Stolonicaulon: A Section-Puzzle within Marsupella (Gymnomitriaceae, Marchantiophyta)

Marsupella sect. Stolonicaulon is not speciose and is a commonly neglected section within the genus, which currently includes three species with somewhat similar morphologies (wiry shoots with distanced leaves) and distributions in the mountains of tropical and subtropical regions (SE (Southeast) Asia, the Venezuelan Andes, and the high mountains of SE Brazil). After studying materials that were found to be dissimilar to the “traditional” Marsupella that were collected in the last decade by the authors of this article, it was found that these plants belong to three new-for-science species, and all of these species should be included in Marsupella sect. Stolonicaulon. The newly described species have expanded the boundaries of morphological variability, not only for the section itself, but also for the genus based on two findings: (1) the leaves of Marsupella sect. Stolonicaulon can be either appressed and entire or spaced and deeply divided (thus, the plants could occasionally be similar to Cephaloziella or Anastrophyllum); (2) some species of the section possess regular underleaf production. The first discovery of regular underleaves in Marsupella, as noted in two of the three newly described taxa, is the main morphological novelty described in this paper. The development of regular underleaves is a presumable relict character that brings Marsupella closer to Nardia, which was recently transferred to the Gymnomitriaceae and occupies an isolated position within its own subfamily, Nardioideae.


Introduction
Marsupella sect. Stolonicaulon (N. Kitag.) Váňa is perhaps the most mysterious and peculiar section within the genus due to its controversial morphological features and southern distribution, strongly contradicting other sections of the genus. Two efforts to make the system of the sect. Stolonicaulon more sound were made in the 20th century. The first was by Kitagawa [1] who described Marsupella stoloniformis N. Kitag. as well as sect. Stolonicaulon by itself. Kitagawa, however, was limited to researching the only known species belonging to the monotypic section. The next advance was made by Schuster [2], who described Marsupella stoloniformis subsp. vermiformis R.M.Schust. The lack of both fresh and old herbarium materials has prevented research on this section for several years, although there were several other attempts to clarify the systematic classification of Marsupella based on a new approach, including molecular genetic studies. The most valuable results in this field were accessed by Vilnet et al. [3] and Shaw et al. [4]. Several species were transferred from Marsupella to Gymnomitrion, while Nardia (previously placed in Jungermanniaceae) was placed within Gymnomitriaceae, but in its own subfamily [5]. Meanwhile, the exhaustive worldwide compendium of Gymnomitriaceae (based mostly on data obtained in the 'pre-molecular era') was published by Váňa et al. [6]. The last regional attempt to clarify the Marsupella-Gymnomitrion complex in East Asia involving recently collected materials was made by the revision of Korean Gymnomitriaceae [7]. All of these updates made the infrageneric structure of Marsupella sound but did not further our understanding of sect. Stolonicaulon due to the aforementioned lack of material.
Some newly collected materials belonging to Marsupella sect. Stolonicaulon that were collected in East and Southeast Asia have permitted scholars to discuss the taxonomy of this section again, such as the discussions by Bakalin et al. [8] in a review of Asian amphi-Pacific Marsupella. In summary, the results of the work by Bakalin  It is worth mentioning that M. microphylla is from Venezuela and Brazil and may refer to the section that is confined to the mountains of tropical and subtropical Asia eastward to Melanesia with some reservations.
Continuing the floristic exploration in Southeast and East Asia, the coauthors of the present work independently discovered strange plants, which were occasionally identified under a dissecting microscope or hand lenses as Cephaloziella or Anastrophyllum. Later, after attempting to understand the systematic position of these plants, these assessments revealed that all of the plants are even representatives of the same section (Stolonicaulon) despite strong morphological differences. The revealed diversity of morphotypes significantly refines both the morphological boundaries of the section itself (and partly of the genus as a whole) and forces us to take a different look at the distinguishing morphological criteria of the species within Marsupella. As is evident from that noted below, the representatives of sect. Stolonicaulon possess a span of morphological variability that is not narrower than that of all other Marsupella taxa taken together. The description of new data on the taxonomy of sect. Stolonicaulon was the main goal of this work.

Results
The ITS1-2 and trnL-F sequence data were obtained for two specimens from Taiwan, and ITS1-2, trnL-F and trnG-intron were obtained for specimens from Vietnam. Seven newly generated accessions were deposited in GenBank. The combined alignment of two genomic regions for 33 specimens revealed 1291 character sites, among which 888 positions belonged to ITS1-2 and 403 belonged to trnL-F.
The ML analysis resulted in a single tree with an arithmetic mean of the log likelihood of −5676.544929. In the Bayesian analysis, the arithmetic means of the log likelihoods for each sampling run were −5525.73 and −5525.83. The tree topologies that were obtained by both methods became congruent. Figure 1 demonstrates the ML topology with an indication of bootstrap support values (BS) and Bayesian posterior probabilities (PP). The backbone phylogeny of the genus Marsupella differs in the position of sections Hyalacme, Ustulatae, and Stolonicaulon compared with the phylogeny that was published by Bakalin et al. [8]. In both cases, some nodes received slight support. This finding may be explained by the influence of different outgroup sampling. The phylogenetic schema obtained previously [8]    The p-distance estimation revealed ITS1-2 variability in two sequenced specimens of M. vermiformis that could be incorrect readings (Table 1). Two specimens from remote regions of Taiwan exhibited variability in both ITS1-2 (0.1%) and trnL-F (0.3%) loci, and this finding is consistent with the level of intraspecific variability in the genus Marsupella [9]. The divergence among taxa varies from 2.5 to 4.4% for ITS1-2 and from 3.9 to 8.9% for trnL-F. A 4.6% difference in trnG-intron counts between the Chinese specimen that was first published as M. stoloniformis and the putative Vietnamese specimen was observed (data not shown in Table 1). Thus, phylogenetic affinity and the level of sequence divergence revealed the existence of three new science species from the genus Marsupella section Stolonicaulon: M. taiwanica, M. praetermissa and M. anastrophylloides.

Discussion
A phylogenetic analysis yielded a topology that was similar to that published by Bakalin et al. [8]. Previously, we wrote [8] (p. 63) the following: "Sect. Stolonicaulon is highly distinctive from the other bulk of Marsupella in a number of features, including . . . very rigid shoots, strongly distanced, scale-like leaves appressed to the stem and mesoxerophilous habitat that contradicts to the majority of other Marsupella". This phrase requires clarification. Indeed, all representatives of the section (as it is known today) are small plants, and most of them are wiry. However, their leaves may not be appressed to the stem and they are not highly distanced.
The basal branch to all other members of sect. Stolonicaulon is formed by Marsupella anastrophylloides, which definitely justifies its name based on its resemblance to small representatives of Anastrophyllum or Sphenolobopsis, as well as deeply incised bilobed leaves with narrow lobes. This species is the most dissimilar to the other representatives of the section. In this case, the genetic distance correlates well with morphology.
Another well-circumscribed taxon described here is Marsupella taiwanica, which is similar to Cephaloziella due to incised leaves and small regular underleaves and therefore strongly dissimilar to other Marsupella. However, the generative features of this taxon definitely possess the character of other representatives of sect. Stolonicaulon, including a long perigynium, which is partially exposed due to relatively short leaves.
Three other species are superficially (under a dissecting microscope) similar to each other morphologically (wiry shoots, distanced leaves, and a stem that is almost as wide as the left shoot). However, these species are not as closely related in the phylogenetic tree as could be suggested based on an initial look. The two species that are most closely related to each other are Marsupella stoloniformis and M. vermiformis. Their morphological differences have been discussed previously [8]. These differences mainly involve short emarginate to entire leaves and somewhat hygrophilous ecology in M. stoloniformis versus mostly bilobed, sharply pointed leaf lobes and somewhat xerophilic ecology in M. vermiformis. The third species is similar to both aforementioned species and is described here as M. praetermissa. The original specimen (treated here as holotype) was originally confused with M. stoloni- formis and appeared in the paper by Shaw et al. [4] under the latter name. However, the comparison of this specimen with a type of M. stoloniformis (KYO, Mizutani 2788!) revealed that it is not a conspecific of the type. Indeed, M. praetermissa is characterized by a morphology that is similar to that of M. stoloniformis (distanced, entire-to-shallowly emarginate leaves, and a comparatively thick stem); however, the leaves of this species narrowly spread from the stem, even under dry conditions, which clearly contradicts the always appressed leaves in the type and other known plants of M. stoloniformis. The main distinguishing feature is regular, albeit small underleaves that are present in M. praetermissa. The latter to a certain extent explains its relationships with M. taiwanica in the phylogenetic tree (despite the superficial morphology, the two taxa are strongly dissimilar).
The occurrence of underleaves was not known in Marsupella prior to the present work. As currently reported, two described species, M. taiwanica and M. praetermissa, show regular underleaves. Moreover, it is worth mentioning that M. anastrophylloides rarely possess the ability to develop 1-2 subulate underleaves just above the ventral branch (although very irregular), whereas underleaves are completely absent in unbranched shoots. Therefore, in the taxon that is most genetically distanced from others (M. anastrophylloides), the presence of underleaves is rare. In contrast, underleaves are regularly noted in the two other taxa (M. taiwanica and M. praetermissa). In M. stoloniformis and M. vermiformis, underleaves are completely absent.
The stem cross section also possesses noticeable variation across sect. Stolonicaulon. Two earlier known taxa (M. stoloniformis and M. vermiformis) have no hyaloderm cells (the cells are thick-walled and not-or slightly-different in size from the cells inward in both). Marsupella taiwanica is an intermediate variant in this feature. Specifically, its outer cells in the stem cross section are larger than the inner cells and have distinctly thinner walls; however, these cell walls are much thicker than those that are observed in the two remaining taxa. Another extreme (in contrast to M. vermiformis and M. stoloniformis) is the pair M. praetermissa and M. anastrophylloides. These two taxa exhibit strongly different leaf morphologies but possess distinct hyalodermis of thin-walled cells that are strongly different from scleroderm cells inward.
Therefore, Marsupella sect. Stolonicaulon possesses wide variation in several morphological features: (1) Leaf shape and comparative stem size (with the stem diameter). Leaves may be contiguous to highly distanced, ranging from deeply divided and appressed to the stem to narrowly spreading; (2) Stem cross-section features. Although all species have thick-walled scleroderm cells and more or less thick-walled cells in the inner part, the outer cells of the stem can be either thick-walled or distinctly thin-walled and form a distinct hyaloderm (although they commonly only slightly differ in size from the inner cells); (3) Underleaf presence. This feature makes the section unique among Marsupella. Regular underleaves are present in M. praetermissa and M. taiwanica. Underleaves are also present in M. anastrophylloides as a rare exception. Underleaves are generally a very unusual characteristic within Gymnomitriaceae. This family in the "old" sense did not include genera with underleaves at all [6]. The only presumable 'member' of Gymnomitriaceae with regular underleaves (as long as leaf length!) is monotypic Paramomitrion R.M. Schust., which has only been reported in sterile conditions from the only collection that was made by R.M. Schuster in Venezuelan Andes. It is questionable whether Paramomitrion belongs to Gymnomitriaceae at all. In the original description of the genus and species, Schuster [2] (p. 138) wrote the following: "I place it in the Gymnomitriaceae, next to Eremonotus-but without any conviction that this is the final resting place". Eremonotus clearly belongs to Jungermanniaceae, whereas the molecular-phylogenetic affinity of Paramomitrion is not known.
However, the molecular revision of the suborder Jungermanniineae [4] has necessitated the transfer of Nardia to Gymnomitriaceae (albeit within the isolated subfamily Nardioideae Váňa) among other things. Thus, Gymnomitriaceae became partly an 'amphigastrious' family. The present account has shown that distinct regular underleaves are also possible in the subfamily Gymnomitrioideae in addition to the atavistic underleaves, which were indicated for some species of Gymnomitrion [10]. Underleaves are occasionally present in G. laceratum (Steph.) Horik. The presence of regular underleaves in Marsupella is most likely a plesiomorphic trait, emphasizing its relationship to Nardia. In this regard, two species of Nardia have been described that are new to science within recent years (Nardia minutifolia Furuki and Nardia grollei Váňa et D.G. Long) and should be mentioned. Both are characterized by entire leaves that only slightly exceed the width of the stem, and both were not tested genetically. Such species may need to be tested to assess their relationship to Marsupella sect. Stolonicaulon. However, these species are most likely not related to the genus Marsupella sect. Stolonicaulon, which is characterized by cell sizes which are much smaller than that noted in both mentioned Nardia species.
In addition to morphology, Marsupella sect. Stolonicaulon shows a peculiar distribution pattern that distinguishes it from other sections of the genus. All representatives of the section are distributed in the mountains in the tropical and subtropical regions of Pacific Asia to Melanesia (the northernmost occurrence of M. vermiformis is noted in the mountains of Jeju Island, where the vegetation of the lowlands is subtropical). One species, namely, Marsupella microphylla R.M. Schust., is known from the Neotropical Floristic Kingdom (Venezuela and Brazil), but its assignment to sect. Stolonicaulon can be questioned. This poorly understood species has not been studied using molecular genetic methods and its systematic position in Marsupella is unclear.
Holotype: Vietnam, Hà Giang Province, Vi Xuyên District, Cao Bo Commune, Tay Con Linh Range, Tay Con Linh Nature Reserve (22 •  Etymology: Its name was given based on morphological similarity to small Anastrophyllum. Ecology: The species was only collected once, therefore, its ecology is incompletely known. It was gathered in a southern subtropical mountain forest over open mesic cliffs as an admixture in the mat of Kurzia makinoana. The plants were creeping over Kurzia. The same patch contains a sparse admixture of the xeromorphic modification of Anastrophyllum bidens. Comment: Due to its prominent external characteristics, including contiguous obliquely spreading and deeply divided leaves, the species could be confused with all other known Marsupella. However, in the sterile state (as it was only collected), the species may be mistaken for depauperate plants of Anastrophyllum bidens, from which it differs based on a distinct tint of copper or pinkish pigmentation and distinctly smaller leaf cells (7-13 µm wide, versus 15-20 µm wide in the midleaf) with thick walls, concave trigones (versus trigones large, prominently bulging), and fragile shoots (versus shoots more elastic). Disregarding the copper and pinkish pigmentation, the species somewhat resembles Sphenolobopsis pearsonii, but the leaf cuticle is virtually smooth and regular underleaves are absent in Marsupella anastrophylloides. The main difference from the latter potentially occurs in generative structures, which are unfortunately unknown.
Given the strong molecular dissimilarity to other taxa and no distantly similar regional taxa, we hypothesized that the species has undergone a long period of isolation and speciation. If so, the expectation of its current distribution is very speculative since the 'old-existing' taxon might have a wide original area and be preserved in several restricted areas as a geographic relict.
Illustrations in present paper: Figures 2 and 3.     Etymology: the name 'praetermissa' is translated as 'overlooked' from Latin due to its superficial similarity to M. stoloniformis.
Ecology: The species is known from the single specimen that was collected from granitic cliffs on the alpine lake shore (a small boulder under a shady dripping cliff) at an elevation of 3455 m a.s.l. (Figure 4, left corner). This is a primarily forestless landscape on gentle NE-facing slopes and a complex of lakes. The nearest forests are approximately 3 km downstream at the S-facing slopes at elevations of 3200-3300 m a.s.l.; however, scattered trees are present near the collecting locality. The associates within the specimens included Riccardia sp., Gymnomitrion sp. (belonging to the morphological group 'Apomarsupella'), and Sphenolobopsis pearsonii (Spruce) R.M. Schust.
Comment: The distribution of the taxon could hardly be expected given the availability of only one record. Although the collecting locality is formally located in the Salween (Nu Jiang) River catchment, the collecting locality is only 500 m eastward from the pass to the Irrawaddy (Ayeyarwady) River catchment, the largest Myanmar River. Therefore, in a broad context, this is an area where several main river upper courses meet one another (Salween, Irrawaddy, Brahmaputra, and Mekong). Although strong differences in the species content between closely situated sites have been known here for over a century [11,12], the actual distribution of the taxon may be not very narrow, and additional records, including those from remote areas (around Sino-Himalaya), are highly possible.
Superficially, when assessing a small sample under the dissection microscope, the species is morphologically very similar to M. stoloniformis, which may explain the original misidentification of the specimen. The specimen was subjected to sequencing by Shaw et al. [4]. In the various topologies we attempted to combine, this specimen occupied an isolated position from the accession of M. stoloniformis from Vietnam that required further
Ecology: The species is known from the single specimen that was collected from granitic cliffs on the alpine lake shore (a small boulder under a shady dripping cliff) at an elevation of 3455 m a.s.l. (Figure 4, left corner). This is a primarily forestless landscape on gentle NE-facing slopes and a complex of lakes. The nearest forests are approximately 3 km downstream at the S-facing slopes at elevations of 3200-3300 m a.s.l.; however, scattered trees are present near the collecting locality. The associates within the specimens included Riccardia sp., Gymnomitrion sp. (belonging to the morphological group 'Apomarsupella'), and Sphenolobopsis pearsonii (Spruce) R.M. Schust. morphological investigation. Shortly thereafter, it was noted that even in dry conditions, the superficial morphology of the taxon was different from that of M. stoloniformis (both type in KYO and our vouchers from Vietnam). The leaves of M. stoloniformis are always (even when wet) appressed to the stem, such that the stem diameter is almost the same as the plant width. However, as described here, M. praetermissa has leaves that narrowly spread from the stem, even when the plants are dry. Another very distinctive feature that is evident under the dissecting microscope is the presence of small but regular underleaves-a feature that is unknown in M. stoloniformis. The third difference helps to distinguish the species from M. stoloniformis and is evident in the stem cross section in the microscope slide. Marsupella praetermissa has distinctly hyalodermatic outer cells with very thin walls, while the outer cells of M. stoloniformis are distinctly thick and hardly different from medullary cells.
Illustrations in present paper: Figures 5 and 6.  Comment: The distribution of the taxon could hardly be expected given the availability of only one record. Although the collecting locality is formally located in the Salween (Nu Jiang) River catchment, the collecting locality is only 500 m eastward from the pass to the Irrawaddy (Ayeyarwady) River catchment, the largest Myanmar River. Therefore, in a broad context, this is an area where several main river upper courses meet one another (Salween, Irrawaddy, Brahmaputra, and Mekong). Although strong differences in the species content between closely situated sites have been known here for over a century [11,12], the actual distribution of the taxon may be not very narrow, and additional records, including those from remote areas (around Sino-Himalaya), are highly possible.
Superficially, when assessing a small sample under the dissection microscope, the species is morphologically very similar to M. stoloniformis, which may explain the original misidentification of the specimen. The specimen was subjected to sequencing by Shaw et al. [4]. In the various topologies we attempted to combine, this specimen occupied an isolated position from the accession of M. stoloniformis from Vietnam that required further morphological investigation. Shortly thereafter, it was noted that even in dry conditions, the superficial morphology of the taxon was different from that of M. stoloniformis (both type in KYO and our vouchers from Vietnam). The leaves of M. stoloniformis are always (even when wet) appressed to the stem, such that the stem diameter is almost the same as the plant width. However, as described here, M. praetermissa has leaves that narrowly spread from the stem, even when the plants are dry. Another very distinctive feature that is evident under the dissecting microscope is the presence of small but regular underleaves-a feature that is unknown in M. stoloniformis. The third difference helps to distinguish the species from M. stoloniformis and is evident in the stem cross section in the microscope slide. Marsupella praetermissa has distinctly hyalodermatic outer cells with very thin walls, while the outer cells of M. stoloniformis are distinctly thick and hardly different from medullary cells.
Ecology: Acidophilic meso-xerophyte. This is a rare and poorly known species, and the data on its ecology may be incomplete. Marsupella taiwanica occupies well-exposed habitats, including bare soil on roadsides and rock fissures in high mountains where it forms blackish-brown pure patches.
Comment: Marsupella taiwanica is most similar to several Cephaloziella species, especially C. divaricata (Sm.) Schiffn.; C. grimsulana (J.B. Jack ex Gottsche & Rabenh.) Lacout.; and C. varians (Gottsche) Steph., due to (1) small shoots with brownish coloration forming blackish brown patches; (2) remote and transversely inserted, erect spreading leaves; (3) the presence of regular underleaves. All the mentioned Cephaloziella species occasionally occur in sterile conditions. However, in patches of Marsupella taiwanica, fertile plants are also occasionally absent. These circumstances make it difficult to distinguish M. taiwanica from the aforementioned Cephaloziella species. Under sterile conditions, M. taiwanica can be distinguished from these species based on the following indirect features. In Cephaloziella, the sterile leaves are generally more deeply bilobed for 0.5-0.8 their length, especially near the shoot apex [14] (p. 36, Figure 498: 10). However, in M. taiwanica, the sterile leaves have a shallower sinus, usually ca. 0.3 of the leaf length. In Cephaloziella, the underleaves are mostly located on stems between the bases of opposite leaves cf. [14][15][16], whereas the underleaves are usually located very close to the base of one of the leaves in a leaf pair in M. taiwanica. The fertile (gynoecial) shoots of M. taiwanica easily differ from Cephaloziella based on the shape of the perianths (hidden within bracts, abruptly narrowed to the mouth), the presence of distinct perigynium, and the shape of the bracts (lobes acute to obtuse or rounded at the apex with shallow sinus and entire margins). The present species differs from all Marsupella based on its Cephaloziella-like habit. Another feature is the presence of regular underleaves shared within Marsupella with M. praetermissa that are strongly dissimilar in other morphological features.
Illustrations in present paper: Figures 7 and 8. M. taiwanica. The fertile (gynoecial) shoots of M. taiwanica easily differ from Cephaloziella based on the shape of the perianths (hidden within bracts, abruptly narrowed to the mouth), the presence of distinct perigynium, and the shape of the bracts (lobes acute to obtuse or rounded at the apex with shallow sinus and entire margins). The present species differs from all Marsupella based on its Cephaloziella-like habit. Another feature is the presence of regular underleaves shared within Marsupella with M. praetermissa that are strongly dissimilar in other morphological features. Illustrations in present paper: Figures 7 and 8.     Figures 9 and 10) 3. Leaves always divided by semi-crescentic sinus with lobes acute … M. vermiformis ( Figure  11)  *-The taxon is known from a very few specimens; the concept of this taxon is derived from literature sources and needs critical testing.

Taxa Sampling
The study involved 6 specimens of Marsupella sect. Stolonicaulon: one voucher of M. vermiformis (with sequences that were obtained twice) from Jeju Island; one voucher that was identified as M. stoloniformis from North Vietnam; two accessions of unknown species from Taiwan (named below as M. taiwanica); one accession that was identified as M. stoloniformis from Yunnan Province of China (named below as M. praetermissa); and the specimen of unknown species from Hà Giang Province in North Vietnam (named below as M. anastrophylloides) ( Table 2). All specimens were compared with each other, although they were so significantly different that we were occasionally inclined to compare them with the taxa from other sections, but not sect. Stolonicaulon itself.
The sequence data of ITS1-2 nrDNA and the trnL-F cpDNA data for the specimens of M. vermiformis (Republic of Korea) and M. stoloniformis (Vietnam) were published by Bakalin et al. [8], and the trnL-F and trnG-intron data for M. stoloniformis (China; as it is shown below, the identification was incorrect and belongs to M. praetermissa described below) were published by Shaw et al. [4]. In this study, ITS1-2 and trnL-F sequence data were obtained for two specimens from Taiwan and a single specimen from Vietnam, and the last trnG-intron was also sequenced. To reveal the phylogenetic affinity of these morphologically putative specimens, the alignment involving ITS1-2 and trnL-F nucleotide sequence data from twenty molecularly studied Marsupella species with representatives

Taxa Sampling
The study involved 6 specimens of Marsupella sect. Stolonicaulon: one voucher of M. vermiformis (with sequences that were obtained twice) from Jeju Island; one voucher that was identified as M. stoloniformis from North Vietnam; two accessions of unknown species from Taiwan (named below as M. taiwanica); one accession that was identified as M. stoloniformis from Yunnan Province of China (named below as M. praetermissa); and the specimen of unknown species from Hà Giang Province in North Vietnam (named below as M. anastrophylloides) ( Table 2). All specimens were compared with each other, although they were so significantly different that we were occasionally inclined to compare them with the taxa from other sections, but not sect. Stolonicaulon itself.

DNA Isolation, PCR Amplification and DNA Sequencing
DNA was extracted using a DNeasy Plant Mini Kit (Qiagen, Germany) according to the manufacturer's protocol. For the amplification and sequencing of ITS1-2 and trnL-F, the pairs of primers that were suggested by White et al. [17] and Taberlet et al. [18], respectively, were used. The trnG-intron of the Vietnamese specimens was obtained with primers from Shaw et al. [19]. A PCR was performed in a 20-µL volume using the following amplification cycles: 3 min at 94 • C; 30 cycles of 30 s 94 • C, 40 s 56 • C, 60 s 72 • C; and 2 min of final extension time at 72 • C. The amplified fragments were visualized on 1% agarose TAE gels by EthBr staining, purified using the Cleanup Mini Kit (Evrogen, Moscow, Russia), and then sequenced using the ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Waltham, MA, USA) following the standard protocol that was provided for the 3100 Avant Genetic Analyser (Applied Biosystems, Waltham, MA, USA) (Genome Centre, Moscow, Russia).

Molecular Analysis
Two datasets for ITS1-2 and trnL-F loci were automatically aligned with the ClustalW option and then manually corrected in BioEdit 7.0.1 [20]. All positions were taken in estimation, and the absent data were coded as missing. Both datasets revealed congruence after preliminary phylogenetic analyses and were combined into a single ITS1-2+trnL-F alignment for subsequent estimations. The maximum likelihood (ML) with PhyML v.3.0 [21] and Bayesian approach with MrBayes v. 3.2.1 [22] were used to test phylogeny. The TN+I+G model was selected as the best-fit evolutionary model of nucleotide substitutions by Mod-elGenerator [23]. The 300 bootstrap replications were found to be sufficient for reaching BS convergence with a Pearson average of ρ100 = 0.995450 in RAxML v7.2.6 [24]. Thus, ML was run with the TN+I+G model with 300 bootstrap replicates, and the rate heterogeneity among the sites was modelled using a gamma distribution with four rate categories. For the Bayesian analysis, each partition of the combined alignment (ITS1-2, trnL-F) was separately assigned to the GTR+I+Г model, as suggested by the program creators, and gamma distributions were approximated using four rate categories. Two independent runs of the Metropolis-coupled MCMC were used to sample the parameter values in proportion to their posterior probability. Each run included three heated chains and one unheated chain, and two starting trees were chosen randomly. Chains were run for five million generations, and the trees were sampled every 100th generation. The software tool Tracer v.1.7.1- [25] revealed an effective sample size of 18,380.9242 and an autocorrelation time of 489.6489. The first 12,500 (25%) trees in each run were discarded as burn-in. Thereafter, 75,000 trees were sampled from both runs. The average standard deviation of split frequencies between the two runs was 0.001548. The Bayesian posterior probabilities were calculated from the trees that were sampled after burn-in. The majority rule (MJ) consensus tree was calculated after combining the runs minus burn-in of 25%, and the topology was illustrated with The sequence variability among the specimens of the section Stolonicaulon was evaluated as the p-distances for each DNA locus in Mega 11 [27] using the pairwise deletion option for counting gaps.

Conclusions
It was somewhat difficult even for a specialist to recognize the simple Marsupella in two of three newly described species and, even more so, to refer these taxa to the section Stolonicaulon without a molecular-genetic analysis. Indeed, after this work was performed, the genus Marsupella became more morphologically inconsistent, whereas the family Gymnomitriaceae itself became more balanced morphologically. Based on the newly described species of Marsupella, the subfamily Gymnomitrioideae evidenced distinct morphological connections in the presence of underleaves with an isolated subfamily Nardioideae. This work shows that even in "long-existing" genera, new species can be found that completely differ morphologically from previously known representatives. To speculate based on the obtained conclusions, the Sino-Himalayas and even the seemingly well-studied Taiwan Island may hide many undescribed species.