OUP user menu

Highly diverse community structure in a remote central Tibetan geothermal spring does not display monotonic variation to thermal stress

Lau Chui Yim, Jing Hongmei, Jonathan C. Aitchison, Stephen B. Pointing
DOI: http://dx.doi.org/10.1111/j.1574-6941.2006.00104.x 80-91 First published online: 1 July 2006


We report an assessment of whole-community diversity for an extremely isolated geothermal location with considerable phylogenetic and phylogeographic novelty. We further demonstrate, using multiple statistical analyses of sequence data, that the response of community diversity is not monotonic to thermal stress along a gradient of 52–83°C. A combination of domain- and division-specific PCR was used to obtain a broad spectrum of community phylotypes, which were resolved by denaturing gradient gel electrophoresis. Among 58 sequences obtained from microbial mats and streamers, some 95% suggest novel archaeal and bacterial diversity at the species level or higher. Moreover, new phylogeographic and thermally defined lineages among the Cyanobacteria, Chloroflexi, Eubacterium and Thermus are identified. Shannon–Wiener diversity estimates suggest that mats at 63°C supported highest diversity, but when alternate models were applied [Average Taxonomic Distinctness (AvTD) and Variation in Taxonomic Distinctness (VarTD)] that also take into account the phylogenetic relationships between phylotypes, it is evident that greatest taxonomic diversity (AvTD) occurred in streamers at 65–70°C, whereas greatest phylogenetic distance between taxa (VarTD) occurred in streamers of 83°C. All models demonstrated that diversity is not related to thermal stress in a linear fashion.

  • Chloroflexus
  • geothermal springs
  • hot springs
  • Roseiflexus
  • Synechococcus
  • thermophiles


Terrestrial geothermal springs support prokaryotic thermophilic communities and these have been the focus of significant research attention due to their suggested use as analogues for early Earth environments (Pentecost, 1996; Cavichiolli, 2002) and in bioprospecting for thermostable biomolecules (Demirjian et al., 2001; Schiraldi & De Rosa, 2002). In order to adequately make use of extant thermophiles for either of these research aims and to further our knowledge of thermophilic ecology we first need to understand community structure thoroughly in relation to thermal stress. Most studies to date, however, have focused on identifying novel diversity among selected taxonomic groups, notably the Cyanobacteria (Ward et al., 1998).

Three distinct community shifts can be observed in geothermal springs. Mats comprising moderately thermophilic filamentous cyanobacteria dominate in some waters at lower temperatures of 40–55°C (Castenholz, 2000). These support morphologically (Sompong et al., 2005) and genetically (Jing et al., 2005a, b) diverse filamentous and unicellular taxa. At higher temperatures the form genus Synechococcus is the only cyanobacterium encountered in geothermal springs. This taxon occurs as a surface layer on microbial mats or in streamers at temperatures of up to 75°C (Ferris & Ward, 1997). Synechococcus is generally encountered in association with green nonsulphur bacteria of the Chloroflexi within mats, and lower mat layers of chemoorganotrophs (Ward et al., 1998). At temperatures above 75°C, where photosynthesis is not possible, chemolithoautotrophic communities appear as gray or pink filaments/streamers in waters up to boiling temperature, often tolerating high dissolved sulphide levels (Reysenbach et al., 1994). It can be assumed that these shifts lead to a change in diversity, and within-assemblage shifts may also occur as, the existence of temperature-defined phylotypes has been demonstrated for certain thermophilic taxa within the Cyanobacteria and Chloroflexi (Ferris & Ward, 1997). There has, however, been no statistically meaningful study of the effects of thermal stress on diversity to date, and so the widely held view in ecology that increasing environmental stress leads to a monotonic reduction in diversity remains untested for geothermal environments. This is confounded by the focus of much research to date on relatively few taxonomic groups within thermophilic communities rather than the community as a whole.

Thermophilic communities are also highly isolated from one another geographically due to thermal barriers between habitats. Indeed, distinct phylogeographic groups of uncultivated cyanobacteria from mats have been shown to exist within continental USA, Japan and New Zealand (Papke et al., 2003; Jing et al., 2005a, b). Genetic diversity between spatially separated populations of Synechococcus appears to be independent of several physico-chemical parameters and is generally attributed to the level of geographic isolation between geothermal springs (Papke et al., 2003). This has been shown to occur over relatively short distances and also on a regional and intercontinental scale (Papke et al., 2003; Jing et al., 2005a, b). Conversely, thermophilic Chloroflexi sequences recovered from mats in continental USA and Japan did not group separately (Nubel et al., 2002). Streamer communities have received less attention in terms of phylogeography but certain genera such as Thermocrinus appear common to continental USA, Japan, Icelandic and Russian geothermal waters (Huber et al., 1998; Hjorleifsdottir et al., 2001; Blank et al., 2002; Nakagawa & Fukui, 2003).

Our ability confidently to address questions concerning diversity in relation to thermal stress and phylogeography can be improved by the study of new geothermal locations in previously uninvestigated regions, and using approaches that target the broadest possible spectrum of the community. Because the cultivation of thermophiles is demonstrated to yield diversity data that does not reflect community composition (Ferris et al., 1996b), a molecular approach is favoured. We report uncultured prokaryotic community diversity in the remote Daggyai Tso geothermal region of southern central Tibet. These geothermal springs are geochemically comparable, and yet highly isolated from previously studied geothermal regions and so represent an ideal location for further resolving thermophilic diversity and understanding the effects of geographic isolation in determining microbial diversity. We report rRNA gene-defined archaeal and bacterial diversity in mats and steamers along a temperature gradient of 52–83°C, using a combination of domain- and division-specific PCR primers, to more fully appreciate community structure. Phylogenetic analysis of sequence data is used to illustrate taxonomic and biogeographic lineages. We also use a polyphasic statistical analyses of sequence data to demonstrate the response of biodiversity to thermal stress.

Materials and methods

Daggyai Tso geothermal region

The Daggyai Tso geothermal field is located in remote southern central Tibet (29°35.413′N, 85°44.486′E), at an altitude of 5050 m. Springs are developed on coarse-grained conglomerates (Aitchison et al., 2002), which lie above an erosional unconformity developed on volcanic rocks (Williams et al., 2004). The dominant feature is a sinter-depositing boiling geyser several metres in height in near-continuous eruption. Studies of geothermal water chemistry indicate negligible sulphide, high levels of dissolved silica, boron and lithium with low concentrations of calcium and magnesium; chemical geothermometers indicate sources in the range of 200–220°C (Grimaud et al., 1985).

Thermophilic community sampling

The temperature and pH for each mat and streamer sample taken were determined at the exact point of growth using a combined pH/temperature electrode with automatic temperature compensation facility (Orion, Boston, MA). The electrode was recalibrated for pH at each sampling temperature, and digital temperature readings cross-checked using an alcohol thermometer. Hydrogen sulphide levels were determined titrimetrically using methylene blue (HS-WR, Hach, Loveland, MA). Mat sections were excised from the substrate using a scalpel, and streamers were plucked from their point of attachment using forceps. Biomass was then placed into sterile glass McCartney bottles. All samples were stored in darkness at ambient temperature during transport from the remote field location (5 days), then transferred to 4°C storage upon reaching the laboratory and until processed. A total of 30 separate samples were collected, representing three randomly chosen replicates at each of 10 thermally defined locations supporting biomass along a transect of c. 20 m in the outflow channels associated with the main geyser. Intermediate temperatures either did not exist or did not support biomass. The multiple 70°C samples reflect the occurrence of biomass in three discreet channels at this temperature. Samples were collected in May 2003.


The structure of mats and streamers, and morphotype composition were determined using stereo and compound microscopy (SZH10 stereo microscope and BX50 compound microscope, Olympus, Tokyo, Japan).

PCR, DGGE and sequencing

DNA recovery from biomass was achieved by lysis in CTAB with lysozyme and RNAse A incubation, and phenol : chloroform extraction at 60°C. 16S rRNA genes were amplified by PCR using the following domain-specific primers: Archaea 344F-905R (Cassamayor et al., 2000), and Bacteria 341F-907R (Muyzer et al., 1993, Rolleke et al., 1998). The following division-specific primers were also used: Cyanobacteria 359F-781R (Nubel et al., 1997), and green nonsulphur bacteria 77F-907R (Rolleke et al., 1998; Boomer et al., 2002). Amplification protocols were as previously described (Schafer & Muyzer, 2001). Each PCR amplicon (2 μg amplified DNA) was separated by denaturing gradient gel electrophoresis (DGGE) (Myers et al., 1988) in a urea/formamide denaturing gradient in a 7% acrylamide gel, run at 150 V in 1 × TAE buffer (pH 8) at 60°C (DGGE-2001, CBS Scientific, Del Mar, CA). Gradients for each gel are indicated in Fig. 1. Bands were excised, soaked overnight in TE buffer (pH 8) at 4°C, reamplified and purified (GFX, Amersham, Bucks, UK) prior to automated sequencing (ABI 3730 Genetic Analyzer, Applied Biosystems, Foster City, CA). Sequence data have been submitted to the NCBI GenBank database under accession numbers DQ01343DQ01400.


Denaturing gradient gel electrophoresis (DGGE) fingerprints of 16S rRNA gene alleles from mats (a–d) and streamers (e–h) along a thermal gradient. A single representative lane is shown although three independent replicates were analyzed by DGGE for each location (all 120 treatments are shown in Fig. S2 of the Supplementary material). Primers specific to Archaea (a, e), Bacteria (b, f), Cyanobacteria (c, g) and green nonsulphur bacteria (d, h) were used to PCR amplify community template. Denaturing gradient profiles for each gel are shown to the left of each gel. Band positions are numbered to the right of each gel. Arrows indicate bands that were sequenced.

Phylogenetic analysis

Approximate phylogenetic affiliations were determined by BLAST searches of the NCBI GenBank database (http://www.ncbi.nlm.nih.gov/). Multiple alignments were then created with reference to selected GenBank sequences using clustal x v.1.81 (Thompson et al., 1997). Maximum likelihood analysis using paup* 4.0b8 (Swofford, 2001) was used to illustrate the relationship of sequences to representative taxa. Bayesian posterior probabilities (Rannala & Yang, 1996) and bootstrap values (1000 replications) were calculated and are shown for branch nodes supported by more than 50% of the trees.

Statistical analysis and calculation of diversity estimates

Parametric one-way anova tests were performed using SPSS v12.0 (LEAD Technologies Inc., Chicago, IL). Diversity estimates (Shannon–Wiener, Average Taxonomic Distinctness, Variation in Taxonomic Distinctness) were carried out using Primer 6 Beta (Primer-E Ltd, Plymouth, UK). The full set of sequences obtained in this study were used in all analyses. Nine taxonomic levels were identified for taxonomic distinctness estimates: kingdom, phylum, subphylum, class, order, family, genus, subgenus, species. Weighted analysis was performed in which half-weighting was assigned to taxonomic categories of each sublevel. All plot generations were based upon 1000 simulations. Confidence funnels were generated using sublist sizes of m=5–20, and the elliptical plot from selections of m=5, 10 (Fig. S1a, Supplementary material) and m=10, 15, 20 (Fig. S1b, Supplementary material).

Results and discussion

The thermophilic community at Daggyai Tso

The geothermal habitat at Daggyai Tso supported prokaryotic microbial mats that were ubiquitous from 52 to 69°C, and morphospecies composition did not change with temperature. Lower temperatures did not support filamentous cyanobacterial mats. Most mats comprised a thin green surface layer predominantly of rod-shaped Synechococcus, over a relatively thicker pink/red layer of Chloroflexi filaments that formed the bulk of mat biomass. The bottom of the mats supported thin discontinuous dark patches of biomass, consisting of coccoid, filamentous and rod-shaped morphotypes. A few mat samples also supported less easily defined layers in some replicates, due to incorporation of mineral grains into the mat matrix. In faster-flowing water at temperatures of 65–70°C, biomass extended from the substrate as green/yellow filaments in a ‘string of sausages’ growth form. In the near-boiling water channel of 83°C, grey/white filaments occurred. The pink and black streamers usually associated with sulfidic geothermal springs were not encountered. Dissolved sulphide levels were negligible (<0.01 mg L−1) at all temperatures, and the slight variation in pH (pH 8.1–8.4) between sampling locations was not related to changes in diversity. Changes in other aqueous geochemical parameters are unlikely to have been significant due to the common source at the main geyser and rapid flow rates.

16S rRNA gene-defined diversity

The 16S rRNA gene-defined community diversity at Daggyai Tso was resolved using a combination of domain- and division-specific PCR primers. As a result we have identified sequences reflecting a broader spectrum of the total uncultured diversity in thermophilic mats and streamers than previously reported for a single geothermal location. A total of 58 unique sequences were obtained from 74 band migration classes (Fig. 1), with phylogenetic affiliations spanning 12 Divisions within the Archaea and Bacteria. All but three of these shared less than 98% sequence similarity with published sequences, indicating novel species-level diversity, and 24 sequences shared less than 88% similarity, indicating novel diversity at the genus level or higher (Stackebrandt and Gobel, 1994) within the Crenarchaeota, Euryarchaeota, Acidobacteria, Bacteriodetes, Cyanobacteria, Chlorobi and Chloroflexi (Table S1, Supplementary material). Our data show that Archaea accounted for a significant portion of total diversity (19% of sequences) in phototrophic mats and streamers at Daggyai Tso, in addition to being present in chemolithotrophoic streamers. This is interesting, as Archaea-specific primers have not previously been used to assess diversity in thermophilic mats, and so this component of mat assemblages in the community has previously been overlooked. Among the Bacteria, 22% of recovered sequences were cyanobacterial, with the Bacteriodetes, Chlorobi, Chloroflexi and Thermotogae accounting for 9–10% each. All archaeal (Fig. 2) and bacterial (Fig. 3) sequences resolved phylogenetically into known familial groups with high statistical support. The recovery of common and also several relatively rare phylotypes indicates that DGGE-based community analysis was an appropriate tool for resolving diversity in this study.


Phylogenetic relationships among the Archaea based upon maximum likelihood analysis of partial 16S rRNA gene sequence data. Sequence codes with a prefix M are derived from mats, whereas those with a prefix S are derived from streamers. The latter are italicized for clarity. Numeric suffixes refer to temperatures from which the sequence was recovered. Tree topology is 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 0.1 nucleotide changes per position.


Phylogenetic relationships among the Bacteria based upon maximum likelihood analysis of partial 16S rRNA gene sequence data. Sequence codes with a prefix M are derived from mats, whereas those with a prefix S are derived from streamers. The latter are italicized for clarity. Numeric suffixes refer to temperatures from which the sequence was recovered. Tree topology is 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. * denotes a branch node not visible in this plot. Scale bar represents 0.1 nucleotide changes per position. The portion of the tree enclosed by a dashed line is further expanded in Figs 4 and 5.

The occurrence of cyanobacteria-, green nonsulphur bacteria- and green sulphur bacteria-like sequences at temperatures up to 70°C indicates the capacity for photoautotrophic/photoheterotrophic production in mats and streamers, albeit at different light optima and oxygen levels as dictated by the known physiological ranges for taxa within these divisions. Synechococcus-like phylotypes were recovered from all samples, reflecting their near-ubiquity in geothermal habitats (Castenholz, 2000). Interestingly, the occurrence of Chloroflexi phylotypes may be partly dependant upon growth form rather than temperature, as Chloroflexus- and Roseiflexus-like sequences were recovered from mats only, whereas streamers within the same temperature range supported Anaerolinea-like sequences only. It may be that Chloroflexus and Roseiflexus are capable of detectable growth in mats only, as they have been recorded only from mats elsewhere (Nubel et al., 2002). Anaerolinea is reported previously only from industrial thermal biofilms (Sekiguchi et al., 2003). The Chlorobi identified in this study clearly inhabit a narrower temperature range than other phototrophic phylotypes, as they were not encountered at the two lowest temperatures. As the Chlorobi have a lower temperature range than other phototrophs, this might appear strange, but probably relates to the greater availability of anoxic niches within higher temperature mats in our study.

Chemoorganotrophic diversity identified in Daggyai Tso mats generally comprised groups well-represented in other mats (Ward et al., 1994; Nold 1996; Santegoeds et al., 1996), although sequences belonging to Candidate Division OP9 and the Thermosdesulfoacteria encountered in our study are not previously recorded from thermophilic mats. An interesting pattern arose for the Proteobacteria, where Alphaproteobacteria-like sequences were recovered only in streamers of 65–70°C, whereas at lower temperatures Gammaproteobacteria-like sequences occurred in mats only. This is in contrast to the recovery of Betaproteobacteria only as environmental sequences from mats in Yellowstone National Park, USA (YNP) (Nold et al., 1996), although enrichment culture studies revealed greater diversity among the Proteobacteria in YNP samples (Santegoeds et al., 1996).

Above 70°C, phototrophy is not possible and unsurprisingly no phototrophic bacterial phylotypes were recovered from these higher temperatures. Streamers at 83°C supported sequences with highest affinity to chemolithotrophic and chemoorganotrophic archaea and bacteria, although it is noteworthy that Aquificales-like sequences were not recovered as they have been recorded for several other streamer communities in YNP (Blank et al., 2002) and Japan (Nakagawa & Fukui, 2003). Instead, Thermus-like sequences were most diverse, although this does not necessarily mean they are the dominant taxon in these streamers. Thermus-like sequences have also been recovered from streamers in Iceland and YNP (Hjorleifsdottir et al., 2001; Blank et al., 2002). Interestingly, archaea were found to be absent from YNP streamers in one study (Reysenbach et al., 1994), although Korarchaeota-like sequences were retrieved from another YNP site (Reysenbach et al., 2000). The latter were not encountered at Daggyai Tso, although in our study we detected several crenarchaeal and euryarchaeal sequences.

The fact that such high diversity appears to exist among so many groups at Daggyai Tso is probably partly due to sampling effort, with previous studies largely focused on a single taxonomic group and sequencing of relatively few bands from DGGE fingerprints. It is also conceivable, however, that the Daggyai Tso geothermal region represents an extraordinarily biodiverse habitat.

Phylogeography of thermophiles

The existence of many published sequences for thermophilic cyanobacteria and green nonsulphur bacteria from various locations also allowed us to draw some conclusions about the phylogeography of the most abundant biomass components in phototrophic mats and streamers from Daggyai Tso. We maximized our recovery of sequences from these two groups by using PCR primers specific to cyanobacteria and green nonsulphur bacteria in addition to universal bacterial primers, although those for the latter group showed little specificity in our study and probably require modification before they can be regarded as truly specific to green nonsulphur bacteria. Cyanobacterial sequences all resolved unambiguously into a Synechococcus clade (Fig. 4). Several sequences affiliated with C1 (mostly Japanese) or C9 (cosmopolitan) lineages and considerably expand known diversity within these lineages. Conversely, none resolved into the A/B lineage containing sequences known only from north America, although a phylogenetically distinct group of Daggyai Tso sequences shared closest affinity with the A/B lineage yet is sufficiently distinct to warrant proposing a new lineage, designated T1 (Fig. 4). The remaining sequences could not be placed in any existing lineage and so we also propose novel lineages for these, which are named T2 and T3 (Fig. 4). Together, these new lineages significantly expand known diversity within the thermophilic Synechococcus clade. There is no evidence that these sequences separate according to temperature as suggested for A/B lineages (Ward et al., 1998), or due to occurrence in mats vs. streamers. It is likely, therefore, that they do represent phylogeographic lineages and this concurs with phylogenetic analysis of the more highly variable ITS region sequences of the genus Synechococcus from Japan, New Zealand and the USA (Papke et al., 2003), where differences in physicochemical factors at geothermal locations were found not to be significant in influencing the observed tree topology.


Phylogenetic relationships among the Cyanobacteria based upon maximum likelihood analysis of partial 16S rRNA gene sequence data. Sequence codes with a prefix M are derived from mats, whereas those with a prefix S are derived from streamers. The latter are italicized for clarity. Numeric suffixes refer to temperatures from which the sequence was recovered. Tree topology is 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 0.1 nucleotide changes per position. Novel lineages proposed from this study are shown as T1–T3 (in bold).

The green nonsulphur bacterial sequences recovered were mainly Roseiflexus-like and this reflects the observed pink/red morphotypes in mats. Our sequences clearly form a novel lineage with very high support (Fig. 5) and this may be, at least in part, a result of phylogeography. Some site specificity within the Roseiflexus A and B types was observed between sites in YNP (Boomer et al., 2002); however, as samples also arise from different temperatures this may also be a factor (Nubel et al., 2002). Our sequences grouped closest to YRL B/Type C sequences, which were recovered from within the same temperature range as those in this study, so they may be temperature-defined, although more information is needed to confirm this. In the case of Thermus-like sequences, those from Daggyai Tso grouped separately from those recovered from thermal environments in Iceland and YNP (Fig. 3). Similarly, Eubacterium-like sequences grouped separately from YNP counterparts (Fig. 3). It remains to be seen whether these too represent phylogeographic lineages. Given that the above taxa occur as part of an assemblage at any given time within mats and streamers, they may coevolve as suggested for Synechococcus and Chloroflexus (Ward et al., 1998). Thus we can expect similar evolutionary pressures for all these groups; namely, that genetic drift due to isolation is more important than variation in physicochemical parameters in determining location-specific phylogenetic lineages (Papke et al., 2003).


Phylogenetic relationships among green nonsulphur bacteria based upon Maximum Likelihood analysis of partial 16S rRNA gene sequence data. Sequence codes with a prefix M are derived from mats, whereas those with a prefix S are derived from streamers. The latter are italicized for clarity. Numeric suffixes refer to temperatures from which the sequence was recovered. Tree topology is 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 0.1 nucleotide changes per position. The novel lineage proposed from this study is shown as Type T (in bold).

The effects of thermal stress on diversity

Conventional microbial ecology theory dictates that diversity will decrease in response to increasing environmental stress such as temperature (Atlas & Bartha, 1997). Indeed, a visual assessment of changes in DGGE banding pattern for YNP samples at temperatures from 49 to 75°C suggested that this may be the case in geothermal environments (Ferris & Ward, 1997), and when comparing morphological and molecular data from studies of different sites, a reduction in diversity with increasing temperature does appear to occur in mats (Ferris et al., 1996a; Ward, 1998; Norris et al., 2002). For streamers occurring at different temperatures, patterns are less evident based upon current information (Blank et al., 2002; Nakagawa & Fukui, 2003). Statistical treatment of diversity data from thermal environments is lacking, although other extreme communities have received attention: a recent study of DGGE-derived diversity data from nonthermophilic intertidal mats demonstrated that species richness (R) and Shannon–Wiener diversity (H′) for archaeal and bacterial phylotypes determined from DGGE banding patterns decreased in a near-linear fashion to desiccation stress (Rothrock & Garcia Pichel, 2005). A general agreement between diversity estimates obtained using morphological, biochemical and molecular techniques has also been recorded for hypersaline mat communities (Nubel et al., 1999). In this study, we used multiple statistical analyses of our sequence-based dataset to demonstrate that diversity does not vary in a monotonic fashion to thermal stress at Daggyai Tso, and that increased diversity occurs under moderate thermal stress.

Shannon–Wiener (H′) diversity estimates (Fig. 6a) show a clear nonlinear relationship to thermal stress with significant differences among sites across the entire temperature range (one-way ANOVA, F=38.2, P<0.001). The greatest H′ value was obtained from mats at 63°C, with diversity decreasing irrespective of growth form as mats or streamers at both higher and lower temperatures. Conventional estimates of diversity (e.g. Shannon–Wiener, Simpson) can be misleading because they take no account of the taxonomic hierarchy within an assemblage, and so cannot reflect taxonomic diversity. As organisms that are distantly related to each other contribute more to taxonomic diversity than those that are closely related, an alternative diversity estimate is desirable. We therefore carried out an analysis of Average Taxonomic Distinctness (AvTD) to present a measure of diversity that considers the expected taxonomic distance apart of any two taxa, and Variation in Taxonomic Distinctness (VarTD) that also considers complexity of phylogenetic branching patterns between these taxa. This was possible due to the large number of sequences obtained in our study, and has the advantage over estimates based upon gel fingerprints that actual taxa (phylotypes) are used to obtain relative taxonomic information. Such analysis is also independent of sampling effort (Clarke & Warwick, 1998), thus making it a useful technique for DGGE-derived sequences where there is no established correlation between band intensity and abundance of a given phylotype. To the best of our knowledge, this is the first report of such analyses applied to community molecular data among microorganisms.


Thermophilic community diversity indices along a thermal gradient. (a) Shannon–Wiener Index (dashed lines) and Average Taxonomic Distinctness (solid lines), for mats (diamonds) and streamers (squares); (b) 95% confidence funnel plot of Average Taxonomic Distinctness for mats (TM1–5) and streamers (TS1–5). For temperatures refer to (c). (c) Relative proportion of diversity represented in mats (TM1–5) and streamers (TS1–5) by each archaeal and bacterial division; (d) 95% confidence funnel plot of Variation in Taxonomic Distinctness for mats (TM1–5) and streamers (TS1–5). For temperatures refer to (c). Shaded areas highlight mat samples, and streamer samples are unshaded.

Average Taxonomic Distinctness (AvTD) reflects the average taxonomic breadth within a sample. When interpreting diversity data along the thermal gradient in this way, the overall trend in diversity generally tracks that obtained for H′ (Fig. 6c). Significant differences occurred between the temperature group of 63–70°C and other temperatures as assessed by one-way anova (F=11.5, P<0.001). Importantly, the most diverse temperature has shifted from 63°C using H′ to a range within 63–70°C (maximum at 70°C) using AvTD, and crucially streamers are revealed as more diverse than mats. This is because although fewer phylotypes were resolved from each of the latter temperatures, they accounted for greater taxonomic diversity, a fact not apparent from H′ estimates. Furthermore, when AvTD values are plotted against a confidence funnel of expected values (Fig. 6b), all but one replicate (out of 30) were within the 95% confidence limits for expected diversity within the thermal gradient. Thermal stress, therefore, does not reduce (or increase) diversity significantly beyond expected limits based upon the total observed diversity. The low AvTD values at lower temperatures can be explained partly by the absence of thermophilic taxa with higher minimum temperature requirements, coupled with an absence of mesophilic/moderately thermophilic taxa. At the highest temperature it is likely a reflection of the reduced number of taxa able to cope with extreme thermal stress and the necessity for chemolithotrophic metabolism. It is interesting that a survey of previous community molecular diversity studies indicates that diversity estimates in terms of the simple number of taxa recorded suggests that mats are more biodiverse than streamers (Ward et al., 1998, Blank et al., 2002), yet our study clearly shows that at the taxonomic level this is not the case, even considering the heterogeneity among some samples as illustrated by streamers recovered from several 70°C locations.

An additional measure of diversity can be achieved with VarTD (Fig. 6d). Here, differences in phylogenetic structure can be resolved, as even AvTD cannot distinguish between assemblages where some genera are very species rich yet others are represented by one or few species. The bar chart format often used to illustrate relative community diversity between taxonomic groups (Fig. 6c) is not phylogenetically informative, and highlights how such plots are of limited use when used to infer this type of diversity data. The funnel plot of VarTD illustrates that the most biodiverse temperatures as measured using H′ or AvTD actually support the least phylogenetic variation, within 95% confidence limits (Fig. 6d). This is likely due to the presence of guilds comprising phylogenetically related taxa that operate at different physiological optima and so allow assemblage function across a broader range of environmental conditions. Such guilds probably exist for thermophilic Synechococcus (Ferris & Ward, 1997) and temperature-adapted strains of green nonsulphur bacteria may also co-occur (Nubel et al., 2002). Our VarTD indices suggest that this phenomenon probably extends to all phylogenetic branches within a given thermally defined assemblage where AvTD is high (as VarTD is correspondingly low, implying similar levels of phylogenetic structure within each group in the assemblage). Conversely, where VarTD is high, the low AvTD values imply that extremes of stress limit the survival of large numbers of closely related guild taxa. Indeed, VarTD values varied almost inversely to AvTD for our data, although they are independent measures and so this is not implicit. Bivariate plots were therefore constructed in order to verify that no positive correlation between AvTD and VarTD (high AvTD combined with low VarTD and vice versa) required rejection of the analysis (Fig. S1, Supplementary material). Despite the improved image of diversity that can be generated by these indices, one caveat to the use of AvTD and VarTD indices may relate to the fact that a combined score for both measures cannot be generated and so a single unified diversity estimate based upon taxonomic hierarchy of a community is not possible.


This study has resolved a highly diverse 16S rRNA gene-defined community at Daggyai Tso, including many novel phylotypes and a domain (Archaea in mats) and several divisions not previously recorded from molecular diversity assessments of mats or streamers. Whether this diversity is a result of sampling effort or represents a true biodiversity ‘hot spot’ in Tibet remains to be seen, but the high phylogenetic diversity among recovered sequences has resulted in the formation of several new phylogeographic and possibly thermally defined lineages. This adds further evidence to the concept that thermophilic communities evolve in isolation. Statistical analysis of sequence data is used to demonstrate empirically for the first time that diversity does not vary in a monotonic fashion to thermal stress, and diversity assessments that take into account taxonomic structure give a different and probably more representative view of diversity than traditional diversity estimates. These findings will undoubtedly influence the focus of future ecological and bioprospecting research in geothermal locations and other extreme environments such as hypersaline waters and sulphide springs, where microbial mats and streamers also occur, and targeting the most biodiverse niches is desired.

Supplementary material

Fig. S1. Fitted 95% probability contours for joint Average Taxonomic Distinctness (AvTD) and Variation in Taxonomic Distinctness (VarTD) distributions. Table S1. Identity of sequences obtained from community DNA of mats and streamers along a thermal gradient.


The authors are grateful to the Tibet Ministry of Geology for fieldwork assistance. This work was supported in part by funding from the Research Grants Council of Hong Kong, grant number HKU 7573/05M and the Stephen S.F. Hui Trust Fund.


View Abstract