Multienvironment Quantitative Trait Loci Mapping and ...

SYMPOSIA

Multienvironment Quantitative Trait Loci Mapping and Consistency across Environments of Resistance Mechanisms to Ferrous Iron Toxicity in Rice I. Dufey, M.-P. Hiel, P. Hakizimana, X. Draye, S. Lutts, B. Koné, K.N. Dramé, K.A. Konaté, M. Sie, and P. Bertin*

ABSTRACT Lowland rice (Oryza sativa L.) is often affected by iron toxicity, which may lead to yield losses. One important constraint in the study of the inheritance of resistance strategies to this stress is the inconsistency of gene expression across different environments. This study aimed to determine the stability of quantitative trait loci (QTL) across several environments. Quantitative trait loci mapping for traits related to resistance mechanisms had been previously performed using 164 recombinant inbred lines derived from ‘Azucena’ and ‘IR64’ screened in hydroponics in a phytotron. In the present study, this population was tested under excessive ferrous iron conditions in three additional environments: in hydroponics in a greenhouse, on washed sand, and in the field. Altogether, 44 putative QTL were identified in the four single QTL analyses for morphological, physiological, and agronomic traits. From these 44 QTL, 20 were found in overlapping regions for the same or related traits in different environments, identifying six genomic regions of great interest for the inheritance of resistance to iron toxicity. Quantitative trait loci stability across environments was also checked by performing a joint QTL analysis, which confirmed the position of nine QTL previously found in the same or adjacent regions by at least one single analysis. Combining both single and joint analyses helps in separating QTL specific to a particular environment from generally expressed ones thus is more suitable for marker-assisted selection.

I. Dufey, M.-P. Hiel, P. Hakizimana, X. Draye, S. Lutts, and P. Bertin, Earth and Life Institute, Université catholique de Louvain, Croix du Sud, 2 bte 11, B-1348 Louvain-La-Neuve, Belgium; B. Koné, K.N. Dramé, and M. Sié, Africa Rice Center 01 B.P., 2031 Cotonou, Benin; K.A. Konaté, Institut national de l’environnement et de recherches agricoles (INERA), 01 B.P., 476 Ouagadougou, Burkina Faso. Received 23 Sept. 2010. *Corresponding author ([email protected]). Abbreviations: AfricaRice, Africa Rice Center; Fv:Fm, photosystem II maximum quantum efficiency; INERA, Institut National de l’Environnement et de Recherches agricoles; LBI, leaf bronzing index; LOD, logarithm of odds; QTL, quantitative trait locus/loci; RIL, recombinant inbred line; UCL, Université catholique de Louvain.

R

ice (Oryza sativa L.) grown on flooded acid soils is often subjected to iron toxicity. Under anaerobic conditions, iron is reduced by microorganisms into its ferrous form, which is soluble in water and readily taken up by plants (Gross et al., 2003). In West Africa, where 43% of the rice fields are located in lowlands (Defoer et al., 2004), this nutrient disorder is considered a major constraint that severely limits yield (Audebert and Fofana, 2009; Cherif et al., 2009). When available at high concentrations in the soil, it may lead to nutrient imbalance by limiting the absorption of other nutrients (Genon et al., 1994; Sahrawat, 2004; de Dorlodot et al., 2005). Moreover, in plant tissues excessive amount of ferrous iron may catalyze the synthesis of hydroxyl radicals, causing severe damage to the membrane lipids, proteins, and nucleic acids (Bode et al., 1995). Typical iron-toxicity symptoms are leaf “bronzing,” formation of a root iron plaque, reduced plant development, and yield loss (Ponnamperuma et al., 1955; Bode et al., 1995). To adapt to this constraint, plants have developed several strategies: oxidation of Fe2+ in the rhizosphere, exclusion or Published in Crop Sci. 52:539–550 (2012). doi: 10.2135/cropsci2009.09.0544 © Crop Science Society of America | 5585 Guilford Rd., Madison, WI 53711 USA All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.

CROP SCIENCE, VOL. 52, MARCH– APRIL 2012

539

retention at the root level, storage in the apoplasm and vacuole, adsorption by ferritin, and detoxification of the active oxygen species by enzymes (Becker and Asch, 2005; Majerus et al., 2007b, 2009). Selecting resistant genotypes is a promising way to improve yield under iron-toxic conditions (Becker and Asch, 2005; Audebert, 2006). Because resistance to iron toxicity is a complex trait, controlled by several genes, quantitative trait loci (QTL) mapping combined with marker-assisted selection appears to be a viable approach for selecting against iron toxicity (Mackill et al., 1999). Several QTL-mapping exercises for morphological and physiological traits involved in iron-toxicity-resistance mechanisms have been conducted (Wu et al., 1997, 1998; Wan et al., 2003a, b, 2004, 2005; Shimizu et al., 2004; Shimizu, 2009; Dufey et al., 2009). However, to the best of our knowledge, no QTL mapping has been performed for yield components under iron-toxic conditions. Moreover, results of a QTL analysis depend on the environmental conditions of the experiment (Jansen et al., 1995; Lu et al., 1996; Hittalmani et al., 2003; Li et al., 2003; Senthilvel et al., 2008). According to Lander and Kruglyak (1995), a QTL should be considered reliable only when it has been identified in at least two independent experiments. The objective of this study was to determine the reliability of QTL identified in a previous study performed under highly controlled conditions (Dufey et al., 2009) and to identify new putative QTL, inclusive of yield-component parameters. Therefore, the segregating population used in the study of Dufey et al. (2009) was tested under excessive ferrous iron conditions in three additional environments whose level of control differed widely. The results of the four experiments constituting a large panel from highly controlled (hydroponics) to field conditions were analyzed in this paper.

MATERIALS AND METHODS Phenotypic Characterization of the Population A segregating population of 164 recombinant inbred lines (RILs) of F7:8 generation—derived from a cross between ‘Azucena’ (O. sativa L. subsp. japonica), which is moderately resistant to ferrous iron toxicity, and ‘IR64’ (O. sativa L. subsp. indica), which is sensitive to ferrous iron toxicity (Wu et al., 1997)—was used for the QTL analyses. This population was obtained by single-seed descent method at Institut de recherche pour le développement (IRD) in Montpellier, France (Ahmadi et al., 2005). The 164 RILs and the parents had been tested in a previous study (Dufey et al., 2009) under fully controlled conditions (hydroponics) in a phytotron at Université catholique de Louvain (UCL) in Louvain-la-Neuve, Belgium, for their response to excessive ferrous iron in the culture medium (Exp. 1). Morphological parameters had been quantified on all 164 RILs and both parents whereas physiological parameters, whose measurement is more complex, had been determined on extreme RILs only. These extreme lines had been selected on the basis of their leaf bronzing

540

index (LBI) (IRRI, 1996), considered as a good indicator of sensitivity to iron toxicity (Audebert and Sahrawat, 2000). In the present study, physiological parameters were measured for the same reason on both parents and 40 contrasting RILs—the 20 most resistant of Fe (II) and 20 most sensitive, selected on the basis of bronzing score—under control and Fe-treated conditions in hydroponics in a greenhouse at UCL (Exp. 2) and on washed sandy soil in a greenhouse at Africa Rice Center (AfricaRice) in Cotonou, Benin (Exp. 3). The entire population was also tested under iron-toxic conditions at Institut national de l’environnement et de recherches agricoles (INERA) in the Kou Valley in Burkina Faso and agronomic parameters were determined (Exp. 4).

Overview of Experiment 1: Hydroponics in a Phytotron at Université catholique de Louvain The experimental conditions of this previously published experiment (Dufey et al., 2009) are presented below to allow a thorough comparison with the present experiments. Thirteen days after germination, six seedlings of each of the 164 lines and 12 seedlings of each parent were planted in perforated polystyrene plates floating on a standard rice nutrient solution (Yoshida et al., 1976) in 26-L plastic tanks in a growth chamber at UCL (30/25°C day/night, 85–95% relative humidity, and 12 h photoperiod with 360 μmol m–2 s–1 light intensity). The pH of the solution was adjusted daily to 4.5 and solution was replaced once a week (Colmant and Bertin, 2004; Dufey et al., 2009). The ferrous iron treatment, consisting of 250 mg Fe2+ L–1 added as FeSO4.7H2O to the nutrient solution (de Dorlodot et al., 2005), was applied to half of the plants after 2 wk acclimation for a period of 4 wk. For each experimental condition (i.e., control and treated conditions) and each replication in space (3 replications), the 164 RILs and two plants of each parent were randomly distributed among seven tanks (24 plants per tank) so that the entire experiment amounted to a total of 42 tanks. Treatments and plants were randomized in the environmental conditions by moving the tanks every 2 d. The experiment was repeated twice in time. Several parameters potentially linked to resistance mechanisms to iron toxicity were quantified on all or part of the RILs, depending on the measurement complexity. The LBI, ranging from 1 (no symptoms) to 9 (IRRI, 1996), shoot dry weight, and shoot water content were determined on each of the 164 lines and the parents. The iron concentrations were determined in the shoots and roots of the 48 most extreme lines and the parents by atomic absorption spectrophotometer after calcination at 450°C and digestion in 6 mL chlorhydric acid and 2 mL nitric acid. Physiological parameters—stomatal resistance, chlorophyll content index, and chlorophyll fluorescence—were measured on the middle upper face of the youngest fully expanded leaf of the 20 most extreme lines and the parents. The stomatal resistance was measured with an automatic porometer (AP3, Delta-T Device, Cambridge, UK). The chlorophyll content index was measured using a Chlorophyll Content Meter (CCM-200 model, Opti-Sciences, Hudson, NH) and the chlorophyll fluorescence measurements were made with a fluorescence induction monitor (FIM 1500, ADC BioScientific Ltd, Herts, UK), which provides the photosystem II maximum quantum efficiency (Fv:Fm) after adaptation of the leaves to darkness.

WWW.CROPS.ORG


Experiment 2: Hydroponics in a Greenhouse at Université catholique de Louvain The screening was conducted in a greenhouse at UCL using only the two parents and 40 extreme lines. The experimental design was similar to that of the first experiment with the following modifications: (i) each plate contained all 40 lines and two repeats of each parent, amounting to a total of 44 lines per plate; (ii) seeds were germinated directly on a grid added on the underside of the plate; (iii) three seeds were placed in each hole of the plate and a selection among the germinated lines was conducted after 12 d, keeping only one plant per line; (iv) the Fe-treatment was applied after 20 d of culture for a period of 3 wk; and (v) mean values of the environmental conditions were about 28/20°C day/night, 70% relative humidity, 12 h photoperiod, and 200 μmol m–2 s–1 light intensity. However, a large variability was observed in these growing conditions because of external climate variations. The following parameters were measured on each plant, using the same methodology and equipment as in Exp. 1: LBI, shoot dry weight, shoot water content, chlorophyll content index, and stomatal resistance. Chlorophyll fluorescence was measured only on 20 extreme RILs and both parents.

Experiment 3: Washed Sandy Soil in a Greenhouse at Africa Rice Center This experiment was conducted using the two parents and the same 40 extreme lines as in the experiments conducted in the greenhouse at UCL. The material was direct seeded in 7-L plastic pots containing 5 kg of washed sand supplemented with iron. The ferrous iron was added at once to the washed sand, in the form of FeSO4.7H2O at 300 mg Fe2+ L –1 for the control plants and 900 mg Fe2+ L –1 for the Fe-treated plants. These doses were chosen on the basis of previous experiments on iron toxicity conducted at AfricaRice using the same protocol (Koné et al., 2008). At the same time, P and K fertilizers were added to the substrate as super triple (250 mg P per pot) and potassium chloride (250 mg K per pot), respectively. Nitrogen fertilizer was supplied in the form of urea (3x 100 mg N per pot) as follows: before planting, at tillering, and at panicle initiation. The washed sand supplemented with iron and fertilizers was then flooded and seeds were sown after 2 d. Two weeks after emergence, plantlets were thinned to four per pot. There were three independent replications, each of them consisting of a combination of both treatments (control and treated) and the lines (40 RILs plus the two parents). Therefore, the whole experiment included a total of 252 pots and 1008 plants. Treatments and lines were randomly distributed in each replication. Plants were watered every day keeping 4 cm of water above the sand level in each pot. Plants were harvested after 6 wk of culture. The experiment was conducted in a mosquitonetted greenhouse covered by semitransparent plastic, with an average temperature of 41/27°C day/night and relative humidity of 80%. The following parameters were measured on each plant: LBI, shoot dry weight, shoot water content, chlorophyll content index, stomatal resistance, and chlorophyll fluorescence.

Experiment 4: Field Conditions at Institut National de l’Environnement et de Recherches agricoles The 164 RILs and the parents were grown during the dry season in the Kou Valley in Burkina Faso (11°22´ N, 4°22´ W) in CROP SCIENCE, VOL. 52, MARCH– APRIL 2012

an irrigated field affected by iron toxicity. Soil analyses were performed in the top horizon indicating strong acidity (pHH20 = 4.7 and pHKCl = 3.6), carbon content of 1.3%, and cation exchange capacity of 7.9 cmol kg–1, with base saturation (Ca2+, Mg2+, K+, and Na+) of 1.5%. This soil is classified as a Gleyic Ferralsol, rich in Fe-oxides and kaolinite. After flooding for rice cultivation, ferrous iron cations are released into the soil solution and accumulate on the exchange complex as a result of decreasing redox potential and increasing pH, which leads to iron toxicity for plants. Cultivars CK4 and Bouaké189, taken as resistant and sensitive to iron toxicity, respectively (Audebert, 2006), were used as check plants. Two weeks after germination in a nursery, the RILs, parents, and check cultivars were transplanted at the rate of three seedlings per hill in plots of 0.60 m2 (three rows of 1 m with 0.20 m spacing between and within plots, respectively). A lattice square design (Williams et al., 1986), where lines, parents, and check cultivars were randomized into three replications, was used. Basal fertilizer (15–25–15 N–P–K) was applied at sowing (200 kg ha–1). Nitrogen fertilizer was also added as urea (46% N) 2 wk after transplanting (35 kg urea ha–1) and at panicle initiation (65 kg urea ha–1). The LBI was determined on each line and cultivar 60 d after emergence. Several agronomic traits were also measured: total plot biomass (straw and panicles together), mean panicle dry weight (mean of 10 panicles), total number of spikelets per panicle, fertility rate (percentage of filled spikelets per panicle), growth cycle length (from sowing to heading), and 100-grain weight at 14% grain moisture content.

DATA ANALYSES AND QUANTITATIVE TRAIT LOCI IDENTIFICATION The normal distribution of data and residuals was checked by the Kolomogorov-Smirnov method using the Univariate procedure of the SAS (version 8.2; SAS Institute, 1999) and using the QStat procedure of the QTL Cartographer software package version 1.17 (Basten et al., 1994, 2004). According to Doerge (2002), violation of the normality assumption of the trait distribution has an impact on the distribution of the statistic used to test for a QTL, affecting the accuracy of the QTL detection. For this reason, nonnormal distributions were transformed by logarithmic functions. Effects of lines, treatments, and replications on the parameters were tested by performing crossed mixed model ANOVA using the GLM procedure of SAS with two fi xed factors (lines and treatment) and one random factor (replication). The most resistant RILs and cultivars were compared to the most sensitive ones using the contrast method. As the degree of leaf bronzing is considered a straightforward indicator of the iron-toxicity intensity (Audebert and Sahrawat, 2000), correlation analysis was performed between the LBI and the other parameters measured under high ferrous concentration (Exp. 4). Quantitative trait loci analyses were performed using a saturated genetic map based on 228 microsatellite marker loci, with average and maximum distance between markers of 7 and 23 cM, respectively. Allelic composition of the 164 RILs and the parents was determined at each marker locus according to Ahmadi et al. (2005). For each experiment, QTL were fi rst identified independently by composite interval mapping (Zeng, 1994) using the

WWW.CROPS.ORG

541

QTL Cartographer software package version 1.17 (Basten et al., 1994, 2004). The threshold for accepting a QTL was calculated for each trait by performing 1000 repeats of a permutation test (Churchill and Doerge, 1994; Doerge et al., 1997). The support interval of a QTL was determined by the logarithm of odds (LOD) drop-off method (Lander and Botstein, 1989), defi ned by the points on the genetic map that corresponds to a decrease in the LOD score of 1 unit from the maximum. When two LOD peaks fell in a common support interval, only one QTL was considered to be present, and its approximate position was taken as the highest peak. According to Jansen et al. (1995), the chance of identifying a QTL simultaneously in several independent experiments is low. To minimize the effect of the environment on the QTL analysis, each trait measured in more than one experiment was also analyzed jointly by composite interval mapping for multiple traits ( Jiang and Zeng, 1995) using QTL Cartographer (Basten et al., 1994, 2004). The minimum LOD score for accepting a joint QTL was determined for each trait by the highest threshold across all the experiments analyzed together (four in the case of LBI and three for the other parameters).

RESULTS AND DISCUSSION Preliminary Results In Exp. 3, conducted on washed sand at AfricaRice, high mortality rate occurred in plants exposed to the iron treatment (900 mg Fe2+ L –1), particularly in replications 2 and 3 where 44% of plants died before harvesting. Accordingly, only the first replication, which showed 12% mortality rate only, was retained for the analyses. Analysis of variance and QTL analyses were thus performed on the mean of four plants per RIL, as each pot of the replication

Figure 1. Morphological parameters measured in control and Fetreated plants across four different experimental conditions. 1: in hydroponics in a phytotron at Université catholique de Louvain (UCL); 2: in hydroponics in a greenhouse at UCL; 3: in washed sandy soil in a greenhouse at Africa Rice Center (AfricaRice); 4: in field conditions in the Kou Valley at Institut National de l’Environnement et de Recherche agricole (INERA). Means and standard deviations of the parents Azucena and IR64, 20 resistant recombinant inbred lines (RILs), and 20 sensitive RILs.

542

contained four plants of one RIL. To take all plants into account in the QTL analysis, a value of 9 on the LBI scale (IRRI, 1996) was assigned to the plants that died before the end of the experiment. For the other characters, values were fi xed at half of the minimal value of the character. Reasons for this early mortality were not clear, but several hypotheses can be proposed. First, plants were attacked by acarids [Oligonychus pratensis (Banks)] and by a bacterial infection, principally plants located in replications 2 and 3. Second, it appeared that the concentration of 900 mg l–1Fe2+ applied to the Fe-treated plants was high and led to early mortality of the most sensitive plants.

Phenotypic Response to Ferrous Iron Treatment The high concentration of Fe (II) in the substrate led to the appearance of bronzing symptoms in all four experiments together with a decrease in the shoot dry weight in Exp. 1, 2, and 3 (Fig. 1). For physiological parameters (Fig. 2)—measured in Exp. 1, 2, and 3—significant decreases in shoot water content, chlorophyll content index, and photosystem II efficiency were observed in the Fe-treated plants vs. control plants in all three experiments. The stomatal resistance significantly increased in Fe-treated plants in Exp. 1 and 3, but no significant effect was detected in Exp. 2, maybe because of the small number of lines tested. The shoot and root iron concentrations—measured in Exp. 1 only—increased highly significantly in Fe-treated plants. Agronomic parameters (Fig. 3)—measured in Exp. 4 only—were also affected by the stress. Correlation analysis showed a significant negative linear correlation between the LBI—whose value increases with the degree of sensitivity to iron toxicity—and each agronomic parameters measured. The Pearson correlation coefficient indicated highly negative significant correlation between LBI and total plot biomass (r = –0.275), mean panicle dry weight (r = –0.215), number of spikelets per panicle (r = –0.327), fertility rate (r = –0.218), growth cycle length from sowing to heading (r = –0.462). A significant negative correlation was observed between LBI and 100-grain weight (r = –0.184). The values of the Pearson correlation coefficient, although low, were significant because of the high number of observations (n = 168). As shown in previous studies, iron toxicity can cause severe physiological damage to the plants through oxidative processes (Bode et al., 1995; Becana et al., 1998; Majerus et al., 2007a; Dufey et al., 2009) and subsequently lead to yield reduction (Becker and Asch, 2005; Audebert and Fofana, 2009; Cherif et al., 2009). Our results confi rmed the occurrence of iron-toxic conditions in all four experiments and highlighted the interest of using these parameters in a QTL mapping for resistance to iron toxicity.

WWW.CROPS.ORG


Figure 2. Physiological parameters measured in control and Fe-treated plants across three different experimental conditions. 1: in hydroponics in a phytotron at Université catholique de Louvain (UCL); 2: in hydroponics in a greenhouse at UCL; 3: in washed sandy soil in a greenhouse at Africa Rice Center (AfricaRice). Means and standard deviations of the parents Azucena and IR64, 20 resistant recombinant inbred lines (RILs), and 20 sensitive RILs. CCI, chlorophyll content index; Fv:Fm, photosystem II maximum quantum efficiency.

Phenotypic Variation The response of both parents to the Fe (II) treatment significantly differed from each other in all four experiment. Azucena was generally more resistant to iron toxicity than IR64 for all morphological and physiological parameters.

However, if compared to the control cultivars CK4 and Bouaké189 under field conditions (Exp. 4), both parents of the RILs can be regarded as sensitive to iron toxicity. Indeed, the resistant parental line Azucena presented the same degree of leaf bronzing as the sensitive control

Figure 3. Agronomic parameters measured in the field under excessive iron conditions in the Kou Valley at Institut National de l’Environnement et de Recherches agricoles (INERA). Means and standard deviations of the control cultivars CK4 (iron resistant) and Bouaké189 (iron sensitive), the parents Azucena and IR64, 20 resistant recombinant inbred lines (RILs), and 20 sensitive RILs. CROP SCIENCE, VOL. 52, MARCH– APRIL 2012

WWW.CROPS.ORG

543

Bouaké189. For the agronomic parameters—measured in Exp. 4—IR64 performed better than Azucena, which presented lower values than IR64 for total plot biomass, mean panicle dry weight, and fertility rate despite having more spikelets per panicle and higher 100-grain weight (Fig. 3). This may be explained by the better adaptation of the indica cultivar IR64 to lowland irrigated conditions in comparison to the japonica cultivar Azucena that is usually cultivated in the uplands. Moreover, indica cultivars tend to produce more tillers than japonica ones, which could explain the high biomass production by IR64. The RILs also responded differentially to the Fe (II) treatment. The contrast analysis revealed that the means of the 20 most resistant and 20 most sensitive RILs differed highly significantly for all morphological, physiological, and agronomic parameters except for root iron concentration in Exp. 1 and photosystem II efficiency (Fv:Fm ratio) in Exp. 1 and 2. The iron-toxicity symptoms were generally more pronounced in the sensitive RILs than in the resistant ones. These results confirmed that both groups of extreme lines differed from each other in response to excessive ferrous iron in the culture medium depending on their different resistance capacities. Thus, the variability in the response of the different genotypes to the stress suggests that resistance mechanisms to iron toxicity are genetically controlled and may be improved via breeding. Finally, a large amount of variability was observed in the mean phenotypic values of the control and Fe-treated plants in all four environmental conditions (e.g., substrate, temperature, light intensity, and humidity). However, the main tendencies stayed relatively constant across the four experiments.

Quantitative Trait Loci Mapping for Resistance to Ferrous Iron Toxicity The results were first subjected to four separate QTL analyses by composite interval mapping (single QTL analysis): one for each environmental condition. Altogether, the four single QTL analyses identified 44 and 26 putative QTL for all parameters examined using a significance level (α) of 0.05 and 0.01, respectively. These QTL were located in 26 nonoverlapping regions on chromosomes 1, 2, 3, 4, 7, 8, 9, 10, and 11 (Table 1; Fig. 4). From the 44 significant putative QTL identified, 12 QTL were found in Exp. 1, 14 in Exp. 2, two in Exp. 3, and 16 in Exp. 4. In Exp. 3, conducted in washed sand in Benin, no highly significant QTL was detected. This may be because only the first replication was analyzed—as explained above— which affected the power and accuracy of the analysis. It is important to note that the estimated QTL additive effects of the same trait measured in different environments were not comparable because data were measured at different scales for the analyses of nontransformed and log-transformed data. Moreover, the population size was different between experiments. Experiments 2 and 3 were 544

performed only on extreme lines relative to their response to a high level of ferrous iron in the culture medium. Various studies have shown that the effect of the QTL is biased when using small-size populations (Beavis, 1998; Melchinger et al., 1998; Dufey et al., 2009). For these reasons, estimated QTL effects found in Exp. 2 and 3 are to be taken with caution when interpreting results.

Influence of the Environment on Quantitative Trait Loci Detection The large variation observed in the expression of the genes across different environments is a significant hindrance to the development of marker-assisted selection. For this reason, only QTL that are stable and consistent across several environments can be of interest for the improvement of the selection of the studied character (Dudley, 1993). In our study, QTL detection was greatly influenced by the environment as only two QTL were detected independently for the same trait in two different experiments: for stomatal resistance on chromosome 1 in region RM473a to RM443 and for chlorophyll content index on chromosome 7 in region RM234 to RM118. However, the QTL detection in a single environment depends on the threshold value chosen for accepting the QTL. Thus to compare accurately the different analyses in the four experiments, a composite interval mapping for multiple traits (joint QTL analysis) was performed for each parameter measured in more than one environment, that is, LBI, shoot dry weight, shoot water content, stomatal resistance, chlorophyll content index, and chlorophyll fluorescence. This analysis gives the most probable position of the QTL, taking into account the correlation between several measures of the same parameter in different environments. For all parameters analyzed, the joint QTL analysis identified 29 highly significant putative joint QTL on chromosomes 1, 2, 3, and 7 (Table 2; Fig. 4). This analysis confirmed the position of seven QTL that were detected by single QTL analysis in only one environment but whose LOD peaks were too low to reach the significance threshold in other environments: three joint QTL for LBI (on chromosome 1 near RM034 and RM403 and on chromosome 3 near RM186), one for shoot dry weight (on chromosome 1 near RM014), one for stomatal resistance (on chromosome 1 near RM246), and two for Fv:Fm ratio (on chromosome 2 near RM465c and RM240). It is note worthy that both QTL identified for the same trait in Exp. 1 and 2 separately—on chromosome 1 for stomatal resistance and on chromosome 7 for chlorophyll content index— were not positioned by the joint QTL analysis at the same locations but in adjacent regions. This might be because of the higher accuracy of the joint QTL analysis to position the QTL in comparison with single QTL analysis (Jiang and Zeng, 1995), because it integrates the LOD peaks of all experiments in the analysis. Finally, 20 putative joint QTL

WWW.CROPS.ORG


Table 1. Putative quantitative trait loci (QTL) detected by composite interval mapping in four different experimental conditions for leaf bronzing index (LBI), shoot water content (SWC), shoot dry weight (SDW), shoot iron concentration (SIC), stomatal resistance (SR), chlorophyll content index (CCI), total plot biomass (TPB), mean panicle dry weight (PDW), number of spikelets per panicle (NSP), fertility rate (FR), growth cycle length (GCL), and 100-grain weight (100GW). Number of QTL 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Trait† LBI

SWC, % SDW, g SIC, mg g–1 SR, s cm –1 CCI

Number of Chromosome genotypes‡ number§

Marker interval¶

LOD score and Position# significance level††

Additive or multiplicative effect‡‡ Direction§§

R2¶¶

QTL identified in Exp. 1: In hydroponics in a phytotron at the Université catholique de Louvain (UCL) 166 1 RM034–RM246 85.99 7.62** 0.85 x 1 RM443–RM403 96.82 7.92** 0.85 x 1 RM265–RM315 121.46 3.78* 1.11 x 11 RM224–RM144 99.43 3.75* 0.90 x 166 2 RM452–RM324 46.15 1.42 + 6.23** 166 3 RM468–RM143 146.19 3.66* 1.16 x 4 RM417–RM142 45.40 3.78** 0.85 x 50 2 RM324–RM550 46.15 8.53** 0.76 x 11 RM004b–RM019 4.27 5.24** 1.23 x 22 1 RM246–RM473a 88.44 7.59** 1.28 + 1 RM473a–RM443 92.98 6.49* 1.28 + 22 7 RM234–RM118 59.47 4.58 + 6.71**

IR64 IR64 Azucena IR64 Azucena Azucena IR64 IR64 Azucena IR64 IR64 Azucena

18.5 18.1 8.1 7.5 13.1 9.1 10.9 35.7 17.6 40.5 40.5 33.9

QTL identified in Exp. 2: In hydroponics in a greenhouse at UCL RM186–RM055 116.26 4.26* RM349–RM280 122.32 3.81* RM590–RM147 96.33 4.59** RM206–RM254 80.07 4.93** RM165–RM014 156.06 4.38** RM338–RM473d 71.80 4.83** RM479–RM287 38.24 4.38** RM473a–RM443 90.98 5.78** RM210–RM080 94.59 3.83* RM209–RM229 51.16 6.28** RM465c–RM561 50.55 7.79** RM318–RM240 119.41 5.41* RM250–RM166 130.33 8.94** RM020b–RM004b 0.01 8.63**

Azucena IR64 Azucena Azucena Azucena Azucena IR64 IR64 Azucena IR64 IR64 IR64 IR64 Azucena

19.5 17.6 26.7 30.5 20.8 21.4 19.0 38.1 16.1 32.2 31.9 44.8 42.7 39.5

Azucena Azucena

22.6 27.0

IR64 IR64 IR64 IR64 Azucena IR64 Azucena Azucena IR64 IR64 Azucena IR64 Azucena Azucena Azucena IR64

7.2 19.0 8.7 19.2 9.0 7.7 6.6 8.0 8.9 13.8 30.5 5.8 6.2 10.2 6.2 7.6

LBI

42

SWC, % SDW, g

42 42

SR, s cm –1 CCI

42 42

Fv:Fm##

22

1 2

CCI Fv:Fm

42 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

LBI

3 4 10 11 1 3 11 1 8 11 2 2 2 11

1.23 x 0.81 x 1.30 x 1.98 + 0.067 + 0.119 + 0.066 + 1.38 x 3.10 + 4.26 + 0.33 x 0.37 x 0.37 x 2.72 x

QTL identified in Exp. 3: In washed sand in a greenhouse at the Africa Rice Center

TPB, g PDW, g NSP FR, %

GCL, days 100GW, g

7 2

RM234–RM118 RM526–RM221

59.47 99.57

3.88* 4.47*

1.66 x 0.14 +

QTL identified in Exp. 4: In the field at the Institut National de l’Environnement et de Recherches agricoles 166 2 RM263–RM526 96.50 3.39* 0.38 + 3 RM132–RM231 4.15 7.95** 0.62 + 166 7 RM505–RM234 57.52 3.95* 30.64 + 8 RM230–RM281 124.21 6.53** 47.36 + 166 3 RM132–RM231 8.15 3.82** 1.33 x 8 RM321–RM409 28.01 3.61* 0.73 x 166 3 RM132–RM231 2.15 1.10 x 3.23* 166 3 RM060–RM132 0.01 4.44** 5.14 + 7 RM234–RM118 63.47 4.33** 5.46 + 8 RM342b–RM042 75.02 7.32** 6.82 + 166 3 RM132–RM231 2.15 16.13** 4.77 + 7 RM2819–RM445 17.94 3.35* 2.09 + 166 1 RM034–RM246 87.99 3.04* 0.07 + 2 RM221–RM318 103.63 5.14** 0.09 + 3 RM442–RM085 170.59 3.19* 0.07 + 10 RM590–RM147 96.33 3.36* 0.08 +

*Significant QTL at the 0.05 probability level. **Significant QTL at the 0.01 probability level. † Parameter analyzed. ‡ Number of recombinant inbred lines (RILs) measured for each trait. The two parents were also analyzed. § Chromosome number where the QTL were detected. ¶ Marker interval in which is located the most probable position of the QTL (maximum logarithm of odds [LOD] score). # Most probable position of the QTL (in centimorgans). †† LOD, logarithm of odds. Likelihood ratio estimated for each trait by 1000 repeats of permutation test. ‡‡ Effect of substituting a single allele from one parent to another. Additive effects are indicated in the table by the sign “+.” Multiplicative effects, indicated in the table by the sign “x,” arise when the QTL analysis was performed on log-transformed data. §§ Direction of the QTL. Azucena indicates that the allele of the parent Azucena contributes to increase the value of the parameter and IR64 indicates that the allele of the parent IR64 contributes to increase the value of the parameter. ¶¶ Part of the phenotypic variation explained by the QTL (%). ## Fv:Fm, photosystem II maximum quantum efficiency.


WWW.CROPS.ORG

545

Figure 4. Location of putative quantitative trait loci (QTL) detected by composite interval mapping in four different experimental conditions for traits potentially linked to iron toxicity resistance mechanisms, that is, leaf bronzing index (LBI), shoot water content (SWC), shoot dry weight (SDW), shoot iron concentration (SIC), stomatal resistance (SR), chlorophyll content index (CCI), chlorophyll fluorescence (photosystem II maximum quantum efficiency [Fv:Fm]), total plot biomass (TPB), mean panicle dry weight (PDW), number of spikelets per panicle (NSP), fertility rate (FR), growth cycle length (GCL), and 100-grain weight (100GW). Thin arrows (→) indicate the most probable position, direction, and part of phenotypic variation (R2) explained by the putative QTL; an arrow toward the left means that the allele of the parent IR64 increased the value of the trait and an arrow toward the right means that the allele of the parent Azucena increased the value of the trait. Thick arrows (⇒) indicate, for each trait measured in more than one experiment, the most probable position of the joint QTL identified by composite interval mapping for multiple traits.

546

WWW.CROPS.ORG


Table 2. Putative joint quantitative trait loci (QTL) for traits potentially linked to resistance mechanisms to iron toxicity, measured in different environmental conditions: leaf bronzing index (LBI), shoot dry weight (SDW), shoot water content (SWC), stomatal resistance (SR), chlorophyll content index (CCI), and chlorophyll fluorescence (photosystem II maximum quantum efficiency [Fv:Fm]). Joint QTL were identified by composite interval mapping for multiple traits. Number of QTL 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

Trait†

Chromosome‡

Marker interval§

LBI

1 1 3 1 1 1 2 1 1 1 1 3 3 1 1 1 2 2 2 7 1 1 1 2 2 2 3 7 7

RM034–RM246 RM443–RM403 RM186–RM055 RM034–RM246 RM443–RM403 RM165–RM014 RM324–RM550 RM476a–RM084 RM272–RM490 RM009–RM005 RM246–RM473a RM016–RM135 RM503–RM2334 RM476a–RM084 RM084–RM220 RM490–RM576 RM154–RM110 RM492–RM452 RM5651–RM106 RM429–RM172 RM129–RM157a RM034–RM246 RM443–RM403 RM452–RM324 RM465c–RM561 RM318–RM240 RM338–RM473d RM234–RM118 RM429–RM172

SDW

SWC SR

CCI

Fv:Fm

Position¶ 83.98 96.82 116.26 85.99 96.82 154.06 46.78 5.22 25.84 72.09 88.43 95.80 107.98 5.22 17.09 33.58 0.01 36.64 85.40 72.12 60.46 79.98 96.82 42.15 50.55 119.41 79.80 59.47 74.12

Joint LOD score# 8.52 5.11 6.28 7.03 6.85 7.02 7.04 8.79 7.80 8.89 7.94 6.76 7.39 16.96 8.37 9.04 7.31 9.90 9.70 6.03 16.26 9.96 12.93 10.03 10.25 7.21 7.45 7.06 7.15

Exp. 1 IR64 IR64 Azucena Azucena Azucena Azucena IR64 IR64 IR64 Azucena IR64 IR64 IR64 IR64 IR64 IR64 IR64 Azucena Azucena Azucena IR64 Azucena IR64 Azucena Azucena Azucena IR64 Azucena Azucena

Direction†† Exp. 2 Exp. 3 IR64 IR64 Azucena Azucena Azucena Azucena Azucena IR64 Azucena Azucena Azucena Azucena Azucena Azucena Azucena Azucena Azucena Azucena Azucena Azucena Azucena Azucena Azucena IR64 IR64 IR64 IR64 IR64 IR64

IR64 IR64 IR64 IR64 IR64 Azucena

Exp. 4 IR64 IR64 Azucena

IR64 IR64 IR64 Azucena IR64 IR64 IR64 Azucena Azucena IR64 IR64 IR64 Azucena Azucena Azucena IR64 Azucena Azucena Azucena Azucena Azucena Azucena

†

Parameter analyzed. Chromosome number where the joint QTL was detected (centimorgans). § Marker interval of the most probable position of the QTL (maximum logarithm of odds [LOD] score). ¶ Most probable position of the joint QTL. # LOD, logarithm of odds. Likelihood ratio for the joint analysis. †† Direction of the QTL. Azucena indicates that the allele of the parent Azucena contributes to increase the value of the parameter and IR64 indicates that the allele of the parent IR64 contributes to increase the value of the parameter. ‡

were found in regions where no QTL was identified by the single QTL analyses. Indeed, a stable QTL with small effect may not be detected in a single environment but may be well detected in the joint QTL analysis. Combining both analyses—single and joint QTL analysis—would be useful to distinguish general from environment-specific gene expressions. Indeed, QTL linked to general gene expression are expressed in all kinds of environments and are therefore more likely to be identified by the joint analysis. Conversely, QTL detected by the single analysis but not by the joint analysis should be linked to a gene whose expression varies according to the environment.

Reliability of Quantitative Trait Loci Detection The present study aimed to detect QTL that are stable across different environments. Therefore, reliability of QTL detection was checked in two ways: (i) the results of each single QTL analysis were compared together and (ii) CROP SCIENCE, VOL. 52, MARCH– APRIL 2012

QTL found in our study were also compared with QTL identified under iron-toxic conditions in previous studies. According to the classification of Lander and Kruglyak (1995), a highly significant QTL detected by two independent studies is regarded as a confirmed QTL. In the present study, both QTL for stomatal resistance (on chromosome 1 in region RM473a to RM443) and chlorophyll content index (on chromosome 7 in region RM234 to RM118) can be considered to be confirmed according to this classification. Moreover, the comparison of the QTL found in the present study with previous ones confirmed four additional QTL, all for LBI: on chromosome 1 in regions RM034 to RM246 (Wu et al., 1997, 1998) and RM265 to RM315 (Wan et al., 2005), on chromosome 2 in region RM263 to RM221 (Shimizu, 2009), and on chromosome 3 in region RM060 to RM231 (Wan et al., 2003b). When considering all parameters together as indicators of the response of plant to the iron stress, the composite

WWW.CROPS.ORG

547

interval mapping identified 20 QTL in overlapping regions under different environmental conditions on chromosomes 1, 2, 7, 10, and 11, highlighting six regions of great interest for the inheritance of resistance to iron toxicity (Fig. 4): (i) on chromosome 1 in the region RM034 to RM403, six putative QTL were identified for LBI, stomatal resistance, and 100-grain weight; (ii) on chromosome 2, in the region RM492 to RM561, three putative QTL were found for shoot water content, shoot iron concentration, and Fv:Fm ratio; (iii) on chromosome 2, in the region RM263 to RM318, three putative QTL were found for the Fv:Fm ratio, LBI, and 100-grain weight; (iv) on chromosome 7, in the region RM455 to RM429, four putative QTL were detected for the chlorophyll content index, total plot biomass, and fertility rate; (v) on chromosome 10, in the region RM590 to RM147, two putative QTL were found for LBI and 100-grain weight; and (vi) on chromosome 11, in the region RM020b to RM332, two putative QTL were identified for shoot iron concentration and Fv:Fm ratio. The joint QTL analysis highlighted interest in three of these regions. Indeed, this analysis revealed the presence of seven joint QTL in region 1 (on chromosome 1 in region RM034 to RM403) for LBI, Fv:Fm ratio, shoot dry weight, and stomatal resistance. In region 2 (on chromosome 2 in region RM492 to RM561), four joint QTL were found for Fv:Fm ratio, stomatal resistance, and chlorophyll content index. In region 4 (on chromosome 7 in region RM455 to RM429), the joint QTL analysis identified three joint QTL for Fv:Fm and chlorophyll content. The interest of region 1 is still highlighted by the results of Wu et al. (1997, 1998), which identified in this region putative QTL for LBI, relative decrease in shoot dry weight, concentration of dehydroascorbate and ascorbate, and enzyme activities (ascorbate peroxidase and glutathione reductase) using a doubled-haploid population consisting of 123 individuals derived from Azucena and IR64. These enzymes are involved in the scavenging system of H2O2 responsible for the formation of damaging active oxygen species through the Fenton reaction in the presence of excess iron. Ascorbate acts as electron donor in the peroxidase reaction to scavenge H2O2 and is regenerated by either monodehydroascorbate reductase or dehydroascorbate reductase using nitotinamide adenine dinucleotide (phosphate) [NAD(P)H] or reduced form of glutathione as the electron donor (Wu et al., 1998). No joint QTL was found in our study in regions 3, 5, and 6 where QTL had been found by the single QTL analyses. This may be because QTL found in these regions were not expressed in all environments. However, in region 3, one QTL for LBI was identified by Shimizu (2009) and one QTL for shoot iron concentration was reported by Wu et al. (1998). In regions 5 and 6, the marker density is low, which may have affected the power of QTL detection. The implication of the putative QTL found in regions 5 and 6

548

in the inheritance of the resistance to ferrous iron toxicity needs to be confirmed by further studies. Finally, several QTL identified in our study in one single environment only were found for related traits in previous studies. On chromosome 1, Wu et al. (1997, 1998) report QTL for leaf bronzing, relative decrease in shoot dry weight, and shoot iron concentration on the same chromosome in the region of RM014, where a QTL for shoot dry weight was detected in Exp. 2 in the present study. In the adjacent region, between RM318 and RM240, Wan et al. (2005) identified a QTL for LBI in the location of our QTL for Fv:Fm ratio. On chromosome 3, in the region RM060 to RM132, Wan et al. (2003b) also identified a QTL for LBI at the same location as our QTL for LBI, mean panicle dry weight, fertility rate, number of spikelets per panicle, and growth cycle length. On chromosome 11, the QTL identified for shoot water content in our study near the marker RM254 was detected by Wan et al. (2003b) for shoot dry weight.

Pleiotropism and Linked Genes The detection of putative QTL for different traits in common regions under the same or different environmental conditions highlights the relation between these parameters and suggests that these regions are involved in several aspects of the inheritance of resistance mechanisms to iron toxicity. However, several QTL detected in the single and joint QTL analyses showed a positive effect linked to the allele of the parent Azucena in one experiment and to the allele of the parent IR64 in another experiment at the same location. Classical quantitative genetics assumes that trait correlation can be attributed to the effect of pleiotropy or to the tight linkage of genes (Hittalmani et al., 2003). If pleiotropism is the major reason, the coincidence of both the locations of QTL for related traits and the directions of their genetic effects can be expected. If the close linkage of genes is the major reason, the directions of the genetic effect of QTL for different traits may be different although the coincidence of the locations of QTL can still be expected (Zhuang et al., 1997). In our study, pleiotropism would explain the correlation of related traits in most regions concerned, because the direction of the QTL was generally consistent for all related traits. However, gene linkage would explain apparent inconsistencies in the direction of overlapping QTL with antagonistic effects of the alleles of both parents: on chromosome 2 in the region RM492 to RM561 between the QTL for Fv:Fm ratio and the QTL for shoot iron concentration and shoot water content, on chromosome 7 in the region RM455 to RM429 between the QTL for chlorophyll content and the QTL for yield components, on chromosome 10 in the region RM590 to RM147 between the QTL for LBI and the QTL for 100grain weight, and on chromosome 11 in region RM020b

WWW.CROPS.ORG


to RM332 between the QTL for shoot iron concentration and the QTL for Fv:Fm ratio. Most QTL found in this study for morphological and physiological parameters were associated with a negative effect of the IR64 allele. However, in a few cases the positive effect of the QTL on the resistance to iron toxicity was attributed to the allele of the sensitive parent IR64, suggesting a complementary gene action, which is probably the cause of the transgressive segregation observed for morphological and physiological characters (Masahiro and Sasaki, 1997). Conversely, the allele of IR64 influenced positively most QTL identified for agronomic parameters except 100grain weight, which could explain the high performance of IR64 in most agronomic parameters under our experimental conditions where this cultivar is well adapted. This highlights the difficulty of screening complex traits such as resistance mechanisms to iron toxicity, which cannot be directly measured. Quantitative trait loci analyses have been performed on indirect traits taken as indicators of the ferrous iron toxicity stress. For this reason, in the perspective of a marker-assisted selection, QTL have to be classified on their linkage to resistance mechanisms to iron toxicity. Quantitative trait loci for LBI, as direct symptom of the stress, should be considered in priority in further fine-mapping studies.

CONCLUSION Quantitative traits detection is greatly influenced by the environment. For this reason, stability and consistency of QTL across several environments have to be checked before using them in marker-assisted selection. In the present study, six regions controlling traits potentially linked to resistance mechanisms to iron toxicity were identified by comparing the results of the four single QTL analyses (one analysis per environment). Moreover, the joint QTL analysis and the comparison with QTL found in previous studies confirmed the interest of three of these regions: region 1 (on chromosome 1 in region RM034 to RM403); region 2 (on chromosome 2 in region RM492 to RM561), and region 4 (on chromosome 7 in region RM455 to RM429). These findings open the way to further approaches to understand genetic mechanisms involved in resistance to ferrous-iron toxicity. These above regions will thus be the targets of further fine-mapping and candidate genes studies. Acknowledgments We thank the Institut de Recherche pour le Développement (IRD) and Centre de Cooperation internationale en Recherche agronomique pour le Développement (CIRAD) in Montpellier (France) for their collaboration to this study by providing the segregating population and the genetic map of the markers for the RILs – European project EGRAM. We are also thankful to the Fonds national de la Recherche scientifique (FNRS, Belgium, FRFC 2.4556.00) and the Fonds scientifique de Recherche (FSR, UCL, Belgium) for their fi nancial contribution.


References Ahmadi, N., C. Dubreuil-Tranchant, B. Courtois, D. Foncéka, D. This, S.R. McCouch, M. Lorieux, J.C. Glaszmann, and A. Ghesquière. 2005. New resources and integrated map for IR64 × Azucena, a reference population in rice. In IRRI (ed.) Int. Rice Genet. Symp., 5th, Manila, Philippines. 19–23 Nov. 2005. Los Baños, Philippines. Audebert, A. 2006. Diagnostic of risk and approaches to iron toxicity management in lowland rice farming. p. 6–17. In A. Audebert et al. (ed.) Iron toxicity in rice-based system in West Africa. Africa Rice Center, Cotonou, Benin. Audebert, A., and M. Fofana. 2009. Rice yield gap due to iron toxicity in West Africa. J. Agron. Crop Sci. 195:66–76. Audebert, A., and K.L. Sahrawat. 2000. Mechanisms for iron toxicity tolerance in lowland rice. J. Plant Nutr. 23:1877–1885. Basten, C.J., B.S. Weir, and Z.-B. Zeng. 1994. Zmap – A QTL cartographer. p. 65–66. In C. Smith et al. (ed.) Proc. World Congr. Genet. Applied Livestock Prod.: Computing Strategies Software, 5th, Guelph, ON, Canada. 7–12 Aug. 1994. Dep. of Animal and poultry Sci., Univ. of Guelph, ON, Canada. Basten, C.J., B.S. Weir, and Z.-B. Zeng. 2004. QTL cartographer, Version 1.17. Department of Statistics, North Carolina State University, Raleigh, NC. Beavis, W. 1998. QTL analyses: Power, precision, and accuracy. p. 145–162. In A.H. Paterson (ed.) Molecular dissection of complex traits. CRC Press, Boca Raton, FL. Becana, M., J.F. Moran, and I. Iturbe-Ormaetxe. 1998. Irondependent oxygen free radical generation in plants subjected to environmental stress: Toxicity and antioxidant protection. Plant Soil 201:137–147. Becker, M., and F. Asch. 2005. Iron toxicity in rice – Conditions and management concepts. J. Plant Nutr. Soil Sci. 168(4):558– 573. Bode, K., O. Döring, S. Lüthje, H.-U. Neue, and M. Böttger. 1995. The role of active oxygen in iron tolerance of rice (Oryza sativa L.). Protoplasm 184:249–255. Cherif, M., A. Audebert, M. Fofana, and M. Zouzou. 2009. Evaluation of iron toxicity on lowland irrigated rice in WestAfrica. Tropicultura 27(2):88–92. Churchill, G.A., and R.W. Doerge. 1994. Empirical threshold values for quantitative trait mapping. Genetics 138:963–971. Colmant, G., and P. Bertin. 2004. Ferrous iron toxicity in rice: Impact of dissolved oxygen in nutrient solution. Belg. J. Bot. 137(1):105–120. de Dorlodot, S., S. Lutts, and P. Bertin. 2005. Effects of ferrous iron toxicity on the growth and mineral composition of an interspecific rice. J. Plant Nutr. 28:1–20. Defoer, T., M.C.S. Wopereis, M.P. Jones, F. Lançon, O. Erenstein, and R.G. Guei. 2004. Rice-based production systems for food security and poverty alleviation in sub-Saharan Africa. Int. Rice Commission Newsl. 53:85–96. Doerge, R.W. 2002. Mapping and analysis of quantitative trait loci in experimental populations. Nat. Rev. Genet. 3:43–52. Doerge, R.W., Z.-B. Zeng, and B.S. Weir. 1997. Statistical issues in the search for genes affecting quantitative traits in experimental populations. Stat. Sci. 12(3):195–219. Dudley, J.W. 1993. Molecular markers in plant improvement: Manipulation of genes affecting quantitative traits. Crop Sci. 33:660–668. Dufey, I., P. Hakizimana, X. Draye, S. Lutts, and P. Bertin. 2009. QTL mapping for biomass and physiological parameters

WWW.CROPS.ORG

549

linked to resistance mechanisms to ferrous iron toxicity in rice. Euphytica 167:143–160. Genon, J.G., N. de Hepcee, J.E. Dufey, B. Delvaux, and P.A. Hennebert. 1994. Iron toxicity and other chemical soil constraints to rice in highland swamps of Burundi. Plant Soil 166:109–115. Gross, J., R.J. Stein, A.G. Fett-Neto, and J.P. Fett. 2003. Iron homeostasis related genes in rice. Genet. Mol. Biol. 26(4):477–497. Hittalmani, S., N. Huang, B. Courtois, R. Venuprasad, H.E. Shashidhar, J.Y. Zhuang, K.L. Zheng, G.F. Liu, G.C. Wang, J.S. Sidhu, S. Srivantaneeyakul, V.P. Singh, P.G. Bagali, H.C. Prasanna, G. Mclaren, and G.S. Khysh. 2003. Identification of QTL for growth- and grain yield-related traits in rice across nine locations of Asia. Theor. Appl. Genet. 107:679–690. IRRI. 1996. Standard evaluation system for rice. 4th ed. International Rice Research Institute, Manila, Philippines. Jansen, R.C., J.W. Van Ooijen, P. Stam, C. Lister, and C. Dean. 1995. Genotype-by-environment interaction in genetic mapping of multiple quantitative trait loci. Theor. Appl. Genet. 91:33–37. Jiang, C., and Z.-B. Zeng. 1995. Multiple trait analysis of genetic mapping for quantitative trait loci. Genetics 140:1111–1127. Koné, B., S. Diatta, M. Sie, M. Fofana, and O. Ayoni. 2008. Criblage des variétés de riz de bas-fond en condition de toxicité ferreuse. (In French.) Africa Rice Center, Cotonou, Bénin. Lander, E.S., and D. Botstein. 1989. Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121:185–199. Lander, E., and L. Kruglyak. 1995. Genetic dissection of complex traits: Guidelines for interpreting and reporting linkage results. Nat. Genet. 11:241–247. Li, Z.K., S.B. Yu, H.R. Lafitte, N. Huang, B. Courtois, S. Hittalmani, C.H.M. Vijayakumar, G.F. Liu, G.C. Wang, H.E. Shashidhar, J.Y. Zhuang, K.L. Zheng, V.P. Singh, J.S. Sidhu, S. Srivantaneeyakul, and G.S. Khush. 2003. QTL × environment interactions in rice. I. Heading date and plant height. Theor. Appl. Genet. 108:141–153. Lu, C., L. Shen, Z. Tan, Y. Xu, P. He, Y. Chen, and L. Zhu. 1996. Comparative mapping of QTLs for agronomic traits of rice across environments using a doubled haploid population. Theor. Appl. Genet. 93:1211–1217. Mackill, D.J., H.T. Nguyen, and J. Zhang. 1999. Use of molecular markers in plant improvement programs for rainfed lowland rice. Field Crops Res. 64:177–185. Majerus, V., P. Bertin, and S. Lutts. 2007a. Effects of iron toxicity on osmotic potential, osmolytes and polyamines concentrations in the African rice (Oryza glaberrima Steud.). Plant Sci. 173:96–105. Majerus, V., P. Bertin, and S. Lutts. 2009. Abscisic acid and oxidative stress implications in overall ferritin synthesis by African rice (Oryza glaberrima Steud.) seedlings exposed to short term iron toxicity. Plant Soil 324:253–265. Majerus, V., P. Bertin, A. Swenden, A. Fortemps, S. Lobréaux, and S. Lutts. 2007b. Organ-dependent responses of the African rice to short-term iron toxicity: Ferritin regulation and antioxidative responses. Biol. Plant. 51(2):303–312. Masahiro, Y., and T. Sasaki. 1997. Genetic and molecular dissection of quantitative traits in rice. Plant Mol. Biol. 35:145–153.

550

Melchinger, A.E., H.F. Utz, and C.C. Schön. 1998. Quantitative trait locus (QTL) mapping using different testers and independent population samples in maize reveals low power of QTL detection and large bias in estimates of QTL effects. Genetics 149:383–403. Ponnamperuma, F.N., R. Bradfield, and M. Peech. 1955. Physiological disease of rice attributable to iron toxicity. Nature 175:265. Sahrawat, K.L. 2004. Iron toxicity in wetland rice and the role of other nutrients. J. Plant Nutr. 27(8):1471–1504. SAS Institute. 1999. SAS/STAT user’s guide, version 8. SAS Inst. Cary, NC. Senthilvel, S., K.K. Vinod, P. Malarvizhi, and M. Maheswaran. 2008. QTL and QTL × environment effect on agronomic and nitrogen acquisition traits in rice. J. Integr. Plant Biol. 50(9):1108–1117. Shimizu, A. 2009. QTL analysis of genetic tolerance to iron toxicity in rice (Oryza sativa L.) by quantification of bronzing score. J. New Seeds 10(3):171–179. Shimizu, A., S. Yanagihara, S. Kawasaki, and H. Ikehashi. 2004. Phosphorus deficiency-induced root elongation and its QTL in rice (Oryza sativa L.). Theor. Appl. Genet. 109:1361–1368. Wan, J.L., H.Q. Zhai, and J.M. Wan. 2005. Mapping of QTLs for ferrous iron toxicity tolerance in rice (Oryza sativa L.). Yi Chuan Xue Bao 32(11):1156–1166. Wan, J.L., H.Q. Zhai, J.M. Wan, and H. Ikehashi. 2003a. Detection and analysis of QTLs for ferrous iron toxicity tolerance in rice, Oryza sativa L. Euphytica 131:201–206. Wan, J.L., H.Q. Zhai, J.M. Wan, H. Yasui, and A. Yoshimura. 2003b. Mapping QTL for traits associated with resistance to ferrous iron toxicity in rice (Oryza sativa L.), using japonica chromosome segment substitution lines. Yi Chuan Xue Bao 30(10):893–898. Wan, J.L., H.Q. Zhai, J.M. Wan, H. Yasui, and A. Yoshimura. 2004. Detection and analysis of QTLs associated with resistance to ferrous iron toxicity in rice (Oryza sativa L.), using recombinant inbred lines. Acta Agron. Sin. 30(4):329–333. Williams, E.R., D. Ratcliff, and P.H. van Ewijk. 1986. The analysis of lattice square designs. J. R. Stat. Soc., B 48(3):314–321. Wu, P., B. Hu, C.Y. Liao, J.M. Zhu, Y.R. Wu, D. Senadhira, and A.H. Paterson. 1998. Characterization of tissue tolerance to iron by molecular markers in different lines of rice. Plant Soil 203:217–226. Wu, P., A. Luo, J. Zhu, J. Yang, N. Huang, and D. Senadhira. 1997. Molecular markers linked to genes underlying seedling tolerance for ferrous iron toxicity. Plant Soil 196(2):317–320. Yoshida, S., D.A. Forno, J.H. Cock, and K.A. Gomez. 1976. Laboratory manual for physiological studies of rice. 3rd ed. International Rice Research Institute, Manila, Philippines. Zeng, Z.-B. 1994. Precision mapping of quantitative trait loci. Genetics 136:1457–1468. Zhuang, J.-Y., H.-X. Lin, J. Lu, H.-R. Qian, S. Hittalmani, N. Huang, and K.-L. Zheng. 1997. Analysis of QTL × environment interaction for yield components and plant height in rice. Theor. Appl. Genet. 95:799–808.

WWW.CROPS.ORG
