Seeds of C. sativa (46 accessions), C. alyssum (3 acc.), C. microcarpa (13 acc.), C. rumelica (2 acc.), C. hispida var. grandiflora (1 acc.) and C. laxa (1 acc.) were kindly provided by the IPK (http://gbis.ipk-gatersleben.de/), USDA (http://www.ars-grin.gov/) and the Arche Noah (Austrian Seed Savers Association, Austria) genebanks (Table 1 and Table 2). All accessions were sown at the end of winter (10 March 2009) in order to define their life form (i.e. winter or spring form).

Genomic DNA was extracted from young leaf tissue of all Camelina accessions using the ‘‘GenElute Plant Genomic DNA Miniprep Kit’’ (SIGMA-Aldrich) in accordance with the manufacturer’s instructions. Intron2 of the β-tubulin gene family (Figure 1), chosen as the most informative molecular marker, was PCR amplified, as already described by Breviario et al. [23] , using 10 ng of template genomic DNA and the following forward and reverse oligonucleotide primers combination: TBPfin2 (5’-GARAAYGCHGAYGARTGYATG-3’) and TBPrin2

(a) Life form as reported in the genebanks; (b) Life form as verified in this work. Unkn= unknown life form or country of origin. ^*IPK, Leibniz Institute of Plant Genetics and Crop Plant Research, Gatersleben, Germany; ^±USDA, United States Department of Agriculture, USA. ^∞Arche Noah, The Austrian Seed Savers Association, Austria; ^{^}URSS, Union of Soviet Republics.

^*IPK Leibniz Institute of Plant Genetics and Crop Plant Research, Gatersleben, Germany. ^±USDA United States Department of Agriculture, USA.

(5’-CRAAVCCBACCATGAARAARTG-3’). For each accession, a total of 3 up to 15 individual plants were analysed. Amplified products were separated on a 6% non-denaturing acrylamide gel and bands were visualized by silver nitrate staining as reported in Breviario et al. [23] . Experiments were independently repeated two or three times, to guarantee the consistency of the genomic profile attributable to each species.

Tubulin nucleotide sequences identification and isolation was carried out as reported in Galasso et al. [17] . For each species only one accession was chosen, selected as the most representative when not the only one available. Selected accessions were: C. sativa CAM134, C. microcarpa CAM47, C. rumelica CAM244, C. hispida PI650133 and C. laxa PI633185. The procedure is briefly described from here to follow. First, the β-tubulin gene family of each accession was PCR amplified from target genomic DNA using the primers combination TBPfex1 (5’- AACTGGGCBAARGGNCAYTAYAC-3’) and TBPrin2 (5’-CRAAVCCBACCATGAARAARTG-3). TBPfex1 anneals at the end of the first exon while TBPrin2 matches the complementary target sequence at the beginning of the third exon (Figure 1). The PCR products resulting from this amplification step were purified and cloned into the pGEM-T easy vector (Promega). About 100 clones for each accession were sequenced in both directions by Macrogen (Seul, Korea). Search on nucleotide sequence homologies was carried out consulting the National Center Biotechnology Information (NCBI) database, with the use of the BLAST algorithm (http://www.ncbi.nlm.nih.gov). A search for tandem repeat motifs and microsatellite sequences was performed using the Tandem Repeats Finder (http://tandem.bu.edu/trf/trf.html) and Sputnik (http://www.cbib.u-bordeaux2.fr/pise/sputnik.html) programs. Nucleotide sequence has been deposited in the EMBL (European Molecular Biology Laboratory) Nucleotide Database with its own specific accession number (from LN811270 to LN811335) (ESM-Table 1). Nucleotide sequences of all the investigated Camelina species were multialigned with the Clustal Omega programme (http://www.ebi.ac.uk/Tools/msa/clustalo/) using default parameters. The evolutionary genetic distance was estimated using Tamura-Nei method and a Maximum Likelihood phylogenetic tree was inferred using the Nearest-Neighbour-Interchange method using MEGA version 5 [26] .

Genomic DNA was extracted from fresh leaf tissue, according to the method reported by Doyle and Doyle [27] . For Southern hybridisation, 5 mg of C. sativa CAM134, C. alyssum CAM21, C. microcarpa CAM47, C. rumelica CAM244, C. hispida PI650133 and C. laxa PI633185 genomic DNA were digested with the restriction enzyme EcoRI, size-separated on a 1% agarose gel and transferred onto a nylon membrane. The DNA sequence of exon2, amplified by PCR using the primers Tubex-2F: 5’-GATTCCAAGTGTGTCACTCGTTG-3’ and Tubex- 2R: 5’-TTACAGCTAGGAGTGGTGAGCTT-3’, designed on the beginning and the end of exon2, was used as probe (Figure 1). Once labelled with a-[³²P]-dCTP, using a random primer DNA labelling kit (Fermentas, Life Sciences), the probe was used for filter hybridisation. Filters were washed in 1.5 mM sodium citrate pH 7.0, 15 mMNaCl, 0.5% SDS (0.1xSSC), at 65˚C, before exposure to X-ray films (Biomax XAR, Kodak).

Flower buds were used for chromosome preparations according to Schwarzacher and Heslop-Harrison [28] . Briefly, after a treatment with 2 mM 8-hydroxyquinoline for 1 h at room temperature followed by 1 hour at 4˚C, the buds were fixed in ethanol-acetic acid (3:1, v/v) and stored at −20˚C until use. Before squashing, flower buds were washed twice for 10 min in 1 x enzyme buffer solution (0.01 M citric acid-sodium citrate buffer, pH 4.8) and then digested at 37˚C for 45 min with an enzyme solution containing 1% (w/v) cellulase (Calbiochem), 1% (w/v) Onozuka R-10 cellulase (Serva) and 20% (v/v) pectinase (Sigma). Chromosome preparations were stained with DAPI (4’, 6-diamidino-2-phenylindole) and counted under a fluorescence microscope (Axiovert 200, Zeiss) in twenty complete metaphases of somatic cells of C. rumelica, C. hispida, C. laxa and Camelina spp.

In order to verify their life form, all Camelina accessions (Table 1 and Table 2) were sown in small plots of 1 square meter in an open field at the end of winter. Among the 46 C. sativa accessions analysed, 39 showed a spring life form (spring type) because they started stem elongation after 44 - 50 days from sowing. Accordingly, they have all been classified as C. sativa ssp. sativa. On the contrary, the remaining 7 accessions (CAM31, CAM37, CAM76, D9952, CAM132, PI650168 and PI650167) showed a winter life form (winter type) since they remained in a prolonged vegetative stage, producing many leaves, with no stem elongation. In fact, these accessions require a vernalization period to start stem elongation and flowering. According to Plessers et al. [21] these seven accessions can thus be classified as C. sativa ssp. pilosa. With reference to the C. sativa accessions (CAM269, CAM45, CAM46, PI650168, PI650146, PI650142, FF084, FF006, FF004 and PI650167), obtained from the germplasm banks as C. sativa ssp., they all turned out to be of the spring type with the exception of PI650168 and PI650167 that resulted winter types. Finally, all the accessions of the species C. microcarpa, C. rumelica, C. hispida and C. laxa (Table 2) exhibited a strict winter life form while C. alyssum showed both winter (CAM176) and spring types (PI650132 and CAM21).

The cTBP method was applied to all the Camelina species and accessions listed in Table 1 and Table 2. cTBP amplifies the genomic region that contains only the second intron of the different members of the β-tubulin gene family. After PCR amplification, performed with the combination of primers TBPfin2 and TBPrin2 (Figure 1), the products were analysed on non-denaturing acrylamide gels (Figure 2 and Figure 3). The cTBP profiles detected for each species are described in more detail from here to follow.

C. sativa: forty-six accessions of different origin were genotyped using cTBP. With the exception of PI650167, all C. sativa accessions showed a very similar cTBP pattern with few polymorphic bands detected at about 500 bp, 750 bp and 800 bp (Figure 2). These polymorphisms were found in both C. sativa ssp. sativa and ssp. pilosa. Accession PI650167 showed a cTBP profile remarkably different compared to all the others C. sativa accessions, strongly resembling that of the majority of the accessions of C. microcarpa (such as PI633186 or CAM60, see Figure 3). This suggests that accession PI650167 has most likely been misclassified.

C. alyssum: all three accessions showed a cTBP fingerprint substantially identical to C. sativa (Figure 3). As observed in C. sativa also the three C. alyssum accessions show a polymorphic band around 500 bp sizes. This band appears slightly higher in CAM21 and PI650132 respect to CAM176 (Figure 3).

C. microcarpa: thirteen accessions with different provenience were analysed (Table 2). According to the cTBP profiles, some of the analysed C. microcarpa accessions appear to have been misclassified. More spec- ifically, accessions PI633191 and PI633134 revealed a cTBP profile that is, respectively, much more similar to C. sativa and C. rumelica than to C. microcarpa (Figure 3), whereas the accession PI650135 showed a unique cTBP profile, very different from any other Camelina species. Besides these three, all the remaining C. micro- carpa accessions shared a very similar cTBP profile, characterized once more by the presence of a pronounced DNA polymorphism around 500 bp. The sizes of this polymorphic band is of 550 bp in the majority of the C. microcarpa accessions, but PI633186, CAM60 and one individual of accession CAM71 showed a lower size fragment (Figure 3). Comparison of the cTBP profile of the majority of the C. microcarpa accessions with that of C. sativa and C. alyssum, reveal many bands in common. This suggests that these three species might share one common progenitor, at the least (Figure 3). Accession PI650135, of French origin (Table 2) stored at the IPK genebank as C. microcarpa, turned out to be a mixture of genetically distinguishable individuals. In fact, cTBP analysis carried out on 15 single individual plants demonstrated that only 9 showed a cTBP profile characteristic of C. microcarpa (data not shown), while 6 individual plants showed a distinct profile with a low number of cTBP bands, as that shown in Figure 3. Because of the distinctiveness of their profile we will refer to this group of samples as Camelina spp. In addition, the low number of cTBP amplified bands is suggestive of a diploid status.

C. rumelica: two accessions were analysed, with origin from Iran and Union of Soviet Republics (IPK genebank), respectively. Both showed a very similar cTBP pattern with no evidence for polymorphic bands (Figure 3).

C. hispida: only one accession of this species could be recovered from the genebanks. This accession resulted contaminated with seeds of other species, in particular C. microcarpa. (Figure 4(A)). Once cleaned from the

contaminating seeds, C. hispida was nevertheless characterised by a high number of intra-accession polymor- phic bands. This is in contrast with the other Camelina species which showed very few polymorphic bands between the single individuals of the same accession.

C. laxa: only one accession of this species was retrievable from the genebanks. Four single plants were analysed, showing some cTBP polymorphic bands (Figure 4(B)). Camelina laxa showed an intra-accession varia- bility lower than C. hispida but higher than that of the other Camelina species.

In order to isolate and sequence the complete β-tubulin gene family from each Camelina species only one accession was chosen and analysed. The accessions selected were those that showed the most representative cTBP profile of the species. In accordance, the β-tubulin gene families from C. microcarpa CAM47, C. rumelica CAM244, C. hispida PI650133 (individual number 2), C. Laxa PI6331858 (individual number 2) and a sample of Camelina spp., were PCR amplified and cloned. Nucleotide sequences of the C. sativa (CAM134) β-tubulin gene family, already available at the EMBL Database [17] were retrieved for sequence comparison and phy- logenetic analysis. By using the forward TBPfex1 and the reverse TBPrin2 primers, located at the end of the first exon and at the beginning of the third exon (Figure 1) respectively, the following numbers of partial β-tubulin gene sequences were isolated from the different Camelina species: 19 from genomic DNA of C. microcarpa (CmTUB1..CmTUB19), 15 from C. rumelica (CrTUB1..CrTUB15), 12 from C. hispida (ChTUB1..ChTUB12), 11 from C. laxa (ClTUB1..ClTUB11) and 9 from Camelina spp. (CameTUB1..CameTUB9). As predicted, each of the camelina β-tubulin nucleotide sequence contained 97 bp of the coding exon1 (partial), 270 bp of exon 2 (complete) and 147 bp of exon3 (partial) together with the two full-length introns, 1 and 2 (ESM-Table 1). Intron length varied in all the β-tubulin isotypes of the Camelina species. The shortest (80 bp) and the longest (936 bp) intron1 nucleotide sequence were both found in Camelina spp. (CameTUB1 and CameTUB9, respectively), whereas the shortest (82 bp) and longest (592 bp) intron2 sequences were detected in C. laxa (ClTUB3) and C. hispida (ChTUB12), respectively (ESM-Table 1). A search for tandem repeat motifs and microsatellite sequences revealed the presence in all species, with the exception of C. laxa, of one direct head-to-tail tandem repeat located in either the intron1 or the intron2 of some β-tubulin isotypes. Camelina microcarpa is the only species that showed the presence of one tandem repeat motif in both introns (CmTUB19). The tandem repeat located in intron2 was the longest detected (189 bp) (ESM-Table 1). Microsatellites such as trinucleotides (CTT)₆ and (CTT)₇ and dinucleotide stretches (TA)₅ and (TA)₆ were detected in introns 1 and 2, respectively. Compound microsatellites (TA)₅ (TC)₃ and (TA)₇ (TC)₃ were observed only in intron1 of Camelina spp. CameTUB9 and C. laxa ClTUB11, respectively. The multialigned nucleotide sequences of all the camelina and Arabidopsis thaliana β-tubulins were used to infer a Maximum Likelihood phylogenetic tree. As shown in Figure 5, the 95 different β-tubulin genes (86 from Camelina and 9 from A. thaliana) are grouped in two main clusters (I and II) which are further subdivided in sub-clusters (A, B, C and D). With the exception of the sub-cluster D, all the others are further subdivided in two groups (A1-A2; B1-B2; C1-C2). The phylogenetic tree spreads the analysed sequences throughout the clusters, sub-clusters and groups without distinction between Camelina species and A. thaliana. The only exception is represented by group C2 that contains only β-tubulins from Camelina species (Figure 5). Furthermore, each group includes at least one β-tubulin sequence per analysed species, except B1 that do not contain any β-tubulin of C. hispida and C. microcarpa.

To obtain additional information on the genomic organization and the number of the β-tubulin genes present in the different Camelina species, we performed a Southern blot analysis on the EcoRI restricted genomic DNA of C. sativa, C. alyssum, C. microcarpa, C. rumelica, C. hispida, C. laxa and Camelina spp. A radioactively-labelled exon2 fragment was used as a probe. EcoRI endonuclease was chosen since no restriction sites for this enzyme are present in any of the exon2 sequences of all β-tubulins. This implies that each of the detected hybridization fragments should correspond to one gene. As shown in Figure 6 the number of hybridization bands is higher in C. sativa, C. alyssum, C. microcarpa and C. rumelica compared to that of C. hispida, C. laxa and Camelina spp. This is in agreement with the number of the isolated β-tubulin genes that was lower in C. hispida, C. laxa and Camelina spp. than in the other species. As already observed comparing the cTBP profile of C. sativa and C. alyssum (see Figure 3) also the Southern blot hybridization showed very few differences between these two species (Figure 6).

The chromosome number resulted 2n = 14 in C. hispida and 2n = 12 in C. laxa and Camelina spp. (Figure 7) suggesting a diploid status of these three species respect to C. sativa, C. alyssum and C. microcarpa, that are all polyploids with a chromosome number 2n = 40 [14] [29] , and to C. rumelica characterized by a chromosome number 2n = 26 (Figure 7).