Quantitative trait loci from identification to exploitation for crop improvement

Advancement in the field of genetics and genomics after the discovery of Mendel’s laws of inheritance has led to map the genes controlling qualitative and quantitative traits in crop plant species. Mapping of genomic regions controlling the variation of quantitatively inherited traits has become routine after the advent of different types of molecular markers. Recently, the next generation sequencing methods have accelerated the research on QTL analysis. These efforts have led to the identification of more closely linked molecular markers with gene/QTLs and also identified markers even within gene/QTL controlling the trait of interest. Efforts have also been made towards cloning gene/QTLs or identification of potential candidate genes responsible for a trait. Further new concepts like crop QTLome and QTL prioritization have accelerated precise application of QTLs for genetic improvement of complex traits. In the past years, efforts have also been made in exploitation of a number of QTL for improving grain yield or other agronomic traits in various crops through markers assisted selection leading to cultivation of these improved varieties at farmers’ field. In present article, we reviewed QTLs from their identification to exploitation in plant breeding programs and also reviewed that how improved cultivars developed through introgression of QTLs have improved the yield productivity in many crops.


Introduction
Mendel's laws of inheritance have been used to identify genes for a trait of interest by studying trait phenotypes in segregating populations. Subsequently, linkage studies between genes have helped to map the genes encoding a phenotype on chromosomes or in linkage groups. This genetic approach is utilized to map many qualitatively inherited traits leading to development of genetic linkage maps, especially for the genes controlling disease resistance in several crops (Kessel and Rowe 1974;Magato and Yoshimura 1998). Genetic improvement of crop plants does not rely only upon the manipulation of qualitative traits, as many agronomically important traits especially yield and yield contributing traits are inherited quantitatively. Many genes control these complex traits and each gene has small and cumulative effect on target trait. Hence, decoding the genetic architecture of agronomically important traits was a challenge before the 1980s. Recent advances in gel-based molecular technology and next generation sequencing have made it possible to map genomic regions controlling quantitative traits, i.e., QTL (Martienssen et al. 2005;Johannes et al. 2008). Moreover, genetical genomics has helped to map phenotypes for thousands of traits leading to reconstructing gene networks at transcript level that used to explain the relationship among traits (Jansen and Nap 2001;Li et al. 2008;Chaibub Neto et al. 2008). Now, genetic linkage analysis of quantitative trait loci (QTLs) has become a common technique (Kearsey and Farquhar Abstract Advancement in the field of genetics and genomics after the discovery of Mendel's laws of inheritance has led to map the genes controlling qualitative and quantitative traits in crop plant species. Mapping of genomic regions controlling the variation of quantitatively inherited traits has become routine after the advent of different types of molecular markers. Recently, the next generation sequencing methods have accelerated the research on QTL analysis. These efforts have led to the identification of more closely linked molecular markers with gene/ QTLs and also identified markers even within gene/QTL controlling the trait of interest. Efforts have also been made towards cloning gene/QTLs or identification of potential candidate genes responsible for a trait. Further new concepts like crop QTLome and QTL prioritization have accelerated precise application of QTLs for genetic improvement of complex traits. In the past years, efforts have also been made in exploitation of a number of QTL for improving grain yield or other agronomic traits in various crops through markers assisted selection leading to cultivation of these improved varieties at farmers' field. In present article, we reviewed QTLs from their identification to exploitation in plant breeding programs and also reviewed that how improved cultivars developed through introgression of QTLs have improved the yield productivity in many crops.  Holland 2007) to decipher the networks of QTL underlying a complex trait (Doerge 2002;Jansen 2003). It has resulted in identification of a number of QTLs for many traits in crop plants and so far more than 2500 studies have been published on QTL analysis including phenotypic QTL (phQTL) and expression QTL (eQTL), protein QTL and metabolic QTL (Gilad et al. 2008;Jansen et al. 2009;Ben-Ari and Lavi 2012). There are reports on using QTLs for genetic improvement in crop plants and several reviews have been published on marker assisted use of QTL (Xu and Crouch 2008;Gupta et al. 2010;Kumar et al. 2011a;Hanson et al. 2016;Shamsudin et al. 2016). The focus of this review is on the developments made in QTL analysis for agronomically important traits during the past two and a half decades and current status of their exploitation in genetic improvement for yield and yield contributing traits in crop plants.

Qualitative vs. quantitative traits
Plant phenotype is a sum of expression of a group of characters each of which has variable expression from one population to another population. Traits are variable features of a character. For example, plant height is a character that can express as tall or dwarf trait in a population. Such traits can be inherited qualitatively or quantitatively in nature. Qualitative traits are the classical Mendelian traits, which occur in discrete or distinct phenotypic classes and exhibit discontinuous variation in a population. Inheritance studies have shown that variation for each qualitative trait in a population is under the genetic control of two or more alleles of a single gene with high heritability as environmental conditions have little or no effects to obscure the gene effects (Mather 1941). Quantitative traits, which are also known as metric traits, are generally economically important traits. These traits show continuous variation in a population and individuals cannot be categorized into distinct classes. Nilsson-Ehle (1909) and East (1916) were the first to describe quantitative inheritance. They observed that several genes (perhaps 10 or more genes) are involved in the inheritance of a quantitative trait and each gene contributes small and cumulative effect to the total phenotype. These traits are significantly influenced by environmental factors and thus have lower heritability. Inheritance of quantitative traits is also called polygenic inheritance and can only be deciphered by statistical methods (Mather 1941).

Sources of genetic variation
Genetic variation for a trait in a population can either be existed naturally or induced artificially through mutation. Therefore, both natural and induced variations are important for studying the genetics of a trait and for making genetic improvement in crop plants. In natural population, genetic variation is generated through spontaneous mutation, recombination and migration of genes. However, mutation is the ultimate source of all variation in natural population. Recombination releases new variability by recombining already exiting variation and migration changes gene frequency of a phenotype in natural population through migration of genes from one population to another.
Induced mutation is an important discovery of genetics for crop improvement (Muller 1930). In conventional breeding programs, it uses to generate rare variability, and therefore, it is used preferably over the other advanced mutagenic tools such as transposon elements (Bingham et al. 1981). Mutation is the main source of genetic variation in qualitative traits. It creates a series of alleles at a locus and each allele can have large phenotypic effect. Therefore, induced mutation uses to generate mutant populations having variation for a number of qualitative traits. These mutant populations are used as donors for desirable trait(s) and as genetic resources for knowing gene function. Point mutations in single nucleotide all over the genomes are also responsible for creating variation in quantitative traits. Therefore, mutant populations have been generated in several model plants and different crop species (Till et al. 2004;Krishnan et al. 2009;Tadege et al. 2009;Xin et al. 2009Xin et al. , 2015Thompson et al. 2013;Henry et al. 2014). These mutant populations are now important genomic resources for knowing the function of genes using TILL-ING approach in several crop species including sorghum (Xin et al. 2008;Jiao et al. 2016), maize (Till et al. 2004), chickpea (Muehlbauer and Rajesh 2008), wheat (Slade et al. 2005), rice (Till et al. 2003(Till et al. , 2007, soybean (Cooper et al. 2008), pea (Dalmais et al. 2008;Le Signor et al. 2009) and Medicago trancatula (Le Signor et al. 2009a).

Quantitative trait loci (QTLs)
Quantitative trait loci (QTLs) are phenotypically defined chromosomal regions that contribute to allelic variation for a biological trait. In literature, quantitative trait locus has been mentioned as QTL, while quantitative trait loci have been described as QTLs. Most of the agronomically important traits follow quantitative inheritance and are the breeding target for genetic improvement. Polygenic control with small additive or dominance effects of each individual gene on the expression of a quantitative trait suggests its complex inheritance. Consequently, the genetics of complex traits was not known for a long time and biometrical assumptions that are known as 'statistical fog' were used to estimate sum number of genes controlling a quantitative trait (Mauricio 2001). Thus, knowledge of individual genes themselves was unknown. However, studies carried out during the first half of the twentieth century helped to resolve complexity of quantitative traits. These studies suggested that genes with a major effect on quantitative trait do exist and can be experimentally mapped on chromosomes by establishing correlation between quantitative trait value and allelic states at linked genetic markers (Sax 1923;Thoday 1961). Sax (1923) studied for the first time the inheritance of a quantitative trait (seed weight) with the help of a major gene locus controlling a qualitative trait (seed color) in Phaseolus vulgaris. Segregation of seed color in 1(PP):2(Pp):1(pp) ratio in F 2 population indicated qualitative inheritance of seed color with a single dominance gene P/p while seed weight showed a continuous distribution. In F 2 population of 166 plants, mean weight of the three seed color classes was measured. The mean of individuals with 'PP' genotypes had heavier seeds compared to the mean of individuals with 'pp' genotypes while average seed weight of heterozygotes 'Pp' was exactly intermediate. In this way, for the first time, a quantitative trait was linked to a major gene locus. Using the same approach in subsequent studies, polygenes were located on chromosome (Thoday 1961). This has led to identification of a quantitative trait locus (QTL) or genetic locus at which functionally different alleles segregate and cause significant effects on phenotypic expression of a quantitative trait (Salvi and Tuberosa 2005).

Type of QTLs
Quantitative trait loci have been described in many types in literature on the basis of their degree of expression effects on phenotype and contribution to phenotypic variation. These are discussed below.

Major vs. minor QTL
Marker-trait association analysis identifies genomic regions (i.e., QTLs) that control phenotypic variation of a quantitative trait available in a population. A number of QTLs contribute to the total phenotypic variation but each QTL explains a part of the total phenotypic variation. The amount of variation explained by a QTL is used to characterize it as major or minor QTL. A QTL that explains less than 10% of the total phenotypic variation has been categorized as minor QTL while a QTL with more than 10% of total phenotypic variation has been called as major QTL. This criterion has been used in several QTL studies conducted in rice, wheat and other crops (Hartman et al. 2013;Uga et al. 2013;Gao et al. 2015). However, in other studies, a criterion of 20% of phenotypic variation has been used for classifying major and minor QTLs (Davey et al. 2006). Therefore, threshold level for classification of major and minor QTL has been variable from case to case studies. During QTL analysis, minor and environment specific QTLs are ignored while the recognized "stable" QTLs are few. As a result, genetic architecture of complex traits is only partially revealed. Therefore, a concept of microreal QTL (MR-QTLs) has been reported in Brassica napus for recognizing and emphasizing the environment specific QTLs and protecting minor QTLs (Long et al. 2008). MR-QTLs refer to QTLs under the threshold (P ≤ 0.05) but above certain standard (P ≤ 0.5) in multi-environment trials. MR-QTLs accounts for 10-15% of the total examined QTLs in B. napus (Long et al. 2008).

Pleiotropic QTLs
QTL analysis sometimes locates more than one trait within the same genomic region. Clustering of QTLs for many traits within a small genomic region may be due to the presence of a single pleiotropic gene (Cai and Morishima 2002;Peng et al. 2003;Doebley 2004;Gyenis et al. 2007;Miller et al. 2014). Hence, identification of a QTL that affects more than one trait is called pleiotropic QTLs. However, it would be required to separate such QTLs from those QTLs, which affect more than one trait due to genetic linkage. Pleiotropic QTLs become advantageous in those cases where agronomically important traits are positively associated. As a result, it helps to improve more than one trait simultaneously. On the other hand, pleiotropic QTLs cannot be used in improvement if they control negatively correlated traits. In rice, STRONG CULM2 (SCM2) QTL for culm strength has shown to be identical to ABERRANT PANICLE ORGANIZATION1 (APO1) gene for panicle structure and located on the same genomic region. The pleiotropic effects of SCM2 QTL could be proved in a nearisogenic line carrying this QTL, which showed enhanced culm strength and increased spikelet number (Ookawa et al. 2010).

Epistatic QTL
Epistasis involves non-allelic interactions between a pair of loci where an allele of one locus modifies the effect of another locus. It was recognized by Bateson (1909) and studied extensively for qualitative traits where many phenotypes were described on the basis of epistatic interaction due to occurrence of modified segregation ratios in segregating populations (Eshed and Zamir 1996;Carlborg and Haley 2004). It is known that quantitative traits are under the control of several genes/QTLs and phenotypic value of a quantitative trait is resulted by sum effects of individual QTLs. However, sometimes due to epistatic interaction between loci phenotype of a given genotype deviates from the sum of its component individual QTL effects. In several studies, it was observed that an individual QTL does not have its own effects but when it comes together with other QTLs, it contributes to the phenotype as a net effect. Such QTLs have been described as epistatic QTLs. Evaluation of interactions between individual candidate genes helps to know that complex traits are also regulated by epistasis. QTL mapping approaches had initially focused on identification of the main genetic (i.e., additive and dominance) effects QTL(s) and ignored epistatic QTLs underlying the complex traits. Hence, no epistasis was assumed in various genetic models used for QTL mapping which could lead to a biased estimation of QTL parameters. However, in recent years, focus has been directed on identification of the epistatic QTLs along with the main effect QTLs using different analytical tools (Xing et al. 2002;Bocianowski 2012a, b, c;Krajewski et al. 2012). Many researchers in their studies have considered the importance of epistatic effects for complex traits (Lark et al. 1995;Eshed and Zamir 1996;Cockerham and Zeng 1996;Yu et al. 1997;Conti et al. 2011;Jiang et al. 2011;Li et al. 2011a;Mao et al. 2011;Upadhyaya et al. 2011;Bocianowski 2012a, b, c).

Expression QTLs (eQTLs)
Quantitative expression of a gene can vary from one individual to another. This variation in the expression of a gene has been identified heritable and the genomic region that governs the expression of genes is known as expression quantitative trait locus, i.e., eQTL (Mangion et al. 2006;Wang et al. 2014b). Expression QTL helps to understand the genetic regulation of gene expression. In expression QTL analysis, relationship between genome and transcriptome is established by studying the quantification of transcript levels of genes underlying a complex trait (Hansen et al. 2008). When gene expression variation is corresponded with phenotypic variation for a trait among individuals of a mapping population, it helps to identify candidate genes for phenotypic QTLs (Holloway and Li 2010). Also, the use of array analysis in e-QTL analysis can help genetic mapping and phenotyping simultaneously (Rostoks et al. 2005;West et al. 2006;Luo et al. 2007).
Expression QTLs have characterized into cis-or trans-eQTLs on the basis of their presence in the genome (Hansen et al. 2008). Cis-eQTL is physically located near the gene itself because differential expression of a gene is occurred due to the polymorphism in a promoter region. Earlier, a number of QTLs for glucosinolate biosynthesis and activation Kliebenstein et al. 2001;Kroymann et al. 2003;Zhang et al. 2006), phosphate sensing (Svistoonoff et al. 2007) and flowering time (Johanson et al. 2000;Caicedo et al. 2004;Werner et al. 2005;Clark et al. 2006;Salvi et al. 2007) have been cloned through QTL mapping approach. Actually, these QTLs are cis-eQTL because polymorphism in a promoter sequence has a direct influence on expression of the gene, which leads to larger phenotypic effect on a trait. In contrast to this, trans-eQTLs are not physically located near to actual position of the gene and hence trans-eQTL causes variation in the expression of genes due to a polymorphism in a regulatory factor. These regulatory factors can be located elsewhere in the genome and many regulatory genes are involved to control a gene. Therefore, change in one regulatory factor can cause small effect on phenotypic expression of a gene. Many times, a regulatory factor controls multiple genes and hence trans-eQTL works as pleiotropic QTLs. It has been reported that large-effect mutations in pleiotropic genes are likely to be deleterious and there might be a constraint on the effect size of trans-eQTLs (Wagner 2000;Jeong et al. 2001;Fraser et al. 2002;Yu et al. 2004;West et al. 2007).
The expression value of a gene in e-QTL analysis is used as a quantitative trait and hence it is denoted as e-trait ). QTL analysis for e-traits could be a useful strategy to decipher regulatory relationships between genes (Gilad et al. 2008;Kliebenstein 2009) and assist in identifying the existing potential regulator in a genomic region for a gene (Hansen et al. 2008;Kliebenstein 2009). Further, refinement in the resolution of e-QTL analysis has been made by identifying regulator candidates underlying e-QTLs by combining eQTLwith co-expression analysis (Terpstra et al. 2010;Flassig et al. 2013). Using this strategy, the regulatory network for many e-traits has been identified in rice. This has resulted in identification of 37.2% cis-eQTLs and 62.8% trans-eQTLs, which regulate the expression variation of many genes ).

Protein or agronomic or phenotypic QTLs (pQTLs) and metabolic QTLs (mQTLs)
Recent scientific advances have led to the development of new technologies for analysis of protein profile of individuals. These protein profiles have been used as phenotypic trait in linkage or association mapping based QTL analysis. QTL that controls variation in protein profiles is known as protein QTL (pQTL) (Jansen et al. 2009). Designation pQTL has also been used for agronomic QTL  or phenotypic QTL (Song et al. 2007;Sutton et al. 2007;Wang et al. 2014c). The pQTL analysis has an important role to understand the complexity of quantitative traits as they play an intermediate role between the eQTL and mQTL. Several studies have been conducted to make inter-level inferences between eQTL and mQTL as well as among eQTL, pQTL, mQTL and phenotypic traits in Arabidopsis (Wentzell et al. 2007;Fu et al. 2009a) and relationship of mQTL with phenotypic traits in tomato (Schauer et al. 2008). QTL analysis for protein expression (pQTL) did not receive the same attention, although proteins are the major determinants for grain quality in crop plants. As protein quality is an important trait in barley, a few studies have been conducted on identification of pQTL in this plant species (Witzel et al. 2011;Robison et al. 2015). Combining pQTL analysis with e-QTL and mQTL has resulted in better understanding of the regulatory network of genes (Wang et al. 2014c) and helped to dissect the genetics underlying the complex quantitative traits of agronomic importance. Metabolites were used as intermediary traits for cloning agronomic or phenotypic QTLs and biomarkers for molecular breeding . It has been shown that the level of a particular metabolite at a particular stage of a crop plant is directly associated with phenotypic trait. For example, a QTL for ear leaf length in maize was exactly co-localized with the QTL, which was identified for controlling the level of fructose in the leaf at reproductive stage and the level of xylulose in leaf at seedling stage . This co-localization of mQTL with phenotypic QTL indicates that both these traits are genetically co-regulated . In Brassica, colocalization analysis of pQTLs and eQTLs resulted in identification of four cis-regulated genes (i.e., BrFLC2_ A02 BrKRP2_A03 BrER_A09 and BrLNG1_A10) and several trans-eQTLs. Co-localization of above cis-regulated genes with colocalized phenotypic QTLs (copQTLs) suggested pleiotropic regulation of leaf development in Brassica rapa (Xiao et al. 2014).

Molecular markers development
Use of morphological traits as a genetic marker has paved the way for mapping genetic loci controlling both qualitative and quantitative traits. However, a population derived from a cross of two contrasting parents segregates only for a limited number of loci. As a result, a large number of segregating populations are required for developing genetic linkage map in crop plants. To overcome this problem, efforts had been made to develop tester lines through wide hybridization for getting many markers and that is applied successfully only in a few extensively studied species (Tanksley et al. 1982). Furthermore, genetic stocks such as aneuploids, chromosome substitution lines and A/B translocation stocks have also been used in wheat, tomato and maize to map genes on a specific chromosome (Roman and Ullstrup 1951;Sears 2008;Rick 1975). These efforts have led to the construction of detailed genetic linkage maps only in a few crop species (King 1975;O'Brien 1984;Vlaming et al. 1984;Weeden 1985) and could not be made available in other plant species due to lack of genetic markers and difficulties in developing tester lines. However, in 1980, restriction fragment length polymorphisms (RFLPs) reported as DNA based genetic markers (Botstein et al. 1980). Simple Mendelian inheritance and co-dominant nature of these markers provided an opportunity to develop detailed genetic maps from a limited number of crosses and used extensively in maize and tomato for saturating already existing genetic maps (Helentjaris et al. 1986;Bernatzkyand;Tanksley 1986). These markers were further used to develop linkage map for mapping quantitative traits (Lander and Botstein 1989). RFLP markers were very popular before the invention of PCR based markers. PCR based marker technology developed a number of DNA markers such as random amplified polymorphic DNA, i.e., RAPD (Williams et al. 1990), simple sequence repeat, i.e., SSR (Jacob et al. 1991) and amplified fragment length polymorphism, i.e., AFLP (Vos et al. 1995). These mediumthroughput markers replaced the use of RFLP markers in crop plants and were categorized as the second generation markers. However, among these markers only SSRs have emerged as "marker of choice" because of reproducibility, high polymorphism and codominance nature. These markers have been used in crop plants for mapping and tagging of genes/QTL (Powell et al. 1996). On the other hand, RAPD and AFPL markers could not be used widely in molecular breeding program as they have poor reproducibility and lengthy and laborious detection method. However, in the recent past, high-throughput sequence-based SNP markers have shown their dominance over the SSR markers because of the fast and cost effective availability of genome sequence information .
Developments in next-generation sequencing (NGS) technologies have revolutionized the genome sequencing of crop plants since 2005. Solexa is the first next generation sequencing technology and after that several sequencing methods have been developed. These methods can largely be grouped into three main types including sequencing by synthesis (Roche 454, pyrosequencing, Illumina, Ion Torrent), sequencing by ligation (SOLiD, Polonator) and single-molecule sequencing (Helicos, Pacific BioSciences, Emerging technologies, Life Technologies, Nanopore sequencing) (Egan et al. 2012). These second and thirdgeneration sequencing methods or NGS platforms can sequence an entire genome in less than 1 day. As a result, today NGS technology has become a powerful tool for rapid development of large numbers of DNA markers. In all NSG methods, one or more restriction enzymes are used to digest multiple samples of genomic DNA. After that a final set of fragments having less than 1 kb in size are selected for NGS owing to the read-length limits of current NGS platforms (Davey et al. 2011). Polymorphisms in the resulting sequenced fragments can be used as genetic markers. These methods include reduced-representation libraries (RRLs; Gore et al. 2009;Hyten et al. 2010) or CRoPS (complexity reduction of polymorphic sequences; Mammadov et al. 2010); RAD-seq (restriction-site associated DNA sequencing; Miller et al. 2007;Pfender et al. 2011), low coverage genotyping (multiplexed shotgun genotyping; Andolfatto et al. 2011), genotyping by sequencing (GBS; Elshire et al. 2011) and sequence based polymorphic marker technology (SBP; Sahu et al. 2012). In RAD sequencing, dominant DNA markers are generated from the variation within restriction sites while those maerkers developed from sequence variation adjacent to the restriction sites are to be codominant in nature (Pfender et al. 2011). These markers have been used for generating SNP markers and in development of linkage maps in crop plants (Pfender et al. 2011;Chutimanitsakun et al. 2011;Barchi et al. 2011;Wang et al. 2015;Marubodee et al. 2015;Zhang et al. 2015a). Similarly, GBS marker technology is also being utilized in crop plants for the development of SNP markers and their use in mapping of genes/QTLs for agronomically important traits in crop plants (Poland and Rife 2012;Peterson et al. 2014;Ariani et al. 2016).

Understanding the genetic basis of quantitative traits through QTL mapping
Quantitative traits have complex inheritance as a network of genes is involved for controlling these traits. In the past years, majority of genes controlling domsticated traits in staple food crops have been identified using classical Mendelian genetic approach. For example, in rice, initially a single recessive gene sd1 has been identified for semidwarfism; but now as many as 61 genes (d1 to d61) are known to dwarfism in rice (Singh et al. 1979;Cho et al. 1994;Ashikari and Matsuoka 2002;Neeraja et al. 2009). Similarly, in wheat, a number of genes controlling reducing plant height (Rht1 to Rht 13) have been identified (Flingtham and Gale 1983;Borojevic and Borojevic 2005; Bachir and Yin-Gang 2014). However, subsequently classical linkage mapping or association mapping based on molecular markers helped to locate these genes/QTLs on chromosomes and to identify the functional role of these genes (Pearce et al. 2011). Moreover, molecular marker applications have efficiently addressed the issues related to genetic linkages and genotype × environment interactions for understanding the genetic basis of complex traits that could not resolve through a classical Mendelian approach (Comstock 1978;Vikram et al. 2015). Thus, gel based molecular markers or recent NGS technology has significantly helped to resolve the genetic architecture of complex traits and developed the diagnostic markers for breeding programs. Moreover, present genotyping methods produce SNPs that provide higher density genome coverage compared to gel based molecular markers. This information can be utilized in QTX mapping which involves use of four -omics data bases including quantitative trait SNPs (QTLs), quantitative trait transcripts (QTT) (Mackay et al. 2009), quantitative trait proteins (QTP) and quantitative trait metabolites (QTM) in analysis of QTLs for complex traits (Zhou et al. 2014a). In QTX mapping, G × G (epistasis) and G × E interactions can also be detected and these interaction effects may explain a considerable proportion of the missing heritability associated with QTL based on individual molecular marker loci (Zuk et al. 2012). As a result, new knowledge on QTLs has significantly contributed to understand the genetic basis of quantitative traits (Vikram et al. 2015). For example, in rice fine molecular mapping identified possible linkages of the drought grain yield QTLs with co-locating flowering and/ or plant height QTLs and its interaction effects with flowering QTLs (Vikram et al. 2016). Similarly, QTL mapping helped to understand adult plant resistance to rust and powdery mildew diseases in wheat Asad et al. 2014;Zhou et al. 2014b;Bajgain et al. 2015).
During the past years, different approaches have been used to map QTL for complex traits. These QTL mapping approaches involve family based mapping populations such as biparental and multiparental populations or natural population based mapping (Xu et al. 2016).

Identification of QTL using family based biparental mapping population
In biparental mapping, genes/QTLs for a target trait are mapped in a population developed by crossing between the parents carrying contrasting traits of interest (i.e., wild × mutant or tall × dwarf or transgenic × non-transgenic). The diverse parents are selected from either natural or mutant populations. This conventional linkage mapping usually results in the identification of flanking markers harbouring the gene(s)/QTL within 2-20 cM. Primary analysis usually maps a QTL within a chromosome region (known as QTL supporting interval) of 10-30 cM. However, estimates of the total gene number present within this interval have been studied in Arabidopsis and maize genome and it has been calculated that a 10-cM chromosome interval has on average 440 genes in Arabidopsis and 310 genes in maize (Salvi and Tuberosa 2005). Thus, many genes are underlying behind the identified QTL for a target trait. Therefore, further fine mapping of targeted genomic regions is required to identify more closely linked markers within the proximity of 1 cM distance to know functional QTLs/genes controlling a quantitative trait. These closely linked flanking markers are then used to identify the candidate genes. Thus, a genetic approach initiated with specific phenotypes is eventually led to identify the genes encoding the target trait. In this way, a number of QTLs have been mapped for agronomically important traits in several crops Mohan et al. 2009;Xu et al. 2016) and many of them have been characterized functionally. Next generation sequencing has revolutionized in-depth genetic dissection of major QTL(s)/gene(s) controlling important agronomic traits. QTL-seq is one of the NGS based strategies that provides a rapid high-resolution genome-wide mapping. It was used first time in chickpea and to identify a major genomic region harboring a robust 100-seed weight QTL (CaqSW1.1) in a biparental intra-specific mapping population (ICC 7184 × ICC 15061). This QTL was further validated by SNP and SSR marker-based QTL mapping (47.6% R 2 at higher LOD > 19), which reflects the reliability and efficacy of QTL-seq strategy. Use of this strategy in combination with classical QTL mapping has also narrowed down genomic region from 1.37 Mb (comprising 177 genes) to 35 kb genomic interval on desi chickpea chromosome 1 containing six genes (Das et al. 2015). More recently two novel anthracnose resistance loci have been mapped in Sorghum bicolor on the basis of SNP markers generated by GBS method (Felderhoff et al. 2016).

QTL analysis using family based multi-parent populations or diverse natural populations
Biparental mapping populations are being widely utilized in marker-trait association analysis. As a result, it could detect thousands of markers associated with QTL for a variety of quantitative traits in a number of crops. However, these markers could only partly be used in marker assisted breeding programs because parents used in crossing program might not be differed for targeted QTL (Bernardo 2008). The lack of tight association of marker with QTL and inefficient costly process of fine mapping have also discouraged use of biparental populations in identification of QTL for traits (Parisseaux and Bernardo 2004). Alternatively, LD-based association mapping (AM) approach has been used to overcome the above problems of biparental mapping population. In association mapping, a set of diverse genotypes derived from known/unknown ancestry is used to constitute an association panel for mapping of the genes/QTLs for target trait. Association panel can be comprised of natural germplasm lines, breeding lines developed in breeding programs, individuals of multiple biparental populations and/or populations developed by crossing of multiple parents Wurschum 2012;Gupta et al. 2014). As phenotypic data on breeding lines are routinely recorded, their use can make association mapping more cost-effective. Therefore, use of diverse breeding lines as association panel is more useful compared to other kind of diverse panels (Gupta et al. 2014). Moreover breeding populations are generally narrow in genetic base. Therefore, identified QTLs in elite breeding material can be used directly in marker assisted selection. In contrast, those QTLs identified from wide (elite × exotic) crosses are useful only for the introgression of traits from exotic material (Jannink et al. 2001). After several recombination cycles, association panel covers most of the genetic variability available in a gene pool for the trait of interest (Mackay et al. 2009). Therefore, in association mapping, multiple alleles at a locus are involved for QTL analysis of a trait in contrast to two alleles in biparental mapping population. As a result, frequency of a particular allele in population provides an opportunity for identification of tightly linked makers for trait of interest. However, in association mapping, marker-trait association is often detected spurious because higher frequency of a particular allele may be due to population stratification and relatedness among individuals rather than due to tight linkage of markers with trait of interest (Gupta et al. 2014). Now statistical methods and software are available, which can analyze combined data derived from biparental mapping populations and association panels and thus help to derive the benefit of both biparental mapping and association mapping for genetic dissection of complex traits. Statistical methods are also available for the analysis of multiparental populations (Xu 1998;Rebai and Goffinet 1993;Jannink and Wu 2003). As a result, family based mapping populations have been suggested to overcome the problem of spurious marker-trait associations (Stich et al. 2006;Rosyara et al. 2009). Different mating designs including diallel, partial diallel, random mating, three way and four way cross have been used to develop family based mapping populations for QTL analysis in several crop plants including maize, wheat and rice (Rosyara et al. 2009;Reif et al. 2010;Liu et al. 2011;Bandillo et al. 2013;Gupta et al. 2014). In recent years, multiparental advanced generation intercrosses (MAGIC; Cavanagh et al. 2008) and nested association-mapping (NAM; Yu et al. 2008) strategies are becoming popular for LD-based association mapping. In MAGIC population, 4, 8 or 16 parents are involved to cross in biparental fashion and followed by crossing between F 1 hybrids in subsequent generations. These populations have been developed and used in QTL mapping for many traits (i.e., plant height, hectoliter weight and presence or absence of the awn) or for varietal development in several crops including wheat, rice and tomato Mackay et al. 2014;Bandillo et al. 2013). Recently, a MAGIC population comprising of 397 MAGIC lines was developed in tomato and it was suggested to be a useful permanent genetic resources for identification of candidate genes (Pascual et al. 2015). This mapping population has been used to identify SNP markers through re-sequencing related to QTL for quality traits (Pascual et al. 2016). The MAGIC population may also be developed through intermating of parental inbreds (PIs) for four generations ). In rice, MAGIC populations were developed for indica and japonica ecotypes to capture the broadest genotypic diversity. Further, MAGIC plus population has been developed by inter-crossing for two more cycles within the recombinant lines of an indica MAGIC population for enhancing recombination. Bandillo et al. (2013) developed global MAGIC populations by inter-crossing between the recombinants of the indica and japonica base populations to increase the overall diversity. More recently, three new MAGIC populations were used for QTL analysis of plant height (PH) and heading date. This resulted in identification of three QTL for PH and two QTL for heading date and among these, four QTL were very close to known genes. Also, a novel QTL for PH was identified in this association panel (Meng et al. 2016). The efforts made in the development of MAGIC populations, which are under progress and are not published so far have reviewed recently .
In NAM population, one common parental inbred line is crossed with a number of founder parental inbreds. The first NAM population was developed in maize consisting of 25 families each with 200 RILs and released as a genetic resource for identification of functional markers (FMs) . Similarly, another NAM population comprising of 5000 members was also developed in maize to identify QTLs for important traits such as flowering time disease resistance plant architecture and 12 metabolites Buckler et al. 2009;Kump et al. 2011;Poland et al. 2011;Tian et al. 2011;Peiffer et al. 2013;Zhang et al. 2015b). More recently, Nice et al. (2016) suggested advanced backcross-nested association mapping (AB-NAM) population that was developed from wild × cultivated barley to maximize the allelic diversity in the association panel. Association analysis has resulted in mapping of three qualitative traits: glossy spike, glossy sheath and black hull color with high resolution to loci corresponding to known barley mutants for these traits. Besides, ten QTLs were identified for grain protein content. This NAM population has been identified as an important tool for high-resolution gene mapping and discovery of novel allelic variation using wild barley germplasm (Nice et al. 2016). In another study, in barley, natural variation in exotic germplasm was exploited by taking a genome-wide association approach using NAM population. This study dissected genetic architecture of flowering time under high salinity and identified putative genes affecting this trait and salinity tolerance. Moreover, a locus was identified under saline conditions and homozygous lines for wild allele at this locus yielded 30% more than lines homozygous for the Barke allele. Thus introgression of wild allele into elite cultivars could markedly improve yield under saline conditions (Saade et al. 2016).
NAM populations developed in several other plant species including Arabidopsis thaliana (Buckler and Gore 2007), barley (Schnaithmann et al. 2014), sorghum  and soybean (Guo et al. 2010) are now available publicly. The first barley NAM population HEB-25 developed by crossing 25 wild barleys with one elite barley cultivar was used for dissection of the genetic architecture of flowering time (Maurer et al. 2015). The important advantage, which the NAM population offers, is that one can conduct joint linkage association mapping (JLAM), which is considered superior to QTLIM and AM for QTL detection. However, it is not necessary to develop aNAM population in all crops. NAM populations using more than one common parent or fewer founder parents have also been developed in Arabidopsis (Bentsink et al. 2010;Brachi et al. 2010).
Although LD-based association mapping provides high resolution between marker and trait, there are several issues including population structure, cryptic familial relatedness, multi trait and multilocus analysis, marker with rare alleles and rare variants and missing heritability that are required to consider in LD-based association mapping (Gupta et al. 2014). Diversity panels used in association mapping often have substantial sub-population genetic structure because they are mixture of geographically distinct lines with varying levels of pedigree relationships (Myles et al. 2009). As a result, subgroups within the diversity panel can differ for mean trait values and also for allele frequencies at many loci. This population substructure can lead to identification of false-positive marker trait associations. Although advancement in statistical methods helps to remove the confounding effects of population structure on association tests but it reduces the power of marker-trait association (Holland 2015). LD-based association analysis also missed the heritability because it could not detect association of a marker with rare variation for the trait present within population. On the other hand, a rare marker allele present in a population is difficult to associate with a trait of interest due to low frequency of this allele in population (Gupta et al. 2014). Genome-wide association scans/studies (GWAS) and candidate gene (CG) approaches are being utilized in crop plants for genetic dissection of complex traits. These approaches have been briefly outlined in Fig. 1 and discussed below in the following text.

Genome wide associations scan (GWAS)
This approach is often used to scan all those genomic regions that may control the trait of interest. In this case, however, we also often obtain statistically significant results with the loci, which are not actually associated with targeted trait. As a result, it leads to the identification of spurious marker-trait association. However, in spite of this, this approach has been used widely in several plant species in marker-trait analysis using improved statistical methods, which minimize the risk of spurious marker-trait association (Table 2). This is a phenotypic driven approach where we used phenotypic data and molecular marker data for reaching the gene controlling a target trait. In barley, a tightly linked marker to the gene controlling β-glucan content was identified using GWAS (Mezaka et al. 2011). Genome wide association mapping in tomato has not only validated previously identified several candidate genes and quantitative traits but also identified other new associations for important traits among the accessions of Solanum lycopersicum var. cerasiforme (Ranc et al. 2012). In rice, elite genes within the landraces were mined for 12 agronomic traits through GWAS and verified the previous results of association mapping .

Candidate gene (CG) association mapping
The CG approach is used when genes controlling a trait under study are known in related or model crop species. This information on genetics of a trait of interest is used to study in the target crops for identification of candidate genes. It has been used first in maize for flowering time (Remington et al. 2001;Thornsberry et al. 2001) and subsequently in many other crops for several agronomically important traits Li et al. 2011b; Table 3). It has been used separately and also in parallel with genome wide approach. In sorghum, 73 SNPs from 26 brassinosteroid candidate genes belonging to BR (brassinosteroid) pathway were found associated with the desired   Winter survival

1-FEH-A 1-FEH-B
Drought tolerance response to abscisic acid and fructan 1-exohydrolase Edae et al. (2013) traits and also many SNPs showed their association with more than one traits (Mantilla Perez et al. 2014). Similarly, in potato, SNPs in eight genes controlling key functions in starch-sugar inter-conversion showed association with the natural variation of tuber starch and sugar content. An allele of the plastidic starch phosphorylase 'PHO1a' responsible for increasing tuber starch content was also cloned in this study (Schreiber et al. 2014).
The parallel use of CG approach with GWAS is found more useful in those cases when no marker showed association with a trait of interest after applying the false discovery rate (FDR) correction in genome wide association analysis. For example, in maize, none of the 51741 SNPs showed any significant association with the trait after a multiple test for FDR (Cook et al. 2012). However, on the other hand, CG association mapping could detect a significant association of mutant gene with oil content. Earlier this mutant gene has been identified for increasing the oil content due to insertion of Phe in a DGAT1-2 gene (Zheng et al. 2008). Moreover, two additional SNPs located within the DGAT1-2 gene have been identified to be associated significantly with oil content. Use of this approach in parallel with GWAS has also increased the power and precision of QTL detection in maize (Lipka et al. 2012). In chickpea, candidate gene association mapping was taken along with genome wide association mapping. This resulted in identification of SNPs in 5 drought responsive genes that were found to be associated with several drought and heat traits (Thudi et al. 2014). In carrot, CG approach has been used in an unstructured population for reducing the chance of detecting the false positive association. This study used an original unstructured population with a broad genetic base, which was developed through intercrossing of 67 diverse cultivars for identification of the candidate genes responsible for accumulation of carotenoids (important quality trait in carrot). The genotyping of 384 individuals with 109 SNPs located in 17 candidate genes belonging to carotenoid biosynthesis has identified genes zea xanthin epoxidase (ZEP), phytoene desaturase (PDS) and carotenoid isomerase (CRTISO), which were significantly associated with total carotenoids and β-carotene contents. Moreover, association of CRTISO and plastid terminal oxidase (PTOX) genes with α-carotene and ZEP gene with color components was also observed in this study (Jourdan et al. 2015).

Functional characterization of agronomically important genes/QTL
Mapping of QTLs using different approaches conducted for the more than two decades identified a large number of genes/QTLs for yield, disease tolerance and seed quality traits in crop plants. Further efforts have also been made to identify functional role of these QTLs/genes governing the targeted traits. Initially, map-based cloning of genes has been used for this purpose but progress remained slow because this approach is relatively time consuming. Therefore, in the recent past, several other methods have been developed to accelerate the identification of function of gene/QTL underlying the trait of interest. These methods followed both forward and reverse genetic approaches including insertional mutagenesis (loss of gene function cause change in phenotype) using TDNA and transposable elements (Meissner et al. 2000;Hayes 2003;Twyman and Kohli 2003;Acosta-García et al. 2004;Tierney and Lamour 2005;Lazarow and Lütticke 2009;Chudalayandi 2011), homologous recombination mediated gene transfer (Hanin and Paszkowski 2003;Iida and Terada 2004;Tierney and Lamour 2005;Fu et al. 2012), gene silencing via RNAi (Waterhouse et al. 1998;Baulcombe 2004;Kusaba 2004;Vaucheret 2006;Eamens et al. 2008;Hebert et al. 2008;Ghildiyal and Zamore 2009;Fu et al. 2011;Mann et al. 2011;Duan et al. 2012;Ré et al. 2012;Rubinelli et al. 2013), virus induced gene silencing (Deng et al. 2012;Kachroo and Ghabrial 2012;Hosseini et al. 2012), TILLING (Ramos et al. 2009;Chen et al. 2012) and candidate gene association mapping (Wilson et al. 2004;Weber et al. 2007Weber et al. , 2008Kloosterman et al. 2010). These approaches could establish the function of genes/QTLs, which was not known previously. In the past years, major QTLs have been cloned in maize for flowering time (Salvi et al. 2007), carotenoid content (Huanget al. 2010) and aluminium tolerance (Famoso et al. 2011), rice for yield components (Xing and Zhang 2010), submergence tolerance (Septiningsih et al. 2009), aluminium tolerance (Famoso et al. 2011), phosphorus starvation and root architecture (Gamuyao et al. 2012;Steele et al. 2013), sorghum for aluminium tolerance (Famoso et al. 2011) and wheat for boron toxicity (Pallotta et al. 2014) (Table 4). Jadhav (2015) reviewed the cloning of important functionally characterized genes / QTLs of agronomic importance in crop plants. More recently, next generation sequencing facilities have further accelerated the identification of functional genes associated with agronomically important traits. For example, in wheat, RNA-seq analysis of near isogenic lines differing for homozygous alleles at a QTL having low and high value of a trait identified PM19-A1 and A2 as candidate genes for a major dormancy QTL (Barrero et al. 2015). Novel chitinase genes associated with the sheath blight resistance QTL (qSBR11-1) have been functionally characterized in rice line Tetep (Richa et al. 2016). Similarly, PEBP family genes have been characterized functionally in upland cotton (Zhang et al. 2016).

Prioritization of candidate genes within QTL regions
QTL mapping establishes genotype-to-phenotype relationships in which variation at the trait level is linked to the variation at the genomic level. However, a QTL region identified for a trait typically contains tens to hundreds of genes and among them only a few candidate genes actually control that trait (Bargsten et al. 2014). Methods have been developed to prioritize the candidate genes in the QTL regions of the target trait on the basis of their over representation in the biological processes (gene functions). Prioritization method has been applied to rice QTL data (Bargsten et al. 2014). In this case, gene functions predicted on the basis of sequence-and expression-information. This led to on an average tenfold reduction in the number of genes. A detailed analysis of flowering time QTLs illustrates that genes with completely unknown function are likely to play a role in this important trait (Bargsten et al. 2014). In chickpea, candidate genes in "QTL-hotspot" region for drought tolerance have been prioritized (Kale et al. 2015).

The crop QTLome
A more accurate and comprehensive characterization of QTLs based on recent progress made in genomics and phenomics is known as QTLome. In the QTLome, we have knowledge of the map position, allele identity and the additive vs. dominant genetic effect and their magnitude for each QTL is targeted in breeding programs. It can also be restricted to specific QTL alleles, which are accessible to breeders in a germplasm collection (Salvi and Tuberosa 2015). In crop QTLome, information on all experimentally studied QTLs for all traits in a crop is included while only trait specific QTL information in one species is used in a trait QTLome. In a crop's QTLome, generally the entire collection of genetically mapped loci and their allelic variation influencing any quantitative trait are used. As a result, we can have knowledge about the substitution effect of one haplotype with other in any given genomic region on a trait (Salvi and Tuberosa 2015). For example, in maize, many QTLs that control yield and related traits are available in public database, which provides as the QTLome database.
This maize QTLome database possesses 44 published studies including 32 independent mapping populations and 49 parental lines for assembling the yield QTLome for carrying out QTL meta-analysis. This resulted in identification of 84 meta-QTLs among 808 unique QTL on the 10 maize chromosomes and 74% of QTLs explained only <10% proportion of phenotypic variance. Consequently, this study helped to conclude that high genetic complexity is involved to control grain yield in maize. This QTLome study also suggested that gene density is the main driver for distribution of yield QTLs on chromosomes. It has also been observed that distribution of dominant and overdominant yield QTLs is not different from the additive effect QTLs (Martinez et al. 2016). In wheat, trait QTLome has been studied and dissected the QTLome governing root system architecture features in durum wheat. Among three major QTL clusters identified for root length and number, five QTLs for RGA have been prioritized that particularly appeared suitable for a possible deployment in markerassisted selection and positional cloning . QTLome information helps in marker assisted breeding for QTLs controlling the targeted traits because it also uses information of environmental conditions under which the QTL alleles were detected or tested (Salvi and Tuberosa 2015).

Productivity improvement using QTLs through marker assisted selection
In the past, a number of major-effect QTLs have been mapped validated and cloned as discussed above. In many cases, QTLs with major effects have also been successfully exploited through marker assisted selection for development of improved cultivars by introgression of QTLs for many traits including drought tolerance in chickpea (Varshney et al. 2013), stay green in sorghum (Borrell et al. 2014), cyst nematode resistance in soybean (Concibido et al. 2004) and Fusarium or salt tolerance, grain protein content and preharvest sprouting in wheat (Anderson et al. 2007;Kumar et al. 2010Kumar et al. , 2011bJames et al. 2011). These efforts have led to the development of improved lines. Multi-environment evaluation of these lines showed significant yield advantage over check varieties and released for  Rawat et al. (2016) cultivation. The most significant progress has been made in rice, wheat and common bean. In these crops, the impact of QTLs introgressed through marker assisted selection has been studied on productivity enhancement. In rice, a marker-assisted introgression breeding program improved drought tolerance in Indian upland variety Kalinga III by introgressing QTLs for root traits from Azucena, an upland japonica variety. In this marker assisted breeding program, four genomic segments carrying QTLs for root traits (root length and thickness) and one segment carrying a recessive QTL for aroma were introgressed. As a result, a set of pyramided lines with different numbers of target QTLs were selected and evaluated in eastern India over 6 years under field conditions (Steele et al. 2006). The lines carrying QTL 9 demonstrated an effect on increasing root length had a significant effect of 0.2 t/hm 2 on grain yield (Steele et al. 2007). Moreover, lines with two or more Azucena alleles at root QTLs showed a mean increase of 0.4 t/hm 2 and QTL 7 was associated with a mean increase of 0.9 t/hm 2 in combination with the other QTLs (Steele et al. 2007). The combination of all four QTLs in one of the lines has resulted in an increased grain yield of 1.0 t/hm 2 (Steele et al. 2007). This line was released for cultivation in Jharkhand state as Birsa Vikas Dhan 111. This is the first released drought-tolerant rice cultivar developed through MAS for improved roots (Steele et al. 2013). In IRRI's drought breeding program, four major drought QTLs (qDTY2.2, qDTY4.1, qDTY9.1 and qDTY10.1) were successfully pyramided in different lines having different combinations of these QTLs (Swamy and Kumar 2011). Those introgressed lines having three or two QTLs gave yield advantage of 1.2-2.0 t/hm 2 in drought conditions. However, these lines had yield and other quality traits similar to improved background variety IR64 under normal irrigated conditions (Swamy and Kumar 2011). In addition to this, QTLs have also been introgressed in the background of improved lines for a number of traits including primary and secondary branches (Ando et al. 2008), grain size and heading date , yield related traits (Zong et al. 2012) and drought tolerance (Li et al. 2005). Moreover, a major QTL for grain number (Gn1) has also been combined with a plant height QTL (Ph1) (Ashikari et al. 2005). The pyramided lines with these QTLs increased grain production (23%) and reduced plant height (20%) compared with Koshihikaiin rice (Qian et al. 2007). In rice, near isogenic lines carrying QTLs for secondary (SBN1) and primary (PBN6) branch number individually or in combination (SBN1 + PBN6) produced more spikelets per panicle compared to the recurrent parent (Ohsumi et al. 2011). However, more number of spikelet per panicle did not increase yield significantly and pyramided lines had only 4-12% higher yield. In another study, marker assisted introgression of desirable allele at qHD8 locus increased the yield per plant (>50%) and enlarged the leaf size (~30%) in improved lines . Phenotypic selection among the lines developed through pyramiding of eight QTLs for spikelet number per panicle and 1000-grain weight in the background of a single line using four RILs allowed identification of a few lines with increased panicle and spikelet size in rice (Zong et al. 2012). A gene (OsPPKL1) controlling the grain length in rice was used in marker assisted selection breeding program. It resulted in development of a NIL containing the desirable allele qgl3. The evaluation of this NIL under the field conditions showed an increase in yield (16.20%), grain length (19.68%), grain width (1.15%), grain thickness (8.25%), filled-grain weight (37.03%) and panicle length (11.76%) .
In rice, an improved line designed as TS4 was developed by introgression of semi-dwarf gene sd1, low amylase content gene Wx b and bacterial light resistance genes Xa4 using marker assisted selection. This line showed semidwarf phenotype with reduced growth duration and produced high yield along with good grain quality and broadspectrum resistance to Xoo strains. This line achieved higher grain yield in field trials conducted in Indonesia and China and was found better in terms of degree of chalkiness and amylose content along with grain quality compared to recurrent parent .
In sorghum, marker assisted selection was able to release improved varieties for cultivation in sub-Saharan Africa. These lines have been developed through introgression of Striga resistance QTLs and selected with 180-298% yield superiority over the checks along with Striga resistant after multi-location evaluation (Mohamed et al. 2014).

Concluding remarks
During the past years, significant advancement was made in genomics and phenomics of crop plant species. This led to the identification of quantitative trait loci controlling complex agronomic traits. Even next generation sequencing and other advanced genomic tools could help to identify potential candidate genes underlying the complex agronomic traits. These developments have helped to use the molecular marker technology to develop improved cultivars through marker assisted selection and major QTLs were exploited in breeding programs. As a result, a number of traits were improved by exploiting QTLs through marker assisted selection and developed lines had better yield potential and biotic and abiotic stresses tolerance. These lines were evaluated over multilocations for yields and other traits over the years and improved varieties were released for farmers' cultivations. However, impact of these introgressed QTLs on the productivity of a crop species could not be documented in a big way and it is still difficult to assess the impact of these studies on overall productivity in a crop species. Moreover, exploitation of QTLs in crop improvement is still a challenge and is not routinely being utilized in most of the breeding programs. Therefore, for actual potential of these QTLs, there is urgent need to identify tightly linked molecular markers and also need to identify the network of genes/QTL that contribute for yield and yield contributing traits. Moreover, complex traits are highly influenced by environmental conditions. Therefore, focus is required on identification of genes/QTLs that express in a particular environmental condition so that such QTLs could be exploited for developing improved cultivars for that particular environmental niche. Moreover, we also need to identify the interactive QTLs of which expression is governed by interaction between the genotypes and environments. Now, it has become possible to sequence a genome of crop species more cost effectively and quickly. This has led to availability of genome sequences of several crop species and in coming years more sequences would be available in more number of crop species. Therefore, in future, QTL analysis would be more precise and it would be possible to identify the actual network of genes governing the targeted traits. As a result, these would be utilized routinely in breeding programs and it will boost the productivity of crop plant species.
Author contribution statement JK concieved the idea and drafted the manuscript. DSG and SK edited thoroughly the manuscript. SG, SD and PG helped in collection of literature and preparation of bibliography.