Karyotypic evolution of the Medicago complex: sativa-caerulea-falcata inferred from comparative cytogenetic analysis

Background Polyploidy plays an important role in the adaptation and speciation of plants. The alteration of karyotype is a significant event during polyploidy formation. The Medicago sativa complex includes both diploid (2n = 2× = 16) and tetraploid (2n = 2× = 32) subspecies. The tetraploid M. ssp. sativa was regarded as having a simple autopolyploid origin from diploid ssp. caerulea, whereas the autopolyploid origin of tetraploid ssp. falcata from diploid form ssp. falcata is still in doubt. In this study, detailed comparative cytogenetic analysis between diploid to tetraploid species, as well as genomic affinity across different species in the M. sativa complex, were conducted based on comparative mapping of 11 repeated DNA sequences and two rDNA sequences by a fluorescence in situ hybridization (FISH) technique. Results FISH patterns of the repeats in diploid subspecies caerulea were highly similar to those in tetraploid subspecies sativa. Distinctly different FISH patterns were first observed in diploid ssp. falcata, with only centromeric hybridizations using centromeric and multiple region repeats and a few subtelomeric hybridizations using subtelomeric repeats. Tetraploid subspecies falcata was unexpectedly found to possess a highly variable karyotype, which agreed with neither diploid ssp. falcata nor ssp. sativa. Reconstruction of chromosome-doubling process of diploid ssp. caerulea showed that chromosome changes have occurred during polyploidization process. Conclusions The comparative cytogenetic results provide reliable evidence that diploid subspecies caerulea is the direct progenitor of tetraploid subspecies sativa. And autotetraploid ssp. sativa has been suggested to undergo a partial diploidization by the progressive accumulation of chromosome structural rearrangements during evolution. However, the tetraploid subspecies falcata is far from a simple autopolyploid from diploid subspecies falcata although no obvious morphological change was observed between these two subspecies. Electronic supplementary material The online version of this article (doi:10.1186/s12862-017-0951-x) contains supplementary material, which is available to authorized users.


Background
Polyploidy is very common in plant evolution. It plays an important role in adaptation and speciation of plants [1]. According to different chromosome set origins, polyploidy is generally classified into autopolyploid and allopolyploid [2]. The structural changes of genome including chromosome fusions, chromosome number reduction, and a variety of chromosome rearrangements were a significance event during polyploidy formation [3]. It has been illustrated in many allopolyploid species, such as Nicotiana [3,4], Tragopogon [5], Gossypium [6,7] and Brassica [8,9]. In Nicotiana, intergenomic translocations have been detected in natural N. tabacum genotypes and this translocation was considered to be significant in tobacco fertility [3]. Compared with in allopolyploid, structural changes were more difficult to be discovered due to homologous genomes were duplicated in autopolyploid. However, chromosomal rearrangements were reported in induced autotetraploid Lathyrus sativus [10] and Arabidopsis thaliana [11].
The Medicago sativa complex includes both diploid (2n = 2× = 16) and tetraploid (2n = 2× = 32) subspecies [12]. Tetraploid subspecies M. sativa ssp. sativa L., an important world forage legume, and diploid subspecies M. sativa ssp. caerulea (Less. ex Ledeb.) Schmalh. have a similar morphology with violet flowers and coiled pods [13,14]. Subspecies M. sativa ssp. falcata (L.) Arcang. comprises both diploid and tetraploid forms which differ morphologically from the previous two taxa by having conspicuous yellow flowers and straight to sickle-shaped pods [13,14]. With the similar ploidy level, they intercross easily and produce viable hybrids [15]. Tetraploid ssp. sativa and ssp. falcata have been considered to be autotetraploidy due to appearance of quadrivalents at meiosis and tetrasomic inheritance [16][17][18]. Two diploid taxa ssp. caerulea and ssp. falcata in the complex were hypothesized to be the direct progenitor of tetraploid ssp. sativa and ssp. falcata, respectively [13,14]. However, recent molecular evidence of chloroplast suggested M. prostrata may have introgression into the tetraploid ssp. falcata in past. Therefore, the Medicago complex is an interesting model for polyploidy evolutionary study especially for autopolyploid [19].
Heterochromatin distributions of diploid ssp. falcata, ssp. caerulea and tetraploid ssp. sativa have been analyzed by C-banding and N-banding techniques [20][21][22][23][24]. Comparing results showed that diploid ssp. caerulea had similar heterochromatin distribution with tetraploid ssp. sativa: constitutive heterochromatic was distributed mainly around the centromeres, telomere and interstitial region of short arms of the chromosomes and partly presented at the interstitial region of long arms of chromosomes [20][21][22][23][24]. On the contrary, there were few heterochromatic distributions on the telomere and interstitial region in diploid ssp. falcata except centromere regions [20,21,25]. Bauchan and Hossain's unpublished data mentioned that there were a larger number of C-bands in tetraploid ssp. falcata than that had been discovered in diploid ssp. falcata [12].
Compared with the traditional banding techniques, fluorescence in situ hybridization (FISH), a valuable molecular cytogenetic tool, can display the molecular information on the chromosome more directly, more accurately, and more stably [26,27]. It has been widely applied to the study of plant genomic organization, chromosome identification, and species evolution by physical mapping repetitive genes or other sequences directly onto chromosomes [27][28][29][30][31][32]. In our previous study [33], 11 tandemly repetitive sequences (nine of which were novel) were isolated from a Cot-1 library in alfalfa and a FISHbased molecular cytogenetic karyotype was well developed for tetraploid ssp. sativa. In this study, we present an indepth comparative molecular cytogenetic analysis between diploid and tetraploid subspecies in Medicago sativa complex using repetitive sequences and FISH. Chromosome changes will be described in detail in evolution process of autotetraploidy ssp. sativa. The relationship of tetraploid and diploid ssp. falcata will be discussed.

Plant materials
Four diploid ssp. caerulea, four diploid ssp. falcata, and six tetraploid ssp. falcata samples were used as materials in this study. Accessions beginning with 'PI' were obtained from the National Plant Germplasm System (NPGS) of the United States Department of Agriculture (USDA). Two tetraploid ssp. falcata accessions, XiaNH-072X-824 and Lizj0944, were acquired from the China Germplasm Bank of Wild Species. Accession 2-6 was collected from a wild population in Xinjiang, China. A list of materials with ploidy levels and origins is given in Table 1.

Chromosome preparation
Root tips with a length of 1-2 cm were harvested from germinated seeds or growing plants and pretreated in ice-cold water at 4°C for 20-24 h. Root tips were then fixed in ethanol:glacial acetic acid (3:1, v/v) for 4 h at room temperature. Each root tip was squashed in a drop of 45% acetic acid. Finally, the slides were stored at −80°C before use.

FISH and microphotometry
FISH procedure was based on Mukai's description [35] with minor modifications. Chromosome DNA denaturation was carried out in 0.2 M NaOH in 70% ethanol at room temperature for 8 min and then dehydrated with the cold ethanol series. The probe mixture (25 ng of each labeled probe DNA, 5-10 mg of sheared salmon sperm DNA, 50% formamide, 2 × SSC, and 10% dextran sulfate) was denatured for 5 min at 95°C and cooled on ice. Then, the denatured probe mixture was applied on dehydrated chromosome slide. The slides were incubated in a humid chamber at 37°C overnight. After hybridization, the slides were washed in 2× SSC three times for 5 min at room temperature and briefly dried. Chromosomes were counterstained with 4′, 6-diamidino-2-phenylindole in Vectashield mounting medium (Vector Laboratories, Burlingame, CA, USA). Images were acquired with a cooled charge-coupled device camera (Photometrics CoolSNAP) under a fluorescence microscope (Leica) and were processed with the MetaVue Imaging System. Finally, images were adjusted with Adobe Photoshop 6.0 for contrast and background optimization.

Results
Physical mapping of repetitive sequences on mitotic chromosomes In Medicago sativa ssp. caerulea Physical mapping of the 11 repetitive sequences in diploid ssp. caerulea accession PI 464715 was conducted by FISH (Additional file 1: Figure S1a-h). Repeat sequence MsCR-1, MsCR-2, MsCR-3, MsCR-4, and MsCR-5 were physically mapped on pericentromeric regions of 14, 8, 6, 10, and 9, respectively, of the 16 chromosomes of ssp. caerulea (Additional file 1: Figure S1a-d). Double-target FISH further revealed that MsCR-3 overlapped with MsCR-2, MsCR-4, and MsCR-5 on four, two, and three chromosomes, respectively (Additional file 1: Figure S1b-d). All three subtelomeric sequences (MsTR-1, clone 65, and clone 74) were co-localized on one end of 12-13 chromosomes (Additional file 1: Figure S1e, f). At the same time, a variation of two end of one chromosome was also detected (Additional file 1: Figure S1e). Probes E180, clone 68, and clone 87 displayed hybridization signals on 16, 15, and 14 chromosomes, respectively (Additional file 1: Figure S1g, h). Double-target FISH revealed different FISH patterns between E180 and clones 68 or 87. The physical mapping results revealed E180 produced the greatest number of information of hybridization signals on each chromosome. Thus, double-target FISH between each repetitive sequence and E180 were carried out. Consequently, we used E180 FISH patterns, previous double-target FISH results, and chromosome arm ratios as references to allocate each sequence to a particular chromosome (Additional file 1: Figure S2a To development a standard molecular karyotype among different ssp. caerulea accessions, three other accessions (PI 212798, PI 577551, and PI 577548) were also used in cytogenetic analysis. Because 18S-26S rDNA, 5S rDNA, E180, MsCR-3, and MsTR-1 repeats showed a strong ability to distinguish chromosomes according to the results of chromosome allocation, two FISH cocktails-one consisting of 18S-26S rDNA, 5S rDNA, and E180 (Additional file 1: Figure S3a Fig. 3. Although polymorphic FISH patterns were detected on a few chromosomes among accessions, a relatively conserved karyotype was still described in Fig. 6 (a).

In diploid Medicago sativa ssp. falcata
Physical mapping of the 11 repetitive sequences in diploid M. sativa ssp. falcata accession PI 631808 was also conducted using FISH (Additional file 1: Figure S1i-p). Repeat sequence MsCR-1, MsCR-2, MsCR-3, MsCR-4, and MsCR-5 were physically mapped on pericentromeric regions of 16, 10, 0-1, 9, and 8 of the 16 total chromosomes, respectively (Additional file 1: Figure S1i-l). In addition, MsCR-4 and MsCR-5 showed an extra band on one and two chromosomes, respectively. The three subtelomeric probes (MsTR-1, clone 65, and clone 74) were co-localized on only one end of one chromosome (Additional file 1: Figure S1m, n). Probe E180 was mostly localized at a single site (mainly around the centromere) of 10 to 11 chromosomes rather than the multiple sites observed on nearly all chromosomes in ssp. caerulea. Similarly, clones 68 and 87 were also mostly co-localized in centromeric regions on 16 chromosomes (Additional file 1: Figure S1o, p). Doubletarget FISH revealed that E180 overlapped with clones 68 and 87 on nine chromosomes.
To further characterize the chromosomes of ssp. falcata, hybridizations were also carried out using probe E180 and each repetitive sequence (Additional file 1: Figure S2i-p). Each sequence was allocated to a particular chromosome as follows ( The same two FISH cocktails with ssp. caerulea were applied to diploid ssp. falcata PI 631808 and three other accessions (PI 234815, PI 502447, and PI 631813) to develop a standard molecular karyotype (Additional file 1: Figure S3e-h and m-p). The polymorphic distributions were presented in Fig. 3. A relatively conserved karyotype pattern was described in Fig. 6 (b).
According to hybridization results between E180 and each repeat sequences, the chromosomal distribution of each sequence was allocated as follows (Fig. 4)  The same two FISH cocktails with ssp. caerulea were applied to tetraploid ssp. falcata accession XiaNH-072X-824 and five other accessions (PI 634023, PI 634118, PI 634117, Lizj0944, and 2-6) to develop a standard molecular karyotype (Additional file 1: Figure S5). Marked variability was detected among the six tetraploid ssp. falcata accessions (Fig. 5). Thus only polymorphism schematic diagram was built in Fig. 6 (c).
Comparative cytogenetic analysis between diploid and tetraploid subspecies M. sativa ssp. caerulea and ssp. sativa The comparative results of chromosome distribution of each repeat sequence ( Table 2) showed that signals of all probes had similar chromosomal locations between ssp. caerulea (PI 464715) and ssp. sativa (Zhongmu No. 1). Moreover, the signal numbers of each probe were nearly twice between ssp. caerulea (PI 464715) and ssp. sativa (Zhongmu No. 1). Cocktail FISH results revealed the chromosome distributions of repeat sequences were highly conserved in four ssp. caerulea accessions and four ssp. sativa accessions, respectively ( Fig. 6(a) and (d)). The high similar FISH patterns between ssp. caerulea and ssp. sativa facilitated the recognition of homoeologous chromosomes. Furthermore, autopolyplidization from ssp. caerulea to ssp. sativa was tentatively reconstructed. The reconstructed results showed though collinearity was well maintained in most chromosomes between diploid and tetraploid by FISH patterns, significant and stable variations were also detected in a few chromosomes ( Fig. 7 (a)). Compared with chromosome 3 of ssp. caerulea, chromosome 6′ of ssp. sativa was missing the MsCR-3 signal at pericentromeric region. Compared with chromosome 8 of ssp. caerulea, chromosome 16′ of ssp. sativa was missing the MsCR-3 signal at pericentromeric region. Compared with chromosome 5 of ssp. caerulea, chromosome 9′ of ssp. sativa had an extra MsCR-3 signal at pericentromeric region. Compared with chromosome 6 of ssp. caerulea, chromosome 12′ of ssp. sativa had 5S and MsTR-1 repeat signals on long arm instead of short arm. Compared with chromosome 7 of ssp. caerulea, chromosome 14′ of ssp. sativa had one of E180 signals near the centromere of short arm instead of long arm. Compared with chromosome 8 of ssp. caerulea, chromosome 15′ of ssp. sativa had E180 signals on short arm instead of long arm. Combining all chromosome changes, chromosome deletion was speculated to occur on the long arm of chromosome 12′ of ssp. sativa during evolution. And pericentric inversions were speculated to occur in chromosome 14′ and chromosome 15′ of ssp. sativa during evolution. Putative chromosome changes were presented on Fig. 7 (b). Furthermore, significant variations between four groups of homologous chromosomes of tetraploid alfalfa were also detected ( Fig. 7 (a)). Compared with homologous chromosome 8′, chromosome 7′ was missing the MsTR-1 signals at subtelomeric region of short arm. Compared with homologous chromosome 11′, chromosome 12′ had 5 s signal on short arm instead of long arm. Compared with homologous chromosome 13′, chromosome 14′ had an extra E180 signals at the pericentromeric region of short arm. Compared with homologous chromosome 15′, chromosome 16′ was missing MsCR-3 signal at pericentromeric region and had signals of MsTR-1and E180 repeats on long arm instead of short arm.  [33]. Chromosomes of tetraploid ssp. sativa are marked with "′" on chromosome numbers. A small black dot next to the FISH signal indicates that the signal is polymorphic across accessions Diploid ssp. falcata and tetraploid ssp. falcata The comparative results of chromosome distributions of each repeat sequence ( Table 2) showed that signals of ten probes had similar chromosomal locations between diploid ssp. falcata (PI 631808) and tetraploid ssp. falcata (XiaNH-072X-824). However, signal distributions of E180, clone 68, and clone 87 probes were more abundant in tetraploid ssp. falcata than in diploid ssp.
falcata. The signal numbers of MsCR-1, MsCR-2, MsCR-4, MsCR-5, E180, clone 68, and clone 87 probes were nearly twice between diploid ssp. falcata (PI 631808) and tetraploid ssp. falcata (XiaNH-072X-824). However, the signals of MsCR-3, MsTR-1, clone 65, and clone 74 probes was only located on one chromosome in diploid ssp. falcata rather than on many chromosomes in tetraploid ssp. falcata. Moreover, the cocktail  Fig. 7 a Reconstructing of chromosome-doubling process of diploid ssp. caerulea. Chromosomes of tetraploid ssp. sativa are marked with "′" on chromosome numbers. b Putative chromosome changes during polyploidization process. A small black dot next to the FISH signal indicates that the signal is polymorphic across accessions FISH results revealed highly variable karyotypes across different tetraploid ssp. falcata accessions. Thus, chromosome collinearity analysis between diploid and tetraploid ssp. falcata could not be conducted as did between ssp. caerulea and ssp. sativa.

Discussion
Genomic differentiation of M. sativa ssp. caerulea and diploid ssp. falcata Medicago sativa ssp. caerulea and diploid ssp. falcata are sympatrically distributed, with naturally occurring hybrids recorded between them [13,14,36]. The genetic affinity of the two species has been demonstrated by cytological research [15]. In addition, chromosomal differentiation between M. sativa ssp. caerulea and diploid ssp. falcata has been well described by analyses of both C-and N-banded chromosomes [20,21]. C-and N-banding has revealed that chromosomes of diploid ssp. falcata possess only centromeric bands. In contrast, all chromosomes of ssp. caerulea have a centromeric band and a telomeric band in the short arm; in addition, most of the chromosomes of this subspecies have interstitial bands in the short arm, with a few chromosomes featuring prominent interstitial bands in the long arm. C-and N-bands reflect constitutive heterochromatic DNA in chromosomes [23]. Our molecular cytogenetic analysis revealed the heterogeneous nature of the constitutive heterochromatin among centromeric, interstitial, and subtelomeric regions. In both ssp. caerulea and diploid ssp. falcata, centromeric bands were revealed to be a heterogeneous mix of MsCR-1, MsCR-2, MsCR-3, MsCR-4, MsCR-5, clone 68, clone 87, and E180 sequences, along with a few 5S rDNA sites. The interstitial bands comprised E180 sequences along with 18S-26S rDNA and 5S rDNA sites, and the subtelomeric bands were represented by MsTR-1, clone 68, clone 87, and E180.
Chromosomal differences between ssp. caerulea and diploid ssp. falcata as revealed by FISH were similar to those uncovered by C-or N-banding. The repetitive sequences were physically mapped onto centromeric, subtelomeric, or interstitial regions in ssp. caerulea, whereas the mapped sequences were mainly on centromeric regions in diploid ssp. falcata. Furthermore, conspicuous differences in distribution patterns were observed between ssp. caerulea and diploid ssp. falcata, even though the repetitive sequences detected in centromeric regions of both species displayed similar levels of heterogeneity. Unlike ssp. caerulea, more than half of the chromosomes of ssp. falcata contained E180 sequences in centromeric regions. In addition, centromeric sequences of MsCR-3 were detected on one or no chromosomes of diploid ssp. falcata, whereas they were found on 3-5 pairs of chromosomes in ssp. caerulea.
Genetic differentiation between ssp. caerulea and diploid ssp. falcata has been previously revealed by nuclear markers [37][38][39]. Moreover, relationships uncovered among diploid members of the M. sativa species complex based on chloroplast DNA sequence analysis supports the recognition of ssp. caerulea and diploid ssp. falcata as distinct taxa [40]. Our study has revealed distinct genomic differentiation between ssp. caerulea and diploid ssp. falcata and supports their taxonomic differentiation at the chromosome level.
Chromosome evolution after polyploidization of diploid ssp. caerulea Violet flowered diploid ssp. caerulea is postulated to have given rise to tetraploid ssp. sativa (alfalfa) [13,16,17]. The identical C-banding patterns of tetraploid alfalfa and ssp. caerulea support tetraploid alfalfa as an autotetraploid derived from diploid ssp. caerulea [21,23]. Sequencing of chloroplast DNA has demonstrated that the two taxa have very closely related chloroplast haplotypes, with most individuals sharing the same haplotype, and are thus undifferentiated genetically for this characteristic. Similar to the Cbanding analysis, chloroplast data supports a simple autopolyploid origin for ssp. sativa from diploid ssp. caerulea [19]. In our study, a putative chromosome doubling process from diploid ssp. caerulea to tetraploid alfalfa was reconstructed according to similar FISH patterns. The results strongly supported the simple autotetraploid origin of ssp. sativa from diploid ssp. caerulea.
It is generally believed that polyploid plants may have unstable genomes in a long term due to a genome-wide gene redundancy [41]. Ma and Gustafson [42] summarized the evolution of an allopolyploid species is a process of both cytological and genetic diploidization. Rapid genomic rearrangement such as chromosome insertion, chromosome deletion, and chromosome rearrangement, which would lead to diploidization of genome structure, has been investigated in some allopolyploid plant species [41,43]. However, the occurrence of similar changes remains to be studied in detail during the generation of autopolyploids [44]. The limited data available so far imply that autopolyploids experience less genome restructuring than allopolyploids [44]. In our study, putative genome changes were discovered after polyploidization of diploid ssp. caerulea. Elimination of repetitive DNA was detected in pericentromeric regions of chromosome 6′ and chromosome 16′ of tetraploid ssp. sativa. Increase of repetitive DNA was detected in pericentromeric regions of chromosome 9′ of tetraploid ssp. sativa. Chromosome deletion was postulated to occur in the long arm of chromosome 12′ of tetraploid ssp. sativa. Pericentric inversions were postulated to occur in chromosome 14′ and chromosome 15′ of tetraploid ssp. sativa. Furthermore, significant diversification was recognized in four groups of homologous chromosome of tetraploid ssp. sativa including chromosome 7′ and chromosome 8′, chromosome 11′ and chromosome 12′, chromosome 13′ and chromosome 14′ and chromosome 15′ and chromosome 16′. Thus, we concluded that autotetraploid alfalfa had undergone a partial diploidization by the progressive accumulation of chromosome structural rearrangements during evolution. A previous study of pachytene karyotype reported that at least four groups of the tetraploid chromosomes appear sufficiently alike to be able to form quadrivalents in ssp. sativa, and three of these were seen to form quadrivalents at pachytene [45]. Subsequently, Armstrong summarized the quadrivalent frequency at pachytene ranged from 0.89 to 2.93 in tetraploid ssp. sativa [46]. It was considerably below theoretical expectations "5.34 quadrivalents per cell" for an autotetraploid [46]. This chromosome behavior in meiotic has confused the origin of tetraploid ssp. sativa. Partial diploidization of homologous chromosome groups in tetraploid M. sativa found in our study should be an explanation for the low quadrivalent frequencies at meiotic.

Phylogenetic relationships between diploid and tetraploid forms of ssp. falcata
Diploid and tetraploid forms of ssp. falcata have been traditionally treated as a single species, M. sativa ssp. falcata. Diploid ssp. falcata and tetraploid ssp. falcata have been recognized as diploid and tetraploid cytotypes on the basis of chromosome counting and morphology. Diploid ssp. falcata is hypothesized to be the ancestor of autoploid tetraploid ssp. falcata [13,14]. As revealed by C-and N-banding, centromeric bands are a distinct feature of the chromosomes of diploid ssp. falcata [20,21]. Because of this assumption of autopolyploid origin, the C-banding pattern of tetraploid ssp. falcata was expected to be similar to that of diploid ssp. falcata. Thus, the results of a preliminary study of six accessions of tetraploid ssp. falcata were surprising. Most of the plants possessed chromosomes that had C-bands in addition to normal centromeric bands [12]. Highly variable C-banding patterns were detected in these accessions. The accession containing the fewest number of additional bands had four pairs of chromosomes with an extra telomeric band on their short arms, whereas the remaining chromosomes had only centromeric bands. At the other extreme, two accessions had multiple bands on each chromosome, similar to doubled-diploid ssp. caerulea. Even though the studied accessions had yellow flowers with sickle-shaped pods, the accessions were speculated to be the product of hybridization with ssp. sativa [12]. Similarly, the six tetraploid ssp. falcata accessions used in our study-three acquired from the NPGS USDA germplasm bank and three collected in situ in China on the basis of morphological identification-also showed highly variable molecular karyotypes. We found that hybridization sites of MsTR-1 and E180, which frequently produce subtelomeric and interstitial bands, respectively, in C-banding analyses, were highly variable across different individuals of tetraploid ssp. falcata. Along with the results of C-banding analysis [12], our results suggested the actual genomic characteristics of tetraploid ssp. falcata.
In an earlier analysis of chloroplast DNA, morphologically identical diploid and tetraploid cytotypes of ssp. falcata were found to possess very different chloroplast haplotypes. The most common haplotype of tetraploid ssp. falcata was shared with M. prostrata rather than diploid ssp. falcata, suggesting past introgression from M. prostrata into the polyploid. The evolutionary trajectory of ssp. falcata does not appear to have involved a simple autopolyploid origin as seen in ssp. sativa [19]. A high variability in the number of chromosomes with multiple E180 sites, which are frequently lacking in diploid ssp. falcata, was uncovered in our study. Although information on the chromosomal distribution of E180 in M. prostrata is not available, multiple E180 hybridization signals have been detected in species of section Medicago, such as M. glutinosa, M. hemicycla, and M. polychroa [47]. This finding suggests that the variable chromosomes in tetraploid ssp. falcata could have been introduced from M. prostrata. Our data indicate that the origin of tetraploid ssp. falcata from diploid ssp. falcata is far from simple. Elucidation of the evolutionary history of ssp. falcata will require a large amount of additional data.

Conclusions
The comparative cytogenetic results provide reliable evidence that diploid subspecies caerulea is the direct progenitor of tetraploid subspecies sativa. And autotetraploid ssp. sativa has been suggested to undergo a partial diploidization by the progressive accumulation of chromosome structural rearrangements during evolution. However, the tetraploid subspecies falcata is far from a simple autopolyploid from diploid subspecies falcata although no obvious morphological change was observed between these two subspecies.