Intriguing microbial diversity associated with metal-rich particles from a freshwater reservoir


*Corresponding author. Department of Environmental Sciences, Geology 2217, University of California Riverside, Riverside, CA 92521, USA. Tel.: +1 (909) 787 2704; Fax: +1 (909) 787 3993.


During the late summer to early fall, Horsetooth Reservoir in Fort Collins, CO, USA is fully stratified and exhibits seasonally high fluxes of iron, manganese, and metal-rich particles into the water column. Particles were collected from the mid-region of the hypolimnion and examined for metal content. Nucleic acids extracted from the particles were used to construct bacterial and archaeal 16S rDNA clone libraries. Surprisingly, 50% of cloned bacterial genes were closely related to a coherent cluster within Candidate Division OP10. To our knowledge, this is the first report of an environmental gene clone library that exhibits a dominance of OP10-related clones. Several other sequences, many with long branch lengths, clustered within eight separate bacterial divisions and the diatom chloroplasts. Most of these divisions are commonly found in freshwater environments. However, gene sequences from characterized metal-oxidizing or metal-reducing bacteria were not identified. The archaeal gene clone libraries contained diverse sequences, most with close homology to previously characterized gene clones of methanogens or uncultivated Crenarchaeota from soil and lacustrine environments. This study identified a unique environment where OP10 bacteria are potentially abundant. Furthermore, we demonstrated that the metal-rich particles from this reservoir support a diverse and interesting community of microorganisms.


Metal cycling occurs frequently in environments containing an aerobic/anaerobic interface, such as freshwater lakes and marine sediments [1,2]. These activities are critical for a variety of geological and environmental processes, such as biomineralization and organic carbon turnover [3,4]. The cycling of metals between reduced and oxidized states, with the ensuing changes in their solubility, often involves microbial activities [1,5]. However, the identity and physiology of microbial communities involved in metal cycling remain poorly understood.

Horsetooth Reservoir is located in a sandstone formation, which is rich in iron and manganese oxides. In the summer months, stratification of the water column leads to development of anoxic conditions in the hypolimnion. One consequence of this anoxia is anaerobic respiration of metal oxides, releasing large amounts of reduced iron and manganese into the water column [6]. As reduced metal diffuses upward and reaches the oxic region of the water column, particles rich in metal oxides begin to form. Removal of manganese from reservoir water has been an ongoing challenge for the City of Fort Collins Water Treatment Facility. To further our understanding of seasonal metal infiltration into the reservoir, we instigated a study of microbial diversity associated with metal-rich particles that form in the lake.

Molecular phylogenetic analyses of 16S ribosomal RNA genes are increasingly expanding our awareness of microbial diversity [7]. Environments examined within the past several years, such as hot springs [8], ultra-oligotrophic lakes [9], flooded soils [10], mobile mud deposits [11], contaminated aquifers [12], and activated sludge [13], have uncovered the existence of novel lineages of Bacteria or Archaea. Although the known molecular diversity of microbes is continually increasing, little is known about the metabolic diversity within these new phylogenetic lineages. 16S rDNA clones that do not associate with previously characterized phylogenetic clusters are usually only minor components of environmental gene clone libraries, making physiological studies of these organisms virtually impossible to conduct. Thus, environments that appear to have substantial populations of novel microbial lineages offer precious opportunities for attempted cultivation or for constructing large insert gene clone libraries to determine partial genome sequences of uncultivated organisms [14].

Here, we continue our previous studies of microbial diversity associated with metal-rich particles from freshwater systems [5]. In this particular system, we report an unusual dominance of cloned 16S rRNA genes related to the bacterial Candidate Division OP10. To date, no members of OP10 bacteria have been cultivated, and nothing is known about their physiology. Our study suggests that metal-rich particles from Horsetooth Reservoir may provide an interesting and fruitful environment for furthering our understanding of OP10 bacteria as well as other diverse microbial groups.

2Materials and methods

2.1Site description

Horsetooth Reservoir is located 5 km to the west of the city of Fort Collins, CO, USA with the north-most point at 40°35.4′N, 105°10.3′W. The maximum length of the reservoir is ca. 10.6 km, the maximum width is ca. 1.5 km, and the maximum depth is ca. 65 m when filled to capacity [6].

2.2Collection of particles

Water was collected from an outlet 16 m above the bottom of the reservoir through a pipeline located in the City of Fort Collins Water Treatment Facility. Particles from the water were continuously collected over an 18-h period on a nominal 1-μm pore size Versapor filter (Pall-Gelman) attached directly to the pipeline outflow at the treatment facility. The concentrated particles were washed off of the filter with lake water, and total material was collected by centrifugation for 10 min (at 10 000×g). The pellet was shipped overnight on dry ice to Pasadena, CA, USA, and immediately frozen at −80°C until further analysis.

2.3Limnological analyses

Measurements were made by the City of Fort Collins Water Quality Laboratory from water samples collected at the pipeline outlet. Units are reported as mg l−1 or μg l−1 unless indicated otherwise, since both soluble and solid phases were measured simultaneously. Temperature, pH, and dissolved oxygen (DO) were measured by individual probes connected to the pipeline outlet. Turbidity was measured by the nephelometric method using a formazin polymer suspension as a standard reference [15]. Ammonia concentration was measured by the automated phenate method using alkaline phenol, hypochlorite, and sodium nitroprusside to catalyze a colorimetric reaction with ammonia [15]. Nitrate was measured directly by ion chromatography. Chlorophyll a was measured fluorimetrically. Total phosphate was analyzed by reaction with ammonium molybdate and antimony potassium tartrate under acidic conditions, followed by reduction with ascorbic acid in a colorimetric reaction [15]. Microscopic counts were conducted from a calibrated amount of filtered water (Envirocheck filter, Pall Gelman). Total particles were enumerated on a per ml basis by averaging counts from 10 fields of view. Photographs of particles imaged with differential interference contrast optics were taken with a camera (onto 35 mm Kodak film) directly attached to the microscope.

2.4Analysis of metals

Manganese, iron, and copper contents of unfiltered lake water were analyzed by atomic absorption spectrometry without sample filtration at the Fort Collins Water Quality Laboratory [15]. The collected particles (described above) were thawed and sedimented (10 000×g), and the resulting supernatant was analyzed for manganese, iron and copper by inductively coupled plasma atomic emission spectroscopy (ICP-AES) using single-element standards (J.T. Baker) [15]. Ascorbate-extractable iron and manganese from the particles were also determined, as described previously [5].

2.5Clone library construction

Sub-samples of the stored particles were thawed and centrifuged and the supernatant was discarded. Two separate samples were processed for the bacterial 16S rDNA clone libraries (libraries referred to as HTL1 and HTL2 below) and one sample was processed for the archaeal gene clone library. Total nucleic acids were extracted from 2 g of each sample using the FastPrep® kit (Bio 101, Vista, CA, USA) according to the manufacturer's instructions. The template was diluted 10-fold to avoid inhibition of the polymerase chain reaction (PCR) due to the presence of metal ions. PCR products were produced by amplification with either the bacterial-specific primer pair 27F (AGA GTT TGA TCM TGG CTC AG) and 1518R (AAG GAG GTG ATC CAN CCR CA) [16], or the archaeal-specific primer pair A21F (TTC CGG TTG ATC CYG CCG GA) and A958R (YCC GGC GTT GAM TCC AAT T) [17]. Reaction conditions were as described previously [5]. Amplified products were ligated into the pCR2.1 vector and cloned into competent Escherichia coli cells as per the manufacturer's instructions (Invitrogen). Plasmid-containing colonies were randomly selected and 16S rDNA inserts were amplified and subjected to restriction digestion with either HhaI or RsaI (37°C for 3 h). Restriction fragment length polymorphisms (RFLP) were analyzed on agarose gels and similar patterns were grouped together.

2.6Phylogenetic analysis

One or two representatives from each RFLP group were bi-directionally sequenced and phylogenetically analyzed. Bi-directional sequencing was completed by MWG Biosciences (High Point, NC, USA) using both LiCor 4200 and Applied Biosystems 377 automated gene sequencers. Full-length gene sequences were compiled using STADEN software [18]. Sequences were manually aligned to their nearest relatives using the program ARB (1999 release plus manual uploading of closely related sequences based on individual BLAST results) [19]. Only full-length sequences were included in the analysis. Alignments were visually checked using the Genetic Data Environment (GDE) version 2.0 sequence analysis software (Steve Smith, Millipore Corporation), as described elsewhere [20]. Regions of sequence with ambiguous alignment were not included in the phylogenetic analysis. No chimeric sequences were identified using the program CHECK_CHIMERA (Ribosomal Database Project) [21]. Phylogenetic inferences were determined by neighbor-joining analysis with Kimura two-parameter genetic distances (2:1 transition:transversion ratio). Neighbor-joining and maximum parsimony bootstrap values were calculated separately from 100 re-sampled data sets [22]. Sequence data from this project are within the GenBank database under accession Nos. AF418924–AF418977.

3Results and discussion

3.1Appearance of metal-rich particles

The turbidity and metal concentration in Horsetooth Reservoir greatly increase during stratification of the water column from late summer through autumn months [6]. Water and particles were collected for analysis from an intake 16 m above the reservoir floor – approximately the mid-region of the hypolimnion. At the time of sample collection, there was almost twice as much total iron as manganese in the water column, and filamentous particles were beginning to appear (Table 1). These large filamentous particles (ca. 10 μm diameter; Fig. 1) have been observed and monitored by the City of Fort Collins at the sample intake point since the late 1980's. During stratification in 1999, filamentous particles first appeared at sample collection depth in mid-August just following the onset of stratification in mid-July (Fig. 2). The numbers of filamentous particles peaked in mid-September, and disappeared completely by the end of January following turnover of the reservoir in mid-October.

Table 1.  Characteristics of reservoir water
  1. aMeasurement techniques described in Section 2.

  2. bReported error represents standard deviation of 3–7 samples.

  3. cNephelometric turbidity units.

  4. dMetal concentrations in water prior to filtration measured by AAS.

  5. eMetal concentrations measured in sample after sedimentation of particles by ICP-AES.

  6. fN.D., not determined.

  7. gCounts reported for Aug. 1998 only.

MeasurementaAug. 1999Avg. Aug. 1992–98
Temperature (°C)14.0014.5±2b
Turbidity (NTU)c6.703.55±1.2
Dissolved oxygen (mg l−1)6.015.2±1.2
Ammonia (mg l−1)0.020.02±0.02
Nitrate (mg l−1)0.210.15±0.02
Chlorophyll a (μg l−1)2.330.12±0.14
Total phosphate (mg l−1)0.020.03±0.01
Total Mn, raw waterd (μg l−1)93.0057.00±23
Total Mn, samplee (μg l−1)213.00N.D.f
Total Fe, raw waterd (μg l−1)162.00211.00±60
Total Fe, samplee (μg l−1)998.00N.D.f
Total Cu, raw waterd (μg l−1)12.006.00±1.5
Total Cu, samplee (μg l−1)106.00N.D.
Total particles (number ml−1)72073490g
Filamentous morphotypes (number ml−1)601516g
Planctomycetales (number ml−1)8647g
Figure 1.

Micrograph of particles collected from filtered reservoir water using differential interference contrast optics. The scale bar is divided into 1-μm increments. The objects of ca. 10 μm diameter indicated by the arrows are the ‘filamentous particles’.

Figure 2.

(A) Profile of total iron, (B) profile of total manganese, (C) profile of total particles (•) and filamentous particles only (□). All data were collected at the sample collection point from January 1999 to January 2000 as described in Section 2. Arrows indicate sampling date.

The bulk of total particle formation in the reservoir occurred much earlier in the season, before stratification (Fig. 2). These earlier-forming particles were a mixture of algae, diatoms, and amorphous forms of 5 μm or greater. However, unlike the majority of particles in the reservoir, the appearance of the filamentous particles always coincides with seasonal increases in total manganese in the water column, and they always decrease following turnover of the reservoir. This pattern indicates an association of the filamentous particles with metal cycling in this system. Similar filamentous particles have been described in numerous freshwater systems that undergo stratification and that support the cycling of both iron and manganese [23–26]. These same studies suggest that the formation of such particles requires the presence of a biological agent.

In preparation for phylogenetic studies, stored particles (containing ca. 8% filamentous particles) were thawed and collected by centrifugation. Metal content of the resulting supernatant was much more concentrated in iron, manganese, and copper than unfiltered reservoir water (Table 1). This was likely due to leaching of loosely associated metal ions, especially iron, from the particles. Extraction with ascorbate indicated that the particles still retained 0.9% iron and 2.7% manganese by dry weight (data not shown), suggesting that manganese is more tightly associated with the particles than iron. In other lacustrine environments, the formation of manganese oxide-containing particles at oxic/anoxic interfaces is thought to be governed by the activities of manganese-oxidizing bacteria; however, it is unclear whether the same is true in this particular system [27].

3.2Phylogenetic diversity: bacteria

The diversity of bacterial populations associated with the particles was investigated through phylogenetic comparisons of cloned 16S rDNA sequences to related sequences in the GenBank database. Bacterial libraries were generated from two separate sub-samples to verify the homogeneous distribution of populations within the total collection of particles. A total of 96 clones from library HTL1 and 48 clones from library HTL2 were randomly selected and analyzed. In HTL1, 12 clones did not contain 16S rDNA inserts, and four had no inserts in HTL2. Thus, we analyzed a total of 84 and 44 clones by RFLP and sequenced 22 and six genes from HTL1 and HTL2, respectively (Table 2). Analyzed gene clones clustered within eight established bacterial divisions plus the diatom chloroplasts (Table 2; Figs. 3 and 4). Clusters related to the OP10 candidate division, the Planctomycetales, the Actinobacteria, the low G+C Gram-positives, the α-Proteobacteria, the β-Proteobacteria, and the chloroplasts were identified in both libraries (Table 2). Of the remaining clusters, only the γ-Proteobacteria were well represented with five clones in library HTL1, whereas only single clones related to the Verrucomicrobia, Green Non-Sulfur, and Cytophagales were identified in either HTL1 or HTL2. Based on the overall composition of clones identified in both HTL1 and HTL2, it appears that the majority of bacteria were homogeneously distributed throughout the particles.

Table 2.  Abundance of RFLP groups and phylogenetic relationships of cloned bacterial 16S rRNA genes
  1. aFrom Figs. 3 and 4.

  2. bA total of 84 gene clones were analyzed from library HTL1 and 44 from library HTL2. Abundance of clones within each library is based on groupings of RFLP patterns as described in Section 2.

  3. cSimilarity determined by neighbor-joining algorithm.

  4. d>0.97=same species [36].

Group affiliationClone designationaAbundancebNearest relativecSimilarity indexd
Candidate Division OP10HTB74321Clone GC55 (AJ271048)0.89
 HTD12  same as above0.89
 HTG11  same as above0.82
PlanctomycetalesHTA210Crat. Lake CL500–3 (AF316767)0.85
 HTC1211HTA2 (AF418943)0.86
VerrucomicrobiaHTBB201Verruco. spinosum (X90515)0.91
Green Non-SulfurHTH410Clone SBR 2022 (X84565)0.81
ActinobacteriaHTG51010Crat. Lake CL500–29 (AF316678)0.93
 HTG6  same as above0.93
 HTDD301Mycobacterium sp. (U30661)0.99
Low G+C Gram-pos.HTE552Clostridium fallax (M59088)0.96
ChloroplastHTAA824Sk. pseudocostatum (X82155)0.93
 HTB2  same as above0.99
 DD6  same as above0.99
CytophagalesHTF210Chrys. joosteii (AJ271010)0.92
γ-ProteobacteriaHTB1050Uncult. γ-Proteobact. (AF445683)0.95
 HTA410Legionella sp. IA55 (AB058909)0.92
β-ProteobacteriaHTA1020Rhodoferax fermentans (D16211)0.97
 HTA110Burkholderia sp. N3P2 (U37344)0.88
 HTBB601Chromo. fluviatile (M22511)0.87
 HTCC211Bordetella avium (U04947)0.96
 HTG2  same as above0.93
α-ProteobacteriaHTC610Uncult. sludge bact. (AF234717)0.97
 HTF110HTC6 (AF418951)0.89
 HTB950Sphingo. sp. B28161 (AJ001052)0.97
 HTD1012Agro. tumefaciens (D01257)0.96
 HTH610Env. Isolate LD12 (Z99997)1.00
 HTD910Green Bay MNG7 (AF292997)0.88
Figure 3.

Phylogenetic positions of particle-associated, bacterial, 16S rDNA sequences. GenBank accession numbers are indicated for all sequences. These unrooted phylogenetic trees were constructed using the neighbor-joining method with the Kimura two-parameter model for nucleotide change, as described in Section 2. All non-ambiguously aligned nucleotide positions were considered in the analyses and are included parenthetically: (A) OP10 (1309), (B) Planctomycetales and Verrucomicrobium (1189), (C) Green Non-Sulfur (1230). Bootstrap values (100 replications each) are shown using the neighbor-joining algorithm above and the maximum parsimony algorithm below the nodes (**values from 80 to 100%; *values from 50 to 79%). Only values over 50% are marked. Fully italicized names represent sequences from bacterial isolates, whereas all others represent environmental gene clones.

Figure 4.

Phylogenetic position of particle-associated, bacterial, 16S rDNA sequences. This phylogenetic tree was constructed similarly to the trees in Fig. 3. The outgroup 16S rDNA sequence used for this tree was from Sulfolobus acidocaldarius (D14876). All non-ambiguously aligned positions (1041 nucleotides) were analyzed. Phylogenetic inference, bootstrap analyses, accession numbers, and naming conventions are as in Fig. 3.

The most striking result from the bacterial gene clone libraries was that the vast majority of recovered genes (51% in HTL1 and 48% in HTL2) were most closely related to a single cluster within Candidate Division OP10 (Fig. 3A) [13,28]. OP10 gene clones were originally characterized in 16S rDNA libraries from Obsidian Pool, Yellowstone National Park [8]. Since then, sequences related to this group have appeared as minor components in 16S rDNA clone libraries from numerous and diverse environments [9,12,13,28–32]. The sequenced clones were very closely related to each other (similarity index from 0.92 to 1), but less related to their nearest neighbor (similarity index from 0.82 to 0.89) (Table 2). Thus, our sequences represent a new sub-group within a previously identified cluster of OP10, which is currently composed of genes cloned from sludge systems [28]. It is well known that extraction, amplification, and cloning of genes from environmental samples are not necessarily quantitative in relation to actual members of the target population [33]. However, the extensive dominance and equivalent abundance of OP10 clones in both HTL1 and HTL2 indicate that the organisms were likely abundant throughout the collected particles.

No member of the OP10 group has ever been cultivated, and, therefore, no physiological function has been ascribed to any branch of the group. The possible abundance of OP10-related bacteria makes Horsetooth Reservoir an ideal site for future attempts at cultivation and characterization of members of this group. Additionally, the association of OP10 gene clones with particles formed during the active cycling of manganese suggests that metal cycling could be a reasonable physiological role for this particular OP10 cluster. Alternatively, since the sample was taken early in the metal-cycling period, the OP10 bacteria may thrive just before the large influx of iron or other factors into the system. Other factors potentially dictating the composition of the microbial community include large concentrations of chlorophyll a, copper, and manganese that appeared during August 1999 in comparison to average values for August from 1992 to 1998 (Table 1). Further sampling expeditions over multiple stratification seasons will be required to verify correlations between OP10 appearance and chemical or physical factors in the reservoir.

Aside from the OP10 group, several other interesting gene clusters were identified in the bacterial clone libraries. A deeply branching cluster related to the Planctomycetales and one related to the Verrucomicrobia were identified (Fig. 3B). Additionally, a deeply branching sequence clustered with members of the Green Non-Sulfur Bacteria (Fig. 3C). Gene clones clustering with one or more of these phyla have been identified in other freshwater systems [9,34]. The Planctomycetales cluster associated with Horsetooth Reservoir particles was most closely related to gene clones from ultra-oligotrophic Crater Lake (Table 2). The findings from the present study along with those from other freshwater systems imply that these three bacterial groups may be common to freshwater systems, although the sequence diversity within each group is quite vast.

The second most abundant group in our analysis (16%; Table 2) clustered within the Actinobacteria, specifically with a sequence from Crater Lake [9]. This group also appears to be cosmopolitan in freshwater systems, as are all of the other clusters identified in this study (Fig. 4), except for low G+C Gram-positives and OP10 [9,34]. Thus, the bacteria associated with particles in Horsetooth reservoir have similar phylum-level diversity to other freshwater systems, but the overall population is unique in terms of potential abundances and sequence diversity within specific groups. For example, many of our sequences had deep branch lengths and low similarities to their nearest neighbors (<0.9), e.g. HTF1 and HTD9 in the α-Proteobacteria, HTA1 and HTBB6 in the β-Proteobacteria, and all of the sequences related to Candidate Division OP10, Planctomycetales, and Green Non-Sulfur Bacteria (Figs. 3 and 4; Table 2). Long branch lengths often indicate the presence of novel taxonomic clusters, which are commonly observed in gene clone libraries from complex environments with multiple, interacting, chemical gradients or with unusual chemistry [8,9,11]. The presence of several deeply branching clusters in our analysis suggests that the metal-rich particles in this system are chemically complex and sustain a unique, diverse, microbial community [7].

3.3Phylogenetic diversity: Archaea

A separate 16S rDNA gene clone library was analyzed to determine diversity of Archaea in association with the metal-rich particles. A total of 96 clones were randomly selected and analyzed: 13 clones contained no inserts, the remaining 83 were grouped by RFLP, and 17 representatives were sequenced. Phylogenetic analysis revealed two dominant groups: the uncultivated Crenarchaeota (36 clones) and the methanogens (47 clones) (Table 3; Fig. 5). This level of diversity was similar to our previous study of archaeal populations associated with ferromanganese particles from freshwater sediments [5]. The majority of Crenarchaeota-related gene clones (83%) were closely related to genes cloned from soil environments (Table 3) [35]. The majority of clones clustering with the methanogens (81%) were related to clones from freshwater systems (Table 3). All archaeal gene clones were related to sequences originally identified from soil or lacustrine samples; none associated with taxa from either marine or hydrothermal systems. These results suggest that Archaea in this system are likely endemic to freshwater and terrestrial systems and not cosmopolitan among all environments.

Table 3.  Abundance of RFLP groups and phylogenetic relationships of cloned archaeal 16S rRNA genes
  1. aFrom Fig. 5.

  2. bDetermined by RFLP patterns as described in Section 2.

  3. cSimilarity determined by neighbor-joining algorithm.

  4. d>0.97=same species [36].

  5. eBased on GenBank submission information.

Clone designationaAbundanceb (out of 83 clones)Nearest relativec by similarity indexSimilarity indexd(avg. for mult. clones)Environmente
HTA-C5/E718SCA1145 (U62811)0.99soil
HTA-H88FRD9 (AY016505)0.99alpine tundra soil
HTA-B103vandinDC69 (U81774)0.96anaerobic digestor
HTA-F43ARC12 (AF293019)0.98Green Bay (freshwater)
HTA-G62SCA1166 (U62816)0.96soil
HTA-B61SCA1170 (U62817)0.98soil
HTA-D61SCA1154 (U62814)0.99soil
HTA-A1/D5181Af–1100Ar (AF056363)0.99Lake Soyang (freshwater)
HTA-C111ARF3 (AF293014)0.98Green Bay (freshwater)
HTA-C75WCHD3–33 (AF050619)0.97contaminated aquifer
HTA-H95WCHD3–16 (AF050618)0.89contaminated soil
HTA-B33ARF3 (AF293014)0.98Green Bay (freshwater)
HTA-F13Methanosarcina siciliae (U20153)0.98anaerobic lake sediment
HTA-B121WCHD3–03 (AF050611)0.82contaminated aquifer
HTA-D21Methanosarcina thermophila (M59140)0.97anaerobic, 55°C
Figure 5.

Phylogenetic positions of particle-associated, archaeal, 16S rDNA sequences. The outgroup 16S rDNA sequence used for this tree was from E. coli (J01695). All non-ambiguously aligned positions (712 nucleotides) were used for the analysis. Phylogenetic inference, bootstrap analyses, accession numbers, and naming conventions are as in Fig. 3.


Overall, our sequence analysis revealed relationships to 11 phylum-level groupings between the Bacteria, diatom chloroplasts, and the Archaea; however, we found no gene clones related to groups of characterized metal oxidizers or reducers. This result is in contrast to our previous study of microbial diversity associated with ferromanganese nodules from freshwater sediments of Green Bay [5]. In this previous study, gene clones related to both metal-oxidizing and metal-reducing bacteria were found associated with manganese- and iron-rich sediment particles. Metal-rich particles in Horsetooth Reservoir are suspended in the water column, which represents an entirely different chemical and physical environment. However, gene clones related to those from Green Bay were identified: a relative of the α-Proteobacteria (HTD9), three related to the Crenarchaeota (HTA-F4), and 14 related to the methanogens (HTA-C1 and HTA-B3). Together, our studies suggest that similar groups of Archaea are found in freshwater, metal-rich environments, while many bacterial groups, such as those responsible for metal cycling, are endemic to specific chemical niches.

Many previous studies of metal-rich particle formation in freshwater systems imply that biological agents are necessarily involved [2]. Because we identified no known metal-cycling organisms in this study, it is possible that: (1) metal-rich particles form abiologically in this system; (2) metal-rich particles form and dissolve through the activities of novel organisms; (3) known groups of metal oxidizers or metal reducers were not sufficiently abundant to be observed with the methods used; and/or (4) organisms responsible for metal respiration were not attached to the surface of the particles. Further studies of microorganisms associated with metal-rich particles in this system will likely address these possibilities. In the meantime, verification of the dominance of OP10-related bacteria during stratification and active metal cycling in Horsetooth Reservoir remains a primary focus.


We wish to thank Sue Martin for water analysis, Myron T. La Duc for assistance with clone library construction, Gary Plett for assistance with the ICP-AES analysis, Brian D. Lanoil for assistance with phylogenetic analysis and interpretation, and two anonymous reviewers for helpful comments and suggestions. This work was supported by NASA grant 050000 and Department of Energy grant DE-FG03-00ER62721 to K.H.N.