Introduction

Sparganium L. (Typhaceae) are aquatic plants widely spread in temperate and subtropical zones of the Northern and Southern Hemispheres1,2. Sparganium species represent an important part of freshwater vegetation. They serve as edificator plants and influence the geomorphology of river courses3,4.

The genus Sparganium is divided into two subgenera, Sparganium (or Melanosparganium) and Xanthosparganium Holmb.1,5, and three sections6. The subgenus Sparganium includes the section Erecta Aschers. et. Graebner, and the subgenus Xanthosparganium includes sections Natantia Aschers. et. Graebner and Minima Aschers. et. Graebner2,5,6.

The number of species and subspecies within the genus Sparganium and their taxonomic rank is the subject of longstanding scientific discussions1,2,7,8,9. Researchers estimate the number of Sparganium species as between 15 and 202,6. Similar to a number of other aquatic genera, some Sparganium species and subspecies lack clear morphological differences1,5,7,9,10,11,12. For this reason, some researchers consider the morphotype/group of certain natural populations in the rank of species, while others assign them to the rank of subspecies or forms only. For example, the taxa S. emersum Rehm. (Eurasia and western North America), S. acaule (Beeby ex Macoun) Rydb. (eastern North America) and S. rothertii Tzvelev (Eurasia) can be considered as individual species12,13 or subspecies of the polymorphic species S. emersum s.l.1,14. Another example is the taxa S. neglectum Beeby and S. microcarpum (Neum.) Domin; they can be treated as a different species5 or subspecies of S. erectum L.9,11.

Intermediate morphological forms are very common in Sparganium. It appears that reproductive isolation between species in the genus is not complete and allows interspecific hybridization1,2,9,15,16,17. Sparganium hybrids can be sterile, partially fertile, or fertile1,17,18,19. The existence of seven hybrid taxa in the genus Sparganium2,9,12,17,20 has now been proved by molecular analysis methods. However, the importance and distribution of interspecific hybridization in the speciation of Sparganium have been poorly studied so far.

A valuable tool for detecting plant hybrids in nature is the study of intragenomic rDNA polymorphism21,22. rDNA fragments are the most commonly used molecular markers in plant phylogeny23,24 thanks to their unique properties. 35S rDNA contains both coding conserved fragments (18S, 5,8S, and 26S rRNA genes) and noncoding highly variable fragments (internal transcribed spacers ITS1 and ITS2). Plants contain from 200 to 22,000 35S rDNA cistrons per haploid genome25 arranged as long tandem repeats on one or more chromosomes26.

First-generation interspecific hybrids always have both parents' rDNA in the genome27,28,29,30. In the next generations, some species retain rDNA31,32,33, while others undergo rDNA homogenization, resulting in the loss of most of the rDNA of one of the parents and only a minor number of 35S rDNA loci remaining in the genome, which can be detected via sequencing cloned PCR products22,28,33,34,35 or by using next-generation-sequencing (NGS)16,21,36,37,38,39,40,41. Analysis of the intraindividual polymorphism of 35S rDNA fragments has been successfully used to study hybrids in various groups of plants29,30,31,32,33. ITS1 and ITS2 fragments are most variable at the intraindividual level42. We chose the ITS1 fragment of the 35S rDNA43 locus as a marker, because it was shown that homogenization was usually lower in ITS1 than in ITS2, suggesting that concerted evolution operates less efficiently on the former44.

The main objective of this study was to investigate the role of hybridization events in the recent evolution of Sparganium. We applied the NGS method to analyze the intragenomic polymorphism of the 35S rRNA gene ITS1 region. Fifteen Sparganium taxa were studied, including both morphologically well-identified species and subspecies, some of likely hybrid origin, as well as plant specimens that exhibited morphological traits intermediate between the two species (i.e., probable hybrids). We set out to address the following questions: (1) Does the molecular analysis prove the hybrid origin of morphologically intermediate forms identified on the basis of the morphological analysis? (2) Can well-identified species have characters of hybrid origin in their genomes? (3) Is the hypothesis of hybrid origin of S. probatovae Tzvelev and S. rothertii put forward by Tsvelev13 validated? (4) Is hybridization between species from different sections and species from different subgenera possible in the genus Sparganium?

Results

Identification and phylogenetic analysis of the major ribotypes

To assess the intragenomic diversity of Sparganium species rDNA, we analyzed ITS1 polymorphism in 37 samples belonging to 15 species and subspecies using Illumina high-throughput sequencing. Four species and subspecies from section Erecta, two species from section Minima, and nine species from section Natantia were studied. The USERCH45, and UNOISE46 algorithms were employed to process Illumina high-throughput sequencing data to identify the unique variants of the ITS1 sequences flanked by 18S rDNA and 5.8S rDNA as well as to calculate their comparative frequency (number of read pairs uniquely mapped to one variant). Obtained denoised sequences are called ZOTUs (zero-radius OTUs)46, or ribotypes21,39,40. ZOTUs with a read count of 10 or more were used in the further analysis. In our study, we divided ribotypes into major ones, which accounted for at least 2% of the reads in the sample, and minor ones. A total of 64 major ribotypes and about 150 minor ribotypes were identified. Data on the occurrence and abundance of each ribotype in the studied plants are presented in Supplementary Table S1, with each major ribotype given a unique name. Some major ribotypes differed from each other by 1–2 SNPs; we designated these variants by letters, combining them into families (e.g., Negle-3A & Negle-3B). The table of SNPs and indels (Supplementary Table S2) shows each family of these variants as a single line. The data for each family are summarized in the graphs (Fig. 2, 3, 4). Some sequences have large indels; for example, Roth-1 and Long-3 have a 70 bp deletion, Negle-4 has a 13-nucleotide deletion, and the sequences Hyper-1 and Hyper-2 contain 3-nucleotide CCC insertions (Supplementary Table S2).

Major ribotypes along with NCBI Sanger sequences of species of Sparganium and Typha genera were used to perform phylogenetic analysis (Fig. 1). In the phylogenetic tree, most of the major ribotypes of each section (Supplementary Table S1 and S2) are clustered together, forming three clades. ITS1 sequences of the subgenus Sparganium appear to be placed closer to the genus Typha, thus occupying a basal position in the genus Sparganium. The longest branch separates two subgenera, Sparganium and Xanthosparganium, and has high bootstrap value. In the subgenus Xanthosparganium, the major ribotypes of section Minima also form one clade with a high bootstrap value. The major ribotypes of section Natantia are clustered together but it is difficult to obtain accurate phylogenetic relationships between them due to low bootstrap support.

Figure 1
figure 1

Phylogenetic tree of Sparganium major ribotypes built by maximum likelihood method. Bootstrap values: green: 80–100, light-green: 60–80, yellow: 40–60, orange: 25–40.

It is worth noting that some species have species-specific ribotypes that are related to the ribotypes of other sections. For example, "Natan-3" ribotype found in S. natans (section Minima) cluster with ribotypes of section Natantia, and "Prob-1" ribotype found in S. probatovae (section Natantia) clusters with ribotypes of section Minima.

Number of major ribotypes across species

The number of major ribotypes detected in the genomes of different species varies (Fig. 2, 3, 4). The species studied can be divided into three groups containing one, two, and more than two major ribotypes. Thus, S. angustifolium Michx., S. fallax Graebn., and S. subglobosum Morong in section Natantia of the subgenus Xanthosparganium (Fig. 2), have one major ribotype. Samples of S. glomeratum Laest. ex Beurl.) Beurl. have both one and two major ribotypes. Chinese samples of S. emersum (No. 4 and No. 5), S. gramineum Georgi, and S. probatovae No. 20 show two major ribotypes (Fig. 2). Samples of S. × longifolium Turcz. ex Ledeb., European S. emersum (No. 6), S. probatovae No. 21 and S. rothertii, have from three to five major ribotypes (Fig. 2). Samples of both species of S. hyperboreum Laest. ex Beurl. and S. natans L. in section Minima of the subgenus Xanthosparganium contain from two to six major ribotypes (Fig. 3). In section Erecta of the subgenus Sparganium, samples of S. erectum subsp. erectum L. have from five to seven major ribotypes, S. erectum subsp. microcarpum (Neuman) Domin from one to three, S. erectum subsp. neglectum (Beeby) Schinz & Thell. from two to seven, and S. stoloniferum subsp. choui (D. Yu) K. Sun contains three ribotypes (Fig. 4).

Figure 2
figure 2

Stacked bar plots showing the relative abundance of the major ribotypes across the species of subgenus Xanthosparganium section Natantia. Minor ribotypes in a sample are collectively termed "Minors".

Figure 3
figure 3

Stacked bar plots showing the relative abundance of the major ribotypes across the species of subgenus Xanthosparganium section Minima. Minor ribotypes in a sample are collectively termed "Minors".

Figure 4
figure 4

Stacked bar plots showing the relative abundance of the major ribotypes across the species of subgenus Sparganium section Erecta. Minor ribotypes in a sample are collectively termed "Minors".

Heterogeneity in ribotype composition

Many ribotypes are not species-specific and occur concurrently in a few species. For example, S. glomeratum from section Natantia has three ribotypes belonging to section Minima. Samples of S. × longifolium contain most of the ribotypes of S. emersum No. 6 and one minor ribotype of S. gramineum. S. rothertii has both S. emersum ribotypes No. 4 and No. 5, and S. emersum No. 6. Samples of S. probatovae have phylogenetically diverse ribotypes belonging to all three sections of the genus Sparganium (Fig. 2). Therefore, S. probatovae No. 20 contains ribotypes found in S. emersum / S. rothertii, the "Prob-1" ribotype of section Minima, which was not observed in other species, a minor number of S. hyperboreum / S. natans ribotypes from section Minima, and a minor number of rDNA of S. erectum subsp. erectum from section Erecta. S. probatovae No. 21 has the ribotypes of S. gramineum, S. emersum / S. rothertii and S. erectum s.l.

The species S. emersum and S. probatovae of section Natantia show considerable intraspecific differences between the samples in terms of rDNA composition. Therefore, sample S. emersum No. 6 from Yaroslavl Oblast, a region in the European part of Russia and samples S. emersum No. 4 and 5 from Inner Mongolia, China, do not have a common ribotype. Samples S. probatovae No. 20 from the Magadan Oblast and S. probatovae No. 21 from the Tyumen Oblast contain a total of six major ribotypes, with only two of them being common (Fig. 2).

Species of section Minima differ greatly from each other in rDNA composition and have only two common ribotypes (Fig. 3). Both species include rDNA ribotypes of section Natantia in their genome, probably originating from S. emersum, with S. hyperboreum also having ribotypes of S. glomeratum. Besides, S. hyperboreum No. 22 shows a minor quantity of S. erectum s.l. rDNA from section Erecta.

The subspecies of S. erectum differ in the set and abundance of ribotypes, with the most notable differences observed in S. erectum subsp. neglectum, which has only one ribotype in minor quantities shared with the other subspecies (Fig. 4). Furthermore S. erectum subsp. microcarpum has a more homogenized rDNA compared to other subspecies. It is worth noting that some samples of S. erectum contain rDNA variants of section Natantia. For example, S. erectum subsp. erectum has minor quantities of S. emersum rDNA, with S. erectum subsp. neglectum containing minor quantities of S. glomeratum rDNA. S. stoloniferum subsp. choui, in addition to species-specific ribotypes, carries ribotypes common to S. eretum subsp. erectum and S. erectum subsp. microcarpum.

Discussion

Organization and evolution of 35S rDNA in plants

Ribosomal 35S rRNA genes in plant genomes exist in a few thousands of copies that are reiterated tandemly at one or more loci which are visible as secondary constrictions or nuclear organizers (NORs) when transcriptionally active26. A noteworthy property of 35S rRNA genes is that the sequences of internal (ITS) and external (ETS) transcribed spacers are highly variable between species yet relatively homogeneous within the species genome and often species-specific47,48. This makes the comparative study of ITS and ETS sequences in different species a fruitful approach in the molecular phylogeny and DNA barcoding47,48,49. It is not well understood how the difference between the repeated 35S rRNA genes of sister species arises in the course of species divergence. This phenomenon is best consistent with the "birth-and-death" model of evolution of tandem repeats according to which one of the multiple copies of the rDNA locus with its typical set of SNPs is multiplied, while the other rDNA variants of this locus disappear40,50,51. The mechanisms for the rapid loss of many rDNA copies and the compensatory amplification of one of the copies are unknown. They are assumed to be in some way correlated with the emergence and repair of DNA double-strand breaks52,53, rolling replication of one of the chromosomal rDNA copies54 and/or extrachromosomal circular rDNA molecules55.

A logical consequence of rDNA homogenization in the plant genome is that only one type of rDNA sequences is detected in the genomes of many species by Sanger genomic sequencing, as observed in Zingeria trichopoda (Boiss.) P.A. Smirn.56, diploid and polyploid Avena L.57, some species of Gossypium L.58, and Brassica L.59. Generally, the phenomenon of nucleolar dominance is observed in hybrids and allopolyploids47,60. Here, the rDNA sequences of the repressed genome rapidly accumulate SNPs and deletions and are gradually lost34,36,38,40,61. However, a great many hybrids and species of hybrid origin show greater or lesser intragenomic heterogeneity of ITS21,22,31,32. There are a few reasons for the heterogeneity. The first is the retention of part of the rDNA ancestral species in hybrids29,30,31,32. The second reason could be pseudogenization of transcriptionally inert rDNA loci with their slow homogenization; it has been shown that in transcriptionally inert rDNA copies the accumulation rate of SNPs and deletions is 10 times higher than in rDNA of transcribed loci38,40,61,62. Homogenization of sequences within a single NOR proceeds faster than in different NORs, where it also occurs at varying rates63,64.

Sparganium species with homogenized rDNA

We assumed that Sparganium species with a single major ribotype and low intragenomic polymorphism, such as S. angustifolium, S. fallax and S. subglobosum, might have diverged earlier than other representatives of section Natantia. This assumption is consistent with the earlier phylogenetic study based on cpDNA and Sanger rDNA sequences2. These species differ greatly from each other in the morphology of the generative sphere. The population system of the species typically consists of geographically or ecologically isolated or semi-isolated populations. It should be emphasized that populations of aquatic plants are considered to be more reproductively isolated in geographical and ecological terms as compared to those of terrestrial plants15,65,66. It is very likely that even if they are of hybrid origin, the process of rDNA homogenization in these plants is complete. It is worth mentioning that most minor variants of S. fallax prove to be phylogenetically close to the major one, except for a small number of distant variants that may testify to ancient hybridizations. Note that only one pair of Ag-NORs loci67 was revealed in the S. fallax genome. Unfortunately, this is the only species of the genus Sparganium with a known number of active NORs. Other species may have from one to two secondary constrictions, as we see in Typha species that are closely related to Sparganium68.

rDNA composition and hybridization in different sections of the genus Sparganium

Section Natantia

The environmental or ecological isolation in the genus Sparganium is often due to the trophic status of water bodies where a particular species occurs. For example, S. gramineum inhabits oligotrophic and oligo-mesotrophic water bodies1,69, while S. angustifolium tends to dystrophic and oligotrophic ones8,70,71. Although the ranges of these species overlap, they generally do not share the same water body. S. emersum usually favors mesotrophic and eutrophic waters, though it can be found in the same water body as S. angustifolium and S. gramineum1,8,69. The latter indicates a wide ecological amplitude of this species. Adjacent habitats of certain species lead to the formation of potential hybrid zones. These zones often produce hybrids in their phenotype combining characters of parental species (S. × engleratum Graebn. (S. angustifolium × S. emersum)2,12 and S. × longifolium (S. emersum × S. gramineum))16,17,19.

The S. emersum samples No. 4 and No. 5 collected in China differed from the European S. emersum sample No. 6 by narrower leaves (with a poorly developed keel) and closely spaced male and widely spaced (often extra-axillary) female capitate inflorescences. The same characters are diagnostic for S. rothertii13. It should be noted that identification of the samples from China was handicapped by the fact that the plants had no above-water leaves (only those floating on the water surface).

Considering the morphological differences between S. emersum European and Baikal-Far East-China populations, Tsvelev13 suggested dividing S. cf. emersum into different species, S. emersum and S. rothertii13. S. rothertii was identified by Tsvelev from S. emersum13 on the basis of the morphological traits of leaves, anthers, styles, and fruits, as well as geographic criteria. However, S. rothertii is currently not a widely recognized species, and Chinese and Japanese researchers treat S. rothertii as S. emersum72,73.

Since the set of rDNA variants of the European S. emersum sample differs fundamentally from the Chinese samples, the latter can be regarded as a different or new cryptic species. Moreover, S. rothertii can be considered as a separate species since its rDNA composition is considerably different from S. emersum, containing ribotypes of both Chinese and European samples. It is worth mentioning that the ribotypes of the European S. emersum sample are also found in S. hyperborem and S. natans (section Minima), as well as in S. erectum subsp. erectum (section Erecta).

Another species with high intragenomic and intraspecific rDNA polymorphism is S. probatovae. This species was described by Tsvelev13. Earlier, collectors referred it to S. hyperboreum Laest., S. minimum Hill (= S. natans), S. simplex Huds (= S. emersum), or S. friesii Beurl. (= S. gramineum)74. According to Tsvelev's13 hypothesis, S. probatovae may have formed as a result of hybridization between S. emersum and S. hyperboreum. The hybrid origin of S. probatovae is proved by the absence of ripe fruits in the available herbarium specimens13,74,75. Other arguments in favor of the hybrid origin of the samples identified as S. probatovae are, as follows: 1) scattered habitats of the collected samples within the overlapping ranges of the supposed parental species74,75 and 2) deviation of the morphological characters towards a particular parental species75.

Six highly diverse ribotypes were found in two studied samples of this species that had similar morphological characters (sample No. 20 from the Magadan Oblast and sample No. 21 from the Tyumen Oblast). At the same time only two of the ribotypes are shared by the two samples. Therefore, these samples apparently have different origins.

S. probatovae sample No. 20 contains the ribotypes of S. rothertii, S. hyperboreum, S. natans, one unique ribotype ("Prob-1") assigned to section Minima on the basis of phylogenetic analysis, and one ribotype found in S. erectum subsp. erectum. Thus, this sample may also have evolved through a number of several rounds of hybridization that probably involved S. rothertii, one of the species of section Minima, as well as S. erectum subsp. erectum.

S. probatovae sample No. 21 contains the ribotypes of S. gramineum, S. rothertii, one unique ribotype ("Prob-2") and one ribotype dominating in S. erectum s.l. from section Erecta of Sparganium subgenus. Based on the data, we can assume that S. probatovae sample No. 21 emerged as a result of several rounds of hybridization involving S. gramineum, S. rothertii and S. erectum s.l.

Thus, the found rDNA diversity suggests a hybrid origin of S. probatovae samples, which is consistent with the morphological and geographical data.

Based on the set of ITS1 variants, S. glomeratum, containing ribotypes of S. hyperboreum, is presumably an intersectional hybrid. Morphological characters also indicate a possible hybrid origin of this species. S. glomeratum has an intermediate position (in terms of style length of the fruit) between section Natantia and section Minima (for example, S. glomeratum has a style length of 1.3 ± 0.2 mm, S. fallax—2.3 ± 0.05 mm, S. hyperboreum—0.3 ± 0.1 mm) and features closely spaced upper pistillate inflorescences often observed in section Minima representatives.

It was previously shown that S. ×longifolium combines the characters of the parental species, S. emersum and S. gramineum17,18,19. The hybridogenic nature of this species has now been proven by molecular methods17. According to our data, S. ×longifolium contains at least four ribotypes of S. emersum No. 6, including two major ones. At the same time, the rDNA of S. gramineum is contained in only one sample of S. ×longifolium in minor amounts. This may indicate that S. ×longifolium is an ancient hybrid, and the rDNA of the S. gramineum was repressed and later lost. Yu et al.17 showed that hybridization between S. emersum and S. gramineum is asymmetrical, and S. gramineum mainly acts as maternal species. Thus, we can assume that paternal rDNA is predominantly present in the samples studied.

Compared to its ancestors, S. ×longifolium has a greater ecological amplitude of occurrence, i.e., in the context of adaptation it is superior to the assumed parental species (based on our data). This allows it to occupy new habitats and sometimes even displace "parental" species from their native ones19. It is our observation that S. ×longifolium is generally sterile and propagates vegetatively. However, the presence of a small number of fertile populations of this hybrid18,19 appears to testify to the potential formation of backcrosses, with some of them being fertile.

Section Minima

Both species of section Minima show clear differences from each other in the set of ribotypes and are highly heterogeneous. Of all the ribotypes identified in this section, only two are common to both species. Thus, our data do not evidence hybridization within the section. We found the remarkable differences in the set of ribotypes between S. hyperboreum and S. natans as unforeseen, since earlier specimens with intermediate morphological characters occurred in nature in the areas where the ranges of S. hyperboreum (circumpolar hypoarctic species) and S. natans (circumpolar boreal species) overlapped18,20,76,77. We believed that the morphologically intermediate specimens might have resulted from introgressive hybridization1,14,76. Based on a very different set of rDNA in S. hyperboreum and S. natans, we developed a new hypothesis that plants of section Minima featuring a "mosaic" of species-specific traits may not be introgressive hybrids but rather a manifestation of Vavilov's law of homologous series in hereditary variability78,79.

It appears from the presence of ribotypes common to section Natantia that S. hyperboreum and S. natans are of hybrid origin, which involved species of this section. Some of the ribotypes are shared with S. emersum and S. glomeratum. S. natans has also two ribotypes phylogenetically related to the section Natantia, though not found in other species. They may have come from a species that was not included in our sampling. Furthermore S. hyperboreum, has a minor ribotype predominant in section Erecta, suggesting possible hybridization with representatives of subgenus Sparganium.

As follows from the above, reproductive barriers between some species of the subgenus Xanthosparganium are weakened, and when the ranges overlap, transfer of genetic material occurs.

Section Erecta

We studied S. stoloniferum subsp. choui and three subspecies of S. erectum from section Erecta of the subgenus Sparganium. These taxa cannot be differentiated by morphological criteria in vegetative and flowering state and may be distinguished only by ripe fruits1,5,9.

rDNA of the representatives of this section is not well homogenized. S. erectum subsp. microcarpum is the only exception, showing one to three major ribotypes across specimens. S. erectum subsp. neglectum differs greatly from the other two subspecies of S. erectum in the set of ribotypes, and share only one common ribotype. S. erectum subsp. erectum and S. erectum subsp. microcarpum are closer to S. stoloniferum, having several common ribotypes. Note that S. erectum subsp. erectum and S. erectum subsp. neglectum each have one minor ribotype from the section Natantia. The former probably inherited it from S. emersum and the latter from S. glomeratum. Minor quantities of rDNA of section Natantia in S. erectum s.l. may indicate ancient hybridization between the subgenera.

It is rather difficult to make any specific assumptions about possible hybrids and their parents from the set of ribotypes within this section. However, it should be noted that the presence of verified hybrid combinations (S. erectum subsp. erectum× S. erectum subsp. microcarpum and S. erectum subsp. microcarpum× S. erectum subsp. neglectum)9 may testify to possible hybridization of S. erectum subsp. microcarpum with other subspecies. Additionally, the presence of morphologically similar fruits in S. stoloniferum specimens from the Rostov, Volgograd, and Astrakhan oblasts (in our observations) may be evidence for the possible hybrid combination of S. stoloniferum× S. erectum subsp. erectum in the Astrakhan oblast.

Píšová and Fer9 concluded that the barriers of reproductive isolation between subspecies of S. erectum. have not yet been finally determined. However, according to our observations, all these subspecies often occupy different ecological niches in water reservoirs. For example, in the European part of Russia, S. erectum subsp. microcarpum is predominantly a cold-water (European-Eastern Siberian plurizonal) species occurring in small rivers, whereas S. erectum subsp. erectum (European-Siberian boreo-meridional species) preferably inhabits well-heated still and slowly flowing waters (water storage basins, ponds, etc.). S. erectum subsp. neglectum is a more southern (European-Pereasian-North African sub-oceanic) species found both in plain areas of cool southern rivers and streams and in well-heated marshy reservoirs. Differentiation of ecological niches leads to ecological isolation of these subspecies and can be an important factor of further speciation.

The existence of already proved hybrids9, as well as the presence of common ribotypes in the subspecies S. erectum and S. stoloniferum may testify to a more or less regular gene flow between these taxa. In our opinion, the differences in the set of ITS1 variants, different genome sizes and verified hybrid combinations, as well as ecological and geographic patterns of distribution, may serve as an argument for attributing S. erectum subspecies to the status of individual species.

Hybridization within and between subgenera

Some species contain in their genomes rDNA of several sections at once. In particular, within the subgenus Xanthosparganium, S. glomeratum and S. probatovae No. 20 from the section Natantia have ribotypes of section Minima. Whereas S. hyperboreum and S. natans have ribotypes of S. emersum from the section Natantia.

Two Sparganium subgenera also have common ribotypes. Thus, S. probatovae and S. hyperboreum from the subgenus Xanthosparganium contain minor amounts of S. erectum s.l. ribotypes from the subgenus Sparganium. On the other hand, some specimens of S. erectum s.l. contain minor amounts of rDNA of S. emersum and S. glomeratum from subgenus Sparganium.

Note that while some researchers doubted the possibility of the formation of hybrids between representatives of both subgenera (e.g., S. emersum × S. erectum s.l.)10,80, others did not deny it6,11,81.

Conclusion

A comparative analysis of the intragenomic polymorphism of the 18S rDNA-ITS1-5.8S rDNA region in 15 Sparganium species and subspecies has showed that the taxa vary in the number and composition of ITS1 variants. Specimens of S. probatovae contain different sets of diverse ribotypes, therefore they can be considered as hybrids of different origin. The Chinese specimens of S. emersum have a principally different set of ribotypes compared to that of European and can be considered as a distinct or new cryptic species. S. rothertii differs from all S. emersum specimens, having ribotypes of both populations and so can be treated as a separate species. Furthermore we suggest that subspecies of S. erectum can be treated as separate species as well.

Species such as S. angustifolium, S. fallax and S. subglobosum have a homogenized rDNA bearing one major ribotype. Species with two major ribotypes include S. gramineum and Chinese specimens of S. emersum. Most taxa, however, contain three or more major ribotypes. Taxa with two or more major ribotypes are probably of hybrid origin. Their reproductive barriers may be weakened, particularly when the ranges of the taxa overlap. In most cases, hybridization is observed between the taxa at the subgeneric level. The fact that each subgenus contains a minor amount of rDNA from another subgenus implies the possibility of rare hybridization events between them.

Methods

Nomenclature and sampling

We have studied a total of 37 specimens belonging to 15 species and subspecies of Sparganium collected in Eurasia (Supplementary Table S3, Fig. 5). Six specimens of two species were selected from the section Minima: S. hyperboreum Laest. ex Beurl., S. natans L. 21 specimens of 9 species were sampled from the section Natantia: S. angustifolium Michx., S. emersum Rehm., S. fallax Graebn., S. glomeratum (Laest. ex Beurl.) Beurl., S. gramineum Georgi, S. probatovae Tzvelev, S. rothertii Tzvelev, S. subglobosum Morong, S. × longifolium Turcz. ex Ledeb. 10 specimens of 4 species and subspecies were sampled from the section Erecta: S. erectum L. subsp. erectum, S. erectum subsp. microcarpum (Neuman) Domin, S. erectum subsp. neglectum (Beeby) K. Richt.), S. stoloniferum subsp. choui (D. Yu) K. Sun. Previously S. stoloniferum was treated as a subspecies S. erectum subsp. stoloniferum (Graebn.) H. Hara due to its fruit morphological features11. But we accept the S. stoloniferum subsp. choui as subspecies of S. stoloniferum based on the recent molecular data82. The herbarium specimens are stored in the Herbarium of Biological Faculty of the Moscow State University (MW), Herbarium of the Institute of Biology of Inland Waters RAS (IBIW), Skvortsov Herbarium of the Main Botanical Garden, Russian Academy of Sciences (MHA) and the personal collection of E.A. Belyakov (PCB). Information on collection locations, collectors, and other collection details of the samples studied is given in the Supplementary S3.

Figure 5
figure 5

The Map showing location (black dots with numbers) of the Sparganium specimens analyzed in this study. Specimens are numbered as in Supplementary S3. (The basis of the maps were taken from d-maps.com (http://data.danetsoft.com/d-maps.com/) Europe (A) (https://d-maps.com/carte.php?num_car=2233&lang=en) and Asia (B) (https://d-maps.com/carte.php?num_car=55&lang=en), the image is changed using Free Software for Digital Photo Editing Paint.NET 4.3.11 https://paintnet.ru/download/).

DNA isolation, NGS sequencing and data analysis

DNA from herbarium specimens was extracted using the CTAB-method83. All specimens were divided into three parts and each part was processed independently (See Supplementary S4). ITS1 library preparation and pair-end sequencing on the Illumina MiSeq were performed at the Core Centre of “Genomic Technologies, Proteomics and Cell Biology” of the All-Russia Research Institute for Agricultural Microbiology (St. Petersburg, Russia) in turn for each of the three parts. The obtained reads were processed using the USEARCH 11.0 software45 according to the recommended protocol. In brief, all samples were pooled, pair-end reads were merged, trimmed and quality filtered according to the maximum expected error threshold of 1.0. Then reads were dereplicated and denoised by the UNOISE3 algorithm46 accompanied by the removal of chimeric sequences. Obtained ZOTUs with a size of at least 10 were blasted against GenBank Sparganium records to filter out off-target sequences and used to construct the ZOTU table. FASTA file containing the ZOTU sequences that were used to construct the ZOTU table is given in Supplementary S5. Table cells that had less than 15 read counts were nulled and the percentage of each ribotype in the total number of reads per specimen was calculated for all specimens. Ribotypes were divided into minor and major with a threshold value of 2%. Supplementary S6 contains FASTA file with the major ribotypes.

Thirty-two NCBI Sanger sequences of the genera Typa and Sparganium were added to the dataset (Supplementary S7). Sequence alignment was obtained by using MUSCLE84 and then checked manually in MegaX85. The maximum likelihood phylogenetic tree was built in IQ-TREE v.1.6.186 under the best-fitted model with 1000 bootstrap replicates and visualized in iTOL87.

Experimental research and field studies on plants, including the collection of plant material, comply with relevant institutional, national, and international guidelines and legislation.