The genus Pulmonaria is karyologically highly variable [e.g. 3, 6], with 16 different somatic chromosome numbers reported (Kobrlová unpubl.) and about 30 taxa recognized, growing in Europe and northeastern and eastern Asia [5, 7, 8, 9]. However, the mechanism, origin and evolutionary consequences of genome size and karyotype variability remain unexplored [but see 10]. Apparently, chromosomal rearrangements have played an important role in the evolution of this genus [cf. 10, 19, 39], but how and to what extent has never been clearly demonstrated.
In this work, we studied the P. officinalis complex (Fig. 1), which includes two morphologically similar, closely related species that differ in chromosome number [3, 10, 18, 59, 60]. The present study confirmed previously reported chromosomal data, which were consistent across the whole of Europe. The predominant 2n = 16 reports correspond to P. officinalis s.l., whereas only 2n = 14 were documented in P. obscura populations (Fig. 1). Only in one case, 2n = 17 was documented from the “pure” population of P. officinalis [22], but no explanation was given. Despite the morphological and karyological differences observed, the origin of this species group remains unknown. To uncover the differences in the karyotype evolution, we used complex methodological approaches involving partial Illumina sequencing followed by bioinformatic analysis and characterization of repeatomes in the P. officinalis group. We identified a new set of chromosome-specific cytogenetic landmarks and performed comparative karyotyping within and between the two species, their putative natural hybrid from a population where both species occur, and ornamental cultivars morphologically similar to P. officinalis, which are also rarely found in nature.
Impact of DNA repeats dynamics on genome size
Genome size can reflect some aspects of the evolutionary history of taxa by allowing us to understand the influence of DNA gain/loss between related species [e.g. 70, 71]. Despite the apparent karyological variability of several Boraginaceae genera [see e.g. 14, 72, 73], there have been almost no complex analyses of genome size variation and the evolutionary pathways behind the observed diversity [but see 74, 75, 76]. Given the sparse DNA content records in the Boraginaceae family, the only comprehensive study has been published, providing the first genome size reports for most taxa [12].
Our study represents the first large-scale investigation of interspecific genome size variation in Pulmonaria. As already shown in a pilot study by Kobrlová & Hroneš [12], genome size is effective in delimiting morphologically similar taxa of the Boraginaceae, which is also true for the P. officinalis group. We found a significant difference in genome size between P. obscura and P. officinalis, corresponding to the number of chromosomes (Fig. 2), confirming previous results [cf. 12], but on a larger geographical scale. The suitability of using flow cytometry to revise the distribution of the P. officinalis group (i.e. relative genome size) has already been documented in the Bohemian Forest and adjacent foothills [60].
So far, the genome size has only been estimated for eight Pulmonaria taxa, including P. officinalis and P. obscura, ranging from 2.27 to 4.27 pg (i.e. very small/small genomes according to the categories of Leitch et al. [77], Supplementary Table 3). Only minor differences were observed when comparing previously analyzed genome sizes of P. obscura and P. officinalis with our data, most likely due to different methodologies used (i.e. nuclei isolation buffer, reference standard, plant organ [cf. 12, 39]). The only exception is the study by Šmarda et al. [78], where almost the same 2C values are presented for P. obscura and P. officinalis, probably as a consequence of taxa misidentification.
DNA transposons have been shown to be the major contributor to the enormous variation in genome size in plants [e.g. 27, 79, 80, 81, 82, 83]. To shed light on genome size dynamics and relationships between P. officinalis and P. obscura species, we performed a genome-wide comparison of their repeatomes. We found that most of the repetitive elements in the genomes of the Pulmonaria taxa studied were dispersed repeats represented by LTR retrotransposons [cf. 27, 81], with higher proportion of Ty3/Gypsy elements, which were twice more abundant than Ty1/Copia. Ty3/Gypsy retroelements were almost exclusively represented by Tekay retrotransposons (Chromoviridea clade), whereas SIRE elements were the most abundant types of the Ty1/Copia superfamily (Table 2). Ty3/Gypsy elements represent one of the major classes of LTR retrotransposons and are dominant in many plant groups, such as the family Poaceae [e.g. 82, 84, 85, 86] or the tribe Fabeae [27, 87]. Unfortunately, a genome-wide analysis of DNA repeats and their impact on genome size has not been performed in any other species of the Boraginaceae family. However, the higher proportion of Ty3/Gypsy retroelements have also been found in genera of the closely related Solanaceae family, such as Solanum, Nicotiana and Capsicum [88, 89, 90, 91, 92, 93). In contrast, recent studies in the genus Salvia, a member of the closely related Lamiaceae family, have shown that the nuclear genomes of different species contain different proportions of Ty3/Gypsy and Ty1/Copia retroelements [94, 95, 96], indicating a proliferation of different types of DNA repeats during the evolution of individual species. In comparison, the Pulmonaria species analyzed contained a similar proportion of the repeat lineages and individual clusters were represented by reads from all specimens analyzed. This indicates a high degree of genome homology within the P. officinalis complex, suggesting that the evolution of this species group was not accompanied by a dramatic diversification of DNA transposons, as previously shown in other plant species [e.g. 82]. To better understand the proliferation of DNA repeats during genome evolution and its impact on genome size variation and speciation, analysis of a larger data set of Pulmonaria species from different phylogenetic groups is required.
Satellite DNAs and their use in comparative karyotyping
Repeatome analysis using the RepeatExplorer pipeline also allows the identification of putative satellite DNAs, which together with tandem organized ribosomal genes are the best cytogenetic landmarks. SatDNAs are usually species- or subspecies-specific, provide chromosome-specific labeling patterns and can therefore be used not only to generate and compare molecular karyotypes, but also to identify putative chromosomal structural changes [42, 45, 97]. The use of the FISH technique revealed different patterns of chromosomal localization of the tandem repeats and rDNA loci examined, both between and within P. obscura and P. officinalis (Fig. 5). The almost identical cytogenetic pattern of satDNAs and rDNA sequences in P. obscura (2n = 14), collected from three different populations, suggests karyotype stability in this diploid species (Fig. 5). In comparison, the chromosome structure in P. officinalis appears to be more dynamic, as individuals from two different populations differ slightly in the cytogenetic pattern of the satDNAs and rDNA sequences. Odd number of signals of some satDNAs as well as of rDNA sequences, and interstitial 45S rDNA loci were found in both diploid accessions (2n = 16), indicating chromosomal structural changes involved in the origin and evolution of P. officinalis (Fig. 5). It is generally accepted that n = 7 is the basic chromosome number in Pulmonaria [e.g. 2, 3, 4, 5, 6], which raises the question of how the species represented by different chromosome numbers arose. In the case of P. officinalis, there are two possible scenarios. There could have been chromosome fission leading to 2n = 16, or a more complex karyotype evolution with polyploidization and further diploidization by chromosome rearrangements. Large differences between the molecular karyotypes of P. obscura and P. officinalis suggest that diploid P. obscura most probably could not give a rise to P. officinalis by chromosome fissions (Fig. 5). Thus, we can only speculate, that the evolution and speciation in the P. officinalis group could be influenced by polyploidy and/or hybridization followed by post-polyploid diploidization process, which can result in numerous chromosomal rearrangements [see 98, 99, 100] and the origin of Pulmonaria species differing in their basic chromosome number. A similar evolutionary scenario involving chromosome multiplication, hybridization and, in particular, structural rearrangements of chromosomes (dysploidy) has been outlined by Liu et al. [10] for the origin of the P. hirta complex on the Italian peninsula and the Swiss endemic P. helvetica [19], respectively.
Evidence of hybridization within the P. officinalis complex
The evolution of the genus Pulmonaria remains unresolved, due to a lack of rigorous phylogenomic studies. However, several molecular studies have been published highlighting the important role of hybridization and introgression in the evolution of the genus [10, 18, 19]. The high level of chromosomal variation may also support the hypothesis of ancient hybridization events and subsequent chromosomal rearrangements [cf. 6, 7, 10, 18]. In addition, some species groups exhibit weak ecological and geographic isolation, near-synchronous phenology and pollinator sharing, all of which may facilitate the hybridization [cf. 18]. This is particularly true for the P. officinalis complex, which is widespread in Europe and therefore often in secondary contact with other Pulmonaria species (cf. 56, 101]. As the ranges of P. obscura and P. officinalis partly overlap (Fig. 1A), the co-occurrence of both species in the same habitat can be expected. Some authors have occasionally reported mixed populations, with morphological intermediates rarely observed [21, 22]. The extent of hybridization between these two species is still controversial. Nevertheless, several karyological data referring to as P. obscura × P. officinalis with an intermediate number of chromosomes (i.e. 2n = 15), may provide convincing evidence of an ongoing hybridization between these two species [18, 21, 22].
In our study, we analyzed presumed hybrids from a mixed population (B481) of P. obscura and P. officinalis. Chromosome counting in all three analyzed individuals confirmed 2n = 15, and their hybrid origin was also supported by their genome sizes, halfway between those of the parents (Table 1). The cytogenetic mapping of the set of satDNAs and rDNA sequences also strongly supports the hybrid origin, by the presence of P. obscura and P. officinalis species-specific satDNAs (PulTR_305 and PulTR_420) in haploid state, and also their pattern on chromosomes (Fig. 7). The hybrid status of these individuals was further confirmed by detailed analysis of the 5S rDNA sequences. As recently shown, graph-based clustering of the RepeatExplorer pipeline enables reconstruction of complete 5S rDNA sequences from partial Illumina sequencing data and provides clues to the evolutionary history of interspecific hybrids and allopolyploids [69, 102]. The detailed analysis of the 5S rDNA cluster shapes of all Pulmonaria accessions examined revealed the presence of both P. obscura- and P. officinalis-specific 5S rDNA sequences in the genome of a putative hybrid clone B481.1 (Fig. 4).
Given the overlapping ranges of some Pulmonaria species groups [see 56, 101], the possibility of interspecific hybridization might be expected to be relatively common. So far, however, natural hybrids have only occasionally been identified by chromosome counting [3, 21] or distinguished on the basis of intermediate morphology [e.g. 103]. Questions also remain about the hybrid's fertility and their longevity in populations. Although the hybridization is apparently rare, hybrids can potentially persist and spread clonally at the locality [cf. 104], as Pulmonaria species partly reproduce by vegetative propagation (i.e. creeping rhizomes).
Origin of ornamental cultivars morphologically similar to P. officinalis
As a valuable medicinal and ornamental plant, P. officinalis is represented in horticulture by several cultivars and has also been used to generate new artificial hybrids [cf. 20]. This seems to be the case for plants with distinctly white-spotted leaves, cordate at the base, which are sometimes offered commercially as P. saccharata. However, they are not “true” P. saccharata sensu Miller [see 10, 66, 67]. These plants often escape into the wild and are sometimes confused with P. officinalis. The origin of these cultivars is unknown, they only resemble P. officinalis complex in their morphology (i.e. cordate bases of leaves, Fig. 1D).
In total, we analyzed three populations of these garden escapees (B15, B465 and B472). Our cytogenetic analysis and detailed examination of the reconstructed 5S rDNA sequence indicate that two analyzed P. saccharata-like accessions (B465.1 and B472.1) are derived from the P. officinalis. Both plants had 2n = 16 as P. officinalis. They also shared the general cytogenetic pattern of satDNAs and rDNA sequences typical of P. officinalis, and contain only P. officinalis-specific, not P. obscura-specific, 5S rDNA sequence types (Fig. 4, 5). In addition, the genome size value of the B465.1 plant was also in the range of P. officinalis, only in the B472.1 plant was slightly larger (Table 1). This may confirm the assumption that these plants are derived from P. officinalis.
On the other hand, an interesting cytogenetic pattern was observed in the third P. saccharata-like plant analyzed (B15.1). The karyotype of this plant was similar to that of the interspecific hybrid B481.1, with the same chromosome number 2n = 15 and P. obscura- and P. officinalis-specific 5S rDNA sequences in its genome (Fig. 4). Furthermore, even the cytogenetic pattern of the probes and the presence of P. obscura- and P. officinalis-specific satDNAs (PulTR_305 and PulTR_420) in the haploid state also suggest a hybrid origin (Fig. 5). In contrast, the genome size of B15.1 (the whole population, respectively) was considerably larger, the largest found in the whole data set presented (Table 1). However, unlike the population B481, population B15 was collected in the area where only P. obscura occurs naturally and where no population of P. officinalis has been confirmed (Kobrlová, pers. obs.). The morphology of the plants was also typical for cultivated P. saccharata-like plants. Their origin therefore requires further investigation, although the cytogenetic data presented suggest a hybrid origin between P. obscura and P. officinalis (e.g. phylogenetic revision and analysis of a larger data set of Pulmonaria species from different phylogenetic groups). As this population is a garden escape, its geographical origin is unclear and it cannot be ruled out that it was originally collected from a mixed population of both species. Our data undoubtedly demonstrate that molecular karyotyping is a powerful method for identifying the mode of karyotype evolution and the hybrid origin of Pulmonaria taxa.