Rediscovery of Roesleria subterranea from Japan with a discussion of its infraspecific relationships detected using molecular analysis

Roesleria subterranea, a distinctive hypogeous fungus, was collected from unidentified deciduous plant roots in red pine forests. The fungus had been documented several times in the past in Japan, but with no description. A description is given here based on specimens collected in Japan. The sequence of the D1-D2 region of the 28S rDNA obtained from the isolate was identical to those of the European and American specimens. Maximum parsimony analysis incorporating the present data and all other available ITS-5.8S sequences for R. subterranea showed that there are two infraspecific groups. One of them, composed of the isolates from Vitis spp. in Germany, Italy, and USA, was monophyletic. The other group, composed of isolates from deciduous trees in various countries, including Japan, was paraphyletic. The phylogenetic patterns indicate that the host may be more important than geographical distance for the genetic diversification of R. subterranea.


introduction
Roesleria subterranea (Weinm.) Redhead is a distinctive hypogeous fungus currently placed in Roesleriaceae (Yao and Spooner 1999) or Helotiaceae Helotiales (Neuhauser et al. 2011). It is recognized as a facultative root parasite that causes damage to grape vines (Neuhauser et al. 2011). Ecologically, it occurs on the plant roots of various deciduous trees. Its morphology characterized by stalked apothecia with spores produced in evanescent asci in dried masses (mazaedia), shows a similarity to mazaediate lichens such as Calicium Pers. It is closely related to helotiaceous fungi, in particular Hymenoscyphus Gray and Cudoniella Sacc. (Kirchmair et al. 2008).
Roesleria subterranea is distributed primarily in Europe and North America, with rare reports from Asia. In Japan, three presumptive occurrences of R. subterranea have been reported. Roesleria hypogaea Thüm. & Pass. was first documented in Japan by Shirai (1894) as a possible agent of root rot in grapes. Hara (1930) later cited the fungus as of an uncertain occurrence in Japan. Togashi (1950) listed the name Coniocybe pallida (Pers.) Fr. as a causative agent of grape root rot and reported its occurrence in Hokkaido and Akita prefecture. This name was cited in fungi from the Tohoku area (North East of Japan) by Sawada (1952) as a synonym of Roesleria pallida (Pers.) Sacc., (Sawada 1952, p. 143). The fungus was thought to be a root rot agent of Vitis vinifera in Morioka. Currently, Roesleria hypogaea is treated as a synonym of R. subterranea, while R. pallida is recognized as Sclerophora pallida (Fr.) Yao & Spooner (Coniocybaceae based on Calicium pallidum Fr.), which was revealed to be a lichen name (Redhead 1984, Yao andSpooner 1999). Therefore, R. subterranea occurs in the northern part of Japan, but none of the authors provided a definite description of the species, nor are voucher specimens available.
A specimen of R. subterranea was recently collected from unknown deciduous tree roots in the red pine forests in Sugadaira heights, Nagano prefecture, Japan, from which an isolate was obtained. Because R. subterranea is a subterranean fungus with a unique phylogenetic position as a helotiaceous fungus, it is documented here with molecular data and reference to the specimens in Japan.

Collection and isolation
A specimen was collected on roots of unknown trees beneath the log of Pinus densiflora, among the conifer forests of 30-year-old trees mixed with other deciduous trees such as Betula platyphylla, Cerastrus orbiculatus, Ligustrum tschonoskii, Rhododendron japonicum, Swida controvers.Although Vitis coignetiae is present in this forest, no individual was found nearby the collection site of the fungus, so it seems likely plausible that the host was not Vitis. The material was air-dried at 23°C for 24 hours. A small amount of spores were picked up by a fine needle and transferred to plates of malt extract agar (MEA; Nissui, Tokyo, Japan), cornmeal agar (CMA; Nissui), and half-strength malt extractyeast extract agar (MEYE; malt extract 3 g; yeast extract 3 g, peptone 5 g, dextrose 10 g, agar 15 g, DW 1 L). Germination was observed under a light microscope to obtain a pure culture. The specimen and isolate were deposited in the mycological herbarium of the National Museum of Nature and Science (TNS) and the National Institute of Technology and Evaluation, Biological Resource Center (NBRC). The color names and codes in the description followed the Pantone color bridge (Pantone Inc., Carlstadt, NJ, USA), adopting the CYMK color system. Additional specimens collected in Japan were investigated in the TNS fungal herbarium.

DNA extraction, polymerase chain reaction (PCR), and sequencing
The isolates were incubated in 2% malt extract broth for 2 weeks, and the mycelium was harvested. Approximately 50 mg of mycelium was mechanically lysed by a Qiagen Tissue Lyser Kit (Qiagen Inc., Mississauga, ON, Canada), using ceramic beads following the manufacturer's instructions. The DNA was extracted using a DNeasy Plant Mini Kit (Qiagen Inc.) following the manufacturer's instructions. To amplify the internal transcribed spacer (ITS1 and ITS2) and 5.8S rDNA regions (ITS-5.8S), the primer pair ITS1F and ITS4 (White et al. 1990) was used. To amplify the D1-D2 region of 28S rDNA (nLSU rDNA; D1-D2), the primer pair NL1 and NL4 (O'Donnell 1993) was used. The DNA was amplified using 40-μl PCR reactions containing 0.2 μM of each primer, 1 unit of TaKaRa Ex Taq DNA polymerase (Takara Bio, Otsu, Japan), and a deoxynucleoside triphosphate (dNTP) mixture containing 2.5 mM of each dNTP and ExTaq buffer containing 2 mM MgCl 2 . PCR was carried out using a Gene Amp PCR system 9700 (Applied Biosystems, Foster City, CA, USA). The DNA was denatured for 3 min at 95 °C, followed by 30 cycles of denaturation at 95 °C for 30 s, annealing at 50 °C for 30 s, and extension at 72 °C for 2 min, followed by final extension at 72 °C for 10 min. Residual primers, and unincorporated dNTPs were removed to prevent the inhibition of the following sequencing reaction using a ExoSAP-IT purification kit (USB Corp., Cleveland, OH, USA). Total DNA samples extracted for the present study were deposited in the Center for Molecular Biodiversity Research in National Museum of Nature and Science (ED16207). The sequencing reaction was carried out using a Big-Dye terminator cycle sequencing kit (Applied Biosystems) following the manufacturer's instruction using the same primers as those used for DNA amplification. Sequencing was conducted using an ABI 3130x Genetic Analyzer (Applied Biosystems).

Molecular phylogenetic analysis
The available sequences for Roesleria subterranea were obtained from GenBank (Table 1). The obtained sequence data for the ITS-5.8S region were aligned by Clustal W (Thompson et al. 1994) and edited manually when necessary using BioEdit v.7.0.5.2 (Hall 1999). Phylogenetic analysis was conducted by a maximum parsimony (MP) method. An MP heuristic search was carried out using the PAUP* version 4b10 (Swofford 2002) with 1,000 replications, each with the option of random sequence additions. Branch swapping by tree bisection-reconnection (TBR) and MulTrees were in effect. Support for the individual nodes was tested with bootstrap analysis under the equally weighted parsimony criterion. The bootstrap analysis was based on 1000 bootstrap replicates using the heuristic search option (TBR and MulTrees options, on) of ten replicates with random addition sequences. Based on the previous analysis (Kirchmair et al. 2008), Hymenoscyphus epiphyllus (Pers.) Rehm ex Kauffman (AY348580), H. immutabilis (Fuckel) Dennis (AY348584), and H. scutula (Pers.) W. Phillips (AY789432) were used as outgroups. The trees were visualized using the Treeview program, version 1.6.6 (Page 1996). The alignments were deposited into TreeBASE as TB2:S11877.

Results and discussion
Description and taxonomy Note. For the detailed synonymy, see Redhead (1984). We found two specimens (TNS-F-185301 and 185302) of R. subterranea deposited in TNS as Calicium pallidum, which were reported by Togashi (1950). Although Calicium pallidum is a lichen name, it is also a misapplied name for R. subterranea. We did not attempt DNA extract analysis of TNS-F-185301 and TNS-F-185302 because the specimens lacked sufficient material to guarantee a DNA, and in our experience, such DNA is damaged due to repeated fumigation.

Genetic variations and geological distribution
The sequence of the D1-D2 region of 28S rDNA was almost identical to the available sequences from European specimens. The present sequence differed from EF608073, Figure 1. Roesleria subterranea (TNS-F-38701). 1 Growing habits of two mature mazaedia on the root tip of an unidentified tree 2 Hymenium of mazaedium, showing protruding paraphyses 3 Enlarged hymenium showing developmental stages of asci and paraphyses 4 Mature ascus containing eight ascospores 5 Paraphysis 6 Discharged matured ascospores; on the right, a transversely one-septate spore is shown in side view 7 Germination of ascospores on MEYE, showing that germination tubes from one of the two-celled spores 2-6 Mounted in Meltzer's solution 7 Mounted in lacto-aniline blue. Bars: 1 = 2 mm; 2 = 50 μm; 3-7 = 10 μm. EF608074, and EF608075 at three, one, and four sites, respectively, among the 554 sites that were compared. The ITS-5.8S sequence was almost identical to the available sequences, and the identification based on morphology was confirmed.
In the MP analysis, we incorporated the available ITS-5.8S sequences of R. subterranea, using the three species of Hymenoscyphus used in Kirchmair et al. (2008) as outgroups. The aligned dataset included 466 characters, of which 28 characters were excluded from the analyses due to ambiguous alignment. The remaining 438 characters were used for the analysis, including a total of 23 parsimony-informative characters.
A single most parsimonious tree was obtained (Fig. 2). Within the tree, a monophyly of isolates from Vitis spp. from Germany, Italy and USA was strongly supported. The remaining group of this clade were found to be paraphyletic and composed of the isolates from various deciduous trees. The sequence obtained from the Japanese isolate (NBRC 108276) was included in this group (Fig. 2). Although the sequence of the latter group seemed to be identical in Fig. 2, the original sequence obtained from the Japanese isolate (AB628057) was not identical with several nucleotide differences to the other three. However, these differences were present only in the ambiguously aligned region, which was excluded from the analysis. Although the tree topology depends on the choice of outgroup (data not shown), the monophyly of the species as a whole and the monophyly of the clade of isolates from Vitis ("Vitis-clade") were highly stable (BS values were 100 in the present data and also in Kirchmair et al. (2008)). However, a group of isolates from other hosts besides Vitis ("non-Vitis group") was paraphyletic. This result suggests that the host specificity was more important than the geographical distance as a factor for infraspecific genetic diversification.
All of the known isolates of the "Vitis clade" were obtained from cultivated grape samples. On the other hand, isolates of "non-Vitis grade" including our isolate were from the roots of Malus sylvesris, Populus sp. and other unknown wild deciduous trees. It is therefore hypothesized that isolates highly specific to Vitis are more widely distributed in the cultivated grape fields than is currently recognized. To clarify whether such a host specificity is present or not, inoculation experiments are indispensable. In addition, further isolates from various hosts and localities should be collected and examined to know the infraspecific relationships of the species. Тhe taxonomic status of one isolate (GE09) is addressed, for which data were previously submitted to GenBank (five ITS cloned from this single isolate; EU414755-EU414759) under the generic name Anaeromyces (Edwards et al. 2008). More detailed genetic analysis has been conducted to clarify its taxonomic position, along with its morphological characterisation. Microscopic examination and phylogenetic reconstruction, using both the ITS1 and LSU regions of the rRNA locus, showed that GE09 and three other similar isolates (isolated from different host species) were clearly distinct from the main Anaeromyces clade. We therefore conclude that these four isolates represent a new genus, which we name Buwchfawromyces gen. nov.
Freshly voided faecal samples (ca. 20 g) were collected and transported to the laboratory within 1 h. Aliquots (ca.10 g fresh matter) of these samples were then homogenised for 2 min with 90 ml of pre-warmed (39 °C) basal medium (without wheat straw, yeast extract or tryptone) in a pre-sterilised Stomacher 400 Circulator Bag (polythene; 177 × 305 mm) using a Seward Stomacher 400 Circulator Paddle Blender (Seward Ltd., Worthing, W. Sussex, UK). A 10-fold serial dilution of this faecal homogenate was then prepared in pre-warmed basal medium (1 ml transferred to 9 ml basal medium in 15 ml Hungate tube). The 10 -4 , 10 -5 and 10 -6 dilutions were used to inoculate (1 ml) the tubes of basal medium (9 ml) supplemented with wheat straw. A mixed solution of penicillin, ampicillin and streptomycin sulphate in 50% (v/v) ethanol (5 mg.ml -1 of each; 10 ml.L -1 added to give a final medium concentration of 50 µg.ml -1 ) was also added to tubes before they were recapped, in order to inhibit bacterial growth. Tubes were incubated in the dark at (39 °C), and routine subculture was conducted at 3-5 d intervals. Exposure of the samples to oxygen was prevented by undertaking the manipulations in a box flushed with CO 2 . Cultures were also grown on basal medium containing cellobiose (5 mg.ml -1 ) instead of wheat straw. Purity of the isolates was ensured by three cycles of cultivation in roll tubes (Joblin 1981), with well-separated colonies being excised for subculture with a mycological spear.
For cryopreservation of cultures, a method based on the procedures suggested by Sakurada et al. (1995) and Yarlett et al. (1986) was devised. A 5× cryopreservation solution was prepared by mixing 49.7 g ethylene glycol with 155 ml clarified rumen, 200 µl of 0.1% resazurin, 0.2 g L-cysteine and 1.2 g NaHCO 3 under anaerobic conditions (total volume 200 ml; 3.2 M ethylene glycol). This solution was bubbled with CO 2 for ca. 3 h, prior to anaerobically dispensing 10 ml aliquots into Hungate tubes which were autoclaved and stored at -20 °C until use. The 5× cryopreservation solution was added to 3-5 d old wheat straw cultures (2.5 ml per 10 ml culture) under anaerobic conditions in Hungate tubes. After mixing, tubes were chilled in ice water for 15 min and then anaerobically dispensed in 2 ml aliquots into sterile 2 ml cryovials. Cryovials were placed at -80 °C overnight, before being transferred to liquid nitrogen for longer-term storage. Storage of cryopreserved cultures at -80 °C is possible for up to a few months but for prolonged storage, liquid nitrogen was found to be much more reliable.

Microscopy
Microscopy was conducted on cultures, grown on either wheat straw or cellobiose, using an epifluorescence microscope (Olympus BX50) with images recorded using a Nikon Coolpix 995 digital camera. For visualisation of nuclei, DAPI (0.3 mg.ml -1 in 50 mM Tris-HCl, pH 7.2) was added, and for enhanced definition of cell walls and septa, Calcofluor white M2R (100 µM; Day et al. 2002) was used (UV-W filters: 365 nm excitation/420 nm emission).

Phylogenetic analysis
DNA extraction was carried out using the CTAB (hexadecyltrimethylammonium bromide) method of Doyle and Doyle (1987), with modifications as described by Griffith and Shaw (1998). Cultures were harvested after 3-6 d incubation, and the biomass washed three times with sterile distilled water before being freeze-dried and ground to a powder. Ground biomass (ca. 50 mg) was used for DNA extraction, and the purified DNA resuspended in 50 µl TE buffer (pH 8.0) before being stored at -20 °C.
Phylogenetic reconstruction was conducted using TOPALi (v2.5) (Milne et al. 2004). For LSU analysis, maximum likelihood analysis was conducted using PhyML (Guindon et al. 2010) and the TrN+gamma substitution model recommended by TOPALi. For analysis of the ITS, only the ITS1 region was used since the great majority of available sequences cover only this region (and not ITS2). To establish the phylogenetic position of isolate GE09, ITS1 sequences derived from five clones were aligned with environmental and isolate sequences belonging to the four closely related monoflagellate genera. The resulting MAFFT alignment was trimmed to include only 15 bp of the flanking 18S and 5.8S regions, and duplicate sequences were removed (retaining sequences relating to cultured fungi, if present). For ITS1 analysis, maximum likelihood analysis was conducted using PhyML (Guindon et al. 2010) and the GTR substitution model recommended by TOPALi. Both the ITS1 and LSU alignments have been submitted to TreeBase (http://purl.org/phylo/treebase/phylows/ study/TB2:S16672).

Morphology
Thalli of isolate GE09, when grown on wheat straw or cellobiose as a carbon source, were consistently monocentric, with rhizoids radiating from a single developing sporangium. Mature sporangia were spherical to ovoid 30 to 80 µm long and 20 to 60 µm wide ( Fig. 1). No apical projections, as found in Anaeromyces mucronatus (Breton et al. 1990) or Piromyces mae (Li et al. 1990) (referred to as a mucro or papilla respectively by these authors), were observed.
Zoospores (spherical; diameter 9-11 µm) were readily observed in 3-5 d old cultures grown on wheat straw and consistently bore a single flagellum (30-40 µm long; 3-4× longer than the length of the zoospore body). However, the process whereby zoospores were released from the sporangium was not observed. On a single occasion, a large (30 µm diameter) zoospore-like structure bearing numerous flagella, each emerging from a different point on the zoospore body, was detected (Suppl. material 1A). This is possibly the result of the agglomeration and fusion of numerous zoospores (hence the numerous flagella that are visible), similar to the structure previously observed by Orpin (1975) in N. frontalis (Orpin's Fig. 2;Suppl. material 1B).
The most distinctive feature of the thalli of GE09 were the swollen sporangiophores (40-80 µm long and 15-50 µm wide), occasionally comparable in volume to the sporangium they supported ( Fig. 1E-H). Also visible was a septum at the point where the sporangium joined the sporangiophore (Fig. 1A, E-I). The sporangiophore was contiguous with the rhizoids which tapered (from 20 µm to 5 µm) and branched (Fig. 1B). The proximal rhizoids (within 100 µm of the sporangium) were often contorted (Fig. 1B). DAPI staining was used to observe the location of nuclei within the thalli (Fig. 1F, H, I); these were abundant in sporangia but none were observed in the sporangiophores or rhizoids.
The swollen sporangiophores and twisted rhizoids observed here are very similar to those noted by Ho et al. (1993b) in Piromyces spiralis (Suppl. material 1C,D). Also similar are the swollen sporangiophores reported in P. mae (Li et al. 1990) (isolate PN11 from horse; Fig. 25). It is also noteworthy that Anaeromyces (formerly Piromyces) polycephalus (Suppl. material 1E) (Chen et al. 2002;Kirk 2012), whilst forming multiple rather than single sporangia, also forms a distinctly swollen sporangiophore. It is possible that such structures play some role in physical disruption of the substrate, as is the case for the bulbous holdfasts formed by Caecomyces and Cyllamyces spp.
Cultures of GE09 maintained viability, and could be subcultured, after incubation at 39 °C for periods of several weeks. This raised the possibility that these cultures may form long-term survival structures (McGranaghan et al. 1999;Struchtemeyer et al. 2014). Thick-walled and septate structures (3-4 septa; 30-40 µm long × 10-15 µm wide), very similar to those previously observed in Anaeromyces sp. EO2 (Brookman et al. 2000) were seen in wheat straw cultures incubated for 28 d (Suppl. material 1F, G) and never observed in younger (3-5 d old) cultures. However, detailed examination of the development of these putative resting structures was not undertaken.

Phylogenetic analysis
Detailed examination of isolate GE09 was not undertaken when it was first isolated. However, it was used as a reference sample in a study of the colonisation of forage by anaerobic fungi (Edwards et al. 2008), in the course of which the ITS1/2 spacer regions were amplified and cloned. The sequences of the five clones were submitted to GenBank (EU414755-EU414759) under the generic name Anaeromyces, since these sequences clustered close to the Anaeromyces clade at that time. However, more detailed analysis has since suggested that this isolate is quite distinct from Anaeromyces (Kittelmann et al. 2012).
The sequence of the D1/D2 domains (ca. 750 bp) of the LSU of GE09 and the three other isolates were identical (submitted to GenBank as KP205570). These sequences were aligned with 36 other LSU sequences (from GenBank) covering all the known genera of Neocallimastigomycota (700 bp alignment; 188 phylogenetically informative sites), and including the outgroup taxon Gromochytrium mamkaevae (Chytridiomycota). Phylogenetic reconstruction consistently recovered Buwchfawromyces isolates as a distinct clade (85% bootstrap support). LSU sequences from Anaeromyces, Neocallimastix and Orpinomyces were also recovered as distinct clades with strong (≥80%) bootstrap support (Fig. 2). Neocallimastix and Orpinomyces were more closely related to each other than to other genera, consistent with the occurrence of polyflagellate flagella in these genera, a feature not found in other flagellate fungi (James et al. 2006). The genera forming bulbous holdfasts (Caecomyces, Cyllamyces) also formed a distinct clade, and Piromyces isolates occupied a basal position but without strong bootstrap support.
Whilst analysis of the LSU proved informative to confirm the distinctiveness of the GE09 clade, there are relatively few LSU sequences available in GenBank for comparison. Therefore, phylogenetic analysis of the ITS1 internal transcribed spacer region, for which there are hundreds of published sequences, was conducted (357 bp alignment; 271 phylogenetically informative sites, 101 sequences) (Fig. 3). Since analysis of the LSU sequences had shown members of the genera Neocallimastix and Orpinomyces to be quite distinct, these were omitted from analyses of the ITS1 region in order to improve the quality of the alignments (fewer gaps). Additionally, 'environmental' sequences obtained from clone library studies, (mostly labelled as "uncultured Neocallimastigales") were also included, as recommended by Nilsson et al. (2011).
The Buwchfawromyces (GE09) clade was again recovered with high (92%) bootstrap support, as was the Anaeromyces clade. Five environmental sequences, all from New Zealand (from cow or red deer; Fig. 2 and listed in Suppl. material 4) also fall into the Buwchfawromyces clade. This clade was also identified (and denoted as clade SK2) in the recent study by Koetschan et al. (2014), who were able to create a reliable phylogeny by using predictors of ITS1 folding to decrease the effects of gaps in anaerobic fungal ITS1 alignments.
The largest survey of anaerobic fungi conducted to date is that of Liggenstoffer et al. (2010). They obtained ca. 250,000 ITS1 sequences, from the faeces of 30 herbivore species kept in Oklahoma zoos (Liggenstoffer et al. 2010), including three of the four herbivore species from which we cultured Buwchfawromyces. It was surprising that such a large dataset should yield no sequences falling into the Buwchfawromyces clade. However, examination of the primers used by Liggenstoffer et al. (2010) revealed the presence of mismatches between the forward primer (MN100modified; 5'-TCCTACCCTTTGTGAATTTG-3') and the cognate sequences found in members of the Buwchfawromyces clade (TCCTACCCTTTGTGAATTGT or TC-CTTACCCTTTGTGAACTGA) (Suppl. material 2). These mismatches would very likely have caused significant primer bias and thus poor amplification of any Buwchfawromyces spp. present.
Five cloned ITS1/2 PCR amplicons of GE09 were originally submitted to Gen-Bank (EU414755-EU414759). These reveal an extremely high level of sequence divergence (<27 polymorphisms within the ca. 200bp ITS1 region; 87.1%-99.5% identity, (Suppl. material 3). High levels of intragenomic variation has also been found in other anaerobic fungi (Ozkose 2001), and also in some other fungal taxa, for instance phylum Glomeromycota (Pawlowska and Taylor 2004;Pringle et al. 2000). For GE09, the most divergent of these clones (EU414756) was more distantly related to the other GE09 clones than were sequences from New Zealand (Fig. 3). A cut-off level of 97% identity is often used to define species or OTUs (operational taxonomic units) in mycology (Nilsson et al. 2008;Yamamoto and Bibby 2014). However, such high levels of intragenomic variation, make it very difficult to generate reliable species hypotheses (Koljalg et al. 2013) for the Neocallimastigomycota based on ITS sequences alone, although the delineation of different genera is still possible.
We consider that isolate GE09 and the three other similar fungi also isolated in the Aberystwyth area represent a new genus Buwchfawromyces within the phylum Neocallimastigomycota. It is not possible for Buwchfawromyces to be placed within the related genus Anaeromyces, since it forms a monocentric thallus not consistent with the circumscription of this genus (Breton et al. 1990) which comprises species with polycentric thalli. Close to Buwchfawromyces and Anaeromyces, is the species currently known as Anaeromyces (formerly Piromyces) polycephalus. This species has a distinctive morphology (Suppl. material 1E), with multiple sporangia arising from a swollen spo-rangiophore, and with an anucleate rhizoidal system. It was originally isolated and described from buffalo in Taiwan (isolate W-33; (Chen et al. 2002)) and has since been reported from India (isolate CTS-47 from zebra faeces; GenBank EU330178); this clade was denoted DT1 (Fig. 3) by Koetschan et al. (2014). Given that 'A. polycephalus' is both morphologically and genetically distinct, this species should be renamed.
The unusually high level of intragenomic variation in ITS1 makes it difficult to nominate a single reference sequence, therefore two are presented (EU414755 and EU414756).
Etymology. From the Welsh words for large cow ('buwch fawr'), since the original isolate GE09 was isolated from a buffalo for which there is no Welsh word. The specific epithet in honour of our former colleague Gary Easton who isolated this fungus.
Nomenclature. Buwchfawromyces Callaghan, Tony & G.W. Griff., gen. nov. Strictly anaerobic fungus forming a monocentric thallus with a single sporangium, usually borne on a swollen sporangiophore connected to a branching and twisted rhizoidal system. Zoospores are spherical and uniflagellate. "Buwch fawr" means large cow in Welsh.

Conclusions
It has been apparent since the widespread use of DNA sequence data to identify anaerobic fungi that many of the sequences currently lodged with GenBank do not fall into any of the currently recognised genera of the Neocallimastigomycota. The situation is further complicated by the presence of many sequences from isolates which are very likely misidentified, a phenomenon which is also problematic for other groups of Fungi (Schoch et al. 2014).
A new genus is described based on on a pure culture which was isolated a decade ago and for which ITS1 sequence has been lodged with GenBank since 2008. The RefSeq project aims to resolve this, and other related issues, by linking ITS and LSU sequences from vouchered reference specimens to accepted names (Schoch et al. 2014). As shown above, ITS sequences for anaerobic fungi can be problematic due to intragenomic sequence variation. Similar problems have not been found for the more conserved LSU region of the rRNA locus (Dagar et al. 2011;Eckart et al. 2010), highlighting the synergy of using both loci. We also note that analysis of LSU data does have the distinct advantage of yielding robust alignments across a wide range of basal fungal taxa (including Chytridiomycetes) and being amenable to direct sequencing of PCR products.
In this paper, we describe and illustrate a new species of Jahnula that was found on submerged wood collected from a freshwater river on Martinique Island, Lesser Antilles.

Taxonomy
Etymology. From Latin "purpureus" referring to the characteristic staining of the substrate purple by this species.
Anamorph. Not known. Known distribution. Martinique, Lesser Antilles (Known only from type locality thus far).

Discussion
Jahnula purpurea differs from all other species of Jahnula in that it stains the wood on which it grows purple. In addition, it is one of the species of Jahnula, which has minute ascomata (125-185 µm diameter). The only other species reported to have minute ascomata is Jahnula marakotii Sivichai & Boonyeun, a species reported from a submerged wood test block of Azadirachta indica from a peat swamp in Thailand (Sivichai and Boonyeun 2010). Jahnula purpurea differs from J. marakotii in a number of morphological characters such as shape and size of asci and ascospores. The asci of J. purpurea are clavate to obclavate (90-98 × 22.5-25 µm), while those of J. marakotii are cylindrical (107.5-120 × 9-11.5 µm). The ascospores of J. purpurea do not possess apical appendages and are larger in size (24-28 × 8-9 µm), while those of J. markotii are equipped with bipolar, hyaline apical appendages and are shorter (17.5-20 × 5-6.5) in size.
Jahnula purpurea should also be compared to the type species of the genus, J. aquatica, in that the ascospores of the two species look morphologically similar at first glance. The two species are however, quite distinct. Jahnula purpurea has smaller ascomata, clavate asci, and smaller ascospores, while J. aquatica has larger ascomata, cylindrical asci, and larger ascospores (Hawksworth 1984;Hyde and Wong 1999;Raja and Shearer 2006). In addition, J. purpurea stains its wood substrate purple (Fig. 1A-C), which has never been reported for substrates on which J. aquatica occurs (Hawksworth 1984;Raja and Shearer 2006). Another species of Jahnula that is morphologically similar to J. purpurea in overall ascomata and ascospore morphology includes J. australiensis. On closer examination, however, the species are quite distinct. Jahnula purpurea stains subtending wood purple, a character not observed in J. australiensis; the asci in J. purpurea are clavate to obclavate, while those of J. australiensis are cylindrical; the ascospores of both the species are some what similar in size (24-28 × 8-9 µm in J. purpurea vs. 19-30 × 6-8 µm in J. australiensis) but those of J. purpurea are slightly smaller and wider.
Several species of freshwater ascomycetes in the family Amniculicolaceae are capable of staining underlying wood substrates purple (Zhang et al. 2009a;Zhang et al. 2009b;Zhang et al. 2012), a characteristic similar to J. purpurea. However, Amniculicolaceae is phylogenetically related to the Pleosporales, while based on morphological data presented herein J. purpurea belongs to the Jahnulales (Pang et al. 2002;Campbell et al. 2007;Suetrong et al. 2011).
In addition, species of Massariosphaeria such as M. phaeospora (E. Müll.) Crivelli has the ability to stain wood purple (Zhang et al. 2012). Lophiostoma purpurascens (K.D. Hyde & Aptroot) Aptroot & K.D. Hyde reported from submerged wood in freshwater habitats in Australia and Papua New Guinea also stains wood purple (Hyde and Aptroot 1998;Hyde et al. 2002). However M. phaeospora, and L. purpurascens are phylogenetically unrelated to the Amniculicolaceae (Zhang et al. 2012). Another recently described species, Massariosphaeria fridae M. Spooren stains its substrate (dead stalks of Alisma plantago-aquatica) red (Spooren 2007). This suggests that unrelated freshwater Dothideomycetes have the ability to stain the vicinity of their substrate red or purple. It would be interesting to study the secondary metabolites from these taxa in the future to understand if there is an underlying ecological significance to the production of bright pigments by species of both terrestrial and freshwater ascomycetes within the Dothideomycetes. Additional collections and molecular sequence data from J. purpurea in the future would certainly shed light on the hypothesis of purple pigment being a true phylogenetic informative character at the family-level within the Dothideomycetes as suggested by Zhang et al. (2012). The production of purple pigment on wood by J. purpurea suggests that pigment production might be a case of convergent evolution that could be functionally significant within the Dothideomycetes. It is also likely that different colors might be characteristic of different phylogenetic clades and might have similar or dissimilar ecological functions. Zhang Y, Fournier J, Crous PW, Pointing SB, Hyde KD (2009a)  factors such as climate and dispersal ability. Geographical partitioning may be the rule for basidiomycete taxa, rather than the exception (Taylor 2008;Taylor et al. 2000).
In agaric systematics, discrepancy among parameters used to make taxonomic judgments at species rank is becoming more widely recognized. Three such standards, DNA sequence data, sexual compatibility and morphological characters of basidiocarps have evolved as important in taxonomic judgments, including proposal of new taxa [Lentinus (Grand et al. 2011); Megacollybia ; Lentinellus Petersen and Hughes 2004); Sparassis crispa complex ; Artomyces (Lickey et al. 2003)], and revision of generic or species complexes [(Panellus stypticus (Jin 2001); Marasmius scorodonius (Gordon and Petersen 1998); M. androsaceus (Gordon and Petersen 1997); Strobilomyces (Sato et al. 2007); Omphalotus (Kirchmair et al. 2004;Petersen and Hughes 1998); Pleurotus (Vilgalys et al. 1993;Vilgalys and Sun 1994); Gymnopus s. l. ]. As phylogenetic analyses based on molecular data have increased, it has become increasingly clear that genetic differentiation of fungi may proceed in the absence of observable morphological change in basidiocarps. Taylor et al. (2006) note that eukaryotic microbes (predominantly fungi) have few discriminating morphological characters on which to base species assessments. Further, the basidiomycete fungal mating system acts to slow or prevent establishment of reproductive barriers (James et al. 1999). In the absence of morphological change and reproductive isolation, delineation of taxa may rest on evaluation of observed genetic change among populations.
Few primary literature sources are available for identification of Gymnopus s.str. Peck's proposals of new species, often under Marasmius, were included by Halling (1983) in a publication intended to summarize the taxa (under Collybia) of northeastern USA and adjacent Canada. Murrill (Kimbrough 1972) accepted Gymnopus Roussel to represent the "collybioid" taxa of Florida and described several taxa of Gymnopus (i.e. Collybia) and Marasmius s.l. from central Florida [see Kimbrough (1972); Halling (1983)]. This left a geographical hiatus between the New England region and Florida. Without publishing a summary of his study, Hesler included observations on Murrill specimens in his (Hesler) notebooks (http://trace.tennessee.edu/utk_hesler/), and Smith [see ] also examined Murrill's type specimens. Lennox (1979) published her dissertation on Pacific Northwest collybioid genera, but included only a single species under Collybia s. str. Desjardin (1987) published a summary of collybioid fungi from California which further expanded geographical and taxonomic coverage of Gymnopus.
A molecular discontinuity between European and eastern North American populations of Gymnopus confluens has been known for some years ), most recently implied in a phylogeny placing G. eneficola from Newfoundland, Canada . Additional data now permit a more detailed report as part of a wider project on transatlantic disjunctions in fleshy fungi.
In this study, we evaluate morphology, ability to dikaryotize in vitro and ITS-LSU sequence divergence to determine whether intercontinental allopatric populations of Gymnopus confluens represent separate taxa at some rank. In this paper, we consider the term population to comprise a group of currently interbreeding individuals. This paper is third in a series of papers exploring the nature of transatlantic disjunctions. The first in the series was Hughes et al. (2014), the second is Petersen et al. (in press).

Collections
Field collections made by the authors were dried overnight on the day they were collected. Prior to drying, a fragment was stored in silica gel for later DNA extraction and spores were deposited on malt extract agar (MEA) plates to obtain cultures. Collections were accessioned into TENN, cultures into CULTENN.

Pairing experiments
Establishment of single basidiospore isolates (SBIs) and pairing experiments were performed as described in Gordon and Petersen (1992). All SBIs were examined microscopically to determine monokaryon status before mating experiments.

Macromorphology
Macromorphological characters observed were stipe length, stipe vesture, color of living and dried material, and lamella structure (distance between lamella, number, narrow vs. broad, attachment). "Complete" lamellae (those which extended from pileus margin to attachment juxtaposed to stipe) were interspersed with numerous lamellulae, usually of at least two and occasionally three ranks. The most accurate assay counted all lamellae (and lamellulae) which reached the pileus margin. This tally was performed by counting lamellae for approximately ¼ pileus circumference and multiplying by four.

Micromorphology
Basidiospore statistics were gathered for "European" vs. "American" collections. Two spore metrics were especially examined: Q m (median ratio of spore length to width) and L m (median spore length). Pileipellis hyphae were examined for presence of short side branches reported by Antonín and Noordeloos (2010). Terminal cell shape of these side branches was noted. Cheilocystidia vary within subg. Vestipedes, and were evaluated for G. confluens in both size and shape (i.e.non-strangulate to strangulate).

Molecular studies
DNA was extracted from dried specimens and/or cultures, and ITS and LSU sequences were amplified and sequenced as described in Hughes et al. (2013). Sequences were deposited in GenBank (Table 1). Sequences were manually aligned using GCG (2000). PhyML with 100 bootstrap replicates was performed in Geneious V9. Gymnopus eneficola ) was selected as the outgroup because it is the most closely related species and is sister to G. confluens. G. eneficola is often mistaken for G. confluens in the field. Trees were visualized in TreeView (Page 1996) and deposited in Dryad (Petersen and Hughes 2015).

Species-delineation metrics
Several species-delineation metrics including Rosenberg's P AB statistic (Rosenberg 2007), P ID (strict) and P ID (liberal) (Ross et al. 2008), P RD (Rodrigo et al. 2008) and PTP (Zhang et al. 2013). P AB , P ID (strict), P ID (liberal) and P RD were implemented in Geneious V9 (Geneious 2005;Masters et al. 2011). PTP was implemented at http://species.h-its. org/ using web default settings for generations (100,000) burn-in (0.1, MCMC convergence was reached) and thinning (100). Rosenberg's P AB statistic is the probability that a putative species will be monophyletic with respect to a sister clade under the model of random coalescence. The null hypothesis is that monophyly is a chance outcome of random branching. The P ID statistics provide the frequency with which a member of a putative species can be correctly identified given a specific alignment of sequences. P ID (strict) requires that an unknown specimen falls within but not sister to the species clade. P ID (liberal) requires that an unknown specimen falls either sister to or within the species clade. P RD (Probability Randomly Distributed) is the probability that a clade has the observed degree of distinctiveness due to random coalescent processes. A probability value less than 0.05 rejects the null hypothesis of random coalescence and suggests that the clade is a cryptic species. PTP (Poisson Tree Processes; Zhang et al. 2013) estimates the number of species using both maximum likelihood and Bayesian approaches.

Data resources
The data underpinning the analyses reported in this paper are deposited in the Dryad Data Repository at doi: 10.5061/dryad.8239h.

Results
Gymnopus confluens basidiomata from North America and Europe are shown in Figs 1, 2.

Morphological parameters -macromorphology
Macromorphological characters readily distinguish G. confluens (at least in Europe and North America) from other Gymnopus taxa. In nature and in herbarium specimens, basidiomata generally exhibit long stipes compared to pileus diameter. Lamellae: Lamellae in G. confluens appear to be quite consistent; crowded, narrow and significantly seceding upon drying. In an attempt to statistically measure the first two items, lamellae in numerous collections were carefully examined for breadth (rarely exceeding two mm) and number. The number of lamellae reaching pileus margin ranged from 116-147, consistently more than 120, and with no discernible intercontinental difference. In similar morphological taxa (i.e. G. subnudus, G. eneficola, etc.) this number ranged from 65-87, significantly fewer than in G. confluens.  Stipe: Pileus Ratio: Overall, stipe length:pileus diameter usually exceeded 3:1 (with range from 1.5:1 to 6:1). This ratio varied little between North American and European populations, but both clades exhibited some ratios downward, usually explainable due to dry weather or poor nutrition.
Stipe vesture: In most cases, stipe vesture is most sparse upward on the stipe and there (at 10X) exhibiting densely distributed spikes. Downward, vesture becomes denser, and toward the stipe base, a felty subiculum subsumes individual spikes and often is strigose. In dried material, stipe vesture takes on a gray coloration, sometimes with a very slight olive tint. Stipe vesture varied considerably in both populations/ clades. In fact, vesture variation within the major populations exceeded that between clades. No suitable metric was devised to summarize this situation, but macroscopic vesture characters did not prove distinctive.

Morphological parameters -micromorphology
Variation in spore shape is shown in Figs 3 and 4. Spores of at least three European collections were shorter (and somewhat narrower) than those of eastern North American material, and also tapering more obviously proximally (TENN-F-67865 GE, TENN-F-67882 GE, TENN-F-59212 FR). Spores of most European collections approach the metrics of eastern North American material (Table 2). A few collections of otherwise mature basidiomata were devoid of spores. Whether this is a function of in vivo drying followed by mechanical but not biological resuscitation is not known. Counter to this hypothesis is the presence on such basidiomata of mature (sterigmate), turgid basidia which appear to be fecund.
When spore statistics from numerous collections were compared, little difference was apparent, and spore statistics were concluded to be inconclusive for morphological separation of the phylogenetic clades of G. confluens. Median spore length for North American collections was 7.72 μm (n=10 collections); for European collections it was 7.52 μm (n=10 collections) ( Table 2). This was not significantly different based on a 2-tailed T-test (P=0.40, df=19). Q m for North American collections was 2.34 (n=10 collections); for European collections it was 2.30μ (n=14 collections). Q m values were not significantly different based on a two-tailed T-test (P=0.71). The two collections from Alaska (MICH139598 and MICH139602) which clearly fell within the European clade by ITS sequence, were included in calculations of spore statistics for Europe.
Cheilocystidia: Cheilocystidum size and shape vary within Gymnopus subg. Vestipedes, and this variation was closely examined for numerous collections of G. confluens from both continents. Variation in both size and shape (i.e.non-strangulate to strangulate) was expected based on previous reports and illustrations. As expected, cheilocystidia varied in abundance, size and shape, but without correlation to geographic origin. Although cheilocystidia of European collections generally appear to be longer and longer-stalked than those from eastern North American specimens, the range of sizes and complexities seems parallel across the two populations. Overall variation of cheilo-  croscopic exudate of slime which conserves water as a microscopic moist chamber in which the apical portion of cheilocystidia can proliferate.
Caulocystidia: In observing caulocystidial hyphae composing stipe vesture, especially those gathered to form the characteristic vesture "spikes," some variation was perceived in the shape of the terminal cells, whether equal (parallel-sided and bluntly rounded at apex), tapering distally (and therefore narrowly rounded at apex), or some variation in shape (i.e. subsagitate, lobed, etc.). Once noted, special care was taken to observe this character. Caulocystidial hyphae are invariably clamped within their emergent length. Caulocystidia, correctly depicted by Antonín and Noordeloos (2010)  often include small side lobes with narrow attachments to parent hyphae. It is easy to observe the erect hyphae which compose the spikes, but more difficult is observation of the subicular hyphae which produce the hyphal complexes which elongate into the spikes. The complexity and depth of the subiculum vary from arachnoid (and then revealing the color and glassy surface of the dried stipe cortex) (TENN-F-59282 SZ; TENN-F-59219 FR) to opaque-felty with subsumed spikes (TENN-F-67865 GE) to shaggy (TENN-F-67882 GE). The felty subiculum increases downward and appears as a sheath toward the stipe base, somewhat reminiscent of Connopus acervatus. Pleurocystidia: Whether pleurocystidia are commonly present has not been thoroughly investigated. Structures resembling cheilocystidia were observed in three collections (TENN-F-59578 RU, TENN-F-67865 GE, TENN-F-59212 FR) among basidia rather than being clustered at the temini of lamellar tramal hyphae. Basidia: Basidia varied little across both continents. Rarely, an individual twospored basidium was detected, and four-spored basidia accounted for almost all mature basidia observed (Figs 5, 6). A basidium is produced as a terminal cell of a subhymenial hypha. The hyphae then proliferates through the subtending clamp connection and another basidium is produced in the same fashion. After several such proliferations and basidial discharge (usually leaving little or no residue), the subhymenial hypha appears asymmetrically notched, and superficially resembles some cheilocystidia (TENN-F-67865 GE), for which they are easily mistaken, especially as seen in considerable numbers in older hymenia.
Side branches from pileipellis: The side branches from pileipellis hyphae reported and illustrated by Halling (Halling 1983) and Antonín and Noordeloos (2010) were consistently observed. Shapes ranged from short, simple and digitate (i.e. TENN- FI) in European specimens than in American. It may be that this proliferation is aided by microscopic local moisture idiosyncratic to individual conditions (including some mucoid matrix if present). The variation was wide both within and between clades, and no clade separation was possible based on this character (Figs 7, 8). Morphological summary: Morphological characteristics based on lamellae, side branches from pileipellis hyphae, cheilocystidia, caulocystidia, and pleurocystidia, varied considerably but observed differences were not continent-specific. Basidia and basidiospores showed little variability and did not differ between continents.

Sexual recognition experiments
Results of three self-crosses of G. confluens (collections TFB 7219, NC; 9048, CA; 11400, SZ) were reported  as showing tetrapolarity, not unusual in Omphalotaceae. This study added TENN-F-69053 (TFB14389), New Brunswick, Canada, which we also determined to be tetrapolar. In all cases, distribution of mating types was unbalanced and in most of these self-crosses, some SBIs were found which exhibited unexplained mating results, mostly ability to dikaryotize two opposing mating types, most easily explained by either harvesting of two hemicompatible basidiospore germlings together, or occurrence of two hemicompatible nuclei within a single basidiospore (Petersen 1995). Such results have also been found in other members of Gymnopus subg. Vestipedes, such as G. subnudus (Murphy 1992;Murphy and Miller 1993;Petersen 1995) which has been reported as bipolar and tetrapolar (Petersen, ined.) In the sexual compatibility study using G. confluens ); Fig. 14], a total of 11 collections were used, seven from Europe and four from North America. Five intercontinental intercollection pairings were performed, all universally sexually compatible. If clamp connection production is accepted as a proxy for sexual reproduction, no apparent prezygotic reproductive barrier exists between the continents. It was concluded, based on limited evidence, that only a single sexual recognition species was involved. No attempt to segregate morphological entities was offered in that study.

Phylogenetic analyses
A PhyML tree based on ribosomal ITS sequences is given in Fig. 9. North American and European collections segregated largely into two distinct clades which share an average sequence identity of 96.75%. Collections from Alaska were placed in both clades. A collection from California was also affiliated with the European clade. For a smaller data set, both ribosomal ITS and LSU sequences were available and were concatenated. Results of the PhyML analysis are given in Fig. 10.
For the ITS data set, Rosenberg's P AB statistic for both European and North American clades was P=1.6 × 10 -8 . Thus, a null hypothesis of reciprocal monophyly under a random coalescence model can be rejected. The probability of correctly identifying an unknown member of a putative species is given by P ID statistics. P ID (strict) European clade = 0.95 (σ=0.89,1.00) and P ID (strict) North American clade = 0.98 (σ=0.93, 1.00). P ID (strict) is the more stringent of the P ID statistics. The probability P RD that a clade has the observed distinctiveness under a null hypothesis of random coalescence for North Figure 9. PhyML tree based on ribosomal ITS sequences. Bootstrap values based on 100 bootstrap replicates are at the left of the supported node. Analysis assumed the GTR model of evolution with the transition/transversion ratio, number of invariable sites and shape of the gamma distribution estimated. The log likelihood of the tree was -1776.7. Bold type = holotype of Gymnopus confluens subsp. campanulatus. Percent identity was based on the entire ITS1-5.8S-ITS2 sequence.
America is 0.05 and for Europe is 0.21. Neither of these probabilities reject the null hypothesis. PTP species-delimitation results produce both maximum likelihood and Bayesian estimates of the number of species. PTP for both analyses partitions G. confluens European and North American populations into two species groups but without significant support (bootstrap support for both analyses was 0.23 for North America and 0.55 for Europe. Outgroup taxa are also partitioned into two species groups. Analysis assumed the GTR model of evolution with the transition/transversion ratio, number of invariable sites and shape of the gamma distribution estimated. The log likelihood of the tree was -3135.8. Bold type = holotype of Gymnopus confluens subsp. campanulatus.

Discussion
Allopatric speciation may be the most common mode of speciation in fungi and other organisms, and separation of populations on different continents would, in the absence of significant gene flow, lead gradually to accumulation of genetic differences and ultimately speciation. The mode and tempo of speciation, however, must vary with reproductive strategies and selection pressures. The point at which two allopatric populations become new species is often a matter of judgment but important in terms of evaluating conservation status and estimating species diversity for a given region. Numerous studies suggest that in basidiomycete fungi, ability of allopatric populations to intercross in vitro is conserved as a function of the unique multiple allelic mating systems even while genetic divergence (usually indicated by differences in nuclear ribosomal ITS sequences) proceeds (Gordon and Petersen 1997;Grubisha et al. 2012;James et al. 1999;Lickey et al. 2002;Lickey et al. 1999;Taylor et al. 2006;Vilgalys and Sun 1994). It should be noted, however, that in vitro compatibility examines only one aspect of reproductive intercompatibility and that failure to produce F 1 fruitbodies, reduced fertility of F 1 hybrids, lack of or inviability of F 2 progeny and competition failure at any level may also be involved in reproductive isolation. These factors have not been evaluated as isolated mechanisms between basidiomycete populations to date.
In Gymnopus confluens, two of three criteria used to evaluate delineation of species (morphology and ability to intercross in vitro) show no significant intercontinental separation. ITS sequences, however, are divergent (3.25% base pair difference), a level often used to suggest different species (Hughes et al. 2009), and European and North American clades are well-supported. The combination of a barcode gap, Rosenberg's P AB statistic results and P ID (strict) suggest that North American and European clades are monophyletic, that the observed differences are not due to coalescence in gene trees and that they are distinct phylogenetic species. In contrast, P RD and PVP do not reject the possibility that the observed result ( Fig. 9) is due to random coalescence. The question then becomes whether to assign these populations nomenclatural rank on the basis of ribosomal ITS and LSU sequences alone. To identify these populations as comprising a single species ignores significant ITS + LSU sequence divergence indicative of speciation processes, and underestimates diversity. To identify these populations as separate species may overstate the degree of genetic divergence. A middle-ground solution seems to be to assign these populations subspecies rank, thus recognizing genetic differentiation. We do so below adhering to, insofar as possible, procedures suggested by Tripp and Lendemer (2014) for naming taxa when molecular evidence is the only evidence available.
The finding that ITS sequences for collections from Alaska represented two distinct ITS entities, one of which falls within the European clade (Fig. 9) suggests a dual origin for collections from this region. An ITS sequence of a collection from California also falls within the European clade. Possibly, these collections represent humanmediated transfer of material from Europe to North America. Alternately, movement from Eurasia to Alaska thence to California may have been feasible via the Bering land bridge during periods of glaciation, but without an understanding of G. confluens from Asia, neither hypothesis be substantiated. Intra-continental geographical partitioning is not clearly evident for either European or North American populations of Gymnopus confluens. This contrasts with findings in some other basidiomycete taxa (Geml et al. 2008;Hughes et al. 2014;Zhao et al. 2013) but intracontinental biogeographical distributions have not been extensively examined and are likely to be species-specific. Taxon diagnosis. 1) ITS nrDNA sequence significantly different from sequence of Gymnopus confluens subsp. confluens; 2) basidiomata densely gregarious to subcespitose; 3) basidiomata apparently persistent beyond spore production and discharge; 4) stipe:pileus diameter ration from 2-5:1 (stipe significantly longer than pileus diameter); 5) pileus hygrophanous, brown where moist, pallid tan to pinkish buff where dry, drying to more uniform pallid color; 6) lamellae very crowded (total lamellae at pileus margin 110-140), shallow, seceding upon drying; 7) lamellar edge entire (smooth) to delicately fimbriate; 8) stipe grooved or compressed, stiff, with brown cortex (rind); 9) stipe vesture concolorous with pileus when moist and fresh, easily bleaching on drying to pallid gray shades; 10) basidiospores generally elongate-ellipsoid to sublacrymiform; 11) cheilocystidia stalked, usually lobed or strangulate, sometimes branched; 12) pileipellis hyphae smooth, firm-walled, with occasional to common side branches appearing digitate to long and branched. 13) Distribution in North America.

Conclusions
Gymnopus confluens in Europe and North America shows intercontinental but not intracontinental divergence in ITS and LSU sequences but European and North American populations do not differ morphologically and retain the ability to dikaryotize in vitro. Intercontinental ITS/LSU sequence divergence is sufficient to recognize differences taxonomically. The North American population is described as G. confluens subsp. campanulatus.