Chapter 20 Marker Assisted Breeding Michael J. Thomson, Abdelbagi M. Ismail*, Susan R. McCouch, and David J. Mackill MJT, AMI, DJM: International Rice Research Institute (IRRI), DAPO Box 7777, Metro Manila, Philippines SRM: Department of Plant Breeding and Genetics, Cornell University, 162 Emerson Hall, Ithaca, NY 14853-1901, USA Summary...............................................................................................................................................................    I Introduction.................................................................................................................................................... II Molecular Markers as Tools for Dissecting Quantitative Traits....................................................................... A Dissecting Complex Traits Using QTL Mapping....................................................................................... B Gene Discovery: Genomics and Positional Cloning................................................................................. C Strategies for Marker-Assisted Selection................................................................................................ III Case Studies from a Model Crop: MAS for Abiotic Stress Tolerance in Rice................................................. A Flooding................................................................................................................................................... B Salinity..................................................................................................................................................... C Phosphorus Deficiency............................................................................................................................ D Drought.................................................................................................................................................... IV Future Perspectives........................................................................................................................................ A Association Mapping for Abiotic Stress Tolerance................................................................................... B Variety Development and Gene Deployment........................................................................................... C Bioinformatics Supporting Molecular Breeding........................................................................................ 1 Integrating Marker Genotype and Plant Phenotype Data................................................................... 2 Using a Gene and Plant Ontology...................................................................................................... 3 Databases for the Next Generation of Plant Breeders.......................................................................    V Conclusions.................................................................................................................................................... Acknowledgements............................................................................................................................................... References............................................................................................................................................................ 452 452 453 454 454 455 457 458 458 460 461 462 462 463 464 464 464 465 465 466 466 * Author for Correspondence, e-mail: abdelbagi.ismail@cgiar.org A. Pareek, S.K. Sopory, H.J. Bohnert and Govindjee (eds.), Abiotic Stress Adaptation in Plants: Physiological, Molecular and Genomic Foundation, pp.451–469. DOI 10.1007/978-90-481-3112-9_20, © Springer Science + Business Media B.V. 2010. 451 452 Michael J. Thomson et al. Summary Recent advances in understanding molecular and physiological mechanisms of abiotic stress responses, along with breakthroughs in molecular marker technologies, have enabled the dissection of the complex traits underlying stress tolerance in crop plants. Quantitative trait loci (QTLs) controlling different abiotic stress traits form the basis for a precise marker-assisted backcrossing (MABC) strategy to rapidly transfer tolerance loci into high-yielding, but stress-sensitive varieties. Case studies are presented to demonstrate the progress and potential for MABC programs to develop rice varieties with increased tolerance to flooding, salinity, phosphorus deficiency and drought, amongst others. Future opportunities exist for employing association genetics for more efficient allele mining for abiotic stress tolerance from germplasm collections, as well as leveraging the power of bioinformatics and genomics data for more efficient trait dissection and use in breeding. Plant breeders now have a wealth of information and tools available to tackle these serious constraints posed by abiotic stresses, with the promise of delivering stable, high yielding varieties, able to thrive in the increasingly degrading soils and the ominously changing environment. Keywords I abiotic stresses • association mapping • Oryza sativa L. • QTLs • rice Introduction As reviewed in previous chapters, abiotic stress responses consist of complex and dynamic systems which allow plants to deal with the manifold stresses encountered in variable environments. The pattern of plant diversity across the various ecosystems around the world represents the consequences of adaptation of each plant species interacting with its environment over a period of time. Out of this assortment of diversity, few Abbreviations: AGI – Arabidopsis Genome Initiative; BAC – bacterial artificial chromosome; CSSLs – chromosomal segment substitution lines; EPSO – European Plant Science Organization; ERF – ethylene responsive factors; FNP – functional nucleotide polymorphism; HKT – transporters highaffinity K+ transporter; IRGSP – International Rice Genome Sequencing Project; IRIS – International Rice Information System; LD – linkage disequlibrium; LOD – scores logarithm of the odds ratio; MABC – marker-assisted backcrossing; MAS – marker-assisted selection; NILs – near-isogenic lines; OsHKT8 – Oryza sativa cation transporter HKT8; PUP1 – phosphorus uptake 1; QTLs – quantitative trait loci; RFLPs – restriction fragment length polymorphism; RILs – recombinant inbred lines; ROS – reactive oxygen species; SOS – salt overly sensitive; SSR – simple sequence repeat; SNP – single nucleotide polymorphism; SKC1 – shoot potassium content 1; SUB1 – submergence 1 selected species have undergone the process of domestication, whereby intense selection pressure was imposed for traits important to humans. While domestication brought about larger fruits, nonshattering grains, and higher harvest index, each species was constrained by the innate responses for dealing with various abiotic stresses. However natural and artificial selection still played a notable role during this domestication, particularly for adaptation to various environmental perturbations. At the same time, the production environment was altered to allow more stable crop production through land leveling, tillage, irrigation, weeding, soil amendments and mitigation strategies, designed to limit stress-induced yield losses. The fact that landraces of most crops are relatively tolerant to a wide array of abiotic stresses demonstrates the success of early plant selectors; however, it is due to recent advances in genetics and molecular biology technology that plant breeders have begun to develop new strategies for developing tolerant, high yielding varieties of various crops. The ability of crop plants to succeed in stressprone environments is becoming increasingly important with the need to •• Produce more food from marginal resources as a response to increasing human population pressure •• Allow crops to adjust to the adverse global climate changes 20 453 Marker Assisted Breeding •• Improve the lives of the poorest farmers who depend on stable yields in marginal environments •• Produce higher yields on shrinking resources such as water and agricultural lands, which are rapidly degrading Fortunately, within major crop species, potential donors exist in the available germplasm pool for tolerance to most abiotic stresses, although with varying degrees of accessibility. Many traditional landraces have higher levels of adaptation to stresses as compared to modern high yielding varieties, but they also have many undesirable traits and lower yields. Wild crop relatives may even have higher tolerance, but also pose greater problems in their use for breeding. Plant breeders have faced difficulties in transferring abiotic stress tolerance from exotic germplasm because of negative linkage drag and also due to the polygenic nature of many abiotic stress tolerance traits. Furthermore, breeding programs need to take into account that crops often face multiple abiotic stresses over the course of a growing season, such as an early drought followed by flooding later in the season, and salinity stress followed by drought as in many coastal and inland areas. These stresses could even occur simultaneously, like P- and Zn-deficiencies and Fe- and Al-toxicities often found in acid and alkaline soils (Ismail et al. 2007). These challenges have forced breeders to search for innovative strategies to make further progress on the seemingly intractable problems that have continued to hinder conventional breeding efforts. Recent advances in understanding the molecular mechanisms of abiotic stress responses, along with the breakthroughs in molecular marker technologies, have now enabled the dissection of complex traits underlying many types of stress tolerance in crop plants. The process of genetic linkage mapping as applied towards polygenic traits has led to the identification of quantitative trait loci (QTLs) that control complex traits in plants. Furthermore, by using natural genetic variation to investigate the elaborate systems plants have evolved to deal with a host of abiotic stresses, geneticists can now identify superior tolerance alleles and transfer these alleles into high yielding, stress sensitive varieties. These advances have paved the way towards a markerassisted breeding approach that employs the latest technologies to further improve the performance of varieties and elite breeding lines developed through conventional crossing. While transgenic technologies will ultimately play an important role in developing abiotic stress tolerant plants, a marker-assisted approach provides a useful alternative when the required traits are available within the species gene pool, and especially in situations where genetically modified organisms face difficulties in approval and are yet to be widely accepted. In this chapter, we will cover the use of molecular markers as tools for dissecting the complex traits associated with tolerance of abiotic stresses in major crop plants through QTL mapping, gene discovery and marker-assisted selection (MAS). We will then provide several case studies of how these techniques are currently being employed in the case of rice, to enhance tolerance towards abiotic stresses, including salinity, flooding, phosphorus deficiency, and drought. Lastly, we will provide some thoughts on how future advances, such as bioinformatics and association genetics, might empower marker-assisted breeding techniques for developing high yielding, stress-tolerant varieties in a more efficient way. II Molecular Markers as Tools for Dissecting Quantitative Traits Plant adaptation to variable environments is reflected by an interrelated set of complex physiological and morphological traits, each with an intricate regulatory system. By integrating physi­ological and genetic strategies, we can obtain a deeper understanding of the underlying molecular mechanisms, which opens the way towards a more targeted breeding approach for higher stress tolerance in crop plants. The breakthrough that has made this approach possible was the introduction of easy-to-use DNA markers that brought QTL mapping into the mainstream, making it possible to efficiently map the genetic loci controlling complex traits. This was made possible through genetic linkage analysis, allowing the construction of linkage maps, and the identification of QTLs controlling particular traits based on statistical methods that help establish the association between molecular markers and phenotypic data. 454 A Dissecting Complex Traits Using QTL Mapping Although the theoretical underpinnings of modern QTL mapping were introduced earlier in the twentieth century, the method was limited in application due to the dependence on morphological markers to tag genes (Sax 1923; Thoday 1961). It was not until the introduction of molecular markers, starting with isozymes, that QTL mapping could provide comprehensive coverage of the genome in scanning the loci that control complex traits (Tanksely 1993). Once DNA markers such as RFLPs and SSRs became widely available for most plant species, QTL mapping was quickly adopted. To date, there are over 10,000 mapped QTLs reported for rice and maize in the Gramene database (www.gramene.org). One of the key advantages of QTL mapping is the ability to map genes underlying many different traits and trait components using the same mapping population and the same genetic linkage map. For abiotic stress tolerance, it becomes possible to test different physiological components and compare the QTL locations for these with the QTLs for tolerance or yield under stress to identify the causal factors. Furthermore, use of permanent mapping populations, such as recombinant inbred lines (RILs) or chromosomal segment substitution lines (CSSLs) enables testing stress tolerant traits in replicated experiments across different environments, which can help differentiate the QTLs based on their effectiveness at different stress levels. Once important QTL targets are identified, i.e., large-effect QTLs from the donor that provide increased stress tolerance, these can be captured as single introgressions in a set of nearisogenic lines (NILs), which can help unravel the complexity of different traits by limiting the variation between lines to focus only on the locus of interest. The NILs then provide the foundation for further physiological characterization, finemapping, and ultimately cloning of the QTL to identify the causal gene. B Gene Discovery: Genomics and Positional Cloning Not long after QTL mapping became commonplace, another breakthrough arrived that brought genetic mapping to the DNA sequence level, i.e., Michael J. Thomson et al. the first high-quality complete sequencing of a plant, Arabidopsis (Arabidopsis Genome Initiative 2000), which was followed soon after by the complete sequencing of rice (International Rice Genome Sequencing Project 2005). Having the complete DNA sequence was an instant boon to genetic mapping as it presented an opportunity to make a universal consensus map that can bring together genetic mapping data from disparate sources into a single physical map based on the DNA sequence. This eliminated problems with ambiguous marker orders and variable map distances, and allowed previously mapped genes, QTLs, and markers to be integrated regardless of the original mapping population. In addition, it provided many new markers across the genome, which is essential for fine-mapping. For example, the rice microsatellite map grew from 500 simple sequence repeat (SSR) markers using conventional techniques (Temnykh et al. 2001); to 2,740 SSRs using limited sequence data (McCouch et al. 2002); to 18,828 SSRs using the complete rice genome sequence (International Rice Genome Sequencing Project 2005). In the future, single nucleotide polymorphism (SNP) markers will further increase the number of available markers and will enable more cost-effective high-throughput genotyping techniques (Box 20.1). This wealth in the number of markers is also helpful when screening large numbers of background markers to offset low polymorphism rates, when dealing with closely related parents. Box 20.1 How Will New Marker Technologies Impact Marker-Assisted Breeding? Present Technology: Currently, SSRs are predominantly being used to map and introgress agronomically important QTLs into popular varieties using MABC. However, their use is still limited by the lack of sufficient polymorphism particularly within related genotypes, labor requirements and cost of application. In addition, SSR markers have a low potential for multiplexing and require lengthy periods of time to genotype many markers (as during the initial background selection in the markerassisted backcrossing protocol; see Fig. 1). This results in a short window of time available to obtain the necessary data to select which individuals to backcross. 20 455 Marker Assisted Breeding Box 20.1 (continued) Future Technology: New marker technologies, such as single nucleotide polymorphism (SNP) markers, promise to greatly increase the number of available markers and the speed of genotyping while lowering the cost per data point. For example, Illumina SNP-bead arrays can be used to simultaneously genotype 1,536 SNP markers across the genome for each DNA sample. Impact: SNP markers will enable genomewide screens of thousands of markers simultaneously, which will allow all donor introgressions to be quickly identified during the background selection steps in the marker-assisted backcrossing protocol (see Fig. 1). This information will allow more selective genotyping in subsequent steps, since only the known introgressions need to be tracked. Likewise, the high resolution SNP genome scans can be used for the final step to confirm the background conversion to the recurrent parent. Outlook: As new marker technologies bring down the cost per data point, it makes more sense to integrate high resolution genome scans into marker-assisted breeding programs to increase the overall efficiency by reducing the time spent on marker genotyping and providing more robust data with fewer errors and more confidence in the location of the donor introgressions. These improvements will likely result in more effective use of markers in breeding for abiotic stress tolerance in major crop plants. These advances have improved the efficiency of fine-mapping and have made cloning a QTL, to isolate the casual gene, much easier in Arabidopsis (Lukowitz et al. 2000; Jander et al. 2002) and in rice (Ashikari and Matsuoka 2006), in spite of the challenges still existing (Salvi and Tuberosa 2005). Most of the QTLs cloned to date were associated with morphological attributes such as plant height, fruit characteristics or flowering time (Paran and Zamir 2003), and similar approaches can be used successfully to identify genes controlling key steps for other physiological traits. Several major QTLs associated with tolerance of abiotic stresses have been identified and fine-mapped in rice, with the first examples of map-based cloning being the salinity tolerance QTL SKC1 (Ren et al. 2005) and the submergence tolerance QTL SUB1 (Xu et al. 2006). The fine-mapping and cloning of QTLs has also given more confidence in the results of the primary QTL studies, which have proven to be highly accurate upon retrospect (Price 2006). Once a QTL is cloned, knowledge of the underlying sequence allows further probing into allelic variation at the causal gene level, which can help identify the functional nucleotide polymorphism (FNP) that controls the change in phenotype. The FNP can then be used to develop a functional or perfect marker that directly assesses the desired phenotype at the molecular level. For example, the cloning of the fragrance gene led to a perfect marker for aroma in rice at an 8 bp deletion (Bradbury et al. 2005), and the cloning of the gene for red pericarp in rice led to the identification of a 14 bp deletion that has been developed into a marker for red rice (Sweeney et al. 2006). These functional markers can have several advantages in marker-assisted breeding, especially as they will always co-segregate with the desired phenotype, eliminating the danger of recombination between the linked markers and the target gene. This also allows for rapid diagnosis of the allele state at that gene across diverse germplasm accessions (Andersen and Lubberstedt 2003; Mackill and McNally 2005). In a number of cases, however, the FNP may be elusive, especially if multiple sequence changes can result in the same phenotype. In this case, the FNP marker may work well with a particular source of the gene, but may not work across all germplasm accessions that have the trait. C Strategies for Marker-Assisted Selection The recent advances in genomics have paved the way for clear and reliable methods for MAS in plants: from QTL identification, NIL development and fine-mapping to transferring the QTL into popular varieties using a precise markerassisted backcrossing (MABC) strategy (Mackill 2006; Collard et al. 2005; Collard and Mackill 2008; Collard et al. 2008). MABC involves the manipulation of genomic regions involved in the expression of particular traits of interest through DNA markers, and combines the power of a conventional backcrossing program with the ability to differentiate parental chromosomal segments. 456 Michael J. Thomson et al. Fig. 1. Example of the marker-assisted backcrossing scheme used to transfer the SUB1 QTL for submergence tolerance into six mega-varieties, showing recommended numbers of plants and markers for each step to develop a Sub1-converted near-isogenic line (NIL), at either the BC2F2 or BC3F2 generation, depending on the number of background introgressions remaining at the BC2F1 generation and the size of the target introgression desired (Neeraja et al. 2007; E. Septiningsih, unpublished). The efficiency of a MABC program depends on a number of factors, including the size and reliability of the target QTL effect, the precision of the target gene/QTL fine-map, the rate of polymorphism when identifying background markers, as well as the cost, speed, and failure rate of the markers employed in each customized MABC system. For each set of parents and for each target QTL, a cus- tomized MABC package needs to be developed with optimized foreground markers to select the QTL target, recombinant markers flanking the locus of interest to reduce linkage drag and an evenly-spaced set of polymorphic background markers across the genome to select the recurrent parent background (Neeraja et al. 2007; Collard and Mackill 2008). QTL mapping has been 20 Marker Assisted Breeding progressing at an accelerating pace over the past decade, but few products using this technology have been released to farmers. Constraints to the use of marker-assisted selection for quantitative traits include: •• Poor resolution of QTLs on the genetic map •• Small effects of many QTLs •• Interaction of QTLs with environment or genetic background •• Poor selection of appropriate parents for mapping populations •• The expense of genotyping, limiting the number of samples that can be processed If these constraints are carefully addressed, breeders would be much more likely to use MABC to develop stress-tolerant varieties. The proper selection of QTL targets, combined with the development of an optimized MABC package consisting of tested markers and appropriate donors, is essential for the successful implementation of MABC for any breeding targets. The selection of the QTL target and appraisal for its usefulness in a MABC program needs to weigh the benefits of a markerassisted program versus conventional selection. For example, often trait components can be used directly for selection in plant breeding depending on their degree of association with plant adaptability or yield under specific environments, the cost and precision of their assessment and their interaction with the environment. The genetic control of these traits can be affected by factors such as the number of genes involved, extent of association with undesirable pleiotropic effects, or adverse genetic linkage. Traits such as yield, nutrient acquisition and tolerance of abiotic stresses consist of several underlying components that need to be combined to achieve higher performance. At present, inadequate progress has been made in using these physiological criteria for the largescale breeding needed to combine multiple traits of importance, and their use has been essentially limited to the identification of parental lines. The value of using markers as a surrogate for direct selection of stress tolerance components or yield under stress will often depend on how difficult are the phenotyping techniques and the amount of replicated trials required versus how reliable are the linked markers in predicting the phenotype after its transfer to a recipient variety. Practically, 457 it is advisable to introgress QTLs of important agronomic or adaptive values into varieties that are well known to farmers and are covering large areas. This will ensure that the new varieties will be used immediately by farmers, millers and marketing channels, who are usually cautious when taking on new varieties (Mackill 2006). Provided that popular varieties normally have limited lifespan, MABC also needs to be integrated with conventional breeding to incorporate useful QTLs into elite breeding lines. So far, the greatest success in MABC for improving tolerance of biotic and abiotic stresses has been achieved with QTLs proven to provide high levels of tolerance in many different genetic backgrounds and environments (Collard and Mackill 2008; Collard et al. 2008). A good example in cereals is the introgression of SUB1, the major QTL for submergence tolerance, into several popular rice varieties (Xu et al. 2006; Neeraja et al. 2007). Future breeding objectives, however, may require more complex situations, such as the pyramiding of multiple QTLs having more subtle effects that are effective during different developmental stages, or the combining of QTLs for different abiotic stress tolerance into the same genetic background. Up to now, MABC has been successful in transferring traits whose expression is controlled by a single gene or by a gene that controls most of the phenotypic variance of the trait. However, the effective use of MABC in combining several genomic regions which control a single trait or a few independent traits required for a desired phenotype still awaits further development of more efficient technologies and innovative strategies. An example of a successful approach used in the transfer of a single QTL into several popular rice varieties is presented in Fig. 1. III Case Studies from a Model Crop: MAS for Abiotic Stress Tolerance in Rice Abiotic stresses seem to offer unique opportunities for the application of markers because of the identification of major QTLs coupled with the progress made in understanding the biology of tolerance to these stresses and the availability of efficient phenotyping systems, which hold promise for tackling such challenging traits. The most common stresses adversely affecting rice 458 Michael J. Thomson et al. production are excess or deficiency of water, extremes of temperature, and mineral deficiencies or toxicities. We will briefly review the progress made in several key stresses that limit rice production worldwide. A Flooding Excess water stress is a serious problem for rice in flood-prone areas, with damage resulting from water logging during germination, and partial or even complete submergence for varying durations during the growing season. Various tolerance traits are necessary for high and stable productivity in these areas. Direct seeding of rice is increasingly being practiced in both rain-fed and irrigated areas because of labor shortage for transplanting. However, poor crop establishment remains a major obstacle facing its large-scale adoption in areas where flooding is anticipated due to rain or uneven leveling. Using a backcross population developed from a tolerant landrace “Khao Hlan On”, and a sensitive variety IR64, six QTLs were detected, two each on chromosomes 1, 7 and 9, explaining 7–31% of the phenotypic variation, and with LOD scores in the range of about 5–20 (Angaji et al. 2009). Current efforts focus on finemapping a few of these QTLs for exploitation through MABC (E. Septiningsih, unpublished). Complete submergence affects more than 10 million hectares of rice lands in Asia. This stress received considerable attention in the past few decades, and a few tolerant landraces were identified that can withstand inundation for up to 2 weeks. The physiological bases of tolerance were also extensively studied, and among the traits identified (as critical for tolerance) are high energy reserves, limited underwater growth and retention of chlorophyll (Jackson and Ram 2003; Ella et al. 2003; Sarkar et al. 2006). The Indian cultivar FR13A is the most widely used source of submergence tolerance, and a major QTL, designated SUB1, was identified that controls most of the submergence tolerance of this genotype (Xu and Mackill 1996). FR13A also has additional QTLs that contribute to its tolerance (Nandi et al. 1997; Toojinda et al. 2003). SUB1 was subsequently fine-mapped and cloned, and three genes encoding putative ethylene responsive factors (ERF), SUB1A, SUB1B, and SUB1C, were identified. SUB1A was recognized as the primary contributor to submergence tolerance (Xu et al. 2006). Cloning of SUB1 provided an excellent opportunity to gain a better understanding of the molecular mechanisms and to unravel the pathways underlying submergence tolerance, and it also helped in designing precise gene-based markers for more accurate genotyping. SUB1 has been successfully introgressed through MABC into a popular high-yielding variety, Swarna, within a 2-year time frame (Neeraja et al. 2007). “Swarna-Sub1”, the first example of a submergence tolerant mega variety, is being evaluated in submergence-prone areas of India and Bangladesh. In the absence of submergence, there were no significant differences in agronomic performance, grain yield or quality between Swarna and Swarna-Sub1 (Sarkar et al. 2006; Neeraja et al. 2007), but a substantial enhancement of survival (Fig. 2a) and two to three-fold increase in yield over the intolerant parent was observed after submergence for 12–17 day period during vegetative stage in the field (Singh et al. 2009). Due to its large effect in providing tolerance, this QTL is an excellent candidate for the application of MAS, and progress has been made to successfully convert several popular rice varieties to submergence tolerant varieties using marker-assisted backcrossing with SUB1 (Septiningsih et al. 2009). The SUB1 QTL provides a marked improvement of submergence tolerance in all genetic backgrounds and environments tested so far. Yet, the level of tolerance is still below that of the original donor FR13A. There is a need to identify and combine additional genes from FR13A and probably from additional donors, including those that confer rapid recovery after submergence, along with SUB1. Varieties that combine tolerance to submergence during germination as well as the vegetative stage, together with tolerance to partial long-term stagnant flooding, would have a major advantage for achieving higher and more stable productivity in flood-prone areas. Molecular markers specific to SUB1 and to QTLs associated with tolerance of anaerobic conditions during germination are currently being used to select new lines combining both traits. B Salinity Salt stress negatively affects growth and productivity of most crop plants and recent research 20 Marker Assisted Breeding 459 Fig. 2. (a) Performance of Sub1 introgression lines under field conditions. Fourteen day old seedlings were transplanted in the field and completely submerged 14 days later, for 17 days. Photo was taken about 2 months after desubmergence. (1) IR64, (2) IR64-Sub1, (3) Samba Mahshuri, (4) Samba Mahshuri-Sub1, (5) IR42 (sensitive check) and (6) IR49830 (tolerant, used as SUB1 donor) (Photo courtesy of IRRI); (b) a rice farmer and his wife showing the performance of their local variety (right) and an improved salt tolerant breeding line (left) in a highly alkaline soil in Faizabad district, Uttar Pradesh, India (Photo by A. Ismail taken on Oct. 7, 2007) [See Color Plate 12, Fig. 19]. has started to unravel the complexities of the traits involved in its tolerance, such as control of sodium transport, Na+ and K+ ion homeostasis, and salt response signaling (Zhu 2003; Horie and Schroeder 2004). The fundamental knowledge of salt response mechanisms in plants forms the basis for developing strategies to improve salt tolerance in crop species such as rice (Sahi et al. 2006; Ismail et al. 2007). Although rice is relatively salt-sensitive, it is the only cereal that can grow on many salt-affected soils because it can survive recurrent floods in coastal areas and can thrive in standing water that can help leach salts from top soils in inlands. Tolerance of salt stress in rice is complex and varies with the stage of development, being relatively tolerant during germination, active tillering and towards matu- rity, but sensitive during the early vegetative and reproductive stages (also see Chapter 18). Salinity tolerance at these two sensitive stages is only weakly associated (Moradi et al. 2003). Hence, discovering and combining suitable tolerance traits at both stages is essential for developing resilient salt-tolerant cultivars. Despite this complexity of traits associated with salinity tolerance in rice, substantial progress has been made in developing salt tolerant breeding lines that are being evaluated and selected in farmers fields (Fig. 2b), some of which have been also released as varieties. Tolerance during the early vegetative stage involves a number of contributing traits, including salt exclusion and control of ion homeostasis, higher tissue tolerance by compartmentalizing 460 salt into vacuoles, responsive stomata that close faster upon exposure to salt stress, up-regulation of antioxidant systems for protection against reactive oxygen species (ROS) generated during stress, and vigorous growth to dilute salt concentration in plant tissue, amongst others (Yeo and Flowers 1986; Moradi and Ismail 2007). During the reproductive stage, tolerant genotypes tend to exclude salt from flag leaves and developing panicles (Yeo and Flowers 1986; Moradi et al. 2003). Developing tolerant varieties will entail combining these various component traits into a high yielding genetic background, which is hard to achieve through conventional methods. The recent advances in understanding the physiological and molecular bases of tolerance are providing better tools to overcome these hurdles and can substantially enhance progress by enabling more precise genetic manipulation. Recently, sodium transporters have been shown to play key roles in maintaining ion homeostasis in plants under salt stress, through several mechanisms that remove sodium from the cytoplasm by either compartmentalizing it into vacuoles or extruding it out of the cell (Horie and Schroeder 2004). The salt overly sensitive (SOS) pathway is well characterized in Arabidopsis as being involved in signal perception and ion homeostasis under salt stress (Zhu 2003). Recently, the role of this system in controlling salt stress in case of rice has been further elucidated (Martinez-Atienza et al. 2007). SOS pathway genes have also been identified in Brassica (Kumar et al. 2009). In addition, the HKT family of transporters has been shown to play important roles in sodium and potassium uptake as well as homeostasis in a number of plant species including rice (Horie et al. 2001; Golldack et al. 2002). Recently, cloning of the rice QTL SKC1, originally detected by its effect on K+ concentration, identified the causal gene as the sodium transporter OsHKT8 (Ren et al. 2005). Several mapping studies identified QTLs associated with salinity tolerance in rice. For example, a study employing the tolerant Indica landrace Nona Bokra with the susceptible japonica Koshihikari, identified several large-effect QTLs, including the SKC1 QTL and a QTL for shoot Na+ concentration on chromosome 7 (Lin et al. 2004). Similarly, a RIL population between the highly tolerant landrace Pokkali and sensitive IR29 identified a major QTL, designated Saltol, on chromosome 1, which accounts for about 45% Michael J. Thomson et al. of the variation for seedling and shoot Na+/K+ ratio (Bonilla et al. 2002). While the salt-tolerant landraces Pokkali and Nona Bokra were routinely used in the past for breeding, the level of tolerance attained by new lines is always below that of the traditional donors (Gregorio et al. 2002), and the existing tolerant varieties seem to be superior in only a few of the traits known to be associated with tolerance. Combining superior alleles underlying these traits could potentially result in higher levels of tolerance, a task difficult to achieve through conventional methods. More recently, the application of QTL mapping provided the means to genetically dissect tolerance traits into discrete QTLs that can then be pyramided into high-yielding rice varieties using DNA markers. By integrating physiological trait dissection with these genetic and genomic tools, a more complete picture of the complex mechanisms of salt tolerance in rice is beginning to emerge. These advances in turn provide the foundation for efficient deployment of tolerance QTLs through MAS. Near-isogenic lines have been developed for Saltol, which is the major QTL on chromosome 1. The locus is currently being fine-mapped and annotated for further candidate gene analysis and more precise gene-specific markers are being developed. A MABC strategy for Saltol was developed and is currently being used to incorporate the Pokkali allele into popular salt stress-sensitive varieties. Furthermore, other QTLs were identified on chromosomes 3, 4, 10 and 12 for salinity tolerance at the seedling stage. Genetic stocks of RILs and backcross populations were developed to allow further analysis of these QTLs to evaluate their usefulness in breeding. Mapping populations are being developed to identify QTLs associated with tolerance during the reproductive stage, to ultimately combine tolerance at both stages for more stable performance in salt affected areas. After identifying a number of QTLs controlling different mechanisms and providing tolerance at different stages, MABC can be used to develop rice varieties adapted to any specific target conditions based on the extent and time of stress, during the season when stress is anticipated. C Phosphorus Deficiency After nitrogen, phosphorus is the most important inorganic plant nutrient but the least available in most soils because of its tendency for tight 20 461 Marker Assisted Breeding binding. As a consequence, phosphorus deficiency is widespread in many rice-growing areas, particularly where farmers do not have access to phosphate fertilizers and, in most cases, because these soils have high P-fixing capacity. Breeding efficient cultivars capable of effectively mining the large pool of P already fixed in most soils will help increase and sustain yields in low-input agricultural systems, particularly for cereal crops. Two QTL mapping studies have been reported in rice. Wissuwa et al. (1998) used a backcross inbred population, with the recurrent parent Nipponbare (japonica, sensitive) and the landrace Kasalath (indica, tolerant). They detected a major QTL on chromosome 12 for P uptake, P-use efficiency, shoot dry weight, and tiller number. For P uptake, this QTL had a LOD score of 10.7 and explained about 28% of the phenotypic variation. Ni et al. (1998), using RILs from the cross of IR20 (tolerant) with IR55178-3B-9-3 (sensitive), found a similarly strong QTL in the same location. They measured P uptake efficiency as relative tillering ability, relative shoot dry weight, and relative root dry weight. Moreover, an intermediate QTL on chromosome 6 and several other minor QTLs were mapped to several chromosomes. The QTL on chromosome 6 accounted for 25–34% of the variance for the above traits in the Ni et al. (1998) study, but has much less effect (R2 = 9.8%) in the field study of Wissuwa et al. (1998). Subsequent studies focused on the major QTL for phosphorus uptake, located on chromosome 12, designated “PUP1”. Wissuwa and Ae (2001a) transferred this QTL by three backcrosses into the japonica variety Nipponbare. The resulting lines containing the tolerant allele showed a 170% increase in P uptake and 250% increase in yield when grown under low-P conditions. The NILs with the PUP1 allele from Kasalath had increased root growth under low-P conditions, but the differences in root growth and P uptake were not observed under anaerobic soil conditions (Wissuwa and Ae 2001b). This QTL explained close to 80% of the phenotypic variation in a secondary mapping population (Wissuwa et al. 2002). Additional cycles of fine mapping further reduced the PUP1 interval to about 145 kb (Heuer et al. 2009). Subsequent sequencing of the corresponding chromosomal region in the donor parent “Kasalath” showed that PUP1 locus in Kasalath is much larger (278 kb), with large numbers of transposon- and retro-transposon-related elements (Heuer et al. 2009). None of the genes annotated in the PUP1 locus were found to be related to previously known genes involved in P uptake or metabolism, and detailed analyses of the putative candidate genes are currently ongoing. A marker assisted backcrossing system was developed and is being used to transfer this QTL into three popular upland rice varieties that were sensitive to phosphorus deficiency, particularly in acid soils (A. Ismail, M. Wissuwa, S. Heuer, unpublished). The significance of this QTL in enhancing P-uptake efficiency will further be validated after completing the development of these near isogenic lines. D Drought Drought is the most widespread and damaging of abiotic stresses, but improving the drought tolerance of rice has been hindered by the low level of genetic variability and the complex inheritance of the trait. One of the most serious constraints to improving drought tolerance is the difficulty of accurately measuring the level of tolerance. Stress symptoms such as leaf death and rolling are the easiest to measure, but these traits are not always related to yield under stress or to yield in the target environment, which would include yield under stress as well as yield potential without stress. Molecular approaches to drought tolerance have been widely applied to rice, beginning with QTL analysis. Numerous QTLs were identified for secondary traits that are expected to be associated with drought response, such as root characteristics (depth, volume, thinness, penetration ability), leaf rolling and death, membrane stability, and osmotic adjustment (Lafitte et al. 2006). However, very few studies have mapped QTLs related to the actual objective of enhanced yield under drought. Babu et al. (2003) found important QTLs related to grain yield under stress on chromosomes 4 and 12. More recently, two major QTLs for yield under drought were mapped. Bernier et al. (2007) identified a major QTL (qtl12.1) for drought tolerance in a Vandana × Way Rarem mapping population. This QTL improves yield under drought by 47%, and explained more than 50% of the genetic variance. Furthermore, this QTL co-localized with PUP1, the P uptake QTL on chromosome 12. Fine-mapping of this locus is currently ongoing to establish whether the drought QTL is 462 Michael J. Thomson et al. pleiotropic with PUP1 or whether the two are just closely linked. Understanding this association is important because the PUP1 locus was found to enhance root growth and is effective only in aerobic soils. A second major QTL was mapped on chromosome 4 from an IR55419-04 (tolerant) × Way Rarem mapping population (A. Kumar et al., unpublished). This QTL is currently being fine mapped at IRRI. Both QTLs hold greater promise as targets for marker-assisted breeding to enhance drought tolerance in rice. IV Future Perspectives A Association Mapping for Abiotic Stress Tolerance Association or linkage disequlibrium (LD) mapping represents an alternative approach to identifying genes or genomic regions associated with quantitative phenotypic variation (Buckler and Thornsberry 2002; Gupta et al. 2005). Similar to QTL mapping, association mapping exploits natural diversity and recombination within a population to correlate polymorphisms with measurable phenotypic variations. However, in contrast to linkage or QTL mapping, which depends on recombination events generated over a fixed number of generations following a bi-parental cross, association mapping exploits larger number of historical recombination events in a population or diverse set of lines over the course of evolution. Linkage disequilibrium mapping has several advantages over traditional QTL mapping approaches (Thornsberry et al. 2001). First, it can survey the range of allelic variation present in a natural population, and is not restricted to a set of alleles found in the two progenitors of a mapping population. Second, by relying on historical recombination, it is often possible to localize QTLs in a genome to a higher degree of resolution than is possible with the same number of individuals using traditional QTL linkage analysis. Third, the technique can be used without developing new mapping populations. Thus, LD mapping can potentially achieve higher resolutions with greater efficiency than linkage-based QTL mapping techniques by taking advantage of both the array of molecular diversity within a species and the large amounts of historical recombina- tion that has occurred within and between populations during evolution. It is these features that have convinced geneticists and breeders to focus attention on LD mapping as an efficient strategy for identifying genes associated with traits of interest, including abiotic stress resistance, based on an exploration of the rich collection of rice genetic resources. To undertake an association mapping experiment, a collection of accessions is genotyped for markers that span either the entire genome, or a genomic region of interest (Wilson et al. 2004; Szalma et al. 2005). The markers are then tested against a specific phenotype to determine whether a statistical correlation exists between marker genotypes and a particular trait. A significant association between marker(s) and trait may arise either because the nucleotide polymorphism causes the phenotypic difference, or because the marker is in LD with the causal or functional polymorphism (Thornsberry et al. 2001). The resolution of an association mapping experiment depends on the extent of LD, which is the correlation between polymorphic loci within the test population (Flint-Garcia et al. 2003). When a mutation arises in a population, it is automatically associated or comes in “disequilibrium” with all the alleles present in the genome of the individual that gave rise to the mutation. If the mutation persists during evolution, associations with other alleles are gradually eroded by segregation and recombination, so that over a period of time, the mutation remains in LD only with alleles which are closely-linked to it physically (Hartl and Clark 1997). The distance over which LD persists in a species or population determines the number and density of markers required for association mapping, with large variations observed both within and between genomes. The first study on LD in rice reported an LD decay of ~70 to 100 kb around the bacterial blight resistance locus, xa5, in the aus sub-population (Garris et al. 2003). More recent studies confirmed that LD generally decays at the rate of ~50 to 100 kb in landraces of indica, while it decays more slowly in japonica (~150 kb in tropical japonica and >500 kb in temperate japonica; Mather et al. 2007; Rakshit et al. 2007). LD generally extends over significantly larger distances in elite varieties than in landraces (Morrell et al. 2005; Remington et al. 20 463 Marker Assisted Breeding 2001; Tenaillon et al. 2001) due to inbreeding, selection, bottlenecks and population admixtures (Nordborg and Tavare 2002; Weir 1996). The different rates of LD decay in different sub-populations of rice offer opportunities to consciously move between low and high-resolution mapping. In the first instance, the use of breeding lines and elite varieties with extensive LD means that a modest number of markers would be sufficient to identify region(s) of the genome containing a gene or QTL (Agrama et al. 2007; Zhang et al. 2005). While the LD mapping resolution using elite germplasm may not be significantly better than QTL mapping, it does, nonetheless, provide an opportunity to identify associations between a phenotype and a larger set of alleles. In contrast, the use of wild species or landrace varieties that exhibit more rapid LD decay require a larger number of markers to define the recombination profiles of the accessions and provide significantly higher LD mapping resolution. In cases where LD decay is very rapid it may not be cost-effective to saturate the entire genome with closely linked markers. In these situations, markers are targeted to a candidate gene or QTL region across the association mapping panel in an effort to narrow down the interval containing the target gene(s) (Garris et al. 2003; Kruglyak 1999; Olsen et al. 2006; Sweeney et al. 2007). Association mapping is most productive when used in association with QTL mapping. Together, these approaches provide rice researchers with numerous possibilities for establishing meaningful associations between phenotypes and genes. There are currently efforts underway to develop a group of recombinant inbred populations for QTL analysis, along with an immortal association mapping panel that will offer the genetics and breeding community many opportunities to explore genetic diversity and the basis of phenotypic variation in rice. The association mapping panel will consist of several thousand diverse, purified genetic stocks of wild, landrace and elite rice accessions, and this collection can be expanded at will. Purification of the lines is required to ensure good quality and reproducibility of the genotyping and phenotyping effort that provides the data for association analysis. Use of genetically identical material will enable the rice research community to leverage its collective strengths to phenotype the lines in a distributed manner, focusing on specific sub-populations and traits that are most interesting or important to a given group of researchers in different institutions and environments. The lines will be genotyped in a central facility and both the genotypic data and the purified seed stocks will be made publicly available. Using this coordinated approach, abiotic stress tolerance can be rigorously evaluated on a common set of materials over years and environments, and data collected by different groups of researchers can be analyzed together. This would provide new opportunities to unravel the relationship between genotype and phenotype, and will deepen our understanding of the diverse genetic mechanisms that allow plants to respond to a wide range of environmental stresses. Based on estimates of LD, it is suggested that polymorphic markers will be needed approximately every 50 kb to cover the genome for association mapping in O. sativa. If this is true, then ~8,000 well-distributed polymorphic markers would provide a good chance of performing genome-wide association mapping in case of O. sativa. However, to have a reasonable chance of finding polymorphic markers across the different subpopulations, and to take advantage of the more rapid LD decay in some regions of the genome, it is recommended that a set of ~24,000 to 40,000 well-distributed markers be deve­ lo­ped for association mapping in rice. A fixed genotyping array consisting of 44,100 SNP markers is currently under development for rice (www.ricediversity.org). B Variety Development and Gene Deployment As plant breeders have achieved great success in developing high-yielding varieties, it has become increasingly difficult to significantly improve on these varieties for the basic agronomic traits of interest. Mega varieties that are widely grown and liked by farmers have been ensconced in the agricultural system and it is increasingly difficult to displace them due to their suite of desirable features. These varieties not only have high grain yield, but often have improved quality traits meaning that they can be easily sold and marketed, ensuring farmers of consistent demand. Furthermore, the seed systems in rice-growing countries cannot easily cope with multiple varieties 464 and this means that only one or a few varieties are easily available to the farmers. This situation explains why the varietal upgrade path through marker-assisted backcrossing is a reasonable one that is likely to achieve more impact through rapid adoption of improved varieties. As discussed above, this approach is most effective when major QTLs are available for the traits of interest, and this has been the case for several abiotic stresses. Nevertheless a similar approach could be feasible for smaller QTLs. Even a relatively small-effect QTL could have a big impact if deployed over a large area. One of the advantages of the MABC approach is that linkage drag can be minimized through the use of recombinant selection with flanking markers (Collard and Mackill 2008). When multiple QTLs are being transferred, this could make a substantial difference in the ability to retain the desirable features of the mega variety. A wider use of this strategy by rice breeders underlines the importance for fine-scale mapping and positional cloning of QTLs. The wide applicability of this approach should not be seen as a substitute for more conventional breeding approaches that place emphasis on developing new mega varieties. The MABC strategy is a relatively conservative approach aimed at getting incremental improvements in the best varieties, while plant breeders will continue to aim at the goal of new mega varieties through multiple approaches relying on conventional breeding and MAS. C Bioinformatics Supporting Molecular Breeding 1 Integrating Marker Genotype and Plant Phenotype Data Over 8,000 QTLs for a wide array of traits have been reported in rice over the last 15 years. Today, a coordinated body of information about the genomic location of these QTLs, as well as marker trait associations is available in the Gramene database (www.gramene.org). This information resource added value to individual mapping studies by aligning all the QTLs to the rice genome sequence using sequenced markers as anchors. As a result, the rice research community can now readily access information about linkage relationships among QTLs for different Michael J. Thomson et al. traits, derive hypotheses about the stability of QTLs across genetic backgrounds and environments based on co-localization of QTLs across studies, and rapidly identify numerous sources of favorable alleles for diverse traits. The alignment of QTLs across multiple studies has also facilitated the effort to fine-map and clone genes underlying many of these QTLs, providing plant breeders with “functional markers” for use in MAS, and providing geneticists with new information about the genes and alleles that are critical to agricultural performance. As agronomically useful genes are identified, they can be mapped to biochemical and regulatory pathways (Shimizu et al. 2007). Pathway information helps researchers to predict how variation at a particular locus may affect or be affected by other genes in the same or different pathways (European-Plant-Science-Organization 2005). It offers a framework for understanding epistasis, and can help identify multiple genetic factors that collectively determine the phenotype of a plant. It also provides opportunities to implement a reverse genetics approach for finding new genes associated with a trait of interest, and for helping to identify parents that may contribute useful variation to a breeding program. 2 Using a Gene and Plant Ontology The rapid accumulation of genetic and genomic information today has led to a rapid change in our understanding of the biological world. We must scramble to identify and make use of relevant findings. Despite the abundance of new discoveries, we are challenged to find an appropriate method for screening through the mountain of data to identify meaningful bits of information. Genome databases are designed to address this problem by providing a set of tools for data browsing and data mining that are tailored according to the needs of the biological community. While not specifically designed for plant breeders, these databases are indispensable to the modern plant breeding community because they provide information about genes, alleles, germplasm, pathways, phenotypes and environments (Bruskiewich et al. 2003; Jaiswal et al. 2006b). To be useful, data must be entered into the database in a timely way, and it must be structured and organized so that disparate pieces of 20 465 Marker Assisted Breeding information can be retrieved computationally to answer a relevant question. Underlying the success of data mining activities is the use of ontologies and controlled vocabularies (sets of terms with defined relationships to each other), that provide a structured set of hierarchical relationships allowing independent pieces of information to be associated with each other in meaningful ways. In the context of genetics and plant breeding, ontologies have been developed for genes, phenotypes, traits, environments, etc. (Clark et al. 2005; Ilic et al. 2007; Jaiswal et al. 2006a; Pujar et al. 2006). These ontologies make it possible to compare data from diverse organisms and experiments within a data domain (i.e., genes), and they also make it possible for the computer to identify meaningful associations between data domains based on the relationships defined by the ontologies (i.e., genes and phenotypes). The development of ontologies and controlled vocabularies is a rapidly evolving science. It represents a cross-road of expertise involving biologists, agriculturalists, computer scientists and software engineers, whereby essential biological relationships are examined and described in a clear and logical manner. In the Gramene database, several ontologies are currently employed to facilitate browsing and data mining activities, including the gene ontology, the plant anatomy ontology, the plant growth stage ontology, the trait ontology and the environment ontology (Clark et al. 2005; Ilic et al. 2007; Jaiswal et al. 2006b; Pujar et al. 2006; www.gramene.org). Use of these ontologies requires that data, or biological observations, be catalogued using structured as well as controlled vocabularies. The use of controlled vocabularies means that scientists are offered a menu of terms that can be used to describe or annotate observations and they have to choose from among these terms, rather than entering free text at the time of entering data into a database. The terms are defined in ways that allow a computer to recognize and rapidly retrieve them when queried to do so (Bruskiewich et al. 2003; Jaiswal et al. 2006b). The use of controlled vocabularies is necessary so that new observations can be readily placed into appropriate relationships with existing data by a computer and it subsequently facilitates retrieval of a meaningful set of related data points when prompted by a query. 3 Databases for the Next Generation of Plant Breeders Because of their power and utility, genome databases have become part of the essential toolkit of modern plant breeders and geneticists. Not only do databases make it possible to identify candidate genes associated with phenotypes of interest, or to identify markers linked to a particular QTL for use in MAS, but they efficiently leverage information from an experiment conducted on one genome or in one environment as the basis for predicting relationships between traits, genes and pathways in another genome evaluated in a different environment. The role of comparative genomics and comparative biology as a paradigm for addressing questions of basic biological significance has begun to impact the applied plant breeding community as well. Questions that were once debated with species-specific and regionally localized communities are now addressed in a much broader, comparative context, and the perspective of many plant breeders has been transformed by the recent availability of relevant genome data and the power of genome databases and data mining tools to retrieve and assemble meaningful information in real time. Nonetheless, there is still a gap between our ability to generate digital information about genes and alleles and corresponding phenotypic variation of interest, and a plant breeder’s ability to utilize that information to develop a new variety. More emphasis needs to be placed on characterizing diverse genetic resources, both genotypically and phenotypically, and high throughput phenotyping methodologies are needed to keep pace with the flow of information from genome sequencing centers. Most importantly, we need greater integration between databases that specialize in germplasm information management and those that specialize in the management of gene-based knowledge. It is at the intersection of these two worlds where the greatest gain for plant breeders will lie, and where some of the most exciting scientific questions remain to be explored. V Conclusions For rice, some of the best QTLs identified for abiotic stress tolerance have significantly larger 466 effects than those identified for yield and related agronomic traits (Mackill 2006). This suggests that measurable progress in improving productivity under unfavorable conditions could be achieved by transferring these loci into elite genotypes. QTLs that have a large effect on the phenotype and are relatively stable across genetic backgrounds and environments are most desirable for applications in marker-assisted selection. These QTLs could also aid in further dissection of the physiological basis of tolerance of these stresses. Isogenic lines that differ in the introgression of a particular QTL could be generated to help in functional analysis and in evaluating the effect of a particular QTL for yield improvement or for general adaptability. The past two decades witnessed substantial progress in our understanding of plant functions and plant adaptations to different environments. This knowledge was aided by the integration of new tools of molecular biology with the conventional phenotypic methods of plant physiology and biochemistry. Most of the abiotic stress genes in case of rice have been detected based on visual symptoms and while these symptoms correspond fairly well to the actual damage that is observed under field conditions, more quantitative estimates of plant stress tolerance are needed. While phenotyping methodologies have not yet been automated, genotyping platforms are becoming increasingly automated, efficient and cost effective. This makes it possible for plant physiologists to dissect complex phenotypes into distinct factors that are associated with genetic loci or QTLs. The phenotypic impact of each QTL can subsequently be studied in specific genetic backgrounds by developing NILs for the locus under investigation. The ability to map QTLs also provides information about the association between physiological traits, particularly those that are inter-dependent, as well as in understanding the relationships of physiological and metabolic processes with other developmental and morphological traits. Molecular marker technology has greatly accelerated the progress in gene discovery through map-based cloning. The progress in understanding gene function will further be aided by expression analysis and evolutionary studies. Complementing conventional methods of plant breeding with MAS for favorable genes and QTL has already enhanced our breeding efficiency. Michael J. Thomson et al. Indeed, it is an exciting era for plant breeders and physiologists who can bring new knowledge and technologies to bear on the development of more productive, stress-tolerant varieties as we strive to tackle the diverse challenges facing agriculture in the future. Acknowledgements We gratefully thank E. Septiningsih and A. Kumar for sharing unpublished data. Research activities summarized here were supported in part, by the German Federal Ministry for Economic Cooperation and Development (BMZ), the CGIAR Generation Challenge Program, the Bill and Melinda Gates Foundation, and the International Rice Research Institute. References Agrama HA, Eizenga GC, Yan W (2007) Association mapping of yield and its components in rice cultivars. Mol Breed 19:341–356 Andersen JR, Lubberstedt T (2003) Functional markers in plants. Trends Plant Sci 8:554–560 Angaji SA, Septiningsih E, Mackill DJ, Ismail AM (2009) QTLs associated with tolerance of flooding during germination in rice (Oryza sativa L.). Euphytica, DOI 10.1007/ s10681-009-0014-5 Arabidopsis Genome Initiative (2000) Analysis of the genome sequence of the flowering plant Arabidopsis thaliana. Nature 408:796–815 Ashikari M, Matsuoka M (2006) Identification, isolation and pyramiding of quantitative trait loci for rice breeding. Trends Plant Sci 11:344–350 Babu RC, Nguyen BD, Chamarerk V, Shanmugasundaram P, Chezhian P, Jeyaprakash P, Ganesh SK, Palchamy A, Sadasivam S, Sarkarung S, Wade LJ, Nguyen HT (2003) Genetic analysis of drought resistance in rice by molecular markers: association between secondary traits and field performance. Crop Sci 43:1457–1469 Bernier J, Kumar A, Ramaiah V, Spaner D, Atlin G (2007) A large-effect QTL for grain yield under reproductive-stage drought stress in upland rice. Crop Sci 47:507–518 Bonilla P, Dvorak J, Mackill DJ, Deal K, Gregorio GB (2002) RFLP and SSLP mapping of salinity tolerance genes in chromosome 1 of rice (Oryza sativa L.) using recombinant inbred lines. Philipp Agric Sci 85:68–76 Bradbury LMT, Henry RJ, Jin Q, Reinke RF, Waters DLE (2005) A perfect marker for fragrance genotyping in rice. Mol Breed 16:279–283 20 Marker Assisted Breeding Bruskiewich RM, Cosico AB, Eusebio W, Portugal AM, Ramos LM, Reyes MT, Sallan MA, Ulat VJ, Wang X, Mcnally KL, Sackville Hamilton R, McLaren CG (2003) Linking genotype to phenotype: The International Rice Information System (IRIS). Bioinformatics. 19(Suppl 1): 63–65 Clark JI, Brooksbank C, Lomax J (2005) It’s all GO for plant scientists. Plant Physiol 138:1268–1279 Buckler ES IV, Thornsberry JM (2002) Plant molecular diversity and applications to genomics. Curr Opin Plant Biol 5:107–111 Collard BCY, Mackill DJ (2008) Marker assisted selection: an approach for precision plant breeding in the twentyfirst century. Phil Trans R Soc Lond B Biol Sci 363: 557–572 Collard BCY, Jahufer MZZ, Brouwer JB, Pang ECK (2005) An introduction to markers, quantitative trait loci (QTL) mapping and marker-assisted selection for crop improvement: the basic concepts. Euphytica 142:169–196 Collard BCY, Vera Cruz CM, McNally KL, Virk PS, Mackill DJ (2008) Rice molecular breeding laboratories in the genomics era: current status and future considerations. Int J Plant Genomics p 25, doi: 10: 1155/2008/524847 Ella E, Kawano N, Yamauchi Y, Tanaka K, Ismail AM (2003) Blocking ethylene perception enhances flooding tolerance in rice seedlings. Funct Plant Biol 30:813–819 European-Plant-Science-Organization (EPSO) (2005) European plant science: a field of opportunities. J Exp Bot 56:1699–1709 Flint-Garcia SA, Thornsberry JM, Buckler IV ES (2003) Structure of linkage disequilibrium in plants. Annu Rev Plant Biol 54:357–374 Garris A, McCouch S, Kresovich S (2003) Population structure and its effect on haplotype diversity and linkage disequilibrium surrounding the xa5 locus of rice (Oryza sativa L.). Genetics 165:759–769 Golldack D, Su H, Quigley F, Kamasani UR, Muñoz-Garay C, Bladeras E, Popova OV, Bennett J, Bohnert HJ, Pantoja O (2002) Characterization of a HKT-type transporter in rice as a general alkali cation transporter. Plant J 31: 529–542 Gregorio GB, Senadhira D, Mendoza RD, Manigbas NL, Roxas JP, Guerta CQ (2002) Progress in breeding for salinity tolerance and associated abiotic stresses in rice. Field Crops Res 76:91–101 Gupta PK, Rustgi S, Kulwal PL (2005) Linkage disequilibrium and association studies in higher plants: present status and future prospects. Plant Mol Biol 57:461–485 Hartl DL, Clark AG (1997) Principles of population genetics. Sinauer Associates, Sunderland, MA Heuer S, Lu X, Chin J-H, Tanaka JP, Kanamori H, Matsumoto T, De Leon T, Ulat VJ, Ismail AM, Wissuwa M (2009) Comparative sequence analysis of the major quantitative trait locus phosphorus uptake 1 (Pup1) reveal a complex genetic structure. Plant Biotech J 7:456–471 467 Horie T, Schroeder JI (2004) Sodium transporters in plants: diverse genes and physiological functions. Plant Physiol 136:2457–2462 Horie T, Yoshida K, Nakayama H, Ymada K, Oiki S, Shinmyo A (2001) Two types of HKT transporters with different properties of Na+ and K+ transport in Oryza sativa. Plant J 27:115–128 Ilic K, Kellogg EA, Jaiswal P, Zapata F, Stevens PF, Vincent LP, Avraham S, Reiser L, Pujar A, Sachs MM, Whitman NT, McCouch SR, Schaeffer ML, Ware DH, Stein LD, Rhee SY (2007) The plant structure ontology, a unified vocabulary of anatomy and morphology of a flowering plant. Plant Physiol 143:587–599 International Rice Genome Sequencing Project (2005) The map-based sequence of the rice genome. Nature 436: 793–800 Ismail AM, Heuer S, Thomson MJ, Wissuwa M (2007) Genetic and genomic approaches to develop rice germplasm for problem soils. Plant Mol Biol 65:547–570 Jackson MB, Ram PC (2003) Physiological and molecular basis of susceptibility and tolerance of rice plants to complete submergence. Ann Bot 91:227–241 Jaiswal P, Avaraham S, Ilic K, Kellogg E, McCouch S, Pujar A, Reiser L, Rhee S, Sachs M, Schaeffer M, Stein L, Stevens P, Leszek V, Ware D, Zapata F (2006a) Plant ontology (PO): a controlled vocabulary of plant structures and growth stages. Comp Func Gen 6:388–397 Jaiswal P, Ni J, Yap I, Ware D, Spooner W, Youens-Clark K, Ren L, Liang C, Zhao W, Ratnapu K, Faga B, Canaran P, Fogleman M, Hebbard C, Avraham S, Schmidt S, Casstevens T, Buckler E, Stein L, McCouch S (2006b) Gramene: a bird’s eye view. Nucleic Acids Res 34:D717–D723; doi: 10.1093/nar/gkj154 Jander G, Norris SR, Rounsley SD, Bush DF, Levin IM, Last RL (2002) Arabidopsis map-based cloning in the postgenome era. Plant Physiol 129:440–450 Kruglyak L (1999) Prospects for whole-genome linkage disequilibrium mapping of common disease genes. Nat Genet 22:139–144 Kumar G, Purty RS, Sharma MP, Singla-Pareek SL, Pareek A (2009) Physiological responses among Brassica species under salinity stress show strong correlation with transcript abundance for SOS pathway-related genes. J Plant Physiol 166:507–520 Lafitte HR, Ismail AM, Bennett J (2006) Abiotic stress tolerance in tropical rice: progress and future prospects. Oryza 43:171–186 Lin HX, Zhu MZ, Yano M, Gao JP, Liang ZW, Su WA, Hu XH, Ren ZH, Chao DY (2004) QTLs for Na+ and K+ uptake of the shoots and roots controlling rice salt tolerance. Theor Appl Genet 108:253–260 Lukowitz W, Gillmor CS, Scheible WR (2000) Positional cloning in Arabidopsis. Why it feels good to have a genome initiative working for you? Plant Physiol 123:795–805 Martinez-Atienza J, Jiang X, Garciadeblas B, Mendoza I, Zhu JK, Pardo JM, Quintero FJ (2007) Conservation of 468 the salt overly sensitive pathway in rice. Plant Physiol 143:1001–1012 Mather KA, Caicedo AL, Polato NR, Olsen KM, McCouch SR, Purugganan MD (2007) The extent of linkage disequilibrium in rice (Oryza sativa L.) Genetics 177: 2223–2232 McCouch SR, Teytelman L, Xu Y, Lobos KB, Clare K, Walton M, Fu B, Maghirang R, Li Z, Zing Y, Zhang Q, Kono I, Yano M, Fjellstrom R, DeClerck G, Schneider D, Cartinhour S, Ware D, Stein L (2002) Development and mapping of 2240 new SSR markers for rice (Oryza sativa L.). DNA Res 9:199–207 Mackill DJ (2006) Breeding for resistance to abiotic stresses in rice: the value of quantitative trait loci. In: Lamkey KR and Lee M (eds) Plant breeding: the Arnel R. Hallauer International Symposium. Blackwell, Ames, IA, pp 201–212 Mackill DJ, McNally KL (2005) Molecular markers in rice: a model crop species. In: Lörz H, Wenzel G (eds) Molecular marker systems in plant breeding and crop improvement. Springer Verlag, Heidelberg, Germany, pp 39–54 Moradi F, Ismail AM (2007) Responses of photosynthesis, chlorophyll fluorescence and ROS scavenging system to salt stress during seedling and reproductive stages in rice. Ann Bot 99:1161–1173 Moradi F, Ismail AM, Gregorio GB, Egdane JA (2003) Salinity tolerance of rice during reproductive development and association with tolerance at the seedling stage. Ind J Plant Physiol 8:105–116 Morrell PL, Toleno DM, Lundy KE, Clegg MT (2005) Low levels of linkage disequilibrium in wild barley (Hordeum vulgare ssp. spontaneum) despite high rates of self-fertilization. Proc Natl Acad Sci USA 102:2442–2447 Nandi S, Subudhi PK, Senadhira D, Manigbas NL, SenMandi S, Huang N (1997) Mapping QTLs for submergence tolerance in rice by AFLP analysis and selective genotyping. Mol Gen Genet 255:1–8 Neeraja CN, Maghirang-Rodriguez R, Pamplona A, Heuer S, Collard BCY, Septiningsih EM, Vergara G, Sanchez D, Xu K, Ismail AM, Mackill DJ (2007) A marker-assisted backcross approach for developing submergence-tolerant rice cultivars. Theor Appl Genet 115:767–776 Ni JJ, Wu P, Senadhira D, Huang N (1998) Mapping QTLs for phosphorus deficiency tolerance in rice (Oryza sativa L.). Theor Appl Genet 97:1361–1369 Nordborg M, Tavare S (2002) Linkage disequilibrium: what history has to tell us? Trend Genet 18:83–90 Olsen KM, Caicedo AL, Polato N, McClung A, McCouch SR (2006) Selection under domestication: evidence for a sweep in the rice waxy genomic region. Genetics 173:975–983 Paran I, Zamir D (2003) Quantitative traits in plants: beyond the QTL. Trends Genet 19:303–306 Price AH (2006) Believe it on not, QTLs are accurate!. Trends Plant Sci 11:213–216 Michael J. Thomson et al. Pujar A, Jaiswal P, Kellogg E, Ilic K, Vincent L, Avraham S, Stevens P, Zapata F, Reiser L, Rhee S, Sachs M, Schaeffer M, Stein L, Ware D, McCouch S (2006) Whole plant growth stage ontology for angiosperms and its application in plant biology. Plant Physiol 142:414–428 Rakshit S, Rakshit A, Matsumura H, Takahashi Y, Hasegawa Y, Ito A, Ishii T, Miyashita NT, Terauchi R (2007) Large scale DNA polymorphism study of Oryza sativa and O. rufipogon reveals the origin and divergence of Asian rice. Theor Appl Genet 114:731–743 Remington DL, Thornsberry JM, Matsouka Y, Wilson LM, Whitt SR, Doebley J, Kresovich S, Goodman MM, Buckler ES 4th (2001) Structure of linkage disequilibrium and phenotypic associations in the maize genome. Proc Natl Acad Sci USA 98:11479–11484 Ren ZH, Gao JP, Li LG, Cai XL, Huang W, Chao DY, Zhu MZ, Wang ZY, Luan S, Lin HX (2005) A rice quantitative trait locus for salt tolerance encodes a sodium transporter. Nat Genet 37:1141–1146 Sahi C, Singh A, Kumar K, Blumwald E, Grover A (2006) Salt stress response in rice: genetics, molecular biology and comparative genomics. Funct Integr Genomics 6:263–284 Salvi S, Tuberosa R (2005) To clone or not to clone plant QTLs: present and future challenges. Trends Plant Sci 10:297–304 Sarkar R, Reddy JN, Sharma SG, Ismail AM (2006) Physiological basis of submergence tolerance in rice and implications for crop improvement. Curr Sci 91:899–906 Sax K (1923) The association of size differences with seedcoat pattern and pigmentation in Phaseolus vulgaris. Genetics 8:552–560 Septiningsih EM, Pamplona AM, Sanchez DL, Maghirang-Rodriguez R, Neeraja CN, Vergara GV, Heuer S, Ismail AM, Mackill DJ (2009) Development of submergence-tolerant rice cultivars: the Sub1 locus and beyond. Ann Bot 103:151–160 Shimizu T, Satoh K, Kikuchi S, Omura T (2007) The repression of cell wall- and plastid-related genes and the induction of defense-related genes in rice plants infected with rice dwarf virus. Mol Plant Microbe Interact 20:247–254 Singh S, Mackill DJ, Ismail AM (2009) Responses of SUB1 rice introgression lines to submergence in the field: Yield and grain quality. Field Crops Res 113:12–23 Sweeney MT, Thomson MJ, Pfeil BE, McCouch SR (2006) Caught red-handed: Rc encodes a basic helix-loop-helix protein conditioning red pericarp in rice. Plant Cell 18:283–294 Sweeney MT, Thomson MJ, Cho YG, Park YJ, Williamson SH, Bustamante CD, McCouch SR (2007) Global dissemination of a single mutation conferring white pericarp in rice. PLoS Genet 3:e133 Szalma SJ, Buckler ES, Snook ME, McMullen MD (2005) Association analysis of candidate genes for maysin and chlorogenic acid accumulation in maize silks. Theor Appl Genet 110:1324–1333 20 Marker Assisted Breeding Tanksely SD (1993) Mapping polygenes. Annu Rev Genet 27:205–233 Temnykh S, DeClerck G, Lukashova A, Lipovich L, Cartinhour S, McCouch SR (2001) Computational and experimental analysis of micro-satellites in rice (Oryza sativa L.): Frequency, length variation, transposon associations and genetic marker potential. Genome Res 11:1441–1452 Tenaillon MI, Sawkins MC, Long AD, Gaut RL, Doebley JF, Gaut BS (2001) Patterns of DNA sequence polymorphism along chromosome 1 of maize (Zea mays ssp. mays L.). Proc Natl Acad Sci USA 98:9161–9166 Thoday JM (1961) Location of polygenes. Nature 191: 368–370 Thornsberry JM, Goodman MM, Doebley J, Kresovich S, Nielsen D, IV BES (2001) Dwarf8 polymorphisms associate with variation in flowering time. Nat Genet 28: 286–289 Toojinda T, Siangliw M, Tragroonrung S, Vanavichit A (2003) Molecular genetics of submergence tolerance in rice: QTL analysis of key traits. Ann Bot 91:43–253 Weir B (1996) Genetic Data Analysis II. Sinauer Associates, Sunderland, MA, pp 376 Wilson LM, Whitt SR, Ibanez-Carranza AM, Goodman MM, Rocheford TR, Buckler ES (2004) Dissection of maize kernel composition and starch production by candidate gene association. Plant Cell 16:2719–2733 Wissuwa M, Ae N (2001a) Genotypic variation for tolerance to phosphorus deficiency in rice and the potential for its exploitation in rice improvement. Plant Breed 120:43–48 View publication stats 469 Wissuwa M, Ae N (2001b) Further characterization of two QTLs that increase phosphorus uptake of rice (Oryza sativa L.) under phosphorus deficiency. Plant Soil 237:275–286 Wissuwa M, Yano M, Ae N (1998) Mapping of QTLs for phosphorus-deficiency tolerance in rice (Oryza sativa L.). Theor Appl Genet 97:777–783 Wissuwa M, Wegner J, Ae N, Yano M (2002) Substitution mapping of Pup1: a major QTL increasing phosphorus uptake of rice from a phosphorus-deficient soil. Theor Appl Genet 105:890–897 Xu K, Mackill DJ (1996) A major locus for submergence tolerance mapped on rice chromosome 9. Mol Breed 2:219–224 Xu K, Xu X, Fukao T, Canlas P, Maghirang-Rodriguez R, Heuer S, Ismail AM, Bailey-Serres J, Ronald RC, Mackill DJ (2006) Sub1A is an ethylene-response-factor-like gene that confers submergence tolerance to rice. Nature 442:705–708 Yeo AR, Flowers TJ (1986) Salinity resistance in rice (Oryza sativa L.) and a pyramiding approach to breeding varieties for saline soils. Aust J Plant Physiol 13: 161–173 Zhang N, Xu Y, Akash M, McCouch S, Oard J (2005) Identification of candidate markers associated with agronomic traits in rice using discriminant analysis. Theor Appl Genet 110:721–729 Zhu JK (2003) Regulation of ion homeostasis under salt stress. Curr Opin Plant Biol 6:441–445