Range extensions for the rare moss Plagiothecium handelii, and its transfer to the resurrected genus Ortholimnobium

Plagiothecium handelii is newly recorded for Europe (Austria and Romania) and eastern North America (Tennessee, USA). This dainty species was previously known only from Yunnan and Sichuan Provinces, China. Morphologically, the disjunctive populations belong to a single species. A 27-taxon phylogeny of Plagiothecium based on nuclear ITS and plastid rpl16 intron DNA sequence data resolved Austrian and Chinese populations of P. handelii as sisters, in a clade with P. paleaceum, a julaceous Himalayan species with cochleariform leaves. In contrast, P. handelii is a filiform plant with distant, ovate-acuminate leaves. In sequence identity the three terminals have a similar level of variation, suggesting that the geographic disjunction between the two populations of P. handelii is quite old. Morphologically and genetically the clade is a well defined lineage (Ortholimnobium) that is transitional between Plagiothecium s.str. and Struckia. The new combinations O. paleaceum and O. handelii are made. In Europe, O. handelii should be classified as vulnerable.

Plagiothecium handelii Broth. is a poorly known pleurocarpous moss. It was described by Brotherus (1929) from several collections made by Heinrich von Handel-Mazzetti in northwestern Yunnan Province, China, in 1915 and1916. It resembles an etiolated form of P. cavifolium (Brid.) Z. Iwats. with concave, ovate-acuminate leaves with attenuate tips. It differs from the latter species in its pseudo-stipitate habit, narrower leaf cell net, and cortical hyalodermis composed of rectangular stem cells that detach with the leaves alongside the large leaf decurrencies that are typical of the genus (Fig. 1). In these respects, P. handelii also resembles an etiolated form of the closely related Sino-Himalayan species P. paleaceum (Mitt.) A. Jaeger, which differs in its slightly larger stature and broader, circular-ovate, concavecochleariform leaves. The erect sporophytes of P. handelii and P. paleaceum are also very similar.
In the course of a recent phylogenetic study of Plagiothecium Bruch & Schimp. (Wynns et al. in press), the first author received several collections of a delicate, feltlike moss collected by the second author in Austria and identified by him as Plagiothecium cavifolium var. gracile Breidl. After comparison with several collections of P. handelii from Yunnan, China, including three syntypes, we identified the Austrian moss as the latter species. A herbarium study of global collections of Plagiothecium (Wynns 2015) uncovered additional specimens of P. handelii from Romania and the eastern United States.
Collections of Plagiothecium handelii from Austria and China were included in Wynns et al.'s molecular study of the genus. Here, we performed phylogenetic analyses of combined nuclear ITS and plastid rpl16 intron DNA sequence data from 27 collections of Plagiothecium and related pleurocarpous mosses (Table 1), in order to establish the conspecificity of Chinese and Austrian populations of P. handelii, and to place the species in a phylogenetic framework.

Methods
For the DNA analyses we included 20 collections of Plagiothecium (sensu Zuo et al. 2011), six collections of other Plagiotheciaceae (Isopterygiopsis Z. Iwats. and Platydictya Berk.), and one collection of Fabronia pusilla Raddi, which was used as the outgroup (Table 1). DNA extraction, PCR amplification and DNA sequencing were performed with the protocol of Wynns et al. (in press). Sequence alignment was performed manually in MEGA4 (Tamura et al. 2007). Next, indel data were generated for each alignment in SeqState ver. 1.4.1 (Müller 2005) using simple indel coding (Simmons and Ochoterena 2000). A single data file including both nucleotide and indel data for each DNA region was assembled in NEXUS format and analyzed by maximum parsimony (branch-and-bound search) in PAUP ver. 4.0.10b (Swofford 2002). A bootstrap (BS) analysis was also performed (2000 replicates) using branch-and-bound. A partition homogeneity test was performed (1000 branch-and-bound replicates), to confirm that the two data sets were congruent and appropriate for a combined analysis.
In addition to the parsimony analysis, a Bayesian analysis was performed using the program MrBayes ver. 3.2 (Ronquist et al. 2012). The data were divided into four partitions, two of nucleotide sequence data (rpl16 and ITS) and two respective partitions of binary indel data. The partitions were unlinked and were allowed to evolve at different rates. Based on the Akaike information criterion and the results of the hierarchichal likelihood ratio tests, the program Modeltest ver. 3.06 (Posada and Crandall 1998), selected the K81uf+I+Γ (Kimura 1981) model of DNA sequence evolution. However, this model is not implemented in MrBayes, so for the nucleotide data partitions we used the GTR+I+Γ model (cf. Lecocq et al. 2013), which in fact had the best overall log likelihood score. For the indel data partitions the default model was used, a Γ-shaped rate variation was assumed (Yang 1993), and the coding bias was set to variable. A Markov chain Monte Carlo (MCMC) analysis was then run for 11 000 000 generations under the default settings. The results of the Bayesian analysis including posterior probability (PP) support values were visualized as a 50%-compromise phylogram based on average branch lengths using the program TreeGraph 2.1.0-386 beta (Stöver and Müller 2010).

Results
The rpl16 data set included 986 characters, of which 54 were indel characters. The ITS data set included 682 characters, of which 22 were indel characters. Thus, the molecular analyses included 1744 characters in total, of which 1668 were nucleotide characters and 76 were indel characters (4.4%); 1384 characters were constant, 153 were variable but not parsimony-informative and 207 were parsimony-informative (20.6% variable characters). The strict consensus of two equally-parsimonious trees of 550 steps was well-resolved, with consistency index (CI, Kluge and Farris 1969) = 0.735, CI excluding uninformative characters = 0.627, retention index = 0.790, and rescaled consistency index = 0.581 (Farris 1989). The 50%-compromise Bayesian tree ( Fig. 2) was identical to the parsimony tree in topology, with an additional unsupported node. The results of the partition homogeneity test (p = 0.190) indicated that the plastid and nuclear data sets reflect the same underlying phylogeny.

Discussion
In Austria, Plagiothecium handelii colonizes humus-rich, cool and moist scree slopes in the montane belt dominated by spruce forests. It occurs in small cavities filled by humus under boulders (Fig. 1, A). The bedrock is mostly Table 1. Specimens used in the DNA study. Barcode numbers are for the herbaria where the specimens are housed. Herbarium acronyms are from Index Herbariorum (Thiers, continuously updated). The last two columns are GenBank accession numbers. * We lacked an ITS sequence for Ortholimnobium paleaceum, and substituted a consensus of two Chinese O. paleaceum ITS sequences from GenBank. , which also has a main distribution in Asia and occurs disjunctively in the Alps, grows in colder, more dynamic sites with larger boulders on the same mountainsides. Unlike P. neckeroideum, P. handelii also grows on (less acidic) shale, and thus the distribution of P. handelii in Europe is likely to be slightly wider. The sister relationship between the Chinese and Austrian populations of Plagiothecium handelii that was found in the molecular analyses (Fig. 2) corroborates their conspecificity. Although the populations from Romania and USA were not included in the molecular sampling, the unusual growth form of this species (felt-like mats of etiolated stems, cf. Fig. 1, A, B) is quite easy to recognize, so we have confidence in their identity. Plagiothecium handelii thus has an unusual distribution pattern. While there are several bryophytes that occur disjunctively between the Alps and the Sino-Himalayan region, including P. neckeroideum, Distichophyllum carinatum Dixon & W.E. Nicholson (Dixon andNicholson 1909, Redfearn et al. 1994), Tayloria rudolphiana (Garov.) Bruch & Schimp. (Koponen 1992), and (possibly) Brotherella lorentziana (Molendo ex Lorentz) Loeske (Frahm 2013), none of these species occur disjunctively in the eastern USA. Similarly, some bryophytes occur disjunctively between the eastern USA and southwest China, for example Entodon macropodus (Hedw.) Müll. Hal. (Iwatsuki and Sharp 1967), Brothera leana (Sull.) Müll. Hal., Grimmia pilifera P. Beauv. and Acrobolbus ciliatus (Mitt.) Schiffn. (Iwatsuki 1972), but these species are not found in Europe. This suggests either the European or American populations of P. handelii may have arisen through long-distance dispersal (cf. Frahm 2013).
On the other hand, Plagiothecium handelii is a dioicous species that is seldom collected in fruit, reducing the likelihood of dispersal by spores. Furthermore, it is restricted to pristine habitats which apparently have a similar ecology at all of the stations. Also, the plants occur in discontinuous localities, both in Europe and America. Last, the relatively large genetic divergence (Fig. 2) between Austrian and Chinese collections is suggestive of a very old disjunction. These facts support a hypothesis that extant populations of P. handelii are relicts of a pre-Pleistocene flora that once had a much broader distribution (Steere 1937, Iwatsuki 1972, Schuster and Damsholt 1974, Manos and Stanford 2001, Hedenäs 2008, Patiño et al. 2016. We believe this explanation is more likely. Many authors (Herzog 1926, Müller 1954, Schuster 1983, Schönswetter et al. 2005, Damsholt 2013 suggest that certain small Alpine localities must have remained uncovered during the Pleistocene glaciations and thus served as refugia for older floras.

Phylogenetic position of Plagiothecium handelii
The sister relationship between Plagiothecium handelii and P. paleaceum was first identified by Zuo et al. (2011), as was the sister relationship between this lineage and Struckia. Huttunen et al. (2013) subsequently transferred P. handelii and P. paleaceum to Struckia. Wynns et al. (in press) found P. handelii and P. paleaceum morphologically closer to Figure 2. 50%-compromise phylogram based on average branch lengths generated from a Bayesian analysis of of plastid rpl16 intron and nuclear ribosomal ITS data from 27 collections of hypnalean mosses (Table 1). Gaps were coded as separate characters, using simple indel coding (Simmons and Ochoterena 2000). For P. paleaceum, we lacked an ITS sequence and substituted a consensus of two sequences from a public database (GenBank). Posterior probability support values are indicated above the branches, and bootstrap support values (2000 branch-and-bound replicates) from the parsimony analysis are indicated below the branches. The consensus parsimony tree (not shown) lacked one node (indicated with *), but was otherwise identical in topology. The node without a bootstrap value near the bottom of the tree was present in the parsimony tree but not in the bootstrap tree.
Plagiothecium than to Struckia, and like Zuo et al. they included Struckia in Plagiothecium.
Struckia sensu Ignatov et al. (2007) includes two species. Struckia argentata (Mitt.) Müll. Hal. is not uncommon on tree trunks in the Sino-Himalayan region, while S. enervis (Broth.) Ignatov, T.J. Kop. & D.G. Long occurs on rocks, and has a restricted distribution in boreal Asia with a disjunctive occurrence in southwest China (Tan et al. 1990, Hedenäs 1996. Struckia differs from Plagiothecium in several respects. In general, species of Plagiothecium are prostrate satiny mosses with spreading leaves, whereas both species of Struckia are small erect mosses with appressed leaves and a shaggy appearance (produced by the long-attenuate leaf tips). Microscopically, the leaves are hardly decurrent in Struckia, while this is a defining feature of Plagiothecium. Also unlike Plagiothecium, the leaves in Struckia have a large number of quadrate alar cells (reminiscent of Entodon Müll. Hal.). Struckia argentata further differs from Plagiothecium in its epiphytic habit, denticulate upper leaf margins, and shiny, small-mouthed capsules with conic-mammillate opercula. On the other hand, S. argentata resembles the sympatric species P. handelii and P. paleaceum in having short erect capsules and ovate-acuminate, concave, subplicate leaves that are often subtended by cells of the cortical hyalodermis when detached.
Struckia enervis is peculiar in that it forms erect defoliated shoots topped with clusters of shortly ligulate, costate gemmae (reduced leaves) (cf. Abramova and Abramov 1981). The specialized morphology of S. enervis promotes vegetative spread and probably arose with the transition away from the epiphytic habitat. It is presumably the more derived species of Struckia (Tan et al. 1990).
Paraphyletic taxa are the real result of evolution (Hörandl 2006) and should not automatically be rejected as the basis for names. In this case Plagiothecium would be rendered paraphyletic by retaining P. handelii and P. paleaceum but excluding Struckia as a distinct genus that evolved rapidly under adaptive pressures associated with a change to epiphyty. This is a plausible hypothesis that is consistent with the morphological and genetic (Fig. 2) differences between Plagiothecium and Struckia. However, this classification hides the close relationship between the P. handelii + P. paleaceum clade and Struckia. Morphologically P. handelii and P. paleaceum share features of both genera and are clearly a transitional taxon that can itself be treated as a distinct genus. The type of the genus Ortholimnobium Dixon belongs to P. paleaceum (Enroth et al. 1992), so this name is available.
Plants glossy, pseudo-stipitate; branches more or less terete, with cortical hyalodermis; leaves imbricate to distant, symmetric, ovate-acuminate, concave, weakly plicate, decurrent, with short narrow cells; capsules short, erect.  Enroth et al. (1992) reported that the lectotype specimen was probably collected by T. Thomsen, not J. D. Hooker. Here we have simply followed the information given by Mitten (1859), which corresponds with the label data.

Variation
In general Ortholimnobium handelii is quite stenotypic. However, a form with wider cells occurs at high elevations. We noticed this both in Austrian (Breidler s.n., 2100 m, GJO 0071156) and American (Anderson 10524, ca 2000 m, DUKE no. 78480) populations. Occasional collections of most species of Plagiothecium s.l. have a laxer leaf areolation, perhaps in response to environmental factors.

Conservation
In Austria, Ortholimnobium handelii is apparently rare to scattered in mountainous areas with siliceous bedrock. It has a wider distribution than the sympatric Plagiothecium neckeroideum, which is rare or absent in the western Alps. Most populations of O. handelii are not currently threatened, but the complex local climate of scree slopes can certainly be disturbed by road construction and logging. Although numerical population data are missing, O. handelii can be categorized as VU D1 (Vulnerable) under current IUCN criteria, on the basis of small population sizes. Now that O. handelii is known to occur in Europe, future floristic studies should improve our knowledge of the frequency and distribution of this species.