by Wu and Luo (1996) and qchl2b (R26-C499) reported by Yang et al. (2003). Similarly, qcc4a (RM349-RM127) and qcc4b (RM255-RM349) located on chromosome 4 are very similar to qchla4b (Xpsr901-Y8026L) and qchlb4b (Xpsr901-Y8026L) reported by Wang et al. (2003). In this study, several QTLs were mapped on adjacent chromosomal locations in water stress and well-watered conditions (qcc2a, qcc2b; qcc3a, qcc3c; and qcc4a, qcc4b). QTLs 12

mapped onto chromosomes 1 and 4 were also quite similar to those reported by Wang et al. (2003). These suggest that chromosomes 1, 2, 3, and 4 play key roles in the genetic control of CC. Our results suggested the main effect QTLs made the major contribution to controlling CC, and the epistatic QTL pairs took a lesser effect. The main effect QTLs contributed 56.19% to CC in well-watered condition, and 47.39% under water stress condition. The epistatic pairs contributed 38.49% to CC in well-watered condition and 18.57% in water stress condition. However, we did not detect any QTL-by-environment interactions (AE) or any epistasis-by-environment interactions (AAE) for CC, regardless of whether plants were in water stress or well-watered conditions. It only explained that Q E interactions could not be detected in the present setup even when best contrasting environments were provided. Because the locations of CC QTLs on the chromosomes were the same in different populations and environments, their precise location should be investigated. This will enable marker assisted breeding programs to improve photosynthetic traits, and therefore serve directly on agricultural production. In addition, it may be helpful to get better results by separating chlorophyll a and b on fine locating in future. Although there have been several studies on the basic genetics of rice PR (Hayashi et al.1977; Liu and Liu 1984; Nagamine 1991; Wang 1995), there is only one report on locating PR s QTLs because of the difficulties with measuring photosynthetic rate (Teng et al. 2004). However, because Teng and coworkers used different populations and different marker methods, their results differed to those in our study. They located PR QTLs on chromosomes 4 and 6, whereas we located PR QTLs on chromosomes 2, 10, and 11. They only reported the mapping results of main effect QTLs to PR in a DH population of rice, but we analyzed the detailed circs of a RILs population of rice, and included analyses of main effect QTLs, epistatic QTL pairs, and environment interactions under well-watered or water stress conditions. Our results showed that epistasis mainly controls PR in rice, while main effect QTLs has a lesser effect. Under water stress condition, epistasis contributed 46.80% to PR while main effect QTLs contributed 34.36% (Tables 3 and 5). In well-watered condition, epistasis contributed 54.71% and main effect QTLs contributed 18.41%. Moreover, under water stress condition, the number of main QTLs increased from 1 to 3, and epistatic QTL pairs increased from 13 to 15. In water stress condition the main QTLs contribution increased from 18.41% to 34.36%, while the contribution of epistasis 13

decreased from 54.71% to 46.80%. This phenomenon is difficult to explain, because in general PR decreases during stress conditions such as drought or high temperature (Xu 1990; Chen et al. 1995). However, considering the CC results, the performances of PR s QTLs and epistasis during water stress showed a similar pattern to that of the QTLs and epistasis controlling CC under the same condition. The PR of rice is sensitive to environmental factors (Xu 1990; Wang 1995), so we expected to find Q E interactions (AE and AAE). However, no significant Q E interactions (AE or AAE) were detected in the well-watered and water stress conditions. We couldn t provisionally find a reasonable elucidation to this kind of phenomenon, but the same results were also observed in our previous studies (Zou et al. 2005; Liu et al. 2008), perhaps there were some inevitability existing in them, we would explore them from the types of populations, experimental facilities and measuring apparatuses etc. Although there were fewer PR QTLs than CC QTLs mapped in this study, one PR QTL, qpr10 (RM596-RM271), was detected in both well-watered and water stress conditions, and showed high LOD values and contribution values (4.39 and 13.18, respectively, in water stress condition; and 5.38 and 18.41, respectively, in well-watered condition). The strong reliability of this QTL makes it a suitable candidate for use in map-based cloning and/or MAS breeding programs. At the same time, none of the PR QTLs overlapped with those QTLs controlling grain yield (Zou et al. 2005). This result agreed with the findings of Ishimaru et al. (2001), and confirms that PR does not influence grain yield of crops (Teng et al. 2004). High photosynthetic efficiency is the main target of rice breeding in the 21 st century (Wang 1995; Peng 2000). Identifying the locations of genes or groups of genes controlling CC and PR in well-watered and water stress conditions is useful to further understand the molecular genetic mechanisms of photosynthesis. These of information can be practically applied in breeding programs that utilize MAS or mapping. Materials and Methods Plant materials A set of 195 F 10 RILs of rice were developed from a cross between paddy rice Zhenshan97B (drought sensitive) and upland rice IRAT109 (drought tolerant) and their parents. Zhenshan97B was an improved indica rice variety adapted to the lowland ecosystem in China, while IRAT109 14

was an upland japonica rice variety originally developed in WARDA (Africa rice center). Each RIL was planted in the drought tolerance (DT) screening facility with a total of 6 rows and 90 hills. When the wet season (May 10-24) is coming, the seeds of RILs were directly sowed into soil (sandy loam) with 15 hills in each row and a space of 20.00 cm between rows and 18.00 cm between hills. Materials were divided into three groups for sowing date with an interval of one week to synchronize flowering time, i.e. on May 10, 17 and 24. The groups in the population and lines in each group were arranged randomly with three replications in the DT Screening Facility in Shanghai, China. Experimental facilities and water stress treatments The DT Screening Facility is oriented in a north-south direction. It has an accurate water control system and a rain-proof roof that can be opened on sunny days. The field within the facility was divided into two plots, each 76.00 m long and 6.00 m wide. The plots were irrigated by overhead sprinklers, and by drip irrigation installed in a north-south direction along the middle of each plot. The plots were surrounded by a 2.00 m deep canal; Holes were designed on a north-south oriented wall of the canal at the same intervals so that the soil water could be leaked into the canal. The time domain reflectometry (TDR) system was installed on the island within the two plots, and was used to measure soil water content (SWC) at different soil depths. The criteria of well-watered in this experiment was 100.00%of SWC at 25.00 cm depth; and the criteria of drought stress were 24.00%,30.00% and 37.00% of SWC at 25.00 cm, 50.00 cm and 75.00 cm depth, respectively. The illuminometer probe and the temperature and humidity transducer were set in different positions within the DT facility to record climatic data such as illumination, temperature, and humidity (Fig. 3). Surface irrigation was applied during the vegetative growth period. Additional water was provided via a sprinkler system to provide plants with well-watered condition. Sprinkler irrigation was stopped when 60.00% of the rice lines reached the early stage of panicle initiation, and then the water level in the canal was decreased from the normal ground water level to 1.80 m below the soil surface. Consequently, soil water leaked into the canal through the holes on the canal wall. Because the rates of water loss differed between the middle and the edge of plots, a water gradient was formed. To enhance the soil water gradient, drip irrigation was applied every day for several hours to supply additional water during the water stress duration. As a result, plants growing at the 15

edge of the plots close to the canal were subjected to water stress condition, while those growing in the center of the plot close to drip irrigation were in well-watered condition (Fig. 3). E N W S Drain Channel TDR tube Illuminometer probe Drip irrigation 6m 6m Temperature & humidity transducers Gradient of soil water content 76 m Fig.3. Configuration of drought tolerance facility Sprinkler irrigation was stopped on July 22 (at the early stage of panicle initiation) and water stress condition was allowed to develop. Drip irrigation supplied approximately 4 tons water per day to an area of 900.00 m 2. As soon as the sprinkler irrigation was stopped (July 22), the drip system enhanced the soil moisture gradient, allowing it to last for a longer period. Full irrigation was resumed on September 8 (at the filling period, i.e. duration of imposed drought stress was 46 days) after severe leaf rolling was observed in the sensitive lines. All other agronomic treatments, e.g. fertilizer and pesticide application, were as per common practice in rice cultivation. Phenotyping and genotyping Susceptible lines showed leaf rolling 2 weeks after water stress condition was established (when rice plants were at the mid- or middle-late stage of panicle initiation). At that time, CC and PR 16

were measured simultaneously. A chlorophyll meter (SPAD-502, Minolta, Japan) was used to measure chlorophyll content of the reciprocal second leaf on the primary stalk at 9:00 12:00 am and 14:00 16:30 pm. Within each genotype, three leaves were measured, and each leaf was measured at three points; upper, middle, and lower. CC was calculated from SPAD values as described by Wu and Luo (1996): Y=0.0996X-0.152 (X=SPAD value, Y=CC mg/dm 2 ). The PR of rice population s leaves was measured using a China Agricultural University (CAU) photosynthesis measuring system (Beijing, China). We measured three leaves within each line of every treatment, at 9:00 12:00 am and 14:00 16:00 pm. Each trait was measured over 1 or 2 days. A total of 213 microsatellite markers were used to genotype the population. A linkage map was constructed using MapMaker/Exp V3.0 (Lincoln et al.1992) to span 1825.00 cm of genome with an average distance of 8.6 0cM between adjacent markers. Data analysis ANOVA and phenotypic correlation analysis were performed using a fixed effect model on S-Plus for Windows V6.1 (Insightful Corporation 2001). The mixed-model QTL analysis was conducted by QTLMapper V1.6 (Wang et al. 1999). Means of three replications were used as input data. Data collected from plants under water stress and well-watered conditions were analyzed separately during QTL mapping. A threshold of P 0.005 was designated as a significant main effect QTL (M-QTL), digenic epitasis (E-QTL), and Q E interaction. Contribution rate (h 2 ) was estimated as percentage of variance explained by each locus or epistatic pair in proportion to total phenotypic variance. M-QTLs were named following established nomenclature (McCouch et al. 1997) but using alphabetic order for QTLs on the same chromosome. Acknowledgements This study was jointly supported by grants NSFC (The International Cooperation Key Program 30040025) and Chinese Ministry of Science and Technology (2003AA207010 and 2004B17200), the Rockefeller Foundation, the key program of nature science foundation of Hunan province (090XHN) and the personnel foundation of Jishou University. 17

References Alberte RS, Thornber JP, Fiscus EL 1997.Water stress effects on the content and organization of chlorophyll in mesophyll and bundle sheath chloroplasts of maize. Plant Physiol.59,351-353. Cao SQ, Zhai HQ, Yang TN, Zhang RX, Kuang TY(2001).Studies on photosynthetic rate and function duration of rice germplasm. Chinese J Rice Sci 15(1), 29-34. Chen WF, Xu ZJ, Zhang BL(1995). Physiological bases of super high yield breeding in rice. Liao Ning science and technology publishing company, Shenyang, China. Dey MM, Upadhyaya HK (1996). Yield loss duo to water stress, cold and submergence in Asia. In: Evenson RE, Herdt RW, Hossain M(eds) Rice research in Asia: progress and priorities. CAB International, Wallingford, 291-303. Hayashi K, Yamanoto T, Nakagahra M (1977). Genetic control for leaf photosynthesis in rice, Oryza sativa L. J Japan Breed 27, 49-56. Insightful Corporation (2001). S-plus 6 for windows, User s guide. Seattle, WA, USA. Jiang MY, Yang WY, Xu J, Chen QY (1994). Active oxygen damage effect of chlorophyll degradation in rice seedlings under osmotic stress. Acta bot. si.36 (4),289-295. Kannangara CG (1991). In: Bogorad L,Vasil I K (eds.).the Photosynthetic Apparatus, Academic Press Inc. California, 302-321. Ishimaru K, Yano M, Aoki N, Ono K, Hirose T, Lin SY et al.(2001). Toward the mapping of physiological and agronomic characters on a rice function map: QTL analysis and comparison between QTLs and expressed sequence tags. Theor Appl Genet 102,793-800. Lincoln SE, Daly MJ,Lander E (1992). Constructing genetic maps with MapMaker/EXP3.0. Whitehead Institute Technical report, 3 rd edn. Whitehead Institute, Cambridge. Liu GL, Mei HW, Yu XQ, Zou GH, Liu HY, Hu SP et al.(2008). QTL analysis of panicle neck diameter, a trait highly correlated with panicle size, under well-watered and drought conditions in rice (Oryza sativa L.). Plant Science 174, 71-77. Liu ZY, Liu ZQ(1984). Genetics and breeding study on photosynthesis. Guizhou people s publishing company, Guiyang, China. McCouch SR, Cho YG, Yano M, Paul E, Blinstrub M (1997). Report on QTL nomenclature. Rice Genet Newslett 14,11-13. 18

Nagamine T(1991).Genetic analysis of photosynthetic capacity of single leaf analyzed by oxygen polarography in rice, Oryza sativa L. J Japan Breed 41, 301-307. Ohno Y(1976). Varietal differences of photosynthetic efficiency and dry matter production in indica rice. Tro Agri 53,115-123. Peng S (2000). Single-leaf and canopy photosynthesis of rice. In: Sheehy JE, Mitchell PL, Hardy B(Eds.), Redesigning rice photosynthesis to increase yield, IRRI, Los Bannos, Philipines, 213-228. Shen YG (1980). Photosynthesis and matter production. China agriculture publishing company, Beijing, China, 31-32. Teng S, Qian Q, Zeng DL, Kunihiro Y, Fujimoto K, Huang DN et al.(2004). QTL analysis of leaf photosynthetic rate and related physiological traits in rice(oryza sativa L.). Euphytica 135, 1-7. Wang B, Lan T, Wu WR, Li WM (2003). Mapping of QTLs controlling chlorophyll content in rice. Acta Genet. Si. 30(12),1127-1132. Wang DL, Zhu J, Li ZK, Paterson AH (1999). Mapping QTLs with epistatic effects and QTL environment interactions by mixed linear model approaches, Theor Appl Genet 99,1255-1264. Wang YR(1995).Physiological breeding in rice. Beijing science and technology literature publishing company, Beijing, China, 23-44. Wu P, Luo AC (1996). Investigation on genetic background of leaf chlorophyll content variation in rice under nitrogen stressed condition via molecular markers. Acta Genet. Si. 23(6),431-438. Xu DQ (1990). Ecology, physiology and biochemistry of midday depression of photosynthesis. Plant Physiol Commun 26(6),5-10. Xu DQ, Shen YG(1994). Progress on physiology of crop high production and high efficiency. Science publishing company,beijing,china,17-23. Xu W, Subudhi P K, Crasta O R, Rosenow DT, Mullet JE, Nguyen HT (2000). Molecular mapping of QTLs conferring stay-green in grain sorghum (Sorghum bicolor L.Moench). Genome, 43(3):461-469. Yang QH, Lu W, Hu ML, Wang CM, Zhang YX, Yano M et al.(2003) QTL and epistatic interaction underlying leaf chlorophyll and H 2 O 2 content variation in rice (Oryza sativa L.) Acta Genet. Si. 30 (3),245-250. 19

Zou GH, Mei HW, Liu HY, Liu GL, Hu SP, Yu XQ et al.(2005), Grain yield responses to moisture regimes in a rice population: association among traits and genetic markers. Theor Appl Genet 112,106-113. 20