Correspondence: Christine Hallmann, Department of Experimental Phycology and Culture Collection of Algae (SAG), Albrecht-von-Haller-Institute for Plant Sciences, Georg-August-University, Untere Karspüle 2, 37073 Göttingen, Germany. Tel.: +49 551 39 7866; fax: +49 551 39 7871; e-mail: firstname.lastname@example.org
Composition and diversity of aeroterrestrial phototrophic microbial communities are up to now poorly understood. Here, we present a comparative study addressing the composition of algal communities on sandstone substrata based upon the analysis of rRNA gene clone libraries from environmental samples and crude cultures. From a west-facing, shaded wall area of the mediaeval castle ruin Gleichen (Thuringia, Germany), sequences mainly related to the green algae Prasiococcus and Trebouxia (Trebouxiophyceae) were retrieved. A south-west-facing, sun-exposed wall area was mainly colonized by Apatococcus and a Phyllosiphon-related alga. Just a few species, in particular Stichococcus-related strains, were ubiquitous in both areas. Samples from a basement vault exposed to low irradiance exhibited Chlorophyceae like Chromochloris and Bracteacoccus. Thus, most green algae on the daylight-exposed walls were affiliated to Trebouxiophyceae, whereas Chlorophyceae were dominant in samples taken from the site kept under low irradiance. Accordingly, cyanobacterial communities were different: the sun-exposed area was dominated by Synechococcus-related organisms, while on the shaded wall area, cyanobacteria were almost absent. The filamentous Leptolyngbya dominated samples from the basement vault. Scanning electron microscopy revealed endolithic algal morphotypes (coccoid algae and diatoms) dominant in open pores between mineral particles. Here, the organisms may be also involved in biogenic weathering of stone.
Phototrophic microorganisms are important primary producers on hard rock substrata as well as on building facades (e.g. Karsten et al., 2007a; Horath & Bachofen, 2009). Eukaryotic microalgae and cyanobacteria, along with fungi and lichens, have also been recognized as important factors for rock weathering and stone decay (e.g. Welton et al., 2003; Büdel et al., 2004; Gorbushina, 2007). The rock substratum itself mostly provides harsh environmental conditions. Temperature may vary by several tens of degrees Centigrade during a day, accompanied by rapid desiccation (or freezing); that is, the availability of water is extremely limited (e.g. Walker & Pace, 2007). The availability of nitrogen and sulphur compounds as well as phosphate strongly depends on the substratum and other nutrient sources in the immediate surrounding, for example precipitation, soil, bird droppings and volatile emissions (e.g. Karsten et al., 2003).
It has been recognized that in extreme habitats, the productivity of the organisms must be close to lowest possible limits (e.g. Johnston & Vestal, 1991). Despite adverse conditions, the mineral substrata may be colonized on the surface (epilithic) or inside the substratum (endolithic) by relatively highly diverse communities of phototrophic and heterotrophic microorganisms (mainly fungi and bacteria).
Several studies identifying algae according to their morphology reported differences in algal diversity depending on diverse substrata including stone (e.g. Bellinzoni et al., 2003; Crispim et al., 2003; Uher, 2008; Macedo et al., 2009; Khaybullina et al., 2010). However, so far only few studies based on a culture-independent approach using rRNA gene as phylogenetic marker were performed for mineral substrata (e.g. Horath & Bachofen, 2009; Cuzman et al., 2010; Ragon et al., 2012). Also knowledge on factors that may determine the algal diversity on different but closely neighbouring sites is still lacking. It is obvious that – apart from irradiance – composition of the algal communities is determined by other physical parameters as well. In soil, the pH appears to be one factor that influences the dominance of the major groups of photoautotrophic organisms: cyanobacteria are known to prefer neutral and alkaline soils (Shields & Durell, 1964; Brock, 1973), whereas green algae prefer acidic soils (Starks et al., 1981; Lukešová & Hoffmann, 1995). In addition, cyanobacterial and green algal communities in soil may be influenced by soil type (Garcia-Pichel et al., 2001) and land use (Zancan et al., 2006). Previous studies of phototrophic communities on various building stones and rock substrata demonstrated that nutrients have less influence on the community structure than, for example, UV radiation, pH and aspect (Bellinzoni et al., 2003; Furey et al., 2007).
Microalgal and cyanobacterial resting stages may easily resist adverse environmental conditions (e.g. Häubner et al., 2006; Lennon & Jones, 2011). These resting stages and other kinds of propagules may be present as a ‘seed bank’ (Lennon & Jones, 2011). They may turn active after environmental conditions change. Thus, it may be assumed that a multitude of organisms will be detectable and even culturable from habitats with adverse environmental conditions, despite these organisms may be inactive and just present in low numbers in their natural habitat.
In this study, we show that the composition of phototrophic microbial algal communities including cyanobacteria differed markedly between apparently similar substrata. This is mainly due to differences in the exposure to sunlight (and hence water availability) and the occurrence of gypsum crusts. We used a molecular approach that allowed the identification of microalgae including cyanobacteria down to generic or even species level. We also discuss the results in view of possible biodeterioration mechanisms.
Materials and methods
The sampling sites were several wall sections of the castle Gleichen, near Gotha, Thuringia, Germany (50°52′49″N, 10°50′20″E). In this location, the average regional annual temperature is 8.1 °C, the sunshine duration is about 1500 h per year, and annual precipitation is about 560 mm (data taken from Deutscher Wetterdienst Offenbach, Germany). In the year of sampling (sampling date May 15, 2009), precipitation was higher than the long year average (March–May 173 mm, compared with 120 mm long year average in these months). Average temperatures in March (4.2 °C), April (11.2 °C) and May (13.2 °C) were slightly higher than the respective long year averages in the first two months (2.8 °C, 7.5 °C and 13.2 °C, respectively).
Samples were taken from two walls. A south-west-facing wall facade (wall area A; slope 90°, aspect value 210°) was exposed to direct sunlight for 8–10 h during these months. A west-facing facade (wall area B; slope 90°, aspect value 275°), due to the existing architectural structures, was reached by direct sunlight just in the afternoon for 3 h (from 4.00 h pm onwards, without direct sunlight in March). A third sampling site (basement vault) was not reached by direct sunlight. The walls consisted of various types of dimension stones (Stück et al., 2011; see below).
Sampling and cultivation
Samples of approximately 100 μL dry volume were collected in May 2009 from south-west-facing wall area A (Fig. 1a and b), west-facing wall area B (Fig. 1c and d) and a basement vault in a distance of 5 m from the entrance (Fig. 1e and f). The samples were scraped off with a sterile scalpel and collected in sterile 2-mL reaction tubes. Biofilm samples used for establishing the clone libraries were randomly taken from each wall area. All samples in area A were taken from sandstone (Gleichenberger Rhätsandstein) (c.f. Stück et al., 2011). Wall joints were mostly closed, that is, filled with gypsum mortar. Area B exhibited a variety of limestone and sandstone lithologies (travertine, Grenzdolomit, Rhätsandstein). Most wall joints were open. Again, just samples from sandstone (Rhätsandstein) were taken into consideration for this study. Samples from green biofilms grown in the inner faces of forming scales were collected from the basement vault (Fig. 1e–h).
For establishing crude cultures, aliquots of the biofilm samples were suspended in flasks with 20 mL 3N BBM+V medium (Starr & Zeikus, 1993) for green algae, BG11 medium (Rippka & Herdman, 1993) for cyanobacteria and Diat medium for diatoms (c.f. http://www.uni-goettingen.de/de/186449.html). The crude cultures were incubated at 18 °C on a 14:10-h light:dark cycle at 25 μmol photons m−2 s−1 from white fluorescent light for four weeks.
General procedures for handling and examination of DNA were performed according to Sambrook et al. (2000). Genomic DNA was extracted from collected environmental biofilm samples and crude cultures. If applicable, cultures grown on 3N BBM+V or BG11 were mixed prior to DNA extraction.
Cells were disrupted by shaking in a Mini-Beadbeater (Biospec Products, Bartlesville, OK) in the presence of equivalent amounts of acid-washed glass beads (120–200 μm and 425–600 μm in diameter; Sigma-Aldrich, St. Louis, MO) and vortexed briefly. The samples were treated in the beadbeater for 30 s at 5000 r.p.m. DNA was extracted with the Invisorb® Spin Plant Mini Kit (STRATEC Molecular, Berlin, Germany), following the manufacturer's instructions.
The MoBio PowerSoil® DNA isolation Kit (MoBio Laboratories Inc. Carlsbad, CA) was used for extraction of genomic DNA from samples of endolithic biofilms according to the manufacturer's instructions. Extraction results were evaluated after electrophoresis on a 1% (w/v) agarose gel. Isolated DNA was stored at −20 °C until further processing.
For amplification of eukaryotic rRNA genes from DNA preparations, PCR was performed as follows: 18S rRNA genes were first amplified using eukaryotic specific primers 20F (5′ GTAGTCATATGCTTGTCTC 3′) and 18L (5′ CACCTACGGAAACCTTGTTACGACTT 3′; Hamby et al., 1988) followed by a second amplification (semi-nested PCR) with the primers 20F and the newly developed CH1750R (5′ CTTCCTCTARTGGGAGG 3′), complementary to positions 1734-1751 of the 18S rRNA gene sequence of Chlorella vulgaris SAG 211-11b (accession number X13688), specific for green algae (this study). For cyanobacteria from environmental samples and crude cultures, the primers PCR1 and PCR18 (Wilmotte et al., 1993) were used for amplification of 16S rRNA genes.
Approximately 30 ng of the extracted DNA was used as template in each amplification reaction. The reaction mixture (25 μL) contained each dNTP at a concentration of 0.1 mM, 2.5 μL of 10× reaction buffer, 2 mM MgCl2, each primer at a concentration of 0.2 μM, 2 U of Taq DNA polymerase (Bioline, Luckenwalde, Germany) and 4% (v/v) dimethyl sulfoxide (DMSO) solution. PCR was performed in a thermocycler Primus 96Plus (MWG-Biotech, Ebersberg, Germany) using the following programme for the primer set 20F/18L: initial denaturation at 95 °C for 5 min, followed by 35 cycles of denaturation at 94 °C for 1 min, annealing at 50 °C for 1 min, extension at 72 °C for 3 min and final extension at 72 °C for 10 min. For the semi-nested PCR with the primer set 20F/CH1750R, a 1 : 25 dilution of the primary PCR product was used as template. The following programme was used: initial denaturation at 95 °C for 5 min, followed by 25 cycles of denaturation at 94 °C for 1 min, annealing at 54 °C for 1 min, extension at 72 °C for 3 min and final extension at 72 °C for 10 min. For the cyanobacterial primer set PCR1/PCR18, initial denaturation was at 95 °C for 5 min, followed by 35 cycles of denaturation at 94 °C for 30 s, annealing at 52 °C for 30 s, extension at 72 °C for 2 min and a final extension at 72 °C for 30 min.
From two crude cultures obtained from basement vault scales, diatom rRNA genes were amplified with specific primers as described in Pniewski et al. (2010).
All PCR products were purified using the Invisorb® DNA CleanUp Kit (STRATEC Molecular). Aliquots of 2 μL of the purified amplicons were analysed by electrophoresis on a 1% (w/v) agarose gel.
rRNA gene cloning and sequencing
Cloning was carried out with the TOPO TA cloning kit (Invitrogen, Carlsbad, CA) with TOP 10 chemically competent One Shot Escherichia coli cells (Invitrogen), as supplied by the manufacturer. In the plasmid blue/white screening, white E. coli colonies containing correct DNA insertions were further identified by direct amplification of the inserted DNA fragment with a vector-specific primer set M13F/M13R (Invitrogen). Positive clones were cultivated overnight in 2-mL reaction tubes with 1 mL LB medium containing 100 μg ampicillin. Plasmid DNA was purified with the Invisorb® Spin Plasmid Mini Two kit (STRATEC Molecular) and stored at −20 °C.
Sequencing reactions were performed with the Dye Terminator Cycle Sequencing v3.1 kit (Applied Biosystems, Darmstadt, Germany) and an ABI Prism 3100 (Applied Biosystems) automated sequencer. All eukaryotic clones were sequenced with the 18S standard sequencing primer 895R (5′ AAATCCAAGAATTTCACCTC 3′), resulting in partial sequences, including the hypervariable regions V2–V4 (Neefs & De Wachter, 1990; Hodač et al., 2012; Lee & Gutell, 2012). Prokaryotic clones were sequenced with PCR1 (Wilmotte et al., 1993), resulting in partial sequences including the hypervariable regions V1–V3 (Santamaria et al., 2012).
Phylogenetic and statistical analysis
The sequences were edited and assembled using SeqAssem (Hepperle, 2004). Sequences shorter than 400 bp were excluded from further analysis. The remaining sequences were compared with available sequences in NCBI by blastn (Altschul et al., 1990; http://www.ncbi.nlm.nih.gov/). Next, relative sequences were imported into the ARB program (Ludwig et al., 2004; http://www.arb-home.de). In addition, internal sequences provided by SAG Culture Collection of Algae (University of Göttingen) were included in the comparisons. To determine preliminary phylogenetic affiliations, the sequences were aligned with homologous rRNA gene sequences using the automatic alignment tool of the ARB program package.
Potential chimeras were checked by Bellerophon (Huber et al., 2004); in addition, the first and the last 300 bp of putative chimeras were compared with similar rRNA gene sequences in NCBI. Chimeric sequences were excluded from the data set.
Rarefaction curves and operational taxonomic units (OTUs) were calculated with MOTHUR (Schloss et al., 2009). OTUs were defined on the basis of ≥ 98% sequence similarity for 18S and 16S rRNA gene sequences (Romari & Vaulot, 2004; Marande et al., 2009; Michaud et al., 2012; Stock et al., 2012). Representative sequences of each OTU were selected and sequenced completely (Moon-van der Staay et al., 2001; Ragon et al., 2012) with standard sequencing primers.
Representative sequences were deposited in GenBank under the following accession numbers: JX127160–JX127192.
For phylogenetic analyses, alignments of rRNA gene sequences were performed using mafft, version 6 (Katoh & Toh, 2008), and small corrections were made by eye. Complete rRNA gene sequences were subjected to phylogenetic analyses using the maximum likelihood (ML) method by RAxML (Stamatakis et al., 2008), in conjunction with the GTR+Γ+I model with 100 bootstrap replicates. In addition, Bayesian posterior probabilities (MB) were calculated with MrBayes 3.2 (Huelsenbeck & Ronquist, 2001). Two parallel Markov chain Monte Carlo (MCMC) runs for two million generations each with one cold and three heated chains were conducted using the GTR+Γ+I model, with trees sampled every 100 generations.
To quantify differences between groups of samples, SIMPER (Similarity percentages) analysis was conducted using the program past, version 1.98 (Hammer et al., 2001). A similarity matrix was calculated based on the abundances (in percentages) of the algal OTUs. As a similarity measure, Bray–Curtis distance index was used. The significance of differences was tested by one-way ANOSIM using the same similarity measure.
Light and electron microscopy
Light microscopic observations were performed using an Olympus BX60 microscope (Tokyo, Japan) with Nomarski DIC optics equipped with a ColorView III camera (Soft Imaging Systems, Münster, Germany). Micrographs were processed using the Cell^D image software (Soft Imaging Systems).
For scanning electron microscopy (SEM), samples were fixed immediately after sampling in 2% (w/v) glutardialdehyde (EM grade, Sigma-Aldrich, Deisenhofen, Germany) and stored at 4 °C until further processing. Samples were dehydrated in an ascending ethanol series (15–99%), mounted on SEM sample holders and sputtered with Au–Pd (7.3 nm for 120 s). Samples were visualized in a SEM LEO 1530 Gemini (Zeiss, Oberkochen, Germany) combined with an INCA X-ACT EDX. Electron micrographs were colorized with Hornil StylePix (www.hornil.com/en/products/stylepix). Methods for transmission electron microscopy were performed as described in Hallmann et al. (2011a).
Macro- and microscopic observations
For this study, samples were taken from two south-west- and west-facing wall areas (A and B) and from scales formed on sandstone on a basement vault. All sampled sites exhibited obvious colonization by cryptogams, including algal (‘green’) biofilms. Thalli of endolithic lichens were abundant on sandstone at the base of wall area A. In addition, a thin gypsum crust (W. Wedekind, person. comm.) covered small sections of wall area A. Green biofilms were occasionally found under these crusts. On wall area B, thalli of crustose lichens were also present, but endolithic lichens were not observed. Along the basement vault, thick green coverings in the inner faces of scales (Fig. 1g and h) were observed.
SEM of the fracture faces of the scales from the basement vault revealed a dense cover by either filamentous (Fig. 2a and inset) or coccoid (Fig. 2b) morphotypes and diatoms (Fig. 2c and d). The coccoid cells were located in open pores and between small chips of the mineral particles (clay particles intermixed with gypsum according to EDX analysis). The filamentous morphotypes (putatively cyanobacteria, see below) built up a dense biofilm on the mineral particles (Fig. 2a).
Diatoms were identified as Diadesmis contenta D.G.Mann (Fig. 2c) and Achnanthidium minutissimum Czarnecki (Fig. 2d) (Krammer & Lange-Bertalot, 1986–2004). Additional diatom phylotypes were retrieved from two crude cultures by rRNA gene analysis (see below).
Various morphotypes of green algae were detected by light microscopy of the crude cultures, for example the cell package–forming green alga Prasiococcus calcarius (Fig. 3a). Also filamentous cyanobacteria were observed in these cultures (Fig. 3b).
rRNA gene analysis of the phototrophic community
Analyses of 16S and 18S rRNA genes aiming at detection of eukaryotic algae and cyanobacteria were performed for environmental samples and for crude cultures. Enrichment in crude cultures allows detection of organisms that are present in the original sample in just extremely low numbers of individuals, which may leave these species undiscovered in the environmental biofilm.
Cyanobacteria. 16S rRNA gene clone libraries were established from four environmental samples of wall area A and six of wall B. Four clone libraries from basement vault scale samples were established from crude cultures. A total of 11 cyanobacterial OTUs were recovered. With respect to cyanobacteria, wall areas A and B differed markedly (Fig. 4). On wall area A, 47 of 75 clones were assigned to cyanobacteria, with the majority of them (46) representing Synechococcus-like OTUs (cyanobacterial OTU 3 and 11, Table 1, Fig. 4). On wall area B, just seven of 70 sequenced clones were represented by cyanobacteria.
Table 1. Distribution of cyanobacterial OTUs on different wall areas
No. of clones
Closest relative (% similarity)
Representative full-length sequence in phylogenetic analysis.
From basement vault scale samples (Fig. 1g and h), 14 clones were retrieved, but with a dominance of clones representing Leptolyngbya-like cyanobacterial OTU 2 (Fig. 3c). This was in accordance with a high frequency of sheathed filamentous cyanobacteria as observed by electron microscopy of respective samples (Fig. 3d and e). In addition, three OTUs could be assigned to uncultured cyanobacteria (OTU 4, 5 and 6), three OTUs related to Chroococcus, one OTU related to Microcoleus vaginatus and one OTU representing Leptolyngbya-like cyanobacterial OTU 1 (Table 1, Fig. 5).
Green Algae. DNA preparations from environmental samples and from crude cultures were analysed with general eukaryotic primers and specific green algal primers as described in the methods section. In total, 648 18S rRNA partial sequences including 482 green algal clones were recovered from all sampling sites. These resulted in 22 green algal OTUs at a cut-off of 98% sequence homologies (Table 2).
Table 2. Distribution of green algal OTUs on different wall areas
No. of clones
Closest relative (% similarity)
Representative full-length sequence in phylogenetic analysis.
Rarefaction curves (Fig. 6) calculated for the biofilm samples of wall areas A and B and the basement vault scale samples reveal the clone library coverage for algal sequences. For wall area A, the rarefaction curve reached almost a plateau, and for wall area B, nearly full coverage of OTUs was reached. For the scale samples (biofilm samples and crude cultures), saturation was not reached.
On wall area A, 149 green algal clones representing 11 OTUs and eight fungal clones were found. Samples from wall area B yielded in 253 green algal clones representing 10 OTUs; 24 clones represented fungi. From scales, 214 clones were established, which were distributed on green algae (80 clones) representing 12 OTUs, flagellates (11 clones), fungi (13 clones) and mosses (110 clones). The green algal clones were processed for further statistical and phylogenetic analyses.
The composition of the green algal communities in wall areas A and B was apparently different, according to the analysis of clones from environmental DNA (Table 2, Figs 7-9). Whereas Apatococcus lobatus and Phyllosiphon arisari-related sequences were highly abundant on wall area A but either absent or rare on wall area B, P. calcarius- and Trebouxia asymmetrica-related sequences were detected nearly exclusively on wall area B. Only five green algal OTUs were shared between both wall areas, out of which three Stichococcus-related OTUs were in fact abundant (green algal OTUs 3, 4 and 8; Table 2).
The composition of the algal community from the basement vault scale samples has little in common with the communities from wall areas A and B (Fig. 7). Just the Stichococcus minutus (green algal OTU 4)- and P. arisari (green algal OTU 16)-related OTUs were shared by all three sampling sites.
With respect to wall areas A and B, crude cultures revealed only sequences belonging to Trebouxiophyceae, affiliated to green algal OTUs 2, 3 and 4. From the basement vault scale samples, just a single green algal OTU (a Chromochloris zofingiensis relative, Chlorophyceae) was recovered in the environmental clone library. In contrast to crude cultures from wall area samples, a high number of otherwise undetected OTUs were retrieved from crude cultures of scale samples (Table 2). Again, Chromochloris zopfingiensis-like green alga was abundant in these clone libraries. In addition, Bracteacoccus- and Stichococcus-related sequences were frequent.
In summary, phylogenetic analysis revealed most green algal OTUs belonging to the green algal class of Trebouxiophyceae (16 OTUs; Fig. 8); only 6 OTUs represented Chlorophyceae (Fig. 9). In addition, one member of the Ulvophyceae (Trentepohlia sp.) was found in two clones on wall area A (data not shown). Interestingly, on wall areas A and B, Trebouxiophyceae were dominant. Only one OTU, retrieved from environmental DNA cloning on wall area A, a Jenufa sp.-related clone, could be assigned to the green algal class of Chlorophyceae. All other chlorophycean clones were retrieved from environmental material and crude cultures of scale samples.
Diatoms. From scale samples developed with diatom-specific 18S rRNA gene primers, 39 clones closely related to Nitzschia amphibia (AJ867277, 98% similarity), and 15 clones related to Phaeodactylum tricornutum (AJ269501, 96% similarity) were identified according to blast database queries. However, the latter clones were assigned to Diadesmis sp. in the ARB phylogenies. In addition, one sequence showed 99% similarity to Spumella sp. strain Mbc 3C (Chrysophyceae).
Distinction among sites
The rRNA gene sequence data were used for statistical analysis to compare community compositions between the different locations. SIMPER analysis documented the dissimilarity between the daylight-exposed wall areas and the endolithic biofilm of scale samples (93.4% between wall area A and scale samples, and 96.6% between wall area B and scale samples; ANOSIM significance: P < 0.01). In contrast, both the daylight-exposed wall areas showed 76.4% dissimilarity.
The molecular phylogenetic approach in our study revealed clear differences in the composition of green algal and cyanobacterial communities between two wall areas made of apparently similar substrata (sandstone) located at the same building. The wall areas experienced different exposures to sunlight and differed by the presence or absence of gypsum mortar and crusts. The composition of the fungal microbial communities at the same building was already found to differ at small scales in a previous study (Hallmann et al., 2011b). A previous study on different substrata from geographically distant locations revealed the substrate type as most significant factor to determine prokaryotic communities. For eukaryotic communities, the geographic location appeared to be more significant (locations in France and Ireland, Ragon et al., 2012).
Although we used only partial rRNA gene sequences to assess the community compositions, we employed full rRNA gene sequences for robust phylogenetic analyses. 18S rRNA partial gene analysis of the hypervariable regions V2–V4 or just the V4 region are recommended as DNA barcodes for various algal groups, for example dinoflagellates and diatoms (Zimmermann et al., 2011; Ki, 2012; Pawlowski et al., 2012). Similarly, for 16S rRNA genes, the suitability of the hypervariable regions V1–V3 in phylogenetic analyses turned out to be comparable to full-length sequences (Jeraldo et al., 2011). Consequently, partial rRNA gene sequences are still commonly used in microbial biodiversity studies using a culture-independent approach (Santos et al., 2010).
There is an obvious difference between the two wall areas concerning the abundance of cyanobacterial clones. Whereas just seven cyanobacterial clones could be retrieved from area B, more than 50% of the clones recovered from area A represented two OTUs (OTU 3 and 11, c.f. Table 1), that is, Synechococcus-like cyanobacteria. The latter two OTUs had their next relatives in blastn queries with 98% and 96% similarities in Synechococcus-like cyanobacteria retrieved from a marble sculpture (JQ404415; China, Beijing, Forbidden City). Phylogenetic analysis (Fig. 5) revealed an uncultured cyanobacterium from gypsum crusts (GQ325750 from Tunisia; Stivaletta et al., 2010) as another next-closest relative. Although our Synechococcus-related clones are closely related to each other, generally Synechoccoccus strains belong to several deeply branching lineages within the cyanobacteria. This has been shown recently for a number of isolates mainly obtained from freshwater (Robertson et al., 2001). With respect to our findings, it should be noted that some Synechococcus isolates could be retrieved from saline and hypersaline environments, including gypsum crusts (e.g. Oren & Seckbach, 2001; Crispim & Gaylarde, 2005). The predominance of Synechococcus in our samples may be also well explained by the presence of a thin gypsum layer on the stone surface (c.f. Fig. 1b). It is generally accepted that cyanobacteria instead of green algae dominate alkali – including hypersaline – environments (e.g. Oren & Seckbach, 2001). Gypsum was one component of the used mortar and may have been dissolved by precipitation (c.f. Hoppert et al., 2010). Typically, the cyanobacteria grow inside the crust, in particular at the boundary layer between crust and stone substratum (Saiz-Jiminez et al., 1990). In the dry and cold Antarctic habitat, hypolithic growth at the contact face between stone and underlying ground has been observed for cyanobacteria (see Cary et al. (2010) for a review). This layer of active growth also defines the fracture plane of the gypsum crust (Hoppert et al., 2010). For Synechococcus, in particular, the binding of calcium ions on the negatively charged surface layers of cells has been described, which contributes to dissolution and biogenic weathering of limestone (Schultze-Lam & Beveridge, 1994).
Leptolyngbya (cyanobacterial OTU 2; Table 1) was abundantly recovered under large basement vault scales (c.f. Fig. 1g and h). Here, the fracture face of the scale is composed of clay particles intermixed with gypsum. Moreover, the site is moist and just exposed to dim light for few hours per day. Leptolyngbya has been predominant in phototrophic biofilms from hypogean sites (Zammit et al., 2011), but was also detected in diverse endolithic microbial habitats such as travertine (travertine terraces of the Yellowstone National Park), alpine dolomite, and even deep sea basalt and gypsum crusts in Tunisia (Norris & Castenholz, 2006; Horath & Bachofen, 2009; Stivaletta et al., 2010). Leptolyngbya OTU 2 is closely related to an uncultured cyanobacterium recovered from a hypolithic biofilm on quartz rock (FJ790618; Wong et al., 2010).
Further cyanobacterial OTUs recovered from clone libraries represented M. vaginatus and Chroococcus sp. Both species/genera are common cyanobacteria of terrestrial habitats also colonizing dimension stone (e.g. Ortega-Calvo et al., 1991; Cuzman et al., 2010). The cyanobacterial OTUs 4, 5 and 6 from the clone libraries were phylogenetically closely related to Chroococcidiopsis. Their close relatives were from terrestrial or hypolithic sites (GQ396895, HM241004, FJ790556, HM224428). The dominance of either Synechococcus on wall area A or Leptolyngbya and other cyanobacteria in the scale samples may be due to the different exposure times to direct sunlight, high calcium concentration in gypsum crusts (as compared with scales) or different moisture regimes at these sites.
In fact, filamentous cyanobacteria were only detected on shaded wall area B and in the scale samples. According to a recent study on microbial biofilms on monuments of the Angkor Wat temple complex (Cambodia), filamentous cyanobacteria like Microcoleus vaginatus and Leptolyngbya appeared to be more prevalent in moist areas as well (internal walls; Gaylarde et al., 2012). However, both filamentous cyanobacteria are also important colonizers of soil crusts and deserts with specific adaptations to desiccation stress like exopolysaccharide sheaths (Garcia-Pichel et al., 2001; Büdel et al., 2009; Pereira et al., 2009).
Eukaryotic algal communities
All eukaryotic algae retrieved in our clone libraries belonged to two classes of green algae, either Trebouxiophyceae or Chlorophyceae. Most OTUs from wall areas A and B were members of Trebouxiophyceae (15 out of 16). Three distinct groups of clones (OTUs 3, 4 and 8), phylogenetically related to Pseudostichococcus monallantoides strain SAG 380-1 (Stichococcus mirabilis strain CCAP 379/3 with 99% similarity in BLAST queries), S. minutus, strain NJ-17 and Diplosphaera sp. strain J4028B (Stichococcus sp. strain MBIC10465 with 99% similarity in BLAST queries), were retrieved in high abundance from both walls (c.f. Table 2 and Fig. 8).
Green algal OTU 5 was recovered only from wall area B and in low clone numbers; it was phylogenetically affiliated to P. calcarius (Stichococcus sp. strain MBIC10465, with 98% similarity in BLAST queries). Another P. calcarius-related sequence, OTU 6, was also found only at wall area B, but with high clone numbers. For OTU 6, both phylogenetic analyses and BLAST queries concurred on the species same identification, the cell package forming–P. calcarius (c.f. Table 2, Figs 3a and 8). The species has been described as a widely distributed subaerial alga on moist soil, calcareous rock and stone walls (Ettl & Gärtner, 1995), including the Antarctic region (Belcher, 1969; Broady, 1983). There, P. calcarius was observed to be dominant in epilithic communities at high-salinity sites where irradiation is infrequent or moisture is available (Broady, 1996).
Obviously, the ‘Stichococcus’ morphotypes within the Prasiola clade are phylogenetically diverse and also differ with respect to their abundance in specific sites. In scale samples, the S. minutus-related clones could be retrieved just from crude cultures. No environmental clone from scale samples, though, was related to Stichococcus, which accounts for a minor ‘contamination’ of the site by this clone.
Surprisingly, A. lobatus-related clones could not be retrieved from shaded wall area B. On wall area B, the highly abundant Prasiococcus (beeing absent on wall area A) appears to be better adapted to this site. Apatococcus, like Stichococcus, is known as a genus of cosmopolitan algae. Cell packages of A. lobatus were described from tree bark, wood, walls, rocks (Ettl & Gärtner, 1995) and man-made substrata (Rindi et al., 2009). The sequences revealed in this study were closely related to A. lobatus strain SAG 2037. The resistance of Apatococcus against air pollution in towns and xeric conditions is well known (Barkman, 1969). In this particular case on wall area B, Apatococcus, although exhibiting wide ecological amplitude, might be less competitive. As Apatococcus is mixotroph (Gustavs, 2010) and is able to grow under shaded conditions, eventually lack of organic substrates on this site may be a reason for the absence of A. lobatus-related clones.
Clones closely affiliated to the rather uncommon green alga P. arisari Kühn could be retrieved from wall area A in high numbers (six sequences from area B, one sequence from scale samples). The green alga P. arisari has been described as a plant parasitic alga penetrating Arisarum leaves in coastal Mediterranean and in tropical climates, but was also described for temperate climates (Aboal & Werner, 2011). Due to this fact, it can be assumed that the reference sequence is a misidentification. This reference sequence, in fact, seems to be assigned to Lobosphaeropsis pyrenoidosa Reisigl (T. Darienko, pers. commun.). Lobosphaeropsis pyrenoidosa was discovered in soil (Ettl & Gärtner, 1995), on tree bark (Freystein et al., 2008) and on stone (T. Darienko, person. comm).
From wall area A, only one chlorophycean alga related to Jenufa minuta was retrieved. Isolates were obtained from tree bark in Singapur (Němcová et al., 2011) and from soil in Germany (Hodač et al., 2012), and detected in clone libraries from endolithic samples in the Alps, Switzerland (Horath & Bachofen, 2009). All six chlorophycean OTUs detected in this study were retrieved from crude cultures of scale samples. Interestingly, enrichment cultures of samples from other wall areas did not result in an enrichment of Chlorophyceae, which accounts for a ‘seed bank’ of chlorophycean algae particularly in scales. Except Jenufa, these Chlorophyceae belong to the order Sphaeropleales, which includes vegetatively nonmotile unicellular or colonial taxa (Lewis & McCourt, 2004).
SEM of scales revealed the dominance of one coccoid morphotype (Fig. 2b), along with cyanobacteria and diatoms, from the basement vault scale sample. This morphotype was assigned to an OTU related to C. zofingiensis (GU827478, formerly Chlorella, Muriella, ‘Myconastes’; Hindák, 1982; Ettl & Gärtner, 1995; Krienitz et al., 2011; Fučíková & Lewis, 2012). The same OTU was also abundant in environmental samples from scales. The aerophytic alga has been found on rather sandy soil and moist substrata. This implies that Chromochloris is adapted to rather moist terrestrial environments (Ettl & Gärtner, 1995).
As the other chlorophycean OTUs could only be retrieved from crude cultures, they must be present just in minor proportions in the original sample. These clones are, like Chromochloris, rather soil algae than algae normally found on hard rock substrata and on wall areas A and B. This feature is also pointed out by SIMPER analysis: wall areas A and B are clearly distinct from the scale samples, in particular caused by presence/absence of Chlorophyceae.
The diatoms observed in the scale samples are cosmopolitan species, not necessarily restricted to terrestrial habitats. Diadesmis contenta is known, for example, for cave systems in Austria and Czech Republic (Schagerl, 1991; Poulíčková & Hašler, 2007) and was also found in lichen thalli (Lakatos et al., 2004). Achnanthidium minutissimum was retrieved, for example from aquatic systems (Potapova & Hamilton, 2007) and caves (Poulíčková & Hašler, 2007). Interestingly, one clone related to the flagellate Spumella sp. was retrieved in the scale samples. Chrysophyceaen algae are well known from freshwater and soil habitats (Boenigk et al., 2005), but could be also retrieved from moss (Škaloud, 2009).
Diversity patterns of the phototrophic community and possible biodeterioration mechanisms
It is reasonable to assume that algal communities found in the dark and moist environment of a basement vault differ from communities exposed to sunlight and, hence, to desiccation stress. However, it may not be expected that algal communities on two walls with (seemingly) similar environmental conditions are clearly distinct from each other.
Although undoubted reasons for these distinctions have yet to be elucidated, two differences between the wall areas are obvious: (i) different sun exposure time per day attended by water availability and (2) wall joints filled with gypsum mortar on wall area A in contrast to open wall joints on wall area B, resulting in the formation of thin gypsum crusts on wall area A. This accounts for clearly distinct ecological adaptations of microalgae on either wall area A or wall area B. It should be taken into account that in addition to the crucial criterion of the available moisture, also the chemical properties of the substratum substantially influence the settlement of specific algae (Darienko & Hoffmann, 2003; Rindi, 2007).
However, crude culturing leads to enrichment of other algal genera that may be well present in low numbers in the environment, but readily capable of multiplying under appropriate conditions. Obviously, these ‘seed banks’ differ between the various sampled sites. Although various Chlorophyceae are present in basement vault scales, the same genera appear to be nearly completely absent on walls.
Knowledge on the diversity, ecophysiology and dispersal strategies of these algae are still in its infancy (Karsten et al., 2005; Rindi, 2007). The different ecological adaptations of, for example, Lobosphaeropsis pyrenoidosa (P. arisari) or P. calcarius being dominant on either wall area A or B are largely unknown. Generally, algae exhibit a variety of adaptations to the terrestrial habitat (Häubner et al., 2006). Dry periods (in summer with low precipitation or due to frost drought) require adaptations to desiccation. These adaptations are manifested by certain cytological features such as thickened cell walls (Klebsormidium, Holzinger et al., 2011; Karsten & Holzinger, 2012; Zygogonium, Hoppert et al., 2004) and mucilaginous sheaths (Coccomyxa; Karsten et al., 2005). These structures may retain water for a certain time period and help to withstand high osmotic stress during desiccation. The accumulation of exopolysaccharides (EPS) as a protection against desiccation has already been reported (Shepherd & Beilby, 1999). Equally important are mechanisms protecting against high radiation, such as the accumulation of carotenoids and the formation of MAAs (mycosporine-like amino acids; Karsten et al., 2007b). Additionally, the presence of polyols, for example ribitol, which were considered as effective stress metabolites, was demonstrated in representative aeroterrestrial algae (Gustavs et al., 2011). Moreover, resting stages such as spores persist for years (Karsten et al., 2005).
In a recent study, different wall areas of the castle Gleichen were mapped, and distinct weathering patterns like formation of crusts or flakes and salt efflorescences were reported for wall area A (Stück et al., 2011). In our study, no direct evidence for biodeterioration, such as traces of actively penetrating endolithic organisms or mineral dissolution around single cells by microalgae, could be found, although it must be expected that the mass development of cyanobacterial and algal layers will destabilize pores and clefts just by mechanical forces (Warscheid et al., 1991; Crispim & Gaylarde, 2005). The secretion of organic acids at least by some organisms must be expected. It is obvious that the algae and cyanobacteria use pre-existing fracture planes under crusts or scales (these fracture planes were also observed without any apparent colonization), but may then accelerate detachment of these features. Just because of the high number of individuals (as implied by clone library data), key players of colonization and therefore putative agents of biogenic weathering could be clearly defined: In case of wall area A, the abundant cyanobacterium Synechococcus (-like), but also Phyllosiphon (Lobosphaeropsis pyrenoidosa) and Stichococcus/Pseudostichococcus, must be taken into account. On wall area B, besides Stichococcus, Prasiococcus was highly abundant. Finally, the basement scale sample was dominated by Chromochloris zopfingiensis and Leptolyngbya. These dominating organisms were also accompanied by a set of distinct other species, except from a few generalists (in our case, just Stichococcus). It has to be expected that these organisms also exert different mechanisms of biogenic weathering on the material surface. Thus, also effects of biogenic weathering regimes may differ on a very small scale.
Factors that influence the dominance of specific phylotypes were obviously irradiance, moisture and presence or absence of gypsum crusts. However, a more detailed analysis of the relevance of these determinants will require further quantification of these physical and chemical parameters of the substratum and long-term measurements of microclimatic conditions.
We are indebted to Tatyana Darienko for her helpful discussion concerning the Phyllosiphon-related sequence (Angewandte Ökologie, University of Rostock). Support of Heidrun Stück, Wanja Wedekind and Siegfried Siegesmund (Geoscience Centre, Georg-August-University, Göttingen) with respect to sandstone lithologies is gratefully acknowledged. This project was funded by the Deutsche Bundesstiftung Umwelt (DBU). Parts of this work were also supported by the German Science Foundation (DFG) by a grant extended to T. F. (Fr 905⁄16-1) and by the German Federal Ministry of Education and Research, BMBF (AlgaTerra project, grant 01 LC0026) within the BIOLOG program. This is Courant Research Centre Geobiology publication no. 114.