Introduction

It is generally believed that all the terrestrial and aquatic environments are based on energy and organic carbon that ultimately originated from photosynthesis, that is, by converting light energy into chemical energy used to fix carbon dioxide from the atmosphere. One of the few exceptions are deep-sea hydrothermal vents (reviewed by Nakagawa and Takai, 2008), in which chemoautotrophic bacteria that gain energy from the oxidation of chemicals such as sulfur and hydrogen are the primary producers. Other exceptions are cave ecosystems, such as Movile Cave (Romania). Unlike deep-sea hydrothermal vents, Movile Cave is located a few kilometers from the Black Sea coast. It is a unique groundwater ecosystem in that: (a) the atmosphere is rich in hydrogen sulfide, carbon dioxide and a substantial amount of methane; (b) the redox potential of the cave water (pH=7.5) is low and it contains reduced compounds such as sulfide (0.2–0.3 mM), ammonium (0.2–0.3 mM) and methane (0.02 mM) (Sarbu, 2000). The cave is isolated from the outside world and earlier studies have implied that this unique ecosystem is exclusively based on chemosynthesis, mainly by sulfur oxidation and methane oxidation (Sarbu et al., 1996; Sarbu, 2000).

Movile Cave is unusual for its high productivity and fauna richness. Characterization of invertebrates has been carried out and 48 terrestrial and aquatic species have been identified, of which 33 are endemic (Sarbu, 2000). However, microbial diversity in this cave has not been studied in detail. Earlier microscopic observations have suggested that filamentous sulfur oxidizers and fungi were abundant in the microbial mat (Sarbu et al., 1994). Later, Vlasceanu et al., (1997) isolated a Thiobacillus thioparus strain, which is particularly abundant in the microbial mat and the activity of ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO), a key enzyme in CO2 fixation for many chemoautotrophs, has been detected for this organism (Sarbu et al., 1994). Although sulfur oxidizers are thought to be the major primary producers in Movile Cave (Sarbu et al., 1996; Rohwerder et al., 2003), the presence of up to 1.5% (v/v) methane in the cave atmosphere suggested that methanotrophs may also be present. Hutchens et al., (2004) used DNA stable-isotope probing (DNA-SIP) to analyze the methanotroph populations and a diverse range of methanotrophs belonging to Alphaproteobacteria and Gammaproteobacteria were identified.

In this study, culture-independent analyses using 16S rRNA gene sequences for bacteria and archaea were carried out to determine the microbial diversity present in this unique cave system. Analyses were also carried out using the RuBisCO gene as a functional gene marker for many chemoautotrophs, soxB (sulfate thioesterase/ thiohydrolase) for sulfur oxidizers, which have the Sox-dependent sulfur oxidation pathway and amoA (ammonia monooxygenase) for ammonia oxidizers. SIP incubations with 13C-bicarbonate were also carried out to detect active CO2-fixing bacteria in the cave.

Materials and methods

Environmental sampling and DNA extraction

Movile Cave water and mat samples were collected in September 2007. The pH of the water was 7.5. The samples were kept at 4 °C until they were shipped to the University of Warwick (UK) for further processing. A total of 50 ml water samples (which includes ∼1 g floating microbial mat), well mixed by vortexing, were pelleted by centrifugation, and DNA was extracted using a method described earlier (Neufeld et al., 2007), and then stored at −20 °C for later analyses.

PCR, cloning and restriction length fragment polymorphism analyses

Bacterial 16S rRNA genes were amplified using the 27F/1492R primer set (Lane, 1991). In total, 96 clones were randomly selected from this library and were sequenced using the primer 341F. Archaeal 16S rRNA genes were amplified using a semi-nested approach, with Arch21F/1492R (Delong, 1992) for the first round and Arch21F/Arch958R for the second round (Delong, 1992). The clones were randomly selected and analyzed by RFLP analysis using MspI. Representative clones from each operational taxonomic unit were sequenced. Crude DNA was also used as template to amplify amoA, soxB, form I RuBisCO (green-like and red-like) and form II RuBisCO genes. The primers used are listed in Table 1. In total, 48 clones from each library were randomly selected and sequenced using the M13F/M13R primers. All PCR products were cloned into the pCRTOPO2.1 plasmid using the Invitrogen TOPO TA Cloning Kit (Invitrogen, Paisley, UK).

Table 1 Oligonucleotides used in this study
Full size table

SIP

Time-course SIP incubations were carried out. In total, 20 ml of well-mixed cave water and mat sample were put in 120 ml serum vials and incubated for 1 and 3 weeks, respectively, with 13C sodium bicarbonate (Cambridge Isotope Laboratories, Hook, UK, 13C>99%). A control was set up for each microcosm in exactly the same way except that 12C sodium bicarbonate was used. Sodium bicarbonate was added to a final concentration of 0.5 mM at day 1 of each week for each microcosm. The headspace contained regular air and no attempts were made to adjust the headspace air to that of in situ conditions. Owing to the production of 12C-CO2 from heterotrophs present in the samples (data not shown), the incorporation of 13C carbon was not measured. After incubation, cells and mat material were pelleted by centrifugation and DNA was extracted according to the protocol described earlier (Neufeld et al., 2007). Ultracentrifugation, fractionation and DNA precipitation were carried out according to the procedure described by Chen et al., (2008). The ‘heavy’ 13C-labelled DNA from the 1-week incubation (time point 1) and 3-week incubations (time point 2) was used as template to amplify 16S rRNA genes from bacteria. A clone library for amoA of bacteria was also constructed using the ‘heavy’ DNA from 3-week incubations. A total of 48 clones from each library were randomly picked up and sequenced.

DNA sequencing and phylogenetic analyses

DNA sequencing was carried by cycle sequencing with the BigDye Terminator Cycle Sequencing Kit (PE Applied Biosystems, Warrington, UK) using a 3730A automated sequencing system (PE Applied Biosystems). Bacterial and Archaeal 16S rRNA gene sequences were aligned using the ARB program (Ludwig et al., 2004) and functional genes were aligned using the ClustalX program (Thompson et al., 1997). Phylogenetic and bootstrap analyses were carried out using the Phylowin program (Galtier et al., 1996). Nucleotide sequences from this study were deposited in the GenBank database under the accession numbers FJ604604—FJ604848>.

Results

Microbial diversity in Movile Cave as assessed by 16S rRNA gene clone libraries

To investigate the microbial diversity in Movile Cave, a 16S rRNA clone library for bacteria and archaea, respectively, was constructed and 96 clones were randomly selected and analyzed. The results are summarized in Table 2 and Figures 1, 2 and 3.

Table 2 Classification by the RDP Classifier of 16S rRNA genes retrieved from Movile Cave crude DNA and ‘heavy’ DNA from 13CO2–SIP incubations
Full size table
Figure 1
figure 1

Neighbor-joining phylogenetic tree of bacterial 16S rRNA sequences related to known sulfur-oxidizing and sulfur-reducing bacteria. Sequences were aligned using the ARB program and the tree was constructed by Phylowin (556 bp). Bootstrap values are indicated as blank circles (<50%), gray circles (50–75%) and black circles (>75%). Sequences in bold were retrieved from the crude DNA; sequences in bold and italic were retrieved from ‘heavy’ DNA of the 1-week incubation with 13C-bicarbonate; sequences in bold and underlined were retrieved from ‘heavy’ DNA of the 3-week incubation with 13C-bicarbonate.

Full size image
Figure 2
figure 2

Neighbor-joining phylogenetic tree of bacterial 16S rRNA sequences related to known ammonium- and nitrite-oxidizing bacteria and denitrifying bacteria. Sequences were aligned using the ARB program and the tree was constructed by Phylowin (672 bp). Bootstrap values are indicated as blank circles (<50%), gray circles (50–75%) and black circles (>75%). Sequences in bold were retrieved from the crude DNA; sequences in bold and italic were retrieved from ‘heavy’ DNA of the 1-week incubation with 13C-bicarbonate; sequences in bold and underlined were retrieved from ‘heavy’ DNA of the 3-week incubation with 13C-bicarbonate.

Full size image
Figure 3
figure 3

Neighbor-joining phylogenetic tree of archaeal 16S rRNA gene sequences (779 bp) retrieved from Movile Cave crude DNA. Sequences were aligned using the ARB program and the tree was constructed by Phylowin. Bootstrap values are indicated as blank circles (<50%), gray circles (50–75%) and black circles (>75%). Sequences retrieved from the crude DNA are highlighted in bold.

Full size image

16S rRNA sequences related to known sulfur-oxidizing bacteria were detected in the crude DNA (Figure 1). These were classified into Betaproteobacteria (Thiobacillus, three clones), Gammaproteobacteria (Thiovirga, seven clones; Thiothrix, one clone; Thioploca, 17 clones) and Epsilonproteobacteria (Sulfuricurvum, one clone). Sequences related to sulfur reducers (Desulfobulbaceae) and denitrifiers (Denitratisoma) were also found (Figure 2). No sequences related to known methanotrophs were detected in the crude DNA. However, a number of 16S rRNA gene sequences were associated with known methylotrophs within the Betaproteobacteria (Table 2), including Methylotenera (15 clones), Methylophilus (three clones) and Methylovorus (three clones). Other sequences detected include Verrucomicrobia (one clone), Firmicutes (two clones), Bacteroidetes (15 clones), Alphaproteobacteria from the Rhodobacterales (two clones) and Gammaproteobacteria from the Xanthomonadales (seven clones).

The archaeal 16S rRNA genes sequences were clustered into two major groups (Figure 3). In total, 48 16S rRNA gene clones were classified in the DHVE1 (Deep-sea Hydrothermal Vent Euryarchaeota; Takai and Horikoshi, 1999) group. A total of 42 clones were associated with the MCG (Miscellaneous Crenarchaeotic group; Inagaki et al., 2003) however, these only showed moderate identity (∼92%) to the 16S rRNA gene sequences from MCG. Among those, an uncultured Crenarchaeota clone (B10), which was retrieved from the water column of a sulfurous karstic lake in Spain, was the closest BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) match (Lliros et al., 2008). It is to be noted that no sequences related to known sulfur-oxidizing archaea, ammonia-oxidizing archaea, methanogens or anaerobic methane-oxidizing archaea were detected in this 16S rRNA gene clone library.

Diversity of RuBisCO genes

The diversity of RuBisCO genes was analyzed using three separate primer sets targeting form I green-like, form I red-like and form II RuBisCO genes. Results are shown in Figure 4. It is to be noted that all of the sequences from the form I RuBisCO (red type) clone library showed <90% identity to sequences in the GenBank database and they formed two separate groups. In total, 30 clones were related (∼88% identity) to the RuBisCO genes from Rhodobacter azotoformans and Labrenzia aggregate. The other two sequences were related (84% identity) to the RuBisCO gene from Methylibium petroleiphilum.

Figure 4
figure 4

Neighbor-joining phylogenetic trees of form I red-like (430 bp) ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) (a), form I green-like (571 bp) RuBisCO (b) and form II RuBisCO (238 bp) (c) genes retrieved from Movile Cave crude DNA. Sequences were aligned using the ClustalX program and the trees were constructed by Phylowin. Bootstrap values are indicated as blank circles (<50%), gray circles (50–75%) and black circles (>75%). Sequences retrieved from the crude DNA are highlighted in bold.

Full size image

Similarly, all the clones from the form I green type RuBisCO clone library showed <90% identity to sequences in the GenBank database. They were reasonably closely related (85–88% identity) to the RuBisCO gene sequences from Thiobacillus denitrificans and T. thioparus.

The sequences from the form II RuBisCO clone library fell into two clades. In total, 11 clones had 92% identity to the form II RuBisCO gene from T. denitrificans and the other 25 clones were most closely related (97% identity) to the RuBisCO gene from Halothiobacillus neapolitanus, but they clearly formed a separate branch as supported by bootstrap analyses.

Diversity of soxB and amoA genes

On phylogenetic analysis, it was obvious that the soxB genes retrieved were distributed throughout the Alphaproteobacteria, Betaproteobacteria and Gammaproteobacteria (Figure 5). Within the Betaproteobacteria, four clones were closely related to Thiobacillus and two clones were related to the soxB homolog from M. petroleiphilum. Within the Gammaproteobacteria, five clones were related to soxB genes from Thiothrix and the other 10 clones showed only 65% identity to soxB genes from Halothiobacillus species.

Figure 5
figure 5

Neighbor-joining phylogenetic tree of soxB (sulfate thioesterase/thiohydrolase) gene sequences (412 bp) retrieved from Movile Cave crude DNA. Sequences were aligned using the ClustalX program and the tree was constructed by Phylowin. Bootstrap values are indicated as blank circles (<50%), gray circles (50–75%) and black circles (>75%). Sequences retrieved from the crude DNA are highlighted in bold.

Full size image

amoA genes from bacteria formed two separate clades: two clones were within the Nitrosomonas europaea clade and the other 45 clones grouped within the Nitrosomonas oligotropha clade (Figure 6). Using crude DNA as template with primers for amoA genes from archaea did not yield any PCR products of the expected sizes.

Figure 6
figure 6

Neighbor-joining phylogenetic tree of bacterial amoA gene sequences (328 bp) retrieved from Movile Cave crude DNA and ‘heavy’ DNA from the 3-week stable-isotope probing (SIP) incubation with 13C-bicarbonate. Sequences were aligned using the ClustalX program and the tree was constructed by Phylowin. Bootstrap values are indicated as blank circles (<50%), gray circles (50–75%) and black circles (>75%). Sequences retrieved from the crude DNA are highlighted in bold and those in bold and underlined were retrieved from ‘heavy’ DNA of the 3-week incubation with 13C-bicarbonate.

Full size image

SIP analyses using 13C-bicarbonate

To investigate microorganisms which were active in assimilating CO2 in Movile Cave, SIP incubations were carried out using 13C-bicarbonate. A bacterial 16S rRNA clone library was constructed using the ‘heavy’ DNA extracted after a 1-week incubation and a 3-week incubation, respectively. The results are presented in Table 2 and Figure 1.

After 1 week of incubation with 13C-bicarbonate, it was found that sulfur-oxidizing bacteria from both the Betaproteobacteria (Thiobacillus, three clones; Thiobacter, eight clones) and Gammaproteobacteria (Thioploca, one clone; unclassified Thiotrichaceae, two clones) were labeled with 13C. Sulfuricurvum-related 16S rRNA gene sequences were not found, although they were found in the 16S rRNA gene library constructed with crude DNA. It is interesting to note that 16S rRNA gene sequences related to ammonia-oxidizing bacteria (Nitrosomonas, one clone) and nitrite-oxidizing bacteria (Nitrospira, two clones) were present after 1-week incubation with 13C-bicarbonate. After 3 weeks of incubation with 13C-bicarbonate, sulfur-oxidizing bacteria were still present, including Thiobacillus (four clones), Thiovirga (one clone). Four 16S rRNA clones were associated with the ammonia-oxidizing bacterium Nitrosomonas and 13 clones showed high identity to the 16S rRNA from Candidatus ‘Nitrotoga arctica’, a novel nitrite-oxidizing bacterium isolated from the Siberian Arctic (Alawi et al., 2007).

A number of 16S rRNA sequences in the ‘heavy’ DNA clone library were associated with the Alphaproteobacteria, including Rhodospirillales, Rhodobacterales and Sphingomonadales. A few sequences related to Bacteroidetes, Planctomycetes and Spirochetes were also found in the ‘heavy’ DNA from both the 1-week and the 3-week incubations with 13C-bicarbonate. Sequences related to Aquimonas were also found in the DNA from the 3-week SIP incubation microcosm.

A clone library of amoA genes from bacteria was constructed using the ‘heavy’ DNA from the 3-week incubation with 13C-bicarbonate. In total, 26 amoA clones clustered in the N. europaea amoA clade and the other 21 clones were associated with the N. oligotropha clade. It was not possible to amplify amoA genes from archaea using the ‘heavy’ DNA as template.

Discussion

It has been hypothesized earlier that the Movile Cave system is driven by chemoautotrophy (Sarbu and Kane, 1995; Sarbu et al., 1996; Sarbu, 2000). The cave continuously receives a high flow rate of hydrothermal water rich in sulfide and methane (Sarbu, 2000) and sulfur oxidation was thought to be the major energy source for primary production (Sarbu et al., 1996). Earlier analyses showed that aerobic methanotrophs were also present in this cave, suggesting that methanotrophy could also be one of the major primary production processes (Hutchens et al., 2004; Sarbu, 2000). Although the cave water contains high concentrations of ammonium (0.28 mM, Sarbu, 2000), nitrification as a major energy production route has not been investigated earlier. It has, however, been observed that nitrogen in the microbial mat is isotopically lighter (δ15N=−9.1‰) than ammonium–nitrogen in the cave water (δ15N=+19.9‰) (Sarbu et al., 1996). Lack of further evidence made it difficult to ascertain whether ammonium uptake by microbes or nitrification was the reason for the observed lighter nitrogen isotope content in the microbial mat. Our DNA-SIP experiment using 13C-bicarbonate indicated that ammonia- and nitrite oxidizers were particularly active in the assimilation of 13C-labelled CO2 during microcosm incubations, indicating that nitrification could at least partially account for the shift of nitrogen isotope content.

Further molecular-based analyses using 16S rRNA and amoA genes showed that Nitrosomonas spp. were responsible for ammonia oxidation and Nitrospira and Candidatus ‘Nitrotoga’ were probably the major nitrite utilizers. The community of Nitrosomonas in the crude DNA is predominated by bacteria of the N. oligotropha clade, whereas more amoA gene sequences from the N. europaea clade were seen during SIP incubations (Figure 6). Studies have shown that ammonia availability could select against different Nitrosomonas clades. For example, by studying the ammonia-oxidizing bacterial community in aeration sequencing batch reactors, Otawa et al., (2006) have observed that the N. europaea clade were dominant in reactors in which ammonium concentration was higher, whereas bacteria of the N. oligotropha clade were dominant in those in which the ammonium concentration was lower. It is also possible that the toxicity of nitrite may affect the relative distribution of these two clades. For example, Limpiyakorn et al., (2007) observed that the N. oligotropha clade comprised the majority of ammonia-oxidizing bacterial populations when nitrite did not accumulate, whereas the N. europaea clade became dominant when concentrations of nitrite were high (>2.1 mM). The toxic effect of nitrite may explain the increase in the relative abundance of the N. europaea clade during SIP incubations. The relative change of ammonium concentration during SIP incubations is probably not as dramatic as that of nitrite because of the presence of relatively high concentrations of ammonium (∼0.3 mM) in the Movile Cave water. Temporary accumulation of nitrite may, however, occur during microcosm incubations: this was supported by the appearance of nitrite-oxidizing populations in the ‘heavy’ DNA during 13C-bicarbonate incubations. Clearly, in the future, quantification of ammonium, nitrite and nitrate needs to be followed during SIP incubations to fully understand the nitrogen cycling in Movile Cave and to establish the link between microbial diversity and activity.

The atmosphere in Movile Cave contains a high concentration of H2S (8–12 mg l−1) and highly diverse sulfur oxidizers inhabit the cave water and form a visible microbial mat. Microscopic observations indicated that the floating mat contained Thiothrix- and Beggiatoa-like filamentous bacteria (Sarbu et al., 1994) and subsequently a T. thioparus strain was isolated from Movile Cave water (Vlasceanu et al., 1997). Earlier studies have shown that aerobic and anaerobic sulfur oxidizers are abundant (up to 107 CFU (colony-forming unit) per gram sample) (Rohwerder et al., 2003). Our molecular-based analyses confirmed the presence of filamentous sulfur-oxidizing bacteria related to Thiothrix, Beggiatoa and Thioploca within the Gammaproteobacteria (Figure 1). The 16S rRNA, soxB (Figure 5) and RuBisCO (Figure 4c) sequences analyses also suggest the presence of a novel group of sulfur-oxidizing Gammaproteobacteria, related to Thiovirga and Thiofaba within the Halothiobacillaceae family (Ito et al., 2005; Mori and Suzuki, 2008). They are loosely related to Halothiobacillus spp. in terms of 16S rRNA gene, soxB gene and RuBisCO gene sequence identities. They are also separated from Thiovirga and Thiofaba by their 16S rRNA identity (Figure 1). 16S rRNA gene sequences related to T. denitrificans and T. thioparus within the Betaproteobacteria were also abundant, which has been suggested earlier (Vlasceanu et al., 1997). This group, compared with the filamentous sulfur oxidizers Thiothrix–Thioploca–Beggiatoa, was particularly active during SIP microcosm incubations. The presence of anaerobic sulfur oxidizers in Movile Cave has also been suggested earlier (Rohwerder et al., 2003) and a 16S rRNA sequence related to Sulfuricurvum was detected in our crude DNA clone library, although it was not labeled during SIP incubations, probably because of the fact that the incubation was carried out aerobically, and Sulfuricurvum oxidizes sulfur anaerobically (Kodama and Watanabe, 2004). It is to be noted that no sequences related to known sulfur-oxidizing archaea were observed in the archaeal 16S RNA clone library, which may reflect the fact that the temperature in the cave (20 °C) is constant while all known sulfur-oxidizing archaea to date are thermophilic.

One notable difference in the Movile Cave microbial community compared with other chemoautotrophically based cave systems, such as Lower Kane Cave (USA), is that Epsilonproteobacteria were less abundant in Movile Cave (Engel et al., 2003, 2004). However, as with our study, another chemoautotrophically based cave, Frasassi Cave (Italy), was also dominated by Gammaproteobacteria and Betaproteobacteria rather than Epsilonproteobacteria (Macalady et al., 2006). The most recent research on niche differentiation among sulfur-oxidizing bacteria within Frasassi Cave tends to suggest that Epsilonproteobacteria dominated when sulfide to oxygen ratios were high (Macalady et al., 2008). Clearly, more study of Movile Cave is needed to determine the distribution and niche adaption of sulfur oxidizers in the mat, in different depths of the cave water and in biofilms on the surface of the walls of the cave.

The presence of a high relative abundance of methylotroph 16S rRNA gene sequences in the crude DNA, as revealed by molecular analyses, is surprising given the fact that the cave receives a high flux of methane. It is likely that these methylotrophs feed on methanol produced by and released from methanotrophs that are consuming methane. Earlier studies showed high methane oxidation activity in the cave water and floating mats and a variety of methanotrophs have been detected in this cave (Hutchens et al., 2004). In fact, methylotroph-related 16S rRNA sequences were also observed in the ‘heavy’ DNA from 13C-methane incubations and this has been interpreted as cross-feeding of carbon, presumably methanol or formate, from methanotrophs to methylotrophs (Hutchens et al., 2004). Another possibility is that methanol may come from the degradation of chitin-like substances from cave fauna, as these compounds need to be recycled when cave fauna die, or that the methylotroph sequences that were labeled are in fact from as yet uncultured methanotrophs in the Betaproteobacteria. The methylotrophs were assigned to the Methylophilaceae family (Betaproteobacteria). Sequences related to M. petroleiphilum (a facultative methylotroph in the Betaproteobacteria) were also found in the soxB and RuBisCO clone libraries. An earlier culture-based study also showed that the numbers of culturable methylotrophs in the floating mat were around 106 CFU per gram dry weight of mat (Rohwerder et al., 2003).

Archaeal communities in cave systems, in general, have not been studied in detail. Macalady et al., (2007) showed that the dominant archaea in the wall biofilms from Frasassi Cave (another chemoautotrophy-based cave, but with a low pH) were associated with two groups, namely Ferroplasma and an environmental clone group with sequences from acid mine drainage and Yellowstone hot spring. No Crenarchaeota were found in Frasassi Cave. In contrast, the archaeal community in Movile Cave showed some similarity to deep-sea hydrothermal vent environments (Takai and Horikoshi, 1999), but differed from these in that no 16S rRNA sequences related to known ammonia-oxidizing archaea were found in our study, and Nitrosomonas spp. were probably the major nitrifiers in Movile Cave, whereas it has been suggested that ammonia-oxidizing Crenarchaea, rather than their bacterial counterparts, are the major nitrifiers in deep-sea vents (Lam et al., 2004; Nakagawa and Takai, 2008). Although it is possible that the clone library had bias because of the imperfect 16S rRNA primers for archaea, as well as bias during SIP incubations, our results imply that bacterial chemolithotrophs are probably the dominant primary producers in Movile Cave.

In addition to 16S rRNA analyses, we also used functional marker genes to characterize the microbial diversity in Movile Cave. Great care is needed in the interpretation of the results, as not all chemolithoautotrophs use RuBisCO for CO2 fixation and at least three other CO2 fixation pathways are known (reviewed by Nakagawa and Takai, 2008). Similarly, microorganisms also use a number of different pathways for sulfur oxidation, although the Sox pathway is the most highly characterized (Kelly et al., 1997). Thus, functional gene analyses should always be used in combination with other techniques, such as 16S rRNA gene-based methods. In addition, we did not carry out functional gene analyses of soxB and RuBisCO for ‘heavy’ DNA from SIP incubations because bacteria, which contain these genes, will also be present in ‘light’ DNA. Furthermore, soxB genes also occur in facultative bacteria that could derive most of their cell carbon from multicarbon sources.

In summary, we have studied the microbial diversity in Movile Cave and suggest that ammonia/nitrite-oxidizing bacteria may also contribute to primary productivity in this cave. Further work on the niche distribution of sulfur and ammonia/nitrite oxidizers, as well as the contribution of one-carbon compound oxidation to primary production, is needed for a full understanding of the sulfur, nitrogen and carbon cycles within Movile Cave.