Introduction

Thermophilic mat communities occur in geothermal springs of neutral/alkaline pH and at temperatures of up to ~65°C (Brock 1978). Mat community composition is largely temperature-defined, and mats have been differentiated on the basis of the cyanobacterial taxa involved in primary production. At temperatures above 60°C Synechococcus is the only cyanobacterial genus encountered, and this has been the focus of most research on thermophilic mat diversity (Ferris et al. 1996a, 1996b; Ward et al. 1997, 1998). Studies have generally utilized 16S rRNA gene data from environmental samples to demonstrate that genotypic diversity far exceeds phenotypic diversity as estimated by observation and culture techniques (Ferris et al. 1996a). Although few locations worldwide have been studied, distinct phylogeographic groups have been shown to exist in the continental USA, Japan and New Zealand (Papke et al. 2003).

The moderately hot runoff channels and pools below ~60°C support a range of different cyanobacterial mats comprising filamentous and unicellular taxa and the occurrence of these is determined by temperature plus combined nitrogen and free sulphide levels. Nitrogen-poor waters support mats of diazotrophic cyanobacteria, usually Fischerella or Calothrix and/or Pleurocapsa at lower temperatures, whereas those rich in nitrogen support Synechococcus and Phormidium mats (Ward and Castenholz 2000). In sulphide-rich waters mats comprise Oscillatoria species that are sulphide tolerant/utilizing (Ward and Castenholz 2000). Filamentous mats have been extensively documented morphologically but have received little attention in terms of molecular diversity. Some Oscillatoria sequences are known from cultures (Miller and Castenholz 2001) and environmental DNA (Papke et al. 2003), and a low temperature Synechococcus mat (40–47°C) from Yellowstone National Park was found to support Phormidium, Pseudanabaena and Spirulina-like genotypes in addition to Synechococcus (Norris et al. 2002). It is reasonable to assume, however, that as with Synechococcus mats morphological diversity does not reflect genetic diversity. In addition, little is known about possible phylogeographic patterns among filamentous genotypes. A partial 16S rRNA gene phylogeny for Oscillatoria revealed genotypes from the continental USA, Japan and New Zealand that were similar although some New Zealand genotypes were unique (Papke et al. 2003). The deep branching between lineages that are observed for Synechococcus were, however, absent.

The paucity of molecular data for filamentous thermophilic cyanobacteria as compared to Synechococcus raises some important questions. Do moderately thermophilic cyanobacterial mats harbour greater genetic diversity than morphotypes suggest? Are filamentous thermophilic taxa phylogenetically distinct from their mesophilic counterparts? Are phylogeographic patterns apparent which may yield clues to the evolutionary forces that shape these thermophilic communities? In order to address these questions, at least in part, we set out to examine 16S rRNA gene-defined and morphological diversity in a range of filamentous cyanobacterial mat communities from moderately hot spring waters.

Materials and methods

Sample recovery

Mat samples were collected from geothermal springs in various locations within Asia as indicated in Table 1. For each mat type three samples were taken. Sections (2 × 1 cm) were cut from mats using a scalpel and stored in sterile glass bottles in darkness on ice in the field (<8 h) and then at 4°C until processed (<2 weeks). Mats generally occurred in shallow (2–3 cm depth) water. Temperature and pH in the water channel for each location where mat growth occurred were recorded at several locations over a 5-min period to ensure sampling areas were representative of conditions generally experienced by the mat. The probes/thermometer were also inserted into holes left from biomass sampling, and values were not significantly different from the above. The pH meter (230-A, Orion) was calibrated on-site at sampling temperatures and used with an automatic temperature compensation (ATC) function when sampling. A thermometer was used to verify accuracy of the ATC digital thermometer. Hydrogen sulphide levels were determined titrimetrically using methylene blue (HS-WR, Hach).

Table 1 Moderately thermophilic mats used in this study
Full size table

Microscopy

Identification of cyanobacterial morphotypes was carried out by microscopy using an Olympus BX50 compound microscope with and without Nomarski optics.

Community DNA recovery and PCR

DNA recovery from mat samples was achieved by lysis in CTAB with lysozyme, RNAse A and Proteinase K incubations, and phenol:chloroform extraction at 60°C. Genes of 16S rRNA were amplified by PCR using cyanobacteria-specific primers CYA359F and CYA781R (Nübel et al. 1997) with a (GC)40 clamp added to the forward primer. The following PCR profile was used: 35 cycles of 1 min at 94°C, 50 s at 55°C, 1 min at 72°C; with an initial denaturation step of 3 min and a final extension step of 10 min. The PCR reaction mixture contained 1.5 mM MgCl2, 0.2 mM of each dNTP, 0.3 μM of each primer, and 1.0 U of Taq DNA polymerase in 50 μl total volume. Since this is the first report of a molecular approach to biodiversity assessment of such thermophilic filamentous cyanobacterial mats, we first confirmed specificity of the PCR primers to twenty taxonomically diverse reference taxa from The University of Hong Kong Culture Collection.

DGGE and sequencing

Each PCR amplicon (2 μg amplified DNA) was separated by DGGE (Myers et al. 1988) in a urea/formamide denaturing gradient in 7% acrylamide gel, run at 150 V in 1 × TAE buffer (pH 8) at 60°C (DGGE-2001, CBS Scientific Co.). Bands were excised, soaked overnight in TE buffer (pH 8) at 4°C, re-amplified and purified (GFX, Amersham) prior to automated sequencing (ABI Prism 377, Applied Biosystems). Some bands could not be sequenced directly and were cloned (pDrive, Qiagen) before successful sequencing. All sequences have been deposited in GenBank, under accession numbers AY787591–AY787626.

Phylogenetic analyses

Approximate phylogenetic affiliations were determined by BLAST searches of the NCBI GenBank database. Multiple alignments were then created with reference to selected GenBank sequences using BioEdit version 5.0.9 (Hall 1999). Maximum likelihood analysis using PAUP* 4.0b8 (Swofford 2001) was used to illustrate the relationship of partial 16S rRNA gene sequences to representative cyanobacteria. Bayesian posterior probabilities (Rannala and Yang 1996) and bootstrap values (1000 replications) were calculated and were shown for branches supported by more than 50% of the trees.

Results and discussion

In this study we report 16S rRNA gene-defined community diversity in environmental samples of moderately thermophilic mats from seven different locations within Asia (Table 1). A total of 45 16S rRNA gene-defined genotypes were resolved by visual analysis of DGGE-banding patterns. For all samples studied genotypic diversity was greater than that observed by microscopy (Fig. 1), mirroring trends observed for Synechococcus mats from higher temperatures (Ferris et al. 1996a). Purely filamentous mats generally displayed fewer bands than those containing both filamentous and Synechococcus-like genotypes. Banding patterns were generally consistent between independent replicates from the same mat, although some within-mat variation occurred. This is to be expected when sampling mat communities where different microenvironments can exist within a small area.

Fig. 1
figure 1

DGGE banding patterns of 16S rRNA gene-defined diversity among moderately thermophilic cyanobacterial mats. Thp1, Thp2, Thp3, Pong Dued, Thailand; Phl, Los Banós, Philippines; Ths, Sankhamphaeng, Thailand; Tht, Teppanom, Thailand; Cht, Daggyai Tso, Tibet, China. Percentages refer to urea-formamide gradients employed for each gel. Horizontal lines indicate band migration positions, arrows denote those bands that were successfully sequenced

Full size image

All differentially migrating bands and co-migrating bands from different mat samples were sequenced wherever possible (36 out of 45 bands). Most were sequenced directly from excised bands, but some samples that generated ambiguous sequences were cloned prior to successful sequencing. It is likely that such bands contained mixed genotypes and this highlights a potential limitation to DGGE in community analysis. Nonetheless this can be favourably weighed in this case against the number of bands recovered per sample, which was greater than or similar to that observed for cyanobacteria in similar studies. The only morphotypes not represented in DGGE-derived sequences were the infrequently observed Gloeothece and Pseudanabaena.

All sequenced bands were provisionally identified on the basis of high similarity by BLAST search to cyanobacterial sequences in the NCBI GenBank database (Table 2). All but one Oscillatoria-like sequence (Thp2-1) corresponded to the observed morphotypes for each mat. Sequence divergence between similar genotypes in this study was greater than recorded for intraspecies 16S rRNA gene sequence variation in cyanobacteria (Tourova 2003) and so DGGE-derived genotypes are unlikely to represent multiple 16S rRNA gene copies from a single species. While Fischerella-, Oscillatoria- and Synechococcus-like sequences shared high similarity (95–99%) to published thermophilic cyanobacterial sequences, all other sequences recovered shared relatively low similarity (86–93%), reflecting the lack of previously published thermophilic sequences available for these genera and novel diversity reported here. Bearing in mind that the sequence data from this study is based upon a ~420 bp fragment of the rRNA gene, and that an overall 98% sequence similarity is generally considered to represent the same species, while <88% similarity suggests a new genus (Stackebrandt and Goebel 1994), we can assume that novel diversity at the species level or higher is identified for Calothrix-, Cyanothece-, Phormidium- and Pleurocapsa-like sequences.

Table 2 Identity of sequences obtained from community DNA of moderately thermophilic cyanobacterial mats
Full size table

Phylogenetic analysis was used to further resolve identities and establish relationships between sequences obtained in this study and representative GenBank sequences. Branching patterns within trees generally had high levels of support, with lower bootstrap and Bayesian posterior probability values for a few branches probably reflecting current ambiguities in cyanobacterial taxonomy and sequence length limitations to the analysis. The Fisherella-like sequences grouped closely with unpublished sequences (AY236467–AY236480) from a nearby geothermal pool at the same location. This lineage included thermophilic Fischerella and Mastigocladus (Fischerella) sequences from as far apart as Iceland and the Philippines, and these probably belong to F. muscicola (Fig. 2). The temperature from which Fischerella-like sequences were recovered is within the known range for this genus and Fischerella is known to occur in springs of pH 5 and above (Ward and Castenholz 2000).

Fig. 2
figure 2

Phylogenetic relationships among moderately thermophilic diazotrophic cyanobacteria based upon Maximum Likelihood analysis of partial 16S rRNA gene sequence data. Sequence codes refer to those given in Table 2, where Thp1, Thp2, Thp3, Pong Dued, Thailand; Phl, Los Banós, Philippines; Ths, Sankhamphaeng, Thailand; Tht, Teppanom, Thailand; Cht, Daggyai Tso, Tibet, China. Unrooted trees are supported by Bayesian posterior probabilities (first number) and bootstrap values for 1000 replications (second number), shown for branches supported by more than 50% of the trees. Scale bar represents 0.01 nucleotide changes per position

Full size image

In contrast the Calothrix- and Pleurocapsa-like sequences likely represent novel diversity at the species level or higher, forming distinct and possibly thermophilic lineages within each genus (Fig. 2). Since GenBank sequences available for comparison were all mesophiles from other continents, it is also possible that phylogeographic factors alone may be responsible. The Calothrix- and Pleurocapsa-like sequences were recovered from three locations in Thailand of 42–48°C, which are close to the 35–47°C range recorded for these taxa in other studies (Ward and Castenholz 2000). The range of pH for growth of thermophilic Calothrix and Pleurocapsa is unknown. Calothrix was recorded from only one site and this may reflect the availability of sinter pebbles for colonization, which were absent at all other sites except mat Cht, where temperatures were above the limit for growth of Calothrix. Since Pleurocapsa occurred in mat Ths but not the morphologically similar mats Thp2 (51°C) and Tht (50°C) that were separated by ~100 km but with similar aqueous geochemistry, we may conclude that the upper temperature limit for Pleurocapsa growth probably lies between 48–49°C. This is the first report of thermophilic Calothrix sequence data, and only one Pleurocapsa sequence is previously recorded from a geothermal environment in Japan (Papke et al. 2003). The sequence data for the latter taxon spans a fragment of the 3′ end for the 16S rRNA gene and ITS region and so is not comparable with our data.

Most sequences from the Phormidium mats (Thp2, Ths, Tht) resolved into a phylogenetically uncertain Cyanothece-like group that spanned clades within the Oscillatoriales. The molecular systematics of the Oscillatoriaceae is, however, polyphyletic and not satisfactorily resolved at present (Wilmotte 1994; Litvaitis 2002). Curiously no Phormidium-like sequences were obtained for mats Thp2 and Ths, although sequences from mat Tht resolved unambiguously into a Phormidium lineage (Fig. 3). Further work is required to resolve this particular ambiguity, and sequence analysis of the entire 16S rRNA gene or additional loci may help in this regard. It should be noted, however, that even where near-complete 16S rRNA gene sequences have been used, conflicts between morphological and molecular identification of Phormidium-like sequences have still arisen (Nadeau et al. 2001; de la Torre et al. 2003). On a relatively small spatial scale (~100 km), distinct populations of Cyanothece-like sequences existed within Thailand at geochemically similar sites, indicating that small-scale phylogeographic effects also occur for at least some moderately thermophilic cyanobacteria. Such small-scale variations have been recorded for Synechococcus within north America (Ward et al. 1998).

Fig. 3
figure 3

Phylogenetic relationships among moderately thermophilic Oscillatoriales and Cyanothece-like cyanobacteria based upon Maximum Likelihood analysis of partial 16S rRNA gene sequence data. Sequence codes refer to those given in Table 2, where Thp1, Thp2, Thp3, Pong Dued, Thailand; Phl, Los Banós, Philippines; Ths, Sankhamphaeng, Thailand; Tht, Teppanom, Thailand; Cht, Daggyai Tso, Tibet, China. Unrooted trees are supported by Bayesian posterior probabilities (first number) and bootstrap values for 1000 replications (second number), shown for branches supported by more than 50% of the trees. Scale bar represents 1 nucleotide change per position

Full size image
Fig. 4
figure 4

Phylogenetic relationships among moderately thermophilic Synechococcus based upon Maximum Likelihood analysis of partial 16S rRNA gene sequence data. Sequence codes refer to those given in Table 2, where Thp1, Thp2, Thp3, Pong Dued, Thailand; Phl, Los Banós, Philippines; Ths, Sankhamphaeng, Thailand; Tht, Teppanom, Thailand; Cht, Daggyai Tso, Tibet, China. Unrooted trees are supported by Bayesian posterior probabilities (first number) and bootstrap values for 1000 replications (second number), shown for branches supported by more than 50% of the trees. Scale bar represents 0.01 nucleotide changes per position

Full size image

Oscillatoria-like sequences all shared high similarity to known thermophilic oscillatorians, but formed a distinct phylogenetic group within the thermophilic O. amphigranulata lineage, sharing highest affinity with cultivated thermophilic oscillatorians (Miller and Castenholz 2001). All but one of our Oscillatoria-like sequences were recovered in association with Fischerella-like sequences in free-floating Philippine mats growing in moderately sulphidic conditions. These cyanobacteria are known to be moderately/highly sulphide-tolerant, occur at up to 62°C and at pH 6.5 and above (Ward and Castenholz 2000). A single Oscillatoria-like sequence was also recovered from Thailand (Thp2) in non-sulphidic conditions, but was absent from morphologically similar mats (Ths and Tht). Since all DGGE bands were sequenced for these samples, and assuming the resolution between mats bearing similar morphotypes is comparable, this is probably an accurate reflection of occurrence for this genotype. Oscillatoria-like sequences from Philippines, Thailand and the USA largely grouped independently within the thermophilic Oscillatoria lineage suggesting novel phylogeographic lineages may have arisen in isolation at each location. This view is also supported by analysis of Oscillatoria-like sequences from New Zealand and Japan (Papke et al. 2003). This greater resolution among thermophilic Oscillatoria sequences as compared to other filamentous cyanobacteria in this study may reflect, at least in part, the greater number of known genotypes for this taxon. All sequences recovered in our study were nonetheless phylogenetically distinct and so expand the known limits of this thermophilic group.

The Synechococcus sequences obtained from moderately thermophilic mats were all unique, yet, generally shared high sequence similarity with known Synechococcus from other geothermal environments, although most of these were recovered from hotter habitats than those in our study. Synechococcus is known to occur within the temperature and pH range of mats used in this study. Philippine and Tibetan sequences were most closely affiliated with the C1 lineage, which largely comprises Synechococcus from relatively lower temperatures (as in this study) and a number of sequences from Japanese springs. That the Philippine sequence resolved in this way is understandable bearing in mind the geographic proximity and geochemical similarities between locations, namely that both are sulphidic and of neutral pH, arising from the same fault system. All but one of the Tibetan sequences grouped loosely with C1 lineage sequences, expanding this grouping with at least three novel lineages. Prevailing atmospheric conditions and winds dictate that the most likely dispersal direction would be West–East, and Tibetan geothermal springs have been in continuous existence for longer than most Japanese and Philippine springs, which are associated with recent volcanic activity. We can therefore speculate that the shallow branching Japanese and Philippine sequences may have evolved from Tibetan ancestors. The phylogenetic placement of Synechococcus-like sequences from Thailand is curious since they form a separate group that is most closely affiliated with sequences only previously recorded for geothermal springs in the continental USA, rather than geographically more proximal Asian sites. In addition, a single Tibetan sequence forms a distinct lineage separate from all other A/B and C9 sequences. Although more locations worldwide would need to be studied in order to make robust conclusions, it would appear that distinct phylogeographic groupings independent of temperature and pH do exist, as observed for north American and Japanese springs (Papke et al. 2003). There was no evidence for grouping of Synechococcus-like sequences from four different locations in our study on the basis of pH, although low pH-adapted genotypes distinct from those occurring at pH 6.1 and above (and most similar to our sequences) may exist in north American springs of pH 5 (Ruff-Roberts et al. 1994). The reason why some geochemically similar locations appear to support narrowly defined thermophilic genotypes, while others such as Daggyai Tso have far greater diversity is at present unclear. It may simply relate to greater diversity in predominantly Synechococcus mats compared to mats dominated by other cyanobacterial taxa, but factors such as age of springs and disturbance events may also be a factor. Whether the composition of moderately thermophilic mats is constant over time is currently unknown, although Synechococcus populations are known to be stable within mats over time and recover after disturbance (Ferris and Ward 1997; Ferris et al. 1997; Norris et al. 2002).

In conclusion, we have characterized community molecular diversity of moderately thermophilic cyanobacterial mats from a range of geothermal springs. The 16S rRNA gene-defined diversity of all mats exceeded that observed by microscopy. Genotypes resolved into distinct thermophilic phylogenetic lineages separate from their mesophilic counterparts. Furthermore, some evidence for phylogeographic patterns emerged among moderately thermophilic cyanobacteria, across relatively short regional and also inter-continental distances. These features are in general agreement with observations of more thermophilic Synechococcus mats, which suggests that cyanobacterial communities in moderately thermophilic habitats are affected by similar evolutionary pressures.