Introduction

Thermophilic mat communities develop in geothermal springs at temperatures of up to about 65°C (Ward et al. 1998). Cyanobacteria are the most commonly reported microbial group constituting these mats and they are considered the major primary producers in these habitats (Castenholtz 1973). Other bacteria share these environments with the Cyanobacteria and have important roles within these microbial communities (Ward et al. 1990; Weller et al. 1992; Moyer et al. 1995).

Current results based on molecular methods are showing a high complexity in the microbial communities forming these hot spring mats (Ward et al. 1990). Synechococcus-like representatives are the Cyanobacteria usually found at the highest temperatures (near 60°C) (Hongmei et al. 2005). Members of the bacterial division Chloroflexi are also commonly retrieved from thermophilic mats (Weller et al. 1992; Moyer et al. 1995). While some studies underline the high abundance of Chloroflexus-like bacteria in some mat communities (Polerecky et al. 2007), others have proposed a limited role of these bacteria and a dominance of Cyanobacteria in other high-temperature mats (Roeselers et al. 2007). At present, there is controversy on the presence and activity of Chloroflexi and Cyanobacteria in different mat systems.

A large fraction of microorganisms in natural systems have been reported to be in dormant stages, so that they are inactive members of the community waiting for better conditions to arrive (Nold and Ward 1996). RNA-based molecular methods are being introduced to unambiguously detect the metabolically active microorganisms within a community (Mills et al. 2004; Nogales et al. 1999; Portillo et al. 2008). Recent studies have emphasized the importance of microbial immigration and dispersion in current estimates of microbial diversity and community structure (Sloan et al. 2005; Portillo and Gonzalez 2008a). Thus, a differential distribution of microorganisms in distinctive layers of the mats could be proposed. Distinct colored mats might be a result of the composition and structure of the microbial communities or the metabolic activity of their microbial components (Bauld and Brock 1973; Castenholtz 1973; Ward et al. 1998). Nevertheless, previous studies have been unable to report changes in microbial communities of hot spring mats (Ferris and Ward 1997; Ward et al. 1998).

This study aims to determine the major members of the microbial communities participating in the activity of different microbial mat layers at Boekleung Hot Springs, Western Thailand. Different colored mat layers (green, yellow and red) were observed and studied using molecular methods based on both DNA and RNA. The results confirmed the uniqueness of the microbial communities integrated in each mat layer.

Materials and methods

Sampling sites and sample collection

Samples were collected from microbial mats at a hot spring in Boekleung (Ratchaburi Province), Thailand, located at 13° 31.015′N 099° 14.682′E (Fig. 1). Temperature was measured with a digital thermometer and a thermopar microprobe. Water showed temperatures up to 57°C at the source. A pH of 6.8 was measured at the studied mats. Mats collected at 56.5°C showed two clear layers, an exterior (upper) layer showing green color (a) and an interior (inner) layer with a bright red color (c) in contact with the stone substrate (Fig. 1). A monolayer, green-yellowish colored mat (b) was collected at 50.8°C (Fig. 1) in the surroundings of the previous mat.

Fig. 1
figure 1

Detail of three different layers of microbial mats from Beokleung Hot Springs analyzed during this study. a Green (external layer); b yellowish mat; c bright red (internal layer). Temperature was 56.5°C for a and c, and 50.8°C for b

Full size image

Samples were collected using sterile tubes containing RNAlater (Ambion, Austin, USA) to preserve microbial RNA, maintained on ice until arrival to the laboratory, and then stored at −80°C until processed. Samples collected for the culture of microorganisms were stored in sterile tubes and maintained on ice until processed.

Nucleic acid extractions

DNA was extracted using the Nucleospin Food DNA Extraction Kit (Macherey-Nagel, Düren, Germany) according to the manufacturer’s recommendations. Total RNA was extracted using the RNAqueous4PCR kit (Ambion Inc., Austin, USA). The protocol for total RNA extraction includes DNaseI treatment (37°C for 1 h) to remove any DNA present in the final RNA extract. Controls without sample (as controls for tubes and reagents) were carried out in parallel to the samples.

Reverse transcription and DNA amplification

A reverse transcriptase reaction was carried out to obtain the complementary DNA (cDNA) to the 16S rRNA genes to be amplified. The reverse transcriptase Thermoscript (Invitrogen, Carlsbad, USA) was used in this study with a 16S rRNA gene-specific primer, 518R (5′-ATT ACC GCG GCT GCT GG; Muyzer et al. 1993), at an annealing temperature of 55°C for 1 h. Controls lacking reverse transcriptase were also carried out to check for the presence of DNA traces in the extracted RNA. A standard amplification reaction by PCR followed production of cDNA. Amplification of 16S rRNA fragments from cDNA was performed by PCR using as forward primer 27bF (5′-AGA GTT TGA TYM TGG CTC AG; Portillo et al. 2008) for Bacteria or Cya106F (5′-CGG ACG GGT GAG TAA CGC GTG A; Nübel et al. 1997) for Cyanobacteria and 518R as reverse primer. ExTaq (Takara, Shiga, Japan) was the DNA polymerase used for PCR, following the manufacturer’s recommendations. Thermal conditions for the amplification reaction consisted on the following steps: 95°C for 2 min; 30 cycles of 95°C for 15 s, 55°C for 15 s and 72°C for 1 min; and a final incubation at 72°C for 10 min.

Amplifications from DNA were performed following a similar protocol although the primers 27bF and 907R (5′-CCC CGT CAA TTC ATT TGA GTT T; Lane 1991) for Bacteria and Cya106F and a combination of Cya781Ra (5′-GAC TAC TGG GGT ATC TAA TCC CAT T) and Cya781Rb (5′-GAC TAC AGG GGT ATC TAA TCC CTT T) for Cyanobacteria (Nübel et al. 1997) were used. Amplification of rRNA gene fragments from Archaea and Eukarya DNA was performed using the primer pairs 20bF (5′-YTC CSG TTG ATC CYG CSR GA) and 1492bR (5′-GGY TAC CTT GTK WCG ACT T), and EukA (5′-AAC CTG GTT GAT CCT GCC AGT) and EukB (5′-TGA TCC TTC TGC AGG TTC ACC TAC), respectively (Gonzalez and Saiz-Jimenez 2005).

Microbial community fingerprints

Fingerprints of the total and active microbial communities were obtained by denaturing gradient gel electrophoresis (DGGE) following the method described by Muyzer et al. (1993). Extracted DNA and the generated cDNA were directly amplified by using the primer pair 341F-GC (5′-CCT ACG GGA GGC AGC AG with a GC-rich tail attached to its 5′ end; Muyzer et al. 1993) and 518R for bacterial DGGE analysis. Cyanobacterial-specific fingerprints were obtained by a nested PCR approach from extracted DNA. A first amplification using cyanobacterial-specific primers (Cya106F and Cya781Ra + b; Nübel et al. 1997) was followed by a PCR amplification using the primer pair 341F-GC and 518R. Only bacterial PCR-DGGE analyses were performed following quantitative requirements (Portillo et al. 2008; Portillo and Gonzalez 2008b) while cyanobacterial PCR-DGGE, obtained after a nested PCR were not considered quantitative. Gels obtained by DGGE were digitalized using Kodak 1D image analysis software (Kodak, New Haven, CT). Images were analyzed with the program tnimage (http://entropy.brneurosci.org/tnimage.html) using its densitometry function. Relative quantification of bands from each bacterial community profile was performed following the quantitative procedure described by Portillo and Gonzalez (2008b). According to the described procedure, product amplification was only considered during the exponential phase of PCR. Comparisons between community fingerprints were performed as described by Portillo and Gonzalez (2008b) calculating a Cramér-von Mises-type statistic through a Monte-Carlo test procedure to determine the significance of differences between microbial community fingerprints from different samples and between DNA- and RNA-based analyses.

Environmental clone library construction and screening

PCR products from bacteria and Cyanobacteria were obtained with the 27F-907R and Cya106F-Cya781Ra + b primer pairs, respectively, purified with the PCR purification kit (JetQuick, Germany), and cloned using a TOPO-TA cloning kit (Invitrogen). The 16S rRNA libraries obtained were used to identify some of the microbial components of the community. A previously described screening procedure (Gonzalez et al. 2003) based on discrimination of clones using PCR-DGGE was followed with these libraries to identify the major DNA bands observed in DGGE analyses.

Nucleotide sequence analysis

Sequence data were edited using the software Chromas, version 1.45 (Technelysium, Tewantin, Australia). Homology searches from the nucleic acid sequences were performed using the Blast algorithm (Altschul et al. 1990) at the NCBI (National Center for Biotechnology Information; http://www.ncbi.nlm.nih.gov/Blast/). Sequences were checked for chimeras using the program Ccode as described by Gonzalez et al. (2005). Sequence alignments were carried out with the program ClustalW (Thompson et al. 1994). Aligned sequences were processed with the program Treepuzzle (Strimmer and von Haeseler 1996) to obtain tree topologies, using a maximum likelihood quartet puzzling method, showing the phylogenetic relationships between retrieved sequences and their related homologs. Reliability values were calculated for internal branches indicating the number of times (as percentages) a cluster has been reconstructed during the puzzling steps (generally from 5,000 estimations). In order to obtain a conservative tree topology, multifurcations are drawn where the support value for a bifurcation is lower than 50% (Stott et al. 2008). The number of nucleotides for tree construction ranged from 740 to 860 bp depending on the bacterial group. Thermotoga neapolitana (Accession Number AB039768) was used as outgroup for tree construction.

Nucleotide sequence accession numbers

Nucleotide sequences obtained in this study have been deposited in the GenBank database under the accession numbers reported in Table 1.

Table 1 Estimates of relative intensity for the major bands obtained by denaturing gradient gel electrophoresis (DGGE) analysis and the corresponding bacteria detected during this study from green, yellowish, and red mat layers
Full size table

Results

Cyanobacterial fingerprints

DNA-based PCR-DGGE analysis with Cyanobacteria-specific primers revealed differences between the mat layers described above. Cyanobacterial community fingerprints showed the presence of dominant phylotypes in each of the studied layers (Fig. 2). The green layer (a) of the mat was dominated by members of the Order Chroococcales showing 93% similarity in its 16S rRNA gene sequence with its closest relative. The red layer (c) of the mat was dominated by closed relatives of Thermosynechococcus elongatus BP-1, a cyanobacterium belonging to the Order Chroococcales, with percentages of similarity in their 16S rRNA gene sequences ranging from 95 to 99%. In the yellowish zone (b), the dominant phylotype was related to thermophilic, uncultured Cyanobacteria (i.e. EF032786) (93% similarity) and was clearly different of the phylotypes dominating the green and red layers (a and c).

Fig. 2
figure 2

DGGE molecular fingerprints of cyanobacterial 16S rRNA genes from the analyzed mat samples: A green (external layer) (56.5°C); B yellowish mat (50.8°C); C bright-red (internal layer) (56.5°C) underneath layer A. Migration distances correspond to those reported on Table 1

Full size image

Bacterial fingerprints

Both DNA- and RNA-based PCR-DGGE analysis of the bacterial communities in Boekleung Hot Spring mats (Fig. 3) confirmed the dominance of the Cyanobacteria detected using cyanobacterial specific primers as reported above. Nevertheless, a relative quantification of the major bands detected in the community fingerprints showed other bacterial groups, besides the Cyanobacteria, as important constituents of these communities, both metabolically active members of the communities (analysis based on RNA) and microorganisms present in these communities (analysis based on DNA). Banding patterns of bacteria confirmed that the three studied mat layers were represented by different bacterial communities which was corroborated by statistical comparisons (P < 0.001). Comparisons of banding patterns from DNA- and RNA-based analyses (Fig. 3) also showed significant differences (P < 0.05) between the structure of the bacterial communities when the total and the metabolically active bacterial profiles were compared. However, most bands detected by DNA analysis were also significant constituents of the metabolically active microbial communities (detected by RNA-based analysis) in the three mat layers under study. The quantitative proportion of these bands in the community differ between the DNA- and RNA-based analyses leading to significant differences between the community structure of the total (DNA-based) and metabolically active (RNA-based) microbial communities (Table 1).

Fig. 3
figure 3

Comparative DNA and RNA based molecular fingerprints of bacteria obtained by DGGE based on 16S rRNA genes from the analyzed mat samples: A green (external layer, 56.5°C); B yellowish mat (50.8°C); C bright-red (internal layer, 56.5°C) underneath layer A. Position of migration markers are indicated by arrows. Migration distances correspond to those reported on Table 1

Full size image

Results of the quantification of identified bands from DGGE analyses (Fig. 3) is presented in Table 1 for the green (a), green-yellowish (b), and red (c) mat layers. Based on RNA analyses, the phylotype showing the highest metabolic activity in the green mat layer (a) belonged to the Cyanobacteria, Orden Chroococcales, and was related to the genus Thermosynechococcus representing 15.2% of the quantified RNA (Table 1). Based on DNA, PCR-DGGE profiles showed Roseiflexus-related Chloroflexi as the most representative components of the total bacterial community (DNA-based) of the green mat layer (a) followed by a member of the Candidate Division OP10 and a Synechococcus-related sequence (Cyanobacteria).

PCR-DGGE profiles from the yellow mat layer (b) showed a Thermosynechococcus-related cyanobacterium and a member of the Bacteroidetes as the most important metabolically active components of this mat (Table 1) with over 10% of the total quantified bands from RNA-based profiles. Based on DNA, the total bacterial community was mainly represented by Thermosynechococcus and Synechococcus-related Cyanobacteria (over 15 and 12% of total quantified DNA) and different members of the phylum Chloroflexi.

The red mat layer (c) was dominated by different metabolically active Cyanobacteria and a Roseiflexus-related Chloroflexi (Table 1) as judged by RNA-based PCR-DGGE profiles. Based on DNA, PCR-DGGE profiles showed a major band corresponding to the genus Roseiflexus (Chloroflexi) which represented over 21% of the quantified DNA.

Table 2 summarizes the composition of the studied communities classified in major bacterial groups. These results confirm Cyanobacteria as the dominant phylum in the studied mat layers with average percentages of 37 and 31% of total quantified DNA and RNA, respectively. Chloroflexi was the second major component averaging 28 and 20% of DNA and RNA, respectively. Members of the Candidate Division OP10 and Bacteroidetes represented significant fractions of DNA and RNA (between 5 and 11%) while bacteria within the Planctomycetes and Actinobacteria constituted lower fractions in these community fingerprints. Archaea and Eukarya sequences were not detected through PCR amplification from extracted DNA and RNA.

Table 2 Composition of the bacterial communities in the green, yellow, and red layers of the studied mat based on analysis of banding patterns obtained by PCR-DGGE based on DNA and RNA
Full size table

Phylogenetic relationships

Phylogenetic trees showing the relationship between the retrieved sequences and their closest relatives were constructed for the major bacterial groups detected during this study. Thermophilic Cyanobacteria detected in Boekleung Hot Spring mats grouped in three major distinctive clusters (Fig. 4a). Cluster I was related to Cyanobacteria from other thermophilic sites (i.e. EF032786; 92–94% similarity in the 16S rRNA gene sequence) and relatively divergent from the Cyanobacteria at hot springs in Northern Thailand (i.e. EF452004). This cluster I was constituted mainly by Cyanobacteria from the yellow zone (b) of the mat suggesting a differential distribution of Cyanobacterial phylotypes at the different colored mat layers. Some cluster II and III sequences were only retrieved from the green (a) and red (c) layers matching the cyanobacterial community profiles by DGGE (Figs. 2, 3) although some representatives of the yellow mat layer (b) were also detected. Cluster II was closely related to Cyanobacteria from hot springs (i.e. DQ131174; 95–99% similarity in their 16S rRNA gene sequences) in Asia and America and were grouped with Thermosynechococcus elongatus BP-1. Cluster III represented a divergent group of Cyanobacteria found at Boekleung Hot Spring showing 92–94% similarity with their closest homolog. Sequences EU376442 and EU376417 represented independent lineages of Cyanobacteria within Synechococcus-like clades.

Fig. 4
figure 4

Phylogenetic tree showing the relationships between sequences obtained from Boekleung Hot Spring mats and closely related Cyanobacteria (a), Chloroflexi (b), Candidate Division OP10 (c), and Bacteroidetes (d). The mat layer of origin for each OTU is reported between brackets together with its migration distance during DGGE analysis. Bar indicate changes per nucleotide

Full size image

Within the Chloroflexi, the detected sequences were grouped in two lineages (Fig. 4b). One group is closely related to the genus Roseiflexus. These sequences showed a 95% similarity in the 16S rRNA gene sequences with respect to their closest homolog Roseiflexus castenholzii. Chloroflexi corresponding to this cluster were found in the three studied mat layers. The phylotypes corresponding to this cluster, especially EU376438 and EU376445, were the most representative components of the Chloroflexi in the studied mat layers. A second group includes sequences related (96% similarity in their 16S rRNA gene sequences) to Chloroflexi from geothermal springs at the Tibet area (i.e. DQ001387) and a thermophilic bioreactor (i.e. AY297975).

Closely related sequences belonging to the Candidate Division OP10 have been mainly detected in the green (a) and red (c) mat layers. Their closest homologs (96–97% similarity) are uncultured bacteria from hot springs at Yellowstone National Park (i.e. AF445745 and AF445740) and other thermophilic bacteria within this division (Fig. 4c).

Sequences belonging to the Phylum Bacteroidetes from the studied mat layers clustered with others from geothermal zones at Yellowstone National Park (i.e. AF027008; 91% similarity), China (i.e. DQ340754; 98% similarity), Japan (i.e. AB113584; 96% similarity), Austria (i.e. AM902639), and a hot spring in Ratchaburi Province in Thailand (AY555776) (Fig. 4d).

Discussion

Community fingerprinting by PCR-DGGE provides a representation of the major components of a microbial community (Roeselers et al. 2007; Portillo et al. 2008). In this study, bacterial communities from freshwater hot spring mats were analyzed based both on DNA and RNA aiming to determine the most abundant microbial groups and those actively participating in the development of the different mat layers, respectively.

Several authors have commented on the dominance of Cyanobacteria or Chloroflexi in different thermophilic mats (Ward et al. 1998; Roeselers et al. 2007; Polerecky et al. 2007). During this study, phototrophic bacteria, Cyanobacteria and Chloroflexi, represented the two most abundant and metabolically active members of the analyzed mat communities. Besides, different mat layers were characterized by distinctive bacterial communities, including the members of these two phyla (Cyanobacteria and Chloroflexi). Direct interactions among light-dependent microorganisms, primary producers (Cyanobacteria) and photoheterotrophs or chemoheterotrophs (Chloroflexi), could be hypothezised for Boekleung Hot Spring mats. A coexistence of Cyanobacteria and Chloroflexi at the studied mat layers (Table 2) could represent a preliminary evidence for this mutualism.

Previous studies on the cyanobacterial components of thermophilic mats from Northern Thailand have reported on the presence of members of the genus Thermosynechococcus detected by culturing techniques (Hongmei et al. 2005; Sompong et al. 2005). In Boekleung Hot Spring mats, Thermosynechococcus-related Cyanobacteria were an abundant group at the three studied mat layers. Besides, clear differences in the distribution of cyanobacterial phylotypes were observed. At this respect, for instance, Allewalt et al. (2006) suggested the existence of clear preferences for the level of light intensity and temperature by thermophilic Cyanobacteria belonging to different Synechococcus-like clades, which could explain those differences in cyanobacterial distribution. Kanokratana et al. (2004) showed a scarce representation of Cyanobacteria in sediments from a hot spring in Ratchaburi Province. This information together with the detection during our study of Cyanobacteria through RNA-based analyses suggested that these thermophilic Cyanobacteria are adapted to thrive in hot spring mats rather than the sediments or surrounding waters and the thermophilic mats are where they optimally develop and find their preferred habitat.

Chloroflexi also require specific conditions for growth depending on oxygen, sulfide, and light intensity (Pierson and Castenholz 1974; Hanada et al. 2002). The present study shows dominance of metabolically active Chloroflexi and Cyanobacteria in three mat layers at Boekleung Hot Springs. Results from Kanokratana et al. (2004) studying the sediments of a hot spring at Ratchaburi Province suggested the presence of a low proportion of Chloroflexi. Some previous works have reported a high abundance of Chloroflexi (Polerecky et al. 2007) in thermophilic mat communities while others suggested a scarce representation of this bacterial group (Roeselers et al. 2007). The present study confirms Chloroflexi as a major group of metabolically active bacteria in Boekleung Hot Spring mats and reports on the preferential distribution within the mat of specific members of the Phylum Chloroflexi which tend to avoid the surrounding waters and sediments. A preferential and distinctive distribution of Cyanobacteria, Chloroflexi, and other bacteria confirms the existence of differential community structures between the studied mat layers.

Kanokratana et al. (2004) detected the presence of 5% of phylotypes from DNA analysis belonging to the Candidate Division OP10 while our results suggested the presence on average of around 8–10% of bacteria belonging to this phylum based on DNA and RNA analyses. Since the OP10 sequences detected in both studies belonged to two different clusters (i.e. AY555775) within the phylum (Fig. 4c), a differential distribution between mat and sediments for distinct phylotypes could be suggested. In the present study, the green mat layer (a) represented the preferred environment for members of the Candidate Division OP10 where they constituted up to 15% of the quantified nucleic acids (Table 2). Besides showing a differential distribution of phylotypes belonging to the Candidate Division OP10 in Boekleung Hot Spring mats, we have demonstrated their significant contribution to mat activity since they were metabolically active members of the microbial communities in the analyzed mat layers.

Most Bacteroidetes detected during our study were related to others previously reported in thermophilic environments. From those detected by Kanokratana et al. (2004) in hot spring sediments only one of the sequences (i.e. AY555776) was closely related to the Bacteroidetes related sequences (i.e. EU376449) detected in Boekleung mats. Results confirmed a significant and characteristic representation of Bacteroidetes in hot spring habitats from Ratchaburi Province.

Planctomycetes was another bacterial phylum well represented in our study in mats and that of Kanokratana et al. (2004) in sediments; both studies suggest a differential distribution of phylotypes of this phylum between mats and sediments and the present study indicates they might constitute 5 or 10% of DNA or RNA-based molecular surveys, respectively. The presence of Crenarchaeota, related to the low-temperature clusters, in hot spring sediments (Kanokratana et al. 2004) and their non-detectable presence in mats (this study) also constitutes evidence supporting the existence of differences in the distribution of Archaea in hot spring sediments and mats.

Some studies on hot springs have reported the presence of the Thermus group (e.g., Meiothermus, Thermus) as a significant component of these communities (Ward et al. 1998; Skirnisdottir et al. 2000). For instance, Kanokratana et al. (2004) reported 11% of sequences belonging to this group in sediments from hot springs in Thailand. In the mats analyzed in this study, this group was not detected using molecular techniques although representatives of the genus Meiothermus could be obtained by culturing methods (data not shown). These results suggest the Thermus group is present in these hot springs (i.e. sediments) but represents only a minor component of the mat communities. The strict physico-chemical and biological conditions existing in this environment govern its microbial diversity resulting in communities typically characterized by relatively low diversity dominated by a few microbial groups (Skirnisdottir et al. 2000).

Heterotrophic bacterial communities in the studied hot spring mats showed relatively low phylum diversity although each of these bacterial groups were constituted by numerous, distinctive, phylotypes. The number of bacterial divisions (6; Table 2) found at Boekleung Hot Spring mats is slightly below the values previously reported in similar systems. For instance, Kanokratana et al. (2004) detected up to 11 bacterial phyla in sediments of nearby hot springs. Ward et al. (1998) mentioned about nine bacterial divisions in Octopus Spring mat community (Yellowstone National Park). Interestingly, Proteobacteria and Firmicutes remained undetected through nucleic acid analysis (both DNA and RNA) in this study. Other studies reported over 80% Proteobacteria based on DNA analysis in geothermal communities (Moyer et al. 1995). In hot spring sediments from Ratchaburi Province, Proteobacteria represented about 12% of 16S rDNA sequences (Kanokratana et al. 2004) and their presence was considered as potential contamination from surrounding systems. Proteobacteria and Firmicutes generally constitute a major component of bacterial communities in soils (Fierer et al. 2007). The absence of Proteobacteria and Firmicutes within the detected microorganisms through DNA and RNA analyses suggests Boekleung Hot Spring mats as a system with low influence from external bacterial communities. Thus, Boekleung Hot Spring mats appeared as a relatively closed system as previously suggested for some thermophilic Cyanobacteria (Papke et al. 2003) and opposed to open volcanic environments presenting high influence from allochthonous microorganisms (Portillo and Gonzalez 2008a).

This study confirms the existence of differential bacterial communities constituting green, yellow, and red mat layers at Boekleung Hot Springs. Cyanobacteria and Chloroflexi clearly dominated these communities. Through DNA- and RNA- based analyses the major components present and participating in the activity of the microbial mat have been identified representing diverse communities of bacteria belonging to a relatively low number of phyla.