Greenland and Svalbard glaciers host unknown basidiomycetes: the yeast Camptobasidium arcticum sp. nov. and the dimorphic Psychromyces glacialis gen. and sp. nov.

Sampling campaigns in Greenland and Svalbard were executed to explore fungal diversity in cold habitats. Three very abundant groups of strains were discovered, consisting either of recently described or of yet-undescribed psychrophilic and oligotrophic yeasts and dimorphic fungi, accounting for around 50 % of the total cultivable diversity of basidiomycetes in our studies. The occurrence of these taxa has also been demonstrated by culture-independent methods. Based on phylogenetic analyses of ribosomal gene cluster sequences (D1/D2 domains of 28S (LSU), 18S (SSU), ITS with 5.8S rDNA) and sequences of protein-coding genes for elongation factor one alpha (TEF), cytochrome b (CYTB) and two subunits of the RNA polymerase II (RPB1 and RPB2) obtained from pure cultures, the isolated taxa presented in this study belong to Basidiomycota, subphylum Pucciniomycotina, class Microbotryomycetes, family Camptobasidiaceae. The dataset of the sequences supported the recognition of three species: Camptobasidium gelus, Camptobasidium arcticum sp. nov. (ex-type strain EXF-12713) and Psychromyces glacialis gen. and sp. nov. (ex-type strain EXF-13111). Camptobasidium gelus was found in the Svalbard and Greenland samples, while representatives of the here proposed new species, C. arcticum, were found only in the Greenland Ice Sheet. Psychromyces gen. nov. was erected for the dimorphic/filamentous isolates found in Svalbard and Greenland glacial environments. The taxon, for which the invalid name ‘Rhodotorula svalbardensis’ has been used, belongs to this genus. Based on ribosomal genes, Camptobasidium arcticum and Psychromyces glacialis are related, phylogenetically most closely related to the genera Glaciozyma and Cryolevonia. Seven genes phylogeny restricted to taxa with available sequences, supported the placement of Psychromyces to Camptobasidiaceae.

Sampling campaigns in Greenland and Svalbard in 2017, investigating subglacial and glacial environments [15,16], led to the isolation of 168 yeasts and yeast-like isolates, 150 of them belonging to the phylum Basidiomycota. Approximately half of these isolates could not be reliably identified at the species level. About half of the cultivable taxa were also detected through amplicon sequencing of the fungal internal transcribed spacer 2 (ITS2) [15,16]. In an effort to continue the characterization of yeasts from Greenland and Svalbard collections, a new species of Camptobasidium, C. arcticum, is described as the phylogenetic sister of the recently described Camptobasidium gelus [18]. A new genus and species, Psychromyces glacialis, is also proposed to accommodate isolates identical in ribosomal sequences to 'Rhodotorula svalbardensis' . Its taxonomy and status were discussed in several studies following the application of the 'one fungus=one name' system [27].

SAMPLING SITES AND ISOLATION METHODS
Cryoconite, snow, supraglacial ice (clear ice), supraglacial ice dominated by dark pigmented glacier algae (dark ice) and supraglacial water samples were collected from the southwest margin of the Greenland Ice Sheet during a fieldwork campaign in July 2017 [15] ( Table 1). The fieldwork site was located at around 60 km east of Kangerlussuaq and was within the so-called 'dark zone' , characterized by extensive algal blooms [29]. Subglacial ice and glacial meltwater were collected from three polythermal glaciers (Midtre Lovénbreen, Pedersenbreen and Vestre Brøggerbreen), situated in the Kongsfjorden area, Ny-Ålesund (Svalbard, Norway) in August 2017 [16] (Table 1).
All samples were collected with sterilized tools and handled with sterile nitrile gloves to avoid contamination. Samples were collected in sterile Whirl-Pak plastic bags or sterile plastic bottles and processed within a few hours of their collection at the primary ice camp on the Greenland Ice Sheet or the NERC Arctic Research Station (Ny-Ålesund). The samples were filtered and incubated at 15 °C on six different media, listed in Table S1 (available in the online version of this article), as described by Perini et al. [15,16]. A total of 150 basidiomycetous yeasts and yeast-like fungi were isolated from all the above listed samples (Svalbard and Greenland [15,16]). The majority of the strains reported in this study (53 of 74) was isolated from media with high water activity (a w =1) and poor in nutrients, such as minimal medium (MM) [30] and synthetic nutrient agar (SNA) [31], both supplemented with chloramphenicol (50 mg l −1 ), and from Reasoner's 2A (R2A) agar [32]. The remaining strains (21 of 74) were isolated from media richer in nutrients, such as dichloran rose bengal chloramphenicol agar (DRBC) [33] and media with lower a w (MY10-12, DG18) [34].
All strains were deposited in the Ex Culture Collection of the Infrastructural Centre Mycosmo (MRIC UL), Slovenia ( www. ex-genebank. com) at the Department of Biology, Biotechnical Faculty, University of Ljubljana, Slovenia (Table S1). Around 15 % of the strains (21 of 74) could not be revived after 3 years of preservation at −80 °C using commercial preservation kit.
The holotype of the new species from Svalbard, Psychromyces glacialis, was deposited in the CBS culture collection of the Westerdijk Fungal Biodiversity Institute, Utrecht, as CBS 16467.

PHENOTYPIC CHARACTERIZATION
Morphological characters of pure cultures were observed on potato dextrose agar (PDA) [35], oatmeal agar (OA) [36] and Oxidation reactions on YT plates did not give any positive reaction, and were therefore not considered.

PHYLOGENY
For phylogenetic analyses, genomic DNA of the isolates was extracted using the PrepMan Ultra reagent (Applied Biosystems) according to the manufacturer instructions. DNA of filamentous cultures was extracted after mechanical lysis in CTAB buffer according to the protocol described by Gerrits van den Ende and de Hoog [40]. DNA regions/genes used in the phylogenetic analyses were the small subunit (SSU) rDNA, partial large subunit rDNA including its D1/ D2 domains (LSU), the internal transcribed spacers 1 and 2 including the 5.8S rDNA (ITS), and partial sequences of genes encoding for translation elongation factor one alpha (TEF), cytochrome b (CYTB), RNA polymerase II largest subunit (RPB1) and RNA polymerase II second largest subunit (RPB2). These were amplified and sequenced with the following primer sets: NS1/NS24, NL1/NL4, ITS1/ITS4, EF1-983F/EF1-2218R, E1M4/E2mr3, RPB1-Af/RPB1-Cr, fRPB2-5F/fRPB2-7cR, respectively [27,[41][42][43][44]. Alignments of ITS and LSU sequences of strains described in this study and of the most closely related sequences of type strains and other reference strains, found with the blast algorithm [45] in the non-redundant GenBank nucleotide database, were used for phylogenetic analyses of a concatenated dataset. Partial sequences of the four protein-coding genes, TEF, CYTB, RPB1 and RPB2, were selected from available sequences of the Microbotryomycetes published by Wang et al. [27] and Li et al. [19]. Accordingly, taxa such as Cryolevonia and Meredithblackwellia were excluded from analyses. Ribosomal genes were aligned in mega7 using ClustalX [46], corrections were made by hand. Protein-coding genes were initially aligned in mega7 using ClustalX, and then manually corrected according to the amino acid translations. Introns were excluded from phylogenetic analyses. The best model of nucleotide substitution was estimated with jMod-elTest 2.1.10 [47] and was used as the custom model (model '010230') input to PhyML 3.1 to estimate the phylogenetic trees [48]. aLRT as Chi2-based support was used for calculation of branch supports. The alpha parameter of the gamma distribution of substitution rate categories and the proportion of invariable sites were estimated by PhyML. Additionally, phylogenies were reconstructed from the same alignments using MrBayes 3.2.7 [49]. Two substitution types of the 4by4 model and gamma distributed rates with a proportion of invariable sites (approximated with four categories of gamma distribution) were used for the estimation through 10 million generations (sampling every 100th generation), two runs of 15 chains each, heated at temperature 0.2 (LSU) or 0.1 (ITS, concatenated dataset) and discarding the first 25 % trees from the final consensus tree. The minimum spanning networks were constructed without ties based on a pairwise matrix of bitwise distances derived from the above described alignment of ITS using packages 'ape' and 'poppr' in R [50][51][52]. For the analysis of the fungal community by amplicon sequencing, the total genomic DNA was extracted in triplicate from filtered biomass and 1 g cryoconite sediment as described by Perini et al. [15,16]. The PCR amplification of the ITS2 region was performed using the primers ITS4-Fun and 5.8S-Fun [53] in three reactions per sample. The sequencing of the fungal amplicons was performed by Illumina MiSeq version 3. The sequencing data were processed in QIIME2 [54] as described by Perini et al. [15,16]. Among the set of representative sequences of individual amplicon sequence variants (ASVs) found in the samples, the ones similar to Camptobasidium arcticum, C. gelus and Psychromyces glacialis were identified with stand-alone blast [45]. These sequences were added to the set of sequences recovered from pure cultures and from the databases and aligned in mega7 using ClustalX [46]. The maximum-likelihood tree was estimated using this alignment with PhyML as described above in order to confirm the taxonomic placement of the Camptobasidium and Psychromyces ASVs.
blast searches using ITS and LSU as queries suggested that 32 isolates from our study are identical to Camptobasidium gelus (Pucciniomycotina, Microbotryomycetes, Kriegeriales, Camptobasidiaceae). Twenty-two isolates (group 2) could not be identified to the species level: 21 isolates differed from C. gelus in five positions in LSU and 17 positions in ITS, while EXF-12685, differed from C. gelus in 15 LSU and 45 ITS positions. Sequences of group 3 were almost identical (two and five positions difference in ITS and LSU) to 'Rhodotorula svalbardensis' , described here as Psychromyces glacialis (Table 2).
Phylogenetic analyses of concatenated sequences of the ITS and the LSU placed representatives of the newly sampled strains into well-separated lineages of the Microbotryomycetes. Group 1 isolates were identified as Camptobasidum gelus, recently described based on one strain isolated from Antarctica and two from Greenland [18]. Phylogenetic sister group relatedness of group 2 isolates and C. gelus was highly supported (branch support BS=100 %). The phylogenetic analysis confirmed that group 3 isolates belong to 'Rhodotorula svalbardensis'/ Psychromyces glacialis. They clustered together with other sequences identified as Rhodotorula species. As indicated in Fig. 1, genera Glaciozyma (grouping together with Sampaiozyma and Leucosporidium), Camptobasidium and Cryolevonia form a highly supported monophyletic group (BS=100 %) that is a sister of a joint clade accommodating Psychromyces and Oberwinklerozyma. Genera such as Phenoliferia, Yamadamyces and Meredithblackwellia formed an unresolved, paraphyletic assemblage near the base of the tree, but their relatedness with Oberwinklerozyma, Psychromyces, Camptobasidum, Cryolevonia and Glaciozyma was highly supported (BS=100 %).
Further phylogenetic analyses were based on concatenated sequences of rDNA loci (SSU, ITS, LSU) and partial sequences of the four protein-encoding genes (TEF, CYTB, RPB1 and RPB2 Monophyly of Psychromyces, Camptobasidum, Glaciozyma and K. eriophori is, however, not supported due to the phylogenetic interference of orders, such as Heitmaniales, Leucosporidiales and Heterogastridiales. Accordingly, classification of Psychromyces, Camptobasidum, Glaciozyma in the Kriegeriales is not supported in the present study, which is why the Camptobasidiaceae is for the time being considered as an incertae sedis within the Mycobotryomycetes (Fig. 2). It is to be emphasized, however, that the downloaded amino acid sequences of K. eriophori (CBS 8387) were partly difficult to align with numerous other included taxa (RPB1, RPB2) and that several codons could not be translated into amino acids at all (CYTB). It is clear that these numerous apomorphies are responsible for the long terminal branch that K. eriophori obtained in analyses of the non-translated DNA sequences (Fig. 2).
To analyse the effect of protein-encoding gene sequences on the dataset analysed in Fig. 2 and to allow comparison with already published trees that are based only on nc rDNA data, protein-encoding genes were excluded from the dataset and the remaining alignment was analysed separately (Fig. S1). This analysis also addressed more Mycobotryomycetes as included in Fig. 1 and supported some of the conclusions made on the basis of concatenated dataset presented in Fig. 2. Relatedness of Camptobasidium, Psychromyces and Glaciozyma is almost equally highly supported (BS=92 %), while the monophyly of the so defined Camptobasidiaceae with Kriegeria sensu stricto is not supported. Monophyly of Kriegeria eriophori with Phenoliferia and Yamadamyces (Fig.  S1, BS=89 %) is, however, not in congruence with inferences made on the basis of the dataset including protein-encoding genes (Fig. 2). The inconsistent clustering of K. eriophori is most likely caused by numerous apomorphies that resulted in the above-discussed alignment problems and a long terminal branch for K. eriophori in Fig. 2.  Table 2 had lower similarities and the number of differing nucleotides in ITS was for all drastically larger as in LSU, as already described in other studies [27]. When comparing ITS sequences almost no variability was noted in C. gelus, while some were noted within C. arcticum (Fig. S2). The type strain of C. gelus, CBS 8941 (AY040665) from Antarctica, strain BL58-2 (AB474396) isolated from Russian glacier ice core in Siberia [55] and EXF-12745 from cryoconite in Greenland, had identical ITS sequences. The Antarctic yeast strain CBS 8941, now recognized as C. gelus, was noticed as a potentially new phylogenetic lineage already by Wang et al. [23] on the level of LSU sequences. It was then selected as the type of C. gelus by de Garcia et al. [11], who described this species based on two additional strains from Greenland. The same authors [23] addressed the separate position of CRUB 1733 (GenBank FJ841888), which was recently described as the new species Cryolevonia giraudoae by de Garcia et al. [11]. The genus Cryolevonia, closely related to Camptobasidium (Fig. 2), was described and typified with Cr. schafbergensis by Pontes et al. [17]. Classification of these psychrophilic yeast species was based only on ITS /LSU data [11,17]. The type species of the genus Camptobasidium, C. hydrophilum, is strictly filamentous and forms peculiar anamorphic Ingoldian spores [56], and no yeast phase, which is the only morphology observed in C. gelus. Analyses of ribosomal DNA sequences placed two of our three newly discovered phylogenetic groups in Camptobasidium; however, ribosomal DNA sequences are only restrictively used for phylogenetic analyses of the Microbotryomycetes [23].
No household gene sequences could be generated or are accessible for the type species, C. hydrophilum, which is  nov., C. gelus and representative species of genera of Microbotryomycetes. In the maximum-likelihood estimation, the best model of nucleotide substitution was estimated with jModelTest, other parameters (alpha parameter of the gamma distribution of substitution rate categories, proportion of invariable sites) were estimated within PhyML. aLRT as Chi2-based support was used for calculation of branch supports. In estimation by MrBayes, a total 10 million generations were calculated and the first 25 % were discarded before the shown consensus tree was calculated from trees sampled every 100 generations. Posterior probabilities higher than 0.9 (in green) and maximum-likelihood bootstrap value from 1000 bootstrap replicates larger than 70 % (in blue) are shown near the nodes.
why monophyly of all Camptobasidium species cannot be confirmed yet on the basis of all loci selected here. Camptobasidium cultures are characterized by pink to reddish-grey colonies [11,56], in our case studied on PDA (Figs 3 and  4). According to our observation, strains identified as C. gelus were typically purplish white (14A2), pastel pink (11A3), greyish rose (11B3) or reddish grey (11B2), with either glistening or ridged, membranous surface appearance, raised with entire or undulating margin (Fig. 4a-d).
Sexual reproductive structures were neither observed in single colonies or in mixed cell assays of C. gelus, nor in C. arcticum. No true mycelium and no teliospores were observed. The sexual morph is only known for C. hydrophilum from water-submerged cultures and is characterized by recurved, 1-3 septate metabasidia showing sporogenous loci on their convex sides and bearing rarely single    basidiospores, mostly in groups of up to 20 that are sessile on inconspicuous denticles [56]. Authors also reported typically broadly fusiform or elliptical chlamydospores.
All Camptobasidium strains in our study showed slow colony development suggesting either oligotrophicity or auxotrophy, thus perhaps the need of a particular nutrient, mediating substance or partner relationship. Also, the closely related and psychrophilic Cryolevonia schafbergensis was described as a species with salient physiological characteristics: not growing at 18 °C, growing on a medium containing 16 % NaCl. These characteristics allowed the distinction between C. giraudoae [11] and Camptobasidium species (Table S2).
Differences in assimilation profile among closely related species are listed in Table 3, and within Camptobasidium and Psychromyces in Table S2. Assimilation of raffinose and sucrose was evident in all Camptobasidium species, but also in Psychromyces glacialis. Assimilation abilities of C. gelus isolates were mostly in accordance with de Garcia et al. [11]. Our isolates variably assimilated l-sorbose, d-xylose and nitrite, but were not able to assimilate melibiose and polyols (i-erythritol, xylitol, d-mannitol, d-sorbitol, adonitol) ( Table  S2). The newly proposed C. arcticum differs from C. gelus in the ability to assimilate melezitose, cellobiose, methyl β-d-glucoside and in the inability to assimilate nitrite. The additional differences in assimilation of other, not often used C sources were noticed in C. arcticum: the ability to assimilate gentiobiose, maltotriose and dextrin. It variably assimilated nitrate and showed weak or no assimilation of l-lysine. In some cases, differences in assimilation were observed when comparing results of classical tests in liquid media with the results of Biolog YT. As an example, Camptobasidium gelus showed assimilation of maltose and glycerol in classical tests, but no optical density changes were detected on YT plates (Table S2).

PSYCHROMYCES
The maximum-likelihood analysis of selected ITS and LSU sequences grouped together dimorphic and filamentous isolates of group 3 with isolates named Rhodotorula sp. and Rhodotorula svalbardensis that were all isolated from cold environments of Svalbard and Greenland [12,57]. The strongly supported sister group relationship of this group 3 to Oberwinklerozyma was suggested on the basis of analyses of nc rDNA sequences (BS=100 %) (Fig. 1). The concatenated seven-gene analysis justified and confirmed the isolated position of the genus (BS=100 %), and placed it as sister clade to the clade comprising Camptobasidium and Glaciozyma (Fig. 2). These results suggest that the clade should be placed in the family Camptobasidiaceae. We propose here a new genus, Psychromyces, for this clade for reasons described below. Bayesian phylogenetic inference analysis based on ITS sequences revealed this clade as unresolved (Fig. S2). A number of six nucleotide substitutions out of 526 distinguish two groups of strains, represented by EXF-13111 and EXF-12419 (Fig. S5) Singh et al. [57] erected the name 'Rhodotorula svalbardensis' for strains MLB-I and CBS 12863; however, the name Rhodotorula svalbardensis is invalid because two gatherings, strains MLB-I (CCP-II) and CRY-YB-1, were assigned as type [58], Shenzhen Code, Art. 40.7). Colony characteristics and microscopic characters described by Singh et al. [57] clearly suggested conspecificity of strains MLB-I (CCP-II), CRY-YB-1 and several strains we obtained from Svalbard and Greenland. However, although our strains were isolated as yeast colonies, only a single strain, EXF-13111, retained its ability to grow as a yeast after 2 years of preservation at −80 °C, while the other 16 strains were revived as purely filamentous after deep freezing.
Singh et al. [57] described and illustrated ellipsoidal vegetative cells, 5.8-8.3×4.2-7.5 µm (mean 7×5.9 µm), occurring individually or in groups. According to our observation, these thick-walled cells most likely represent teliospores. In our strains these structures were typically globose to subglobose and had a diameter of 3-4 µm in young cultures; however, depending on the culture condition and age of culture, also spores with a diameter of up to 11 µm and of variable shapes were seen. Sometimes they contained oil droplets. In an attempt to induce their germination, the spores were cut out, placed into sterile H 2 O and kept at 5 °C. After an incubation period of 6 months, they were placed on water agar, however, no germination was observed. Chlamydospores are known in some species of basidiomycetous yeasts, for example, in genera Mrakia, Sporidiobolus, Tilletiaria, Fellomyces and Rhodosporidium, where they can appear in old cultures [59]. They were also observed in Camptobasidium hydrophylum [56]. Singh et al. [57] described, however, not illustrated, unilaterally or occasionally multilaterally budding cells. Yeast colonies were observed only in the strain EXF-13111. The yeast phase was maintained with the subculturing of yeast colonies, however after approx. 1 month of incubation, the colonies started developing hyphae and became strictly filamentous. The observed yeast cells were oblong and measured 7.5±3 (mean±SD; min-max: 5-12) µm×3.5±0.5 (mean±SD; min-max: 3-4) µm. Budding was uni-, bi-or multilateral and occurred solitarily directly on the mother cell or on sympodially proliferating, up to 5 µm long stalks. Daughter yeasts formed on terminal or lateral sympodially proliferating stalks have been so far described for certain Bullera species [59]. Some of the yeast cells in EXF-13111 prolonged to more than 20 µm long stalks, and formed pseudomycelium-like structures. Strains described by Singh et al. [57] formed up to 1.45 µm wide, septate hyphae. The presence of hyphae, with and without clamps, was observed in some strains in our study. Typically, the width of clamped mycelium was 2.5 µm in strain EXF-12419, while hyphae without clamps were narrower (up to 2 µm) and densely septated. In strain EXF-13111, non-clamped mycelium had a width of approx. 1.5 µm. If filamentous colonies were subcultured, they remained filamentous.
The most obvious difference between Psychromyces and other related species was its ability to assimilate Tween 80 (reaction on YT plate), which indicates lipolytic capacity. Camptobasidium gelus isolates varied in assimilation of Tween 80 (negative, positive, week reactions), while C. arcticum was unable to assimilate it. The assimilation ability of Tween 80 is in agreement with Singh's et al. [57] observation of high lipase activity. In contrast, amylase activity reported by the same authors [57] was not detected by the classical assimilation tests performed in this study. The assimilation profile was similar to the closely related genus Camptobasidium: sucrose and raffinose were assimilated, as for all species of Camptobasidium; however, Psychromyces showed also the ability assimilate trehalose (Tables 3 and  S2). It assimilated melesitoze and cellobiose, as C. arcticum. According to Singh et al. [57], Psychromyces glacialis is welladapted to life in glacial environments, in particular to low temperatures. It can synthesize antifreeze proteins, modulate its membrane lipid composition by increasing the unsaturated fatty acids content and produce extracellular enzymes, particularly amylase, cellulase, protease and catalase at 4 °C [57].

ECOLOGY
Ice samples retrieved from the Greenland Ice Sheet harboured a high abundance of yeasts, e.g. up to 900 c.f.u. ml −1 in dark ice, and contained a large number of yeast taxa. The majority of the taxa belonged to the Microbotryomycetes (Basidiomycota), including Camptobasidium gelus, the here proposed C. arcticum, formerly classified as Glaciozyma antarctica-like, Psychromyces glacialis, formerly classified as 'Rhodotorula svalbardensis', Phenoliferia glacialis, Sporobolomyces ruberrimus, and other yet-undescribed basidiomycetous yeasts [15]. A study of glacial environments on Svalbard resulted in the isolation of some of the same taxa, such as C. gelus and Psychromyces glacialis, but also revealed additional species: Glaciozyma watsonii, Leucosporidiella muscorum, Phenoliferia glacialis and P. psychrophenolica, with values up to 5 c.f.u. ml −1 [16]. The phylogenetic and phenotypic analysis of both sets of glacial isolates supported the description of two novel species, herewith proposed as Camptobasidium arcticum and Psychromyces glacialis.
High-throughput amplicon sequencing and analysis of ITS2 sequences from total environmental DNA [15,16] revealed that species of the Microbotryomycetes commonly and abundantly occur in supra-and subglacial environments. The abundance of this group was up to 96 % in supraglacial water, and clear and dark ice (Greenland Ice Sheet) and up to 75 % in subglacial ice (Svalbard). ITS2 sequences identical to those of Camptobasidium gelus EXF-12745, Camptobasidium arcticum EXF-12713, Camptobasidium sp. EXF-12685 and Psychromyces glacialis EXF-13111 were found in total environmental DNA extracts (Fig. S3). Accordingly, these species appeared to commonly occur in diverse glacial environments and different geographical locations. The genus Camptobasidium has a circumpolar distribution and it also occurs in Antarctica [18]. The genus Psychromyces has so far only been recorded in Arctic glacial environments, both in Greenland and Svalbard. None of the species has been found so far in uppermost layers of the glacier consisting of fresh snow.
Slow growth of axenic cultures and unsuccessful preservation of circa 15 % of the isolates might indicate that Camptobasidium species, especially C. arcticum, accommodates a complex ecological niche or that it has complex nutritional requirements. It is possible that they grow optimally only in associations with other glacial organisms. Myco-and plant-parasitism is a frequent trait of the Pucciniomycotina, including Microbotryomycetes [60]. Camptobasidium hydrophilum, a close relative of C. arcticum, was identified as a mycoparasite, since it can form coiling hyphae around hyphae of aquatic hyphomycetes in dual cultures [56]. Parasitic life mode at the host-parasite interface is otherwise known to be supported via special interactive organelles, called colacosomes. These structures were so far detected in mycoparasitic Heterogastridiales, but also, occasionally, in non-parasitic Sporidiobolales and Leucosporidiales within Microbotryomycetes [60], implying their loss in saprobic species. The presence of these structures, indicative of mycoparasitism, have not been studied in Camptobasidium and Psychromyces so far, and should be the focus of future studies.
Statistical analysis of ITS2 NGS sequencing data from Greenland glacial environments revealed high co-occurrence of non-identifiable Microbotryomycetes with Phialophora (Chaetothyriales, Ascomycota) and with unidentified Leucosporidiales (Mycrobotryomycetes, Basidiomycota). While in Svalbard glacial samples high co-occurrence was calculated for non-identifiable Microbotryomycetes with an unidentified species of Leucosporidiales, with Didymellaceae (Pleosporales, Ascomycota), with Libkindia masarykiana (incertae sedis, Microbotryomycetes), and with Kriegeriaceae (Kriegeriales, Microbotryomycetes; data not shown). These co-occurrences might indicate similar ecological preferences of these taxa (e.g. for cold aquatic environments) or interactions between the taxa.
The transition from species only showing a yeast phase, a presumable ancestral morphology to species also having a filamentous phase, co-occurs with ecological niche diversification and adaptations to various lifestyles and environments [60]. Exclusively filamentous species in vitro are found only in Heterogastridium, Pycnopulvinus (both Heterogastridiales), some genera of the Microbotryales, Camptobasidium (type species) and Psychromyces. It is unclear whether C. gelus or C. arcticum, showing only yeast growth in vitro, produce hyphae in their natural environment. However, it is rather likely that their life styles in nature are far more complex than can be predicted by in vitro studies. Occurrence of teliospores and clamped mycelium in P. glacialis, for example, implies that this species might produce basidia and basidiospores in nature, although these structures have not yet been observed on agar plates. Similarly, only a single P. glacialis strain displayed dimorphic characters while all other strains were strictly filamentous.
Although the ecological role of this species is yet unresolved, numerous adaptations to cold environments have been already recognized [53].
This study shows that the glacial environments of Greenland and Svalbard harbour a high abundance of ecologically highly adapted species of the Microbotryomycetes. Additionally, a large yeast diversity remains uncharacterized, which should be investigated in future studies with novel isolation, cultivation and preservation approaches. These studies should also investigate species interactions, which might be particularly important in environments characterized by extremely low temperatures, lack of water and oligotrophic conditions. The presented results also underline the power of combining culture-dependent and independent strategies to assess yeast diversity and to progress towards unravelling the function of these yeasts in glacial ecosystems.
Sexual reproductive structures not observed in single or mixed cell assays. Camptobasidium arcticum (Fig. 3) differed from C. gelus (Fig. 4) in several phenotypic characteristics, such as colony colour and morphology, colony and cell size. Moreover, the newly proposed C. arcticum differs from C. gelus in the ability of assimilation of melezitose, cellobiose, gentiobiose, maltotriose, methyl β-d-glucoside and dextrin, in the inability of assimilation of nitrite, in variable assimilation of nitrate, and in weak or no assimilation of l-lysine. Camptobasidium arcticum was isolated from all sampled environments in Greenland, with no evidence of a clear preference for a specific habitat. Camptobasidium gelus occurred in cryoconite, dark ice, clear ice, supraglacial water in Greenland and in subglacial ice and glacial meltwater in Svalbard. None of the species were recovered from snow. Camptobasidium arcticum was until now found only in Greenland samples of this study. Three identical LSU rDNA sequences (JQ768846, AB558448, AY040647) were deposited at NCBI. The first sequence (JQ768846) was named Basidiomycota sp. TP-Snow-Y73 by Shao and Ma and was isolated from glacier surface snow of the Tibetan plateau (PR China), but no publication has been linked to the strain. The second sequence (AB558448), named Basidiomycota sp. GU54, was produced by Uetake et al. [61], and was isolated from glacial surface ice and snow of the Gulkana glacier (Alaska).