Introduction

The first Thermus species, Thermus aquaticus, was isolated from hot springs in Yellowstone National Park in the late 1960s (Brock and Freeze 1969). Since then, seven species of the genus have been described whose names have been validated. They are Thermus thermophilus (Oshima and Imahori 1974), Thermus filiformis (Hudson et al. 1987), Thermus scotoductus (Kristjansson et al. 1994), Thermus brockianus (Williams et al. 1995), Thermus oshimai (Williams et al. 1996), Thermus igniterra and Thermus antranikianii (Chung et al. 2000). Thermus spp. are relatively easy to isolate and are found in both natural and man-made thermal systems worldwide. All Thermus species except T. aquaticus and T. filiformis have been found in Iceland.

Thermophilic microorganisms encounter the unique situation that they can only grow in relatively few small spots on the planet. Furthermore, their habitats are comparatively rare. Therefore, they can be considered as islands in the ecological sense with large dispersal barriers between them. According to the currently known global distribution of terrestrial Thermus species, there appears to be a distinct difference between widely separated geothermal regions in species composition, with a mixture of unique endemic lineages and more cosmopolitan species. Certain regions appear also to be more diverse than others. Thus, T. aquaticus has only been found in Yellowstone National Park in the USA. Another distinct species, T. filiformis has only been isolated or detected in New Zealand and is so far the only terrestrial species found there. One species, T. thermophilus seems to be confined to marine or coastal hot springs and has a worldwide distribution (Kristjansson et al. 2000; Williams and Sharp 1995).

Many of the methods of bacterial systematics, such as restriction fragment length polymorphisms (RFLP) and multilocus enzyme electrophoresis (MLEE) are both fast and reliable, and a large number of samples or isolates can be processed at the same time (Vandamme et al. 1996). Therefore, it is both practical and feasible as never before to measure various population parameters to describe and compare microbial communities. MLEE is an electrophoretic technique used to map the distribution of different alleles of a number of enzymes and to calulate diversity indices of populations. Genetic relationships are calculated from similarity in genotype on the basis of percentage of shared alleles. MLEE is an inexpensive technique and a great number of isolates can be processed at the same time. It is comparable with DNA:DNA hybridization in delineating species but as a taxonomic tool, it is more appropriate for hierarchical classification as it is very sensitive at and below the species level. In contrast to DNA:DNA hybridization, MLEE is particularly suitable for analyzing spatial and temporal dynamics of microbial populations and has proved its value in epidemiological studies (Goodfellow and O’Donnell 1993; Selander et al. 1986; Vandamme et al. 1996; Whittam 1995; Whittam et al. 1983).

Ideally, a taxonomic study of bacteria should be polyphasic, based on phylogenetic, phenotypic, biochemical and physiologic analysis on a large number of isolates from different regions in order to define species boundaries. Species boundaries should also ideally circumscribe ecologically distinct populations whether they exhibit a large or a narrow range of phenotypic, biochemical or genetic properties (Chung et al. 2000; Goodfellow et al. 1993; Vandamme et al. 1996). We have used the MLEE method and SSU rRNA sequence analysis in hand with typical phenotypic and physiological studies to examine genetic diversity and species boundaries of Thermus species isolated in Iceland. The aim of this study is to contribute to the growing taxonomic body of the genus and help in establishing basis for the study of the ecological factors and evolutionary mechanisms that direct the evolution and divergence of Thermus species.

Materials and methods

Strains

Strains were isolated at different temperatures and with some variations in medium composition and purified by repeated streaking onto medium 160 and 166 agar plates (Hjorleifsdottir et al. 2001). Medium 160 is the same as the Degryse medium 162 with only 1/10 of the phosphate buffer strength. It contains (per liter) 2.5 g of yeast extract, 2.5 g of tryptone, 1.0 g of nitrilotriacetic acid, 0.4 g of CaSO4 2H2O, 2.0 g of MgSO4 · 7H2O, 1.5 ml of 0.2 M Na2HPO4 · 12H2O; 1.0 ml of 0.2 M KH2PO4, 0.5 ml of 0.01 M Fe(III) citrate · 5H2O and 5.0 ml of trace element solution, pH 7.5. Medium 166 contains per liter geothermal tap water, 0.3 g of K2HPO4, 1 g of yeast extract, 1 g of peptone, 1 g of tryptone, 0.5 g of glucose, 0.5 g of starch, 0.6 g of pyruvic acid, 0.3 g of proline and 0.18 g of Na2CO3, 0.5 ml of 0.01 M Fe(III) citrate · 5H2O and 5.0 ml of trace element solution, pH 7.5. The trace solution contains (per liter): MnSO4 · H2O, 0.22 g; ZnSO4 · 7H2O, 0.05 g; H3BO3, 0.05 g; CuSO4 · 5H2O, 0.0025 g; Na2MoO4 · 2H2O, 0.0025 g and CoCl2 · 6H2O, 0.0046 g.

The study also included reference strains deposited in culture collections, which were isolated from different parts of the world. T. aquaticus strain YT-1 (type strain, DSM 625) isolated in Yellowstone National Park, USA, T. filiformis strain Wai33 A.1 (type strain, DSM 4687, ATCC 43280) isolated in New Zealand, T.thermophilus strain HB8 (type strain, DSM 579, ATCC 27634) isolated in Japan, T. thermophilus strain AT-62 (DSM 674, ATCC 33923) isolated in Japan, T. brockianus strain YS38 (type strain NCIMB 12676) isolated from hot springs in Yellowstone National Park, USA, Thermus sp. strain X-1 (ATCC 27978) isolated from man-made geothermal systems in USA and T. scotoductus strain SE-1 (type strain ATCC 51532) isolated from hot tap water in Iceland.

Sampling sites

Snaefellsnes, Oxarfjordur and Reykjanes are coastal geothermal sites and the hot water systems sampled were saline. The sampling sites in Oxarfjordur were runoff streams from boreholes on the coast in Snaefellsnes, and they were typical coastal hot springs. In Reykjanes, the sampling sites were ponds formed by runoffs from boreholes. Salinity values at these three sites were respectively, 0.8, 0.2–0.4 and 2.2–5.8%. The hot water running into the ponds in Reykjanes has a salinity of approximately 3% and temperature between 70 and 80°C. The geothermal diversity was poor and therefore, only one to two hot water systems were sampled in each area (Petursdottir et al. 2000).

Hveravellir, Hrafntinnusker and Hagongur are in the sparsely vegetated highlands (>500 m above sea level). Hrafntinnusker is a high-temperature geothermal field with steam holes and solfatara fields. These regions are characterized by emissions of steam and volcanic gases. The gas is composed primarily of N2 and CO2, but H2S and H2 can be up to 10% each of the total gas fraction. Because of the weak acids, CO2 (pK = 6.3) and H2S (pK = 7.2), the pH of the subsurface steam is near neutrality. On the surface the H2S is oxidized chemically and biologically, first to sulfur and then to sulfuric acid. This lowers the pH, causing corrosion of the surrounding rocks and formation of the typical acidic mud of solfatara fields. The pH often stabilizes around 2–2.5, where sulfuric acid (pK2 = 1.92) is the effective buffering agent. Because of the high temperature, there is little water that comes to the surface and the hot springs are mostly steam holes or fumaroles. These areas are usually unstable, and the individual openings often disappear or move to another location within the geothermal field.

Where groundwater levels are high, the water percolates into the hot areas, warms up, and returns to the surface containing dissolved minerals, such as silica and some dissolved gases, mainly CO2. Usually there is little H2S in such fluids. Also here the subsurface pH is near neutrality, but there is usually much more water and little sulfide so its surface oxidation has no effect on the pH. On the surface, however, the CO2 escapes and the silica precipitates, both resulting in increased pH. The pH is often stabilized at 9–10, where silicate (pK n  = 9.7) and carbonate (pK2 = 10.25) start to act as effective buffering agents (Kristjansson and Hreggvidsson 1995).

The Hagongur region is also in a high temperature region characterized by solfatara fields and steam holes. The two hot springs sampled were untypical for the area having high water turnover rate and relatively high pH for the area, 7.1 and 7.6. Hveravellir is also situated in a high-temperature geothermal field. However, the area has an unusually large number of alkaline hot springs with a high water turnover rate besides having more typical solfatara fields, steam holes and acidic hot springs with little water.

Hveragerdi is a high-temperature geothermal field situated in the lowlands and has abundant vegetation. The groundwater level is high and therefore, neutral and alkaline hot springs with high water turnover are common. The Geysir area is in between the highland and lowland and has characteristics of both. It has a number of highly alkaline water bubbling hot springs and acidic solfatara fields.

Preparation of samples for MLEE analysis

After purification, isolates were grown overnight at 65°C on medium 160 agar plates and harvested by scraping of the plates. The cells were suspended in TE buffer (10 mM Tris–HCl and 1 mM EDTA, pH 8.0) giving about 1 g in 5 ml, and then disrupted in a French Press at 700 psi. The crude extract was centrifuged at 20000g for 30 min at 4°C and the supernatant collected and kept at –80°C until use. Before use, the samples were spun again and the clear supernatant was collected. The samples were run on 7.5% (w/v) polyacrylamide gels and after running, the gels were assayed for alkaline phosphatase (AP), aspartate aminotransferase (AAT), non-specific esterase (EST), glucose-6-phosphate isomerase (GPI), hexokinase (HK), isocitrate dehydrogenase (IDH), malate dehydrogenase (MDH), nucleoside phosphorylase (NP), superoxide dismutase (SOD) and an unspecific dehydrogenase (UDH). Demonstration of the enzyme stainings has been described elsewhere except 0.2 M Tris buffer (pH 8.0) that was used in our study (Manchenko 1994; Petursdottir et al. 2000). Distinctive electromorphs of each enzyme, numbered in order of decreasing anodal mobility were equated with alleles at the corresponding structural gene locus. An absence of enzyme activity was attributed to a null allele. Distinctive combinations of alleles over the ten enzyme loci (multilocus genotypes) were designated as electrophoretic types (ETs).

Statistical analysis

Genetic diversity and relationships among the 101 Thermus strains characterized by MLEE was analyzed by two computer softwares, ETDIV and ETCLUS, kindly provided by the author, Dr. T.S. Whittam (Department of Biology, Pennsylvania State University, University Park, PA, USA). Genetic diversity at an enzyme locus among either ETs or isolates was calculated from the allele frequencies among ETs or isolates as h = (1 − ∑x 2 i )(n/n − 1), where x i is the frequency of the ith allele and n is the number of ETs or isolates. Mean genetic diversity (H) is the arithmetic average of h values over all loci. H S, the within group diversity was calculated as the mean of the diversity values obtained for the separate sampling sites (subpopulations). Total genetic diversity, H T, is the diversity value calculated for the population as a whole. For a subdivided population the total diversity H T will be greater than the diversity within subpopulations. Nei and Chesser (1983) coefficient of genetic differentiation G ST, was then calculated as (H T − H S)/H T using the ETDIV software. This coefficient indicates how big proportion of the overall variation is due to differences between subsamples. A dendrogram based on the average linkage algorithm of all the isolates with the reference strains included was made. Distance was measured as the proportion of mismatched loci between pairs of ETs. Genetic distance between pairs of ETs was expressed as the proportion of enzyme loci at which dissimilar alleles occurred (mismatches) and clustering of ETs was performed from a matrix of genetic distances by the average linkage method (UPGMA) (Caugant et al. 1987; Petursdottir et al. 2000).

Phenotypic characterization

A phenotypic and physiological analysis was carried out on a subset of the isolates from the MLEE analysis representing different clones of the same lineages and from different geographic regions. Growth was examined at temperatures 50, 65 and 78°C on agar plates with medium 160. Growth was analyzed at pH 5.0, 6.0, 8.0, 8.7 and 9.5 on agar plates with medium 160, but the pH was adjusted with HCl and NaOH. Salt and ion tolerance was tested on medium 160 agar plates supplied with 0.5, 1 and 2% NaCl, 50 mM of MgSO4, 50 mM of CaSO4, 2 mM of CuSO4, 50 mM of Na2SO4, 50 mM of Na2SO3 and 50 mM Na2S2O3. Growth in the presence of the chelating agent EDTA was analyzed on medium 160 agar plates supplemented with 2 and 5 mM EDTA. Utilization of single carbon sources was tested on minimal medium agar plates containing 0.2–0.4% of the organic compound as described before (Kristjansson et al. 1994; Petursdottir and Kristjansson 1996). Growth was examined in the presence of 35 carbon sources. Carbon sources, salts and other growth conditions are listed in Table 5. Growth was examined after 1, 3 and 5 days of incubation. Nitrate reduction was tested with 2- and 5-day-old cultures as described before with the modification of using microtiter plates (Smibert and Krieg 1994).

Thiosulfate oxidation

A few representative strains were selected for the thiosulfate oxidation test on the basis of the clustering obtained by the MLEE analysis. The production of sulfate from 16 mM of thiosulfate was tested in liquid minimal medium supplied with 0.2% acetate, pyruvate or proline after 2 and 4 days incubation at 65°C (Skirnisdottir et al. 2001). The concentration of sulfate was determined by using BaCl2 and turbidity (Tabatabai 1974).

Phylogenetic analysis

Strains for the phylogenetic studies were selected on the basis of the UPGMA clustering obtained by the MLEE analysis. The phylogenetic position of a few representatives of each lineage was determined by using the SSU rRNA gene as the phylogenetic marker by partial sequencing. The following strains were analyzed: 165, 220, 346, 2101, 2103, 2121, 2123, 2126, 2127, 2133, 2135, 2789, 2791 and 6230. DNA was isolated with a Dynabeads DNA Direct Kit (Dynal) according to the manufacturer. PCR amplifications of the SSU rRNA has been described elsewhere (Skirnisdottir et al. 2001). The SSU rRNA genes from strains were partially sequenced with an ABI 377 DNA sequencer by using BigDye Terminator Cycle Sequencing Ready Reaction kit according to the manufacturer (PE Applied Biosystems). Primer R805 (5’-GACTACCCGGGTATCTAATCC’-3; 805-785) was used for the sequencing. After BLAST searches, the sequences were manually aligned with other sequences within the Thermus group obtained from the Ribosomal Database Project (Maidak et al. 1999). Evolutionary distances were computed from pairwise similarities by using the correction of Jukes and Cantor (1969). ClustalX (http://bips.u-strasbg.fr/fr/Documentation/ClustalX/) was used to construct the phylogenetic tree by the neighbor-joining algorithm.

The GenBank accession numbers of the SSU rRNA sequences of the organisms used in this analysis are as follows: T. scotoductus IT252 (SE-1) (AF032127), Thermus sp. NMX2 A.1 (L09661), Thermus sp. IT-7254 (AF257219), T. antranikianii HE5 (Y18412), Thermus sp. ac-1 (L37520), T. igniterrae RF-4T (Y18406), unidentified Thermus sp. SRI248 (AF255591), T. aquaticus YT-1 (L09663), unidentified Thermus OPB31 (AF027020), Thermus sp. ac-7 (L37522), unidentified Thermus OPB32 (AF027021), unidentified Thermus OPB19 (AF027019), T. thermophilus HB-8 (X07998), T. filiformis WAI 33 A.1 (X58345), Thermus sp. YS38 (Z15062), T. oshimai SPS-17 (Y18416), Thermus YSPID A.1 (L10070), Thermus ZHGIB A.4 (L10071), T. thermophilus AT-62 (L09660), Thermus sp. Tok8 A.1 (L09666), Thermus sp. Tok20 A.1 (L09665), Thermus sp. W28 A.1 (L10068), Thermus sp. T351 (L09671) Thermus rehai RH99-02 (AF331969), Thermus sp. RH-1514 (AF521188), Thermus yunnanensis RH1003 (AY861385) and Thermus kawarayensis KW11 (AB071811).

DNA:DNA reassociation analysis was performed between T. brockianus YS38 (type strain, NCIMB 12676) and strain 2789. DNA was isolated by chromatography on hydroxyapatite by the procedure of Cashion et al. (1977). DNA:DNA reassociation analysis was carried out as described by de Ley et al. (1970) with modification described by Huss et al. (1983) and Escara and Hutton (1980).

Results

Origin and isolation of strains

The 101 Thermus strains used in this study cover a wide geographic range and sample sites different in physicochemical and ecological characteristics. The sites could be broadly assigned to eight widely separated geothermal areas in Iceland (Fig. 1). The hot springs varied in pH, temperature, salt concentration, sulfide concentration, altitude and water turnover rate. A total of 45 hot springs and three man-made hot water geothermal systems were sampled. Three hot springs were in the pH range 4.5–5.9, nine hot springs in the pH range 6.0–6.9, twelve hot springs in the pH range 7.0–7.9, nine hot springs in the pH range 8.0–8.9 and twelve hot springs in the pH range 9.0–9.9. Most of the strains were isolated from alkaline hot springs, but a number of isolates were also obtained from hot water springs with pH lower than 7 (Table 1).

Fig. 1
figure 1

Location of sample sites in Iceland for 101 Thermus strains used in this study

Full size image
Table 1 Characteristics of the geothermal sampling sites and isolated Thermus species
Full size table

Genetic relationships of ETs

Based on the MLEE analysis of the 101 strains, seven distinct and genetically highly divergent lineages of Thermus were observed (Fig. 2). This grouping was consistent with the SSU rRNA sequencing results (Fig. 3). All except one of the population lineages (represented by isolates 2139 and 2789) could be affiliated to known species. The known species lineages were T. brockianus, T. thermophilus, T. igniterrae, T. scotoductus, T. antranikiaii and T. oshimai.

Fig. 2
figure 2

Genetic relationship between 101 Thermus strains isolated from hot springs in Iceland. The dendrogram was generated with average linkage algorithm of clustering from matrix of genetic distances based on ten enzyme loci. The sample sites are defined according to the colors in Fig. 1

Full size image
Fig. 3
figure 3

Phylogenetic positon of 14 subset Thermus strains used in this study among representatives of all major Thermus lineages

Full size image

The T. thermophilus lineage consisted of marine isolates. These strains were isolated from only two locations, both of them coastal and saline (Oxarfjordur and Reykjanes). Isolates from each region were clustered together (Fig. 2).

The most commonly isolated species was T. scotoductus with 29 isolates. This lineage, in contrast to the other lineages was further divided into six distinct genetic sublineages, designated as A–F for convenience. These sublineages may represent subspecies with different habitat preferences as clustering of regional isolates could be observed. Strains belonging to sublineage A were mainly isolated from hot springs in Hagongur and Hveravellir, whereas strains in the very distinctive sublineage F were mainly isolated in Hrafntinnusker. Colorless strains of T. scotoductus, including the type strain (SE-1) from Iceland and Thermus sp. X1 from USA, clustered together in subgroup C. Subgroups B, D and E are only represented by few strains. In contrast to the T. scotoductus lineage, no inter-regional clustering was observed for isolates in the T. igniterrae and T. brockianus lineages. Only one T. antranikianii isolate was obtained.

One of the lineages consisted of strains that were highly divergent from the other lineages. According to the SSU rRNA sequencing results, bacteria belonging to this lineage were most closely related to T. brockianus, with 98.3% sequence similarity. However, as seen from the MLEE results it was genetically distant from T. brockianus with different alleles at all loci. Results of DNA:DNA reassociation between Thermus 2789 from this lineage and T. brockianus YS38 (type strain) gave a value of 72.9%.

Genetic analysis of isolates from different regions

Thermus brockianus

The T. brockianus group consisted of 21 isolates from three of the most varied inland geothermal areas, that is, Hveragerdi, Hveravellir and the Geysir areas. Thirteen ETs could be distinguished (Fig. 2). Six of the ten enzymes were polymorphic and the number of alleles ranged from 1 to 4 with an average of 2.1. SOD, AP, GPI and the unspecific dehydrogenase were invariant in the population (Table 2). Little genetic diversity was found within this population, H T = 0.254 (Table 3). Some sharing of ETs was observed between the different areas. One ET was shared between the Geysir area and Hveravellir, one ET was represented by four isolates from the Geysir area and one from Hveragerdi. Two ETs were shared between isolates from Hveravellir and Hveragerdi (Fig. 2). The G ST value was noticeably low for the whole cluster, 0.016. This indicates that the different geographic regions harbor the same genetic population.

Table 2 Allele frequencies within lineages for different loci
Full size table
Table 3 Genetic diversity within lineages
Full size table

Thermus thermophilus

This population consisted of 15 isolates from only two coastal locations, eight from Reykjanes and seven from Oxarfjordur. No ET was shared between the populations of these two areas, but 14 ETs could be distinguished. Six of the ten enzymes were polymorphic and the number of alleles ranged from 1 to 6 with an average of 2.8. SOD, HK, AP and IDH were invariant in the population (Table 2). The overall diversity was moderately high, H T = 0.438 (Table 3), but significantly lower within the regions, 0.329 in Reykjanes and 0.340 in Oxarfjordur. The G ST value showed that the inter-regional component of diversity was high between these two areas, 0.242, which strongly indicates that the separate populations are genetically different and are evolving independently. This divergence of the T. thermophilus population was also clearly observed in the UPGMA clustering of the isolates (Fig. 2).

Thermus igniterrae

The population of T. igniterrae consisted of 22 isolates found in all the sampling areas except the three coastal sites. Nineteen ETs could be distinguished (Fig. 2). Eight of the ten enzymes were polymorphic and the number of alleles ranged from 1 to 6, with an average of 2.9. SOD and IDH were invariant in the population (Table 2). The overall population was genetically relatively tight, as reflected by genetic diversity index, H T = 0.376 (Table 3). There was a little sharing of ETs within the lineage from the five geographical areas. One ET was shared between Hveragerdi and Hveravellir. The G ST value was relatively high for the whole population, 0.179, indicating inter-regional genetic differentiation. However, as isolates from all of the locations were relatively few, this high value should be cautiously interpreted.

Thermus scotoductus

This population consisted of 29 isolates, which were found in all the sampling sites except the three coastal sites. The genetic diversity was highest for this Thermus population, H T = 0.583, including T. scotoductus SE-1 (Table 3). All ten enzymes were polymorphic (Table 2), but the number of alleles for each enzyme ranged from 4 to 10, with an average of 5.4. Twenty-three distinct ETs could be distinguished and there was no sharing of ETs between the different geographic regions. The Nei’s coefficient of genetic differentiation for the whole population was very high, G ST = 0.362, indicating inter-regional genetic differences in this population. However, this high G ST value may be misleading as it may be due to the very distinct genetic lineages observed within the population. These lineages may be ecologically distinct subpopulations, which have different habitat preferences and are evolving independently. The lineages B–F consisted of very few isolates. With larger sample size, these lineages might be found in their preferred habitats in other geographical regions as well, and this would consequently lower the overall G ST value.

Thermus oshimai, Thermus sp. and Thermus antranikianii

The low genetic diversity observed for the new Thermus sp. and T. oshimai lineages is based on the low isolate numbers within them (Table 3). Therefore, these values are not significant. Only one T. antranikianii strain was found.

Environmental variables

The origin of the Thermus strains selected for this study was investigated in relation to two different physical parameters of the sample sites, pH and temperature. No clear relationship was found between temperature at the isolation site and the different lineages. Place of origin of three species T. igniterrae, T. brockianus and T. scotoductus showed some relation to pH (Tables 1 and 4). The T. igniterrae strains examined in this study originated from 12 different hot springs. Sixty-seven percent of them were above pH 8 and 50% were in the pH range 9–9.9 (Table 4). This tendency can also be estimated on the basis of sampling effort. The majority of strains belonging to this lineage were isolated from hot springs with pH above 8.1 (up to pH 9.8). Twenty-one of the sampled hot springs had pH at and above 8.0 and 16. T. igniterrae strains were isolated from these hot springs. On the other hand, 21 hot springs were in the pH range 6–7.9 and only six T. igniterrae strains were isolated from these hot springs. Therefore, per sampling effort, T. igniterrae was found three times more common in the highest pH hot springs than in the hot springs with pH range 6–7.9.

Table 4 Percentage of total hot springs in which each species was found, at different pH ranges. Number of isolates isolated at each pH range is also shown
Full size table

The T. brockianus strains originated from 15 different hot springs. Seventy-eight percent were above pH 8 and 40% were in the pH range 9–9.9. None of the hot springs had pH lower than 6 and only two hot springs were below pH 7. This tendency is also clear on the basis of sampling effort where T. brockianus (a total of 21 strains) was found almost three times more common in the hot springs with the highest pH (pH 8–9.9) than in the hot springs with pH range 6–7.9.

T. scotoductus was isolated from 18 hot springs and one man-made system. Approximately 78% of these had pH below 8.0 and 17% had pH below pH 6.0. Also, per sampling effort, T. scotoductus (total of 29 strains) was found more than six times more common in hot springs in the pH range 4.5–7.9 (25 strains) than in hot springs of pH range 8–9.9 (a total of four strains) (Table 4). This is in clear contrast with what was observed with T. igniterrae and T. brockianus.

Information on conductivity was not available for the sampling sites, but the coastal sites were saline. T. thermophilus was only found in the coastal sites and the only other species found in a coastal hot spring was the unaffiliated Thermus lineage represented by isolates 2789 and 2139.

T. oshimai was found in two separate geothermal areas. However, all these hot springs had a striking common characteristic, that is, a bacterial filamentous biomass that is typical of high-sulfide hot springs.

Physiologic and phenotypic characterization

The numerical taxonomy results did not reflect the genetic clustering. No correlation was found between species lineages and growth on specific single carbon source or nitrate reduction. However, T. igniterrae as a group could be distinguished from other lineages, as usually only about half of the strains grew on any particular single carbon source (Table 5).

Table 5 Percentage of positive phenotypic tests for the strain lineages derived from the multilocus enzyme electrophoresis analysis
Full size table

The salt and ion tolerance patterns varied between the different lineages (Table 5). For example, the isolates from coastal habitats belonging to T. thermophilus had the highest salt tolerance of all. All of them tolerated 1% NaCl and the majority tolerated 2% NaCl (67%). Two isolates of the new Thermus sp. (lineage 5) grew at 2% NaCl, but these isolates were also from a saline coastal hot spring. T. igniterrae was on the other hand noticeably more sensitive to salt concentration than the other species. Only 5% of the isolates grew at NaCl concentration as low as 0.5%. Similarly, only 19% of the T. igniterrae strains grew on Na2SO4, none grew on Na2SO3, and 29 and 33% grew on and 2 mM of CuSO4 and 50 mM of CaSO4, respectively. The great majority of strains belonging to all the other Thermus species grew under these conditions (Table 5).

Over 90% of the strains of all the Thermus species grew well at pH 8.7, except strains in the T. igniterrae lineage, where only 29% of strains grew. This was unexpected as they were isolated from the hot springs with the highest pH. Of the terrestrial species, T. brockianus was the other species most often isolated from hot springs with high alkalinity, and 62% of the strains from this lineage grew at pH 9.5. Thermus species appear generally to be adapted to slightly alkaline conditions; however, T. scotoductus was most often isolated from hot springs in the lower pH range. Twenty-five percent of the strains in of this lineage grew at pH 5.0, which is in the higher range compared with other species. Notably, only 14% of the strains grew at pH 9.5, which agrees well with the pH of the hot springs origin of most isolates. The unaffiliated Thermus lineage grew very well at both extremes, pH 5.0 and 9.5 (86 and 100%, respectively), and was also isolated from hot springs in this range.

Ethylenediaminetetra-acetic acid (EDTA) is a scavenger of divalent metal ions but bacteria need these ions, that is, as coenzymes in enzyme complexes and in the cell wall. If these ions are not present, the enzymes does not work and no growth occurs (Madigan et al. 1997). Most strains tolerated 2 mM of EDTA, but 5 mM was too much for all of them.

Thiosulfate oxidation

We tested four to nine strains in each lineage and in some of the lineages (T. thermophilus, T. igniterrae and T. scotoductus), a part of the strains could oxidize thiosulfate (Table 5). However, all of the T. brockianus and T. oshimai strains were able to do this. Furthermore, none of the strains in the new Thermus sp. lineage were thiosulfate-oxidative.

Discussion

Based on MLEE analysis of 101 Icelandic Thermus strains, seven distinct and genetically highly divergent populations of Thermus were observed. This grouping (Fig. 2) was consistent with the SSU rRNA sequencing results (Fig. 3). As expected, the two trees had different tree topologies as the MLEE method can only be used for demarcation of lineages, but not for phylogenetic relationship as the SSU rRNA sequencing method.

Six of the lineages could be assigned to validly described Thermus species that have previously been found in Iceland: T. brockianus, T. thermophilus, T. oshimai, T. scotoductus, T. antranikianii and T. igniterrae. The two species, T. aquaticus and T. filiformis seem to have no close relatives in Iceland. One lineage of seven isolates differed in all loci from other groups and may represent a new Thermus species. According to SSU rRNA sequencing, the isolates 2123 and 2789 that belong to this new lineage are approximately midway between T. brockianus (strain YS38) and T. igniterrae (strain RF-4) giving 98.3 and 97.8% sequence similarity, respectively. However, results of DNA:DNA reassociation between Thermus 2789 and T. brockianus strain YS38 (type strain) gave a borderline value of 72.9% (Amann et al. 1995; Goodfellow and O’Donnell 1993; Stackebrandt and Goebel 1994).

Some Thermus strains are able to oxidize thiosulfate and grow mixotropically (Skirnisdottir et al. 2001). Our results here show that this is a relatively common trait within the genus Thermus.

This study illustrates that the MLEE method is very effective in delimiting Thermus species. It can also be used for a rapid identification of species. For example, the SOD and the IDH were different between the different Thermus species, but almost invariable within the species.

An apparent geographic component was observed in the distribution within the T. thermophilus lineage. Two genetically distinct populations were found in the two widely separated locations sampled. Similar results have been obtained with another species, Rhodothermus marinus, living in coastal hot springs (Petursdottir et al. 2000).

Geographically distinct populations could not be discerned within other Thermus species identified in this study. Genetic diversity was generally high, but isolates of each species were too few and scattered between too many geographic areas. Of the lineages with the greatest number of isolates, genetic diversity was far greatest in the T. scotoductus lineage, moderate in T. igniterrae and T. thermophilus lineages and very low in the T. brockianus lineage. The T. scotoductus lineage appeared also to be divided into very distinct sublineages that may represent different ecotypes of the species. The opposite was true for the T. brockianus lineage. Genetic diversity in this lineage was unusually low for a natural species, despite the fact that the isolates were from widely separate regions and is indicatieve of pandemic distribution. It is noteworthy that the type species isolated in Yellowstone National Park in USA had a very similar allelic pattern to the Icelandic isolates.

Number of studies have shown that phenotypical properties are not very useful in defining taxonomic groups at the species level of Thermus. Furthermore, physiological properties, such as temperature and pH growth optimum and maximum appear to be similar from species to species (Chung et al. 2000; Williams et al. 1995). Our study corroborates previous findings that phenotypic characteristics, especially carbon source utilization patterns, do not discriminate between different Thermus species as no correlation was found between growth on specific single carbon source to the clustering obtained by the MLEE analysis. Thus, no metabolic phenotype could be attributed to the genetic lineages.

Balkwill et al. (2004) have noted that T. scotoductus may be phenotypically more variable than other Thermus species as observed in lipid composition, pigmentation, carbon utilization and not the least in utilization of diverse inorganic compounds, such as iron and sulfur compounds. This could explain the considerable subdivision of these lineages by MLEE in this study where the different lineages may represent quite different ecotypes.

Water activity may be important for habitat preferences of Thermus species, either ability or inability to tolerate salts. Thus, the capacity of T. thermophilus growing at higher concentrations of sodium chloride may be sufficient to explain its habitat preference. In contrast, the majority of T. igniterrae strains appear to be sensitive to ionic strength of various salts. Noteworthy is its sensitivity to sulfur compounds (Table 5).

T. igniterrae shows discontinuous but very unexpected distribution. It has so far, only been found in Iceland and northeast Australia. In Australia it was detected and isolated as the dominant Thermus species at the higher temperatures in nonvolcanically heated hot runoff from a borehole (New Lorne Bore) in the Great Artesian Basin (GBA) aquifer in Blackall, central Queensland (Spanevello and Patel 2004). The water from this borehole is very pure and has a sodium bicarbonate chloride hydrochemistry typically found in bores in this region. The total ionic strength is about 430 mg/L of which CO −3 –HCO −3 is 60%. It has low concentration of Na+, 3.7 mg/L and bivalent cations, 2 mg/L of Ca2+, 0.1 mg/L of Mg2+ and 0.01 mg/L of Mn2+. No sulfide is reported and a low sulfate concentration (4.3 mg/L) observed. The pH is relatively high, 8.5, and temperature at the source is 88°C (Spanevello 2001). The physiochemical characteristic of this T. igniterrae habitat in Australia agrees well with our findings, which indicate that the species prefers high alkaline and high temperature hot springs generally low in mineral salt content and is sensitive to NaCl, bivalent cations and sulfur compounds in the culture media (Tables 1, 4 and 5).

Two important niche parameters that may explain growth in any particular habitat are pH and temperature. In this study specific temperature did not relate the isolation sites of any of the species lineages. Regular temperature fluctuations in hot springs appear to be common. Up to 10°C difference have been observed between the lowest and the highest temperature within a 24–60-h period (Marteinssson, unpublished). Also, due to temperature gradients in hot springs, isolates may originate from preferred neighboring temperature zones that may have higher or lower temperature. The effect of temperature on the distribution of Thermus species may, therefore be masked.

Most hot springs in Iceland are alkaline and high alkalinity hot springs with pH up to at least 9.0 are relatively common in Iceland. Higly acidic hot springs with high water turnover rate are, scarce and only single or very few isolates are usually obtained from these habitats. These isolates are interesting as representatives of strains tolerating or growing at or close to the extreme low pH limit of growth for a particular species. In this study, pH of the sample site indicates a relationship to lineage formation. T. igniterrae and T. brockianus were generally isolated from hot springs with pH > 8.0, but T. scotoductus from hot springs around neutral pH.

In conclusion, our results indicate that some of the genospecies of Thermus may be ecologically distinct populations distinguished by adaptation to different physicochemical conditions. The results of this study also suggest that an ecological approach will perhaps reveal more about the factors that distinguish between different Thermus species or populations than phenotypic analyses in the laboratory. Such an approach should be based on a greater number of isolates randomly selected from each of a selected few characteristic hot springs of different types. The isolates should also be selected over a period of time to see how stable the populations are and environmental conditions carefully monitored for the same reason. In such a study, the MLEE methodology would provide fast and relatively inexpensive means for species identification besides giving valuable genotypic data for population structure studies.