Marker-assisted selection and gene pyramiding for resistance to bacterial leaf blight disease of rice (Oryza sativa L.)

Abstract Marker-assisted selection and gene pyramiding are very important breeding strategies for conferring broad spectrum and durable resistance against diseases causing yield loss in rice. One such disease causing major set backs in rice production is bacterial leaf blight (BLB) caused by the pathogen Xanthomonas oryzae pv. oryzae. Molecular markers are very essential in both marker-assisted selection and pyramiding of genes, hence, many molecular marker techniques have already been developed. Presently, the most commonly used ones are DNA-based markers also known as molecular markers. The molecular markers are classified into two major categories based on the techniques used for detecting them. These are hybridization and polymerase chain reaction-based markers. Other types of markers available include the morphological (traditionally based) and biochemical (enzyme-based) markers. Host plant/varietal resistance is the most suitable means for controlling BLB disease of rice. Marker-assisted gene pyramiding has the potential to accelerate the breeding programmes and guarantee the durability of resistance conferred in the host plant. Therefore, this paper uncovers the utilization, economic importance, limitations and future prospects of marker-assisted selection and gene pyramiding for resistance to BLB disease of rice.


Introduction
Rice serves as staple food crop and source of energygiving food for more than half of the world's population, which is over 3.5 billion people. The demand for rice is still increasing in Asia as the consumption rate is at least 90% and it is globally projected that the demand for rice will rise up to 650 million tonnes by 2050 [1]. At present, the population of the world is increasing geometrically and the demand for staple food like rice is also increasing, so bacterial leaf blight (BLB) caused by Xanthomonas oryzae pv. oryzae (Xoo) constitutes a major threat to rice production and invariably, to food security [2]. In order to guard against the threat posed by the disease, it is important to make use of varieties of rice incorporated with resistance genes. It is also important to consider pyramiding of several genes for durable resistance to BLB of rice in order to guarantee long-term resistance [3]. Biotechnology makes use of two strategies in molecular breeding of rice. This involves marker assisted selection also known as marker assisted breeding, while the other one would require development of genetically modified crops (GMOs) [4]. A genetic marker is any assayable phenotype or visible trait, for which alleles at individual loci segregate following the laws of Mendel. Marker assisted selection is a technique that helps to make conventional breeding more efficient but not a substitute for conventional breeding [1]. It offers genetic tools for selection of target genes from the existing germplasm that would be used in breeding activities but does not include genetic engineering which involves the transfer of isolated gene sequences [5,6].
Molecular or genetic markers are the stable and inheritable variation at DNA sequence level, which could be measured or detected by using appropriate methods and subsequently used to identify a specific genotype [7]. Molecular/DNA-based markers are regarded as steady landmarks in the genetic make-up of plants. They are identifiable sequences of DNA that are found at specific locations of the genome and are transmitted by the standard laws of inheritance from one generation to the next. Molecular markers are tools used to study diversity at DNA level (polymorphism) and help breeders to identify specific chromosome segments that contain genes of interest.

Genetic basis of bacterial leaf blight resistance in rice
The BLB pathogen Xoo belongs to the c-division of Gram-negative proteobacteria. This pathogen grows in the vascular system until the xylem vessels are clogged with bacterial cells and extracellular polysaccharides. There are several races of Xoo. All the races secrete race-specific effectors in the xylem which activate individual response leading to infection. Xoo also release avirulence factors that bind and activate the transcription of genes known as resistance genes, which activate a resistance response [8]. The host specificity is determined by avirulence factors through the interaction of genes. It also reduces the virulence of Xoo as the host recognizes them. Resistance genes have evolved to give resistance to specific Xoo races due to the fact that each race of Xoo produces different virulence and avirulence factors. The hypersensitive response and pathogenicity genes are used to determine the interactions of Xoo with plants. These are required for hypersensitive response (HR) in resistant/non-host plants and for pathogenicity in susceptible/host plants [9]. Substantial attempt has been made to identify genes responsible for pathogenesis of Xoo and roles of each gene in causing disease. The genome sequence of Xoo strains like KACC10331, MAFF311018 and PXO99A has made the knowledge of pathogenesis genes clearer [10].
The inheritance of BLB has been critically studied. Plants usually contain a single locus which confers resistance against a complementary avirulence. This has been shown from genetic analysis of many plant pathogen interactions [10]. Japan, IRRI, Sri Lanka, India and China were among the first to study the inheritance of BLB resistance in rice. Researchers found it difficult to characterize and differentiate the R-genes due to diversity of Xoo strains in various countries. The identical differential standard was set up to compare the genes identified [10]. Currently, 42 BLB R-genes have been discovered in rice [10][11][12]. These genes were identified from mutant populations, cultivated and wild rice species and confer resistance against different Xoo strains ( Table 1). Out of the 42 genes, 14 are recessive while nine have been cloned and characterized. The genes encode different types of proteins which suggest multiple mechanisms of R-gene mediated Xoo resistance [13][14][15]. Majority of these genes give complete race-specific resistance to the pathogen. The genes have either been introgressed singly or pyramided in breeding for BLB resistance in rice [16]. Due to artificial and natural selection of BLB resistant genes, the bacterial races keep evolving. Therefore, it would be required to critically explore the new resistant resources to guard against the evolved races.
Development of molecular markers is one of the most recent advances in DNA technologies and has made it possible to reveal a large number of genetic variations within a population [45]. Consequently, they are used in evaluating the genetic basis of the phenotypic variability. Molecular markers are more or less DNA loci with no function and no impact on plants' performance. However, they are located either near or within major or functional genes so that their presence indicates to a large extent the presence of the functional gene [46,47].

Types of markers
Generally, there are three types of markers according to Akhtar et al. [7].

Morphological markers
The morphological markers are known as the traditional markers. The morphological mutant characters in the given population are mapped. Linkage to a desirable or non-desirable character is determined while indirect selection is done using the physically identifiable mutant for the trait [48,49]. Morphological markers have many non-desirable factors associated with them. Hence, the use of such markers in breeding is labour intensive, space and time consuming.

Biochemical markers
These markers are superior to traditional/morphological markers. Generally, unlike morphological markers, they are independent of growth conditions in the environment. Biochemical markers involve the use of isozymes (enzymes) that are expressed in the plants cells [46]. After extraction, the enzymes are run on denaturing electrophoresis gels while the denaturing component eliminates the enzyme's secondary and tertiary structure and separates the enzymes based on net charge and mass [50]. Polymorphism takes place at the amino acid level enabling the detection of singular peptide polymorphism which is used as a marker (polymorphic biochemical marker). However, the major limitation of biochemical markers is that isozymes do not produce sufficient polymorphism since most commercial breeds of plants are genetically very similar [7,51].

Molecular markers
Molecular markers are DNA-based markers. Molecular markers are practically unlimited and also phenotypically neutral. The use of such markers has enabled scanning of the entire rice genome and assigning landmarks in high density on every chromosome of the plant [7,52]. Examples of molecular markers are restriction fragment length polymorphism (RFLP), cleaved amplified polymorphic sequences (CAPS), sequence characterized amplified regions (SCARs) and other polymerase chain reaction (PCR)based markers. A suitable molecular marker must be polymorphic, randomly and frequently distributed throughout the genome, co-dominant in inheritance, reproducible, easy and cheap to detect.
A rice marker should be closely linked to target loci. Less than 5 cM genetic distance is preferable [53]. The reliability of a rice marker in predicting the phenotype would increase when flanking or intragenic markers are used. Although it may be difficult to obtain, some marker techniques require large quantity and high quality of DNA [54]. This increases the cost of the scheme. The simplicity and the time  [11] requirement of the marker technique are important considerations. Highly desirable methods should be very simple and quick. An ideal rice marker is also expected to be highly polymorphic. This implies that it should be able to discriminate or separate between different genotypes [55]. The cost effectiveness of the marker assay is also important for the marker assisted selection to be feasible. Simple sequence repeats (SSRs) or microsatellites are often used in rice and other cereal crops [56]. The major problem of microsatellite markers is that they particularly need polyacrylamide gel electrophoresis and generally provide information only for a single locus per assay. Breeders have been able to overcome the challenge by selecting SSR markers that have very large size differences for detection in agarose gels. Multiplexing several markers in a single reaction is also helpful. The microsatellites also require enough time and money to develop, and often, some orphan crop species lack sufficient numbers for high density mapping. Other DNA-based markers like the RFLPs that are linked to a target gene or QTL such as SCAR, sequence tagged site (STS) or single nucleotide polymorphism (SNP) markers are also properly utilized for marker assisted selection [57,58]. On the basis of the techniques used for their detection, molecular markers are classified into two major categories: hybridization-based markers and PCR-based markers [46].

Hybridization-based markers
The traditional, or first-generation, RFLPs require the use of an appropriately labelled DNA probe to select the gene of interest from a digested DNA sample by hybridization. The generated DNA fragments are separated by gel electrophoresis and polymorphism, or genetic variation among genotypes, is visualized as hybridization bands, wherein identical genotypes produce similar banding patterns while different genotypes have different banding patterns. Thus, the process involves DNA extraction, digestion of the isolated DNA with restriction enzymes, hybridization of the digested DNA fragments with a labelled DNA probe, separation of the hybridization products on agarose or polyacrylamide gel and analysis of the hybridization bands for DNA variation [59].

PCR-based markers
PCR-based markers are molecular markers that do not require the probe hybridization step. Their development has led to the discovery of several useful and easy-to-screen new generation markers such as random amplified polymorphic DNA (RAPD), amplified fragment length polymorphism (AFLP), microsatellites or SSRs, SNP, expressed sequence tags (EST), etc. [46,60]. Being PCR-based, this category of markers requires the use of primer pairs to select specific regions of the DNA for which genetic variation (polymorphism) is measured [61]. Primers are short sequences of nucleotides designed to be complementary to specific regions of DNA. Consequently, primers are used to select, by base pairing, a specific segment of DNA for amplification by PCR and sequence analysis [38]. They, therefore, initiate the amplification of a particular DNA segment. The amplification products of DNA from different genotypes are separated on a gel to observe size-dependent variation in banding pattern and may be further sequenced to identify sequence variations [62].
PCR-based molecular marker analysis involves DNA extraction from the source material, determination of quantity and quality of the isolated DNA, amplification of the isolated DNA using PCR and gel electrophoresis [63]. If sequencing is required, the amplification products are usually purified, subjected to sequencing PCR and further purified before sequencing [56,64].

DNA-based markers for rice
There are several advantages of the use of DNA markers for trait selection. Collard and Mackill [65] reviewed the application of DNA-based markers in the breeding of BLB disease resistant rice varieties. They both emphasized the application of the markers and their limitations in crop improvement. Akhtar et al. [7] listed the following advantages of molecular markers: Time saving: Any part of the rice crop can be sampled for isolation of genomic DNA and information about the trait of choice can be obtained using tightly linked polymorphic markers before fertilization, thereby facilitating a more informed genetic crossing by breeders.
Consistency: Molecular markers are not affected by variation in the environment. Environmental factors usually affect the phenotypic screening of genetic traits.
Biosafety: Diagnostic tests to confirm the absence or presence of disease resistance traits can be carried out using molecular markers that are tightly linked to the gene of interest without adopting inoculation of pathogen in the greenhouse or field. Also, DNA-based markers would facilitate the introgression of genes into elite cultivars before the epidemics of certain pathogens.
Efficiency: Early generation evaluation of breeding lines with molecular markers can guide breeders in rejecting undesired progenies while improving the genetic quality of desirable breeding materials.
High accuracy in complex traits selection: The use of DNA-based markers makes it easier to select multigenic traits. Molecular markers that are linked to QTLs enable them to be treated like single Mendelian factors. McCouch et al. [66] initially developed RFLPframework map with 250 markers which was later supplemented with additional markers. The additional markers include RAPD, SSR or microsatellite markers, STS and AFLP markers [67]. Tightly linked markers have been used to map numerous agronomically important genes of rice. About 18,828 class-I SSR sequences were listed by the International Rice Genome Sequencing Project [68]. Dai et al. [69] reported that several BLB disease resistance genes were isolated from the plant's genome with the aid of saturated rice molecular map and genome sequencing information. The BLB resistance genes include Xa1, xa5, xa13, Xa21, Xa26 and Xa27 [17,28,70,71].

Simple sequence repeats (SSRs) markers
Microsatellites, also known as SSRs, are tandemly arranged repeats of short DNA motifs (1-6 bp in length) that frequently exhibit variation in the number of repeats at a locus. Litt and Lutty [72] first coined the term 'microsatellite'. SSR markers have high polymorphism and abundance in nature; hence these markers have become a valuable source of genetic information. Microsatellites are ubiquitous in eukaryotic genomes and exhibit highly variable numbers of repeats at a locus, and polymorphism in the repeat motif is referred to as simple sequence length polymorphism (SSLP). Their hyper variability, abundance, co-dominant and multi-allelic nature make them valuable as genetic markers. Using PCR, microsatellites can be amplified by the use of primers that are complementary to the flanking region. When PCR products from different plants vary in length due to variation in the number of repeat units in the SSR, it is taken as length polymorphism [73,74]. Microsatellites, or SSR markers, are regarded as the ideal markers (genetic) for rice out of the several available DNA-based markers. Candit and Hubbell [75] observed heavy presence of SSRs in plants, and that was the first published study on microsatellites in plants. Subsequently, microsatellites have been developed for different plant species including barley [76], rice [77,78], soybean [79], etc.
Initial effort to map microsatellites in rice was made by Zhao and Kochert [78] using (GGC)n, followed by mapping of (GA/AG)n, (ATC) 10 and (ATT) 14 microsatellites by Panaud et al. [80]. Presently, over 20,000 SSR markers (microsatellites), which are well distributed landmarks all over the genome of rice, have been developed [68,[81][82][83]. Significantly greater allelic diversity of microsatellites over random fragment length polymorphism (RFLPs) has been reported [84]. Research has also shown that genetically mapped microsatellite markers cover the entire rice genome with at least one microsatellite per 157 kb (<1 cM) [67,85]. A study comparing 12 Japonica cultivars found microsatellite's mean polymorphism of at least 1.5 times higher than AFLP and RAPD markers [86]. Intra and inter cultivar polymorphism between the cultivated rice and the landrace was detected by Provan et al. [87] using SSR markers. Xu et al. [88] found that genetic diversity is present to a greater extent in landraces compared to improved cultivars in their molecular diversity studies of Philippine landraces and cultivars. Enoki et al. [89] reported the use of SSRs in estimating the diversity among parental lines, inbreds and varieties of rice. SSR markers have also been used for tagging and mapping of target genes and/or major QTL associated with Xa30t for resistance to BLB [12], Xa7 and xa13 gene from Basmati-48, Basmati-51A, Basmati-334, and Basmati-370A [90], dominant Rsb1 for resistance to sheath blight [91], sub1A for tolerance of submergence [92,93], restorer gene for HL type cytoplasmic male sterile lines [94] and fragrance gene in rice [95].

Marker-assisted selection
Marker-assisted selection is the selection for a trait on the basis of genotype using associated markers rather than the phenotype of the trait. Thus, marker-assisted selection of breeding materials involves cultivar identification, assessing the genetic diversity and purity, selection of parents, identifying the genomic regions under selection and study of heterosis [96]. It makes use of DNA-based markers that are tightly linked to the gene of interest to either assist or replace phenotypic evaluation. Identification of plants that carry specific genes or QTLs is based on their genotype and this is done by determining the allele of the molecular (DNA-based) marker [54]. Marker-assisted selection is very promising in the development of rice varieties resistant to BLB disease [5,97]. The use of molecular marker information to identify and select specific genotypes also gives a clear understanding of the molecular marker. Before now, breeders relied on information about how plants or animals perform to make inferences about how good their genes were, a practice that is time consuming, laborious and less effective [98][99][100]. A direct handle on the genes controlling the traits of interest could lead to much faster progress. Some traits are controlled by single genes, but most traits of economic importance are controlled by a large number of genes [4,101]. Such traits are called quantitative traits and the genes locations on the chromosomes are quantitative trait loci (QTL). However, some of the genes in the QTL have larger effects than others. Such genes are called major genes. In effect, QTL actually refers to the major genes following the pattern of inheritance of the QTL or the major genes that might be helpful in selection. In practice, it is not easy to observe inheritance at QTL itself, but it is possible to observe the inheritance of marker sequences located close to or within the QTL. Molecular markers are chosen for their proximity to QTL. Well-designed experiments can be used to identify marker sequences located close to or within major genes on the chromosomes or genomes [102,103].
The molecular markers should be able to predict the phenotype reliably. A marker can either be located within the gene of interest or be linked to a gene determining a trait of interest, which is the most common case. Molecular markers can also be used in confirmation of true F 1 , testing the seeds for genetic purity, protecting the cultivar, establishment of breeding strategies, linkage map construction, gene mapping and also mapping of QTLs associated with biological processes [5]. The following considerations are necessary in applying DNA-based markers in marker-assisted selection [96]: i. Reliability: for a molecular marker to be reliable, it should be tightly linked to target traits (<5 cM genetic distance is preferable). The use of intragenic or flanking markers would increase the reliability of the markers to predict phenotype rapidly. ii. Quantity and quality of DNA: The techniques of some markers require DNA in a large amount with high quality. This increases the cost of the procedure and is sometimes difficult to obtain in practice. iii. Technical procedure: it is required that molecular markers should be transferable among researchers and also have high reproducibility across laboratories. Time and level of simplicity are to be considered. Therefore, a high throughput simple and quick methods are desirable factors to be considered. iv. Polymorphism: DNA-based markers should have a high level of polymorphism, hence, be hypervariable. Such markers should also possess codominant inheritance of homozygotes and heterozygotes in progenies that are still segregating. v. Affordability: the cost of DNA markers ought to be affordable. This implies that the assay should be cheap, user-friendly and easy to apply for efficient screening of large samples/populations. The cost effectiveness of the marker assay should be considered in order to guarantee the feasibility of the marker-assisted selection.
Importance of quantitative trait loci (QTLs) mapping for marker-assisted selection QTL is a gene or chromosomal region that affects a quantitative trait. The QTLs are regarded as the underlying genes controlling quantitative traits. With the aid of QTL mapping experiments, the breeder would be able to identify the DNA-markers that are linked to specific genes and/or QTLs [104]. Although, it was previously thought that markers could be used directly in marker-assisted selection, but research has shown that QTL mapping is actually the basis for developing markers in marker-assisted selection [105]. However, environmental effects, population size and type, number of replication of phenotypic data, genotyping errors, etc. are some of the factors that affect the QTL mapping accuracy [106][107][108]. These factors are very relevant for more complex polygenes which controls quantitative characters comprising several QTLs each with relatively small effects. Therefore, it has been widely accepted in recent years that QTL confirmation, validation and/or additional marker testing steps are necessary after QTL mapping and prior to markerassisted selection [105,109,110]. These steps may include conversion of the marker which is required to improve the reliability and make marker genotyping simpler for marker assisted selection. Confirmation of QTL is necessary to ascertain the accuracy of the results obtained from the primary QTL mapping study [111,112]. In order to verify the effectiveness of the QTLs in different genetic backgrounds, validation of QTL is done. The validation of a marker involves testing the level of polymorphism of most tightly-linked markers and the reliability of markers for predicting phenotype [84,113].

Marker-assisted introgression
Introgression simply means the transfer a specific gene of interest from one species to another through hybridization and repeated backcrossing. Hospital [114] defined introgression as the process where a target gene or QTL from a plant in population 'A' is inserted to another plant in population 'B' by crossing the two of them and then repeatedly backcrossing to 'B' which is known as the recipient and/or recurrent parent. In this case, DNA markers are useful in controlling the presence of the target gene or QTL. This is also useful in accelerating the recovery of the background genome to the recipient type. Introgression using molecular markers is very effective in incorporating genes or QTLs from landraces, because the time required to produce an improved variety and the issue of linkage drag are reduced [60].

Marker-assisted backcross breeding (MABB)
Markers can be applied in backcross breeding at three stages of foreground, recombinant and background selections. At the first stage of foreground selection, markers are used to select for the target trait. The use of markers here are very effective because some traits have laborious phenotypic selection procedures or recessive alleles whose effects could have been masked by the dominant genes. The recombinant selection is the second stage which involves selecting backcross progeny with the specific gene and tightlylinked flanking markers so that linkage drag can be reduced. The third stage is referred to as the background selection and it involves selecting backcross progeny (i.e. progeny already selected for the target trait) using background markers. However, the background markers must not be tightly linked markers but should be ideal rice markers [53,56]. This implies that markers can be used to select against the donor genome, i.e. reduce the genetic contribution of the donor parent while accelerating the recovery ratio of the recurrent parent genome [101].

Marker-assisted pyramiding
The process of simultaneously incorporating multiple genes or QTLs into a single genotype is known as pyramiding [115]. Pyramiding of genes is very essential for broad spectrum resistance to BLB of rice and to guarantee the durability of the disease resistance. Molecular markers facilitate selection because molecular or DNA-based marker assays are non-destructive and markers encoding for several target genes can be tested using a single DNA sample without phenotyping. Combining multiple disease resistance genes or QTLs so as to provide durable disease resistance has been the most widespread application of pyramiding in plant breeding [60,116]. Although it is still possible to use conventional breeding, it is very difficult or nearly impossible at the early generations due to the need to phenotypically screen each plant for all traits being tested. This makes it very hard to evaluate plants from some populations like the F 2 or for traits with destructive bioassays [4,117].

Early generation marker-assisted selection
Early generation stages are the most important stages to use markers for selection of plants. The essence of this is such that the plants that do not carry required genes can be deselected. In breeding of plants for resistance to BLB, the plants to be selected must carry the disease resistance genes in addition to other useful agronomic traits. The implication of this is that there would be an efficient evaluation of fewer breeder lines with a cost effective design at the later stage of breeding [118].

Combined approaches
Sometimes, phenotypic screening can be combined with marker-assisted selection in order to maximize the genetic gain especially where QTL mapping could not identify some QTLs. A combined approach is also useful to find the level of recombination between a marker and a QTL. It also reduces the size of the population for traits [105]. Although marker genotyping seems to be easier and cheaper compared to phenotyping, marker-assisted phenotyping (tandem selection) can be utilized where it is obvious that marker genotyping alone is not completely accurate or that phenotyping alone would cost more money [1,119].

Advantages of marker-assisted selection
Plants can be selected at their seedling stages. There is high reliability of single plant selection. On average, the marker-assisted selection is a simpler process compared to phenotypic evaluation [53]. Given these advantages, marker-assisted selection is capable of increasing the efficiency and effectiveness of a breeding programme. For instance, marker genotyping or DNA marker tests can be used to replace large population field trials. This saves the breeder time and difficult labour thereby saving costs as well. Also, the molecular marker based selection of plants is more reliable due to the variability in the environmental factors which affect field trials [5,120]. The use of foreground markers in marker-assisted selection enables the breeder to discard plants that are not carrying the target genes at the early stage of breeding. This is another advantage of marker-assisted selection because the number of plants to be tested is reduced, thereby increasing the efficiency of target trait selection with effective breeding design. In addition, marker-assisted selection enables the acceleration of recurrent parent genome recovery ratio with the aid of background markers during marker-assisted backcross breeding [7,51,64].

Limitations of marker-assisted selection
The prohibitive cost is one of the limitations of using marker-assisted selection in plant breeding today. In addition, two more factors limiting the use of markerassisted selection in plant breeding are the equipment and consumables. These two are required to establish and maintain a molecular marker laboratory and therefore, their costs should be considered [7,121]. Development of markers also requires large initial costs and should be put into consideration. In some developing nations, electricity supply poses a major challenge to preserve the markers at a steady temperature. Although, few numbers of reports are available in literature on the economics of marker-assisted selection compared to conventional breeding, the cost effectiveness of the two approaches considerably varies in different studies. In marker assisted backcross breeding (MABB), the initial cost implication of adopting markers may appear to be higher in the short run.
In the long run, however, the release of newly developed varieties in a short time as a result of markerassisted selection could turn into higher profits compared to the cost of production [122,123]. Markers have low reliability to determine the phenotype. This is another limitation of using markerassisted selection for line development. The thoroughness of the primary QTL mapping study is always responsible for this. Sampling bias mostly in small populations can still affect the QTLs that are detected with high LOD scores which explain a large proportion of the phenotype, hence, may not be utilized for marker-assisted selection. Also, the genetic make-up of the plant may determine the effect of a QTL. This explains why the effects of QTL and the reliability of markers, otherwise referred to as QTL/marker validation, should be tested before embarking on markerassisted selection [109,124].
Recombination or crossing-over may occur during DNA copying and this is a major problem of markerassisted selection technology, because with recombination, one cannot be sure of which marker variant or allele is associated to each gene variant or allele. This is why molecular markers may be described as direct or indirect markers. A marker is a direct one when it is located within the major gene, and an indirect or linked when it is located near the major gene. Recombination is a function of the distance between a gene or QTL and the marker linked to it. The larger the distance between a maker and a major gene is, the greater the problem of recombination [104,105].
If a marker is located within a major gene, recombination will not occur at all. In this case, knowledge of the marker allele tells us directly about the gene variant. Use of direct markers for selection is straightforward. The problem is how to decide based on linked or indirect markers. To do this, the pattern of inheritance of the major gene and its associated marker is first established. Consequently, progenies that perform well phenotypically may lack the preferred marker allele, while progenies that perform poorly may have the desired marker variant. Generally, linked markers with high recombination frequency (more than 25%) should be avoided, while those with 75% fidelity and above are efficient [7,106]. This means that if 75% or more of the genotypes that performed well phenotypically show the preferred marker allele, the marker is useful in marker-assisted breeding. If the genotypes are less than 75%, the marker is inefficient [52].

Gene pyramiding
Gene pyramiding can be defined as a process of combining two or more genes from multiple parents to develop superior lines and varieties. Pyramiding involves combining or stacking multiple genes leading to the simultaneous expression of more than one gene in a variety. Molecular markers aid in selecting the best plants with which to proceed. The term gene pyramiding is used in agricultural research to describe a breeding approach to achieve disease control and higher crop yield. The development of molecular genetics and associated technology like markerassisted selection has led to the emergence of a new field in plant breeding called gene pyramiding. The gene pyramiding scheme can be distinguished into two parts [116]. The first part is called a pedigree, which aims at cumulating all the target genes in a single genotype called the root genotype. The second part is called the fixation step which aims at fixing the target genes into a homozygous state, i.e. to derive the ideal genotype from the one single genotype. Molecular marker genotyping can facilitate the gene pyramiding process by reducing the number of generations that breeders must evaluate to ensure they have the desired gene combination. Gene pyramiding is an important strategy for germplasm improvement [116,117,125,126].

Main factors affecting gene pyramiding
Characteristics of the target trait/gene Gene pyramiding will prove to be more successful when the genes to be stacked are characterized well functionally, using perfect markers. The presence or absence of the gene of interest can be traced using one or two markers per gene. It is preferable to use the bulk segregant analysis (BSA) method to identify tightly linked markers to a major gene [119,127].

Reproductive traits
The number of seeds produced by a plant determines its propagation capability. Collection of seeds from many F 1 plants from a cross involving two homozygous parents would produce a fairly large F 2 population. The F 3 generation requires that seeds be collected from only a single plant. However, the efficiency of cross fertilization could largely constraint some species of crop. It is also important to note that some other reproduction related problems such as incompatibility in crossing may arise when wild relatives are used as donor parent contributing the desirable genes into the cultivated rice variety [128,129].
It is difficult to pyramid genes using only conventional breeding due to the effect of linkage drag, which is not easy to break after repeated backcrossing. With the introgression of two or more genes, phenotyping alone cannot clearly separate the effect of each individual gene. Each gene confers resistance to several races of the same pathogen. There is also the issue of the masking effect where dominant and recessive genes are involved. Marker assisted pyramiding enables the use of DNA-based markers that are tightly linked with each of the disease resistance genes for selection of plants with multiple genes [4].
Success has been recorded in earlier attempts to generate three-gene pyramids in Pusa Basmati, PR106 and Samba Mahsuri cultivars [129][130][131]. Although susceptible to various pests and disease, hybrid rice is suitable for increasing the productivity of rice. There is a longer period for BLB susceptibility in a hybrid rice compared to rice varieties. 'Kresek' at seedling stage, defoliation and death of the entire plant under severe infection at the stage of maximum tillering are some symptoms of BLB. Some management practices, such as leaf cutting, also make the plant more vulnerable. KMR3 is among the best restorer lines for breeding hybrid rice and it is used as the donor parent of KRH2 (a non-aromatic hybrid). On the other hand, PRR78 is the donor parent of Pusa RH10, an aromatic hybrid which is well known for its aroma and superfine grain quality. Research has shown that these two lines are highly susceptible to BLB causing a yield reduction up to 40% [64]. Many hybrids have been developed using IR58025A, a well known cytoplasmic male sterile (CMS) line, as female line. Examples are KRH2, Pusa 6A and Pusa RH10 [64]. Pyramiding of resistant genes alone in the restorer lines would not be sufficient since the hybrid will carry the resistance genes in a heterozygous state. This would reduce the level of resistance incorporated. Therefore, the recessive xa5 and xa13 genes should also be introgressed in the recipient line. Currently, it is impossible to transfer the BLB resistance genes directly into the 'A' lines because the genes are available in partial restorer backgrounds. This requires transferring the genes to the maintainer backgrounds first. They can be easily transferred to the male sterile background after stabilization. When the genes have been successfully transferred to the CMS and the maintainer line base, it would be easier to transfer multiple resistance to every other CMS line without encountering fertility restoration problems. It is also important to insulate both parents in order to get resistance in the hybrid. This requires that BLB resistance genes be introgressed into the genetic background of the maintainer and the restorer lines. Shanti et al. [64] reported that a four-gene combination (Xa4, xa5, xa13 and Xa21) was the most effective gene combination that confers broad spectrum resistance. The result showed that it did not show any sign of loss of resistance to varying strains of the BLB pathogen.

Marker-assisted pyramiding
The process of introgressing various genes into a single cultivar is known as pyramiding. Pyramiding can be done through conventional techniques, but it is difficult to identify plants containing multiple genes using that approach [132]. Evaluation of all traits tested is required in conventional phenotypic selection. Therefore, it could be difficult to assess traits with destructive bioassays on plants from populations like the F 2 . Molecular markers can enhance selection because molecular markers are not destructive and markers for multiple specific genes could be tested using a single DNA sample without phenotyping [133]. Introgressing various disease resistance genes into a particular variety is the most wide utilization of pyramiding. The aim of pyramiding has been to develop varieties with stable and durable resistance to diseases, since pathogens usually break single gene resistance after some time due to mutations or emergence of new races of pathogen. Previous studies have shown that incorporating multiple genes can guarantee broad spectrum resistance [134].
Initially, it was not easy to pyramid multiple genes of resistance because they displayed a phenotype, which led to progeny tests to confirm the particular plant with multiple genes. Now, by the use of linked molecular markers, the breeder can identify the number of resistance genes in the plant. Pyramiding may involve incorporating resistance genes sourced from over two parents. For instance, Pradhan et al. [117] pyramided three BLB (xa5þxa13þXa21) resistance genes for broad-spectrum resistance in deepwater rice variety, Jalmagna. Three major genes (Pi1, Piz-5 and Pita) were pyramided by Hittalmani et al. [135] using RFLP markers. Hittalmani et al. [135] and Castro et al. [134] combined genes originating from three parents for rice blast and stripe rust in barley, respectively. Marker assisted pyramiding was also proposed as an effective approach to produce three-way F 1 cereal hybrids possessing durable resistance. Strategies for marker assisted pyramiding of linked genes of interest have also been evaluated [61]. It is preferable to pyramid over successive generations in terms of minimal marker genotyping for many linked target loci. In theory, marker-assisted selection could be used to pyramid genes from different parents.

Marker-assisted selection in rice breeding for bacterial leaf blight resistance
Marker-assisted selection enhances the identification of rice cultivar with multiple resistance genes. For example, the IRBB60 has four BLB resistance genes, and namely Xa4, xa5, xa13 and Xa21 [1,62]. There are about 40 known Xa-genes conferring resistance against the BLB of rice and each of them can be singly identified using marker assisted selection. Major genes conferring resistance to various diseases in several crops have also been mapped with linked DNA markers, thereby facilitating marker-assisted selection for disease resistance in such species. The markerassisted selection scheme has also been used successfully in selection for resistance when pathogens were absent [136], multiple genes pyramiding for durable resistance against rice BLB [137] and in developing multiple disease-resistant cultivars [138,139]. Many rice BLB resistance genes have tightly-linked primers while some are already cloned for developing BLB resistant rice varieties. Some examples are Xa1, xa5, xa13, Xa21, Xa26 and Xa27. Excluding recessive xa-5 and xa-13 genes, the bacteria leaf blight disease resistant genes mentioned here are dominant genes. The markers are developed from the primer sequence information, which is broadly utilized in marker-assisted selection [70]. Due to the presence of molecular markers developed from the BLB resistance genes, the breeder can now pyramid multiple genes of resistance into a susceptible rice variety with good agronomic value.
The pyramiding of dominant Xa4 and Xa21 genes led to development of an improved 'indica' rice variety with 'broad spectrum durable resistance' to BLB. Pyramiding of Xa4 þ xa5 þ Xa21 BLB resistance genes expressed effective resistance to virulent BLB isolates of Korea in comparison to single resistance genes which had their resistance broken after a short period of time and had become susceptible [140]. The xa5þxa13þXa21 resistance genes were pyramided through marker-assisted selection into an indica rice (PR106) cultivar, which exhibited effective resistance to Indian races of BLB [129]. Angke and Conde are two commercially cultivated rice cultivars released for cultivation in Indonesia in 2002 with gene pyramids Xa4 þ xa5 and Xa4 þ Xa7. NSIC-Rc142 and NSIC-Rc154 rice cultivars were also released in the Philippines and possess the genes Xa4 þ xa5 þ Xa21 pyramid. The BLB resistance genes were introgressed into the genetic background of the susceptible IR64 cultivar through marker-assisted selection [141]. Genes for basmati quality from PB-1 and BLB resistance from IRBB55 xa13þXa21 have been pyramided. Rice lines pyramided with BLB disease resistance genes and their reactions to various races of Xoo are presented in Table 2.

Conclusions
The future prospects of marker-assisted selection in rice breeding are very promising. The adoption of markers in rice breeding programmes is expected to increase in the future. The successes and challenges recorded in using marker-assisted selection would be critical to determine the future endeavours in markerassisted selection. The major challenges for increased adoption and impact of marker-assisted selection for breeding rice for BLB resistance include the costeffectiveness. However, developing effective strategies for using markers would facilitate a greater level of the adoption. Establishment of facilities for marker genotyping as well as training the members of staff in rice breeding stations in different countries are important strategies for future prospects. Identification of tightly-linked markers, establishment and curation of public databases for QTL/marker data would also help to increase the adoption of marker-assisted selection in breeding for BLB resistance in rice. Hence, the available resources for developing new markers from DNA sequence data as a result of rice genome sequencing and research in functional genomics would be helpful.