A total of 1536 DArT clones were developed from a PstI/TaqI representation generated from a mixture of DNA of two T. monococcum accessions MDR002 and MDR308 35. These two accessions were the parents for a large mapping population, which had previously been genotyped with SSR markers and used for mapping several agronomically important traits 89. In a parallel project, a substantial number of DArT clones were developed from genomic representations of other Triticum species with different ploidy levels (Triticarte P/L, unpublished). We combined clones from all projects to assemble a custom-designed array containing 1536 clones derived from the two T. monococcum accessions, 2304 clones derived from hexaploid wheats (including the Triticarte Wheat 2.3 array), 1536 clones derived from tetraploid durum wheat (including the Triticarte Durum 2.0 array), and 1536 clones derived from 15 Iranian accessions of T. boeoticum Boiss., which is the wild relative of T. monococcum (see Additional file 1; Ali Mehrabi, unpublished data).

The composite array was used to assess the genetic diversity amongst sixteen T. monococcum accessions listed in Table 1. They were pre-selected based on their geographical origins, useful traits, available molecular and genetic tools, as well as genetic relationships assessed using SSR fingerprinting 8.

In total, 846 DArT markers were identified as polymorphic amongst the 16 accessions. Polymorphism Information Content (PIC) values for these markers were relatively low, with only 35.5% of DArTs having PIC values of 0.4-0.5 and 34.6% with PIC values < 0.2 (Table 2). The average mean PIC value was 0.31. When the quality of the DArT markers (measured as the % of total variance in hybridisation intensity between the two clusters: present and absent) was analysed against their performance, which is determined through call rate and PIC values, more than half of the polymorphic DArT markers (n = 439) were in the 90-100% quality category with an average PIC value of 0.34 and call rate of 99.8%, respectively (Table 3). The average PIC value decreased with the average quality value. PIC values were > 0.30 when the marker quality was > 80%. However, PIC values were reduced to 0.24, 0.13 and 0.12 for markers with the quality values of 70-80%, 60-70% and 50-60%, respectively.

This set of polymorphic DArT markers was used to assess the genetic diversity of the 16 T. monococcum accessions. A Jaccard similarity matrix was generated and used to construct a principal coordinate plot deciphering the genetic relationships among the accessions (Figure 1). The first two principal coordinates derived from the scores jointly explained 23.74% of the total data variance. There was a clear separation for most of the T. monococcum accessions and partial correlations between genetic relationships and geographic origins. The accession MDR650 (PI 355520) of Iranian origin was fairly distantly related to other accessions including most of the accessions of European origin. This accession is unique in its ability to produce fertile F₁ hybrids with hexaploid wheat 40. MDR650 is also a known source of resistance to leaf rust 41 and provides resistance to a range of other wheat pathogens (HCJ and KHK, unpublished). Our previous diversity study using SSR fingerprinting has shown that another Iranian accession MDR049 also formed an independent clade 8. Furthermore, in the current study MDR308 and MDR043 were clustered close to each other and were distant to MDR002. Amongst the 14 analysed accessions there were two accessions with unusual pedigrees. MDR050 is an inbred line derived from a cross between T. monococcum and Triticum sinskajae 42, whereas MDR657 (L118) is a recombinant inbred line generated by several rounds of crosses and backcrosses between Triticum boeoticum and Triticum urartu (B. Kilian, personal communication).

As described above, the custom-designed DArT array contained clones developed from the T. monococcum genome and clones derived from the genomes of other Triticum species. This allowed us to assess (1) the degree of genetic similarity/diversity between different T. monococcum accessions, and (2) the relationships between the genome of T. monococcum and the A-genomes of other closely related Triticum species. Analysis of the 16 T. monococcum accessions revealed that 317 of the 1536 of DArT markers (~20%) developed from T. monococcum accessions MDR002 and MDR308 were polymorphic (Table 4). In this set of T. monococcum accessions ~10% of DArT markers derived from the genomes of hexaploid and tetraploid wheat were polymorphic, whereas only ~8.2% of DArTs derived from T. boeoticum were polymorphic. The latter markers also displayed the lowest PIC values. A similar trend was observed for the polymorphic DArT markers in the F₂ mapping population derived from MDR308 × MDR002 cross (see below). Overall, the DArT markers developed from T. monococcum genome were found to be more informative than those developed from the genomes of other related Triticum species.

To test the power of the DArT markers of different origins in resolving the T. monococcum relationships, the 846 polymorphic DArT markers were split according to their genome origins and principal coordinate analyses were carried out using the four subset data. The percentages of total data variance explained by the first two coordinates were 27.75% in hexaploid wheat, 23.96% in tetraploid wheat, 35.26% in genomes of T. boeoticum, and 35.77% in T. monococcum, respectively (Figure 2). These principal coordinate similarity matrices had a correlation coefficient between 0.73-0.94 in describing the relationships among these T. monococcum accessions (Table 5). The Mantel test 43 indicated that the matrices are highly significantly associated (all the p = 0.000, less than 0.001).

To further analyse the relationships of the genomes of T. monococcum and those of hexaploid wheat, we simultaneously hybridised the genomes of nine T. monococcum accessions and 23 hexaploid wheat varieties of European, American and Chinese origins to a customised Triticarte DArT array (see Additional file 2). The ploidy levels of the genomes did influence the scoring and hence the raw data was analysed to identify DArT markers which were not affected by this genome context-dependent scoring. As shown in Table 6, 1036 DArT markers were reliably scored, of which 696 markers were polymorphic in hexaploid wheat genomes but were commonly present or absent in T. monococcum genomes; whilst 238 markers were polymorphic in T. monococcum genomes but were monomorphic in hexaploid genomes. However, there were 102 DArT markers which were polymorphic in the genomes of both ploidy. The principal coordinate plot constructed using these DArT markers showed that 38% of the variation was explained by the two-dimensional analysis (Figure 3). The plot showed a good separation of the genomes with different ploidy levels and the genomes within each ploidy level.

To confirm that DArT markers were inherited in a Mendelian manner, we constructed a linkage map for a cross between T. monococcum accessions MDR308 and MDR002 8. Hybridising the array with a total of 94 F₂ progeny from this cross as well as the two parental accessions identified 300 polymorphic DArT markers: 133, 71, 61 and 35 were derived from genomes of T. monococcum, hexaploid wheat, tetraploid wheat, and T. boeoticum, respectively (Table 4). The same T. monococcum mapping population was also genotyped using microsatellite markers mapped to the A sub-genome of hexaploid wheat 44. Out of 274 microsatellite markers analysed, 90 (32.8%) were polymorphic between the two parental T. monococcum accessions MDR002 and MDR308. Rates of polymorphism for different types of microsatellite markers were as follows: WMC - 41.9% (36 polymorphic markers out of 86 examined), CFA and CFD - 39.3% (11 out of 28), BARC markers - 38.2% (21 out of 55), GDM and WMS - 33.7% (28 out of 83), and DuPw - 13.6% (3 out of 22). The data for DArT and microsatellite markers were merged for construction of a genetic linkage map for T. monococcum.

In T. monococcum, black awns and hairy leaves are two morphological traits known to be controlled by single dominant genes Ba and Hl, respectively. Our previous genetic study revealed a linkage between Ba and the SSR locus Xwmc336 on 1A^m, and between Hl and the SSR locus Xcfd39 on 5A^m 8. These two traits segregated in MDR002 × MDR308 F₂ mapping population as single independent loci and therefore it was possible to refine the positions of Ba and Hl in T. monococcum genome. Previous estimations showed that the genetic distance was 21.5 cM between Ba and Xwmc336, and 17.4 cM between Hl and Xcfd39. Linkage analysis combining SSR and DArT markers confirmed that Ba was located at 80.09 cM on 1A^m between the DArT marker locus wPt-2584 at 60.59 cM and the SSR locus Xgwmd33 at 83.76 cM, and Hl was located at 136.16 cM on 5A^m between the DArT marker loci 376589 at 134.84 cM and 469591 at 138.77 cM.

DArT markers are typically developed from a representation which is generated from a pool of DNA samples from a number of accessions, cultivars or breeding lines which as a group cover the genetic diversity within a species or a group of closely related species 27. In the current study, DArT markers were initially generated from genomic representations of two T. monococcum accessions, parents of an F₂ mapping population. In addition, this study took advantage of the existence of a large number of DArT markers previously developed from hexaploid, tetraploid and other diploid Triticum species.

The inclusion of DArT clones from various genomes meant the average PIC value of the data set (0.31) was lower than in comparable studies. For example, for barley, sorghum and cassava the values obtained were 0.38, 0.41 and 0.42, respectively 283334. However, the data provided useful information for comparison of Triticum genomes of different ploidy levels. In the genetic diversity assessment of 16 T. monococcum accessions, 20.6% (317 out of 1536) of the DArT markers developed from the two T. monococcum accessions were polymorphic. DArT markers developed from genomes of other Triticum species also displayed good polymorphism frequencies (10.68%, 10.22% and 8.2% for markers developed from hexaploid, tetraploid and diploid species, respectively) in T. monococcum. In hexaploid wheat, 15.3% (788 out of 5137) of DArT markers were found to be polymorphic when assessing the genetic diversity of 13 Australian cultivars 35. Similar studies in durum wheat and barley revealed only 9.6% and 10.4% of polymorphic DArTs, respectively 2836. Thus, the group of 16 T. monococcum accession genotyped in this study appeared to be more genetically diverse than the bread and durum wheat collections assayed in those studies. Furthermore, most of the DArT markers whose genome locations have been determined in previous studies were located on homoeologous chromosomes of T. monococcum. Thus, DArT markers from related genomes were also useful in probing genetic diversity in T. monococcum. The custom-designed DArT array developed here can be used in studies focusing on comparison of the T. monococcum genome with genomes of other Triticum species.

Principal coordinate analysis of the 16 T. monococcum accession revealed that the site of collection is only partially correlated with genetic diversity. Most accessions used in this study have been genotyped with SSR markers in a previous study, and a similar pattern had been observed 8. Thus, DArT fingerprints are as useful for resolving the genetic relationships in T. monococcum as non-DArT based markers. One of the accessions, MDR650 (PI 355520) was found to be genetically distantly related to other T. monococcum accessions, which raised a question whether it is a true T. monococcum. However, MDR650 (PI 355520) possesses the three major traits of the domesticated einkorn wheat: larger and plumper seeds, a tough rachis preventing spikelets falling apart at maturity, and relatively easy threshing. These traits are absent in the wild species T. boeoticum. We have crossed MDR650 (PI 355520) with many other T. monococcum accessions and discovered that the crossability could reach 100% (data not shown). Furthermore, MDR650 (PI 355520) did not cluster with hexaploid wheat and was closer to the T. monococcum cluster. Thus, we still consider MDR650 (PI 355520) an accession of T. monococcum.

The relationships between the A^m-genome of T. monococcum and those of hexaploid wheat T. aestivum and its sub-genome donor, T. urartu have been studied in different context. For instance, chromosome pairing and recombination were studied between homoeologous chromosomes 1A and 1A^m, 3A and 3A^m as well as 5A and 5A^m and were shown to be collinear and differentiated at sub-structural level 2046. Good micro-colinearity/colinearity at particular genetic loci has also been reported 4748. Other studies as well as the present one demonstrate that many A-genome molecular markers could also be used for genetic studies in T. monococcum 2425. However, their genome-wide use was not previously reported. The use of DArT markers allowed us to assess the relationships in a new perspective. We showed here that most of the DArT markers which are polymorphic in T. monococcum were either conserved or absent in hexaploid genomes, and vice versa, indicative of divergence of the A^m-genome from the A-genome 25. On the other hand, many DArT markers originated from B- and D-genomes of hexaploid wheat could hybridise to the T. monococcum genomes and provide the polymorphism information in T. monococcum. Thus, at the homoeologous genomes level, there is a complex relationship between T. monococcum and hexaploid wheat.

A T. monococcum genetic map has been previously constructed using MDR308 (DV92) as one of parental accessions by Dubcovsky et al (1996) 21. This map contains 335 markers, including RFLPs, isozymes, seed storage proteins, rRNA, and various morphological loci. The total length of this map is 1071.6 cM, the average chromosome length is 153.1 cM, and the average marker density is one marker per 3.2 cM. More recently, Singh et al (2007) 24 used a T. boeoticum × T. monococcum RIL population to construct another genetic linkage map integrating 177 SSR and RFLP markers, and two morphological trait loci. The total length of this map is 1262 cM, the average chromosome length is 180.3 cM, and the average marker density is one marker per 7.05 cM. In comparison, our current genetic map derived from linkage analysis of an F₂ population from a cross between MDR308 and MDR002 integrates 274 DArT markers, 82 SSR markers and two morphological trait loci. This map spans over 1062.7 cM, with six chromosomes represented by single linkage groups and chromosome 4A^m by three groups of linked markers. The average chromosome length is 151.8 cM, and the average marker density in the genome is one marker per 2.97 cM. Observed differences in lengths and marker densities for the three T. monococcum linkage maps may well be related to differences in the mapping populations, the types and error-rates of the genetic marker systems used, and/or the algorithms and mapping functions used to compute genetic distances. For example, Singh et al (2007) 24 used the Haldane's mapping function, whereas the Kosambi's mapping function was used in other cases.

The current genetic linkage is also very similar in length and in marker densities to the A-genome maps of hexaploid wheat constructed using the Kosambi's map function. In one case, 369 SSR markers were mapped onto a 944 cM A-genome with the marker density of one marker per 2.56 cM 44. In another study, 464 SSR markers were mapped onto a 1231 cM A-genome with the marker density of one marker per 2.65 cM 49. Recently DArT markers have also been developed and integrated with conventional markers i.e. SSRs, RFLPs and AFLPs in hexaploid wheat 35 and tetraploid wheat 36.

The A^m-genome of T. monococcum is closely related to the A^u-genome of T. urartu and to the A-genome of the hexaploid wheat 456. This is reflected by the high transferability of SSR markers from hexaploid wheat to T. monococcum and the overall good colinearity of SSR arrangement along the chromosomes 824. Gaps are frequently observed in hexaploid wheat genetic linkage maps making use of SSR, RFLP and AFLP markers 49505152. Similarly, Singh et al (2007) 24 observed four large gaps on linkage maps of chromosomes 2A^m, 4A^m and 7A^m in T. monococcum. In contrast, in our current linkage map for T. monococcum there was only two gaps on chromosome 4A^m. This suggests that DArT markers can be used to fill the gaps and help generate higher resolution genetic linkage maps.

Strong segregation distortion was observed during construction of our current genetic linkage map for T. monococcum. Over one third of the marker loci across the seven chromosomes displayed allele frequencies skewed from their Mendelian expectations. This was observed for both DArT and SSR markers, with preferences for both parental alleles. Segregation distortion has also been observed by Dubcovsky et al (1996) 21 and Singh et al (2007) 24 during construction of einkorn wheat genetic linkage maps, regardless of the type of populations used. In one study (Dubcovsky et al 1996) 21 15% of the marker loci displayed segregation distortion and were clustered on chromosomes 1A^m and 7A^m, while in other study (Singh et al 2007) 24 two major distorted regions were detected on chromosome 2A^m. Strong segregation distortion has also been noted during the construction of durum wheat and bread wheat linkage maps integrating SSR and DArT markers 36.

The two parental T. monococcum lines, MDR308 and MDR002, used for developing the F₂ mapping population had a number of contrasting traits including awn colour, leaf pubescence, grain hardness, salt tolerance, as well as resistance to the fungus Mycosphaerella graminicola and to soil-borne cereal mosaic viruses 8953. We demonstrated here that the constructed genetic linkage map helped refine the chromosome regions spanning the Ba and Hl trait loci. Also, this map was recently used to refine the TmStb1 locus conferring resistance to M. graminicola isolate IPO323 on chromosome 7A^m, and to identify and map new QTLs conferring salt tolerance (HCJ and KHK, unpublished).

MDR308 (also known as DV92), one of the parental lines of the F₂ mapping population used in this study, has previously been used for developing a range of molecular tools including a genetic linkage map integrating RFLPs, AFLPs and seed storage proteins 54, a BAC library 55, and several populations of chemically- and radiation-induced mutants http://www.wgin.org.uk. The availability of DArT marker information for MDR308 (DV92) further increases the utility of this accession.

In the near future it should be possible to link various available genetic linkage maps of T. monococcum thereby creating a frame work consensus map for wider applications in molecular genetics and genomics studies.

This study used 16 T. monococcum accessions and an F₂ population of 98 individuals derived from a cross between accessions MDR002 and MDR308 8. The detailed information about these accessions is also listed in Table 1. In addition, 23 hexaploid wheat varieties collected from UK, continental Europe, China and America, or developed at CIMMYT, Mexico were used for genetic diversity compassion. The information about these hexaploid wheat varieties is provided in Additional file 2.

The DNA was extracted from leaves of 2-week-old einkorn wheat seedlings using the DNeasy Plant Mini Kit (QIAGEN) according to manufacturer's instructions. A set of 1536 new DArT clones were generated from a PstI/TaqI representation of the MDR002 and MDR308 accessions as described previously 2836. The new clones were printed together with 2304 polymorphism-enriched clones from hexaploid wheat, 1536 from tetraploid wheat, and 1536 from a group of 15 Iranian accessions of diploid T. boeoticum (see Additional file 1). The resulting composite array was then used to genotype the T. monococcum samples using the standard DArT protocol 2836.

A total of 279 SSR markers mapped to the A-genome of hexaploid wheat were tested for polymorphism between the T. monococcum accessions MDR002 and MDR308. These SSRs originated from 5 groups: 57 BARC markers from the Beltsville Agricultural Research Centre, USA 5657, 29 CFA and CFD markers from INRA Clermont-Ferrand, France 525859, 24 DuPw markers from DuPont company (60 et al. 2002, DuPont, unpublished; Dreisigacker et al. 2005), 85 GDM and WMS markers from IPK Gatersleben, Germany 6162, and 89 WMC markers from the Wheat Microsatellite Consortium 63. For each SSR primer pair, the 5'-end of the forward primer was labelled with infra-red dye (IRD700 or IRD800, LI-COR Biosciences UK Ltd). The PCR were carried out using the PCR Master Mix (Promega) in a 10-μl reaction volume containing 1× PCR Master Mix buffer, 0.1 μM forward and reverse primers, and 20-30 ng plant DNA. Amplifications were carried out in 96-well microtiter plates using a G-storm Thermal Cycler (GS4/GS4s, Gene Technologies, Essex, England). The programmes were 2 min at 94°C, followed by 30-35 cycles of 30 s at 94°C, 30 s annealing at 50-60°C (depending on the primer pairs), and 1 min at 72°C, and a final extension of 5 min at 72°C. The final PCR products were diluted to 20-40 times using formamide Li-Cor loading dye, denatured for 5 min at 85°C and stored on ice before 0.5-0.8 μ1 of the reaction mix was loaded on Li-Cor 4300 DNA Analyser.

A group of 16 T. monococcum accessions were genotyped using the custom-built DArT array as described previously 36. The polymorphic DArT markers were scored, and used for clustering analysis using principal coordinate plots. For genome comparison, genome representations of nine T. monococcum accessions and 23 hexaploid wheat varieties were hybridised to a newly built DArT array and scored the same way. The ploidy levels did influence the scoring as envisaged in the raw data of the hybridisation intensity. Therefore, only the DArT markers which were not affected by this context-dependent scoring were scored. The nine T. monococcum accessions and 23 hexaploid wheat varieties were hybridised in duplicate.

T. monococcum accessions MDR308, MDR002 and 94 F₂ individuals derived from a cross between them were genotyped with the custom-built array as described previously 36. In total, 300 DArT markers, 90 SSR markers as well as two morphological trait loci (Ba, black awn; Hl, hairy leaf) were used for genetic linkage map construction using JoinMap^® 4.0 (Van Ooijen, J.W., 2006, JoinMap^® 4.0, Software for the calculation of genetic linkage maps in experimental populations. Kyazma B. V., Wageningen, The Netherlands). Because of the inherent dominant nature of DArT markers which separates the female and male markers in the repulsion phase, the linkage maps for maternal DArT markers (scored as a, c) and for parental DArT markers (scored as b, d) were initially constructed independently with SSR markers. The two maps for individual chromosomal linkage groups were then merged together using SSR markers as bridges. Assignment of markers to linkage groups was achieved using logarithm of the odds (LOD) threshold values ranging from 3.0 to 10.0. The Kosambi's map function was used to estimate genetic distances. In total, 358 markers including 274 DArT markers, 82 SSR markers and 2 morphological trait locus markers were integrated into nine linkage groups. Goodness of fit for all the loci to an expected 1:2:1 or 1:3 segregation ratio was tested using chi-square (χ²) analysis. The graphical representation of the map was drawn using GGT2.2 software 64.

All statistical analyses were carried out using GenStat (10^th edition, VSN International, UK). For principal coordinate analysis Jaccard similarity matrices were generated using the DArT markers. Two-dimensional scores were calculated and used to generate scatter plot matrices of scores. For the Mantel test which looks for association between the off-diagonal values of two similarity (or distance) matrices, the correlation of the matrices were evaluated. The DArT markers were subdivided into four different categories depending on their genome origins and similarity matrices calculated. The similarity coefficients from these matrices were then compared pair by pair to form a new correlation matrix. The significance of each correlation was assessed using a randomisation test (with 1000 random permutations). The p-values were calculated and were always < 0.001 (where the null hypothesis is one of zero association). For the Kolmogorov-Smirnov test, the map position data of the DArT and SSR markers on individual chromosomes were fed into GenStat and the p-values were calculated. The Chi-square goodness-of-fit test was carried out to examine the random distribution of DArT markers of different origins across the genome, by calculating the association between the numbers of the markers of different origins the chromosomes.

The genetic marker and linkage map data, along with details and accession numbers for the deposition of raw data, are freely available at the UK WGIN http://www.WGIN.org.uk and Monogram http://www.monogram.ac.uk/services.php as well as the GrainGene http://wheat.pw.usda.gov/GG2/index.shtml websites.