The occurrence of natural hybrids with high ploidy levels has rarely been documented for hybrid complexes from the Triticeae tribe. For example, there are reports of a heptaploid hybrid between Thinopyrum junceum and Elytrigia repens from Sweden [30] and a nonaploid hybrid of Elytrigia pycnantha and E. repens from France [31]. GISH analysis of the latter hybrid revealed the presence of four haplomes within this nonaploid, which consisted of four chromosome sets from Pseudoroegneria, two chromosome sets from Agropyron, two chromosome sets from Thinopyrum and one chromosome set from Hordeum. These results demonstrate that regular meiosis in higher polyploids may occur in hybrid complexes within Triticeae. However, to the best of our knowledge, the presence of four haplomes has not been shown in hexaploid and heptaploid natural hybrids.
In this study, we show that Elytrigia ×mucronata is an allopolyploid of high complexity, in which all three studied hybrid cytotypes comprised all four haplomes (D, H, St, V) present in the parental species E. repens and E. intermedia. In all but one case (see below), all of the chromosome sets in the hybrids were euploid and consisted of seven chromosomes. While the genomic constitution (i.e., the type and number of specific haplomes/basic genomes) of the E. ×mucronata hybrids reflects the ploidy level of particular cytotypes, it also depends on the type of gametes involved in the origination of particular plants.
If we assume that regular meiosis occurs in the parental species, then the E. ×mucronata hexaploid would harbour three chromosome sets from Pseudoroegneria and one chromosome set each from Hordeum, Aegilops and Dasypyrum (genomic formula StStStHDV). In addition to these chromosome sets, the heptaploid cytotype harboured an additional chromosome set from Pseudoroegneria (StStStStHDV). The genomic constitution of the nonaploid cytotypes differed between plants. Two plants exhibited 35 St + 14 H + 7 D + 7 V chromosomes, while the other nonaploid harboured 28 St + 7 H + 14 D + 14 V chromosomes. We assume that this difference reflects the distinct origins of the nonaploid cytotypes [16] (see below).
Origin of different cytotypes
The genomic constitutions of the analysed plants allowed us to partly infer the types of gametes that gave rise to their origination (Fig. 6). The hexaploid cytotype of E. ×mucronata most likely originated through the merging of two reduced gametes from both parental species.
As mentioned above, nonaploid plants most likely originated through the fusion of reduced and unreduced gametes [15, 16]. Mahelka et al. [16] suggested different scenarios for the origin of the hybrid nonaploids 50–1, 50–7 and 41–5 (plants N7, N6 and N8 in the original article). While plants 50–1 and 50–7 may have arisen from 2n (E. repens) + n (E. intermedia) or 2n (6x E. ×mucronata) + n (E. repens) combinations, nonaploid 41–5 may represent either 2n (E. intermedia) + n (E. repens) or 2n (6x E. ×mucronata) + n (E. intermedia) gamete compositions. Since the alternative gamete combinations result in the same genomic compositions, we are not able to discern which scenario truly led to the formation of the analysed nonaploids by using GISH. The involvement of hexaploid E. ×mucronata hybrids in the formation of the nonaploids seems to be more likely because hybrids might more easily produce unreduced gametes than pure species due to disturbed meiosis [9]. The heptaploid cytotype likely resulted from heteroploid hybridisation; however, the exact mode of its origination is difficult to determine. One possibility is that the heptaploid originated after a cross between a hexaploid and an octoploid (2n = 8x = 56). If the hexaploid parent was either E. repens or E. intermedia (gamete n = 3x = 21 = StStH or StDV, respectively), then the gamete from the octoploid would have to have been n = 4x = 28 = StStDV or StStStH, respectively (Fig. 6). However, no octoploid plants have been recorded from this locality. Alternatively, the heptaploids could have originated after a cross between a hexaploid and a nonaploid, in which the latter was the donor of the gamete comprising four chromosome sets. Such a scenario has been observed in Elytrigia wheatgrasses, where a heptaploid plant was found among progeny of the nonaploid hybrid 50–1 collected in the field [16]. It is likely that the pollen donor was either E. repens or E. intermedia. Unfortunately, the genomic constitution of this particular heptaploid was not analysed.
Chromosomal alterations in E. ×mucronata
Chromosomal alterations occurred in all three cytotypes and involved all four haplomes. Most of the translocations involved St and H chromosomes, while the V chromosome from Dasypyrum was involved in only one translocation. The question is whether this difference simply occurred because St and H chromosomes outnumber chromosomes from the other haplomes, or if it stems from different levels of karyotype stability, which may have a strong effect on chromosome restructuring and aneuploidy in Triticeae [32].
Structural chromosomal alterations have been reported in other perennial species harbouring an H and/or St haplome. Dou et al. [33] found two types of non-reciprocal translocations between H and St haplomes and two types of reciprocal translocations between H and Y haplomes in Elymus nutans (2n = 6x = 42, StStHHYY). Different frequencies of chromosomal alterations between particular haplomes were observed in Kengyilia thoroldiana (2n = 6x = 42, StStPPYY). The frequency of P/Y translocations was higher than that of P/St translocations, while no translocations were observed between the chromosomes of the St and Y haplomes [34].
The question arises of how frequently and at which stage of hybrid formation do translocations occur? Without knowledge on the parental species, it is problematic to infer whether particular structural rearrangements have been inherited from the parental species, or whether they have originated de novo in hybrids. Cytogenetic analyses of local accessions of parental species E. intermedia and E. repens showed that in E. repens, one pair of Hordeum chromosomes carried a centromeric H/St ‘translocation’ [18], corresponding to what we called Type 1 translocation here. In contrast, no translocation that would resemble those observed in the E. ×mucronata plants analysed here were observed in the other parental species E. intermedia [19, 23]. Therefore, we concluded that all but one (type 1) chromosomal alterations in the three cytotypes of E. ×mucronata appeared during hybrids’ formation.
In any case, the presence of only one such translocation in the hexaploid hybrid indicates the occurrence of regular meiosis in the E. repens parent. Thus, it appears that no de novo translocations appeared in the E. ×mucronata hexaploid. The presence of two such translocations in two nonaploids (50–1 and 50–7) and one translocation in the other nonaploid (41–5) is consistent with both alternative origins of these nonaploids.
Furthermore, the presence of the D/H reciprocal translocation (type 5) in the nonaploid 50–1 indicates that this plant is not a primary nonaploid hybrid between E. intermedia and E. repens, since recombination between H and D haplomes (which do not co-exist in any of the parents) must have occurred in the hybrid plant. Therefore, an origin involving a 2n (6x E. ×mucronata) + n (E. repens) combination seems more likely in this plant.
The notably higher frequency of chromosomal alterations observed in heptaploid plants contrasted with the results for the other analysed cytotypes. Three of the four types of translocations (types 3, 4, and 6) were not found in other cytotypes, suggesting that several multivalents between homoeologous chromosomes must have occurred during the formation of the gametes, giving rise to the heptaploid plants.
The concurrent presence of both chromosomes with reciprocally translocated segments indicates alternate segregation (i.e., the translocated chromosomes do not segregate and are present in a single gamete). It is unlikely that this state originated from the fusion of two unbalanced gametes. Non-reciprocal translocations can be non-reciprocal per se or may result from adjacent segregation, where normal and translocated chromosomes segregate into one gamete [35].
Moreover, the numerical chromosomal alteration observed in one heptaploid plant was an example of hidden aneuploidy [27, 32], i.e., the absence of one chromosome is compensated by the acquisition of an extra chromosome from another chromosome set.
Chromosomes were also observed carrying differentiating signals in the centromeric regions in Dasypyrum-like chromosomes. This feature has previously been reported for E. intermedia species [19, 23]. Further research is required to reveal the true nature of these GISH patterns and determine whether they resulted from chromosome restructuring or sequence homology.
Mapping of repetitive DNA in E. ×mucronata hybrids
The evolution of ribosomal DNA genes in relation to allopolyploidy is an intensively studied issue (e.g., [36, 37]). rDNA loci are valuable chromosome markers, and the mapping of rDNA loci using in situ hybridisation allows for the evaluation of the progenitor-derivative patterns and positional dynamics of ribosomal loci within allopolyploid genomes. rDNA genes in allopolyploid species may experience contrasting and barely predictable patterns of evolution, ranging from loss of some loci with respect to their progenitors (Zingeria—[38]) over nearly complete additivity (e.g., Thinopyrum ponticum—[39]; Nicotiana—[40]) to increasing the number of rDNA loci (Triticum—[41]). Triticeae grasses are characterised by the ability to change the positions of rDNA loci [42,43,44], which may occur via the dispersion of minor loci, followed by rDNA array magnification and deletion of the original loci thereafter. Loss of rDNA loci is one potential mechanism accelerating the process of concerted evolution [45].
In all investigated hybrid cytotypes, the 5S and 18S rDNA loci were located on all chromosome sets representing the different haplomes. Therefore, the rDNAs of the E. ×mucronata hybrid reflect the principle of genome additivity to a certain degree. However, because E. ×mucronata is a hybrid between two allopolyploid species, the dynamics of the rDNA loci of this hybrid are traceable, provided that the pattern in both parental species is understood. Thus, we characterised specimens of both E. repens and E. intermedia from the same distribution area from which the hybrids originated ([18, 23], Mahelka, Kopecký, unpubl. data). In both parental species, we encountered some reorganisation of rDNA loci with respect to their diploid progenitors, which likely occurred following the origination of the allopolyploids (for discussion, see [18, 23]). The patterns of the rDNA loci of both E. repens and E. intermedia are shown in Additional file 1: Table S1. Notably, severe elimination of all but one minor 45S rDNA locus likely occurred within the Hordeum subgenome in E. repens (genomic formula StStStStHH). Similarly, the elimination of some 45S rDNA loci occurred within the Pseudoroegneria- and Dasypyrum-like subgenomes in E. intermedia (genomic formula StStDDVV). In both species, 5S rDNA was less heavily affected by the loss of the loci than 45S rDNA.
In this study, we predicted the theoretical numbers of rDNA loci in hybrid E. ×mucronata cytotypes (Additional file 1: Table S1) by assuming complete additivity of the parental rDNA loci. From comparison of the observed data, we can infer the positional dynamics of rDNA loci in the hybrids. For this purpose, we consider the 18S and 45S probes to be equally informative.
The total numbers of 5S rDNA loci fell well into the expected numbers in all hybrid plants examined. In the nonaploid 41–5, the number of 5S rDNA loci only fell within the expected range if the [2n (6x E. ×mucronata) + n (E. intermedia)] scenario of its origin was considered. Under the opposite scenario [2n (E. intermedia) + n (E. repens)], a lower number was observed (11 vs. 13–15 expected). This depletion was mainly caused by a smaller number of 5S rDNA sites on Pseudoroegneria chromosomes than expected (4 sites observed vs. 6–7 expected). However, we must note that the real number of loci in this plant may be higher than was observed because one chromosome was missing in the analysed metaphases. Such an observation is otherwise in agreement with the pattern found for 5S rDNA loci in the parental species, in which no major changes in the 5S rDNA loci were recorded ([23], unpubl. data).
In contrast, the total numbers of 18S rDNA sites were always higher than expected in the hybrids of all analysed cytotypes. This observation is especially interesting if we consider that severe losses of 45S rDNA loci had already occurred in both parental species [18, 23]. Restoration of some loci clearly occurred within the Hordeum and, to a lesser degree, Pseudoroegneria haplomes (see hexaploid 10–1) after the hybridisation events. In particular, while examining the Hordeum haplome, we observed co-localised 5S–18S and/or 18S–5S–18S rDNA loci in all the cytotypes (although in both heptaploids, the co-localised locus had been translocated to a Pseudoroegneria chromosome); however, this locus was not observed in E. repens [18]. This pattern was consistent in all three cytotypes, but neither the mechanism of the re-appearance of the loci nor its cause was studied.
We did not probe either of the parental species with the pSc119.2 probe. In other studies, up to 5 chromosomes with one or two pSc119.2 loci (located in a terminal or interstitial position) have been found in E. intermedia [46]. In E. repens, the total number of pSc119.2 sites ranges between 5 and 10 (one interstitial site and others in the terminal region) [47]. Although these plants come from different geographic regions, it appears that similar to 18S rDNA, the total number of pSc119.2 sites detected in E. ×mucronata was higher than that in the parental species.