Faba bean (Vicia faba L.) is currently the third most important cool-season food legume in the world. Faba bean provides an important source of dietary protein in human diet, edible oil and animal feeds. Like other grain legumes, faba bean contributes to sustainable agriculture in the management of soil fertility and plays an essential role in crop rotation. Faba bean is a diploid with 2n = 2x = 12 chromosomes 12, is partially cross-pollinated ranging from 4 to 84% 3, and possesses one of the largest genomes among crop legumes (~13000 Mb). The development of saturated linkage maps as well as the identification and map-based isolation of important qualitative traits or quantitative trait loci is therefore complex and expensive.

A saturated genetic linkage map provides an invaluable tool in plant genetic studies and practical breeding. One of the first genetic linkage maps of faba bean was constructed by Van de Ven et al. 4 with only 17 markers; followed by successively more detailed genetic maps by Torres et al. 5 with 51 markers, Satovic et al. 6 with 157 markers, Vaz Patto et al. 7 with 116 markers, Roman et al. 8 with 121 markers, Roman et al. 9 with 192 markers and Avila et al. 10 with 103 markers. To date, faba bean genetic maps have been restricted to morphological, isozyme, RFLP, RAPD, a few seed protein genes and four SSR markers. These markers have been limited either in number, transferability or in their ability to provide syntenic information with other legume species.

The recent generation of abundant genomic and genetic resources focussed around the model species Medicago truncatula and Lotus Japonicus has opened up abundant opportunities for creating gene-based molecular markers that are ideal for genetic mapping in general and comparative mapping in particular 111213141516. In this approach, oligonucleotide primers were designed from sequences of conserved regions in gene exons that flank polymorphic regions such as introns or microsatellites. This PCR-based, codominant marker system has remarkably increased the efficiency of transferring genetic information across species. Examples include the comparison of M. truncatula with alfalfa, pea, chickpea, soybean, mung bean, lentil and lupins 12131415161718.

Comparative genomic studies can expose and confirm phylogenetic relationships among species and determine patterns of chromosomal evolution and syntenic relationships. More importantly, comprehensive comparative genomics can facilitate back-and-forth use of genomic resources between different legumes species, and help to reduce cost and increase efficiency in genetic research as well as crop breeding. The use of conserved genome structure to assist in transferring knowledge among related plant species is well established in grasses 1920 where synteny greatly assists in gene identification among related species.

In this paper we report: (1) the application of gene-based markers in faba bean; (2) the development of the first exclusively gene-based genetic and comparative map for the species; (3) analysis of syntenic relationship between faba bean and M. truncatula; and (4) the levels of homology existing between faba bean, M. truncatula and lentil, a closely related species to faba bean.

The first genetic map of faba bean composed exclusively with gene-based co-dominant molecular markers was constructed using a F₆ RIL population between lines Vf6 and Vf27. The map is also the first to enable the establishment of syntenic relationships between faba bean and the model legume M. truncatula, comparison with other legume species, and integration with genetic maps available in faba bean.

The map is composed of 12 linkage groups and 151 genetic markers. Although the number of chromosomes in faba bean has been reported as 2n = 12 2, the number of linkage groups in recent genetic maps in the species range from 13 to 18 78910 and previously as many as 48 have been reported 6. The high number of linkage groups compared to the number of chromosomes may be due to the fact that faba bean possesses one of the largest genomes among cultivated legumes (~13000 Mb). This compares with other well-characterised species such as M. truncatula, chickpea, soybean, lentil and pea which have genomes of ~450 Mb, ~740 Mb, ~1200 Mb, ~4000 Mb and ~4000 Mb respectively 28.

Of the 24 non-orthologous markers found in this study, eight were from primer pairs where more than one PCR gel band was present and where two or more such amplicons were mapped. In each case, at least one amplicon mapped syntenically. The percentages of markers sequenced in faba bean were lower compared to lentil (63%, 26% and 55% compared to 93%, 69% and 65% for MP, ML and MLG markers respectively, Table 1). This was due to higher proportion of markers amplifying multiple bands in faba bean compared to lentil (data not shown), which may imply duplication. Differences in amplification, sequencing and polymorphism rates among different types of markers used for this study reflect the mode of design of the markers. Since 'MP' and 'MLG' markers were often based on the homology of more than two phylogenetically distant species, they are more likely to work in different legume lineages. The same observation was reported for these primer sets in lentil 15.

Despite the large differences in genome sizes between M. truncatula and V. faba, a simple and direct relationship between the two genomes was identified in this study. Given the number of markers used (151), the syntenic regions cover a large proportion of M. truncatula pseudogenome with 90%, 87%, 66%, 62% and 47% for M. truncatula chromosomes 8, 1, 3, 4 and 5, respectively (Table 3). The appearance of clear isoclinic diagonal lines along the linkage groups in Figure 2 also demonstrates strong evidence for the extensive co-linearity between linkage group pairs of the two species. Similar high levels of conservation have also been reported between L. culinaris ssp. culinaris and M. truncatula 15 and other closely related legumes such as L. culinaris ssp. culinaris and P. sativum 29, M. sativa and P. sativum 30, M. truncatula and P. sativum 17, M. truncatula and M. sativa 11. This study also shows markers originally designed from genes on the same BAC clustered in corresponding syntenic areas in lentil and faba bean. The mapping populations were too small to resolve marker order in lentil and faba bean but extensive conservation of gene order,(and microsynteny) has been shown in previous studies between other legume species at similar or greater phylogenetic distances 1231323334, and to some extent between M. truncatula and Arabidopsis 333435.

A higher level of homology between V. faba and L. culinaris ssp. culinaris compared to that between V. faba and M. truncatula could be inferred from this study based on the common markers mapped in the two genomes, common homology with M. truncatula and similar pattern of rearrangements (Figures 3, 4 and Phan et al., 2007 15). This finding agrees with phylogenetic studies that place the genera Vicia, Lens and Pisum within the tribe Viceae while Medicago and Melilotus form a parallel tribe Trifolieae within the Galegoid or cool season legumes 36, and is consistent with different levels of macrosynteny observed between M. truncatula, P. sativum, V. radiata, G. max, and Phaceolus vulgaris dependent on phylogenetic distance 12. However, chromosomal rearrangements were evident (Figures 2 and 3).

Rearrangements involving Mt6 and Mt3 in particular may explain the differences in chromosome number between the two species (M. truncatula: n = 8; V. faba: n = 6). Mt6 might be considered unusual and is largely composed of heterochromatic DNA 37, contains few transcribed genes 11 and a large proportion of resistance gene analogues 38. In this study no corresponding linkage group was detected in faba bean, as found previously in pea 12 and lupin 1314, together with less than five percent estimated coverage by the L. japonicus genome 34. In faba bean FB5 and FB9 appear to correspond to Mt3. This configuration is supported by a similar pattern in lentil (Figure 4B and Phan et al., 2007 15), although a larger number of markers are needed to confirm this.

The faba bean comparative map constructed here is consistent with the pattern of chromosome conservation previously observed, where different levels of conservation were found to be relatively consistent between M. truncatula and other legume species i.e. high conservation of M. truncatula chromosomes 1, 5 and 8; moderate conservation in the M. truncatula chromosomes 2, 3, 4 and 7 and lowest conservation in the M. truncatula chromosome 6 (Figures 3, 4 and 39). As described above, no homology was identified with M. truncatula chromosome 6 in this study. The alignment of this faba bean map with lentil and the current M. truncatula genome based on M. truncatula genome assembly Mt2.0 is slightly different to that based on an earlier assembly 15. The changes can be observed in Figure 3B where orthologous markers which were syntenic to M. truncatula chromosome 6 in lentil are now co-linear with M. truncatula chromosome 1 in common with faba bean.

Genome studies have demonstrated different factors are responsible for genome size variation and speciation. These include ancient polyploidisation events in the case of the Brassicas 40; segmental or region-specific duplication 41; and genetic rearrangements, transposable element amplification, or combination of different genome modifications 42. Large scale rearrangements, duplications, or polyploidisation were not apparent in this study, possibly as a result of the focus on single locus markers, however differences in non-coding repetitive DNA or transposable elements provide a possible explanation for the large differences in genome size. Retroelements are known to account for substantial proportions of these Viceae genomes as shown by extensive studies in pea, for example 4344, and more recently Vicia 4546. Local genic rearrangements similar to that in found in the grasses (duplications, translocations, and insertions or deletions) may explain multiple PCR amplicons 1920.

The shared macrosynteny among the three species demonstrated here and even higher level of homology between L. culinaris and V. faba will undoubtedly facilitate the identification of markers closely linked to traits of interest in V. faba. Alignment of this map with existing faba bean maps containing important traits with polymorphic SSR markers and/or markers developed in this study, coupled with cross-reference to the abundant genetic information from the Medicago genome sequencing and extensive EST libraries available for the model legume species, will undoubtedly assist this process. As the parental line Vf6 has been used in a number of genetic and QTL mapping projects 679, this map can serve as a central reference map. This study has provided a number of significant outcomes for faba bean genomics and legume genomics in general.

SE and RPO designed the research. HTTP and SE wrote the manuscript. HTTP, SE and MG performed marker polymorphism discovery and population genotyping in Perth. JH assisted with informatics procedures. AMT, CMA and SC-I were involved in developing recombinant inbred lines and marker polymorphism discovery of GLIP markers in Córdoba. All authors read and approved the final manuscript.