For over 3 billion years, stromatolites dominated ancient Earth (Byerly et al., 1986; Grotzinger and Knoll, 1999). Today, however, analogues of these lithified microbial accretions are relegated to only a few locations around the globe. One such location is the margin of Exuma Sound, Bahamas, the only known site of actively forming stromatolites in normal marine salinity (Dravis, 1983; Dill et al., 1986; Reid et al., 1995). Within Exuma Sound, the stromatolites at Highborne Cay represent well-laminated microbialites extending over 2.5 km in the shallow subtidal zone (Figures 1a and b). The Highborne Cay stromatolites form large reef-like structures that can sometimes serve as substrates for eukaryotic colonization (Figure 1c). Five key microbial functional groups have been identified in Highborne Cay stromatolite development including: cyanobacteria, aerobic heterotrophic bacteria, sulfate-reducing bacteria, sulfide-oxidizing bacteria and fermentative bacteria (Visscher and Stolz, 2005). Together, the cell–cell interactions of these microbial groups result in the formation of three major stromatolite community mat types (types 1–3 in Reid et al., 2000). These stromatolite types are distinguished by their taxonomic richness as well as the degree of lithification (Reid et al., 2000; Baumgartner, 2007). Type 1 mats are the most abundant in the Highborne Cay ecosystem and are comprised unlithified carbonate sand grains (that is, ooids) tightly bound by extracellular polymeric secretions (EPS) and filamentous cyanobacteria. The EPS material coupled with the carbonate ooids comprises the initial structural framework for stromatolite integrity (Kawaguchi and Decho, 2002a, 2002b). Earlier studies indicated that type 1 mats represent a trapping and binding stage of stromatolite development (Reid et al., 2000). Type 2 mats are characterized by the presence of a thin calcified biofilm (20–60 μm), which forms a micritic crust on the stromatolite surface. The cyanobacteria in type 2 mats are concentrated immediately below the micritic crust within an unlithified matrix of EPS and carbonate ooids. Type 3 mats are typified by an extensive colonization of the ooids by endolithic cyanobacteria classically identified as Solentia sp. (Macintyre et al., 2000; Reid et al., 2000). The colonized ooids are connected by the endolithic microbes resulting in a fused sand grain layer within the stromatolite (Macintyre et al., 2000). Combined, type 2 and type 3 mats represent approximately one-third of the stromatolite ecosystem (Reid et al., 2000). Both types, however, appear to incur fluctuations in their abundance throughout the year (RP Reid, personal observation). The highly energetic nature of the subtidal zone subjects these cyanobacteria-dominated stromatolites to oolitic sand burial by wave action, and these intermittent burial events are critical in determining mat type and stromatolite accretion (Reid et al., 2000; Bowlin et al., 2008).

Figure 1
figure 1

Stromatolites at Highborne Cay, Bahamas. (a) Google Earth image of Highborne Cay showing the extensive fields of stromatolites on the eastern shore of the island (arrows). (b) Windward coast of Highborne Cay depicting the subtidal (laminated) stromatolites (s) in the fore reef and intertidal (unlaminated) thrombolites (t) in the back reef. (c) Underwater photograph of stromatolites colonized by the eukaryotic algae Batophora. A cutaway (arrow) shows the laminated buildups of stromatolites. (d) Micrograph of a vertical section of the type 2 stromatolites. Cyanobacteria (left arrow) form a greenish brown layer concentrated in the upper 5 mm of stromatolite. Micritic crust (m) is visible as a white precipitate on the stromatolite surface. (e) Photomicrograph of a thin section of the mat, visualizing the network of integrated oolitic sand grains (o), with interstitial spaces filled extracellular polymeric secretions (e) and cyanobacteria (c, arrows). (f) Corresponding epifluorescence micrograph of the same view field showing extensive autofluorescence at 650 nm of cyanobacterial colonization (arrows) between the nonfluorescent oolitic sand grains (o).

In all three stromatolite community types, cyanobacteria are critical for the accretion and early lithification of stromatolites (Pinckney and Reid, 1997; Dupraz et al., 2004). Through the production of EPS material, cyanobacteria trap and bind ooids from the surrounding seawater and incorporate these grains into the nascent stromatolite surface (Visscher et al., 1998; Reid et al., 2000). EPS material has been previously shown to be important in calcium carbonate precipitation (Kawaguchi and Decho, 2000, 2001; Braissant et al., 2007) and is a critical component of stromatolite accretion (Reid et al., 2000). In addition to producing EPS, cyanobacteria exhibit other physiological characteristics considered essential for stromatolite development, such as phototaxis. The stromatolite environment is characterized by chemical (that is, oxygen, sulfide) and physical (that is, light) gradients (Visscher and Stolz, 2005). Cyanobacteria respond to such gradients by a vertical phototactic migratory response to optimize photosynthesis rates and minimize potential cell damage (Prufert-Bebout and Garcia-Pichel, 1994; Nadeau et al., 1999; Fourcans et al., 2006).

Previously, most of the identification of cyanobacterial diversity within Bahamian stromatolites has relied on morphological description. These classical approaches identified Schizothrix gebeleinii (Oscillatoriales) as the major cyanobacterial mat builder of the stromatolites in the Exuma Cays (Golubic and Browne, 1996; Stolz et al., 2001). S. gebeleinii is typified by cells (1 μm wide × 2 μm long) organized into filaments containing one to several cellular trichomes that are encased in a common (1–2 μm wide) polysaccharide sheath.

This study examines the cyanobacterial diversity that comprises type 2 Highborne Cay stromatolites by complementing the classical morphological approach with modern molecular tools, an approach that has been successfully applied to studies of stromatolite microbial diversity in Shark Bay, Australia (Burns et al., 2004). Specifically, the objectives of this study were to identify cyanobacterial species composition in type 2 communities by classical microbial isolation and cultivation-independent molecular analysis, then characterize those isolates for phenotypes known to be essential for stromatolite development. Type 2 stromatolites were chosen for this study as they are considered to be a critical intermediate phase of stromatolite development (Reid et al., 2000). For each isolate, physiological characteristics considered important for stromatolite development such as EPS production and phototactic motility were examined.

Materials and methods

Environmental site description and sample collection

The stromatolites samples were collected from the island of Highborne Cay, Exuma, Bahamas (76°49′W, 24°43′N). All samples were collected from the subtidal stromatolite communities in July 2006. Stromatolite site designations (sites 1–12) were using in accordance to Reid et al. (2000). Identification of the stromatolite type (that is, type 2) was made using morphological criteria as previously described (Reid et al., 2000). Ten cores of stromatolites were collected from the site 8 using an 8 mm Harris Uni-Core Punch (Ted Pella, Redding, CA, USA) and briefly allowed to drip dry to remove the surrounding seawater. To stabilize nucleic acids, stromatolite cores were then immediately placed in RNA later (Ambion, Austin, TX, USA) and frozen at −20 °C. All samples were transported to the Space Life Sciences Laboratory at the Kennedy Space Center and stored at −80 °C until further processing. Additional core samples from the corresponding stromatolites were collected and stored in filtered seawater for subsequent culturing and morphological analysis. All chemicals used in this study were from Sigma-Aldrich (St Louis, MO, USA) unless otherwise indicated.

Cyanobacterial cultivation and imaging

Isolates from type 2 stromatolites were cultivated using various methods as previously described (Castenholz, 1988; Prufert-Bebout and Garcia-Pichel, 1994). Briefly, samples of stromatolite were grown in ASN III 1 ×, or with 5 × nitrate (ASN; Rippka et al., 1979) and incubated at 25 °C under fluorescent cool white lights (40 μE m−2 s−1). Culture stocks were maintained with ASN 5 × NO3, however, HBC7 was also maintained on nitrate-free media. Enrichment cultures were vortexed and then serially diluted onto ASN agar plates and untreated 96-well microtiter plates containing ASN media to further isolate individual cultivars of cyanobacteria. Microorganisms were monitored by light microscopy and photomicrographically documented with a Zeiss Axioplan Microscope (Carl Zeiss, Jena, Germany) and a 100 × plan objective using bright and dark field microscopy and epifluorescence. The cell width of each isolate was measured using the Zeiss MicroImaging software (Carl Zeiss). For each isolate, the diameters of 10 cells from three replicate cell cultures were measured and compared using descriptive statistics (MiniTab, State College, PA, USA).

Transmission electron microscopy

To characterize the ultrastructure of the cultivars and visualize the EPS material, isolates were fixed in 2.5% glutaraldehyde/2.5% paraformaldehyde in a buffer containing 0.1 M sodium cacodylate with 0.45 M NaCl, pH 7.4 at 23 °C for 1 h. The samples were then rinsed in the same buffer three times for 15 min. After rinsing, the samples were postfixed with 1% osmium tetraoxide in the cacodylate/NaCl buffer for 45 min to improve the contrast of the samples. The cultivars were then rinsed as before and then dehydrated in a standard ethanol dehydration series. Specimen were infiltrated with 100% propylene oxide for 15 min then incubated overnight in a 1:1 ratio of propylene oxide and accelerated Spurr (Spurr, 1969) for 72 h. After infiltration the cultivars were then embedded in freshly prepared Spurr at 60 °C overnight. Thin sections of the sample blocks were counterstained with a saturated solution of uranyl acetate and lead citrate, then examined with a Hitachi H7000 transmission electron microscope.

Extracellular polymeric secretions: visualization and quantification

To visualize the extent and nature of exopolysaccharide production in type 2 stromatolite in situ, samples were stained with the Alcian Blue (AB)/periodic acid (PA) kit (Richard Allen Scientific, Kalamazoo, MI, USA). To stain for acidic mucopolysaccharides samples were fixed in a 2.5% glutaraldehyde solution for 1 h, washed with ASN media and then stained with AB pH 2.5 for 30 min. Samples were then rinsed in distilled water (diH2O) and then incubated in PA solution (0.5%) for 5 min and washed with diH2O. To visualize PA staining the cells were then exposed to Schiff reagent for 10 min, washed with diH2O and then examined with bright field microscopy (Zeiss, Germany). PA oxidizes the samples to form aldehyde groups that are then visualized by Schiff reagent. Cells that have stored glycogen, glycoproteins or neutral polysaccharides will appear pink where as those cells that are rich in anionic groups such as acidic mucopolysaccharides or mucins appear blue.

To examine the ability of cyanobacterial stromatolite cultivars to produce EPS material in vitro, cultures of equal biomass were vortexed for 20 min to produce small cell colonies approximately 1–2 mm in diameter. The colonies were grown at 30 °C under low fluorescence white light (30 μE m−2 s−1) with shaking until the average colony diameter of the culture was 2–3 mm (approximately 1 week). Equal amounts of biomass (w/v) were exposed to low (30 μE m−2 s−1) and high (100 μE m−2 s−1) white light in triplicate for 1 week. After light exposure cultures were vigorously stirred at maximum speed to break up colonies and shear extracellular EPS material into the medium. Cultures were then centrifuged at 20 000 g at 4 °C for 4 h. Supernatant was collected and heated at 80 °C for 30 min, then precipitated with cold 100% ethanol and stored overnight at −20 °C. Samples were then thawed and centrifuged at 3000 g for 30 min. Supernatant was discarded and the pellet was resuspended in 100 ml of 95% ethanol. The suspension was again centrifuged at 3000 g and then the pellet was air-dried. The pellet was resuspended in 20 ml of sterile water.

Crude extracted EPS material was quantified using a modified anthrone colorimetric assay (Morris, 1948). Briefly, EPS material was serially diluted to a final volume of 40 μl into an untreated 96-well microtiter plate. To each well, 100 μl of concentrated sulfuric acid containing 123.5 mM of anthrone was added. The plate was then incubated at 92 °C for 3 min and the absorbance was read at 630 nm. Absorbance was compared to a standard curve containing concentrations of 50, 100, 150, 200, 250, 300 and 400 mg of glucose.

Gliding motility and phototactic motility assay

To assess whether filamentous isolates demonstrated phototactic movement, cultures were streaked in triplicate on a four-well microtiter plates containing ASN agar. The isolates were then placed in a growth chamber for 7 days with a unidirectional light source (80 μE m−2 s−1) and maintained at 30 °C. Growth was measured as the distance (mm) cells extended over the agar plate every 24 h. Motility of each strain was independently confirmed using a Zeiss Axioplan Microscope in response to a unidirectional fiber optic light source.

DNA extraction

Because of the excessive amounts of EPS material in both field-collected material and cultured isolates, DNA was extracted using a modified bead-beating preparation method that included potassium ethyl xanthogenate (Tillett and Neilan, 2000; Green et al., 2008). Briefly, 10 stromatolite cores and isolates (20–80 mg) samples were pretreated in a 1% (w/v) lysozyme solution (10 mg of lysozyme per 1 ml of ASN media) for 30 min at room temperature with vortexing. The samples were then centrifuged at low speed to remove ooids; no additional processing was required to remove CaCO3. The remaining material was resuspended in extraction buffer (100 mM Tris-HCl pH 8.0; 100 mM potassium phosphate buffer pH 8.0; 1% (w/v) cetyl trimethyl ammonium bromide; and 2% (w/v) sodium dodecyl sulfate) containing a bead cocktail of 0.2 g 0.1 mm; 0.2 g 0.7 mm; and eight 2.4 mm sterile glass beads (BioSpec, Bartlesville, OK, USA). Samples were vortexed, and then briefly centrifuged to reduce the foam. A concentrated xanthogenate solution (2.5 M ammonium acetate, and 3.2% (w/v) potassium ethyl xanthogenate) was added to the samples and the samples were then incubated for 2 h at 65 °C. After heat treatment, the samples were incubated on ice for 30 min then centrifuged for 15 min at 20 000 g. Supernatant containing DNA was then mixed with a solution containing 2.5 M KCl for 5 min so that the final concentration was 0.5 M KCl. The samples were again centrifuged (20 000 g), and then mixed with 80 μl 5 M NaCl solution and 2 volume of 100% ethanol. Samples were then placed overnight at −80 °C to precipitate the nucleic acids. DNA was recovered by centrifugation (20 000 g) at 4 °C. Supernatant was discarded and the DNA pellet was dried using a DNA110 SpeedVac (Savant, Hicksville, NY, USA) for 5 min. DNA pellets were then resuspended in 700 μl of the chaotropic C4 solution (MoBio PowerSoil DNA kit; Mo Bio Laboratories, Carlsbad, CA, USA). The DNA was bound to a silica spin column, washed and recovered using the MoBio PowerSoil DNA kit according to manufacturer's instructions. DNA concentrations were measured spectrophotometrically from the 10 replicate extractions, normalized and pooled.

Generation and screening of 16S rRNA gene clone libraries

Two distinct 16S rRNA gene clone libraries were generated using different cyanobacterial primer sets (Table 1) from the pooled type 2 stromatolite DNA. The first clone library (cHBC1) was generated by PCR amplifying genomic DNA with broad specificity cyanobacterial primers Cya359 and Cya781a, b (Table 1). The second clone library (cHBC2) was generated using a second cyanobacterial primer set, Cya266 and Cya1234, which targeted a narrower range of coccoid and endolithic cyanobacterial species. Other than the primer sets employed, the construction of the 16S rRNA gene libraries was identical. Briefly, genomic DNA isolated from the type 2 stromatolites was PCR amplified using the GoTaq PCR core kit (Promega, Madison, WI, USA), with 1.5 mM MgCl2. The PCR amplifications were performed in an PTC-200 (MJ Research-BioRad, Hercules, CA, USA) and the cycle parameters included an initial denaturation at 95 °C for 2 min, 30 cycles of denaturation at 95 °C for 30 s, annealing at 58 °C for 2 min, elongation at 72 °C for 2 min and a final 10 min elongation step at 72 °C. The amplified PCR products were purified using the Ultra-Clean PCR up kit (Mo Bio Laboratories) and then inserted into the pCR2.1 vector and cloned into TOPO10 competent cells using the TOPO TA Cloning kit (Invitrogen, Carlsbad, CA, USA) according to manufacturer's instructions.

Table 1 PCR primers used in this study

Individual colonies of the clone library were randomly selected and screened for the appropriate sized inserts by PCR, and 192 colonies from library cHBC1 and 96 colonies from cHBC2 were selected for further analysis. Plasmid DNA was isolated using a Qiagen Plasmid Prep kit (Qiagen, Valencia, CA, USA) and sequenced by the University of Florida Interdisciplinary Center for Biotechnology Research sequencing core facility using traditional Sanger sequencing with and ABI 3130 DNA sequencer. The sequences recovered from these two clone libraries were combined with sequences (n=102) derived from a third library generated using a ‘universal’ 16S rRNA gene primer set (Baumgartner, 2007).

16S rRNA gene sequence analysis

Recovered sequences and closely related sequences, as determined by the basic local alignment search tool (BLAST; Altschul et al., 1997), were aligned to known 16S rRNA gene sequences using the ‘greengenes’ 16S rRNA gene database and alignment tool (DeSantis et al., 2006). Clone libraries were screened for chimeric sequences using the Bellerophon chimera checking tool, by the greengenes website (Huber et al., 2004) and the chimera detection program at the Ribosomal Database Project website (Cole et al., 2007). Aligned sequences were imported into in the ARB software package (Ludwig et al., 2004) using the greengenes prokaryote 16S rRNA gene database. The sequences generated from the clone libraries were relatively short (375–900 bp) and located in different sections within the rRNA gene. To produce a single phylogenetic tree the short sequences were inserted into the full ‘greengenes’ 16S rRNA gene tree using the ARB parsimony option, employing a 50% maximum frequency cyanobacterial filter in which highly variable positions were ignored, whereas the overall topology of the tree was maintained. Subsequently, the tree was trimmed and only relevant cyanobacterial sequences were retained.

For phylogenetic analyses of the longer sequences from the cyanobacterial isolates, the aligned sequences and close relatives were exported from ARB into the MEGA software package version 3.1 (Kumar et al., 1994) using a cyanobacterial 50% conservation filter. Neighbor-joining phylogenetic trees were constructed on the aligned 16S rRNA gene sequences using the Kimura 2-parameter substitution model with complete deletion of gapped positions. The robustness of inferred tree topologies was evaluated by 1000 bootstrap resamplings of the data; values greater than 70% are indicated adjacent to each node. In addition, Bayesian analyses were performed on the sequence data (MrBayes ver. 3.1; Ronquist and Huelsenbeck, 2003) by running four simultaneous chains (three heated, one cold) until the average standard deviation of split frequencies was below 0.01 (106 iterations). The selected model was the general time reversible using empirical base frequencies, and estimating the shape of the γ-distribution and proportion of invariant sites from the data. A resulting 50% majority rule consensus tree (after discarding the burn-in of 25% of the generations) was determined to calculate the posterior probabilities for each node. Nodes on the neighbor-joining tree that were supported by Bayesian analysis, with posterior probability values greater than 70%, are indicated by black circles. All sequences identified in this study were filed in GenBank under accession numbers EU248965EU249128 and FJ373355FJ373445.


Morphological diversity of cyanobacteria

Cyanobacteria in the Highborne Cay stromatolites were concentrated in a distinct layer occupying the top 5 mm beneath the stromatolite surface (Figure 1d). This layer was comprised ooid grains embedded in a cyanobacterial mat dominated by a filamentous morphotype consistent with the description of S. gebeleinii. The filaments contained one to several trichomes, 1–1.5 μm wide, with cells several times longer then wide surrounded by a thick, layered, externally uneven sheath. On the stromatolite surface, a thin layer (<1 mm) of white carbonate precipitate was frequently observed (Figures 1d and m). A combination of transmitted light and fluorescence microscopy revealed high concentrations of cyanobacteria in the interstitial spaces surrounding the ooids (Figures 1e and f), including distinct filaments (arrow) and coccoid aggregates on the ooid surfaces. Enrichment cultures were started from stromatolite samples with different media known to be favored by marine cyanobacteria.

Ten cyanobacterial isolates grew in culture; eight of these cultivars were filamentous, one coccoid, and one pseudofilamentous morphotype. All filamentous cultivars produced single trichomes within tubular polysaccharide sheaths. The cell diameter varied greatly between isolates and ranged from 0.98 to 9.58 μm, but showed a narrow range distribution within each isolate (Table 2). Cellular trichomes exhibited gliding movement, frequently crawling out of their sheaths and forming new ones. The polysaccharide sheaths varied accordingly in thickness and were comprised mostly of acidic mucopolysaccharides as visualized by AB staining. The isolates showed variable compositions or photosynthetic pigments with colors ranging from pale blue-green to intense red (Figure 2).

Table 2 HBC culture collection
Figure 2
figure 2

Bright field photomicrographs of Highborne Cay culture library (HBC1–9) taken at the same magnification as indicated by the scale bar in the upper left picture (bar, 10 μm; except HBC6). HBC1–4, and 8 are similar to each in cell shape and size corresponding to morphotypes of the genus Leptolyngbya. In HBC2 the dark red pigmentation suggests a high concentration of phycoerythrin. HBC5 corresponds to the morphotype Symploca sp. Isolate HBC6 is characterized by coccoid cells that appear to be dividing by binary fission (arrow). Isolates HBC7 and HBC9 correspond to the morphotypes Hydrocoleum sp. and Phormidium sp., respectively.

Isolates HBC1–4, and 8 were similar in having very narrow trichomes and elongated cells. They were classifiable by morphotype to Leptolyngbya sp. One isolate, HBC2, was distinguished by dark red pigmentation suggestive of phycoerythrin (Figure 2; HBC2). HBC5 corresponded to the morphotype Symploca sp., as it exhibited large isodiametric cells (5 μm wide/long) and thick smooth sheaths that occasionally appeared with rounded closures and false branching. Isolate HBC7 was identified as Hydrocoleum sp., with calyptrate terminal cells and thick, layered sheaths with single or multiple trichomes. HBC7 cells average 9.58 μm and are several times shorter than wide. Strain HBC9 corresponds to Phormidium sp. and has an intermediate cell width averaging 3 μm in diameter. Although various coccoid cyanobacteria were observed in field-collected samples of stromatolites, only HBC6 was isolated as a coccoid monoculture, which appeared to replicate by binary fission and was morphologically identified as Aphanocapsa sp. (Figure 2; HBC6).

The pseudofilamentous cyanobacteria (Pleurocapsales) observed in this study (HBC10) appeared to occupy microbored tunnels within individual ooid grains (Figure 3) and is consistent with the morphotype Hyella sp. Inside the tunnels they appear to have excavated, these phototrophic endoliths (euendoliths sensu Golubic et al., 1981) appeared to be encased in sheaths that stained with AB, indicating that the sheaths are comprised acidic mucopolysaccharides (Figures 3a and b). The euendolithic cyanobacteria were autofluorescent at 650 nm and in dark field their borings are seen as light refracting outlines (Figure 3c). The envelopes and sheaths of HBC10 stained with PA/Schiff reagent, but not with AB suggesting the presence of neutral polysaccharides. Cultures of the euendolithic Hyella sp. HBC10 persisted in culture within the ooid grains for several months (HBC10; Figure 3d).

Figure 3
figure 3

Colonization of ooid sand grains by euendolithic cyanobacteria. (a) Field of sand grains stained with Alcian Blue (AB) stain and periodic acid (PA)/Schiff reagent (pink color), which stain for mucopolysaccharides and glycoproteins, respectively. (b) Staining by AB and PA reveals extensive production of polysaccharide material by endolithic cells; the surfaces of the organism appear to stain extensively with PA suggesting the endolith is producing neutral polysaccharides. (c) Dark field image showing light refracting microbored tunnels (arrow) within an individual ooid sand grain. (d) Bright field photomicrograph of the pseudofilamentous strain Highborne Cay 10 (HBC10) showing pigmented, photosynthetically active cells still within the sand grain.

Production of extracellular polymeric secretions

Cyanobacteria isolated from the type 2 Highborne Cay stromatolites were tested for their capacity to produce large amounts of EPS material as it was observed in field-collected samples. All 10 isolates were examined under normal culture conditions and found to produce detectible but different levels of EPS material as visualized with AB staining (data not shown). The producer of largest amounts of extracellular EPS material was HBC2 (Leptolyngbya sp.). To examine the effect of light intensity on EPS production, cultures of this strain were maintained under two different light conditions, at low (30 μE m−2 s−1) and high (100 μE m−2 s−1) white light flux (Figure 4). The excreted EPS is seen in agar cultures at low magnification as a diffuse halo surrounding the cyanobacterial colony. At low light intensity HBC2 colony produced little visible EPS material (Figure 4a). Conversely, at high light intensity, the extracellular EPS material was visible as long fibrous tentacle-like structure that extended outward from the cell mass (Figure 4b). At the ultrastructural level, EPS material appeared as an electron dense matrix extending from the cell outward beyond the sheath (Figure 4c). HBC2 cultures exposed to low and high light for 1 week demonstrated a fourfold increase in the amount of extracellular EPS material (Figure 4d). At low light, the mean EPS levels were 52±20 mg of glucose equivalent compared to a mean of 209±43 mg of glucose equivalent in cultures grown in high light. These results indicate that higher light levels can trigger increased EPS production in the HBC2 stromatolite isolate.

Figure 4
figure 4

Light-dependent production of extracellular polymeric secretions (EPS) in stromatolite culture libraries. (a) Representative culture of Highborne Cay 2 (HBC2) grown under low light (30 μE m−2 s−1). HBC2 forms round colonies surrounded by a narrow halo of opaque polysaccharides. (b) HBC2 culture grown at high light (100 μE m−2 s−1) showing extensive production of EPS material extending outward in radiating streaks. (c) Transmission electron micrograph (TEM) of a cross-section of a HBC2 filament grown at high light, showing the cell (c) surrounded by sheath (s) and profuse EPS material (eps) emanating from the cell beyond the cyanobacterial sheath. (d) Comparison of quantified EPS extracts from HBC2 colonies with equal biomass (60 mg) grown under high and low light conditions. The results indicate a fourfold increase in EPS production at higher light intensities for 1 week.

Cyanobacterial phototaxis

To determine whether the isolates were capable of phototactic motility, filamentous isolates were examined using agar plating with a unidirectional light source (Figure 5). A typical positive response of the eight filamentous isolates six had detectible levels of gliding motility, which was also independently confirmed by direct observation using bright field microscopy (data not shown). Significant phototactic response was shown in strains HBC1, 3, 5 and 8. Two filamentous cultivars, HBC7 and HBC9 appeared to move irrespective of light direction. By day 6, HBC9 had evenly covered the entire agar well. The remaining two isolates showed no significant level of motility. After 10 days the density of the original inoculum increased but the colonies did not extend beyond 2 mm of the starting inoculum suggesting growth but no motility.

Figure 5
figure 5

Directional migration assay of cultured cyanobacteria isolated from type 2 stromatolites. (a) Micrograph depicting the positively (+) phototactic response of the filamentous isolate Highborne Cay 8 (HBC8). Very few cells grew or migrated (or grew) away (−) from the light source. (b) Histogram of filamentous cultures indicating the variability in motility (including growth). Isolates HBC1, 3, 5 and 8 demonstrated a significant positive response. HBC7 and 9 showed motility but no response to light direction. HBC2 and HBC4 cultivars showed negligible levels of movement.

Sequence analysis of 16S rRNA gene from cyanobacterial cultivars

The phylogenetic relationship of the cyanobacterial cultivars (HBC1–10) was compared to known cyanobacteria by analysis of 16S rRNA gene sequences. Nearly full-length 16S rRNA genes sequences recovered from the cultivars were used to generate a bootstrapped neighbor-joining tree (Supplementary Figure S1). Sequences of closely related cyanobacteria, identified by BLAST analyses, were included in the phylogenetic tree. Overall the isolates were classified into three cyanobacterial orders including Chroococcales (HBC6), and the polyphyletic Oscillatoriales (HBC1–5, 7–9) and Pleurocapsales (HBC10). The 16S rRNA gene sequences for the cultivars revealed that eight of the ten stromatolite cyanobacterial isolates had BLAST identities greater than the genus level (95%), whereas the remaining two cultivars had BLAST similarities at 94% (Table 2). Of the cultivated strains two were highly similar (99.5%) to each other by 16S rRNA gene analysis (HBC1 and 4), however, these organisms displayed different physiologies, including differences in growth rate and motility.

Sequence analysis of 16S rRNA genes from cyanobacterial clone libraries

In this study, two clone libraries (cHBC1, cHBC2) consisting of 249 clones were generated using cyanobacterial-specific primers (Table 1) and genomic DNA isolated from type 2 stromatolites. These two libraries were compared to 102 Highborne Cay cyanobacterial sequences retrieved from Baumgartner (2007; cHBC3) in which DNA from all three mat types (1–3) as well as the surrounding seawater was amplified using universal Bacteria primers (Table 1; Figure 6). A bootstrapped neighbor-joining tree was generated using the 10 Highborne Cay cultured isolates and close relatives (Supplementary Figure S1). Fourteen major clusters of cyanobacteria, each containing at least four environmental clone sequences were identified. Not all environmental sequences were contained in the major clusters and a total of 29 distinct cyanobacterial phylotypes (97%) were identified from the clone libraries (Figure 6). A comparison of these clone libraries with the 16S rRNA gene of the cultured isolates resulted in the detection a total of 33 distinct phylotypes in the type 2 stromatolites.

Figure 6
figure 6

Phylogenetic tree of cyanobacterial 16S ribosomal RNA gene sequences recovered from clone libraries generated with different primer sets and from isolates, as described in the text. Short clone sequences and longer isolate sequences were inserted into a large tree based on full or nearly full 16S rRNA gene sequences using parsimony and filtering the data with a 50% cyanobacterial conservation filter while maintaining the overall tree topology. Clusters containing at least four environmental sequences recovered in this study were compressed. The number of sequences included in each cluster is enclosed in parentheses. The scale bar represents 0.05 substitutions per nucleotide position. Highborne Cay cultivars are highlighted in gray and cultivars detected in the cultivation-independent clone libraries are indicated with and asterisk.

The clone libraries generated with three different primer sets differed in the detection of each of the major clusters of cyanobacteria (Table 3). Only sequences belonging to cluster 1, which were most closely related to the coccoid cyanobacteria Stanieria cyanosphaera PCC7437, were detected in all three of the clone libraries. These Stanieria-like sequences were also the most abundant, and comprised 43% (151 of 349) of all environmental cyanobacterial sequences recovered. Clusters 4–6, representing roughly 5% of the total clone library, contained taxa only detected by the cHBC3 analysis and were not amplified using cyanobacterial-specific primers. These sequences were closely related to Xenococcus and Chroococcidiopsis. Clusters 2, 7–9, 12 and 13 representing roughly 24% of the total clone library contained cyanobacterial sequences that were detected only in the cyanobacterial primer libraries and not detected in the universal primer cHBC3 library. Eukaryotic chloroplast sequences were detected in two of three clone libraries (cHBC1 and cHBC3); however, these sequences only comprised 3% of detected sequences.

Table 3 HBC cyanobacterial clone library sequences

A number of the cultivars isolated from the Highborne Cay stromatolites were detected in the three clone libraries. The filamentous cultivars Leptolyngbya sp. HBC2 and Leptolyngbya sp. HBC3 and Phormidium sp. HBC9, as well as the euendolithic Hyella sp. HBC10 shared close sequence similarity to clones detected in the cHBC1 and cHBC3 libraries (Figure 6; Supplementary Figure S1). Cultivars HBC1, HBC4 and HBC8 were closely related to each other (96%), however, they were not observed in any of the three clone libraries. This cluster of cultivars appeared to be closely related to the Lyngbya/Phormidium/Plectonema group and to Leptolyngbya sp. consistent with classical morphological identifications.


The results of this study provide evidence that cyanobacterial diversity within the type 2 stromatolites was taxonomically richer than has been previously reported (Golubic and Browne, 1996; Reid et al., 2000; Stolz et al., 2001; Baumgartner, 2007). The combined culture-dependent and culture-independent approaches used in this study detected 33 distinct phylotypes (97%) in the type 2 stromatolites, 14 of which represented greater than 1% of the total clone library. Many of the cultured phylotypes shared morphological characteristics such as cell size, shape and the presence of a polysaccharide sheath, but also exhibited both physiological and molecular differences. These differences included variable levels of EPS production under high and low light conditions, variable motility rates among the cultivars despite sharing high sequence similarity in the 16S rRNA gene. Together these differences suggest that cyanobacteria be involved in the stromatolite ecosystem. These results also show that morphological or molecular analyses alone are inadequate for characterization of microbial communities, and emphasize the need for combined cultivation-dependent and cultivation-independent analyses.

Molecular analysis of both clone and culture libraries indicated there was a total of 33 distinct phylotypes (97%) identified from the clone and culture libraries. Of the ten cultivars three HBC isolates overlapped with sequences recovered from the clone libraries at 97% (Supplementary Figure S1). Leptolyngbya HBC2, Leptolyngbya HBC3 and Phormidium HBC9 each shared 99% sequence similarity to clones recovered from library cHBC1. The 16S rRNA gene sequence of a fourth isolate, the euendolithic Hyella HBC10, was 96% similar to a clone from the cHBC3 library. The incomplete overlap between culture and clone libraries is indicative of the limitations of both cultivation-based and cultivation-independent analyses of microbial communities. The biases associated with cultivation are well known (for example, Amann et al., 1995), but the cultivation-independent molecular analyses are also susceptible to distortions of the in situ community. Particularly relevant to this study is the potential interference in recovering cells encased in thick polysaccharide sheaths and EPS during the nucleic acid extractions that may result in an under estimation of true cyanobacterial diversity within the stromatolites. Low levels of overlap between the culture and clone libraries are, however, not uncommon in Shark Bay stromatolites (Burns et al., 2004) and other microbial diversity studies (Eilers et al., 2000; Munson et al., 2002; Donachie et al., 2004; Munson et al., 2004; Maturrano et al., 2006). For example even in the well-studied oral biofilms, where a century of cultivation efforts have still only isolated approximately 50% of the total number of ecotypes (700; Kolenbrander et al., 2003; Aas et al., 2005), there are significant gaps between clone and culture libraries. In studies where both cultivation and molecular techniques were used simultaneously to examine microbial diversity in endo- and exodontic oral biofilms there was only a 33% and 38% overlap, respectively, between the cultivar and clone libraries (Munson et al., 2002; Munson et al., 2004). Such results from these previous studies are comparable to the 30% overlap detected between the HBC cyanobacterial cultivar and clone libraries, reinforcing the growing need to incorporate these multifaceted approaches concurrently (Donachie et al., 2007).

Phylogenetic comparison of HBC cyanobacterial sequences identified in this study to those in GenBank found that approximately 30% of the cultivars shared sequence similarity (>97%) to cyanobacteria associated with Shark Bay stromatolites, Guerrero Negro hypersaline microbial mats, thermophilic hot springs and Antarctic lakes (Taton et al., 2003; Papineau et al., 2005; Taton et al., 2006; Green et al., 2008) suggesting that there may be a selection for certain cyanobacterial sequences in microbial mat communities. However, some sequences including two isolates (HBC2, 10) shared low sequence similarity to known sequences (<94%) and may be endemic to HBC stromatolites.

Several of the filamentous cyanobacterial isolates contained narrow trichomes (HBC1–4, 8), however, none of these cultivars could be identified with the morphotype S. gebeleinii, the multitrichome filamentous cyanobacteria previously observed to dominate natural stromatolite material. Morphological comparisons were the only means to compare the samples as direct PCR from dissected filaments from natural stromatolites material failed to amplify the 16S rRNA gene, most likely because of the elevated levels of Ca2+ in the stromatolite EPS matrix, and no 16S rRNA gene sequences from S. gebeleinii have ever been submitted to GenBank. It is possible this morphological characteristic did not persist in the culture enrichments and that the property to form filaments with multitrichomes is lost under artificial growth conditions. The HBC stromatolite ecosystem is oligotrophic for some nutrients (that is, Fe, P and trace metals; Pieter Visscher, personal communication) and the conditions of ASN media may suppress the multitrichome phenotype as it has been well documented that cultivation can alter in vivo morphologies (Braun, 1952; Domotor and D’Elia, 1986; James et al., 1995; Alonso et al., 2002). Alternatively, it is possible that the organisms dominant in the natural setting were not competitive when transferred into culture (Ward et al., 1997). The lack of the multitrichome morphology coupled with 16S rRNA gene analyses suggest these isolates belong to the polyphyletic morphogenus Leptolyngbya sp. (Castenholz, 2001). The ability of Leptolyngbya morphotype to grow and dominate standard culture media has been observed in other studies of marine mat-forming cyanobacteria (Abed et al., 2003). A different marine Schizothrix (S. splendida) that dominated the microbial mats in brackish ponds of the Rangiroa Atoll did not grow in culture medium whereas other cyanobacteria from this site were transferred without loss of morphotypic characters (Richert et al., 2006). However, together these observations suggest that filamentous, EPS producing cyanobacteria other than S. gebeleinii are also important for stromatolite development at HBC.

Culturing and sequencing of the euendolithic cyanobacterium of Hyella morphotype was successful, demonstrating its phylogenetic position in the cluster of pleurocapsalean cyanobacteria. Ooids are known for harboring an array of euendolithic morphospecies (Lukas and Golubic, 1983; Al-Thukair and Golubic, 1991a, 1991b, 1996; Al-Thukair et al., 1994). The type 2 HBC stromatolites, which are the subject of the present contribution, are considered an intermediate mat type, that is, in the process of consolidation, following the entrapment of loose, shoaling ooids by cyanobacterial mats. The euendolithic microorganisms of the Hyella morphotype may therefore represent a community that has colonized ooids before their incorporation into stromatolites, in contrast to colonization by Solentia, which was found to contribute to the later consolidation stages (that is, type 3) of stromatolite development (Macintyre et al., 2000; Reid et al., 2000). In addition to the cultured Hyella sp. HBC10, 45% of the recovered sequences from the three clone libraries were Pleurocapsales (clusters 1, 3–5) with the highest number of recovered sequences similar to Stanieria sp. (n=156). The high number of unicellular pleurocapsalean cyanobacteria and the absence of Nostocales may suggest that the Stanieria-like organisms may be involved in nitrogen fixation (Walker et al., 2008) or Fe sequestration (Geiss et al., 2001). These clone library data, however, should be taken with the caveat that the simple abundance of clones can reflect the differences in cell size between endolithic, coccoid and filamentous morphotypes. The cell width of the euendolithic Hyella is 3–6 times larger than the filamentous Leptolyngbya sp. (Table 1) and can result in differences in rRNA gene copy number and different ratios of rRNA gene copy number to total cell biovolume. Such differences have been hypothesized in the hypersaline mats of Guerro Negro, Mexico (Ley et al., 2006; Green et al., 2008) and could bias the amount of DNA recovered from each morphotype and thus PCR amplification. Despite this caveat, morphological characterization of the types 2 and 3 stromatolite types have identified that approximately one-third of the ooids are colonized by euendolithic cyanobacteria. This observation coupled with the clone libraries, and other studies (Havemann and Foster, 2008) suggest that other Pleurocapsales in addition to the previously described Solentia sp., be involved in stromatolite formation and growth.

In addition to examining the cyanobacterial diversity within the type 2 stromatolites, physiological characteristics considered essential for stromatolite stability and development were examined in those cultured isolates including EPS production, motility and phototaxis. Stromatolites must endure the intense erosive nature of wave action by maintaining their cohesive structure (Eckman et al., 2008). Two of the major mechanisms that contribute to stromatolite stabilization in this turbulent environment are lithification and EPS production. Although other key functional groups of bacteria, such as sulfate-reducing bacteria, have been shown to produce detectable levels of EPS in the HBC stromatolites (Decho et al., 2005; Braissant et al., 2007), most studies have shown that the cyanobacteria are the major contributor to EPS production in the stromatolites (for example, Kawaguchi and Decho, 2000; Kawaguchi et al., 2003; Decho et al., 2005). The capacity to produce EPS was tested on all cultivars. Although concentrations of EPS differed between HBC cultivars, all produced detectible levels of EPS with the highest level of EPS detected in Leptolyngbya sp. HBC2. Light-dependent EPS production has been seen in other aquatic ecosystems such as freshwater cyanobacteria (Merz-Preiss and Riding, 1999), and diatoms (Smith and Underwood, 1999; De Brouwer et al., 2002; Wolfstein and Stal, 2002) where diel EPS production has been shown to facilitate cell motility, thus improving growth conditions by increasing nutrient access (Ruddy et al., 1998; Wolfstein and Stal, 2002). Light-driven EPS production may also serve as a protectant against environmental stress conditions that occur during the day in shallow waters. For example HBC2, which does not exhibit a phototactic motility response, produces the highest level of EPS of all tested cultivars. In other systems, EPS material has been shown to be an essential protectant against ultraviolet radiation (Elasri and Miller, 1999; Wang et al., 2007), oxidative stress (Davies and Walker, 2007) and phage invasion (Deveau et al., 2002). EPS has also been shown to actively bind Ca2+ ions, thus directly involve in carbonate precipitation and stromatolite development (Arp et al., 2001; Kawaguchi and Decho, 2001, 2002a, 2002b; Braissant et al., 2007). Light-dependent EPS production could imply a diel cycle or seasonality to stromatolite carbonate deposition and lithification. Such light-driven calcification has been well documented in corals (for example, Chalker and Taylor, 1975) calcifying algae (for example, De Beer and Larkum, 2002) where elevated levels of photosynthesis drive an increase in EPS production and in pH. The alkaline conditions within the stromatolite have been shown to facilitate carbonate deposition and lithification (Dupraz et al., 2004).

Several cultured isolates exhibited another physiological property critical for stromatolite development: motility and phototaxis. In addition to erosion, the hydrodynamic environment of HBC results in the periodic burial of stromatolites by oolitic sands. Of the eight filamentous cultivars, six exhibited positive phototaxis with four of these six exhibiting similar filamentous morphotypes that resemble S. gebeleinii. Vertical migration of photosynthetic cells has been characterized in other microbial mat communities in response to diel fluctuations in UV, oxygen and sulfide concentrations (Prufert-Bebout and Garcia-Pichel, 1994; Fourcans et al., 2006).

This study revealed that the cyanobacterial diversity is far more complex in the type 2 HBC stromatolites than had been previously identified using morphological analysis alone. Results from combined cultivation and molecular techniques imply the presence of a higher number of phylotypes than morphotypes. Both the culture and molecular data indicate that the predominant cyanobacterial morphotypes within the stromatolite are nonheterocystous filamentous and euendolithic, and unicellular coccoid organisms, which together may greatly facilitate structural and ecosystem stability (for example, May, 1973). Additional ecotype complexity may also facilitate stromatolite initiation and development by increasing the communities resilience to perturbations such as wave action and burial (Yannarell et al., 2007). Together these results emphasize the need for an integrated analysis of field populations combined with cultivation and molecular analysis, as the results of either of these approaches alone may be inadequate to delineate the comprehensive role of cyanobacteria diversity in stromatolite initiation and development.