Plant Cell, Tissue and Organ Culture (2005) 82: 317342 DOI 10.

1007/s11240-005-2387-z

Springer 2005

Marker assisted selection in crop plants

` E. Francia, G. Tacconi, C. Crosatti, D. Barabaschi, D. Bulgarelli, E. DallAglio & G. Vale*
C.R.A Istituto sperimentale per la Cerealicoltura, Sezione di Fiorenzuola dArda, Via S. Protaso 302, 29017 Fiorenzuola dArda (PC), Italy (*requests for oprints; Fax: +39-0523-983750; E-mail: gp.vale@iol.it)
Received 27 May 2004; accepted in revised form 15 February 2005

Key words: crop improvement, genetic mapping, genome analysis, marker assisted selection, PCR-based markers, QTLs, synteny

Abstract Genetic mapping of major genes and quantitative traits loci (QTLs) for many important agricultural traits is increasing the integration of biotechnology with the conventional breeding process. Exploitation of the information derived from the map position of traits with agronomical importance and of the linked molecular markers, can be achieved through marker assisted selection (MAS) of the traits during the breeding process. However, empirical applications of this procedure have shown that the success of MAS depends upon several factors, including the genetic base of the trait, the degree of the association between the molecular marker and the target gene, the number of individuals that can be analyzed and the genetic background in which the target gene has to be transferred. MAS for simply inherited traits is gaining increasing importance in breeding programs, allowing an acceleration of the breeding process. Traits related to disease resistance to pathogens and to the quality of some crop products are oering some important examples of a possible routinary application of MAS. For more complex traits, like yield and abiotic stress tolerance, a number of constraints have determined severe limitations on an ecient utilization of MAS in plant breeding, even if there are a few successful applications in improving quantitative traits. Recent advances in genotyping technologies together with comparative and functional genomic approaches are providing useful tools for the selection of genotypes with superior agronomical performancies. Abbreviations: G E genotype environment; LD linkage disequilibrium; MAS marker assisted selection; NILs near isogenic lines; Q E QTL environment; RILs recombinant inbred lines

Introduction Plant breeding, in its conventional form, is based on phenotypic selection of superior genotypes within segregating progenies obtained from crosses. Application of this methodology often encountered diculties related principally to genotype environment (G E) interactions. In addition, several phenotyping procedures are often expensive, time consuming or sometimes unreliable for particular traits (i.e. for some traits related to abiotic stress tolerance).

Molecular marker-assisted selection (MAS) is an approach that has been developed to avoid the problems connected with conventional plant breeding changing the selection criteria from selection of phenotypes towards selection of genes, either directly or indirectly. Molecular markers are clearly not environmentally regulated and are unaected by the conditions in which the plants are grown and are detectable in all stages of plant growth. With the availability of an array of molecular markers and genetic maps, MAS has become possible both for traits governed by

318 major genes as well as for quantitative trait loci (QTLs). Usefulness of a given molecular marker is dependent from its capability in revealing polymorphisms in the nucleotide sequence allowing discrimination between dierent molecular marker alleles. These polymorphisms are revealed by molecular techniques such as restriction fragment length polymorphisms (RFLP), amplied fragment length polymorphisms (AFLP), microsatellite or simple sequence length polymorphisms (SSR), random amplied polymorphic sequences (RAPD), cleavable amplied polymorphic sequences (CAPS), single strand conformation polymorphisms (SSCP), single nucleotide polymorphisms (SNPs) and other (Mohan et al., 1997; Rafalski, 2002). A successful application of molecular markers to assist breeding procedures rely on several factors: a genetic map with molecular markers linked to the major gene(s) or QTLs of agronomic interest; a tight association between the markers and the major gene(s) or the QTLs; adequate recombinations between the markers associated to the trait(s) of interest and the rest of the genome and the possibility of analyzing a large number of individuals in a time and cost eective manner. The success of MAS also depends on the localization of the marker with respect to the target gene. In a rst case, the molecular marker can be located directly within the gene of interest. This kind of relationship is clearly the most favorable and in most cases requires the availability of the target gene cloned. This situation is now possible for several disease resistance genes and for genes related to quality traits in tomato (discussed below). In a second case, the marker is genetically associated to the trait of interest. In this case lower is the genetic distance between the marker and the gene and more reliable is the application of the marker in MAS because only in few cases the selected marker allele will be separated from the desired trait by a recombination event. Most of the successful applications of MAS discussed below rely on this class of molecular markers. In a third case, the target gene(s) can be represented by one or more QTLs. In this case genomic regions to be selected are often chromosome segments; it is therefore preferable either to have two polymorphic markers anking the target QTL, and/or one or more markers within the QTL genomic region. Some examples related to both approaches for the introgression of QTLs into dierent genetic backgrounds are discussed below. To date, MAS has shown to be eective for relatively simple traits that are controlled by a small number of genes and several examples are provided in this review on the advantages and on the successful application of MAS for this class of traits. For more complex traits, MAS has proven to be less eective; several reasons at the basis of these diculties, some examples of successful application of MAS for quantitative traits and perspectives for increasing eciency of MAS for QTLs are discussed. Marker assisted selection aims For plant breeders, the most useful application of MAS is to use DNA-based markers for basically three purposes: tracing favorable allele(s) (dominant or recessive) across generations; in order to accumulate favorable alleles, identifying the most suitable individuals among segregating progenies, based on the allelic composition of a part or of the entire genome and breaking the possible linkage of favorable alleles with undesirable loci. When the expression of a target trait is regulated by a single gene, or by a gene responsible for a high percentage of the phenotypic variance of the trait, the transfer of a single genomic region from a donor to a recipient line can produce signicant trait improvement. MAS is now increasingly employed for accelerating the recovery of the recurrent parent in backcross (BC) programmes. Compared with conventional backcrossing, the use of molecular markers can improve the eciency of BC breeding at least in three ways: for traits that are dicult to phenotype, selection for a marker allele from the donor parent at a locus near the target gene can increase the eciency and accuracy of selection; markers can be used to select BC progeny with less amount of donor parent germplasm in the genome outside the target region and to select rare progenies resulting from recombination near the target gene, thus minimizing the eects of linkage drag and in the transfer of recessive genes through conventional breeding, additional selng genera-

319 tions after every backcross are required, leading to a procedure that is prohibitively low for most breeding purposes. The probability of selecting superior genotypes is low for low to moderate heritability. In classical breeding, plant breeders cope with this problem by producing and testing progeny from numerous crosses, using low selection pressure, using replicate testing and testing advanced generations. Breeders selecting for low to moderate heritability traits have the following dilemmas: when the heritabilities of the traits to be selected are low or moderate and small samples of progeny are tested, the probability of selecting an outstanding genotype is very low; large numbers of progeny must be selected (low selection intensities must be used) to ensure the presence of one or more superior genotypes in the selected sample and even when low selection intensities are used, the most outstanding genotypes produced by a cross might not be present in the selected sample when heritability is low and samples are small. MAS has therefore emerged as a strategy for increasing selection gains with respect to phenotypic selection alone and quantitative genetic theory suggests that the eectiveness of MAS is inversely proportional to the heritability of a given trait (Lande and Thompson, 1990; Knapp, 1998). Knapp (1998) developed a theory to estimate the probability of selecting one or more superior genotypes by MAS and dened a parameter to estimate the cost eciency of MAS relative to phenotypic selection. Depending on the selection pressure, the genotypic superiority target and the trait heritability, it is estimated that a breeder using phenotypic selection must test 1.0 to 16.7 times more progeny than a breeder using MAS to be assured of selecting one or more superior genotypes. Thus, MAS can substantially decrease the resources needed to accomplish a selection goal for a low to moderate heritability trait when the selection goal and the selection intensity are high. The parameter dened by Knapp (1998) predict that MAS is most ecient than phenotypic selection when breeders use high selection intensities and set high selection goals. Selection intensity can be increased to exclude inferior genotypes when the heritability is increased by using MAS. An empirical comparison on the eciency of MAS and phenotypic selection for the simultaneous improvement of quantitative traits involved in seedling emergence and eating-quality characteristics has been carried out in three different sweet corn populations (Yousef and Juvik, 2001). In this study, selection of QTLs aecting seedling emergence, kernels sucrose concentration and kernels tenderness was assessed by using ve molecular markers per QTL. MAS and phenotypic selection have then been compared for the selection gain in each of the three single traits and for the selection gain in the multiple traits (i.e. emergence plus kernel tenderness): in both cases, MAS was superior to phenotypic selection. Selection eciencies were therefore evaluated on the basis of gains over one cycle of selection and estimated evaluation costs. A total of 52 paired comparison were made between MAS and phenotypic selection; in 38% of the paired combinations MAS resulted in signicantly higher gains than phenotypic selection across the three populations, while phenotypic selection was signicantly grater than MAS in only 4%. Even if the estimates of the MAS and phenotypic selection costs provided in this study may not be applicable to every breeding scheme, it appears that MAS can economically compete with phenotypic selection, particularly with the advances in DNA technology and the gains resulting from reduced size and duration of breeding programs. Theoretical and analytical investigations (Lande and Thompson, 1990; Wittaker et al., 1995) have shown that the maximum selection eciency for quantitative traits may be obtained by using a combination of molecular and phenotypic information. For maize grain yield and percent stalk lodging, an index including both phenotypic and marker information predicted performancies of testcrosses better than phenotypic or marker information alone. For grain yield, combining marker information with phenotypic information allowed a reduction of 40% in the number of lines to be tested (Eathington et al., 1997). Thus, DNA markers could allow the identication of high performing genotypes in early generations and therefore have a strong impact on breeding programs by reducing the number of lines that have to be tested and therefore accelerating the breeding process (Eathington et al., 1997; Schneider et al., 1997; Frisch et al., 1999).

320 MAS for improvement of quantitative traits Most of the traits of agronomic importance, such as yield, some classes of disease resistance genes, several abiotic stress tolerance genes and quality traits, are complex and regulated by several genes. Diculties in manipulating these traits are derived from their genetic complexity, principally the number of genes involved, the interactions between genes (epistasis) and environment-dependent expression of genes. Quantitative traits often have a low heritability, with many QTLs segregating for the trait, each with small eect individually. The result is that eects of individual regions are not easily identied, and multiple genomic regions must be manipulated at the same time in order to have a signicant impact. For this reason, replicates of eld tests are required to characterize accurately the eects of QTLs and to evaluate their stability across environments. Although signicant QTL eects should be detected across several environments, variation in expression due to QTL by environmental interactions (Q E) remains a major constraint to the discovery of QTL that will confer a consistent advantage across a wide range of environments; on the other side the identication of G E as well as Q E eects, it may permit identication of genotypes adapted to specic environments (Fox et al., 1997). Given this complexity, integrated approaches are therefore required to increase the probability of an useful application of MAS for QTLs (Figure 1). In fact, despite the proliferation of QTL mapping works in recent years, a number of constraints have determined severe limitations on an ecient utilization of QTL mapping information in plant breeding through MAS. These constraints include the identication of major QTLs controlling the trait of interest; uncertainty of the QTL position, notably for those with a small eect (the condence interval for QTLs location determined with current QTL analysis techniques sometimes is up to 30 cM for small populations); deciencies in QTL analysis leading either to an overestimation or underestimation of the number and eects of QTLs; problems connected with the identication of QTL-marker associations applicable over different sets of breeding materials;
Molecular marker-based genetic maps Phenotypic evaluation of agronomic traits in multi-environments

QTL analysis of input traits Identification of markers linked to QTLs

epistasis

G x E effects

or

Additive effects

Accomodation of the effects of epistasis and GxE

association mapping (haplotype analysis) Estimation of the QTL allele effects

Definition of a target combination of markers alleles

MAS application in breeding programs

Figure 1. An integrated approach for a possible exploitation of QTL data in MAS.

possibility of loosing the target QTL during MAS through double-cross-overs between markers (this possibility is increasing with increased length of the marker interval analyzed); diculty in precisely evaluating epistatic eects and diculty in evaluating Q E interactions. Improved eld designs (Gleeson, 1997) and statistical approaches for QTL can led to a better characterization of the target genes map position. In recently devised mathematical methods, such as composite interval mapping (CIM), eld data from dierent environments can be integrated into a joint analysis to evaluate Q E and thus identify QTLs that are stable across environments (Jang and Zeng, 1995). Besides, with a detailed linkage map, CIM allow a better identication of linked QTLs (in coupling phase) from the same parental line. In addition, analysis methods have been proposed to accommodate QTL mapping data for

321 the eects of G E interactions (Crossa et al., 1999), of epistasis (Boer et al., 2002) and of G E interactions and epistasis at the same time (Podlich et al., 2004). It is possible that such integrated approaches would better allow estimation of QTL eect for MAS application in breeding programs. An emerging approach for identifying genes underlying phenotypic variation in complex traits is represented by linkage disequilibrium (LD i.e. nonrandom association between alleles at linked loci) mapping (association mapping). Linkage mapping carried out on a bi-parental cross, samples only a small fraction of the possible alleles in the population from which the parents originated. LD, on the contrary, infers associations between genotypes (haplotypes) and phenotypic variation by examining genetic polymorphisms that have been generated into dierent genetic backgrounds through many past generations of recombination ` (Nordborg and Tavare, 2002). Because no mapping populations need to be created, the key advantages is that association tests can be performed relatively quickly and inexpensively. In LD mapping, a whole genome may be scanned to identify regions that are associated with a particular phenotype, or alleles at a few selected candidate genes may be tested for association with a phenotype (Rafalski, 2002). The primary obstacle to association studies is represented by the presence, within the population, of subgroups with an unequal distribution of alleles. In such populations, false-positive associations can be detected between a marker and a phenotype, even if the marker is not physically linked to the locus responsible for the phenotypic variation (Buckler and Thornsberry, 2002). Thus, a crucial rst step in LD mapping is to dene a set of subpopulations. An eective method controlling for population structure in association tests, uses a genetic similarity matrix estimated from molecular marker data. This method was successfully implemented on a maize population consisting of three subpopulations, reducing the number of false positive (Thornsberry et al., 2001). The extent of LD is aected by many factors, including population history and the frequency of recombination in the examined genome segment (Rafalski, 2002). In plant species where population bottlenecks have occurred, such as sugarcane and sugar beet, LD extending for several cM was found in RFLP and AFLP marker analysis (Jannoo et al., 1999; Kraft et al., 2000). Similarly, data available for soybean, which also has a very narrow genetic base, show that LD decays at distances of 2.02.5 cM (Zhu et al., 2003). Maize studies have shown that LD decays over 1500 bp distances, but other studies have also shown that LD values may vary considerably (Morgante and Salamini, 2003). LD mapping has successfully be applied in the model plant Arabidopsis (Olsen et al., 2004) and maize (Thornsberry et al., 2001) to identify QTLs associated with variation in owering time. Association methods that incorporate estimates of population structure is therefore providing a powerful approach to identifying alleles responsible for variation in a variety of quantitative traits.

MAS for yield Although some investigations provided examples on the pratical application of MAS to increase yield, it is becoming clear that integrated approaches involving traditional methods of agricultural improvement (Lande and Thompson, 1990) and combination of crop modeling and QTL mapping (Figure 1; Yin et al., 2003) are required to select a crop ideotype for a given environment. In the following sections some selected examples of successful applications of MAS to increase yield in important crop plants like maize, rice, barley and soybean are provided. Maize Marker-mediated backcrossing is a selection scheme used in maize to monitor the transfer of favorable alleles at QTLs (foreground selection) and to hasten the return to the recipient genotype in the remainder of the genome (background selection) (Bouchez et al., 2002; Figure 2). A similar approach of marker-mediated backcrossing has been used to generate series of maize NILs derived from an elite recipient line (the recurrent line) and a exotic donor line (Stuber et al., 1999). Marker facilitated backcrossing and markerfacilitated selng were used for foreground and background selection. As few as two BCs and one selng (to x the introgressed segment) generations were sucient to generate dierent NILs, each with dierent introgressed genomic regions. When crossed to a tester line and evaluated in

322 gene action might prevail for some loci aecting grain yield) the heterozygous QTL genotype would not be the most favorable. For this reason, an eective prediction of hybrid performance based on markers solely would requires knowledge of QTLs linked to the markers. Rice QTL alleles for yield components traits derived from the wild rice relative Oryza rupogon have recently been extensively studied by using advanced BC populations (AB-QTL; Tanksley and Nelson, 1996). In these studies, despite its inferior performance, 53% (Thomson et al., 2003) and 33% (Septiningsih et al., 2003) of the QTLs alleles originating from O. rupogon had a benecial eect for yield and yield components in the recipient rice elite cultivars. The lower percentage reported in the second study may be explained by an higher genetic similarity between the elite line and O. rupogon at the yield QTL alleles or by the fact that in this cross the elite cultivar may have more favorable alleles at most of the identied loci. Some of the O. rupogon yield QTLs identied were not linked to any deleterious negative QTLs and would directly be useful to develop breeding materials. In several dierent instances, the O. rupogon alleles showed the same eect in different genetic backgrounds and environments, supporting the stability of these yield QTLs. A thousand grain weight (TGW) QTL has recently been identied on chromosome 6 by using BC inbred lines derived from a cross between the high-yield rice japonica cv. Nipponbare and the low-yield indica cv. Kasalath (Ishimaru, 2003). The QTL allele increasing TGW is derived from the low yield cv., and when introgressed by MAS into a Nipponbare NIL, this QTL increase TGW and yield per plant by 10 and 15% respectively without any eect on plant type. The genomic region in which this yield QTL is located is tagged by several molecular markers that can be be used to introgress this QTL to increase yield in high yielding rice cultivars. Barley The transfer of high grain yield QTL alleles on chromosomes 2 and 3 from the widely adapted cv. Baronesse to the high malting quality cv. Harrington has involved a marker-assisted selec-

Figure 2. Marker-assisted introgression scheme of favorable alleles at QTL applied in maize by Bouchez et al. (2002).

replicated eld trials, dierent NILs revealed to have received donor segments increasing or decreasing their yield performancies. This breeding scheme not only creates enhanced elite lines, but also provide materials for the identication and mapping of yield QTLs. A possible disadvantage of this approach is that favorable epistatic eects between QTLs may not be identied. Development of a reliable method for predicting hybrid performance in maize, without generating and testing hundreds or thousands of single-cross combinations, has been the goal of numerous studies, using both marker data and combinations of marker and phenotypic data. In order to explore heterosis (hybrid vigor) and G E interaction, Stuber et al. (1992) used a cross between two widely used elite maize inbreds, B73 and Mo17. They identied and mapped QTL alleles that were predicted to increase hybrid yield. Markers were used to introgress the QTLs into the inbred lines, and the hybrids from the enhanced inbred lines yielded better than hybrids from inbred lines that lacked the marker-introgressed QTLs (Stuber, 1994). Whenever a QTL for grain yield was detected, the heterozygote had a higher phenotype than the respective homozygote (with only one exception) suggesting not only overdominance (or pseudo-overdominance) but also that these detected QTLs play a signicant role in heterosis. This conclusion was reinforced by a high correlation between grain yield and proportion of heterozygous markers. However, for traits governed largely by additive gene action (this type of

323 tion scheme (Schmierer et al., 2004). Several BC3 Harrington isolines have displayed yields equal to Baronesse across test sites and years, and in most cases the malting quality of Harrington has been maintained or improved. In this work the chromosome 3 QTL seems to be more eective on yield than the chromosome 2 QTL. Soybean The limited diversity in elite soybean germplasm has prompted an interest in evaluating G. soja, the wild ancestor of soybean, as a new source of genetic diversity for improving the crop. Concibido et al. (2003) mapped a QTL allele from G. soja PI 407305 that was associated with 12% yield increase across testing environments. The QTL allele from G. soja was then introgressed into six genetic backgrounds by marker-assisted BC to assess the adaptability of the G. soja yield QTL across genetic backgrounds. The ecacy of the yield QTL was limited to two out of the six backgrounds, demonstrating the potential of using exotic germplasm to improve soybean yield, at least in some genetic backgrounds. Also Wang et al. (2004) used G. soja as a potential source of useful genetic variation for soybean improvement. They mapped four QTL for yield across environment in a series of BC populations developed using a G. soja line as a donor parent and a soybean cultivar as a recurrent parent. For these yield QTLs, the marker allele of G. max cultivar was responsible for the greater yield than the marker allele from G. soja. tions of resistance-segregating breeding populations. Phenotypic assay to assess resistance to soybean cyst nematode (SCN) in soybean or to leaf stripe in barley takes more than 5 weeks and extensive greenhouse space. Identication of reliable PCR-based markers for MAS of the resistant phenotype has therefore led to an eective improvement of the resistance breeding procedure against these soybean (Cregan et al., 2000) and barley (Arru et al., 2003) diseases. In addition, MAS can provide specic advantages in resistance breeding, allowing faster response to a breakdown in resistance, rapid introgression of multiple genes from diverse germplasm, pyramiding, and selection of rare recombinants between tightly linked resistance genes (Michelmore, 2003). Developing molecular markers Markers linked to a given R gene have been obtained by Bulk Segregant Analysis (BSA Michelmore et al., 1991), or by using fully mapped populations. Although the BSA approach to nding a linked marker in barley has been estimated to be approximately one-third the cost of using fully mapped populations, BSA is limited to traits controlled by one or two major genes (Barr et al., 2000). New tools are providing assistance in determining the map position of R genes; arrays of contiguous genomic Bacterial Articial Chromosomes (BAC) clones are becoming available for rice, corn and soybean. Hybridization to such contigs can provide a rapid and accurate method for mapping cloned sequences and can replace segregation analysis; such a procedure has the additional advantage that it does not require polymorphism between the parents of a mapping population (Michelmore, 2000). This approach has recently been used to map the gall midge R gene Gm7 in rice: an AFLP fragment associated to Gm7 has been used as a probe to screen rice BAC and YAC genomic libraries and the identication of a hybridizing YAC clone has allowed genetic and physical mapping of the R gene (Sardesai et al., 2002). R genes against a variety of dierent pathogens have now been cloned from a variety of crop plants (Baker et al., 1997; Hammond-Kosack and Parker, 2003). Most of these genes have been isolated through a marker-facilitated approach called

MAS in breeding for disease and pest resistance Most of the successful applications of MAS to plant breeding have been those for major disease resistance (R) genes. The introgression of an R gene into an elite breeding line by traditional breeding can take up to 1015 years. It is also complicated by the need of performing timeconsuming and labor-intensive articial inoculation tests to assess the resistant phenotype, which requires maintenance of the pathogens or the pests on the host (or alternate hosts) if they are obligate parasites. Molecular markers tightly linked to R genes can obviate the need for resistance testing to identify resistant individuals from early genera-

324 as map-based or positional cloning. The method rely on the development of large populations segregating for the trait of interest; utilization of markers anking the target gene to screen these populations in order to identify rare individuals derived from one or more gametes containing a crossover in the given interval; analysis of these individuals in relation to all the markers within the region of interest and identication of molecular markers tightly associated or co-segregating with the target gene. These markers are suitable for the screening of a large insert genomic library (BAC, PAC or YAC). In theory, the ideal marker would be the one lying at a physical distance from the target gene that is less than the average insert size of the genomic library from which one expects to isolate the gene. The identication of such markers would avoid the chromosome walking steps via overlapping clones, a long and problematic process, by allowing a direct landing on the large insert genomic clone containing the target gene (Tanksley et al., 1995). Development and mapping of new molecular markers derived from sequencing of the ends of the BAC or YAC clones would ensure if the clones identied contain the target gene. Modications and optimizations of the method have allowed the isolation of several R genes from dierent plant species (e.g. Buschges et al., 1997; Zhou et al., 2001; Feuillet et al., 2003; Song et al., 2003). The vast majority of these cloned R genes encode proteins belonging to the nucleotide-binding/ leucine-rich repeat (NB-LRR), extracellular LRR (eLRR) or LRR-Kinase superfamilies (HammondKosack and Parker, 2003). Because sequences of the cloned R genes and of the immediately surrounding genomic regions (frequently including ESTs) are available, precise (R-gene derived) or physically close molecular markers for these R genes can be easily generated. In addition to known resistance genes, plant genomes contain hundreds of other NBS-LRR encoding genes, called Resistance Gene Analogs (RGAs). Analysis of the Arabidopsis and rice genomes has revealed a content of about 150 and 600 predicted NBS-LRR encoding genes, respectively (The Arabidopsis Genome Initiative, 2000; Go et al., 2002). RGAs may represent undiscovered resistance genes of complete or partial eect. Alternatively, they may perform some function unrelated to resistance (Madsen et al., 2003). RGA sequences identied by a PCR-based approach often map to regions containing known disease resistance genes, as revealed for several crop plants including soybean, potato, maize, barley and tomato (for references see Pan et al., 2000; Madsen et al., 2003). Since RGAs that map close genetically often reect physical proximity (Leister et al., 1999; Wei et al., 2002) markers derived from RGA sequences could also be useful in MAS of R genes during breeding. For instance, an RGA polymorphic marker was recently used to develop a SCAR marker co-segregating with the BYDV resistance gene Bdv2 in wheat and this marker has successfully been used to assess BYDV resistance in a wheat breeding program (Zhang et al., 2004).

Assisted breeding for resistance Several plant disease resistance genes have now been tagged by molecular markers (selected examples are reported in Table 1) and examples of practical application of MAS are now available. Genes conferring resistance to the two most destructive rice diseases, bacterial blight caused by Xanthomonas oryzae and blast caused by the fungus Magnaporthe grisea, have been mapped (Mohan et al., 1997). A MAS assisted BC program has recently been used to pyramid into the high yielding rice cultivar PR106, three bacterial blight resistance genes (Xa21, xa13 and xa5) conferring resistance to all the bacterial races present in the specic cultivation area (Singh et al., 2001). The broad-spectrum blast resistance gene Pi5(t) in rice, derived from the African cultivar Moroberekan, has recently been genetically and physically mapped (Jeon et al., 2003). Analysis of BAC end sequences totalling around 70 kb has revealed a cluster of NBS-LRR sequences at the Pi5(t) locus that may constitute a natural pyramid of resistance genes responsible for the broad-spectrum resistance. A marker-assisted approach based on RFLP and PCR markers has successfully been used to pyramid three major genes for blast resistance (Pi1, Piz5 and Pita); when tested with isolates virulent towards a single R gene, the pyramided lines showed enhanced resistance, demonstrating a complementary eect of the three R genes when present together (Hittalmani et al., 2000).

325
Table 1. Selected examples of gene-marker association for disease resistance in dierent crops Disease/pest Wheat Leaf rust (Puccinia recondita f.sp. tritici) R gene Molecular markers Note References

Stem rust (Puccinia graminis f.sp. tritici) Powdery mildew (Blumeria graminis f.sp. tritici)

Lr34 from T. aestivum Lr 35 from T. speltoides Lr47 from T. speltoides Sr31 Qpm.vt-1A, Qpm.vt-2A, Qpm.vt-2B Pm4a Pm5e Qfhs.ndsu-3BS from T. aestivum cv Sumai3 Yr15 YrMoro

SSR STS and CAPS CAPS STS markers SSRs

Lr34 expresses resistance in a quantitative way Adult plant leaf rust R gene R to a wide spectrum of leaf rust strains Broad spectrum R gene Quantitative resistance from T. aestivum cv. Massey, eective since 1981

Suenaga et al. (2003) Seyfarth et al. (1999) Helguera et al. (2000) Mago et al. (2002) Liu et al. (2001)

Fusarium head blight (Fusarium graminearum) Yellow rust (Puccinia striiformis f.sp. tritici) Rice Bacterial blight (Xanthomonas oryzae pv. oryzae)

CAPS SSR RFLPs

RAPD and SSR STS STS

Sumai is the principal head blight resistance source in wheat R to a wide spectrum of yellow rust strains

Ma et al. (2004) Huang et al. (2003) Anderson et al. (2001)

` Chague et al. (1999) Smith et al. (2002)

xa5

Three bacterial blight R genes were pyramided in the high yielding susceptible cultivar PR106

Singh et al. (2001)

Rice blast (Pyricularia oryzae)

xa13 Xa21 Pi5(t) Pi1, Piz5, Pita Piz, Pizt Pi-b, Pi-k, Pita2

STS STS CAPS RFLP and SAP PCR-based SNP SSRs SA598 SCAR STS

Broad-spectrum resistance Pyramiding of three blast R genes

Jeon et al. (2003) Hittalmani et al. (2000) Hayashi et al. (2004) Fjellstrom et al. (2004) Sardesai et al. (2002) Murai et al. (2001)

Gall midge (Orseolia oryzae) Brown planthopper Barley Barley yellow mosaic virus

Gm7 bph2

rym4/rym5

SSR

rym4 and rym5 on chromosome 3H, are allelic with respect to BaMMV Eective in Europe Broadly eective

Graner et al. (1999) Williams (2003)

Leaf rust (Puccinia hordei)

rym4, rym9, rym11 Rph7 Rph15

SSRs CAPS CAPS

Werner et al. (2003) Graner et al. (2000) Weerasena et al. (2004)

326
Table 1. (Continued) Disease/pest Maize Sugarcane mosaic virus (SCMV) Sugar beet Rhizomania (BYNVV ) R gene Molecular markers Note References

Scm1 and Scm2

SCAR and CAPS

Finely mapped to maize chromosomes 6 (Scm1) and 3 (Scm2) F6 is linked in coupling with the Rr1 allele; dominant allele of N9 is linked to the Rr1 allele in repulsion QTL on chromosome 2 is the most eective A recessive gene conferring R to a soil-borne fungus Incompletely dominant gene

Dussle et al. (2002)

Rr1 introgressed from Beta maritima

SCAR F6 SCAR N9

Barzen et al. (1997)

Tomato Black mold (Alternaria alternata) Corky root rot (Pyrenochaeta lycopersici) Powdery mildew (Oidium lycopersicum) Root knot nematodes (Meloydogyne spp.)

QTLs introgressed from L. cheesmanii py-1 introgressed from L. peruvianum Ol-1 introgressed from L. hirsutum Mi introgressed from L. peruvianum Mi3 introgressed from L. peruvianum Vf introgressed from Malus Floribunda 821 Vm introgressed from M. micromalus Pl1

CAPS and RFLP CAPS

Robert et al. (2001) Doganlar et al. (1998)

SCAR RAPD RAPD and RFLP

Huang et al. (2000) Williamson et al. (1994) Yaghoobi et al. (1995)

Apple Scab (Venturia inaequalis)

Eective against most ALO7-SCAR of the pathogen races cosegregating with M19-CAPS and M18-CAPS STS SCAR

Tartarini et al. (1999) King et al. (1999)

Cheng et al. (1998) Kellerhals et al. (2000)

Mildew (Podosphaera leucotricha) Grapevine Powdery mildew (Uncinula necator)

Run1 introgressed from Muscardinia rotundifolia

AFLP, CAPS, SCAR

Eective in the eld against the most frequent genotypes of the fungus in Bordeaux and Montpellier

Pauquet et al. (2001) Donald et al. (2002)

For pathogens that can easily generate new virulent pathotypes, such as rusts and mildews, high and durable levels of resistance should be achievable by MAS. Of almost 50 known Lr resistance genes against the damaging leaf rust disease of wheat, the slow rusting genes Lr34 and Lr46 have been eective over a long period of time, in dierent environments and against dierent pathotypes of the fungus (Kolmer, 1996; Singh et al., 1998). Several studies have described

enhanced resistance eects derived from combinations of Lr34 with other Lr genes, such as Lr2, Lr12, Lr13, Lr16, which give hypersensitive resistance responses (Kolmer, 1996; Kloppers and Pretorius, 1997). Microsatellite markers linked to Lr34 have been identied (Suenaga et al., 2003). Molecular markers have also been developed for other leaf rust genes including Lr1, Lr9, Lr19, Lr24, Lr25, Lr28, Lr29, Lr32 and the highly eective Lr35 mediating a resistant hypersensitive

327 response (for references see Seyfarth et al., 1999; Huang and Gill, 2001), potentially allowing combinations of slow-rusting genes with others conferring an hypersensitive response. Resistance to the causative agent of the wheat powdery mildew disease (Blumeria graminis f. sp. tritici) is governed by both race-specic resistance genes and adult plant resistance (APR) genes. APR retards infection, growth and reproduction of the pathogen in adult plants but not in seedlings. Three adult plant powdery mildew resistance QTLs, which have been eective for more than 20 years, have been mapped in the wheat cultivar Massey (Liu et al., 2001). Recombinant inbred lines containing Masseyderived alleles at all three loci had a mean disease severity of 3.4%, whereas the RI lines with alleles of the susceptible parent at all three loci had a mean disease severity of 22.3%. Microsatellite markers identied as associated with the three QTLs have therefore potential for use in MAS for APR to powdery mildew, with or without concurrent selection for race-specic powdery mildew resistance genes. Molecular markers linked to major wheat powdery mildew R genes and suitable for MAS have been identied for Pm4a (Ma et al., 2004), Pm5e (Huang et al., 2003), as well as for other Pm genes (for references see Ma et al., 2004; Huang et al., 2003). Sources of resistance to some pathogens are often not present in the cultivated species but in related wild germplasm. Suppression of recombination with the introgressed segments can lead to linkage drag of genes with unwanted eects on the phenotype, persisting through many BCs. Molecular markers can be used to select for an introgression segment of minimum size in order to reduce or eliminate deleterious eects resulting from linkage drag. For example, accessions of Malus with a minimum chromosome segment from M. oribunda containing the scab resistance gene Vf have been selected (King et al., 1999). Similarly, MAS allowed selection of wheat-rye recombinant lines containing the highly eective rust (Lr26, Sr31/SrR, Yr9) and powdery mildew (Pm8) resistance genes from rye without the deleterious grain quality characteristics (sticky dough and a reduction in dough strength) of the closely linked rye Sec-1 locus (Mago et al., 2002). In grapevine (Vitis vinifera) the monogenic dominant powdery midew resistance gene Run1 has been introduced from the wild muscadine grape Muscadina rotundifolia. A map of the Run1 region was made using AFLP markers (Pauquet et al., 2001) and the markers identied are suitable for selecting good quality genotypes with the smallest M. rotundifolia genomic fragment containing Run1. Three AFLP markers were identied as being appropriate for subsequent MAS of Run1 in grapevine cultivars. In potato, introduction of resistance traits also relies mostly on wild germplasm, and is therefore frequently associated with linkage drag. During introgression of resistance to late blight (Phytophthora infestans) from Solanum bulbocastaneum, several resistant individuals showing the recurrent parent alleles at molecular markers on 6 or more chromosomes were identied already from the BC2 populations. These lines were selected as superior parental materials for subsequent BC generations (Naess et al., 2000). Genes conferring resistance to the most important potato diseases have been mapped and molecular markers to assist the selection of the resistant phenotype have been identied (reviewed in Gebhardt and Valkonen, 2001). Linkage drag problems are potentially greater when the resistant phenotype is governed by QTLs. No genetic resistance to blackmold has been reported in cultivated tomato (L. esculetum). QTLs for blackmold resistance have been introgressed into cultivated tomato from the wild species Lycopersicon cheesmanii using MAS mediated by RFLP and PCR-based markers anking and within the chromosomal regions containing the QTLs (Robert et al., 2001). A QTL on chromosome 2 provided a positive eect on blackmold resistance; other two QTLs were associate with increased level of resistance but associated to deleterious agronomical traits. Utilization of molecular markers is a valuable alternative means of diagnosis for assessing disease resistance genes when the phenotypic trait is governed by recessive or incompletely dominant genes. Tomato resistance to corky root rot and to powdery mildew is governed by the recessive py-1 (Doganlar et al., 1998) and the incompletely dominant Ol-1 (Huang et al., 2000) resistance genes respectively. Co-dominant CAPS markers associated to py-1 and dominant SCAR associated in coupling and repulsion phase to Ol-1 are allowing these genes to be incorporated into modern cultivars and speeding up breeding programmes for resistance to these two diseases.

328 The availability of molecular markers to select for resistance in long-generation cultivated tree or woody species can be a great advantage, especially for those species which are self-incompatible or have a long juvenile period. The fungus Venturia inaequalis is the causative agent of apple scab, the most important apple disease. Identied codominant PCR-based markers closely anking the Vf resistance gene allow the identication of seedling homozygous for the resistance gene and thereby make traditional test-cross experiments unnecessary (King et al., 1999; Tartarini et al., 1999). Similarly, a codominant STS marker for the peach Mij gene, which makes rootstocks resistant to the root-knot nematodes M. incognita and M. javanica, can allow the selection of homozygous resistant individuals (Lu et al., 1999). Molecular markers can also provide information on the virulence functions present in the pathogen population of a given cultivation area, and thereby help to assess the likelihood of the pathogen in overcoming the deployed R gene. For example, molecular markers diagnostic for the ability of root-knot nematodes of the genus Meloidogyne (M. incognita, M. javanica, M. arenaria) to overcome the tomato resistance gene Mi have been obtained (Xu et al., 2001). Similarly, at least three mutations associated with high levels of resistance to Bt toxins have been identied in insect pests (reviewed in Tabashnik, 2001). Typical bioassays are not eective in detecting the presence of Bt toxin resistance alleles when these are recessive or rare (as is most often the case). Discovery of genes underlying Bt resistance in key insect pests therefore enables a DNA-based monitoring strategy for these rare mutational events which will play an important role in managing insect resistance to Bt crops. In barley, two tightly linked QTLs for lowtemperature tolerance were identied on chromosome 5H (Francia et al., 2004); these QTLs were coincident with QTLs regulating mRNA levels as well as protein accumulation of two well characterized cold-regulated (COR) genes (Vagujfalvi et al., 2003; Francia et al., 2004 respectively). Several genes with the CBF transcription factor signature mapped in a cluster in this region. Since a CRT/DRE recognition site, a potential site for interaction with a CBF transcription factor, was found in the genomic regulatory sequence of one of the two COR genes, the identied CBF genes represent candidates for the gene underlying the QTL (Francia et al., 2004). Because it has been demonstrated that CBF1 overexpression induces COR genes and enhances freezing tolerance in Arabidopsis (Jaglo-Ottosen et al., 1998), these results support the hypothesis that members of the CBF gene family may regulate the stress responses of a wide range of plant species. PCR-based markers (a RAPD marker and an STS derived from the sequence of a wheat RFLP mapped in the frost tolerance QTLs region on chromosome 5H) have recently been validated for their ability in assessing frost tolerance level in two sets of winter and spring barley genotypes and in doubled-haploid lines derived from a cross between a highly ` tolerant and a susceptible genotype (Toth et al., 2004). These two markers were shown to discriminate eciently between frost-tolerant and frost-susceptible genotypes. Their use in dierent breeding materials will clarify how much would be the gain in frost tolerance obtained by the only MAS respect to phenotypic selection in stressing environments. Some stress-implicated genes have been shown to co-segregate with stress tolerance QTLs in crops. Two Dehydrin loci (Dhn1/Dhn2 and Dhn9) are located in the region of Triticeae chromosome group 5 known to contain QTLs for cold and salt tolerance and ABA accumulation (reviewed in Cattivelli et al., 2002). A potential application of these ndings has derived from a study on cowpea; in this plant the accumulation of the 35 kDa dehydrin was correlated with chilling tolerance during seedling emergence. Allelic variations in the coding region of the dehydrin structural gene map to the same position as the dehydrin protein presence/absence trait, which in turn is associated

MAS for low-temperature stress tolerance Besides a requirement for vernalization, overwintering crops also require frost and cold tolerance. Cold tolerance is recognized as having a complex quantitative inheritance, making therefore problematic MAS approaches to increase tolerance phenotypic values. Nevertheless, some few examples of successful utilization of MAS for improving cold tolerance in crop plants are available.

329 with chilling tolerance/susceptibility during seedling emergence (Ismail et al., 1999). Rice has evolved in tropical and subtropical areas, and hence its cultivation is vulnerable to low-temperature stress in temperate-growing regions and in high-elevation environments. Anthers at booting stage are known to be susceptible to low temperature, so cold stress results in delayed heading or maturation and yield reduction due to spikelet sterility. Abe et al. (2002) reported the tight association of a SNP in a rice alternative oxidase gene (OsAOX1a) with two closely linked QTLs (Ctb1 and Ctb2) for low temperature tolerance of anthers at the booting stage mapped to chromosome 4. They found that the allelic variation in molecular mass of AOX isoforms among varieties diering in low temperature tolerance co-segregate to the presence of the QTL. These results suggest that exploitation of this SNP represent a good tool for MAS of the cold tolerance QTL. Ctb1 locus has recently been physically mapped and seven candidate genes for this QTL have been identied (Saito et al., 2004). et al., 1991). An alternative strategy is represented by the selection of a range of morpho-physiological characters suggested to be indicators of increased grain yield under drought conditions. Even if it is dicult to identify traits that provide a consistent advantage on yield across variable water-limited environments, a catalogue of such traits has been proposed (Turner, 1997). In this sense, OA is believed to be important for allowing plants to maintain turgor and avoid meristem damage when faced with extreme desiccation; in addition, other mechanisms are probably also important for drought tolerance, including: the ability of the roots to exploit deep soil moisture to meet evapotranspirational demand, moderatation of water-use by reduction of leaf area and shortening of growth period, and the limitation of nonstomatal water loss from leaves (i.e. through the cuticle). A plethora of traits are therefore involved in increasing drought stress tolerance, making a molecular breeding approach for drought tolerance a complex task. In this section some selected examples are reported on the mapping and on the possible exploitation of traits involved in drought stress tolerance. Osmotic adjustment When cells are subjected to slow osmotic stress, compatible solutes are accumulated resulting in the maintenance of a higher turgor potential at a given leaf water potential. Genetic variation in OA has been reported in a number of plant species and cultivars and diers with respect to the types of solutes accumulated (i.e. amino acids, sugars, polyols, quaternary amines, ions, and organic acid) (Bohnert and Jensen, 1996). OA of wheat was found to be inuenced by alternate alleles at a single locus on chromosome 7A, with high response being recessive (Morgan and Tan, 1996). Control of OA by the 7A locus was based primarily on potassium accumulation, and secondarily on amino acid accumulation. In rice, a QTL for OA under drought stress was identied on chromosome 8 across rice populations (Lilley et al., 1996; Robin et al., 2003). Comparative mapping indicates that the region of rice chromosome 8 containing the OA QTL is homeologous with a segment of wheat chromosome 7S which contains the OA locus identied by

MAS for drought stress tolerance Drought is by far the most signicant environmental stress in worldwide agriculture. Although plants are exposed to many types of environmental stresses, osmotic stress, whether by drought, salinity or low temperature, constitutes the most serious limitation to plant growth, productivity and distribution. Many studies on drought resistance have monitored the physiological and biochemical status of stressed plants compared with unstressed plants. Important mechanisms of drought resistance deduced from these studies mainly include the following: drought escape via a short life cycle, photoperiod sensitivity and developmental plasticity; drought avoidance via enhanced water uptake and reduced water loss; drought tolerance via osmotic adjustment (OA) and antioxidant capacity; and drought recovery via desiccation tolerance (Zhang et al., 1999). Direct selection for grain yield under water-stressed conditions is dicult due to low heritability and signicant G E interactions (Ceccarelli

330 Morgan and Tan (1996) and with a barley chromosome 1 region where a QTL for relative water content in stressed conditions was identied (Teulat et al., 2003). Similarly, a rice QTL for OA located on chromosome 3 reside in a genomic region syntenic with the homeologous region of maize chromosome 1; in maize this region is associated with various physiological and agronomic traits aecting drought tolerance (Zhang et al., 2001). These results suggest that during cereal evolution, genes in these genomic regions in rice, wheat, barley and maize have been conserved to respond to drought conditions and might therefore contains useful genes for the improvement of drought resistance in cereal crops. The rice QTL region controlling OA on chromosome 3 has recently been subjected to saturation mapping and new markers as well as stress-related EST have been added to this chromosomal region and, following the authors suggestions (Nguyen et al., 2004), these markers can now be used for MAS of the favorable QTL allele. Root penetration and morphology Carbon isotope discrimination Constitutive and adaptative root growth have been implicated in the improved performance of rice under rainfed lowland conditions. A causal relationship between root traits and yield performancies under drought stress was found in the study of Babu et al. (2003). In this work, QTLs controlling production traits under irrigated and drought stress conditions were mapped using a rice DH population. On chromosome 4, a major region controlling grain yield components under drought stress was identied. In this study, root traits (like root penetration index, root thickness at stem base, root pulling force and root morphology) identied in the same mapping population in other studies (references in Babu et al., 2003) showed positive correlations with yield and yield components under drought stress on chromosome 4 region. A marker-assisted BC program was recently used to improve root traits aecting drought stress tolerance in the elite rice cultivar IR64, (Shen et al., 2001). Foreground selection of the Azucena (an upland tropical japonica variety) allele at four QTLs for deeper roots was performed strictly on the basis of the genotypes at the marker loci up to the BC3F2. Selected NILs for the four target QTLs showed, depending of the target QTLs, improved Carbon isotope discrimination (CID) provides an integrated measurement of TE (Transpiration Eciency, the ratio of dry matter produced to water transpired) of C3 crop species (Ehdaie et al., 1991). During photosynthesis, plants favor incorporation of the light carbon isotope (12C) over the heavy isotope (13C), with the result that CID is positively correlated with the ratio of internal leaf CO2 concentration to ambient CO2 concentration (Ci/Ca) and negatively associated with TE (Ehdaie et al., 1991). Thus, a high Ci/Ca leads to a higher CID and a lower TE. The major advantage of using CID in selection is its high heritability, which is primarily due to small G E interactions in dryland areas (Merah et al., 2001). Teulat et al. (2002) have reported the rst QTL study for CID measured in mature grains from plants grown in Mediterranean eld conditions, identifying ten QTLs involved in CID variation. Eight QTLs for CID overlap with QTLs previously identied in the same population and aecting plant water status, OA and/or yield components. These regions are of interest in terms of plant breeding as they control both important droughtadaptative traits for cereals and yield components. root length (by 12 to 27% with respect to IR64) or improved deep root weight (two NILs had the highest phenotypic gain, outperforming IR64 by up to 75%). Eorts for the identications of gene functions responsible for the QTLs aecting root traits have recently involved the mapping of candidate genes for root traits under drought stress (Zheng et al., 2003; Nguyen et al., 2004). Genes, identied as dierentially expressed under drought conditions and/or putatively involved in root growth/modication, were found to map to chromosomal regions that aects root traits. The tight genetic linkage between these candidate genes and the QTLs for root traits does not denitely demonstrate a causal relationship; associations of these data with ne mapping experiments and with the development of NILs for a given QTLs would reinforce a possible causal relationship. Notwithstanding, the availability of additional sequences tightly linked to QTLs conferring drought resistance provide an additional set of markers useful to select the favorable QTL alleles.

331 Phenological traits When drought stress occurs just before or during the owering period, a delay in silking is observed, resulting in an increase in the length of the anthesis-silking interval (ASI) in maize. This asynchrony between male and female owering has been associated with a grain-yield decrease under drought (Ribaut et al., 1997). Under conditions of low yield, selection for secondary traits such as ASI, which are highly correlated with grain yield and have relatively high heritability, may increase selection eciency (Bolanos and Edmeades, 1996). Selection for reduced ASI in tropical open-pollinated varieties has been shown to correlate with improved yields under drought stress. Molecular markers allowed the identication of four genomic regions for the expression of both yield and ASI in maize (Ribaut et al., 1997). In three of these regions, the allelic contributions for short ASI corresponded to grain yield increase, while for one genomic region the allelic contribution for short ASI corresponded to a yield reduction. In the design of a breeding strategy, selection for QTLs involved in the expression of ASI should therefore be combined with selection for grain yield. Stay-green is an important form of drought resistance mechanism in sorghum, conferring resistance to premature senescence under soil moisture stress during the post-owering period. QTL studies with RILs and NILs have identied several genomic regions associated with resistance to pre-owering and post-owering drought stress (Haussman et al., 2002; Sanchez et al., 2002). In the work of Sanchez et al. (2002), three stay green QTLs, Stg1, Stg2 and Stg3, accounting for 20, 30 and 16% of phenotypic variance respectively are described. At least Stg2, considered as the most important QTL, was found to be consistent across several dierent mapping populations and environments; for this QTLs, NILs have been developed by marker-assisted BC breeding and physical contig of sequences have been obtained with sorghum BAC clones. Because the average marker interval for the genomic regions of the QTLs Stg1 and Stg2 is 1.3 and 1.7 cM respectively, several information are therefore available to allow the exploitation of this favorable trait. In addition, sorghum could serve as a bridge species for comparative mapping analysis between the grass relatives (Paterson et al., 2000); breeding for drought resistance is also an important objective for other crops like maize or wheat and the stay-green phenomenon has also been reported in these crops. After ne mapping in the region of the sorghum QTLs, DNA markers could be used to pinpoint the corresponding orthologous regions in maize or wheat. Seed germination An additional approach to minimizing agricultural losses incurred by drought stress is to develop, via genetic means, plant cultivars that can escape or withstand periods of drought. Selection and breeding for drought tolerance is also dicult because tolerance is a developmentally regulated, stage-specic phenomenon (Richards, 1996). Most commercial cultivars of tomato are sensitive to drought stress at all stage of plant development, with seed germination and early seedling growth being the most sensitive. A tomato population segregating for seed germination rate under drought stress has been obtained from a cross between a Lycopersicon esculentum breeding line and a germination stress-tolerant Lycopersicon pimpinellifolium (Foolad et al., 2002). A total of 119 RFLP markers were used in a distributional extreme marker analysis to measure statistical dierences in marker allele frequencies between rapidly germinating and slow germinating lines. This resulted in identication of four QTLs for rate of germination under drought stress. At the QTLs located on chromosomes 1 and 9 the favorable alleles were contributed by the L. pimpinellifolium parent and had large eects, while at the QTLs on chromosomes 8 and 12 the favorable alleles were contributed by the L. esculentum recurrent parent. The overall results indicate that drought tolerance during seed germination in tomato is genetically controlled and could potentially be improved by directional phenotypic selection or MAS.

MAS for salinity and aluminum tolerance Irrigation is a common cause of agricultural land degradation, because salt dissolved in the irrigation water is left in the soil following evapotraspiration. In regard to salinity, rice has been

332 particularly studied, due to its preference for irrigation, its sensitivity to salinity and its relatively small genome. Lee et al. (2003) found that reduction in several growth parameters were signicantly lower in indica (tolerant) varieties than in japonica varieties. Tolerant indica varieties were good Na+ excluders, absorbed high amounts of K+, and maintained a low Na+/K+ ratio in the shoot. Tolerant japonica varieties absorbed less Na+ but were not as good excluders as indica varieties. Taken as a whole, these results indicate that, for all parameters measured, the tolerance level of indica was higher than that of japonica. These results have recently been conrmed in a study where QTL mapping for physiological traits related to rice salt-tolerance has been performed (Lin et al., 2004). In this study QTL analysis has been carried out on a segregating population derived from a cross between a high salt-tolerant indica variety and a susceptible elite japonica variety. Two major QTLs with very large eects, one on chromosome 7 for shoot Na+ concentration (called as qSNC-7) and one on chromosome 1 for shoot K+ concentration (called as qSKC-1), explained 48.5 and 40.1% of the total phenotypic variance respectively; the additive aect of the alleles of the salt-tolerant variety at the qSNC-7 and qSKC-1 QTLs led respectively to a reduction of shoot Na+ concentration and to an increase of shoot K+ concentration. In three F3 lines belonging to the segregating population, the alleles of the salt-tolerant variety at a discrete number of QTLs (three to four, including qSNC-7 and qSKC-1) for physiological traits related to salttolerance were pyramided: these lines showed a level of seedling survival under salt stress equal or superior to the one observed for the resistant parent. These results would indicate that a breeding method of QTLs pyramiding using MAS could be applied for the development of varieties with high level of salt tolerance. Aluminum toxicity is a major limiting factor for agriculture in tropical and acidic soils. Using bread wheat (T. aestivum) recombinant inbred lines, a single locus for Al tolerance (referred to as AltBH) was found on the long arm of chromosome 4D (Riede and Anderson, 1996). A single gene controlling aluminum tolerance was also found in barley on chromosome 4H (Alp; Tang et al., 2000) and microsatellite markers associated to this locus have been identied (Raman et al., 2003); the microsatellite marker Bmag353 has been validated in a F3 population segregating for Al tolerance and the marker was found to predict the Al tolerance phenotype with over 95% accuracy. Previous reports showed that there is a conserved genomic region on the long arm of homoeologous chromosome 4 for Al tolerance among wheat (AltBH), rye (Alt3) and barley (Alp) (Miftahudin et al., 2002). On the basis of common markers it was suggested that the AltBH, Alt3 and Alp genes are orthologous loci because of the high level of synteny among chromosomes 4DL, 4RL and 4HL and they may share common function. One of the mechanisms for Al tolerance in the Triticeae is Al exclusion; this mechanism is mediated by Al-activated release of organic acids (malic acid), which chelate Al3+ in the rhizosphere and prevent its entry into the root apex. A rice major QTL for Al tolerance, contributed by O. rupogon, has recently been mapped on chromosome 3 and, on the basis of comparative map analysis, this QTL appear to be orthologous to the genomic region carrying the major Al tolerance gene on group 4 of the Triticeae (Nguyen et al., 2003). Even if physiological evidences of Al tolerance in rice are still lacking, the fact that collinear genomes pinpoints similar traits to the same chromosomal region raise the suspect that these loci are encoded by dierent alleles of a single gene. If this hypothesis will turn to be true, the Al tolerance locus on the long arm of homoeologous chromosome 4 of the Triticeae will provide a useful source in breeding for Al tolerance across several crop plant species.

MAS for quality traits Most quality traits show continuous variation and are inuenced by environmental conditions. Notwithstanding, some example of quality traits, for which MAS is a reliable approach for selection, are now available for several important crops including tomato, barley, wheat, cotton, and rice (Table 2). Tomato Soluble-solids content is of paramount importance for processing tomatoes, because lines with higher sugar content require less energy input (i.e. less cooking) during the concentration process. To

333
Table 2. Selected examples of important quality traits in plants Species Tomato Tomato Tomato Tomato Barley Barley Wheat (Triticum aestivum) Cotton Rice Trait Total soluble solids (sugar and acids) Total soluble solids (sugar and acids) Fruit elongation and neck constriction Fruit size Malting quality Malting quality Dough strength (HMW glutenin) Fiber strength Eating and cooking quality Locus/gene Brix9-2-5 Brix TA1150 QTL ovate fw2.2 QTL1 (chr. 1H) and QTL2 (chr. 4H) QTLs on chr. 3H, 6H and 7H Glu-1 alleles Molecular markers RFLP, SCAR, CAPS RFLP and SCAR RFLP, SCAR, CAPS RFLP and CAPS RFLP RFLP SCAR References Fridman et al. (2000) Frary et al. (2003) Liu et al. (2002) Frary et al. (2000) Han et al. (1997) Igartua et al. (2000) Radovanovic and Cloutier (2003), Ma et al. (2003) Zhang et al. (2003) Zhou et al. (2003)

QTLFS1 Waxy

SSR and RAPD SSR

uncover the molecular basis of sugar content variation, a QTL for total soluble solids (sugar and acids), named Brix9-2-5, derived from the greenfruited tomato species Lycopersicon pennellii, was characterized (Fridman et al., 2000). The genetic basis of the QTL was dissected by positional cloning and was shown to be the Lin5 gene, coding for a fruit-specic apoplastic invertase hypothesised to modulate fruit sink strength. The L. pennellii Brix allele increased glucose (+28%) and fructose (+18%) contents in various genetic backgrounds of cultivated tomato and across different environments, and confers around 3-fold increase in soluble solid content (up to 15% of the fruits fresh weight). Brix9-2-5 was shown to be partially dominant and independent of fruit weight and yield. Another QTL increasing soluble solids content has recently been introgressed in cultivated tomato from L. chmielewskii chromosome 1, generating a NIL with a 56-cM introgression (Frary et al., 2003). Cross of this NIL with a cultivated tomato line, generate an F2 segregating population. From this population subNILs with a reduced size of the L. chmielewskii introgressed fragment were generated through the identication of recombinant plants identied by RFLP analysis of markers anking the introgression. Additional marker analysis to determine the exact recombination break points allowed the identication of a

subNIL with only 19-cM introgression containing the soluble solids QTL. Another important quality trait in tomato is the fruit shape. The recessive pear-shaped tomato fruit trait ovate identied by early 20th-century geneticists has been mapped to a locus on chromosome 2. Recent genetic analyses have identied ovate as a major quantitative trait controlling fruit elongation and neck constriction in both tomato and eggplant. The major QTL ovate has recently been cloned from tomato and encodes for a protein with regulatory functions (Liu et al., 2002). Sequencing of the OVATE gene both in the wild type (round-fruited) and in an ovate genotype has led to the identication of an early stop codon in the ovate genotype causing the transition from round to pear shaped. Fruit size is another important quantitative trait. In the last 20 years a great number of QTLs for fruit size have been detected in several Lycopersicon species, 28 of which have been conrmed in independent studies (Grandillo et al., 1999). In one tomato population, the QTL fw2.2 was found to be responsible for 30% of the variation in fruit size. The gene underlying the major QTL fw2.2 has been cloned (Frary et al., 2000). The gene product has sequence similarity to the human oncogene c-H-rasp21m, and is early expressed during oral development.

334 A MAS BC scheme (Bouchez et al., 2002; Figure 2) has recently been applied for the introgression of ve QTLs controlling fruit quality traits derived from L. esculentum var. cerasiforme into three dierent genetic backgrounds of cultivated tomato (Lecomte et al., 2004). In this study it was observed that as few as three BCs were sucient to recover most of the recipient genome for chromosomes in which no QTLs eect had been detected. Because quality traits QTLs were not precisely mapped, to reduce the risk of loosing the QTLs, almost no selection for genetic background was performed on chromosomes carrying the quality traits QTLs; large regions of the donor chromosomes were therefore transferred leading to unfavorable linkage drag. It is therefore advisable that QTLs map position should be precisely dened in order to allow a selection pressure also in chromosome regions where target QTLs are located. Nevertheless, in this study, marker-assisted BC was ecient in improving quality traits in the genetic backgrounds used. Malting barley Methods for assessing the suitability of barley for malting involve analyses of grain characters, micromalting and laboratory analyses of malt quality traits. Barley grain and malt quality characters generally exhibit quantitative variation and are inuenced by genetic and environmental factors and by G E interactions (extensively reviewed in Fox et al., 2003). MAS would therefore be desirable in breeding programs to help overcome environmental eects and to minimize costly and time-consuming laboratory analysis. In a six-row Steptoe Morex barley population, major QTLs controlling malt extract percentage, alpha-amylase activity, diastatic power, and malt beta-glucan content, have been identied on chromosomes 1 (QTL1) and 4 (QTL2). The large and consistent eects of these two QTL regions across dierent environments make them good candidates for MAS. RFLP markers Brz and Amy2, WG622 and BCD402B, respectively anking QTL1 and QTL2, were used in a study comparing dierent selection strategies (Han et al., 1997). Genotypic selection (G) alone was used, in addition to phenotypic selection (P), tandem genotypic and phenotypic selection (rst G and then P, GP), and combined phenotypic and The most important quality parameters in wheat relate to physical (rheological) properties of the dough during bread making, such as extensibility and resistance to extension. These properties depend on the endosperm gluten proteins, which comprise two major fractions: gliadins and glutenins (Ma et al., 2003). Generally, high molecular weight (HMW) glutenins have been found to be more important than gliadins and low molecular weight (LMW) glutenins for dough rheological properties. Breadmaking qualities, especially dough strength, are dependent on the composition of HMW glutenin subunits, particularly the alleles Glu-A1b and Glu-D1d. SDS-PAGE of seed proteins is used for screening wheat lines for glutenin polypeptide proles. This method is relatively ecient because allelic variation at multiple loci can be assessed in a single gel lane. PCR-based molecular markers based on sequence variations of the coding and promoter regions of the wheat HMW glutenin genes at the Glu-1 locus have been developed (Radovanovic and Cloutier, 2003). When tested in a DH population segregating for bread making quality, DNA and SDS-PAGE protein markers showed discrepancies of only 2 to 8.5% depending on the marker assayed. genotypic selection (G and P together, G+P). MAS for QTL1 (GP and G+P) was more eective than phenotypic selection, while MAS for QTL2 was not as eective as phenotypic selection due to a lack of QTL2 eects in the background used. In the two-row Harrington (North America malting barley industry standard cultivar) TR306 barley population, QTLs for grain and malt quality traits were identied on chromosomes 3 (3H), 6 (6H) and 7 (5H). Barley lines, derived from same cross where the QTLs had been mapped, were selected on the basis of their marker-genotype at two identied QTLs on chromosome 7 (5H) (Igartua et al., 2000). Selected lines were phenotypically superior and the magnitude of the eects for these regions were closed to the estimates calculated in the mapping population. MAS was therefore eective within this population, but there are not indications of whether these QTLs would be useful beyond this particular cross. Wheat

335 PCR-based molecular markers have also been developed for the Glu-A1 locus in Australian commercial wheat varieties (Ma et al., 2003). These cultivars show only one or two predominant alleles at each HMW glutenin (Glu-1) homoeologous locus. Products of a single multiplexed PCR reaction permitted the discrimination of the major HMW glutenins in one simple assay. These markers are currently used in MAS for HMW glutenins in DH-based wheat breeding programs. Cotton Cotton is a high-value per acre crop that provides a raw material for the textile industry. Heritability of cotton yield components and ber properties is moderate to high (around 4080%), based on current estimates. However, genetic control of ber quality is aected by G E interactions, specically by dierences in water management regimes (Paterson et al., 2003). Molecular markers linked to ber-strength QTLs have been recently identied using a segregating population derived from a cross between a Gossypium anomalum introgression line, with good ber quality properties, and a standard cotton variety (Zhang et al., 2003). Three SSRs and six RAPDs markers were identied to be linked at two QTLs for ber strength. A major QTL (QTLFS1) associated to eight markers explained more than 30% of the phenotypic variation. MAS using two RAPD markers and a SSR marker was used to assist the transfer of QTLFS1 in four dierent genetic backgrounds. The major QTL was genetically stable in the dierent backgrounds and lead to a substantial increase in ber strength of the improved lines. It was concluded that MAS is ecient to increase ber strength, especially using the SSR marker associated to the major QTL, since this marker could identify the homozygous genotype. Rice In China there is a strong emphasis on improving the quality of indica hybrid rice varieties. Zhenshan 97, the female parent of a number of widely cultivated hybrids, is of poor quality because of its high amylose content (AC), hard gel consistency (GC), low gelatinization temperature (GT) and chalky endosperm. These three traits for cooking and eating quality are controlled by the genomic region containing the Waxy locus. The eating and cooking quality of Zhenshan 97A (male-sterile) has been improved by introgression of the Waxy gene region from Minghui 63 (the restorer line) (Zhou et al., 2003). MAS was used during three generations of backcrossing. An SSR marker waxy, representing the Waxy gene, was used to select for the presence of the Minghui Waxy region; two RFLP markers dening a 6.1 cM interval and anking the Waxy locus were used to select recombinants between the anking markers and the Waxy gene (to ensure that the introgressed region was shorter than the interval dened by the two RFLP markers). A total of 118 AFLP fragments were used in background selection to recover the genetic background of Zhenshan at unlinked loci. The obtained selected lines and their hybrids with Minghui 63, or Shanyou 63, showed a reduced AC and an increased GC and GT, coupled with reduced grain opacity. Results from this study also conrmed that the Waxy region has major eects on the three traits for cooking and eating quality.

Conclusions and perspectives The recent progress in the area of plant molecular biology and plant genomics have the potential to initiate a new Green Revolution. However, these discoveries need to be implemented in new cultivars to realize that potential. Some potential limitations of MAS strategies have been highlighted for both simply and complex inherited traits. A cautious optimism has been expressed about MAS of complex traits. Although molecular markers have been successfully associated with QTLs, these associations have actually demonstrated limited usefulness in plant breeding programs. Complex traits are the most dicult to handle during a breeding program, but are responsible for most breeding progress in critical traits such as yield, yield stability and adaptation. Before selection of complex traits can be fully realized, eorts are required to obtain more reliable and comprehensive data about quantitative trait by using correctly sized mapping populations (i.e. small mapping populations are not adequate for QTL mapping) (Young, 1999). Economics is also a key determinant for the application of molecular genetics in genetic improvement

336 programs. The use of MAS can be justied when it replaces more expensive or tedious assays, or results in increased precision in the identication of the desired genotypes. What becomes critical therefore is the balance between added value and additional cost. Dekkers and Hospital (2002) have recently reviewed some of the potential limitations of MAS strategies, and concluded that the use of MAS will be determined by the economic benet relative to conventional selection. In the immediate future, the key for the eciency of MAS in large breeding population will depends from the implementation and integration of dierent points. The rst crucial point is therefore the availability of cost-ecient and high-throughput genotyping methods. A variety of high-throughput genotyping technologies (Rafalski, 2002) are just becoming suciently inexpensive to allow their use in plant breeding. A new generation of molecular markers based on the detection of SNPs promises high-throughput assays at relatively low costs, along with the potential for high levels of multiplexing. Implementation of this multiplexing technology in plant improvement strategies can provide cost-eective tools for selection of multiple traits in breeding populations. Second, exploiting information derived from comparative genetic maps, genomic regions conferring positive traits across syntenic species might be directly applied across species for the improvement of the trait. Comparative genetic maps show that chromosomal segment structure (orthology or conserved synteny) and marker order (colinearity) are conserved across plant species over substantial evolutionary distances (Paterson et al., 2000). Most of the economically important species of the grass have detailed comparative maps such that both gene content and gene order often can be predicted across species (Gale and Devos, 1998); similarly, genetic and genomic information can be shared among many leguminous species (e.g. soybean and mung bean) (Boutin et al., 1995) or among solanaceous species (e.g. tomato, pepper, potato) (Livingstone et al., 1999). When genetic mapping in collinear genomes pinpoints similar traits to the same chromosomal regions, there is a good reason to suspect that these loci are encoded by ortholog genes. Even in the absence of an orthologous candidate gene, the information from the model plant of the family (e.g. rice for the grass family and tomato for the nightshade family) can be applied to increase the marker saturation of a dened chromosome region, as it may be required for the identication of markers suitable for MAS or for the ne mapping of a gene for positional cloning. Third, for improving polygenic traits in a quick time-frame and in a cost eective manner, recent advances in MAS strategies have suggested improved selection schemes in which MAS is applied only once during the breeding process to a range of agronomically important traits (Single ` large-scale-MAS Ribaut and Betran, 1999) or involve the mapping of loci for all agronomic important traits (Breeding by design Peleman and van der Voort, 2003), and selection schemes in which MAS operates by cyclically re-estimating the value of the QTL alleles each time a new set of germplasm is generated during the breeding process (Mapping as you go Podlich et al., 2004). Empirical applications of these improved selection schemes will provide indications on their eciency in improving MAS for complex traits. A fourth important point is represented by the increasing amount of information on the topic that are publicly available. Recent large-scale sequencing projects have produced a large amount of single-pass sequences of cDNAs from dierent plant species. The number of ESTs deposited in gene bank for wheat, maize, barley and soybean has mounted to 562,000, 416,000, 380,000 and 342,000 sequences respectively (http://www.ncbi. nlm.nih.gov/dbEST/). Because SSRs and SNPs based markers can be obtained quite easily from ESTs (Morgante et al., 2002; Rafalski, 2002), the development of molecular markers has recently shifted from anonymous DNA fragments to genes. Transcription-based genetic maps have thus been obtained from dierent crop species including wheat (Gao et al., 2004), barley (Graner et al., 2004) and rice (Wu et al., 2002). The availability of marker-dense transcriptional maps has two important implications for the improvement of complex traits in plant breeding. First, it can contribute candidate genes during the mapping of QTLs. Second, when relatively stringent sequence similarity threshold are used, the EST loci form connecting points between related genomes (e.g. wheat, barley and rice) allowing the exploitation of the information derived from syntenic genomic regions across species, i.e. in the case of the grass

337 family, provides the opportunity for a transfer of genetic information from rice to barley and other Triticeae. Also for the tools directly useful for MAS, publicly available information are increasing. As an example, in the MAS wheat project, funded by the USDA, all the information and protocols used in the MAS-wheat project in wheat MAS for traits related to pathogens resistance and quality are publicly available through the project WEB site (http://maswheat.ucdavis.edu). As a fth point it has to be mentioned that the availability of comprehensive cDNA and oligoncleotide arrays is now providing an option for the development of functional genomics-based strategies for the investigation of quantitatively inherited traits, using at least two strategies. The rst is a functional association strategy: cDNA array can reveal that gene expression within a given tissue varies between genotypes diering for a given trait. Genetic mapping of identied candidate genes can then reveal congruency between the map position of the candidate gene and the presence of a QTL (Graner et al., 2004). The second strategy allow the identication of QTLs by a methodology called as eXtreme array mapping (XAM). This method can estimate the dierences in allele frequency between pools of lines, selected for extreme phenotypes, by hybridization of total genomic DNA to a GeneChips (Wolyn et al., 2004). Thus, because dierent approaches are improving the strategies on which MAS rely on, an increased complementarity between molecular technologies and conventional breeding is expected in the near future for a more ecient improvement of the crop plants.
among rice varieties diering in low temperature tolerance. FEBS Lett. 527(13): 181185 Anderson JA, Stack RW, Liu S, Waldron BL, Fjeld AD, Coyne C, Moreno-Sevilla B, Mitchell Fetch J, Song QJ, Cregan PB & Frohberg RC (2001) DNA markers for Fusarium head blight resistance QTLs in two wheat populations. Theor. Appl. Genet. 102: 11641168 Arru L, Faccini N, Govoni C, Cattivelli L, Pecchioni N, Delogu ` G, Stanca AM, & Vale G (2003) The PCR-based marker MWG2018 linked to the Rdg2a leaf stripe resistance gene is a useful tool for assesing barley resistance in breeding programs. Crop Sci. 43: 10361042 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. Crop Sci. 43: 14571469 Baker B, Zambryski P, Staskawicz B & Dinesh-Kumar SP (1997) Signaling in plant-microbe interactions. Science 276: 726733 Barr AR, Jeeries SP, Warner P, Moody DB, Chalmers KJ & Langridge P (2000) Marker assisted selection in theory and pratice. Proceedings VIII Barley Genetic Symposium, Vol. I (pp. 167178). Barzen E, Stahl R, Fuchs E, Borchardt DC & Salamini F (1997) Development of coupling-repulsion-phase SCAR markers diagnostic for the sugar beet Rr1 allele conferring resistance to rhizomania Mol. Breed. 3: 231238 Boer MP, Ter Braak CJF & Jansen (2002) A penalized likelihood method for mapping epistatic quantitative trait loci with one-dimensional genome searches. Genetics 162: 951960 Bohnert HJ & Jensen RG (1996) Strategies for engineering water-stress tolerance in plants. Trends Biotechnol. 14: 8997 Bolanos J & Edmeades GO (1996) The importance of the anthesis-silking interval in breeding for drought tolerance in tropical maize. Field Crop Res. 48: 6580 Bouchez A, Hospital F, Causse M, Gallais A & Charcosset A (2002) Marker-assisted introgression of favorable alleles at Quantitative Trait Loci between maize elite lines. Genetics 162:19451959 Boutin SR, Young ND, Olson T, Yu Z-H, Shoemaker RC & Vallejos C (1995) Genome conservation among three legume genera detected with DNA markers. Genome 38: 928937 Buckler IV ES & Thornsberry JM (2002) Plant molecular diversity and application to genomics. Curr. Opin. Plant Biol. 5: 107111 Buschges R, Hollricher K, Panstruga R, Simons G, Wolter M, Frijters A, van Daelen R, van der Lee T, Diergaarde P, Groenendijk J, Topsch S, Vos P, Salamini F & Schulze-Lefert P (1997) The barley Mlo gene: a novel control element of plant pathogen resistance. Cell 88: 695705 Cattivelli L, Baldi P, Crosatti C, Di Fonzo N, Faccioli P, Grossi M, Mastrangelo AM, Pecchioni N & Stanca AM (2002) Chromosome regions and stress-related sequences involved in resistance to abiotic stress in Triticeae. Plant. Mol. Biol. 48(56): 649665 Ceccarelli S, Acevedo E & Grando S (1991) Breeding for yield stability in unpredictable environments: single traits, interaction between traits, and architecture of genotypes. Euphytica 56: 169185

Acknowledgements This work was supported by the Italian MiPAF, Project Protezione delle piante mediante luso di marcatori molecolari (PROMAR), the CEREALAB project and the E.C. COST Action 860 SUSVAR. The authors thanks Dr Nicholas Collins for critically reading and improving the review.

References
Abe F, Saito K, Miura K & Toriyama K (2002) A single nucleotide polymorphism in the alternative oxidase gene

338
` Chague V, Fahima T, Dahan A, Sun GL, Korol AB, Ronin YI, Grama A, Roder ME & Nevo E (1999) Isolation of microsatellite and RAPD markers anking the Yr15 gene of wheat using NILs and bulked segregant analysis. Genome 42: 10501056 Cheng FS, Weeden NF, Brown SK, Aldwinckle HS, Gardiner SE & Bus VG (1998) Development of a DNA marker for Vm, a gene conferring resistance to apple scab. Genome 41: 208214 Concibido VC, Vallee BL, Mclaird P, Pineda N, Meyer J, Hummel L, Yang J, Wu K & Delannay X (2003) Introgression of a quantitative trait locus for yield from Glycine soja into commercial soybean cultivars. Theor. Appl. Genet. 106: 575582 Cregan PB, Mudge J, Fickus EW, Danesh D, Denny R & Young ND (2000) Two simple sequence repeat markers to select for soybean cyst nematode resistance conditioned by the rhg1 locus. Theor. Appl. Genet. 99: 811818 Crossa J, Vargas M, van Eeuwijk FA, Jang C, Edmeades GO & Hoisington D (1999) Interpreting genotype environment interaction in tropical maize using linked molecular markers and environmental covariables. Theor. Appl. Genet. 99: 611625 Dekkers JCM & Hospital F (2002) The use of molecular genetics in the improvement of agricultural populations. Nat. Rev. Genet. 3: 2232 Doganlar S, Dodson J, Gabor B, Beck-Bunn T, Crossman C & Tanksley SD (1998) Molecular mapping of the py-1 gene for resistance to corky root rot (Pyrenochaeta lycopersici) in tomato. Theor. Appl. Genet. 97: 784788 Donald TM, Pellerone F, Adam-Blondon A-F, Bouquet A, Thomas MR & Dry IB (2002) Identication of resistance gene analogs linked to a powdery mildew resistance locus in grapevine. Theor. Appl. Genet. 104: 610618 Dussle CM, Quint M, Xu ML, Melchinger AE & Lubberstedt T (2002) Conversion of AFLP fragments tightly linked to SCMV resistance genes Scmv1 and Scmv2 into simple PCRbased markers. Theor. Appl. Genet. 105: 11901195 Eathington SR, Dudley JW & Rufener GK (1997) Usefulness of marker-QTL associations in early generation selection. Crop Sci. 37: 16861693 Ehdaie B, Hall AE, Farquhar GD, Nguyen HT & Waines JG (1991) Water-use eciency and carbon isotope discrimination in wheat. Crop Sci. 31: 12821288 Feuillet C, Travella S, Stein N, Albar L, Nublat A & Keller B (2003) Map-based isolation of the leaf rust disease resistance gene Lr10 from the hexaploid wheat (Triticum aestivum L.) genome. Proc. Natl. Acad. Sci. USA 100: 1525315258 Fjellstrom R, Conaway-Bormans C, McClung AM, Marchetti MA, Shank AR & Park WD (2004) Development of DNA markers suitable for marker assisted selection of three Pi genes conferring resistance to multiple Pyricularia grisea pathotypes. Crop Sci. 44: 17901798 Foolad MR, Zhang LP & Subbiah P (2002) Genetics of drought tolerance during seed germination in tomato: inheritance and QTL mapping. Genome 46: 536545 Fox GP, Panozzo JF, Li CD, Lance RCM Inkerman PA & Henry RJ (2003) Molecular basis of barley quality. Aust. J. Agricult. Res. 54: 10811101 Fox PN, Crossa J & Romagosa I (1997) Genotype environment and adaptation. In: Kempton R & Fox P (eds) Statistical Methods for Plant Variety Evaluation (pp. 117 138) Chapman & Hall, London Francia E, Rizza F, Cattivelli L, Stanca AM, Galiba G, Toth B, Hayes PM, Skinner JS & Pecchioni N (2004) Two loci on chromosome 5H determine low-temperature tolerance in a Nure (winter) Tremois (spring) barley map. Theor. Appl. Genet. 108: 670680 Frary A, Nesbitt TC, Frary A, Grandillo S, van der Knaap E, Cong B, Liu JP, Meller J, Elber R, Alpert KB & Tanksley SD (2000) fw2.2: A quantitative trait locus key to the evolution of tomato fruit size. Science 289: 8588 Frary A, Doganlar S, Frampton A, Fulton T, Uhlig J, Yates H & Tanksley S (2003) Fine mapping of quantitative trait loci for improved fruit characteristics from Lycopersicon chmielewskii chromosome 1. Genome 46: 235243 Fridman E, Pleban T & Zamir D (2000) A recombination hotspot delimits a wild-species quantitative trait locus for tomato sugar content to 484 bp within an invertase gene. Proc. Natl. Acad. Sci. USA 97: 47184723 Frisch M, Bohn M & Melchinger AE (1999) Comparison of selection strategies for marker-assisted selection backcrossing of a gene. Crop Sci. 39: 12851301 Gale MD & Devos KM (1998) Comparative genetics in the grasses. Proc. Natl. Acad. Sci. USA 95: 19711794 Gao LF, Jing RL, Huo NX, Li Y, Li XP, Zhou RH, Chang XP, Tang JF, Ma ZY & Jia JZ (2004) One hundred and one new microsatellite loci derived from ESTs (EST-SSRs) in bred wheat. Theor. Appl. Genet. 108: 13921400 Gebhardt C & Valkonen JPT (2001) Organization of genes controlling disease resistance in the potato genome. Annu. Rev. Phytopathol. 39: 79102 Gleeson AC (1997) Spatial analysis. In: Kempton RA & Fox PN (eds) Statistical Mathods for Plant Variety Evaluation (pp. 6885) Chapman & Hall, London Go SA, Ricke D, Lan T-H, Presting G, Wang R, Dunn M, Glazebrook J, Sessions A, Oeller P, Varma H, Hadley D, Hutchison D, Martin C, Katagiri F, Lange BM, Moughamer T, Xia Y, Budworth P, Zhong J, Miguel T, Paszkowski U, Zhang S, Colbert M, Sun W-L, Chen L, Cooper B, Park S, Wood TC, Mao L, Quail P, Wing R, Dean R, Yu Y, Zharkikh A, Shen R, Sahasrabudhe S, Thomas A, Cannings R, Gutin A, Pruss D, Reid J, Tavtigian S, Mitchell J, Eldredge G, Scholl T, Miller RM, Bhatnagar S, Adey N, Rubano T, Tusneem N, Robinson R, Feldhaus J, Macalma T, Oliphant A & Briggs S (2002) A draft sequence of the rice genome (Oryza sativa L. ssp. japonica). Science 296: 92100 Grandillo S, Ku HM & Tanksley SD (1999) Identifying the loci responsible for natural variation in fruit size and shape in tomato. Theor. Appl. Genet. 99: 978987 Graner A, Streng S, Kellermann A, Schiemann A, Baner E, Waugh R, Pellio B & Ordon F (1999) Molecular mapping of the rym5 locus encoding resistance to dierent strains of the barley yellow mosaic virus complex. Theor. Appl. Genet. 98: 285290 Graner A, Streng S, Drescher A, Jin Y, Borovkova I & Steenson B (2000) Molecular mapping of the leaf rust resistance gene Rph7 in barley. Plant Breed. 119: 389392 Graner A, Kota R, Perovic D, Potokina E, Prasad M, Scholz U, Stein N, Thiel T, Varshney RK & Zhang H (2004) Molecular mapping: shifting from the structural to the

339
functional level. Proc. 9th International Barley Genetic Symposium (pp. 4957). Hammond-Kosack KE & Parker JE (2003) Deciphering plant pathogen communication: fresh perspectives for molecular resistance breeding. Curr. Opin. Biotec. 14: 177193 Han F, Romagosa I, Ullrich SE, Jones BL, Hayes PM, Wesenberg DM (1997) Molecular marker-assisted selection for malting quality traits in barley. Mol. Breed. 3: 427437 Haussmann BIG, Mahalakshmi V, Reddy BVS, Seetharama N, Hash CT & Geiger HH (2002) QTL mapping of stay-green in two sorghum recombinant inbred populations. Theor. Appl. Genet. 106: 133142 Hayashi K, Hashimoto N, Daigen M & Ashikawa I (2004) Development of PCR-based SNP markers for rice blast resistance genes at the Piz locus. Theor. Appl. Genet. 108: 12121220 Helguera M, Khan IA & Dubcavsky (2000) Development of PCR markers for wheat leaf rust resistance gene Lr47. Theor. Appl. Genet. 101: 625631 Hittalmani S, Parco A, Mew TV, Zeigler RS & Huang N (2000) Fine mapping and DNA marker-assisted pyramiding of the three major genes for blast resistance in rice. Theor. Appl. Genet 100: 11211128 Huang CC, Cui YY, Weng CR, Zabel P & Lindhout P (2000) Development of diagnostic PCR markers closely linked to the tomato powdery mildew resistance gene Ol-1 on chromosome 6 of tomato. Theor. Appl. Genet 101: 918 924 Huang L & Gill BS (2001) An RGA-like marker detects all known Lr21 leaf rust resistance gene family members in Aegilops tauschii and wheat. Theor. Appl. Genet. 103: 1007 1013 Huang XQ, Wang LX, Xu MX & Roder MS (2003) Microsat ellite mapping of the powdery mildew resistance gene Pm5e in common wheat (Triticum aestivum L.). Theor. Appl. Genet. 106: 858865 Igartua E, Edney M, Rossnagel BG, Spaner D, Legge WG, Scoles GJ, Eckstein PE, Penner GA, Tinker NA, Briggs KG, Falk DE & Mather DE (2000) Marker-based selection of QTL aecting grain and malt quality in two-row barley. Crop Sci. 40: 14261433 Ishimaru K (2003) Identication of a locus increasing rice yield and physiological analysis of its function. Plant Physiol. 133: 10831090 Ismail AM, Hall AE & Close TJ (1999) Allelic variation of a dehydrin gene co-segregates with chilling tolerance during seedling emergence. Proc. Natl. Acad. Sci. USA 96: 13566 13570 Jaglo-Ottosen KR, Gilmour SJ, Zarka DG, Schabenberger O & Thomashow MF (1998) Arabidopsis CBF1 overexpression induces COR genes and enhances freezing tolerance. Science 280: 104106 Jannoo N, Grivet L, Dookum A, DHont A & Glaszmann JC (1999) Linkage disequilibrium among modern sugarcane cultivars. Theor. Appl. Genet. 99: 10531060 Jeon J-S, Chen D, Yi G-H, Wang GL & Ronald PC (2003) Genetic and physical mapping of Pi5(t), a locus associated with broad-spectrum resistance to rice blast. Mol. Gen. Genom. 269: 280289 Jiang C & Zeng Z-B (1995) Multiple trait analysis of genetic mapping for quantitative trait loci. Genetics 140: 1111 1127 King GJ, Tartarini S, Brown L, Gennari F & Sansavini S (1999) Introgression of the Vf source of scab resistance and distribution of linked marker alleles within the Malus gene pool. Theor. Appl. Genet. 99: 10391046 Kloppers FJ & Pretorius ZA (1997) Eects of combinations amongst genes Lr13, Lr34 and Lr37 on components of resistance in wheat to leaf rust. Plant Pathol. 46: 737750 Knapp SJ (1998) Marker-assisted selection as a strategy for increasing the probability of selecting superior genotypes. Crop Sci 38: 11641174 Kolmer JA (1996) Genetics of resistance to wheat leaf rust. Annu. Rev. Phytopathol. 34: 435455 Kraft T, Hansen M & Nilsson NO (2000) Linkage disequilibrium and ngerprinting in sugar beet. Theor. Appl. Genet. 101: 323326 Lande R & Thompson R (1990) Eciency of marker-assisted selection in the improvement of quantitative traits. Genetics 124:743756 ` Lecomte L, Due P, Buret M, Servin B, Hospital F & Causse M (2004) Marker-assisted introgression of ve QTLs controlling fruit quality traits into three tomato lines revealed interactions between QTLs and genetic backgrounds. Theor. Appl. Genet. 109: 658668 Lee KS, Choi WY, Ko JC, Kim TS & Gregorio GB (2003) Salinity tolerance of japonica and indica rice (Oryza sativa L.) at the seedling stage. Planta 216: 10431046 Leister D, Kurth J, Laurie DA, Yano M, Sasaki T, Graner A & Schulze-Lefert P (1999) RFLP- and physical mapping of resistance gene homologues in rice (O. sativa) and barley (H. vulgare). Theor. Appl. Genet. 98: 509520 Lilley JM, Ludlow MM, McCouch SR & OToole JC (1996) Locating QTL for osmotic adjustment and deydration tolerance in rice. J. Exp. Bot. 47: 14271436 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: 253260 Liu JP, Van Eck J, Cong B & Tanksley SD (2002) A new class of regulatory genes underlying the cause of pear-shaped tomato fruit. Proc. Natl. Acad. Sci. USA 99: 1330213306 Liu S, Griey CA & Shagai Maroof MA (2001) Identication of molecular markers associated with adult plant resistance to powdey mildew in common wheat cultivar Massey. Crop Sci. 41: 12681275 Livingstone KD, Lackney VK, Blauth JR, van Wijk R & Jahn MK (1999) Genome mapping in Capsicum and the evolution of genome structure in the Solanaceae. Genetics 152: 1183 1202 Lu Z-X, Sossey-Alaoui K, Reighard GL, Baird WMV & Abbot AG (1999) Development and characterization of a codominat marker linked to root-knot nematode resistance, and its application to peach rootstock breeding. Theor. Appl. Genet. 99: 115122 Ma W, Zhang W & Gale KR (2003) Multiplex-PCR typing of high molecular weight glutenin alleles in wheat. Euphytica 134: 5160

340
Ma ZQ, Wei JB & Cheng SH (2004) PCR-based markers for the powdery mildew resistance gene Pm4a in wheat. Theor. Appl. Genet. 109: 140145 Madsen LH, Collins NC, Rakwalska M, Backes G, Sandal N, Krusell L, Jensen J, Waterman EH, Jahoor A, Aylie M, Pryor AJ, Langridge P, Schulze-Lefert P & Stougaard J (2003) Barley disease resistance gene analogs of the NBSLRR class: identication and mapping. Mol. Gen. Genomics 269: 150161 Mago R, Spielmeyer W, Lawrence GJ, Lagudah ES, Ellis JG &Pryor A (2002) Identication and mapping of molecular markers linked to rust resistance genes located on chromosome 1RS of rye using wheat-rye translocation lines. Theor. Appl. Genet. 104: 13171324 Merah O, Deleens E, Teulat B & Monneveux P (2001) Productivity and carbon isotope discrimination of dierent durum wheat organs under Mediterranean conditions. CR Acad. Sci. Paris, Science de la vie, Serie III 324: 5157 Michelmore R (2000) Genomic approaches to plant disease resistance. Curr. Opin. Plant Biol. 3: 125131 Michelmore RW (2003) The impact zone: genomics and breeding for durable disease resistance. Curr. Opin. Plant Biol. 6: 397404 Michelmore RW, Paran I & Kesseli RV (1991) Identication of markers linked to disease-resistance gene by bulked segregant analysis: a rapid method to detect markers in specic genomic regions by using segregating populations. Proc. Natl. Acad. Sci. USA 88: 98289832 Miftahudin G, Scoles J & Gustafson JP (2002) AFLP markers tightly linked to the aluminium tolerance gene Alt3 in rye (Secale cereale L.). Theor. Appl. Genet. 104: 626631 Mohan M, Nair S, Bhagwat A, Krishna TG, Yano M, Bhatia CR & Sasaki T (1997) Genome mapping, molecular markers and marker-assisted selection in crop plants. Mol. Breed. 3: 87103 Morgan JM & Tan MK (1996) Chromosomal location of a wheat osmoregulation gene using RFLP analysis. Aust. J. Plant Physiol. 23: 803806 Morgante M & Salamini F (2003) From plant genomics to breeding practice. Curr. Opin. Biotechnol. 14: 214219 Morgante M, Hanafey M & Powell W (2002) Microsatellites are preferentially associated with nonrepetitive DNA in plant genomes. Nat. Genet. 30: 194200 Murai H, Hashimoto Z, Sharma PN, Shimizu T, Murata K, Takumi S, Mori N, Kawasaki S & Nakamura C (2001) Construction of a high-resolution linkage map of a rice brown planthopper (Nilaparvata lugens Stal) resistance gene bph2. Theor. Appl. Genet. 103: 526532 Naess SK, Bradeen JM, Wielgus SM, Haberlach GT, McGrath JM & Helgeson JP (2000) Resistance to late blight in Solanum bulbocastaneum is mapped to chromosome 8. Theor. Appl. Genet. 101: 697704 Nguyen BD, Brar DS, Bui BC, Nguyen TV, Pham LN & Nguyen HT (2003) Identication and mapping of the QTL for aluminium tolerance introgressed from the new source, Oryza rupogon Gri., into indica rice (Oryza sativa L.). Theor. Appl. Genet. 106: 583593 Nguyen TTT, Klueva N, Chamareck V, Aarti A, Magpantay G, Millena ACM, Pathan MS & Nguyen HT (2004) Saturation mapping of QTLs regions and identication of putative candidate genes for drought resistance in rice. Mol. Gen. Genomics 272: 3546 ` Nordborg M & Tavare S (2002) Linkage disequilibrium: what history has to tell us. Trends Genet. 18: 8390 Olsen KM, Halldorsdottir SS, Stinchombe JR, Weinig C, Schmitt J & Purugganan MD (2004) Linkage disequilibrium mapping of Arabidopsis CRY2 owering time alleles. Genetics 167: 13611369 Pan Q, Wendel J & Fluhr R (2000) Divergent evolution of plant NBS-LRR resistance gene homologues in dicot and cereal genomes. J. Mol. Evol. 50: 203213 Paterson AH, Bowers JE, Burow MD, Draye X, Elsik CG, Jang C-X, Katsar CS, Lan T-H, Lin Y-R, Ming R & Wright RJ (2000) Comparative genomics of plant chromosomes. Plant Cell 12: 15231539 Paterson AH, Saranga Y, Menz M, Jiang CX & Wright RJ (2003) QTL analysis of genotype environment interactions aecting cotton ber quality. Theor. Appl. Genet. 106: 384 396 Pauquet J, Bouquet A, This P & Adam-Blondon A-F (2001) Establishment of a local map of AFLP markers around the powdery mildew resistance gene Run1 in grapevine and assessment of their usefulness for marker assisted selection. Theor. Appl. Genet. 103: 12011210 Peleman JD & van der Voort JR (2003) Breeding by design. Trends Plant Sci. 8:330334 Podlich DW, Winkler CR & Cooper M (2004) Mapping as you go: an eective approach for marker-assisted selection of complex traits. Crop Sci. 44: 15601571 Radovanovic N & Cloutier S (2003) Gene-assisted selection for high molecular weight glutenin subunits in wheat doubled haploid breeding programs. Mol. Breed. 12: 5159 Rafalski JA (2002) Applications of single nucleotide polymorphisms in crop genetics. Curr. Opin. Plant Biol. 5: 94 100 Raman H, Karakousis A, Moroni JS, Raman R, Read BJ, Garvin DF, Kochian LV & Sorrels ME (2003) Development and allele diversity of microsatellite markers linked to the aluminium toloerance gene Alp in barley. Aust. J. Agric. Res. 54: 13151321 Ribaut JM & Betran J (1999) Single large-scale marker-assisted selection (SLS-MAS). Mol. Breed. 5: 531541 Ribaut JM, Jiang C, Gonzales-de Leon D, Edmeades GO & Hoisington DA (1997) Identication of quantitative trait loci under drought conditions in tropical maize 2-yield components and marker-assisted selection strategies. Theor. Appl. Genet. 94: 887896 Richards RA (1996) Dening selection criteria to improve yield under drought. Plant Growth Regul. 20: 157166 Riede CR & Anderson JA (1996) Linkage of RFLP markers to an aluminum tolerance gene in wheat. Crop Sci. 36: 905909 Robert M, West MAL, Inai S, Caines A, Arntzen L, Smith JK & St. Clair DA (2001) Marker-assisted introgression of blackmold resistance QTL alleles from wild Lycopersicon cheesmanii to cultivated tomato (L. esculentum) and evaluation of QTL phenotypic eects. Mol. Breed. 8: 217233 Robin S, Pathan MS, Courtois B, Latte R, Carandang S, Lanceras S, Amante M, Nguyen HT & Li Z (2003) Mapping osmotic adjustment in an advanced back-cross inbred population of rice. Theor. Appl. Genet. 107: 12881296

341
Saito K, Hayano-Saito Y, Maruyama-Funatsuki W, Sato Y & Kato A (2004) Physical mapping and putative candidate gene identication of a quantitative trait locus Ctb1 for cold tolerance at the booting stage of rice. Theor. Appl. Genet. 109: 515522 Sanchez AC, Subudhi PK, Rosenow DT & Nguyen HT (2002) Mapping QTLs associated with drought resistance in sorghum (Sorghum bicolor L. Moench). Plant Mol. Biol. 48: 713726 Sardesai N, Kumar A, Rajyashri KR, Nair S & Mohan M (2002) Identication of an AFLP marker linked to Gm7, a gall midge resistance gene and its conversion to a SCAR marker for its utility in marker aided selection in rice. Theor. Appl. Genet. 105: 691698 Schmierer DA, Kandemir N, Kudrna DA, Jones BL, Ullrich SE &Kleinhofs A (2004) Molecular marker-assisted selection for enhanced yield in malting barley. Mol. Breed. Published on line Schneider KA, Brothers ME & Kelly JD (1997) Markerassisted selection to improve drought resistance in common bean. Crop Sci. 37: 5160 Septiningsih Septiningsih EM, Trijatmiko KR, Moeljopawiro S & McCouch SR (2003) Identication of quantitative trait loci for grain quality in an advanced backcross population derived from the Oryza sativa variety IR64 and the wild relative O. rupogon. Theor. Appl. Genet. 107: 14191432 Seyfarth R, Feuillet C, Schachermayr G, Winzeler M & Keller B (1999) Development of molecular marker for the adult plant leaf rust resistance gene Lr35 in wheat. Theor. Appl. Genet. 99: 554560 Shen L, Courtois B, McNally KL, Robin S & Li Z (2001) Evaluation of near-isogenic lines of rice introgressed with QTLs for root depth through marker-aided selection. Theor. Appl. Genet. 103: 7583 Singh RP, Mujeeb-Kazi A & Huerta-Espino J (1998) Lr46: a gene conferring slow-rusting resistance to leaf rust in wheat. Phytopathology 88: 890894 Singh S, Sidhu JS, Huang N, Vikal Y, Li Z, Brar DS, Dhaliwal HS & Khush GS (2001) Pyramiding three bacterial blight resistance genes (xa5, xa13 and Xa21) using marker-assisted selection into indica rice cultivar PR106. Theor. Appl. Genet. 102: 10111015 Smith PH, Koebner RMD & Boyd LA (2002) The development of a STS marker linked to a yellow rust resistance derived from the wheat cultivar Moro. Theor. Appl. Genet. 104: 12781282 Song J, Bradeen JM, Naess SK, Raasch JA, Wielgus SM, Halberlach GT, Liu J, Kuang H, Austin-Phillips S, Buell CR, Hengelson JP & Jang J (2003) Gene RB cloned from Solanum bulbocastaneum confers broad spectrum resistance to potato late blight. Proc. Natl. Acad. Sci USA 100: 91289133 Stuber CW (1994) Success in the use of molecular markers for yield enhancement in corn. Proc. 49th Annual Corn and Sorghum Industry Res. Conf. American Seed Trade Assoc. 49: 232238 Stuber CW, Lincoln SE, Wol DW, Helentjaris T & Lander ES (1992) Identication of genetic factors contributing to heterosis in a hybrid from two elite maize inbred lines using molecular markers. Genetics 132: 823839 Stuber CW, Polacco M & Lynn M Senior (1999) Synergy of empirical breeding, marker-assisted selection, and genomics to increase crop yield potential. Crop Sci. 39: 15711583 Suenaga K, Singh RP, Huerta-Espino J & William HM (2003) Microsatellite markers for genes Lr34/Yr18 and other quantitative loci for leaf rust and stripe rust resistance in bread wheat. Phytopathology 93: 881890 Tabashnik BE (2001) Breaking the code of resistance. Two reports shed light on the molecular basis of resistance to Bt toxins. Nat. Biotechnol. 19: 922924. Tang Y, Sorrels ME, Kochian LV & Garvin DF (2000) Identication of RFLP markers linked to barley aluminum tolerance gene Alp. Crop Sci. 40: 778782 Tanksley SD & Nelson JC (1996) Advanced backcross QTL analysis: a method for the simultaneous discovery and transfer of valuable QTLs from unadapted germplasm into elite breeding lines. Theor. Appl. Genet. 92: 191203 Tanksley SD, Ganal MW & Martin GB (1995) Chromosome landing: a paradigm for map-based gene cloning in plants with large genomes. Trends Genet. 11: 6368 Tartarini S, Gianfranceschi L, Sansavini S & Gessler C (1999) Development of reliable PCR markers for the selection of the Vf gene conferring scab resistance in apple. Plant Breed. 118: 183186 Teulat B, Merah O, Sirault X, Borries C, Waugh R & This D (2002) QTLs for grain carbon isotope discrimination in eldgrown barley. Theor. Appl. Genet. 106: 118126 Teulat B, Zoumarou-Wallis N, Rotter B, Ben Salem M, Bahri H & This D (2003) QTL for relative water content in eldgrown barleys and their stability across Mediterranean environments. Thoer. Appl. Genet. 108:181188 The Arabidopsis Genome Initiative (2000) Analysis of the genome of the owering plant Arabidopsis thaliana. Nature 408: 796815 Thomson MJ, Tai TH, McClung AM, Hinga ME, Lobos KB, Xu Y, Martinez C & McCouch SR (2003) Mapping quantitative trait loci for yield, yield components, and morphological traits in an advanced backcross population between Oryza rupogon and the Oryza sativa cultivar Jeerson. Theor. Appl. Genet. 107:479493 Thornsberry JM, Goodman MM, Doebley J, Kresovich S, Nielsen D & Buckler IV ES (2001) Dwarf8 polymorphisms associate with variation in owering time. Nat. Genet. 28: 286289 Toth B, Francia E, Rizza F, Stanca AM, Galiba G & Pecchioni N (2004) Development of PCR-based markers on chromosome 5H for assisted selection of frost-tolerant genotypes in barley. Mol. Breed. Published on line Turner NC (1997) Further progress in crop water relations. Adv. Agron. 58: 293338 Vagujfalvi A, Galiba G, Cattivelli L & Dubcovsky J (2003) The cold-regulated transcriptional activator Cbf3 is linked to the frost-tolerance locus Fr-A2 on wheat chromosome 5A. Mol. Genet. Genom. 269: 6067 Wang D, Graef GL, Procopiuk AM & Diers BW (2004) Identication of putative QTL that underlie yield in interspecic soybean backcross populations. Theor. Appl. Genet. 108: 458467 Weerasena JS, Steenson BJ & Falk AB (2004) Conversion of an amplied fragment lenght polymorphism marker into a co-dominant marker in the mapping of the Rph15 gene conferring resistance to barley leaf rust, Puccinia hordei Otth. Theor. Appl. Genet. 108: 712719

342
Wei F, Wing RA & Wise RP (2002) Genome dynamics and evolution of the Mla (powdery mildew) resistance locus in barley. Plant Cell 14: 19031917. Werner K, Friedt W, Laubach E, Waugh R & Ordon F (2003) Dissection of resistance to soil-borne yellow-mosaicinducing viruses of barley (BaMMV, BaYMV, BAYMV-2) in a complex breeders cross by means of SSR and simultaneous mapping of BaYMV/BaYMV-2 resistance of var. Chikurin Ibaraki 1. Theor. Appl. Genet. 106: 1425 1432 Williams KJ (2003) The molecular genetics of disease resistance in barley. Aust. J. Agric. Res. 54: 10651079 Williamson VM, Ho J-Y, Wu FF, Miller N & Kaloshian I (1994) A PCR-based marker tightly linked to the nematode resistance gene, Mi, in tomato. Theor. Appl. Genet. 87: 757763 Wittaker JC, Curnow RN, Haley CS & Thompson R (1995) Using marker-maps in marker-assisted selection. Genet. Res. 66: 255265 Wolyn DJ, Borevitz JO, Loudet O, Scwartz C, Maloof J, Ecker JR, Berry CC & Chory J (2004) Light-response quantitative trait loci identied by composite interval mapping and eXtreme array mapping in Arabidopsis thaliana. Genetics 167: 907917 Wu J, Maehara T, Shimokawa T, Yamamoto S, Harada C, Takazaki Y, Ono N, Mukai Y, Koike K, Yazaki J, Fujii F, Shomura A, Ando T, Kono I, Waki K, Yamamoto K, Yano M, Matsumoto T & Sasaki T (2002) A comprehensive rice transcript map containing 6591 expressed sequence tag sites. Plant Cell 14: 525535 Xu J, Narabu T, Mizukubo T & Hibi T (2001) A molecular marker correlated with selected virulence against the tomato resistance gene Mi in Meloidogyne incognita, M. javanica and M. arenaria. Phytopathology 91: 377382 Yaghoobi J, Kaloshian I, Wen Y & Williamson VM (1995) Mapping a new nematode resistance locus in Lycopersicon peruvianum. Theor. Appl. Genet. 91: 457464 Yin X, Stam P, Krop MJ & Schapendonk HCM (2003) Crop modeling, QTL mapping, and their complementary role in plant breeding. Agron. J 95: 9098 Young ND (1999) A cautiously optimistic vision for markerassisted breeding. Mol. Breed. 5: 505510 Yousef GG & Juvik JA (2001) Comparison of phenotypic and marker-assisted selection for quantitative traits in sweet corn. Crop Sci. 41: 645655 Zhang Z, Xu J, Xu Q, Larkin P & Xin Z (2004) Devlopment of novel PCR markers linked to the BYDV resistance gene Bdv2 useful in wheat for marker-assisted selection. Theor. Appl. Genet. 109: 433439 Zhang J, Zheng HG, Aarti A, Pantuwan G, Nguyen TT, Tripathy JN, Sarial AK, Robin S, Babu RC, Nguyen BD, Sarkarung S, Blum A & Nguyen HT (2001) Locating genomic regions associated with components of drought resistance in rice: Comparative mapping within and across species. Theor. Appl. Genet. 103: 1929 Zhang JX, Nguyen HT & Blum A (1999) Genetic analysis of osmotic adjustment in crop plants. J. Exp. Bot. 50: 291302 Zhang TZ, Yuan YL, Yu J, Guo WZ & Kohel RJ (2003) Molecular tagging of a major QTL for ber strength in Upland cotton and its marker-assisted selection. Theor. Appl. Genet. 106: 262268 Zheng BS, Yang L, Zhang WP, Mao CZ, Wu YR, Yi KK, Liu FY & Wu P (2003) Mapping QTLs and candidate genes for rice root traits under dierent water-supply conditions and comparative analysis across three populations. Theor. Appl. Genet. 107: 15051515 ` Zhou F, Kurth J, Wei F, Elliot C, Vale G, Yahiaoui N, Keller B, Somerville S, Wise R & Schulze-Lefert P (2001) Cellautonomous expression of barley Mla1 confers race-specic resistance to the powdery mildew fungus via a Rar1independent signaling pathway. Plant Cell 13: 337350 Zhou PH, Tan YF, He YQ, Xu CG & Zhang Q (2003) Simultaneous improvement for four quality traits of Zhenshan 97, an elite parent of hybrid rice, by molecular marker-assisted selection. Theor. Appl. Genet. 106: 326331 Zhu YL, Song QJ, Hyten DL, Van Tassel CP, Matukumalli LK, Grimm DR, Hyatt SM, Fickus EW, Young ND & Cregan PB (2003) Single nucleotide polymorphisms (SNPs) in soybean. Genetics 163: 11231134