Origin of the Rare Hybrid Genus ×Trisetokoeleria Tzvelev (Poaceae) According to Molecular Phylogenetic Data

In our article, we analyzed new data on the origin of the hybrid genus ×Trisetokoeleria. According to the morphological criteria ×T. jurtzevii is a hybrid between Koeleria asiatica s. l. and Trisetum spicatum, ×T. taimyrica, and originated from Koeleria asiatica s. l. and Trisetum subalpestre, ×T. gorodkowii, a hybrid between Koeleria asiatica and Trisetum ruprechtianum. Later ×T. taimyrica was transferred to Koeleria. Parental taxa are prone to active hybridization themselves, thus, new methods of next-generation sequencing (NGS) were needed to clarify the relationships of these genera. For NGS we used the fragment 18S rDNA (part)–ITS1–5.8S rDNA (totally 441 accessions). We analyzed ITS1–5.8S rDNA–ITS2 region, trnL–trnF and trnK–rps16 from eight samples of the five species, using the Sanger method: ×Trisetokoeleria jurtzevii, ×T. taimyrica, Koeleria asiatica, Sibirotrisetum sibiricum (=Trisetum sibiricum), and Trisetum spicatum. We also studied the pollen fertility of ×Trisetokoeleria and its possible progenitors. Our data partly contradicted previous assumptions, based on morphological grounds, and showed us a picture of developed introgression within and between Koeleria and Trisetum. ×T. jurtzevii, a totally sterile hybrid formed rather recently. We can suppose that ×T. jurtzevii is a hybrid between K. asiatica and some Trisetum s. str. Species, but not T. spicatum. ×T. gorodkowii, a hybrid in the stage of primary stabilization; it has one unique ribotype related to T. spicatum s. l. The second parental species is unrelated to Trisetum ruprechtianum. ×T. taimyrica and is a stabilized hybrid species; it shares major ribotypes with the T. spicatum/T. wrangelense group and has a minor fraction of rDNA related to genus Deyeuxia s. l.

The subtribe Koeleriinae itself has a complex evolutionary history. Previous research established intermingling of the Trisetum and Koeleria in molecular phylogenetic schemes, as well as separation of some of the Trisetum s. l. species within the subtribe [6][7][8][9]. The members of this subtribe, along with all genera of the Aveneae chloroplast group within the Poeae tribe [1,2], are prone to multiple hybridizations.
So, the hybrid nature of the genus ×Trisetokoeleria was assumed on the basis of morphological characteristics. However, molecular phylogenetic analysis that could confirm or refute this assumption has not yet been performed. Only one species, ×T. taimyrica, was studied in previous molecular phylogenetic analyses and, as a result of this work, the species was assigned to the genus Koeleria [8].
Our goal was to find out which species may have participated in the formation of the species of ×Trisetokoeleria and whether the DNA sequence data were consistent with morphological observations.
To identify the origin of the hybrid taxa, the ITS region, as well as chloroplast sequences, were mostly used. In our case, we studied the intragenomic variability of the transcribed spacer ITS1 using Illumina high-throughput next-generation sequencing (NGS). This approach is suitable for revealing the hidden polymorphism of hybrid taxa in the case of multiple crossing [10][11][12]. In addition, we took into analysis ITS1-5.8S rDNA-ITS2 sequences of the nuclear genome, as well as trnL-trnF and trnK-rps16 sequences of the chloroplast genome, obtained by the Sanger method.

Results
Marker sequences obtained via NGS comprised 5 -18S rDNA, partial sequence-ITS1-5.8S rDNA, partial sequence. In total, they had 334 nucleotide positions. Our data on herbarium specimens and results of NGS are shown in Table 1, chromosome numbers of the studied species are given according to [13]. After processing, the ITS1 sequences were sorted into variants, so called ribotypes [14,15], each of which corresponded to a single sequence with a certain number of reads per the whole rDNA pool. The major ribotypes (more than 1000 reads per rDNA pool of amplicons) are shown in Table 1. Table 1. Summary of the possible parental taxa and three Trisetokoeleria species used for next-generation sequencing in the present study and their ribotypes. Chromosome numbers are given according to [13]. The sequences obtained in our study by the Sanger method are given in Table 2. Primary structure of the major ribotypes is given in Table 3. Table 3. Primary structure of the major ribotypes obtained by NGS. The minor ribotypes (55-G15 and 14-G15 (3)) are also provided, representing distant hybridization. As the reference sequence we took the main ribotype of Trisetum spicatum-possible progenitor of ×Trisetokoeleria species. D is a deletion.  We observed that the ribotypes had few distinguishing positions (SNPs) but were group-specific. The ribotype network built in TCS 2.1, and visualized in TCS BU, is shown in Figure 1.
Minor variants and all singletons were derivatives of the major variants (ribotypes) (Figure 1). Table 3 shows two minor ribotypes of ×Trisetokoeleria taimyrica, 55 and 14 reads, respectively, which represented a minor fraction in the hybrid genome, probably related to those of the genus Deyeuxia Clarion ex P.Beauv. s. l. (incl. Cinnagrostis Griseb.). We took as the consensus sequence that of Trisetum spicatum, one of the presumed ancestral species of the hybrid ×Trisetokoeleria. Major ribotypes were divided into those that were common for different species and those that were species-specific ( Figure 1, Table 3). Trisetum spicatum, which had the main ribotype, T1, shared this ribotype with both T. wrangelense (21,565 reads, 78% per its rDNA pool) and ×Trisetokoeleria taimyrica (14,574 reads, 76%). The T1 was the only major ribotype of ×T. taimyrica. In addition, the ribotype T1 was present in the genome of Koeleria asiatica (1505 reads, 7%). Trisetum subalpestre, the possible ancestor of ×T. taimyrica, on the contrary, had the unique major ribotype, Ts (12,538 reads, 47%), and the second major ribotype T2 (8685 reads, 33%), also found in the genomes of Koeleria asiatica (12,064 reads, 54%) and ×T. jurtzevii (2824 reads, 19%). The species Trisetum ruprechtianum (=T. sibiricum subsp. litorale) was only distantly related with the other studied species, and its two major ribotypes, T3 and T4, were unique ( Table 1). The ×Trisetokoeleria gorodkowii had one major ribotype T5 (3492 reads, 29%). This ribotype also occurred in the minor rDNA fraction of Trisetum spicatum and T. wrangelense (18 and 12 reads, respectively). The species ×Trisetokoeleria jurtzevii had the unique main ribotype Tj (7090 reads, 48%). The neighbor-joining phylogenetic tree ( Figure 2) showed almost the same results with an obviously separate clade of Trisetum ruprechtianum and ×Trisetokoeleria jurtzevii major ribotypes. We observed that the ribotypes had few distinguishing positions (SNPs) but were group-specific. The ribotype network built in TCS 2.1, and visualized in TCS BU, is shown in Figure 1.  Table 1. Major ribotypes are larger than others and marked with numbers 1-8. The smallest circles correspond to ITS1 variants that have been read fewer than 1000 times.
Minor variants and all singletons were derivatives of the major variants (ribotypes) (Figure 1). Table 3 shows two minor ribotypes of ×Trisetokoeleria taimyrica, 55 and 14 reads, respectively, which represented a minor fraction in the hybrid genome, probably related to those of the genus Deyeuxia Clarion ex P.Beauv. s. l. (incl. Cinnagrostis Griseb.). We took as the consensus sequence that of Trisetum spicatum, one of the presumed ancestral species of the hybrid ×Trisetokoeleria. Major ribotypes were divided into those that were common for different species and those that were species-specific ( Figure 1, Table 3). Trisetum spicatum, which had the main ribotype, T1, shared this ribotype with both T. wrangelense (21,565 reads, 78% per its rDNA pool) and ×Trisetokoeleria taimyrica (14,574 reads, 76%). The T1 was the only major ribotype of ×T. taimyrica. In addition, the ribotype T1 was present in the genome of Koeleria asiatica (1505 reads, 7%). Trisetum subalpestre, the possible ancestor of ×T. taimyrica, on the contrary, had the unique major ribotype, Ts (12,538 reads, 47%), and the second major ribotype T2 (8685 reads, 33%), also found in the genomes of Koeleria asiatica (12,064 reads, 54%) and ×T. jurtzevii (2824 reads, 19%). The species Trisetum ruprechtianum (=T. sibiricum subsp. litorale) was only distantly related with the other studied species, and its two major ribotypes, T3 and T4, were unique ( Table 1). The  Table 1. Major ribotypes are larger than others and marked with numbers 1-8. The smallest circles correspond to ITS1 variants that have been read fewer than 1000 times.
The ribotypes of ×T. taimyrica were intermingled with those of Trisetum wrangelense and T. spicatum ( Figure 2). The second major ribotype of Koeleria asiatica and derivatives of this ribotype formed separate branches, as well as those of T. wrangelense. All ribotypes of Trisetum ruprechtianum (a member of T. sibiricum = Sibirotrisetum sibiricum group) belonged to the single group sister of all the other species. The neighbor joining network constructed in SplitsTree 4.18 demonstrated a rougher pattern where the ribotypes of ×Trisetokoeleria jurtzevii, ×T. gorodkowii, Trisetum ruprechtianum, and some ribotypes of Koeleria asiatica, were obviously separated ( Figure 3). same results with an obviously separate clade of Trisetum ruprechtianum and ×Trisetokoeleria jurtzevii major ribotypes.     (Tables S1-S3, Supplementary). We used only sequences of ×Trisetokoeleria jurtzevii and ×T. taimyrica in this analysis. According to the tree constructed on sequences of the chloroplast region trnL-trnF, the species of Trisetum, Koeleria, and ×Trisetokoeleria formed a large polytomy within the clade, comprising the taxa of the subtribe Koeleriinae Asch. and Graebn. s. l. (Figure 4).  The tree constructed on ITS data ( Figure 6) shows us a slightly different patte though the Koeleriinae clade was retained in all analyses.

Pollen Fertility Analysis
Our acetocarmine staining of pollen grains of × Trisetokoeleria species and their possible progenitors gave the following results (Table 4).

Pollen Fertility Analysis
Our acetocarmine staining of pollen grains of ×Trisetokoeleria species and their possible progenitors gave the following results (Table 4). The ×T. jurtzevii yielded completely sterile pollen, while ×T. taimyrica had not much more sterile pollen than the non-hybrid species, and ×T. gorodkowii was intermediate with 45% abortive pollen cells. Non-hybrid, possible parental species had a low proportion of abortive pollen (Table 4).

Overview of the Genus
The ×Trisetokoeleria is an example of a hybrid genus in the grass family. The hybrid genera are quite rare among the members of the tribe Poeae s. l. and more common in the tribe Triticeae Dumort. Nevertheless, processes of hybridization occurring within the tribe Poeae s. l. are active, and they are often detected by modern molecular phylogenetic methods.
As a result of multiple reticulation, the entire Trisetum genus was relatively recently regarded as polyphyletic with different clades of Trisetum s. l. uniting with many small genera of the subtribe Koeleriinae [6][7][8][9]. In fact, this multiple division of the genus tells us that its species could be part of an introgressive-interspecies complex, a hybrid swarm [16][17][18][19][20], where morphological similarity of its members is a result of genome absorption in a series of introgressive hybridizations.
The evolution of the hybrid genus ×Trisetokoeleria may be related to the high latitudes of the Northern Hemisphere. Despite the fact that its possible parental species also grow together in Asian mountains, e.g., in the Altai-Sayan region, we have not seen any occurrences of hybridization between Trisetum and Koeleria (both in a narrow sense) outside the Arctic. Arctic conditions are extreme and, thus, facilitate the processes of polyploidization and hybrid speciation for adaptation and surviving [21,22]. The ×Trisetokoeleria species are neopolyploids, in terms of their hybrid state [23][24][25], formed rather recently and probably reproduced mostly vegetatively, at least two of them did. As we see from the pollen fertility analysis, ×T. jurtzevii is a sterile hybrid, and ×T. gorodkowii is most likely in the early stages of hybrid stabilization. However, our data on their origin showed some differences from the evolutionary hypotheses based on morphological criteria.

Origin of Each Species According to the Molecular Data
The nothospecies ×T. jurtzevii had one main ribotype that was not shared with any other of the studied species (Figure 1). It clearly differed from the main ribotype of Trisetum spicatum obtained in our research and ITS sequences taken from GenBank data. As the second major ribotype of ×T. jurtzevii was common with Koeleria asiatica, we could suppose that ×T. jurtzevii originated from the intercrossing of K. asiatica s. l. and some Arctic race of Trisetum close to T. spicatum affinity. From the trnL−trnF gene analysis (Figure 4) we could see that some relative of T. spicatum probably gave the maternal genome to ×T. jurtzevii. The results of the analysis of the ITS sequences (Figure 6), obtained by the Sanger method (longer region than studied by NGS), revealed that ×T. jurtzevii was unrelated to any other species of Trisetum and Koeleria within the common clade of most of the Koeleriinae members. This allowed us to assume not only possible post-hybridization transformation of the sequences via intergenomic conversion [26], but also significant genetic difference between Arctic and Siberian mountain Trisetum species. In addition, one of the ×T. jurtzevii ancestral taxa could probably be extinct, and the hybrid persisted for a long time via vegetative propagation. The fact that in the samples from geographically distant locations different sequences from the whole rDNA pool were amplified by the Sanger method could be the result of polyploidy of T. spicatum and allied species (mostly 2n = 28- [13,[27][28][29][30]).
The ×Trisetokoeleria gorodkowii had only one main ribotype that was shared with Trisetum spicatum and T. wrangelense (the minor fraction in their genomes). According to the marker sequences obtained by NGS, there was no connection between this hybrid and Trisetum ruprechtianum, which most likely belonged to a separate line now named Sibirotrisetum [8]. The ×T. gorodkowii was first described as Koeleria gorodkowii Roshev [31], belonging to the section Caespitosae Domin related to K. pyramidata affinity. From a morphological point of view, it was more reminiscent of the species of Koeleria than those of Trisetum. Our NGS and pollen analyses results showed that ×T. gorodkowii was most probably a modern hybrid, having sequences close to the Koeleria line retained in the allopolyploid genome. However, ×T. gorodkowii was not related to K. asiatica s. l. and, therefore, might be a derivative of the K. pyramidata group instead. The second parental taxon of ×T. gorodkowii was probably some relative of Trisetum spicatum group (Arctic races) but not Trisetum ruprechtianum or any other ally of the newly separated genus Sibirotrisetum [8]. Marker sequences of the second parental species could be deleted from the allopolyploid genome.
The ×Trisetokoeleria taimyrica had one main ribotype common with Trisetum spicatum and T. wrangelense. According to NGS data, this ribotype was unrelated to the ribotypes of T. subalpestre. The analysis of the ITS sequences performed by the Sanger method demonstrated a similar result: ×T. taimyrica was close to T. spicatum but formed a separate subclade on the tree (Figure 6), not connected with Koeleria asiatica. In addition, the chloroplast sequences of the ×T. taimyrica group with Trisetum spicatum and Koeleria asiatica (trnK−rps16 data, Figure 5) also fell within the separate subclade. Pollen fertility in ×T. taimyrica was more pronounced than in ×T. gorodkowii; this might be due to the stabilized hybrid status of the species, allowing it to propagate in seeds as well as in vegetative clones. Recent molecular phylogenetic research placed ×T. taimyrica within the genus Koeleria because of the ITS and cpDNA sequence data showing close affinity of Trisetum spicatum to the Koeleria members [8]. Nevertheless, our ITS data showed more close affinity of ×T. taimyrica to the T. spicatum group than to the most part of the genus Koeleria. According to the NGS data, the main ribotype of ×T. taimyrica also belonged to the Trisetum group and was not closely related to the ribotypes of Koeleria. We tended to retain the hybrid status and generic name ×Trisetokoeleria for ×T. taimyrica although the hybridization in this taxon could have occurred earlier in its evolutionary history, according to the pollen staining data.
Thus, we could assume that ×Trisetokoeleria taimyrica was probably a stabilized introgressive hybrid between Trisetum wrangelense (arctic member of T. spicatum group) and Koeleria asiatica (probably even from multiple introgressions). In addition, ×Trisetokoeleria taimyrica had two minor ribotypes in its rDNA pool that reflected hybridization with some more distant relative (Figure 1). It was close to the ITS sequences of Peyritschia deyeuxioides (Kunth) Finot and Cinnagrostis viridiflavescens (Poir.) P.M.Peterson, Soreng, Romasch. and Barberá (GenBank data). These Central and South American species belong to the separate line including Deyeuxia and Calamagrostis Adans. genera (former Calamagrostis s. l., Figures 4-6). Of course, we could not say with exactitude that these species formed hybrid ×T. taimyrica, but it seemed that some lineage of Calamagrostis s. l. (probably not only one species) might have been involved in the formation of this hybrid. Affinity between arctic and southern species may be an interesting fact that brings to mind the interpolar disjunction in grass evolution [32][33][34][35][36].
Presumably, the species carrying the genome that became the parent of ×T. taimyrica, related to the southern group of Calamagrostis s.l., could have migrated to South America at first through Beringia and, later, further along the Cordillera mountain range. We also could not exclude the possibility that this ancestral species that took place in ×T. taimyrica formation may have become extinct recently, because modern Calamagrostis and Deyeuxia species from boreal and arctic regions usually have ITS sequences unrelated to those of southern species.

Allopolyploid Parental Taxa of ×Trisetokoeleria and Problem of Delimitation of the Genera
It is important to pay attention to the possible hybrid status of the parental taxa of ×Trisetokoeleria. In modern classifications, based on molecular phylogenetic analyses of different nuclear and chloroplast genes of Trisetum, the section Trisetaera (Asch. and Graebn.) Honda to which T. spicatum belongs was moved to the genus Koeleria [8,20]. This decision could be supported by the similarity of some morphological characteristics of these taxa, such as dense and narrow panicles with hairy branches, and, sometimes, subequal glumes [8,37]. Nevertheless, the genus Trisetum (even in the broader sense, including the members of sect. Trisetaera), also has morphological features that are clearly different from the genus Koeleria. The first, and one of the main distinguishing features, is the awn on the lemma. In the genus Trisetum sect. Trisetaera the species have dorsal well-developed lemma awn usually inserted slightly below the apex [8]. The awn is bent basally to sub-basally [8]. In addition, the lemmas of Trisetum have two awn-like teeth on the tip. Koeleria, in its turn, usually has only slightly acute lemma apex without any awns. In some rare cases, the species of Koeleria have very short mucro right on the lemma apex. Lemma callus in the genus Trisetum is acute, whereas that of Koeleria is obtuse [13]. Molecular phylogenetic data (Figures 4-6, see also [9,20]) showed multiple hybridization events in this group. For example, Trisetum spicatum fell into the clade with ×Trisetokoeleria taimyrica, Trisetum oreophilum, T. montanum, T. phleoides, T. andinum, and one Koeleria species, K. capensis, but was distant from K. asiatica, according to the ITS analysis ( Figure 6). Chloroplast gene data (trnL-trnF), on the contrary, placed Trisetum spicatum from the Altai Mountains with ×Trisetokoeleria jurtzevii and T. spicatum, from Mexico (GenBank accessions), formed the separate clade with T. montanum and T. oreophilum (Figure 4). The maternal genome of Trisetum spicatum could be related to Koeleria asiatica (trnK-rps16 data), though not closely ( Figure 5). All sequence data placed T. spicatum rather distantly from Koeleria pyramidata. The ITS and chloroplast sequences clearly showed the possibility of multiple introgressive hybridizations in the sect Trisetaera, since different samples of the same species gave different phylogenies. The ITS1 sequences, obtained by NGS, showed that the main ribotype of Trisetum spicatum was common to that of T. wrangelense and with only minor ribotype fraction of Koeleria altaica. We need to note that Trisetum spicatum was unrelated to the type species of the genus, Trisetum flavescens. However, in its turn, T. flavescens formed a clade with Rostraria pumila, based on both chloroplast and ITS sequences. Previous researchers did not unite Trisetum with annual grasses of the genus Rostraria, pointing to the developed hybridization within the subtribe Koeleriinae [9,20]. Of course, multiple introgression events could take place in our case, in the clade with Trisetum sect. Trisetaera, other Trisetum species (T. drucei, T. preslii), and Trisetaria Forssk. We see here a phylogenetic picture that is a reminder of intergeneric hybrid speciation in the tribe Triticeae. For example, the genus Elymus L. s. l. originated from the intercrossing of St, H, and Y-genome bearers belonging to different genera [38,39]. New taxonomic treatment of Elymus and its relatives proposed generic names based on the genome combinations that the species has: Elymus s. str. (StH), Roegneria K.Koch (StY), Campeiostachys Drobov (StYH), etc. [40,41]. At the same time, genome combinations in the subtribe Koeleriinae are unknown to us yet. Thus, we cannot estimate contribution of each genome in the formation of the allopolyploid Trisetum species as well as in its allies. For our convenience, and taking into account multiple reticulation events between the species and in main lineages of the subtribe Koeleriinae [20], we preferred to retain the previous classification which separates the most part of Trisetum (including sect. Trisetaera) and Koeleria. We also distinguished the specific clade representing the genus Sibirotrisetum. We need to mention that morphological traits in this case more or less reflected the genome combinations.
As can be seen, Koeleria asiatica, as well as Trisetum spicatum (sect. Trisetaera) and T. subalpestre (sect. Agrostidea Prob. close to the section Trisetaera- [8]), were most prob-ably of introgressive hybrid origin (Figures 1-3). K. asiatica had the main ribotype T2 shared with T. subalpestre (the second major ribotype), but also in K. asiatica there was one unique ribotype Ka that was far less frequent in the genome (7%, Figure 1). In addition, K. asiatica had a minor fraction of T1 ribotype that was characteristic for Trisetum spicatum and T. wrangelense. Thus, we could assume that K. asiatica was an allopolyploid from intercrossing of T. subalpestre and some Koeleria species, probably with little participation of T. spicatum s. l. According to our analysis, K. asiatica certainly belonged to the eupolyploid (mesopolyploid) stage of polyploid evolution when the karyotype in meiosis behaves as a classic diploid without any failures, but the parental genomes can still be separated cytogenetically [24].

Conclusions
Our NGS data, together with chloroplast and nuclear sequences obtained by the Sanger method, gave us some new and important information about the origin and possible parental taxa of three species of nothogenus ×Trisetokoeleria from the Poeae tribe. This information partly contradicted morphological assumptions. However, we can also say that the NGS data helped clarify some previous hypotheses built on previous molecular phylogenetic results (because the nuclear sequences obtained by the Sanger method may have lower ability in distinguishing nucleotide positions, due to conversion in allopolyploid genomes). Looking at the polyploid members of Trisetum/Koeleria affinity we needed to take into account possible multiple cases of introgression. Nevertheless, we tended to retain the previous generic names for the most part of this tribe. All species of these taxa may be intergeneric hybrids, and the morphological features used in delimitation of generic borders depended on which proportion of the parental genomes was present in them. Our methods of molecular phylogenetic analysis, along with analysis of pollen fertility, allowed us to successfully assume the order of the formation of nothospecies. The ×T. jurtzevii is the most recent sterile hybrid, originated from Koeleria asiatica and some northern species of Trisetum (not T. spicatum s. str.), whereas ×T. gorodkowii is a hybrid in the stage of primary stabilization, probably formed not long ago from some Trisetum species unrelated to T. ruprechtianum, and ×T. taimyrica is a stabilized ancient hybrid species, originated from Koeleria asiatica and Trisetum wrangelense.

Molecular Phylogenetic Analysis
In our study, we used 441 sequences of 18S-ITS1-5.8S rDNA obtained via NGS from eight species: ×Trisetokoelria gorodkowii, ×T. jurtzevii, ×T. taimyrica and their putative ancestors: Koeleria asiatica, Trisetum ruprechtianum, T. spicatum, T. subalpestre, T. wrangelense (V.V.Petrovsky) Prob. All our samples were taken from the Herbarium of Komarov Botanical Institute (LE). Information on these species with their accession numbers and ribotypes is presented in Table 1.
Genomic DNA was extracted from seeds using a Qiagen Plant Mini Kit (Qiagen Inc., Hilden, Germany), according to the instruction manual. NGS was carried out at the Center for Shared Use "Genomic Technologies, Proteomics and Cell Biology" of the All-Russian Research Institute of Agricultural Microbiology on an Illumina Platform MiSeq. We used 15 µL of PCR mix containing 0.5-1 unit of activity of Q5 ® High-Fidelity DNA Polymerase (NEB, Ipswich, MA, USA), 5 pM of forward and reverse primers, 10 ng of DNA template, and 2 nM of each dNTP (Life Technologies, ThermoScientific, Waltham, MA, USA). The PCR was carried out using primers ITS 1P [42] and ITS 2 [43] under the following parameters: initial denaturation 94 • C for 1 min, followed by 25 cycles of 94 • C for 30 s, 55 • C for 30 s, 72 • C for 30 s, and a final elongation for 5 min. PCR products were purified using AMPureXP (Beckman Coulter, Indianapolis, IN, USA). Further preparation of the libraries was carried out in accordance with the manufacturer's MiSeq Reagent Kit Preparation Guide (Illumina) (http://web.uri.edu/gsc/files/16s-metagenomic-library-prep-guide-15 044223-b.pdf (accessed on 11 May 2020)). The libraries were sequenced, according to the manufacturer's instructions, on an Illumina MiSeq instrument (Illumina, San Diego, CA, USA) using a MiSeq ® ReagentKit v. 3 (600 cycle) with pair-end reading (2 × 300 n). The obtained pool of raw sequences was trimmed with the aid of Trimmomatic [44] included in Unipro Ugene [45] using the following parameters: PE reads; sliding window trimming with size 4 and quality threshold 12; and minimal read length 130. Then paired sequences were combined using the program fastq-join [46], dereplicated and sorted by vsearch 2.7.1 [47]. The resulting sequences formed ribotypes in the whole pool of genomic rDNA; they were sorted according to their frequency. For our analysis, we established a threshold of 20 reads per pool of rDNA. The sequences were aligned using MEGA X [48]; a haplotype network was built in TCS 2.1 [49] and visualized in TCS BU [50]. Obtained ribotypes were used for the neighbor-joining tree constructed in MEGA X [48], by a maximum likelihood algorithm with the parameters GTR + G. We built a neighbor net of the ribotypes using SplitsTree 4.18 [51]. For this network, we set the threshold of 30 reads per the whole pool.
The marker region ITS1-5.8S rRNA gene-ITS2 was amplified using the universal primers ITS 1P [42] and ITS 4 [43]. The amplification parameters were: one cycle of 95 • C for 5 min; 35 cycles: 95 • C for 40 s; 52-56 • C for 40 s; 72 • C for 40 s; and final elongation 72 • C for 10 min. The chloroplast region trnL-trnF was amplified with the primers tabC, tabD, tabE and tabF [52]. Amplification parameters, when working with primer pairs C and D and E and F, were as for ITS amplification, but when using only external pair C and F were as follows: one cycle of 95 • C for 5 min; 35 cycles: 95 • C for 1 min; 52-56 • C for 1 min 10 s; 72 • C for 1 min 10 s; and final elongation 72 • C for 10 min. Region trnK-rps16 was amplified using primers rps16−4547mod [53] and trnK5'r [54] following the same parameters as for the ITS fragment. The sequencing was performed according to the standard protocol provided with a BigDyeTM Terminator Kit ver. 3.1 set of reagents on the sequencer ABI PRIZM 3100 sequencer at the Center for the collective use of scientific equipment "Cellular and molecular technologies for the study of plants and fungi" of the Komarov Botanical Institute, St. Petersburg. Chromatograms were analyzed with Chromas Lite version 2.01(Technelysium co.) and then the sequences were aligned with the aid of the Muscle algorithm [55] included in MEGA X [48].
Additionally, we used 42 sequences of trnL-trnF (including trnL gene, trnL intron, and trnL-trnF intergenic spacer), 32 sequences of trnK-rps16, and 54 sequences of ITS1-5.8S rDNA-ITS2 region, from GenBank database (https://www.ncbi.nlm.nih.gov/nuccore/ ?term (accessed on 11 May 2020)) along with our data (Supplementary, Tables S1-S3). Each dataset was analyzed separately because of the possible hybridization events resulting in different schemes by chloroplast and nuclear gene data. Appropriate evolutionary models were computed with MEGA X [48]. Indels were coded with SeqState 1.4.1 [40] and included into the alignment file as binary data ("restriction" option). Bayesian inference was performed by Mr. Bayes 3.2.2 [41] using GTR + I + G for ITS dataset, T92 + G for trnL-trnF and T92 for trnK-rps16, under the following conditions: 1-1.5 million generations; sampling trees every 100 generations; and the first 25% trees were discarded as burn-in. Maximum likelihood analysis was conducted using the same models with 1000 bootstrap replications using MEGA X [48]. A clade with 100-90% of posterior probability (PP) and bootstrap index (BS) was treated as strongly supported; 89-70% as moderately supported; and 50-69% as weakly supported. Nodes with indices below 50 were treated as unsupported.

Pollen Fertility Detection
In addition, to detect the hybrid status of the species we performed pollen fertility analysis for Trisetum spicatum, T. ruprechtianum, Koeleria asiatica, ×Trisetokoeleria jurtzevii, ×T. gorodkowii, ×T. taimyrica. Three or four anthers were taken from the herbarium material and placed in an Eppendorf tube with 200−250 µL of 45% acetic acid. The required degree of maceration was achieved within 1 h. Then each anther was placed on a glass slide, 50 µL of 10% acetocarmine solution was added, the anther was covered with a coverslip and slightly heated. After that, the material was distributed under the coverslip by lightly tapping with a wooden stick. Then, under a microscope, colored and uncolored pollen grains were counted, taking into account a total of at least 1000 pieces (lens 10×). Unstained pollen grains without contents or slightly colored and deformed ones were considered abortive, while uniformly intensely colored and undeformed ones were considered conditionally fertile. Temporary preparations were also photographed using 10× and 40× objectives.
Supplementary Materials: All supplementary materials are available on https://www.mdpi.com/ article/10.3390/plants11243533/s1. Table S1. TrnL-trnF sequences used in our analysis. Genbank data and sequences obtained in our research are given. Table S2. TrnK-rps16 sequences used in our analysis. Genbank data and sequences obtained in our research are given. Table S3. ITS sequences used in our analysis. Genbank data and sequences obtained in our research are given.