DNA isolation and sequencing

Different laboratories contributed data using various protocols, but most DNA sequence information was produced as follows: genomic DNA was isolated from cultures grown on MEA. Fungal biomass was transferred to a tube with 500 μL of TES buffer and ground with a micro-pestle for 1–2 min, with or without silica-mix (2/3 silica-gel, 1/3 Celite® 545). A volume of 140 μL of 5 M NaCl was then added, followed by 65 μL of 10 % (w/v) CTAB (cetyltrimethylammoniumbromid). After an incubation of 30 min at 65 °C, 700 μL of (24:1) chloroform/isoamylalcohol was added, the tubes were mixed carefully by hand, stored on icy water for 30 min, and centrifuged for 10 min at 4 °C (10 000 x g). The supernatant was recovered and the genomic DNA precipitated using isopropanol. After washing the pellets with 70 % ethanol, they were dried in a vacuum centrifuge and re-suspended in 60 μL of TE buffer (protocol modified from Möller et al. 1992).

Six regions covering five genes were amplified: nucLSU, nucSSU, mtSSU, RPB1 region A–D, RPB2 region 5–7, and RPB2 region 7–11 (see table 2 for primers used). Genomic DNA (1 μL of a 1/10 or 1/100 dilution) was added to a PCR mix comprising 2.5 μL of PCR buffer (buffer IV with 15 mM MgCl₂, Abgene, Epsom, U.K.), 2.5 μL of dNTPs (2 mM), 2.5 μL of BSA (10 mg/mL), 2.0 μL of primers (10 μM), 0.15 μL Taq polymerase (5 U/μL, Denville, Metuchen NJ, U.S.A.), and water for a total volume of 25 μL. Amplification cycles for nucLSU, nucSSU and RPB1 (same conditions applied for RPB2) are described in Gueidan et al. (2007), and in Zoller et al. (1999) for mtSSU. The PCR products were purified using Microcon PCR cleaning kits (Millipore, Billerica MA, U.S.A.). Sequencing was carried out using Big Dye Terminator Cycle sequencing Kits (ABI PRISM version 3.1, Perkin-Elmer, Applied Biosystems) on ABI 3730xl DNA Analyzers (Applied Biosystems, Foster City CA, U.S.A.) from the Duke Center for Evolutionary Genomics (Durham NC, U.S.A.) and the Hubrecht Institute (Utrecht, Netherlands).

Classification of rock fungi related to Dothideomycetes

Although very diverse within Dothideomycetes, RIF have not been included in recent phylogenetic studies of this class (Lumbsch et al. 2001, Schoch et al. 2006). Only very few of these rock-inhabiting species have been taxonomically described (Sterflinger et al. 1997, Bills et al. 2005, Sert et al. 2007b), and the molecular marker available for most of these species (ITS) does not allow their inclusion in large-scale phylogenetic analyses. The few attempts to produce phylogenies involving RIF have shown that they belong to two diverse classes of Ascomycota, namely Eurotiomycetes (particularly the order Chaetothyriales) and Dothideomycetes (preponderantly the orders Capnodiales, Dothideales and Pleosporales) (Sterflinger et al. 1999, Ruibal 2004, Ruibal et al. 2005, 2008).

Our results confirm the placement of RIF in the same orders of Dothideomycetes, although some lineages are shown to belong to additional groups. Based on our results, many RIF should be classified within Dothideales, Pleosporales and Capnodiales, the latter order holding the largest number in rock-colonising species. The genera Elasticomyces and Recurvomyces, as well as the Antarctic genus Friedmanniomyces, were previously attributed to Capnodiales based on nucSSU data (Selbmann et al. 2008). Our multigene analyses confirm this placement, and show that these three genera belong to Teratosphaeriaceae s. str., the family currently showing the highest diversity in RIF (Fig. 3). We also showed that one RIF (TRN 235) previously thought to be related to the Dothideales (Ruibal et al. 2008) actually belongs to Myriangiales, along with Sarcinomyces crustaceus, a species similarly melanised and meristematic, but isolated from plant material (Sigler et al. 1981).

Several well-supported groups of RIF could not be attributed to any known families and orders according to our data. As a consequence, Cryomyces should still be considered as Dothideomycetes incertae sedis, as no close relationship was recovered for this enigmatic Antarctic genus (Selbmann et al. 2005). The positions of RIF-rich genera Coniosporium and Sarcinomyces are also problematic. Previous studies placed them either in Dothideales or Chaetothyriales based on ITS or nucSSU data (de Leo et al. 1999, Sterflinger et al. 1999, Sert et al. 2007a). Yet, the limited taxon and gene sampling on which these analyses were based was probably insufficient to demonstrate clear phylogenetic relationships. Our results show that Coniosporium apollinis (including the type strain CBS 352.97), C. uncinatum (including the type strain CBS 100219) and Sarcinomyces crustaceus belong to Dothideomyceta (Fig. 4). However, a previous multigene analysis showed that two other species, Coniosporium perforans and Sarcinomyces petricola, belong to Chaetothyriales (Gueidan et al. 2008). These anamorphic genera are therefore not monophyletic, and additional research is required to clarify their status.

Among lineages lacking known reference taxa, two groups seem to belong to Dothideomycetes (unknown group 2, a lineage comprising RIF from the Alps, and unknown group 3, a lineage including strains isolated in Arizona; Fig. 2). Another unknown group (lineage 1) clusters outside Dothideomycetes, sister to the Arthoniales (Figs ​(Figs2, 2, ​,4).4). A previous study had noted the problematic placement of this latter group (Ruibal et al. 2008). Many lineages including RIF still need to be named. In the past, several melanised meristematic species and genera have been described such as Lichenothelia (Hawksworth 1981; see also Henssen 1987), which could potentially correspond to some of these RIF lineages. However, little is known about these formerly named taxa, and no molecular data or cultures are available for many of them. Naming RIF will therefore require an extensive study of both rock-inhabiting species and formerly described melanised meristematic species, whether they grow on rock or not.

Rock surfaces: “terroirs” for ancient lineages or reservoirs for plant-associated fungi?

Despite the prevailing extreme conditions, rock surfaces host a large variety of specialised fungi. Fungal colonisation of subaerial rocks can be explained by two non-exclusive hypotheses. Firstly, atmosphere-exposed rock substrates could constitute “terroirs” for ancient fungal lineages. Rock surfaces were among the first terrestrial substrates available for living organisms on earth (Gorbushina & Broughton 2009). It is therefore likely that, early on, some species became adapted to colonise rock surfaces. RIF are persistent to different types of physical stress, but are poor competitors and surrender to more combative organisms (Gorbushina et al. 2008). Increasing competition with other rock-inhabiting organisms living under more permissive conditions may have restricted some of these ancient, morphologically reduced, slow-growing, fungal relicts to extreme habitats. The presence of lineages comprising exclusively RIF that diverged early in the evolution of Dothideomyceta (e.g., Cryomyces and lineage 1, Fig. 2) supports this hypothesis of rock surfaces as substrates for ancient fungal lineages.

Secondly, rock surfaces could form reservoirs for plant-associated or saprobic fungi. Through spore or propagule dispersal, some species of various unrelated groups of plant pathogens, epiphytes or saprobes can reach rock substrates. Their ability to survive in these environments will depend on some key features, namely oligotrophy, melanisation and pleiomorphism (or diversity of growth forms, amongst which meristematic growth). Under extreme conditions prevailing on rock surfaces, fungi possessing these key features can survive due to their slow, meristematic, clumpy growth and thick-walled, heavily melanised cells. These key features seem to have evolved several times in Dothideomycetes, allowing different lineages to colonise rock substrates. In Dothideales, phyllosphere fungi such as Aureobasidium pullulans and relatives, which have a filamentous or yeast-like growth under moist conditions, but convert to a meristematic form when colonising inert substrates, have also been isolated from rock surfaces (Ruibal et al. 2008). The family Teratosphaeriaceae s. l. is another example of a group in which some leaf-colonising species can also grow meristematically and form dark, thick-walled cells. According to our results, this family (as traditionally delimited; i.e., including Teratosphaeriaceae 1 and 2) is also extremely diverse in RIF (Fig. 3). Rocks supporting growth of subaerial biofilms (Gorbushina & Broughton, 2009) may be viewed as a reservoir for all types of melanised meristematic fungi, from where other habitats can be re-colonised. Survival of new comers is probably additionally facilitated by the existing microbial community on rocks (Gorbushina & Broughton 2009) in a fashion known for immigrant bacteria on leaf surfaces (Monier & Lindow 2005).

Alternatively, rock-colonising lichens may supply buffered environments and refugia for RIF or organisms otherwise occupying other niches (Selbmann et al. 2010). Recent studies have shown that lichens harbour an amazing diversity of ascomycetous endophyte-like (endolichenic) fungi (Arnold et al. 2009), and phylogenetic relatedness was found between some endolichenic fungi isolated from saxicolous lichens and RIF (Harutyunyan et al. 2008). If in most cases, species from rock surfaces can still go back to their primary habitats, in some cases, these fungi keep specialising and get trapped in these extreme habitats. This may be the case for groups with no close relationships with plant-associated fungi, such as the genus Friedmanniomyces (Fig. 3).

Geographical distribution of rock-inhabiting fungi

The large majority of rock-inhabiting strains isolated thus far originated from rocks in the Mediterranean region or Antarctica (Sterflinger et al. 1999, Ruibal 2004, Ruibal et al. 2005, 2008 Selbmann et al. 2005, 2008). In Antarctica, RIF tend to grow within rocks, together with the cryptoendolithic lichen communities, finding shelter from extreme conditions prevailing on rock surfaces. In the Mediterranean area, RIF tend to grow on the rock surface or in cracks, causing damages to the substrate (e.g., biopitting of marble). Despite differences in temperature, they share similar morphological and physiological adaptations, such as melanisation, meristematic growth and oligotrophism.

Similarly to previous studies (Selbmann et al. 2005, Ruibal et al. 2008), our results show that Antarctic RIF often share an evolutionary history with RIF from semi-arid areas. In our study, RIF sampled in geographically disjoint localities (Antarctica versus Mediterranean region) cluster together in Davidiellaceae, the two groups of Teratosphaeriaceae, and unknown lineage 1 (Figs ​(Figs2, 2, ​,3).3). In some cases, Antarctic and Mediterranean strains are even phylogenetically very closely related, showing a recent common evolutionary history (e.g., in Teratosphaeriaceae 2, the Mediterranean rock isolates TRN 124 and A73 with the Antarctic strain CCFEE 5489). Likewise, some strains of Recurvomyces mirabilis and Elasticomyces elasticus have been recorded in the Antarctic as well as in high peaks of the Alps and Andes (Selbmann et al. 2008). Therefore, it seems that an efficient mechanism of dispersal, most probably wind-mediated (Gorbushina et al. 2007, Gorbushina & Broughton 2009), have led to a colonisation spanning different continents.

Work performed by C.R. at Duke University was supported by a NSF AToL grant (AFTOL, DEB-0228725) to F.L. Work performed by C.L.S. after 2008 was supported in part by the Intramural Research Program of the National Institutes of Health (National Library of Medicine), and until 2008 by a grant from NSF (DEB-0717476). Work performed by A.A.G. was funded by grants from the National Swiss Foundation (31003A-122513) and the Deutsche Forschungsgemeinschaft (DFG Go 897/3). Work performed by L.S. at the CBS was funded by a Synthesys grant. The authors would like to acknowledge the Italian National Program for Antarctic Research (PNRA) for supporting the collection of samples, the Italian National Antarctic Museum “Felice Ippolito” for supporting the Culture Collection of Fungi From Extreme Environments (CCFEE) and the Alpine guides A. Serafini and M. Heltai for collecting rock samples in the Alps and Aconcagua, respectively. Thanks to William Broughton for his editorial help and to the technical staff of the CBS for their assistance with the cultures.