Construction of two genetic linkage maps in cultivated tetraploid alfalfa (Medicago sativa) using microsatellite and AFLP markers

Background Alfalfa (Medicago sativa) is a major forage crop. The genetic progress is slow in this legume species because of its autotetraploidy and allogamy. The genetic structure of this species makes the construction of genetic maps difficult. To reach this objective, and to be able to detect QTLs in segregating populations, we used the available codominant microsatellite markers (SSRs), most of them identified in the model legume Medicago truncatula from EST database. A genetic map was constructed with AFLP and SSR markers using specific mapping procedures for autotetraploids. The tetrasomic inheritance was analysed in an alfalfa mapping population. Results We have demonstrated that 80% of primer pairs defined on each side of SSR motifs in M. truncatula EST database amplify with the alfalfa DNA. Using a F1 mapping population of 168 individuals produced from the cross of 2 heterozygous parental plants from Magali and Mercedes cultivars, we obtained 599 AFLP markers and 107 SSR loci. All but 3 SSR loci showed a clear tetrasomic inheritance. For most of the SSR loci, the double-reduction was not significant. For the other loci no specific genotypes were produced, so the significant double-reduction could arise from segregation distortion. For each parent, the genetic map contained 8 groups of four homologous chromosomes. The lengths of the maps were 2649 and 3045 cM, with an average distance of 7.6 and 9.0 cM between markers, for Magali and Mercedes parents, respectively. Using only the SSR markers, we built a composite map covering 709 cM. Conclusions Compared to diploid alfalfa genetic maps, our maps cover about 88–100% of the genome and are close to saturation. The inheritance of the codominant markers (SSR) and the pattern of linkage repulsions between markers within each homology group are consistent with the hypothesis of a tetrasomic meiosis in alfalfa. Except for 2 out of 107 SSR markers, we found a similar order of markers on the chromosomes between the tetraploid alfalfa and M. truncatula genomes indicating a high level of colinearity between these two species. These maps will be a valuable tool for alfalfa breeding and are being used to locate QTLs.


Background
Alfalfa (Medicago sativa L.) is the most cultivated forage legume, with about 32 millions hectares over the world [1]. Its agronomical interest is based on its high protein content, suitable feeding value and favourable environmental balance (perenniality and no nitrogen fertilizer required). However the genetic progress in this species is slow because of its autotetraploidy [2], with 2n = 4x = 32 [3,4], its allogamy and the synthetic structure of the varieties. The development of a genetic linkage map would be a significant breakthrough, either to map candidate genes, to search for QTL of agronomical traits, or to make marker-assisted selection. Such analysis requires working on cultivated material, either in terms of genetic background or ploidy level.
In the past years, low density linkage maps were obtained on diploid forms of the Medicago sativa species complex [5][6][7]. The parental plants originated from CADL plants (Cultivated Alfalfa at the Diploid Level, diploid plants produced from cultivated tetraploid alfalfa) [6] or from natural populations, the crosses being made at the subspecific level [5,7]. All these maps comprised from 89 to 153 RFLP, RAPD or isozymes markers, and spanned from 467.5 cM [7] to 659 cM [6]. A denser map based on a cross between M. coerulea and M. quasifalcata, was published in 2000 [8] and covered 754 cM for a haploid genome, with 868 markers (mainly RAPDs and RFLPs). A linkage map of tetraploid alfalfa was constructed with 82 RFLP markers, using two backcross populations of 101 individuals, and covered about 470 cM, with 7 homology groups only [9]. The number of polymorphic RFLP markers is clearly a limiting factor to the development of a saturated map at the tetraploid level in alfalfa.
Until recently, mapping procedures in autopolyploids plants were based on SDRF (single-dose restriction fragments) segregation in mapping populations analysed as pseudo-test crosses [10,11]. Genetic maps were obtained in autotetraploid species such as potato [12,13], alfalfa [9], tea [14], rose [15], birdsfoot trefoil [16] and in autooctoploid species such as sugar cane [17][18][19] and strawberry [20]. Recently, theoretical studies were published to analyse gene segregation and to perform the mapping in autotetraploid species [21][22][23][24] or even autooctoploid species [25]. All types of markers can be used: a dominant marker present in one parent in single dose (simplex or SDRF) or double dose (duplex), a dominant marker present in both parents (double simplex), and codominant markers (Figure 1). It is now possible to develop a map in autotetraploid species, based on all codominant or dominant markers, and to calculate linkage between pairs of markers, either in coupling phase or in repulsion phase. The software TetraploidMap, released in 2002 [26] is suited for this type of calculation. A method aiming at ordering the markers into a linkage map of autotetraploid species using simulating annealing was also proposed [27], but no specific software is presently available.
In autopolyploid species, two facts are usually unknown: (1) does the whole genome show a tetrasomic inheritance, or do some parts of the genome have a disomic inheritance ? (2) what is the frequency of double-reduction ? Indeed, two sister chromatids could segregate in the same gamete after formation of a tetravalent during meiosis. Consequently, double-reduction creates new types of gametes and modifies the probabilities of each type of gametes in a segregating population [28][29][30]. Its frequency can be calculated by TetraploidMap software on codominant markers.
In this article, we report the construction of a genetic map for each of the parents of an F1 population obtained by crossing one individual from each of the Magali and Mercedes alfalfa cultivars and using the mass-revealed dominant AFLP markers together with codominant microsatellite (SSR) markers. We have used a set of SSRs Marker types in an autotetraploid genotype   [31,32]. A genetic map was recently developed using an F2 population [33] and preliminary evidence was given that diploid M. sativa and M. truncatula genetic maps were aligned. Considering the phylogenetic proximity of M. truncatula and alfalfa, we used these EST-based SSR, and some other SSR markers published in M. truncatula ( [34], Santoni et al. unpubl.) or M. sativa [35], to build alfalfa linkage maps and to align all genetic maps.
Based on 599 AFLP and 107 SSR markers, we identified, for both parents, 8 groups of 4 homologous chromosomes each. The total length of the maps covered between 88 to 100% of the genome. We have numbered the homology groups of alfalfa as for M. truncatula [8,33].

Mapping population
Alfalfa is an allogamous species with a strong inbreeding depression. Consequently, plants are highly heterozygous and selfing gives rise to either self-sterility or lethal allelic combinations or very "weak" plants. Homozygous lines cannot be obtained and F2 populations may suffer from a genetic bias induced by the death of some genotypes. So our mapping population is a set of 168 F1 individuals from the cross between Mercedes4.11 used as female parent and Magali2 used as male parent.

AFLP markers
Using 27 AFLP primer pairs, a total of 599 polymorphic markers were scored on the mapping population ( Table  1). Most of them segregated as simplex (Figure 1), but a large number of markers had other segregations, mainly duplex or double-simplex. Among all AFLP markers, the percentage of distorted markers reached 34.8%.

SSR markers
Transfer of SSR to alfalfa The SSR primer pairs mainly originated from M. truncatula: 35 among those developed by Santoni et al. ( [34] and unpubl.), 1 published by Diwan et al. [35] and 187 obtained by Huguet et al. (unpubl. data). Nine SSR primer pairs were developed from an alfalfa genomic library [35] and tested on diploid and tetraploid alfalfa.
The amplification and polymorphism among the 2 parents were tested for these 232 primer pairs (Table 2). Only 19% of the primer pairs (44) gave no amplification. Over the 81% of primer pairs (187) that gave amplification, 34% (78) were monomorphic and 47% (110) were polymorphic between Magali2 and Mercedes4.1. So far, only 87 pairs have been used for genotyping in the mapping population.

Inheritance of SSR markers
With the 87 primer pairs (Table 3), 318 alleles (= bands) were scored, 284 of them being polymorphic in the mapping population. The parental genotypes were determined from the segregation of alleles at each locus using the "Findgeno" procedure from TetraploidMap software. The monomorphic bands were excluded from segregation analyses as a monomorphic allele is present at least in triplex dose in a parent. For 43 loci, the parental genotypes were determined, and the coefficient of double reduction (α) was not significantly different from 0 (Table 4). For 20 loci, α was significantly different from 0 but the genotypes of the parents were also found under the absence of double reduction (Table 4). For 24 loci (Tables 5, 6), the parental genotypes could not be determined. The amplification of more than one locus was suspected from the visual reading of the gels (a large gap between groups of alleles was interpreted as the presence of 2 different loci, Figure 2) and from the segregation of alleles in the mapping population (i.e. the presence of individuals with more than 4 bands, or individuals with 3 alleles from the same parent). From 21 primer pairs, 41 polymorphic loci were identified, and their segregation was analysed ( Table  5). The genotypes of the parents were determined, with double reduction for only 6 loci. Finally for 3 primer pairs (Table 6), a multilocus segregation was evidenced, but it was not possible to group the alleles in loci with an expected segregation. For these 3 primer pairs, the bands were treated as dominant markers to determine the parental genotypes (Table 6). For the loci that showed significant double-reduction (in Tables 4 and 5), all the F1 genotypes could have been formed without double-reduction. So the segregation of these markers in F1 is not different from segregation distortion.
Over the 104 polymorphic SSR loci with codominant segregation, the average number of alleles per locus in both

Map construction
All markers (simplex, duplex, double-simplex AFLPs and multi-allele SSR loci) except the distorted double-simplex AFLP markers were used to build maps. The map construction was performed in 3 steps. First, Tetraploid-Map calculated a χ 2 test of independence between markers, which led to 8 homology groups for each parent. Then within each homology group, TetraploidMap determined the linkage phase, recombination rate and LOD score between markers, whatever their dose. Linkage groups (i.e. chromosomes) within homology groups were manually identified using the linkage phase information. In both Mercedes4.11 and Magali2, for each homology group, 4 linkage groups were found, except for homology group 5 in Magali2 which had only 3 linkage groups. Finally, for each linkage group, the recombination rates and LOD scores were entered in the JoinMap software [36]. The order of the markers was calculated and the maps were drawn. For Mercedes4.11 (Figure 3), the map length was of 3045 cM, with 339 marker positions and an average distance between markers of 9.0 cM. The map of Magali2 ( Figure 4) covered 2649 cM, with 350 marker positions and an average distance between markers of 7.6 cM. In both maps, the distorted markers were spread over the genome.
The repulsion linkages were distributed among each pair of chromosomes within one homology group, and this was observed for each homology group (figure 5, example for homology group 8 in Magali2).

Composite map
The large number of polymorphic SSR loci made it possible to build a composite map, based on the recombination rate and LOD score for each pair of SSR markers in each homology group. Instead of 8 chromosomes per homology group for both parents, a single linkage group was obtained. The total length of the map (figure 6) reached 709 cM, with 107 loci. Conflicting linkages occurred in group 4, so this chromosome was split in 3 parts. The average distance between markers was 6.6 cM.

Discussion
The development of a dense genetic linkage map in cultivated tetraploid alfalfa is the first step in understanding the genetic control (QTLs) of traits of agronomic interest. Agronomic traits are not well evaluated in wild M. sativa because of its prostrate growth habit, so the dense map obtained in wild diploid M. sativa [8] can hardly be used for QTL detection of agronomic traits. As in other heterozygous species, and especially in autotetraploid ones [21], we used a F1 population for mapping. Indeed, even if a F2 generation could be a valuable tool in terms of number of recombination events, a selfing generation could lead to genetic skewing through the death of some inbred plants.
We chose easy to use handy PCR markers: AFLPs and SSRs. AFLPs, because they are mass-revealed, help in covering the whole genome. SSRs mainly originated from EST databases of M. truncatula, are codominant, locus-specific and portable. They were previously mapped in this model species, and thus allow the analysis of synteny between the two species, alfalfa and M. truncatula.
Interestingly, the two parents of the mapping population shared the same characteristics: the number and proportion of simplex, duplex, double-simplex and distorted AFLP markers and the number of SSR alleles (1.8 alleles per locus and per parent).
A large number of null alleles were found with SSRs, possibly arising from mutations in alfalfa compared to M. truncatula in the sequences that match the SSR primers. The number of null alleles using SSR primers directly developed in alfalfa should be lower. Indeed, there was only one null allele in this mapping population for each of both SSR markers developed in alfalfa (AFct32 and AFct45). In Prunus [38] and sugar beet [39] a large number of null alleles were also identified when using SSR markers from different species.
Some SSR primer pairs (21 over 87) gave multi-locus amplification in alfalfa; this was also observed in soybean [37], Prunus sp [38], Brassica sp [39], sorghum [41], mungbean and lablab [42]. Multi-locus amplification was evidenced from allele segregation in the mapping population. This can result from the duplication of zones of genome during the evolution between M. truncatula and M. sativa.

Mapping in alfalfa
Using the AFLP and SSR markers, distorted or not, we identified 8 groups of homologous chromosomes in each parent, each but one group with 4 chromosomes. The homologous status of the chromosomes was established by the codominant SSR loci and the repulsion linkages.
We obtained longer and more saturated maps with AFLPs and SSRs (662 and 761 cM for the haploid genome in Magali2 and Mercedes4.11, respectively) in the tetraploid cultivated alfalfa than in the previously published research with only RFLPs (470 cM for the haploid genome) [9]. Furthermore our map lengths were close to what was obtained in diploid alfalfa [8], with 754 cM.
We propose a portable reference map built with only SSR loci. It spans 709 cM for the haploid genome, covering 94% of the genome of diploid alfalfa [8]. The average distance between markers is 6.6 cM. A limited number of SSR MTIC258  CACCACCTTCACCTAAGAAA  TGAAATTCACATCAACTGGA  1  H  MTIC272  AGGTGGATGGAGAGAGTCA  TCATGAATAGTGGCACTCAA  3  H  MTIC273  TGTTAGCAACTTTGTGATGG  TCCATTACAATACCCAGAGG  7  H  MTIC278  CTTACCCTCCACTGCTACTG  CGCATATAACAGAGGGTTTC  2  H  MTIC289  GCTGGTGTCAAAGAGGTCTA  AGATTTGAAACTGCCCTACA  H  MTIC299  AGGCTGTTGTTACACCTTTG  TCAAATGCTTAAATGACAAAT  8  H  MTIC304  TTGGGCTTAATTTGACTGAT  AGCGTAAAGTAAAACCCTTTC  2  H  MTIC314  TCTAATCCCAACAACACTCTT  GAAGAAGAAGCCATAGTGTGA  1  H  MTIC318  TCAACCAACTCAATGCCACT  TTGTTGTGAAATGGAAAATGG  7  H  MTIC326 GATCACCCTTTATGGAGTTTGAA  The dense maps including AFLPs will be valuable for an accurate QTL analysis in our population. The composite SSR map will be useful to compare QTL position in different populations.
As in published diploid and tetraploid alfalfa maps [43], we have found that AFct32 and MTLEC2A markers were mapped on the same chromosome (group 3) and AFct45 on another chromosome (group 8).

Autotetraploidy in alfalfa
Codominant markers are adequate tools to test the autotetraploid structure of the genome [22]. This study is the first demonstration of the tetrasomic inheritance of chromosomes in alfalfa, using SSR markers. This autotetraploidy pattern is confirmed by the repulsion linkages between markers belonging to all homologous chromosomes. Recently in alfalfa, a preferential chromosome pairing at meiosis was found with RFLP markers in a F1 population [44]. We can hypothesize that this result originates from a choice of mapping population parents that are genetically distant (a non-dormant genotype of M. sativa and a very dormant genotype related to M. falcata).
In our study, some loci showed a significant double reduction frequency. Nevertheless, genotypes only originating from double-reduction were not found. The doublereduction obtained here is not different from a segregation distortion. The rate of double-reduction in alfalfa is presumably low, so its detection would require the   analysis of a large population. This result is in accordance with the observation of meiosis in alfalfa, which shows a low proportion of tetravalent formation [45]. Nevertheless, in mapping procedures, double reduction cannot be taken into account [23].

Segregation distortion
The segregation distortion reached 35% over the AFLP markers. The double-simplex and the duplex were more frequently distorted than the simplex markers. For the SSRs, considering the loci showing significant double reduction as distorted loci, the proportion of distortion was 25% (26 loci over 104). These rates of distortion are similar to that obtained in M. truncatula [33] or in diploid alfalfa [5][6][7]. But in tetraploid alfalfa, Brouwer and Osborn [9] obtained a segregation distortion of only 5.5% for the simplex RFLP alleles, and stated that the autotetraploidy could reduce the effect of deleterious alleles that cause distorted segregation. However, the restriction of the segregation distortion analysis to the simplex alleles could result in an optimistic view of the distortion. The random repartition of the distorted markers over our maps showed that no alfalfa genome region was particularly distorted contrarily to the situation in the M. truncatula genome [33].

Transfer of SSR markers
More than 80% of the SSR markers initially developed in the legume model species M. truncatula gave amplification products in alfalfa. This capacity to give rise to amplification in these species is not surprising since they belong to the same genus. Indeed, such transfers have already been reported across annual and perennial species of Medicago genus for 4 markers [43] and very recently for 89% of a set of 455 primer pairs developed from EST of M. truncatula [46]. The transfer of SSR markers among species within some genera has already been obtained in Glycine sp [37], Prunus sp [38] and Brassica sp [39], in the three cases with SSRs isolated from genomic DNA. The high level of SSR transferability allows the use of a large number of SSRs from public M. truncatula EST on alfalfa.

Comparison with M. truncatula map
Over the 94 SSR loci mapped on both M. sativa and M. truncatula, all but 2 were gathered in the same groups (T. Huguet, unpublished data) suggesting that no major chromosome rearrangement had occurred since the species differentiation. Unexpected mapping was obtained for (1) MTIC272 was mapped in group 3 of M. truncatula but group 6 in Magali2 and (2)  The order of SSR markers along chromosomes is very well conserved between tetraploid alfalfa and M. truncatula (T. Huguet, unpubl. data). The cases of misalignment can be explained by a lack of accuracy in the recombination rate estimation. Indeed a recombination rate has a standard error which varies with the size of the mapping population, the recombination rate and the type and phase of the pairs of markers [21]. In autotetraploid species, for a F1 mapping population of 168 individuals, the standard Amplification pattern obtained with one primer pair (MTIC134), showing 2 loci We were surprised to note that the haploid genome map length in a F2 population of M. truncatula was about 2 times larger than our map on tetraploid alfalfa (1225 vs 709 cM), although DNA content of haploid genome was 1.5 times less in M. truncatula than in tetraploid alfalfa [47].  Group 2

Conclusions
We have obtained genetic linkage maps of cultivated tetraploid alfalfa using a combination of AFLP and SSR markers, with length and marker density close to saturation. Their comparison to M. truncatula maps through the common SSR markers reflects the synteny between these two species. The inheritance of the codominant SSR markers indicates a true autotetraploidy of alfalfa, with complete random pairing of chromosomes at meiosis. The orders of the markers on the chromosomes are very similar. The use of such maps to identify QTLs for agronomical traits is in progress in our laboratory.

Mapping population
The two parental plants were selected within the numerous alfalfa varieties bred in France. One originated from the Provence type cultivar Magali, and was named Magali2. This genotype was previously used in a diallel crossing design [48] and has high general combining ability for forage digestibility and plant height. The other plant, Mercedes4.11, originates from the high yielding Mercedes cultivar, and was kindly provided by B. Bayle, from Limagrain Genetics company. It was selected after 2 cycles of selection for increased disease (Verticillium alboatrum), pest (Ditylenchus dipsaci) and lodging resistance. In year 1999, the two plants were manually crossed, with Mercedes4.11 as the female parent. The flowers were castrated before stamina dehiscence, and the style was cleaned with alcohol at 70% to remove any pollen grains. The pollen of Magali2 flowers was harvested on a brush and applied on the style of female plant. A total of 236 seeds was harvested. In year 2000, a subset of 168 plants out of 230 germinated seeds, was randomly taken to represent the F1 mapping population.

DNA extraction
Young leaflets were harvested on each plant, and DNA was extracted using a standard CTAB (hexadecylatrimethylammoniumbromide) procedure [49].

AFLP markers
AFLP analysis [50] was performed with the Gibco Brl kit AFLP™ Analysis System I, as recommended by the manufacturer, with some modifications designed to optimize the readability of the gels. Pre-amplification products were diluted 1:10 in water instead of 1:50. For some primer pairs, selective amplifications were performed with 33 P-labeled EcoRI selective primers as recommended by Gibco Brl. Reaction products were separated on 6% denaturing polyacrylamide sequencing gel with a 1X TBE buffer with 7 M urea. Electrophoresis was performed at 85 W, for approximately 3 h 30 min, on a 38 × 53 cm gel apparatus. The gel was subsequently placed for 3 h at 75°C in a gel dryer and exposed to a Kodak Biomax MR X-ray film for 2-5 days.
For primer pairs analyzed without 5'end-labeled EcoRI primer, products from selective amplifications were separated on denaturing polyacrylamide sequencing gels as above and visualized using a silver staining method [51].
Each AFLP marker was identified by the last 2 nucleotides of the two primers, followed by a number according to the size of the band. Only polymorphic bands were scored. Three cases of polymorphism were considered: bands present in Magali2 and absent in Mercedes4.11, bands absent in Magali2 and present in Mercedes4.11, and Genetic linkage map of Magali2, homology groups 1 to 4 Figure 5 Genetic linkage map of Magali2, homology groups 1 to 4. Magali2 is the male parent of a F1 mapping population in tetraploid alfalfa. See legend in Figure 3. In group 4, the linkage group numbered "3/4" indicated that it was linked to both chromosomes 3 and 4.  Genetic linkage map of Magali2, homology groups 5 to 8 Figure 6 Genetic linkage map of Magali2, homology groups 5 to 8. See legend in Figure 3.  other PCR conditions were the same as for silver staining. The amplification products were separated on 6.5% denaturing polyacrylamide gel (LI-COR) as recommended by LI-COR.
From test experiments on the 2 parental plants plus 2 F1 plants, we classified the markers as showing no amplification (or a non specific amplification showing a multibanding profile), showing amplification but with a monomorphic profile, and showing amplification with polymorphic profiles. The whole mapping population was then genotyped for 87 of these SSR loci.

Marker detection and inheritance
We used the "Findgeno" subroutine from the Tetraploid-Map software [26]. It gives the segregation type for each marker. The distorted AFLP markers were thus identified. For the SSR markers, the autotetraploid inheritance was tested. The software identifies the parental genotypes, from the phenotypes of the parents and of the F1 plants at each locus, without or with double reduction. For 24 loci, the number of alleles exceeded 4 for some F1 individuals, and the gels patterns clearly indicated that 2 or 3 loci were amplified ( Figure 2). In these cases, several loci were scored for each primer pair and the segregation was analysed as described above. The locus names were thus composed of the name of the primer pair followed by a number (i.e. MTIC84-1 and MTIC84-2). When the parental genotype identification failed, the alleles amplified by the primer pairs were scored as dominant markers.

Map construction
The "Cluster" subroutine of the TetraploidMap software was used to create groups of markers that do not segregate independently. We wanted to obtain 8 groups with many markers, according to the number of base of chromosomes in alfalfa. We had to ask for 9 to 11 groups, as 1 to 3 of the groups had very few markers, and so had to be discarded. All the markers were included, distorted or not, except the AFLP markers present in both parents that did not segregate in 3:1 ratio.
Then for each homology group, the linkage for each pair of markers was analyzed ("Twopoint" subroutine). The markers can be unlinked, linked in a coupling phase or linked in a repulsion phase. A linkage was considered only if the LOD score was above 2. From the linkage analysis, the markers that composed each linkage group (chromosome) in each homology group were identified. The Twopoint procedure produced an output file containing the recombination rate and the LOD of each pair of markers. The order of the markers on each chromosome was determined using JoinMap 3.0 software [36], using the ouput file of the Twopoint procedure. We used the Kosambi distance, and the default parameters of JoinMap.
The alfalfa map was accepted if the χ 2 test in JoinMap was low, and so, some AFLP markers were excluded.
A composite map was built with only the SSR loci. The SSR previously assigned to the groups of homologous chromosomes, and the recombination rates between pairs of loci were used to build the map of each group using JoinMap. Three parameters of JoinMap were changed: "the number of maximum linkages to show per locus" was decreased from 2 to 1, "use linkage with REC smaller than" was changed from 0.40 to 0.49 and "LOD larger than" was changed from 1 to 0.5.
The cover of the maps was calculated by the ratio between the length of our maps and the length of the dense map of diploid alfalfa [8] (754 cM).

Author's contributions
BJ carried out the statistical and mapping analyses and drafted the manuscript, SF carried out most molecular analyses, TH, GC and SS provided information on M. truncatula SSR markers and contributed to the comparisons of alfalfa and M. truncatula maps, PB gave technical advice and contributed to the design of the study, CH coordinated the study and contributed to the discussion. All authors read and approved the manuscript.
Repulsion linkages in the homology group 8 of Magali2 parent Figure 7 Repulsion linkages in the homology group 8 of Magali2 parent. The red lines indicate alleles of a common SSR locus or duplex AFLP markers, and the blue lines show the repulsion linkages between markers. Composite map with the SSR markers, for a F1 mapping population in tetraploid alfalfa Figure 8 Composite map with the SSR markers, for a F1 mapping population in tetraploid alfalfa.