1. Introduction
Wheat (Triticum aestivum L.) is a major food crop of the world and the second most important crop in India. It is a segmental allopolyploid containing three distinct but genetically related (homoeologous) genomes: A, B and D. It is a hexaploid containing 42 chromosomes. The haploid DNA content of bread wheat genome is approximately 1·7×1010 bp.
In bread wheat several genetic linkage maps have been published either in the form of separate homoeologous groups, such as groups 1 to 7 (Phillips & Vasil, Reference Phillips and Vasil2001), or as complete maps (Liu & Tsunewaki, Reference Liu and Tsunewaki1991; Gale et al., Reference Gale, Atkinson, Chinoy, Harcourt, Jiu, Li, Devos, Li and Xin1995; Messmer et al., Reference Messmer, Keller, Zanetti and Keller1999). Owing to the poor levels of polymorphism often encountered in wheat, mapping strategies most often used wide crosses involving either a synthetic wheat and a variety such as Chinese Spring (Gale et al., Reference Gale, Atkinson, Chinoy, Harcourt, Jiu, Li, Devos, Li and Xin1995) or Opata (Nelson et al., Reference Nelson, Van Deynze, Autrique, Sorrels, Lu, Merlino, Atkinson and Leroy1995a–Reference Nelson, Van Deynze, Sorrels, Lu, Atkinson, Bernard, Leroy, Faris and Andersonc) as parents, or crosses between Chinese Spring and Triticum spelta (Liu & Tsunewaki, Reference Liu and Tsunewaki1991; Messmer et al., Reference Messmer, Keller, Zanetti and Keller1999).
The development of genetic maps is a prerequisite for the understanding of QTLs governing complex agronomic traits and their use in plant breeding via marker-assisted selection. The first intervarietal map of bread wheat, based on restriction fragment length polymorphism (RFLP) markers, was published by Cadalen et al. (Reference Cadalen, Boeuf, Bernard and Bernard1997). An updated version of this Chinese Spring–Courtot genetic map was published by Sourdille et al. (Reference Sourdille, Cadalen, Guyomarc'h, Snape, Perretant, Charmet, Boeuf, Bernard and Bernard2003). More recently three intervarietal maps based on Australian bread wheat varieties were reported by Chalmers et al. (Reference Chalmers, Campbell, Kretschmer, Karakousis, Henschke, Pierens, Harker, Pallotta, Cornish, Shariflou, Rampling, McLauchlan, Daggard, Sharp, Holton, Sutherland, Appels and Langridge2001) and other intervarietal maps by Paillard et al. (Reference Paillard, Schnurbusch, Winzeler, Messmer, Sourdille, Abderhalden, Keller and Schachermayr2003), Liu et al. (Reference Liu, Anderson, Hu, Friesen, Rasmussen and Faris2005), Quarrie et al. (Reference Quarrie, Steed, Calestani, Semikhodskii and Lebreton2005), Suenaga et al. (Reference Suenaga, Khairallah, William and Koisington2005) and Torada et al. (Reference Torada, Koike, Mochida and Ogihara2006). Some of these maps have also been used for QTL analysis (Sourdille et al., Reference Sourdille, Cadalen, Guyomarc'h, Snape, Perretant, Charmet, Boeuf, Bernard and Bernard2003; Liu et al., Reference Liu, Anderson, Hu, Friesen, Rasmussen and Faris2005; Quarrie et al., Reference Quarrie, Steed, Calestani, Semikhodskii and Lebreton2005; Suenaga et al., Reference Suenaga, Khairallah, William and Koisington2005; Semagn et al., Reference Semagn, Bjornstad, Skinnes, Maroy, Tarkegne and William2006).
The major trait governing plant stature is plant height (culm length). Several major genes reducing plant stature have been identified in wheat. Introduction of genes Rht-B1b (Rht1) and Rht-D1b (Rht2) for reduced height (Peng et al., Reference Peng, Richards, Hartley, Murphy, Devos, Flintham, Beales, Fish, Worland, Pelica, Duralalagaraja, Christou, Snape, Gale and Harberd1999) into commercial wheat cultivars resulted in the Green Revolution. The other Rht genes in hexaploid wheat include Rht4, Rht5, Rht7 and Rht12.
Leaf size has a positive effect on biomass and yield of plants. In wheat, the flag leaf makes important contribution of photosynthates, particularly during grain filling. Several flag leaf morphogenetic parameters have been identified which contribute to the moisture stress tolerance of wheat. Flag leaf length and flag leaf breadth are components which contribute to photosynthesis. Searching for loci controlling these quantitative traits will be useful as they are of agronomic importance (Lupton, Reference Lupton1987).
The aim of this study was to obtain a genetic linkage map of wheat (Triticum aestivum L.) using a cross between the Indian bread wheat varieties Kalyansona and Sonalika and to obtain a QTL map for three metric traits: length and breadth of flag leaf blade, and culm length. For this purpose an F2 population was used. Most often a genetic linkage map is prepared using populations obtained from two highly diverse genotypes. However, the markers from such a map may not be useful in the breeding programme as they may not be polymorphic among the varieties used in breeding. Kalyansona and Sonalika have served as popular cultivars and have been used as parents in a breeding programme, and hence the markers thus obtained would be useful for a future Indian wheat-breeding programme involving parents related to the two varieties.
2. Materials and methods
The mapping population consisting of 150 F2 plants was derived from a cross between the varieties Sonalika and Kalyansona (bread wheat: Triticum aestivum L.). A set of nullitetrasomic lines derived from Chinese Spring were used. All plants were grown at Trombay under field conditions.
(i) DNA extraction and estimation
DNA was isolated and quantitated from leaf tissue by a new method of DNA isolation suitable for long-term storage (Nalini et al., Reference Nalini, Bhagwat and Jawali2004).
(ii) Phenotypic data collection
The data on three agronomic traits, viz. culm length (CL), flag leaf length (FLL) and flag leaf breadth (FLB), among 150 F2 individuals were recorded at different stages of growth.
(iii) PCR analysis
All PCR amplifications were carried out on an Eppendorf Mastercycler-Gradient Thermal Cycler.
(a) AP-PCR analysis
PCR amplification was carried out in a volume of 25 μl containing 100 ng of template DNA, 2 mM MgCl2, 25 pmol of primers, 200 μM each of dNTPs and 1 unit of Taq DNA polymerase. The cycling condition was as follows: 1 cycle of 5 min at 94°C, 5 min at 45°C and 5 min at 72°C, and 35 cycles of 1 min at 94°C, 1 min at 45°C and 1 min at 72°C, followed by a final 10 min extension at 72°C.
(b) RAPD analysis
PCR amplification was carried out in a volume of 25 μl containing 100 ng of template DNA, 2 mM MgCl2, 10 pmol of 10mer primers, 200 μM each of dNTPs and 1 unit of Taq DNA polymerase. The cycling condition was as follows: 1 cycle of 5 min at 94°C, 5 min at 42°C and 5 min at 72°C, and 45 cycles of 1 min at 94°C, 1 min at 42°C and 1 min at 72°C, followed by a final 10 min extension at 72°C.
(c) ISSR analysis
PCR amplification using a 3′ anchored I SSR primer was carried out in a volume of 25 μl. The reaction mixture contained 100 ng of template DNA, 2 mM MgCl2, 25 pmol of ISSR primer, 200 μM each of dNTPs and 1 unit of Taq DNA polymerase. The cycling condition was as follows: 1 cycle of 5 min at 94°C, 5 min at 50°C and 5 min at 72°C, 45 cycles of 1 min at 94°C, 1 min at 50°C and 1 min at 72°C, followed by a final 10 min incubation at 72°C.
The PCR products of AP-PCR, RAPD and ISSR were separated by electrophoresis using 1×TBE buffer on a 2% agarose gel. The DNA fragments were stained with ethidium bromide and viewed under ultraviolet light and photographed.
(d) AFLP analysis
The AFLP analysis was carried out essentially by the method described by Vos et al. (Reference Vos, Hogers, Bleeker, Reijans, Thoe, Hornes, Frijters, Pot, Peleman, Kuiper and Zabeau1995). Genomic DNA (100 ng) was digested with EcoRI and MseI. Ligations of the EcoRI and MseI adapter sequences, the preselective amplifications and the selective amplifications were carried out using the primer pairs EA+3/MC+2 as described previously. Equal amounts of the selective amplification products and formamide loading dye were mixed. The samples were denatured for 3 min at 90°C, chilled on ice and fragments were separated by electrophoresis on a denaturing 5% polyacrylamide gel in a vertical cassette. The DNA fragments were stained by the silver staining method.
(e) STMS analysis
PCR amplifications were carried out using 84 STMS primers for the A, B and D genomes of bread wheat, viz. two each on either arm of the seven A, B, D chromosomes (Roder et al., Reference Roder, Korzun, Wandehake, Planschke, Tixier, Leroy and Ganal1998). The PCR reaction mixture (25 μl) contained 10 mM Tris-HCl pH 9·0, 2 mM MgCl2, 10 pmol of each left and right primer, 200 μM of each dNTP, 1·0 unit of Taq DNA polymerase (Bangalore Genei, India) and 100 ng of template genomic DNA. The cycling condition was as follows: 1 cycle of 3 min at 94°C and 45 cycles of 1 min at 94°C, 1 min at 62°C and 20 s at 72°C followed by a final 10 min incubation at 72°C. The PCR products were separated on 2·5% agarose gel and some STMS were analysed on denaturing 5% polyacrylamide gels.
(iv) Extraction and analysis of seed proteins
Total protein was extracted from five seeds each of Sonalika, Kalyansona and F3 seeds from F2 plants. High molecular weight (HMW) glutenin subunit and other seed proteins were analysed by SDS-PAGE (Payne & Lawrence, Reference Payne and Lawrence1983).
(v) Analysis of gene-specific loci
(a) PCR-RFLP of the ITS region
ITS region from 18S-5.8S-26S rRNA was amplified using the primers G1: 5′-TCCGTAGGTGAACCTGCGG-3′ and C2: 5′-TCCTCCGCTTTATTGATATGC-3′ as detailed by Saini et al. (Reference Saini, Krishna, Reddy and Jawali2000). An aliquot of the PCR product was digested with a restriction endonuclease (4 units) in a 10 μl reaction mixture according to the manufacturer's instructions. Digested fragments were separated on a 3% high-resolution agarose gel in TBE at 8 V/cm for 1·5–2 h and then stained with ethidium bromide.
(b) PCR of puroindoline genes
The wild-type allele of pinA (Pina-D1a) was PCR-amplified using the allele-specific primers Pina-D1F-5′-CCCTGTAGAGACAAAGCTAA-3′, Pina-D1R-5′-CACCAGTAATAGCCAATAGTG-3′ as described by Gautier et al. (Reference Gautier, Aleman, Guirao, Marion and Joudrier1994).
(c) Rht-B1 and Rht-D1 amplification
The wild-type and mutant alleles of Rht-B1 and Rht-D1 were PCR-amplified using allele-specific primers according to Ellis et al. (Reference Ellis, Spielmeyer, Gale, Rebetzke and Richards2002).
(vi) Nullisomic-tetrasomic analysis
A series of nullisomic-tetrasomic lines of Chinese Spring (CS) (Sears, Reference Sears, Riley and Lewis1966) was used to physically map 22 AFLP primer combinations.
(vii) Data analysis and linkage mapping
In the case of a dominant marker the polymorphic band positions were scored as ‘0’ or ‘1’ for absence or presence of band, respectively. In the case of a co-dominant marker, the allele from the female parent was scored as ‘1’, the allele from the male parent was scored as ‘2’ and presence of the two alleles together (heterozygotes) was scored as ‘3’. The observed segregation ratios were tested by chi-square analyses (3:1). The linkage analysis was performed using MapMaker v.3.0b (Lander et al., Reference Lander, Green, Abrahamson, Barlow, Daley, Lincoln and Newburg1987) for the F2 population. Recombination frequencies were converted to centimorgans (cM) using Kosambi's mapping function (Kosambi, Reference Kosambi1944). The linkage groups were constructed using the ‘two-point/group’ command with a LOD threshold of 3·0 and a maximum distance of 50 cM.
(viii) QTL mapping
QTL Cartographer v.2.5 (Wang et al., Reference Wang, Basten, Gaffney and Zeng2005) was used for QTL analysis. Zmap QTL, Model 6 with a window size of 10 cM, was used for composite internal mapping (CIM) analyses. The number of markers for the background control was set to five. For each trait, a minimum LOD value of 2·5 was used for the identification of putative QTLs. Association of a marker with a QTL was analysed by a two-population t-test. The F2 population was divided into two groups based on the alleles of a marker closest to a QTL. The trait means of the two groups were subjected to a t-test for significance. QTL effects (R 2 values), also referred to as phenotypic variation, were obtained from the output file of CIM. Two combinations of the quantitative traits, viz. (1) FLB–FLL, (2) CL–FLB–FLL, were used for joint multitrait CIM (MCIM) using the module JZmapqtl available in QTL Cartographer.
3. Results
(i) Genetic linkage map
A genetic linkage map referred to as an ‘SK’ map consisting of 236 loci with a marker density of 15·4 cM was obtained. The map consisted of 37 linkage groups and spanned 3639 cM, with 1211·2 cM for the A genome, 1669·2 cM for the B genome, 192·4 cM for the D genome and 566·2 cM for unassigned groups. Twenty-four linkage groups were assigned to 17 chromosomes; however, none were assigned to chromosomes 1D, 2D, 4D and 7D (Table 1). The number of markers mapped was highest in the B genome (97) followed by the A genome (72) and D genome (17) (Fig. 1).
Genetic linkage map of bread wheat using an F2 population derived from a cross between Sonalika and Kalyansona. The name of the chromosome or group is given on the top of individual linkage group. The names of markers are given on the right-hand side and the distance (in centimorgans) between markers on the left-hand side of the linkage groups. The arrows on chromosomes 3A1, 5A4 and 6A indicate the markers E16_M5D, E3_M5B and Xgwm1696A, respectively, which are lying closest to the QTLs for the trait culm length (CL). The arrows on chromosome 5A2 indicate the markers E1_M7G and OPAA07C, which are lying closest to the QTLs for the trait flag leaf length (FLL). The black coloration within the linkage group indicates the QTL at these locations. The STMS markers indicated as ‘MS’ are the ‘Xgwm’ series of STMS described by Roder et al. (Reference Roder, Korzun, Wandehake, Planschke, Tixier, Leroy and Ganal1998). The arrows on chromosomes 2B1 and 6B indicate the markers E7_M4H, SS13RA and E2_M7C, respectively, closest to the QTLs for CL. The arrows on chromosomes 1B, 2B1 and 7B indicate the markers E14_M5J, Proi, OPAB18A, VM13RA, and SS26LB, respectively, closest to the QTLs for FLL. The arrow on linkage group 1 indicates the marker E17_M7M closest to the QTL for trait flag leaf breadth (FLB). The arrows on linkage groups 11 and 12 indicate the markers OPAB08A, and E19_M7A closest to the QTLs for traits FLB and CL, respectively.
Number of markers and length of each linkage group
(ii) Segregation distortion
Of the 280 markers analysed, 89 (31%) deviated significantly (P<0·05) from a 3:1 ratio and this was not specific to any marker type. Of the 89 distorted markers, 74 were mapped and 15 remained unlinked. Thirty-nine mapped markers (53%) showed a segregation distortion in favour of Kalyansona and 35 (47%) in favour of Sonalika, indicating no bias towards a particular parent.
(iii) Frequency distribution, ANOVA and correlation among the traits
The frequency distribution of each of the three traits in the segregating population was found to be different. While FLB showed a normal distribution, FLL showed a distribution skewed towards shorter leaf length. A skewed distribution indicates higher frequency of a phenotype. The skewed distribution could arise due to (a) dominance of the alleles responsible for shorter leaf length, (b) epistatic action of leaf length inhibitor as well as (c) genotype×environment interactions. CL showed a double bell-shaped curve distribution. This is due to segregation for the two major semi-dwarfing genes present in the parents, viz. Kalyansona harbours Rht-B1b while Sonalika harbours the Rht-D1b gene. In the absence of major dwarfing genes the population would have shown a normal distribution. Further analysis showed that three combinations, viz. CL–FLL, CL–FLB and FLL–FLB, among the three quantitative traits were positively correlated.
(iv) Composite interval mapping (CIM) of QTLs
Twenty-five QTLs with LOD scores above 2·5 spread over seven chromosomes were detected for the three traits (Table 2). Eight QTLs for CL with LOD scores ranging from 2·5 to 4·3 and phenotypic variation (QTL effect, R 2) ranging from 21·6% to 66·5% were found on chromosomes 2B, 3A, 5A, 6A, 6B, 7B and linkage group 12. Twelve QTLs for FLL with LOD scores ranging from 2·5 to 7·2 and phenotypic variation ranging from 8·2% to 39·0% were found on chromosomes 1B, 2B, 5A, 6A, 6B and 7B. Five QTLs for FLB with LOD scores ranging from 2·5 to 3·3 and phenotypic variation ranging from 10·9% to 34·9% were found on chromosomes 2B, 5A, 6A, and linkage groups 1 and 11.
Composite interval mapping (CIM) for three quantitative traits
*, ** Means for marker allele classes, which differed significantly at P<0·05 and 0·01 respectively. The markers associated with the traits are shown in bold.
a Intervals in cM were obtained by marking positions ±1 LOD from the peak.
b Values in parentheses are the distances (cM) of the marker from the peak.
(v) Multitrait composite interval mapping (MCIM)
Of the various combinations of the quantitative traits that were used for correlation analysis, two combinations, viz. FLB–FLL, CL–FLB–FLL, which showed positive correlation, were chosen for MCIM analysis. The results are described below.
(a) Flag leaf breadth and flag leaf length
The results of MCIM are given in Table 3. Twenty-two QTLs were detected in joint MCIM, of which nine were also detected by CIM.
QTLs detected by joint MCIM involving two correlated quantitative traits: flag leaf breadth (FLB) and flag leaf length (FLL)
a The closest marker is the one lying next to the QTL locus influencing the QTL at the respective LOD value.
(b) Culm length, flag leaf breadth and flag leaf length
The results of MCIM are given in Table 4. Forty-three QTLs were detected in joint MCIM, of which 17 were also detected by CIM. As an example a representative QTL Cartographer plot involving chromosome 6B obtained using MCIM involving three correlated metric traits – trait 1, culm length; trait 2, leaf breadth; trait 3, leaf length – is shown in Fig. 2.
A representative QTL Cartographer plot involving chromosome 6B obtained using multitrait composite interval mapping (MCIM) involving three correlated metric traits: trait 1, culm length (CL); trait 2, flag leaf breadth (FLB); trait 3, flag leaf length (FLL). The LOD value is given on the y-axis and the name of the markers and the distance between them in centimorgans is given on the x-axis.
QTLs detected by joint MCIM involving three correlated quantitative traits: culm length (CL), flag leaf breadth (FLB) and flag leaf length (FLL)
a The closest marker is the one lying next to the QTL locus influencing the QTL at the respective LOD value.
(vi) Association of molecular markers with quantitative traits by t-test
Thirty markers that were closest to the QTLs were analysed further. The means for the trait concerned were estimated for each of the two-allele classes and subjected to a t-test. Among five of the markers, the differences between the allele classes were found to be significant, indicating that these markers show association with the respective traits (Table 2), viz. one marker associated with CL, two for FLB and two for FLL.
4. Discussion
The aim of the study was to develop a linkage map based on varieties which were extensively used in cultivation and in wheat breeding, with the purpose of identifying associations between markers and QTLs. In this study an intervarietal map has been constructed based on the Indian bread wheat cultivars Kalyansona and Sonalika. In general, cultivated wheat varieties exhibit narrow genetic diversity. However, the two varieties used in this study showed differences in 10 agronomically important traits and also exhibited considerable DNA polymorphisms. The differences at DNA level detected as polymorphisms were observed to be mostly associated with QTLs for the observed differences. An F2 population was used as it is available earlier and expected to be unbiased.
(i) Genetic linkage map
The length of the SK map (3639 cM) is comparable with reported wheat maps lengths such as the Courtot×Chinese Spring intervarietal map (3685 cM; Sourdille et al., Reference Sourdille, Cadalen, Guyomarc'h, Snape, Perretant, Charmet, Boeuf, Bernard and Bernard2003) and the Cranbrook×Halberd intervarietal map (4110 cM; Chalmers et al., Reference Chalmers, Campbell, Kretschmer, Karakousis, Henschke, Pierens, Harker, Pallotta, Cornish, Shariflou, Rampling, McLauchlan, Daggard, Sharp, Holton, Sutherland, Appels and Langridge2001), and is less than the Synthetic W7984×Opata 85 (ITMI) map (>5000 cM; for review see Langridge et al., Reference Langridge, Lagudah, Holton, Appels, Sharp and Chalmers2001). The number of markers in the ITMI map is 1065, while it is 659 in the Courtot×Chinese Spring map and 902 in the Cranbrook×Halberd map. The mean interval between two markers on the ITMI map is ∼5·8 cM, while on the Courtot×Chinese Spring intervarietal map it is 5·6 cM and on the Cranbrook×Halberd map it is 4·5 cM and on the SK map it is 15·4 cM. The large difference in the marker frequency could be attributed to the number of markers on the map.
In the SK map, maximum number and proportion of markers were mapped onto the B genome (97, 41%), followed by the A genome (72, 31%) and then the D genome (17, 7%). Chromosomes 1D, 2D, 4D and 7D were not represented at all. A low level of polymorphism in the D genome observed is in agreement with the reports available in the literature and with the hypothesis suggesting a recent and monophyletic introduction of the D genome in bread wheat (Lagudah et al., Reference Lagudah, Appels, Brown and McNeil1991). The higher proportions of markers placed on the chromosomes 1B, 6B and 5A indicated that the two parents could be carrying more variations in these chromosomes than the rest.
The lengths of chromosomes 3A, 2B, 5B, 6B, 7B and 3D in the SK map are comparable to the sizes reported by others. The lengths of chromosomes 1A, 2A, 4A, 6A, 7A, 5D, 6D, 3B and 4B in the SK map are shorter than the lengths reported in the ITMI genetic linkage map. This could be due to the lower number of markers in the map. In contrast, the lengths of chromosomes 5A and 1B in the SK map are longer than the lengths reported in previous maps. Apparently addition of more markers was not the only reason for the increase in the length of chromosomes 5A and 1B, because for chromosome 6B, which had a similar number of markers to 5A and 1B, the chromosome length was comparable to that reported in the literature. The majority of the markers on chromosomes 5A and 1B were AFLP markers, and this may have led to map stretching. To test the effect of the type of marker on chromosome length, the lengths of chromosomes 1B and 5A were estimated using two modifications: (1) selectively withdrawing AFLP markers located on these chromosomes and (2) withdrawing non-AFLP markers. The results showed that withdrawal of AFLP markers resulted in a larger change in chromosome length (Table 5, chromosome 1B, rows 2–6) while withdrawal of non-AFLP markers resulted in a smaller change in length (Table 5, chromosome 1B, rows 7–9). It was also observed that withdrawal of AFLP markers resulted not only in compression at the interval per se but also all over the chromosome (Table 5). For example, when AFLP marker E6_M4A from chromosome 1B was removed the compression at the interval per se was only 7 cM but the overall length of the chromosome reduced by 20 cM. Similar results were found for other chromosomes including chromosome 6B. Stretching of linkage maps with the incorporation of AFLP markers in the map along with other markers such as RAPD, ISSR and SSR has been shown previously in several studies on cereals (Becker et al., Reference Becker, Vos, Kuiper, Salamini and Heun1995; Maheswaran et al., Reference Maheswaran, Subudhi, Nandi, Xu, Parco, Yang and Huang1997; Castiglioni et al., Reference Castiglioni, Ajmone-Marsan and Wijk1999; Lotti et al., Reference Lotti, Salvi, Pasqualone, Tuberosa and Blanco2000; Saal & Wricke, Reference Saal and Wricke2002). The stretching of the map in the case of durum wheat (Lotti et al., Reference Lotti, Salvi, Pasqualone, Tuberosa and Blanco2000), barley (Becker et al., Reference Becker, Vos, Kuiper, Salamini and Heun1995) and rice (Maheswaran et al., Reference Maheswaran, Subudhi, Nandi, Xu, Parco, Yang and Huang1997) was 52·5%, 70·9% and 68·9%, respectively. Map stretching could occur due to the addition of map distances as new markers are discovered and also due to differences in methods used in the construction of an existing map and the new data being superimposed. In the present case the methods used for the earlier mapping and the superimposed markers were same; therefore the increase in map distances could be due to addition. Genetic distances are subject to modification as new loci are discovered between the existing ones.
Effect of removal of a marker on the length of chromosome 1B
Rows 2–6: AFLP markers, Rows 7–9: Non-AFLP markers.
The distortion in segregation ratios for the markers observed (31%) in the SK map is comparable to the segregation distortion reported by others in wheat (27%, Cadalen et al., Reference Cadalen, Boeuf, Bernard and Bernard1997; 35%, Messmer et al., Reference Messmer, Keller, Zanetti and Keller1999). Segregation distortion is reported among F2 progenies of wheat (Liu & Tsunewaki, Reference Liu and Tsunewaki1991). The segregation distortion in the SK map was not biased towards a particular marker type; also, when all markers were considered together, the segregation distortion was not found to be biased towards any parental allele. The segregation distortion observed could be (a) due to the polymorphic band being amplified from more than one loci or (b) the phenotypes associated with the marker may influence selection towards a particular allele. Segregation distortion is also reported among F2 progenies in other plants such as rice (McCouch et al., Reference McCouch, Kochert, Yu, Wang, Khush, Coffman and Tanksley1988), lettuce (Landry et al., Reference Landry, Kesseli, Farrara and Michelmore1987) and tomato (Helentjaris et al., Reference Helentjaris, Slocum, Wright, Schaefer and Nienhuis1986).
Marker order and distances of some regions on the SK map and the reported maps were found to be comparable. Specific gene/loci markers such as Rht-B1b, Glu-B1 loci and the Nor-B1 locus along with the STMS loci allowed the comparison of two different intervals on the SK map with microsatellite (Roder et al., Reference Roder, Korzun, Wandehake, Planschke, Tixier, Leroy and Ganal1998) and consensus genetic (Somers et al., Reference Somers, Isaac and Edwards2004) maps. The distance between Rht-B1 and the microsatellite marker Xgwm368-4B on chromosome 4B in the SK map was 9·2 cM, which is similar to the reported distance of 9·0 cM (Roder et al., Reference Roder, Korzun, Wandehake, Planschke, Tixier, Leroy and Ganal1998). The reported distance between Glu-B1 and Nor-B1 loci on chromosome 1B (22 cM) (Payne et al., Reference Payne, Holt, Hutchinson and Bennett1984; Ellis et al., Reference Ellis, Spielmeyer, Gale, Rebetzke and Richards2002; Ram et al., Reference Ram, Boyko, Giroux and Gill2002) is shorter than the distance estimated in this study (29 cM). This could be due to addition of a marker between the two loci and/or the computational method used.
(ii) QTL mapping
In recent years the availability of DNA markers and powerful biometric analytical tools has led to considerable progress in QTL mapping in plants. There are several types of experimental designs for QTL analysis and the choice of method depends on the mating system of the crop species. Most QTL analyses in plants involve populations derived from pure lines and use several approaches to associate QTLs with molecular markers. In this study an F2 population was used to detect the loci significantly contributing to the traits of interest.
The SK map was used for QTL analysis of three metric traits that differed between the parents. We used CIM and MCIM, which are often recommended for power and precision of QTL analysis. CIM and MCIM have been used recently in bread wheat (Kulwal et al., Reference Kulwal, Roy, Balyan and Gupta2003; Campbell et al., Reference Campbell, Baenziger, Gill, Eskridge, Budak, Erayman, Dweikat and Yen2003; Marza et al., Reference Marza, Bai, Carver and Zhou2006). CIM is an extension of simple interval mapping (SIM) that considers both the markers flanking the QTL and background markers, which could be or need not be linked to the QTL. CIM is said to give more power and precision in the detection of QTLs than SIM. CIM has been used in QTL mapping in wheat (Shah et al., Reference Shah, Gill, Baenziger, Yen, Kaeppler and Ariyarathne1999; Campbell et al., Reference Campbell, Baenziger, Gill, Eskridge, Budak, Erayman, Dweikat and Yen2003; Kulwal et al., Reference Kulwal, Roy, Balyan and Gupta2003). One of the most important advantages of CIM is that the markers can be used as boundary conditions to narrow down the most likely QTL position. The resolution of QTL locations can be greatly increased.
(a) Culm length (CL)
Culm length (or plant height) is an important trait that contributes to the plant's stature. Classical genetic studies have shown that genetic control of CL in bread wheat is complex, and most chromosomes harbour factors (loci) that can affect it (Law et al., Reference Law, Snape and Worland1973). To date 21 loci with major effect on plant height have been identified (Worland et al., Reference Worland, Korzun, Roder, Ganal and Law1998). The two most common semi-dwarfing genes, Rht-B1b and Rht-D1b, are present on the short arms of chromosomes 4B and 4D (Ellis et al., Reference Ellis, Spielmeyer, Gale, Rebetzke and Richards2002), respectively, and are gibberellic acid (GA)-insensitive. Both Sonalika and Kalyansona are semi-dwarf genotypes, and harbour Rht-D1b and Rht-B1b, respectively. Of the eight QTLs that were detected for CL in this study, the QTL on chromosome 6A showed the highest phenotypic variation (66·5%). The QTL with the highest LOD score of 4·3 and a phenotypic variation of 65·3% was on chromosome 3A1 with E16_M5D as the closest marker. The STMS marker Xgwm169-6A is closest to the QTL for plant height on chromosome 6A. Since genes for plant height are known to be present on chromosome 6A, this STMS marker could be linked to one of these genes.
In addition to Rht-B1b and Rht-D1b, a large number of QTLs for CL have been reported by many workers (Table 6). Two major QTLs on chromosome 6B, with LOD scores of 3·2 and 4·3, have been observed in this study. A QTL for CL on chromosome 6B has not been reported, although previous cytogenetic studies have indicated that a gene for plant height is also present on chromosome 6B (Goud & Sridevi, Reference Goud, Sridevi, Miller and Koebner1988). The two parents Sonalika (average CL=57·75±1·6 cm) and Kalyansona (average CL=49·3±2·34 cm) are semi-dwarf genotypes. The phenotype exhibited is contributed to by the eight QTLs in addition to the two dwarfing genes.
List of QTLs detected in the present study and reported in literature for the four quantitative traits
QTLs in bold are detected in this study.
(b) Flag leaf length (FLL) and flag leaf breadth (FLB)
FLL and FLB determine area, which is an important trait. QTLs for leaf breadth reported in the literature are listed in Table 6. Monosomic analysis has shown that chromosomes 1A, 2A, 3A, 3B, 4B, 6D and 7D affect FLB (Iqbal & Vahidy, Reference Iqbal and Vahidy1992). Among the five QTLs identified for FLB in this study, the QTL with the highest phenotypic variation (34·9%) is present on unassigned linkage group 1. STMS marker Xgwm3045A was lying closest to the QTL on 5A showing a phenotypic variation of 10·9%. A QTL on chromosome 5A with a phenotypic variation of 14·9% has been reported by Keller et al. (Reference Keller, Karutz, Schmid, Stamp, Winzeler, Keller and Messmer1999). QTLs for FLB on chromosome 2B have not been reported previously.
Monosomic analysis of FLL in bread wheat had indicated that chromosomes 1A, 2A, 5A, 6A, 2B, 3B, 4B, 5B and 6D affected FLL (Iqbal & Vahidy, Reference Iqbal and Vahidy1992). Of the 12 QTLs for FLL identified in this study, those on chromosomes 5A, 6A and 2B were also identified by monosomic analysis and the one exhibiting maximum phenotypic variation (39%) was on chromosome 6A.
(c) General observation
For CL and FLL, often, more than one QTL for the same trait was identified on a chromosome. However, more than one QTL within the same interval, if present, cannot be identified since CIM does not have the power to resolve such linked QTLs. Among QTLs for a trait located on the same chromosome, two QTLs for CL were located on 6B and three QTLs for FLL were located on 1B, two on chromosome 5A2 and two on chromosome 6B. Such QTLs on the same chromosome were not linked. In all cases QTLs on the same chromosome were separated by long genetic distances ranging from 13 to 280 cM, thus suggesting absence of close linkage or no linkage between the QTLs.
(iii) Multitrait composite interval mapping (MCIM)
MCIM has been used recently as a means of improving the power and precision of QTL detection for correlated traits, as information on the traits acts like repeated measurements. Using MCIM, loci showing pleiotropy on the traits CL, FLB and FLL were analysed.
(a) Flag leaf breadth – flag leaf length
The traits FLB and FLL showed significant positive correlation (r=0·53; P<0·01), indicating that there could be some loci affecting both the traits. Of the 22 QTLs detected by joint MCIM, the nine QTLs which were also detected by CIM in addition to joint MCIM are the QTLs which influences both the traits in the combination. Of these nine, the QTL on chromosome 7B was also detected for CL. Two of these nine QTLs located on chromosomes 1B and 5A1 were the same as those for which the corresponding molecular markers showed significant association with the traits in question (see later).
(b) Culm length – flag leaf breadth – flag leaf length
Positive correlation was observed between CL and FLB (r=0·22; P<0·05) and between CL and FLL (r=0·22; P<0·05), indicating that there could be some loci influencing all three traits. MCIM analysis on FLB, FLL and CL detected a few more QTLs in addition to the ones detected by MCIM analysis on FLB and FLL. In addition to MCIM analysis on FLB and FLL, additional QTLs were detected when all three traits were analysed together for multitrait analyses. Seven QTLs located on chromosomes 3A1, 3B, 5A4, 6B, 7B and linkage group 12 were detected exclusively for CL by CIM but were also detected by joint MCIM for all three traits. Joint MCIM detected more QTLs jointly affecting CL, FLL and FLB than the QTLs jointly affecting only FLL and FLB. Of the 43 QTLs detected by joint MCIM, 17 which were also detected by CIM in addition to joint MCIM were the QTLs which influence both the traits in the combination. Three of the 17 QTLs located on chromosomes 1B, 3A2 and 7B are the same as those for which the corresponding molecular markers showed significant association with the traits in question (see later).
In both the above MCIM analyses the LOD scores for the common QTLs were higher in MCIM than in CIM. This shows that the level of confidence in MCIM for QTL detection is higher even than for CIM. The repeated finding of a QTL by two methods such as CIM and MCIM confirms the QTL only in the available data. However, a separate set of data would be needed to confirm the presence of a QTL. A similar finding has been reported by Kulwal et al. (Reference Kulwal, Roy, Balyan and Gupta2003), although with a different combination of traits. The QTLs detected by joint MCIM could suggest pleiotropy as the possible cause of correlation among the correlated traits. This inference may be taken into account while designing experiments involving molecular MAS aimed at improving more than one trait simultaneously.
(iv) Markers associated with quantitative traits
The significance of marker–trait association was analysed by t-test to check their usefulness in MAS and in detecting probable false QTLs. For instance OPAB18A (FLL) and SS26LB (FLL) were coincident with the QTL positions but did not show a significant association with the trait. It was noticed (Table 2) that the marker–trait association was not always correlated either with the distance of the marker from the QTL position or with the size of confidence interval. Also, no significant correlation between marker–trait associations and the magnitude of the LOD score was detected, in contrast to that reported by Kulwal et al. (Reference Kulwal, Roy, Balyan and Gupta2003). Markers having lower LOD values but showing association could be due to a lower confidence interval and the marker being very close to the QTL position. A few markers with LOD values higher than 3·9 did not show association with the trait in question. This could be due to the lower contribution to phenotypic variation or the larger distance between the marker and the QTL position or larger confidence interval. Of the five marker–trait associations found, four were also detected by MCIM for the trait(s) in question and these markers could prove to be useful in MAS.
5. Conclusions
An intervarietal genetic linkage map based on a cross between the two wheat Indian varieties Sonalika and Kalyansona was developed. These two varieties have served as the starting material for many of the later-developed cultivars and hence the markers thus obtained would be useful for future breeding programmes involving parents related to the two varieties. Several QTLs were detected for three quantitative traits, of which 15 have not been reported previously. Cultivated Indian bread wheat varieties have narrow diversity; however, we found many agronomically important traits which differed among Kalyansona and Sonalika and also sufficient polymorphisms at the DNA level. The four markers that showed association with quantitative traits could be useful in MAS.
Although genetic maps of wheat have been developed previously, many of which using the ITMI populations, there is a need to develop maps for specific populations for actual use. The Indian cultivars, for example, are spring wheats many of which carry cytogenetic variations. Also varieties adapted to certain agroclimatic conditions could carry variations. Hence an independent map based on Indian wheat varieties will thus be more useful.
