Correspondence: Tracy B. Norris, Center for Ecology and Evolutionary Biology, University of Oregon, Eugene, OR 97403, USA. Tel.:+1 541 7298815; fax: +1 541 3462364; e-mail: email@example.com
Molecular and culture based methods were used to survey endolithic, photosynthetic communities from hot spring-formed travertine rocks of various ages, ranging from<10 to greater than 300 000 years. Much of this travertine contained a 1–3-mm-thick greenish band composed mainly of cyanobacteria 1–5 mm below the rock surface. The travertine rocks experienced desiccation in summer and freezing in winter. A total of 83 environmental 16S rRNA gene sequences were obtained from clone libraries and denaturing gradient gel electrophoresis. Small subunit rRNA gene sequences and cell morphology were determined for 36 cyanobacterial culture isolates from these samples. Phylogenetic analysis showed that the 16S rRNA gene sequences fell into 15 distinct clusters, including several novel lineages of cyanobacteria.
Extremophilic microorganisms are by definition those that live at the outer limits of the physical parameters that define the boundaries for life on Earth. One niche that encompasses a number of extreme physical stresses is the endolithic environment of rocks of hot and cold deserts. Here we are using the term endolithic as defined by Friedmann et al. (1967) as meaning ‘occupying a space within the rock tissue without an apparent connection to the outer rock surface’, in contrast to chasmolithic (living in fissures in rocks). The term cryptoendolithic is used by some authors with the same meaning as the more general term endolithic (e.g. Sun & Friedmann, 1999; Hughes & Lawley, 2003). In such environments microorganisms must endure extremes of temperature (often including freezing), desiccation, low nutrient supply, and low photon flux. In most regions these communities are composed primarily of photosynthetic microorganisms (Friedmann, 1982; Bell, 1993; Gerrath et al., 1995; Sigler et al., 2003). Cyanobacteria are commonly the principal phototrophic members of endolithic communities. However, these communities include heterotrophic bacteria and often fungi, protolichens, and micro-algae. Studies of hot desert and Antarctic endolithic communities have demonstrated that, in general, endolithic communities are of low species diversity (Friedmann, 1982; Bell, 1993; Nienow & Friedmann, 1993; Wessels & Büdel, 1995), especially compared with soil crusts that are subject to similar extremes but are generally composed of more complex consortia of microorganisms (Garcia-Pichel et al., 2001; Redfield et al., 2002).
Many endolithic communities are characterized by a 1–4 mm thick greenish layer of phototrophs that commonly occurs 1–5 mm below the upper surface of the rock they inhabit (Fig. 1); however, the depth may vary considerably, and in some cases the visible green layer may be spread over more than 5 mm. The depth is presumably dependent on the penetrance of light mainly (Matthes et al., 2001). Since endolithic microorganisms inhabit the spaces between the mineral particles and crystals they are usually found in relatively porous and somewhat translucent rocks such as limestone, sandstone, gypsum and dolomite. In addition to these periodically dry environments, a thermal, active endolithic community has been reported in extremely acidic, but moist, siliceous sinter and dominated by unicellular red algae of the order Cyanidiales (Gross et al., 1998; Walker et al., 2005).
Endolithic microorganisms, in general, probably favor this environment as it allows them to escape competition by occupying a niche that is uninhabitable for most other microorganisms, although growth may be extremely slow (Sun & Friedmann, 1999). Another possible advantage to living in the rock interior is that it would provide an escape from harsher conditions at the surface, such as high solar irradiance (including UV radiation), rapid desiccation, and abrasion by wind-borne elements. The endolithic environment may act as a refugium, where light, moisture and nutrient availability form a fragile but sufficient balance for life. The combination of these three factors surely dictates the depth to which endolithic communities live within the rocks.
Ancient and recent travertine deposited by hot springs in the temperate climate of Yellowstone National Park, Wyoming, USA, harbor an apparently unique endolithic community. Travertine is a form of calcium carbonate (CaCO3) deposited by calcareous springs or rivers, but with numerous variations in porosity, crystal forms, and mineralogy even within the Mammoth area of Yellowstone (see Fouke et al., 2000). The July high temperature in this area averages about 28°C and often exceeds 34°C but with nighttime lows of ∼4–10°C and a mean monthly precipitation of 2.0 cm. The January mean low temperature is −15°C, with extremes of −45°C. Relative humidity values in summer may commonly drop as low as 14%.
The goal of this work is to provide an initial survey of the photosynthetic microorganisms that inhabit this extreme environment. Much of the previous work on endolithic communities has entailed only microscopy and relatively little cultivation (Friedmann, 1982) or cultivation without molecular evaluation, thus revealing mostly morphotypic information. Only recently have molecular techniques been applied to describe the community diversity of endolithic communities (Sigler et al., 2003; Taton et al., 2003; de la Torre et al., 2003; Walker et al., 2005). The present study uses molecular methods as well as morphotypic characterization of culture isolates to survey the taxonomic composition. We believe studies of endolithic microbial communities are key areas of investigation in furthering the exploration of past life on Mars and, most certainly, exploration of early life on the terrestrial Earth that may have occurred as early as 2.6 × 109 years before the present (Watanabe et al., 2000).
Materials and methods
Field site and collections
Dry travertine rock samples were collected from six different sites in or near the northwestern quadrant of Yellowstone National Park. Replicate samples for molecular work and culture isolation were collected in 2002. Samples for molecular work were frozen at −20°C upon returning to the lab and dry samples to be used for culture isolations were stored in plastic containers at ambient temperature in the dark. Additional collections for cultures were made in 2003. All of the sample sites are located in the Mammoth Upper Terraces of Yellowstone National Park, except for the ‘Hoodoos’, which comprise the travertine rockfall of Terrace Mountain (above the Mammoth Upper Terraces), formed over 350 000 years ago (Sturchio et al., 1994), and the ancient Gardiner Lower Terraces, which are above the city of Gardiner, Montana, and outside the boundaries of Yellowstone National Park. The global positioning system (GPS) coordinates for each collection site are as follows:
Abbreviations following site names were used for naming the environmental clones and denaturing gradient gel electrophoresis (DGGE) bands. The sampling locations included travertine of various depositional ages.
DNA purification from travertine samples was done using a modification of the method of Moréet al. (1994). The green layer occurring 1–2 mm beneath the surface of travertine samples was removed using a sterilized metal file and razor blade, after abrasive removal of the upper few millimeters of rock surface. Two aliquots of ∼1 g filings were collected for each sample and transferred to 2 mL screw-cap microcentrifuge tubes. Next, 0.75 g of 0.1 mm diameter zirconia/silica beads (BioSpec Products, Bartlesville, OK) were added to each tube along with 600 μL 120 mM sodium phosphate (pH 8.0) and 400 μL lysis buffer (10% sodium dodecyl sulfate, 0.5 M Tris-HCl (pH 8.0) and 0.1 M NaCl). Cells were lysed by shaking for 3 min at high speed on a Mini-Beadbeater (BioSpec Products, Bartlesville, OK) and then centrifuged for 3 min at 13 000 g. Supernatant (700 μL) was collected and DNA was precipitated on ice using 2 : 5 (v/v) of 7.5 M ammonium acetate and then centrifuged again. The supernatant was collected and the DNA was isopropanol precipitated. Finally, the pellet was washed with 70% ethanol (−20°C), air dried and resuspended in 100 μL of 10 mM Tris (pH 8.0).
Clone library construction
PCR amplification from total community DNA was used for clone library construction and DGGE. The cyanobacterial specific primers CYA106F and CYA781R (Nübel et al., 1997) were used to amplify 16S rRNA gene sequences, resulting in a product of approximately 675 bp. Each PCR reaction contained 1.5 mM MgCl2, 0.2 mM dNTPs, 0.5 μM of each primer, approximately 10 ng of template DNA, 0.1 mg mL−1 bovine serum albumin, 2.5 U Taq polymerase and 1 × PCR buffer (Promega, Madison, WI) in a total volume of 50 μL. The PCR amplification cycle was as follows: initial denaturation for 5 min at 95°C, then 30 cycles of 1 min of denaturation at 94°C, 1 min annealing at 60°C, and 1 min extension at 72°C followed by a final extension of 7 min at 72°C. PCR amplicons were cloned into the PCR 2.1 cloning vector using the TOPO TA cloning kit (Invitrogen, Carlsbad, CA). Clone libraries of 60–100 clones were created for each site in the study. For each library, 12–15 random clones were selected for re-amplification of vector inserts and commercial sequencing using the cyanobacteria-specific primers CYA359F and CYA781R (Nübel et al., 1997). Sequenced fragments had a length of approximately 380 bp. A GenBank BLAST (basic local alignment search tool) search was performed for each sequence to identify the nearest relatives. The environmental clone sequences from this study were submitted under the GenBank accession numbers AY790390–AY790464.
For DGGE analysis, cyanobacterial 16S rRNA gene sequences were amplified from total community DNA using the cyanobacterial specific primers CYA359F and CYA781R. DGGE analysis was performed using a D-Code system (Bio-Rad, Hercules, CA) essentially as described by Muyzer et al. (1993). Reamplified PCR products from bands were sequenced commercially and a BLAST search of the GenBank database was done to identify species or strains of nearest relatives. The sequences of DGGE bands from this study were submitted to the GenBank database under the accession numbers AY790465–AY790472.
Rock samples that were used for molecular analysis were also used for culture isolations of phototrophs. Fine particulate samples of the green layer in the travertine samples were removed with sterile file and blade as in the molecular analyses. Most culture isolates were obtained by the dilution to extinction method, although direct enrichments were also used. A small portion of the filings (∼0.5 g) was used to initiate a dilution series in BG11 medium, or in some cases, ND medium that lacked combined nitrogen (Castenholz, 1988). Cultures of filamentous morphotypes were isolated from final dilution enrichments by transfer to agar plates. Filaments (trichomes) were allowed to migrate or grow out from the source; then single trichomes (clones) on minute agar blocks were removed with watchmaker's forceps under a dissecting microscope, and transferred to liquid medium in sterile tubes (Castenholz, 1988). To obtain non-filamentous types, the final enrichment was streak-plated and single colonies were transferred to liquid medium. In most cases, these cultures were replated for greater assurance of clonality. In many cases, axenic cultures resulted, but the presence of heterotrophic bacteria in some cultures still allowed molecular cyanobacterial identification with the use of the cyanobacteria-specific primers CYA359F and CYA781R. Culture isolates of green algal strains were sequenced using the 18S rRNA gene primers NS1 and NS2 (White et al., 1990), yielding a 555 bp product. The culture isolate sequences from this study were submitted under the GenBank accession numbers AY790836–AY790874.
The 16S rRNA gene sequences were analyzed for phylogenetic affiliation using the BLAST (Altschul et al., 1997) available on the NCBI website (http://www.ncbi.nlm.nih.gov/). The aligned SSU rRNA gene sequences of six known cyanobacterial relatives were exported from the ARB program (Ludwig et al., 2004) into Clustal X (Thompson et al., 1997) and used as a guide file for profile alignment of both sequences from this study and additional cyanobacterial sequences from GenBank. Alignments of partial 16S rRNA gene sequences corresponded to Escherichia coli sequence positions 412–807. The program PAUP 4.0b10 was used to construct a distance tree by the neighbor joining method (Swofford, 2001). The Kimura two-parameter algorithm was used with pairwise deletion of gaps and missing data. Bootstrap values from 1000 resamplings were calculated and Chloroflexus aurantiacus DSM 637 was used as the outgroup. The tree includes all of the environmental clones, DGGE bands and culture sequences from this study as well as some of the nearest relatives determined by BLAST. Phylotypes that included more than one sequence, defined by groups of sequences with greater than 97.5% sequence identity (i.e. operational taxonomic units, OTUs), were represented by a single sequence from that group for tree construction. Parsimony was used as an alternate method of tree construction to check for robustness of sequence clusters.
Collections and cultures were examined using an American Optical research microscope with phase contrast × 40 and × 100 objectives. The photomicrography was with a Zeiss Axioplan microscope, using a × 100 plan-neofluor objective with Nomarski DIC. Photo-images were made with a Nikon Coolpix 990 digital camera. The taxonomic descriptions and tentative genus names used in Bergey's Manual of Systematic Bacteriology (Boone & Castenholz, 2001) were used for morphotypic identifications.
The greenish bands of about 1–2 mm thickness were usually 1–4 mm below the upper surface of the rocks and visible, especially when the upper surfaces were light colored and not covered with crustose lichens. Most of the samples were quite similar to that shown in Fig. 1 (inset) but, in some cases, vertical fracture lines occurred in the rocks that apparently allowed light penetration and water to greater depths (e.g. 5–10 mm). Such communities are termed chasmolithic. If the rocks were extended as overhanging ledges, the endolithic green layer was also apparent near the lower surface. There was no visual evidence in this study of microorganisms boring into the rocks. Although no quantitative measurements were made, the youngest rocks were the most porous and easily breakable, whereas the older rocks were usually of significantly greater hardness. The collections were made mainly in mid summer. During a typical warm summer day (25 to >30°C) with 14–18% relative humidity immediately above most of the sites, the water content of the rocks was less than 3%, and even lower (0.4%) in less porous rocks of the Hoodoos. In one sample of relatively non-porous travertine, with a maximum noontime rock surface temperature of >35°C, a temperature of 28°C was reached at 5 mm depth by c. 14:00 hours and minimum nighttime temperature of 10–12°C by 04:00 hours AM (Jon Wraith, pers. comm.). Winter temperatures are normally below 0 °C and often plunge to −30°C or lower. Only one of the sites was possibly influenced by water from a neighboring active spring. Narrow Gauge Lower Terrace (NGL) had experienced water flow 3–4 years prior to collecting, and this more porous and delicate deposit retained 8–20% water, probably for several days. Samples were collected from various compass faces of the travertine deposits. Although not quantified, northerly or more shaded exposures appeared to harbor an endolithic green band more frequently and more prominently than in other exposures.
Analysis of community diversity using DGGE
DGGE community profiles of cyanobacterial 16S rRNA gene sequences for each site were generated (data not shown). The profiles for each community were characterized by a few (two to seven) distinct bands over a background of many more poorly resolved, indistinct bands. By visual inspection, it appears that the communities may share some common phylotypes, as evidenced by the equidistant migration of bands from different profiles. Each profile also contained some uniquely migrating bands as well. BLAST search results for the sequences that were retrieved from DGGE bands are included in Table 1. In most cases the nearest relative was an uncultured (and usually unidentified) cyanobacterium or occasionally a chloroplast of a green alga or moss from another study.
Table 1. Summary of rRNA gene sequence data combined from clone libraries, DGGE and cultures
A total of 75 cyanobacterial 16S rRNA gene sequences were determined from the clone libraries, which included 12–15 clones from each site. The clones were divided into phylotype groups (Table 1). We have defined members of a phylotype group as sequences with greater than 97.5% sequence identity to each other (Stackebrandt & Goebel, 1994), although this does not guarantee species or genus identity. The combined sequence data for clones and DGGE bands from all sites can be divided into 28 phylotype ‘groups’, of which 15 have two or more sequences. An additional 13 phylotypes are each represented by a single sequence. Within the phylotype groups there were some sequences that were 100% identical (over ∼380 bp) to GenBank database sequences. From the consolidated clone library there were eight sequences that were retrieved more than once, each of these appearing two or three times. A BLAST search analysis revealed that 16 of the 28 phylotype groups have a nearest relative in the GenBank database with sequence identity of 97% or greater. In almost all cases, the nearest relative was an uncultivated cyanobacterial sequence. The sources of these nearest relatives were wide ranging environments that included hot and cold deserts as well as other cryptoendolithic sites.
The two largest phylotype groups dominating the clone library were phylotypes C (12 clones) and E (13 clones+1 DGGE band). These two phylotype groups were 98–99% identical to uncultivated cyanobacterial clone sequences originating from soil crusts of the Colorado Plateau (Redfield et al., 2002; Yeager et al., 2004). Group C included sequences from Gardiner Lower Terraces (GLT), Narrow Gauge Upper Terrace and Painted Pool Terrace (PAP). Group E included sequences from the same sites in addition to Narrow Gauge Ancient Terrace (NGA). Another major phylotype was group I (6 clones, 1 DGGE band, 1 culture isolate), which was distantly related to a Plectonema sp. F3 strain (94% identity) and therefore may represent a novel cyanobacterial lineage. Interestingly, group I is a cluster composed entirely of sequences from Narrow Gauge Lower Terrace (NGL).
In addition to cyanobacterial sequences, our clone library recovered sequences of representatives of moss and green algal lineages. Phylotype group G (4 clones, 1 DGGE band) is 95% identical to the plastid sequence of the green alga Koliella sempervirens. All of the sequences in this group were recovered from the Hoodoos (HDO). Mosses (epilithic) are also represented in the clone library by phylotype group H (5 clones, 1 DGGE band), which shares 99% identity to a plastid sequence of the moss Physcomitrella patens. Sequences in this phylotype group originated from both the Hoodoos (HDO) and the Gardiner Lower Terrace (GLT).
Isolation of clonal cultures
In addition to molecular methods, culture isolation was used to identify members of the phototrophic endolithic communities. A straightforward approach of dilution to extinction and direct enrichment in BG11 medium (or ND medium) was used to isolate cyanobacterial strains from each of the sites. The most apparent result of cultivation is that the majority of isolates are filamentous cyanobacteria. Subsequent to clonal isolation, the 16S rRNA gene sequence was determined for 36 isolates of cyanobacteria and three green algal isolates (Table 2). Of the 36 cyanobacterial culture sequences, 27 are unique; the rest represent identical isolates that were retrieved more than once. Duplicate isolates are included in Table 2 to indicate the frequency of retrieval. The Table also includes four culture isolates from three additional endolithic sites in Yellowstone National Park for which clone library analyses were not performed.
Table 2. Culture isolates retrieved from cryptoendolithic travertine communities
Nearest GenBank relative
Sequences in bold belong to phylotype clusters that include clones and/or denaturing gradient gel electrophoresis bands.
In general, the cyanobacterial and green algal isolates have high similarity to other cultivated strains. For the 27 unique cyanobacterial isolates, 14 have nearest BLAST search relatives of 97–100% identity to cultivated cyanobacterial strains. The sequence of one isolate from Narrow Gauge Lower Terrace (NGL) was 100% identical to the sequence of Cyanobium gracile, a unicellular cyanobacterium with a global distribution in freshwater lakes and brackish seas (Crosbie et al., 2003). Of the other 13 strains with close cultivated relatives, 10 are related to species of the genus Leptolyngbya and three to Nostoc species. Only two of the cultures, CCMEE 6048 and CCMEE 6034, had nearest relatives that were environmental clones. Seven of the unique culture sequences have 94% or less identity to database sequences and may represent novel lineages. These cultures encompassed the less common morphotypes cultivated in this study, including the genera Synechococcus, Chroococcidiopsis, Gloeocapsa, and Schizothrix. In addition to cyanobacteria, we also retrieved two green algal strains: CCMEE 6027, with 99% identity to Bracteacoccus aerius, and CCMEE 6038/6039, with 99% identity to Paradoxia multiseta.
The majority of the culture sequences could not be assigned to phylotype groups. The 16S rRNA gene sequences from the culture isolates that could be assigned to phylotype groups from environmental clones or DGGE bands are included in Table 1 and are also listed in bold in Table 2. Phylotype group J includes the most culture isolates, with seven isolates retrieved from the Narrow Gauge Lower Terrace, Hoodoos, and Narrow Gauge Upper Terrace. Phylotype group J also includes two environmental clones from the Narrow Gauge Lower Terrace; one with 100% identity to isolates CCMEE 6004/6007. These sequences are 99% identical to Leptolyngbya sp. CNP1-Z1-C2, and microscope observations confirm a Leptolyngbya morphotype for these strains (Fig. 2). Phylotype I is the only other group with overlap between environmental clones and cultures. Within this group the culture CCMEE 6011 has 100% identity to two environmental clone sequences and a DGGE band. This phylotype group is a distant relative (94% identity) of the filamentous cyanobacterium Plectonema sp. F3. There were 23 other culture isolate sequences that were unique and did not fall into phylotype groups composed of multiple sequences.
Figure 2 displays photomicrographs of selected culture isolates that are included in Table 2 and in the phylogenetic tree (Fig. 3). The nearest GenBank relative is indicated in Table 2. The cyanobacteria shown in panels a, b, c, and f have close morphological similarities to the genera to which they are most closely related by sequence identity [i.e. Cyanobium (CCMEE 6012), Nostoc (CCMEE 6104, CCMEE 6110) and Gloeocapsa (CCMEE 6058)]. The filamentous cyanobacteria shown in panels g, i, and j all resemble their closest phylotypes with 97–99% identities, the artificial ‘form-genus’Leptolyngbya (Fig. 2). The Leptolyngbya-type in panel h (CCMEE 6132), however, has only a 94% identity to a loosely named member of the order Oscillatoriales (which includes Leptolyngbya) (Table 2, Fig. 3). Panel e displays a unicellular (or aggregated unicellular) type with some similarity to the morphotype Synechocystis but with only a 93% identity to the nearest relative, the unicellular genus Cyanothece (Table 2). The cultures shown in panels d and k are anomalous with respect to nearest relatives. Culture CCMEE 6048 (panel d) resembles a Chroococcidiopsis morphologically but does not have a close identity to the previously sequenced members of this genus (Table 2, Fig. 3). The unicellular rods shown in panel k resemble the artificial ‘form-genus’Synechococcus, but the closest relative (92% identity) is a well-known marine, filamentous cyanobacterium (Table 2). However, identities below 97–98% have little meaning even at the genus level, and reflect the large gaps in the database at the present time.
Phylogenetic analysis of molecular and culture sequences
A phylogenetic tree constructed by the neighbor-joining method is shown, using the sequence data from the environmental clones, DGGE bands and culture isolates (Fig. 3). Although the basal branching order of the tree is mostly unsupported, the sequences from this study generally fell into well-supported sequence clusters at the tips of these branches and bootstrap values are given where relevant. The sequences are distributed in 15 distinct sequence clusters or cyanobacterial lineages. Phylogenetic trees were also constructed by Parsimony methods, and although the basal branching order differed considerably, no differences were observed in the major sequence clusters (data not shown).
Clusters I and II are novel cyanobacterial lineages that comprise the majority of the environmental clone sequences. Cluster I is the largest, with 45% of the environmental clone sequences, including the two largest phylotype groups. There is only one culture isolate sequence (CCMEE 6048) that belongs to this group. Morphologically this culture resembles a Chroococcidiopsis, but it clearly does not fall into a clade with other known Chroococcidiopsis isolates. Cluster I includes environmental clone sequences from every sample location except Narrow Gauge Lower Terrace (NGL). An additional five environmental clone sequences from this study belong to Cluster II. Clusters I and II are comprised entirely of sequences from this study or GenBank sequences of environmental clones from desert soil crusts; they have no close relatives that are cultivated cyanobacterial species. Cluster VII is another group without close cultivated relatives. This cluster is composed of eight environmental clone sequences and one DGGE band. The major phylotype group in this cluster is 99% identical to a DGGE band retrieved from a cryptoendolithic environment of dolomite in Switzerland. Clusters V and VI are small groups of unique environmental clone sequences that are distantly related to Microcoleus-like cyanobacteria.
There are four clusters of Leptolyngbya-like sequences, based on morphology, sequence similarity or both. These four clusters include a majority of the culture isolates from this study. Cluster IX is composed entirely of culture isolate sequences related to Leptolyngbya-like cultures and Phormidium autumnale. Cluster X is a novel lineage that includes one DGGE band and four culture isolates for which the closest relative is an uncultivated Antarctic microbial mat clone. Clusters XIII and XIV are two more distinct Leptolyngbya-like clusters that each include a phylotype group composed of sequences retrieved by both culture and molecular methods. Cluster III is a well-supported clade of heterocystous type sequences that include four culture isolates of Nostoc-like morphology and one DGGE band.
A few of the 16S rRNA gene sequences do not fall into distinct clusters. Members of Phylotype group K are closely related (99% identity) to an environmental clone sequence from a petroleum-polluted microbial mat, but have no close cultivated relatives. Two culture isolates from this study, CCMEE 6109 and CCMEE 6120, also lack close relatives and represent completely novel culture isolates. Culture CCMEE 6109 is morphologically similar to Synechococcus (Fig. 2, panel K), and culture CCMEE 6120 has a Schizothrix-like morphology.
Advantages of a multifaceted approach
This study has provided the first multifaceted (polyphasic) survey of the photosynthetic, travertine-inhabiting endoliths in a temperate climate and laid the groundwork for future environmental and physiological work on these extreme phototrophs and their habitats. This study includes whole community and culture molecular analyses, as well as morphological information on culture isolates. Molecular methods, such as the cloning of environmental DNA sequences, are now generally regarded as providing a more accurate description of microbial community diversity than culture-based methods. But although molecular methods are extremely useful tools for genetic inventories, they provide limited information about the physiology or ecology of community members and so a more ideal tactic is to use a combination of approaches. The methods employed in this study retrieved different but complementary information about the variety of phototrophs in the travertine habitat. Molecular methods allowed us to propose that unicellular cyanobacteria may be the most abundant type of photosynthetic endoliths in this community, information that will aid the development of methods used in future culture isolation attempts. It is important to note that there is no close similarity at the 16S rRNA gene level between the photosynthetic microorganisms of these communities and those that occur in active travertine-depositing hot springs of the Mammoth area of Yellowstone National Park (Fouke et al., 2003; Pentecost, 2003). None of the sequences that we obtained from the cultures or travertine came close to matching the sequences retrieved from the active springs.
Comments on the methods used
DGGE was found to be of limited usefulness in this particular study because, rather than being dominated by a few sequence types, the DGGE profiles were composed of many unresolved bands, making excision and sequencing almost impossible. This is not an uncommon issue and recent literature documents the limitations of DGGE (Liu & Stahl, 2002). Therefore, we relied more on environmental clone library construction for our molecular survey. Almost all of the sequences retrieved in our environmental clone library analysis were closely related to other environmental clones from analogous environments, rather than to cultivated strains. There are a few reasons for the lack of database sequences of cultured strains of cyanobacteria from environments where desiccation and freezing tolerance are important for survival. Relatively few habitats of this type have been surveyed outside of polar or hot desert regions, and most of the recent studies have been limited to molecular analyses of bulk DNA from environmental samples [e.g. (Redfield et al., 2002; de la Torre et al., 2003)]. Another reason is the apparent difficulty of cultivation of some types. In this study we were only able to isolate a single culture belonging to sequence cluster I, which contained 45% of the environmental cyanobacterial sequences. Morphologically this isolate appears to be similar to Chroococcidiopsis, but its phylogenetic placement is clearly outside the clade of other Chroococcidiopsis strains (Fig. 3, Panel d). Thus, morphological information alone may be misleading, but the combination of genetic and morphological data may be of great help in clarifying taxonomic units.
Culture isolates from this study also complemented information retrieved by molecular methods. Using standard cyanobacterial culturing techniques (Castenholz, 1988), we isolated over 100 strains of cyanobacteria. Of the 36 isolates that have been sequenced, 27 are unique strains of cyanobacteria, four of which belong to phylotype groups defined by our clone library sequences. Although there was some overlap between cultures and environmental sequences, many of the cultures fall into separate sequence clusters apart from the environmental sequences. A majority of these cultures are filamentous cyanobacteria, which are typically easier to isolate, and, therefore, more frequently obtained than unicellular types. Culture isolates CCMEE 6004, 6007, 6133, 6119, 6125 and 6054 had 98–99% 16S rRNA gene identity to cyanobacterial strains or sequences originating from North American desert environments, and the 16S rRNA gene sequence of culture CCMEE 6052 is 99% identical to a DGGE band sequence originating from a hot spring source elsewhere in Yellowstone National Park (Black Sand Pool). Two of the Leptolyngbya sequence clusters, X and XIII, include relatives that are environmental clone sequences from Antarctica. Many of the culture sequences from this study are most closely related (but not identical) to sequences obtained from other extreme environments, including deserts. Little is known about dissemination of cyanobacteria, particularly from one continent to another. However, a convincing case of endemism has been demonstrated in thermophilic cyanobacteria from disjunct transoceanic hot springs (Papke et al., 2003). The method of dispersal of the endolithic microorganisms of our study is unknown, but it is possible that exfoliation of rock surfaces that exposes the microbial layer with subsequent dissemination by wind or insects may result in inoculation of other surfaces (e.g. Friedmann & Weed, 1987; Sun & Friedmann, 1999), but a gradual lateral spread within the rock may also occur (Van Thielen & Garbary, 1999).
Problems using 16S rRNA gene sequence similarities for species identification
It is also important to note that, although 16S rRNA gene sequences may show close relationships (i.e. the sharing of a common ancestor), similarities do not discriminate taxa at the species level and extended periods of evolution may still separate cyanobacteria that have 98–100% identities of 16S rRNA gene sequences. In the present study, an excellent example is that of culture CCMEE 6012 for which a BLAST search indicates 100% 16S rRNA gene identity with a few sequenced strains in the database that have been identified as Cyanobium gracile (Table 2). The fact that the other strains were isolated from lakes (Germany, Hungary) and the Baltic Sea would indicate that the genomes of these strains must differ greatly, at least in the genes related to the obvious ecological and physiological disparities. Two strains of the marine pico-cyanobacterium, Prochlorococcus marinus have 16S rRNA gene sequence identities of over 97% but the complete genomes of both show enormous differences, some of which equate with the different light regime preferences for these two strains (Rocap et al., 2003).
Comparisons with other endolithic microbial communities
Endolithic communities in hot deserts tend to be dominated by unicellular cyanobacteria (Whitton & Potts, 2000). However, cold deserts of Antarctica have endolithic habitats dominated in some cases by eukaryotes (e.g. green algae and protolichens) and in others by cyanobacteria, many of which are unicellular (Van Thielen & Garbary, 1999; de la Torre et al., 2003). de la Torre et al. (2003) found that an environmental clone library derived from one of these Antarctic cryptoendolithic communities (one of the McMurdo Dry Valleys) was, in fact, dominated by a single cyanobacterial phylotype. Cultivation of a cyanobacterial strain with 100% 16S rRNA gene identity to this phylotype revealed that it had a coccoid morphology. The limited information that exists at present would suggest that temperate endolithic habitats tend to have a greater diversity of taxa than those of hot and cold deserts. Surveys of endolithic communities in dolomitic rocks of Switzerland (Sigler et al., 2003) and limestones of the Niagara Escarpment (Gerrath et al., 1995, 2000) in Canada recovered a wide range of cyanobacterial genera, including both filamentous and unicellular types. In both of these studies the green algae Chlorella and Stichococcus made a major contribution to the community structure, and yellow-green algae also occurred in the Niagara limestones. Endolithic communities have also been found in moist acidic geothermal environments (Gross et al., 1998; Walker et al., 2005), in which case, the only phototrophs of the community were members of the acidophilic red algal order Cyanidiales. In contrast, the travertine cryptoendolithic community of Yellowstone was dominated by cyanobacteria, and few green algae were retrieved in cultures or clone libraries. Although our clone library pool was small, it was evident that a similar distribution of phylotypes was retrieved from all of the sampling sites except Narrow Gauge Lower Terrace (NGL), which did not share any phylotypes with other sites. This area is the most recently formed travertine (3–4 years) and is also much softer, and more porous than in the other locations, properties that may allow longer retention of moisture and therefore may be inhabited by less desiccation-tolerant strains. In the travertine system overall, a broad range of cyanobacterial genera are present. This suggests that adaptations to desiccation and freezing tolerance are not limited to a few cyanobacterial genera, but rather arose convergently or, alternatively, early in the cyanobacterial radiation.
A future goal of this research is to describe in some detail the physical and chemical properties of selected ancient travertines of Yellowstone National Park, measurements of rock porosity, water retention, nutrient composition, and light attenuation. In addition, the physiology of desiccation- and freezing-tolerance of cultured community members represents an ultimate objective. However, the tolerances of many culture isolates of this study to desiccation stress at low relative humidity (equivalent to summer values in Yellowstone National Park), to high temperature, to freezing, and to various salinities have already been assayed in a preliminary study. Results indicate enhanced tolerances of travertine isolates to these stresses compared to tolerance levels in most cyanobacteria from aquatic or moister habitats (Norris and Castenholz, unpublished data). The importance of continuing to study endolithic environments and their microbial communities is underscored by their relevance as analogues for life on early terrestrial Earth and for possible past or present life on Mars.
This work was performed while T.B.N. held a National Research Council Research Associateship Award at the University of Oregon in affiliation with NASA-Ames Research Center. Special thanks go to undergraduate Pamela N. Johnston who contributed greatly to the culture isolation work and to undergraduate Julie Fox who assayed tolerances to desiccation and other stresses. This work was supported by a research grant from the Thermal Biology Institute at Montana State University to T.B.N. and by a NASA/ARC Cooperative grant NCC 2-5524 to R.W.C. We would like to thank the Yellowstone Center for Resources, Yellowstone National Park, for permission to collect within the park (Permit No. 0185). R.W.C. is a member of the NASA Astrobiology institute.