Extremotolerant fungi from alpine rock lichens and their phylogenetic relationships

Fungi other than the lichen mycobiont frequently co-occur within lichen thalli and on the same rock in harsh environments. In these situations dark-pigmented mycelial structures are commonly observed on lichen thalli, where they persist under the same stressful conditions as their hosts. Here we used a comprehensive sampling of lichen-associated fungi from an alpine habitat to assess their phylogenetic relationships with fungi previously known from other niches. The multilocus phylogenetic analyses suggest that most of the 248 isolates belong to the Chaetothyriomycetes and Dothideomycetes, while a minor fraction represents Sordariomycetes and Leotiomycetes. As many lichens also were infected by phenotypically distinct lichenicolous fungi of diverse lineages, it remains difficult to assess whether the culture isolates represent these fungi or are from additional cryptic, extremotolerant fungi within the thalli. Some of these strains represent yet undescribed lineages within Chaethothyriomycetes and Dothideomycetes, whereas other strains belong to genera of fungi, that are known as lichen colonizers, plant and human pathogens, rock-inhabiting fungi, parasites and saprotrophs. The symbiotic structures of the lichen thalli appear to be a shared habitat of phylogenetically diverse stress-tolerant fungi, which potentially benefit from the lichen niche in otherwise hostile habitats. Electronic supplementary material The online version of this article (doi:10.1007/s13225-015-0343-8) contains supplementary material, which is available to authorized users.


Introduction
Bare rock surfaces provide little comfort to life. They are poor sources of nutrients and are constantly exposed to a variety of extremes in abiotic conditions. Variations in surface temperatures and water availability can occur at very short time spans and be the source of diverse stresses (Zakharova et al. 2013;Sterflinger et al. 2012), and with enormous amplitudes. In addition, direct exposure to full sunlight includes a threatening level of energy-rich ultraviolet wavelengths. Not many organisms can cope with such surfaces at this Bedge of life^, thus these surfaces are colonized by specialists with particular adaptations (Selbmann et al. 2005(Selbmann et al. , 2013Onofri et al. 2007;Marzban et al. 2013). In fact, some fungal lineages, which are known as Bblack fungi^or Bmicrocolonial fungi^are among the most stress-resistant eukaryotic organisms on Earth and can occur at considerable diversity on rocks (Ruibal et al. 2005(Ruibal et al. , 2009. The adaptations of these rockinhabiting fungi (RIF) include pleomorphic growth, efficient osmolyte management, melanin production, biofilm formation, and survival in cryptobiotic stage (Gostincar et al. 2010(Gostincar et al. , 2011. Black fungi do not form a monophyletic lineage but are members of Dothideomycetes and Chaetothyriomycetidae (Gueidan et al. 2008;Ruibal et al. 2009) which evolved during periods of dry climate in the late Devonian and middle Triassic, respectively . At approximately the same time scale, the symbiotic association whereby a Electronic supplementary material The online version of this article (doi:10.1007/s13225-015-0343-8) contains supplementary material, which is available to authorized users. fungus shelters microscopic algae or cyanobacteria in exchange for fixed carbon and nitrogen helped to ameliorate nutrient deficiencies on rocks. The lichen symbiosis was this key innovation in the evolution of fungi and lichenized mycobionts have since evolved and diversified (Lutzoni et al. 2001;Hawksworth 2015). Particularly, in alpine altitudes where conditions prevent the development of higher plants, lichen thalli express their phenotypic and phylogenetic diversity and shape the landscapes with colorful mosaics on rock surfaces.
In such variably stressed situations, black fungi and lichens can occur side by side on rock, and black fungi also colonize asymptomatic lichens, especially in arid situations (Harutyunyan et al. 2008). Harutyunyan et al. (2008) showed that black fungi may opportunistically infect lichens, but do not cause damage to their host thalli. Some of the fungi resemble hyphomycetous lichenicolous fungi. However, most lichenicolous fungi have a host specific occurrence and are recognized by their phenotypic symptoms and their sexual or asexual spore-producing structures (Hawksworth 1979;Lawrey and Diederich 2003). It is not known whether black fungi, cryptically colonizing lichen thalli, are directly in contact with the photobiont to obtain nutrients. Some studies, however, suggest that black fungi indeed have some affinity to microscopic algae (Brunauer et al. 2007;Gorbushina et al. 2005). Arnold et al. (2009) also used micro-dissection followed by surface sterilization to show that more culturable fungi were associated with the algal layer compared to the medulla and cortex.
In this study, we conducted a comprehensive sampling of saxicolous lichen species (as reported in Fleischhacker et al. 2015), including samples infected by symptoms-developing lichenicolous fungi from different sites of an alpine range, above the tree-line. We prepared culture isolates of the fungi and produced sequence data for phylogenetic analyses. With these we aimed at answering the following questions: i) are there patterns of co-occurrence among cryptic, black extremotolerant fungi, symptomatic lichenicolous fungi and lichen mycobionts?; ii) are lichen-associated fungal communities structured by mycobiont host? ; iii) what is the phylogenetic placement of the isolated strains?

Material and methods
Sampling Lichen thalli were collected on the Koralpe mountain range in the southeastern rim of the Austrian Alps , between the states Styria and Carinthia. The sampling was carried out as in Fleischhacker et al. (2015). Ten collection sites (plots), each further divided into 3 subplots, were selected in alpine habitat, above the timberline, ranging between 1800 and 2100 m a.s.l., and are characterized by big boulders and cliffs of homogeneous size of siliceous-schist/ gneissic rocks separated by wide areas of pastures or dwarf shrub formations. Here winds, in particular from South and West, reach speeds over 120 km/h and the annual temperature averages 0-5°C (http://www.umwelt.steiermark.at/cms/beitrag/10023583/ 25206/). In winter, rocks can remain covered by wind-pressed snow and ice for several weeks; alternatively, in summer the south-exposed rock surfaces receive intense solar radiation.
In these sites, boulders' surfaces are almost entirely colonized by crust-forming (90%), foliose and fruticose (10%) lichens. Crust-forming and foliose lichens were selected for the culture isolation experiment: (i) crustose thalli are composed by contiguous islands of thallus (areoles) which tightly adhere to the substrate with their entire lower surfaces; (ii) foliose thalli adhere only partly to the substrate by a central holdfast (umbilicus) or by root-like appendices (rhizines). About 10% of the lichen thalli in this region are infected by lichenicolous fungi (Fleischhacker et al. 2015). For the isolation of lichenicolous and extremotolerant fungi we selected multiple lichen thalli of different species visibly infected by different species of symptomatic lichenicolous fungi (Tables 1, 2, S3). In doing so, we aimed at including a comprehensive survey of the different lichenicolous fungus-lichen host associations occurring in the area. Within the same subplot, we selected up to four different symptomatically infected thalli. These were either lying close to each other or lying apart up to 50 cm. The lichen thalli were sampled together with their substratum by chiseling the piece of rock. We sampled on both horizontal and vertical positions and at different expositions.
Culture isolation A total of 130 lichen samples, comprising 25 different lichenicolous fungus-lichen host associations, were selected for culture isolations. Thallus areoles or lobes presenting lichenicolous fungal infections were removed with a sterile razor blade and put into an Eppendorf tube. The isolation protocol followed Yamamoto et al. (2002). The pieces, about 2 mm 2 , were washed three times for 15 minutes in distilled sterile water on a shaking bath, followed by a 30 minutes washing step with 500 μl of 1:10 dilution of Tween 80 to remove the possible external contaminations of bacteria and yeast (Bubrick and Galun 1986). A final washing step was carried out twice in distilled sterile water for 15 minutes. The clean fragments were dissected under the stereomicroscope using a sterile razor blade and single pieces were picked with a sterile needle, moistured with distilled sterile water, and transferred into agar tubes. In order to promote the growth of different fungi we inoculated the dissected fragments on six different media: Trebouxia Medium (TM, Ahmadjian 1967), Malt Yeast Extract Medium (MY, Ahmadjian 1967), Lilly and Barnett´s Medium (LBM, Lilly and Barnett 1951), Potato Glucose Agar (PGA; Sigma), Dichloran-Glycerol 18%-Agar (DG18; Sigma), Sabouraud-Agar (SAB; Sigma). Four tubes of the same medium were inoculated, resulting in a total of 24 tubes (inocula) for each original sample. The tubes were  incubated in a growing chamber at 20°C, with a light-dark regime of 14:10 hours with light intensity of 60-100 μmol photons m -2 s -1 and 60% humidity. After three to five months, the inocula reached about 1-3 mm in diameter and it was possible to subculture and to prepare them for DNA extraction, sequencing and morphological analyses. The subcultures were set on agar plates using the same growth medium where the inoculum grew successfully. The cultured strains are deposited at the University of Graz in the culture collection of the first author LM and are preserved as cryostocks.
Morphological analyses Morphological and anatomical characters of the cultured strains were analysed using standard microscopic techniques and documented with photographs. Analyses and photographs were performed on 10 month to one year old subcultures considering the following characters: form of growth, branching of the hyphae and melanization. Small fragments of the mycelia were taken; squashed sections were mounted in water and studied by light microscopy. Images were acquired with a ZeissAxioCam MRc5 digital camera fitted to the microscope. Both images of growth habit and hyphae structure were digitally processed using the CombineZM software (www.hadleyweb.pwp.blueyonder.co. uk/). The photos were slightly refined in sharpness and color tone with Adobe Photoshop 7.0 and the figures were prepared with CorelDRAW X4.
DNA extraction, amplification and sequencing Small parts of the subcultured fungi were taken, transferred into 1.5 ml reaction tubes containing sterile beads for homogenization, frozen and ground using a TissueLyserII (Retsch). The DNA was then extracted following either the C-TAB protocol of Cubero et al. (1999) or using the DNeasy Plant Mini Kit (Qiagen, Austria). The industrial kit was used for those most melanized isolates for which the C-TAB protocol failed in extracting amplifiable DNA. The identity of the cultured fungal strains was studied with sequences of the nuclear large and partial nuclear small ribosomal subunits (nucLSU and nucSSU) and the mitochondrial small ribosomal subunit (mtSSU). The nucLSU fragment was obtained in two pieces using primers SR6R (http://www. botany.duke.edu/fungi/mycolab) and LR5 for the upstream fragment, and LR3R and LR7 (Vilgalys and Hester 1990; http://www.biology.duke.edu/fungi/mycolab/primers.htm) for the downstream fragment. The nucSSU locus was amplified using the primers NS1 (White et al. 1990) and nuSSU0852 (Gargas and Taylor 1992). The mtSSU locus was amplified with primers mtSSU1KL (Lohtander et al. 2002) and MSU7 (Zhou and Stanosz 2001) or mtSSU1 and mtSSU3R (Zoller et al. 1999). PCRs amplifications were carried out with the Illustra™ puReTaq Ready-To-Go PCR Beads (GE Healthcare, UK Limited) with a reaction volume of 25μl and a primer concentration of 0,6 pmol/μl. The Pleosporales amplification of the genes followed touch-down PCR conditions as in previous studies (Muggia et al. 2011(Muggia et al. , 2013. PCR products were cleaned with E.Z.N.A.® Cycle Pure Kit (Omega Biotek, VWR) according to the manufacturer 's instructions. Both complementary strands were sequenced using the same PCR amplification primers by Microsynth (Sanger 3730xl from ABI, Vienna, Austria). Forward and reverse sequences were assembled into contigs and edited manually in BioEdit (Hall 1999).

Alignment and phylogenetic analyses
We checked the identity of the newly generated sequences with sequences available in the GenBank database by blast similarity search (Altschul et al. 1990). Taxa which closest matched our sequences for a value not lower than 95% identity and the further closest related ones (up to 90% identity) were selected for the phylogenetic analyses. As our sequences showed closest matches with representatives of the classes Eurotiomycetes (particularly in the subclasses Chaetothyriomycetidae), Dothideomycetes, Leotiomycetes and Sordariomycetes, we prepared four different datasets representing each lineage (the multilocus sequences alignments are deposit at TreeBASE). We tried to include in each dataset the widest spectrum of taxon diversity by selecting, if possible, at least three taxa representatives of different families or orders of the four classes (Table S1, S2, S3). We based our selection also on previous phylogenetic analyses which considered the aforementioned classes (e.g. Zhang et al. 2006;Wang et al. 2006;Gueidan et al. 2008Gueidan et al. , 2011Ruibal et al. 2009;Schoch et al. 2009;Huhndorf and Miller 2011;Untereiner et al. 2011;Muggia et al. 2013;Hyde et al. 2013;Maharachchikumbura et al. 2015;Suija et al. 2015, ). The datasets of Eurotiomycetes and Dothideomycetes were prepared in summer 2014 whereas those of Leotiomycetes and Sordariomycetes in January 2015. For this reason recent sequence data published subsequently summer 2014 by Gueidan et al. (2014) and Ertz and Diederich (2015) are not included here. For each dataset, outgroup taxa were chosen from the most closely related classes. Sequence alignments for each locus (nucLSU, nucSSU and mtSSU) and for each fungal class (Eurotiomycetes, Dothideomycetes, Leotiomycetes and Sordariomycetes,) were prepared manually in BioEdit (Hall 1999). Introns and ambiguous SNPs were removed from the alignment. For a number of specimens we were unable to generate sequences for all of the selected loci and for other taxa sequences were not available in GenBank. Therefore we present here a three-locus phylogenetic inference for the classes Eurotiomycetes and Dothideomycetes, and two-locus inferences for the classes Leotiomycetes and Sordariomycetes. The final phylogenetic analyses of the Eurothiomycetes dataset included a subset of the isolates, which were selected after having estimated a first phylogeny including all the isolates. As multiple isolates shared identical sequences, we selected for the final analyses as representatives those isolates obtained from different samples of the 25 lichenicolous fungus-lichen host associations which were grown on different media. Combined data of different loci, either fully or partially congruent, have been commonly considered in phylogenetics (Dettman et al. 2003). We performed, therefore, as in previous studies (Kauff and Lutzoni 2002;Miadlikowska et al. 2006;Muggia Perez-Ortega et al. 2014), both single locus and combined datasets analyses. We analysed the single locus datasets with a Maximum Likelihood (ML) approach (Meson-Gamer and Kellogg 1996;Reeb et al. 2004) and the combined dataset using both maximum likelihood (ML) and Bayesian approaches. In both approaches the combined datasets were treated in partitions by genes nucLSU, nucSSU and mtSSU. The ML analyses were performed using the program RAxML v. 7.1.3 (Stamatakis et al. 2005). The GTRMIX model was applied both for the single loci and to each partition in the combined datasets (as only a single model of molecular evolution can be used across gene partitions in RAxML), and 1000 bootstrap replicates were run. The Bayesian Markov Chain Monte Carlo (B/MCMC) analyses were run in MrBayes 3.1.2 (Huelsenbeck and Ronquist 2003;Ronquist et al. 2005). The model of molecular evolution applied in the Bayesian analysis to each gene partition, the GTR+I+G model, was estimated in JModeltest v. 2.1.4 (Darriba et al. 2012) using the Akaike Information Criterion (Posada and Crandall 1998). The B/MCMC analyses were run with six chains simultaneously, each initiated with a random tree. Ten million generations for the Eurotiomycetes and Dothideomycetes datasets and five million generations for Leotiomycetes and Sordariomycetes datasets were run, respectively. Trees were sampled every 100 generations. The log-likelihood scores were plotted against generation time using Tracer 1.4 (Rambaut and Drummond 2007) to determine when the stationarity of likelihood values had been reached (e.g., the burn-in stage; Ronquist et al. 2005). Burn-in was set at half of the generations (the first 50,000 and 25,000 sampled trees for the two datasets groups respectively) and the majority rule consensus trees were calculated from the posterior samples of 50,001 and 25,001 trees, respectively. The convergence of the chains was confirmed by the convergent diagnostic of the Potential Scale Reduction Factor (PSRF), which approached 1 (Ronquist et al. 2005). The phylogenetic trees were visualized in TreeView (Page 1996).

Results
Culture isolation A total of 248 fungal cultures from 77 host lichen thalli were isolated and identified to date: 191 belong to the subclass Chaetothyriomycetidae, 36 to the class Dothideomycetes, 12 to Leotiomycetes and 9 to Sordariomycetes. We obtained 21 additional isolates that corresponded to the lichen mycobionts (not shown). The majority of the strains, 24%, grew on TM, 22% grew on LBM, 20% on SAB, 16% on KGA, 13% on MY and 5% were isolated on DG media. Cultured mycobionts represented 2% of the grown isolates. From these cultures we obtained in total 710 new sequences: 244 for nucLSU rRNA gene, 237 for nucSSU rRNA gene and 229 for mtSSU rRNA gene (Table 1, Table 2, Table S3). The diversity of fungi isolated from lichen thalli, and which did not represent the mycobiont of the lichen symbiosis, varied among the 77 original thalli. The specificity of the isolated fungi neither correlates with the presence of any observed lichenicolous fungus nor with the identity of the lichen mycobiont. Fungi belonging to the same lineage were isolated from multiple thalli representing the same association of lichen and lichenicolous fungus, but also from the same lichen host species infected by different lichenicolous fungi (from hosts not growing in vicinity) and from other different associations of lichen and lichenicolous fungus (Tables 1, 2, S3). For example, we isolated up to five different lineages of fungi in two lichen individuals; fungi of four different lineages were isolated from only a single thallus, fungi of three different lineages were isolated from nine thalli. Fungi representing two different lineages were retrieved from 21 thalli, and fungi representing one lineage were obtained from 40 thalli. P h y l o g e n e t i c a n d m o r p h o l o g i c a l a n a l y s i s o f Chaetothyriomycetidae (Fig. 1, Fig. 2, Table 1 and  Table S1) The phylogenetic relationships recovered in Chaetothyriomycetidae are highly congruent with previous studies of Gueidan et al. (Gueidan et al. 2008 and Diederich et al. (2013). There were no significant incongruences between single locus (not shown) and multilocus trees. The only exception is the clade of Sclerococcum sphaerale, which is placed in our multilocus reconstruction at the base of Chaetothyriomycetidae, possibly due to the availability of only the nucLSU marker (Fig. 1). In this Sclerococcum sphaerale clade we recovered the single isolate A1016. A1016 was isolated from a thallus of Pertusaria corallina infected by Sclerococcum sphaerale, and this placement seems to confirm the identity of the lichenicolous fungus. This isolated strain forms pale pinkish, compact mycelia with thin, hyaline hyphae (Fig. 2 F1-F5). Clade I is represented by six isolates (from three host species), which together with Celothelium cinchonarum are basal to the split between Verrucariales and Chaetothyriales. These isolates are similar in morphology, forming white mycelia composed by thin, hyaline hyphae, which occasionally gather in thick, plectenchymatous strands (Fig. 2 A1-A4, B1-B4). Clade II is represented by three isolates: they come from three different thalli of the same lichen host-lichenicolous fungus association (Rhizocarpon geographicum -Muellerella pygmaea-Rh). These strains also present a pale pinkish mycelium, but hyphae are formed by cylindrical to semi-elliptical cells which are occasionally intercalated by roundish cells (Fig. 2 C1-C7). Two samples, A579 and A1026, are nested within Epibryaceae, the lineage formed by Epibryon and two rock-inhabiting fungi. The mycelium of these isolates is a dense aggregate of roundish, melanised cells containing inclusions, and filamentous hyphae are rarely present (Fig. 2 E1 -E5). Clade III represents the lichenicolous fungus Lichenodiplis lecanorae (Muggia et al. in prep.), which appears here basal to the split between the families Epibryaceae, Chaetothyriaceae and Herpotrichiellaceae. The identity of these isolates is also confirmed by the conidiomata-like structures and the conidia that are observed in the cultures (Fig. 2  D1-D4). Herpotrichiellaceae is here the most represented family of Chaetothyriales and comprises ecologically diverse fungi including human pathogens (Exophiala dermatitidis and Capronia semiimmersa), lichenicolous fungi (Capronia peltigerae and Cladophialophora parmeliae) and rock inhabiting fungi (Gueidan et al. 2008;Gueidan et al. 2011, Gueidan et al. 2014. Four newly cultured isolates are nested in this main Chaetothyriales lineage. A561 is nested in a clade with RIF and Phialophora europaea, and is morphologically similar to other previously isolated black RIFs (Fig. 2 L1-L3), having melanized hyphae frequently budding laterally and apically. Three other samples are nested in a clade with Cladophialophora parmeliae and Capronia semiimmersa. The isolates are characterized by melanized mycelia, with branching hyphae composed by elliptical, subcylindrical and subglobose cells constricted at the septa (Fig. 2 K1-K5).
In Chaetothyriales, the majority of the isolates group into subclades of a fully supported lineage sister to Chaetothyriales. Within this lineage we distinguished the main clades IV, V, VI and VII (as subclade of clade VI, Fig. 1), each represented by more than four isolates. The other isolates are placed on separate smaller clades in this large assemblage of branches. Clade IV and clade V include isolates from six and seven, respectively, different lichen hostlichenicolous fungus associations. Clade VI contains the majority of the isolates which come from 16 different lichen host Fig. 1 Multilocus phylogenetic inference of Eurotiomycetes. The ML and the Bayesian phylogenetic hypotheses were inferred from the combined dataset of nucLSU, nucSSU and mtSSU loci and corresponded in their topologies; the ML analysis is shown. Bayesian posterior probabilities (PP ≥ 95 %) and ML bootstrap support values (≥ 70 %) are reported above branches (PP/bootstrap value). Newly identified clades of isolated fungi obtained from this study are highlighted in bold and are labelled as clade I to VII. Symbols indicate the different lichen host-lichenicolous fungal associations which represent the original thallus from where the fungal strains were isolated. A symbol reported multiple times for a clade indicates the number of different original thalli sharing the same lichen host-lichenicolous fungal association. Fungal life-styles are reported in parenthesis. Samples labelled with an asterisk (*) are those photographed in Fig. 2  Fig. 1 (continued) -lichenicolous fungus associations. These include lichens infected by known but unrelated lichenicolous fungi. With the exception of few isolates, such as A514 (Fig. 2 H1-H5) and A511, which lack melanized mycelium, all the fungal strains included in this big assemblage of lineages are characterized by melanized mycelia. However, two main morphologies are observed among the strains: i) mycelia with filamentous, branching hyphae composed by cylindrical cells, usually aseptate (rarely 1-septate), intercalating by roundish cells, ii) mycelia with hyphae composed exclusively by globose, roundish cells, sometimes 1-septate, forming dense assemblages and budding.
Except for the Sclerococcum clade and the clade III, we do not find clear evidence of correspondence of certain lineages with other lichenicolous fungi infecting the lichen samples. P h y l o g e n e t i c a n d m o r p h o l o g i c a l a n a l y s i s o f Dothideomycetes - (Fig. 3, Fig. 4, Table 2 and Table S2 Hyde et al. (2013). Topological congruence was recovered between the Bayesian and the maximum likelihood analyses and among the single locus analyses. Also in Dothideomycetes the isolates are nested in clades together with fungi of diverse ecological niches and presenting different lifestyles (Fig. 3).
The isolate A930 is recovered within Lichenostigmatales, which includes lichenicolous fungi and RIFs. A930 is morphologically identical to the Lichenostigma cultures isolated by Ertz et al. (2014), presenting yeast-like, budding, melanized cells. Four isolates form a fully supported clade nested in Pleosporales. Also in Pleosporales, four further isolates group together with lichenicolous species of the genus Phoma; however they were isolated from thalli of four different lichen host-lichenicolous fungi associations and none of them showed the symptomatic presence of Phoma species. These isolates form whitish to pale pinkish mycelia, composed by hyaline hyphae distributed to form a dense aggregate (Fig. 4 A1-A5, B1-B3). Seven isolates represent a lineage sister to Lichenotheliales; these isolates originate also from four thalli representing different lichen hosts infected by different lichenicolous fungi. The isolates comprise both melanized and non-melanized fungi (Fig. 4 D1-D6 and E1-E3). Three isolates are recovered in Myriangiales, a lineage of saprobic fungi; they present white mycelium of very thin hyaline hyphae (Fig. 4 F1-F4). The single isolate A559 is recovered as a member of Teratosphaeriaceae I. The remaining isolates group as a single lineage in Capnodiales, being nested among the clades Teratosphaeriaceae I, Teratosphaeriaceae II and Mycosphaerellaceae. In this lineage we identify three subclades, even though all isolates have a similar morphology, with dark, melanized mycelia composed by suglobose to cylindrical cells with rough cell wall and sometimes constricted at the septa (Fig. 4 G1-G6, H1-H6).
Phylogenetic and morphological analysis of Leotiomycetes and Sordariomycetes (Fig. S1, Fig. S2, Table S3 and  Table S4) Only 15 and nine isolates have been identified as Leotiomycetes (Helotiales) and Sordariomycetes, respectively. Within Leotiomycetes none of our isolates is closely related to the lineage Encoelioideae, where recently lichenicolous fungi were identified to belong (Suija et al. 2015). Five isolates are placed with unresolved position at the base of Leotiomycetes; one isolate is closely related to Leotia lubrica (saprotroph among mosses and plant rests, Kuo 2003) and Microglossum olivaceum (a grassland species, Fleming 2001). Eight isolates obtained from three different combinations of lichen host and lichenicolous fungus are closely related to Mitrula paludosa (a species known from swamps and bogs, Wang et al. 2005).
Three isolates are identified in Xylariales within the Sordariomycetes, one isolate is nested in Hypocreales (including insect parasitic species, mycoparasites, endophytes and saprotroph, Gazis et al. 2014), and five isolates, deriving from three different lichen host-lichenicolous fungi associations are recovered in Coniochaetales (saprotrophs, leaf and root endophytes, plant pathogens, Zhang et al. 2006). Strains of both Leotiomycetes and Sordariomycetes form pale pinkish to white mycelia (Fig. S3); melanization was seldom observed and was restricted only to localized parts of the culture (Fig S3 G).

Discussion
Rock-inhabiting alpine lichens are exposed to harsh environmental conditions, with drastic and sometimes sudden changes in temperature and hydration, as well as UV radiation. Conceivably, only fungi that tolerate such fluctuating conditions can persist or grow in lichens. In addition, these fungi must cope with the diverse and highly concentrated extracellular secondary products of their host species. We already found a surprising number of lichenicolous fungi in lichens (Fleischhacker et al. 2015), and evidence for a high number of additional, cryptically occurring fungi. Here we provided a comprehensive set of isolates of the culturable fungal fraction in lichens from an alpine habitat for a survey of their phylogenetic relationships, with special emphasis on members of Dothideomycetes and Chaetothyriomycetes.
Molecular data and the morphological analyses seem to confirm the identity of only two symptomatic lichenicolous fungal species with Eurotiomycetes. The isolates obtained from thalli infected by Sclerococcum sphaerale indeed group within the lineage Sclerococcum (Diederich et al. 2013). The formation of conidiomata and conidocells was observed in multiple cultured fungi from different thalli with infections of Lichenodiplis lecanorae. This proved the identity of the culture with the original infection of the lichenicolous hyphomycete.
Except for the above mentioned clades, we do not find clear evidence of correspondence of certain lineages with other lichenicolous fungal species infecting the lichen samples. Some of the observed lichenicolous fungi cannot be the origin of the sequenced cultures, since these belong to completely unrelated groups (e.g. Arthonia, Carbonea, Cecidonia, Opegrapha, Skyttea, Stigmidium; Ertz et al. 2009Ertz et al. , 2014Schmull et al. 2011;Suija et al. 2015). It is likely that the clade IV and V, and the plenty of clades with few representatives, so far correspond to lineages of still unknown fungi which may occur widespread in lichen thalli from rocks, but are unapparent to the eye.
The present phylogenetic results also show that some of the detected fungi are closely related to lichenicolous fungi as well as to fungi known from diverse other ecological niches. Two isolates are closely related to the genus Epibryon, which was originally described as bryophilous (Döbbeler 1978). It is now emended by non-lichenized lichenicolous species (Zhurbenko and Hafellner 1999;Sérusiaux et al. 1999), which demonstrates cross-kingdom host switches in this monophyletic genus. The host lichens of the Epibryon strain were also visibly infected by the genera Carbonea and Muellerella, respectively. Also Muellerella comprises species on bryophytes and lichens (Döbbeler and Triebel 1985), but its relationship with Epibryon requires further study. Clearly Carbonea, as a member of Lecanoraceae, is unrelated. The results suggest that Epibryon could occur also as a non-symptomatic lichen inhabitant, which agrees with the previous results of U 'Ren et al. (2010), who discovered a group of fungi capable to live cryptically in both lichens and mosses. The cryptic presence of otherwise symptomatic lichenicolous fungi is also demonstrated by isolates placed with the lichenicolous lineage of the anamorph genus Phoma, and those strains which are nested w i t h l i c h e n i c o l o u s s p e c i e s o f C a p r o n i a a n d Cladophialophora. None of these isolates, however, originated from thalli which were visibly infected by either Phoma, Capronia or Cladophialopora. The high similarity (>95%) that the new sequences showed with the already available Phoma sequences suggests that the isolated strain could represent closely related Phoma species.
The majority of the isolates are melanized fungi, which closely resemble previously studied rock-inhabiting fungi (RIF) and in fact are closely related to them. The presence or absence of these fungi in hosts of the same area seems to be largely unpredictable, unspecific and facultative. Rather than indicating host specificity, they seem to be broadly tolerant species whose presence might depend more on physical parameters. Nonetheless, lineage clade VI (Chaetothyriales) seems to be rather ubiquitous in lichens. All selected fungi likely represent the same species occurring in many thalli and in combination with different lichenicolous fungi.
We also isolated fungi which constitute two monophyletic lineages, both closely related to orders and families of lichenicolous and lichenized genera, RIF and pathogens in Dothideomycetes: the first closely related to Lichenotheliales, the second nested in Capnodiales. The first lineage is closely related to species of the genus Lichenothelia, which are known to share multiple lifestyles on rocks (Hyde et al. 2013;Muggia et al. 2013Muggia et al. , 2015. They dwell on bare rock surfaces, but are often found associated with free living algae also present on the rocks. Some species specialize as lichen parasites and seem to associate with the lichen photobiont . Some oligotrophic fungi apparently improve their carbon supply by attaching to microscopic algae. A direct involvement of black fungi in fungalalgal interactions was earlier described as a balanced algal parasitism (Turian 1977). Several rock-inhabiting and lichen-inhabiting microcolonial fungi develop into lichenoid structures within months when co-cultured with algae obtained from lichen thalli (Gorbushina et al. 2005;Brunauer et al. 2007). Gorbushina and Broughton (2009) showed an example with a co-culture of Nostoc and a rock-inhabiting fungus (Sarcinomyces). They observed a specific spatial arrangement of both organisms and growth alterations in the photosynthetic cyanobacteria suggested a specific interaction. Therefore black fungi that loosely associate with algae in nature might be interpreted as Blichenoids^and are considered prime forms of symbiosis (Muggia et al. 2013).
In our survey, fungi of unrelated lineages were recovered several times from individual lichen thalli. This may indicate that there is no competition between the different fungi, which complies with a concept of niche-sharing ), and that the occurrence of certain lineages does not implicate the presence or absence of others. Lichen-associated fungi, which do not develop any diagnostic structure on the thallus host, use the host just for their own cryptic internal life, likely awaiting the most suitable substrate/host to propagate. Perhaps not all isolated fungi grow equally well in lichens, and we cannot exclude that some might be present as spores or small germlings, while others form mycelia networks in their hosts. We often see mycelia of melanized fungi on the lichens and expect their growth is well adapted to the poikilohydric lichen habitat. The symbiotic structures of the lichen thalli function as a shared habitat of phylogenetically diverse stress-tolerant fungi, some of which use their host as protection, while others use it as nutrition sources in otherwise hostile environments.