Culture-independent analysis of the soil bacterial assemblage at the Great Salt Plains of Oklahoma

Abstract

The Great Salt Plains (GSP) of Oklahoma is a natural inland terrestrial hypersaline environment that forms evaporite crusts of mainly NaCl. Previous work described GSP bacterial assemblages through the phylogenetic and phenetic characterization of 105 isolates from 46 phylotypes. The current report describes the same bacterial assemblages through culture-independent 16S rRNA gene clone libraries. Although from similar hypersaline mud flats, the bacterial libraries from two sites, WP3 and WP6, were quite different. The WP3 library was dominated by cyanobacteria, mainly Cyanothece and Euhalothece. The WP6 library was rich in anaerobic sulfur-cycle organisms, including abundant Desulfuromonas. This pattern likely reflects differences in abiotic factors, such as frequency of flooding and hydrologic push. While more than 100 OTUs were identified, the assemblages were not as diverse, based on Shannon indexes, as bacterial communities from oligohaline soils. Since natural inland hypersaline soils are relatively unstudied, it was not clear what kind of bacteria would be present. The bacterial assemblage is predominantly genera typically found in hypersaline systems, although some were relatives of microbes common in oligohaline and marine environments. The bacterial clones did not reflect wide functional diversity, beyond phototrophs, sulfur metabolizers, and numerous heterotrophs.

Introduction

The Great Salt Plains (GSP) of Oklahoma are part of the Salt Plains National Wildlife Refuge in north-central Oklahoma. The Salt Plains Microbial Observatory was established there to isolate and characterize the diverse halotolerant microbes in hypersaline soils at the site. The GSP are barren sandy mud flats covering 65 km², which are crusted with evaporite salt deposits derived from briny aquifers [1, 2]. Areas of the surface have shallow standing ponds of saturated brine, mainly NaCl, while other areas exhibit only transient pools that are associated with shifts in microbial community composition [3]. The environment is not only extreme in salinity, but the surface is exposed to unobstructed UV irradiance, desiccation, alkaline pH, and high temperature (>50 °C in summer).

Previous work at the GSP has focused on halotolerant aerobic bacteria and archaea [4–7]. The GSP collection of bacterial isolates obtained through enrichment culture was dominated by Halomonas and Bacillus species, among the 46 phylotypes represented. In some cases, the isolates were halotolerant strains of bacteria common in oligohaline (low-salt freshwater) soils, but most were from groups typically associated with hypersaline environments. A wide range of salinity tolerances was observed for most of the GSP bacterial isolates, with growth in media from 0.1% salinity to over 20% salinity. Some showed tolerance to a wide range of pHs (pH 5 to 11), and temperatures (4 to 50 °C), and high resistance to UV irradiation (comparable to E. coli). In a salinity environment that changes rapidly with rain events, wide tolerance ranges and flexibility can enhance survival. In contrast to the bacterial isolates, GSP archaea did not exhibit wide environmental tolerances and tended to be halophilic, requiring high salinity. Random clone libraries of archaeal 16S rRNA gene sequences from direct extractions of soil genomic DNA showed relatively low species diversity. However, nearly two-thirds of the sequences were in clusters that did not have close relatives reported in public databases.

The current report extends these observations using culture-independent bacterial 16S rRNA gene clone libraries to describe bacterial diversity at the GSP. We present substantial molecular analyses of bacterial communities from natural inland terrestrial hypersaline soils. Previous related molecular studies have examined salterns and ephemerally wet playa environments [8–11]. Preliminary accounts of this work have been presented previously [12, 13].

Materials and methods

Sample collection

Soil samples were collected from two sites at the GSP between August 2001 and August 2002: WP3, N 36°43.044′, W 98°15.702′ and WP6, N 36°43.851′, W 98°15.561′ (Fig. 1). Grab samples of surface crusts included the top two cm of soil on the unvegetated salt-crusted areas of the GSP with a pH of approximately 7.8 and a salinity of 7.5%, as described previously [4]. Each bulk sample collected with sterile instruments was mixed by hand and aliquots (100 g) for molecular analysis were frozen in the field on dry ice in sterile Whirl-Pak bags, transported to the laboratory on dry ice, and stored at −80 °C for later use.

Figure 1.

Map of the Salt Plains National Wildlife Refuge showing the GSP and sampling sites used for the current study.

PCR, cloning, and DNA sequencing

Genomic DNA extracts were made directly from GSP soils using the bead-beating protocol of Bürgmann et al. [14], with some modifications as previously described [7]. Aliquots of soil (0.5 g) were mixed with 0.75 g of zirconia/silica beads (0.1 mm) and 1.25 ml of extraction buffer (0.2% CTAB, 1 mM DTT, 0.1 M NaCl, 50 mM EDTA, 0.2 M sodium phosphate [pH 8.0]) and subjected to bead beating (Genie; Scientific Industries) at 4 °C for 3 min to disrupt cells. After centrifugation for 5 min at 16,000 × g, the supernatant was extracted with equal volumes of phenol (pre-equilibrated with 0.1 M Tris buffer [pH 8.0]) and then water-saturated chloroform-isoamyl alcohol (24:1, v/v). The aqueous phase (approximately 700 μl) was incubated for 1 h with 750 μl of precipitation solution (20% [w/v] polyethylene glycol 6000 in 2.5 M NaCl), followed by centrifugation for 30 min at 16,000 × g to pellet genomic DNA. After washing with ice-cold 70% ethanol, the air-dried pellet was resuspended in 100–300 μl of TE buffer (1 mM EDTA, 10 mM Tris [pH 8.0]). Purity of the extract was determined as the ratios of A₂₆₀ (DNA) to A₂₈₀ (protein) and to A₂₃₀ (humic acid).

Extracts of genomic DNA were the targets for PCR amplification of 16S rRNA genes using universal bacterial primers (pA: 5′-AGAGTTTGATCCTGGCTCAG-3′ and pH: 5′-AAGGAGGTGATCCAGCCGCA-3′) [15]. PCR was performed in a thermal cycler (Eppendorf Mastercycler) as 25 μl reactions containing 1 × PCR buffer (Takara), 20 μM of each dNTP, 0.2 μM of each primer, 1 U of Ex-Taq DNA polymerase (Takara), and 5 μl of extract (100–400 ng DNA). DNA was denatured at 95 °C for 2 min, followed by 40 cycles of 95 °C for 1 min, 50 °C for 1 min, and 72 °C for 1 min, with a final 5 min extension at 72 °C. Ten separate PCR amplicon populations were purified by band excision from a 2% agarose gel after electrophoresis and pooled. Clone libraries were generated from the pooled amplicons using the TOPO-TA blue-white cloning system in E. coli (Invitrogen) following the manufacturer’s instructions. More than 200 clones from each library were randomly collected and inoculated into 96-well plates, with plasmid isolation and insert sequencing performed by a commercial vendor (Agencourt) using the pA primer. Sequences were trimmed to remove remaining vector regions, leaving sequences of approximately 600 bp for analysis. All sequences appear in GenBank with accession numbers JN122625 to JN122944.

Sequence analyses

All sequences were automatically aligned using Clustal-W [16] and then manually aligned and trimmed in MacClade v4.08 (Sinauer Associates). Contextual 16S rRNA gene sequences were identified in GenBank or RDP using BLAST [17] or from comparison to relevant literature. PAUP 4.0 b10 [18] generated phylogenetic trees using distance analysis with Jukes-Cantor rules and the neighbor-joining algorithm. Sequences were trimmed to equal lengths, with sequences less than approximately 550 bp removed, and positions with gaps and ambiguous bases ignored, giving 450–500 positions for analysis. Jackknife was used to assess the relative support for each branch with a total of 100 bootstrap replicates conducted heuristically using the distance-based neighbor-joining algorithm and the nearest-neighbor-interchange algorithm in PAUP. The trees were rooted using an appropriate sequence as the functional outgroup. Putative chimeras (approximately 15% of the sequences) were identified through several iterative analyses using the BELLEROPHON server [19], manually examined, and removed if necessary. Full phylogenetic trees with GenBank accession numbers can be found in supplementary materials (Figs. S1–S5). Distance files were further analyzed using the MOTHUR statistical package [20] to determine Chao1 estimators, Simpson indexes, non-parametric Shannon indexes, and rarefaction curves, and to define OTUs at various levels of sequence similarity. Libraries were statistically compared for similarity using the LibShuff program within MOTHUR.

Results

Culture-independent clone libraries

Bacterial 16S rRNA gene clone libraries were prepared from direct DNA extracts of WP3 and WP6 soil samples that had been used previously for the enrichment and isolation of archaea and bacteria [4], and from which archaeal 16S rRNA gene clone libraries have been constructed [7]. To reduce biases in the libraries, the amplicons from 10 separate PCR reactions were combined into a single cloning reaction. Nearly 400 bacterial 16S rRNA gene clones were sequenced, roughly divided between the two sampling locations.

Estimation of diversity

The results of statistical estimates [20] of bacterial diversity in GSP soils are given in Table 1 and Fig. 2. The data are presented at several levels of sequence identity, reflecting commonly used thresholds for distinction of strain (99%), species (97%), genus (94%), and division (88%). The number of OTUs increases at higher levels of sequence identity as expected. With a threshold of 99% sequence identity, 125 OTUs were identified within the total bacterial library. The individual libraries from the two sampling sites gave approximately equal numbers of OTUs, however there was little overlap between the species detected. Pairwise comparisons using LibShuff gave a statistically significant difference (p < 0.0001) between the bacterial clone libraries from WP3 and WP6.

Table 1.

Diversity analyses of GSP bacterial clone library sequences.

Parameter	Level of Sequence Identity
	99%	97%	94%	88%
WP3 Library
OTUs	59	51	40	33
Chao1
Average	121	97	71	54
95% CI	83–217	68–178	49–139	39–106
Shannon Index (H)	3.89	3.70	3.36	3.07
Simpson Index (D)	0.031	0.035	0.049	0.069
WP6 Library
OTUs	66	61	52	38
Chao1
Average	123	115	77	52
95% CI	92–193	85–185	62–116	43–81
Shannon Index (H)	4.05	3.90	3.66	3.18
Simpson Index (D)	0.072	0.075	0.078	0.095
Combined Library
OTUs	125	111	91	68
Chao1
Average	257	214	144	101
95% CI	196–372	164–312	115–204	81–149
Shannon Index (H)	4.61	4.43	4.16	3.68
Simpson Index (D)	0.024	0.026	0.031	0.047

Figure 2.

Rarefaction curves based on 16S rRNA gene sequences from the total bacterial clone library. The curves represent different levels of sequence identity.

Chao1 estimates were used to project the total number of OTUs at different levels of sequence identity (Table 1). At the 99% sequence identity level, 257 OTUs were predicted, while 214 OTUs were predicted at the 97% sequence identity level in the combined libraries. By comparing the Chao1 estimates to the OTUs delineated in the combined sequence library, it is estimated that 49, 52, 63, and 67% of the estimated bacterial diversity was detected at 99, 97, 94, and 88% sequence identity, respectively. These values are slightly higher for the WP6 library and slightly lower for the WP3 library. The rarefaction curve for the total library is shown, as it reflects what was observed for the individual libraries (Fig. 2). The Shannon and Simpson indexes both show that bacterial diversity increases at higher levels of sequence identity.

Phylogenetic groups recovered

The taxonomic composition of the WP3 and WP6 clone libraries were very much different from each other, even though the samples come from areas of the high-salt mud flats with similar topography, gross soil quality, pH, and salinity (Fig. 1). The WP3 library was dominated by cyanobacteria, while the WP6 library was richer in anaerobes and heterotrophic bacteria (Fig. 3). Organisms from some clades were found at both sites, specifically, examples of Bacteroidetes and Gammaproteobacteria. While hypersaline soils at the GSP are thalas-sohaline, these are inland areas, not associated with the marine environment. Thus, the microbial community might be expected to include halotolerant relatives of bacteria found in oligohaline soils, rather than organisms more closely related to marine species. Some of the bacterial clones observed were from genera that are not usually associated with hypersaline environments, such as Klebsiella, however, most of the clones were from genera that have halotolerant or halophilic members. There were a few clones that matched marine organisms, such as Leeuwenhoekiella marinoflava and Marinobacter, but the bulk of the clones were from groups that have been isolated from hypersaline soils and were abundant in isolate collections from the GSP, such as Halomonas and Bacteroidetes. Phylogenetic trees based on 16S rRNA gene clone libraries were generated that also include sequences from cultured GSP bacteria and contextual sequences obtained from the GenBank and RDP databases (Figs. 4–8 and S1–S5).

Figure 3.

Distribution of clones into major clades for the libraries from WP3 and WP6.

Figure 4.

Phylogenetic tree for GSP cyanobacteria and related genera based on 16S rRNA gene sequences. Bootstrap values greater than 50% are shown. A full tree with GenBank accession numbers can be found in Fig. S1.

Figure 8.

Phylogenetic tree for GSP deeply branching genera based on 16S rRNA gene sequences. Bootstrap values greater than 50% are shown. A full tree with GenBank accession numbers can be found in Fig. S5.

The WP3 sample was predominantly cyanobacteria, encompassing 83 of 175 clones (47%) (Figs. 4 and S1). A green layer several mm below the salt crust can be seen in this area of the GSP. All of the cyanobacterial clones were related to unicellular organisms, with the majority being Cyanothece and Euhalothece, both found in solar salterns and hypersaline lakes [21–24]. Related sequences were detected in an earlier cyanobacterial rRNA gene clone library from the GSP and one of the Cyanothece clones was isolated previously [25]. The geographic distribution pattern reported here also was observed by this previous study using cyanobacteria-specific PCR primers for comparisons between the Clay Creek (CC) site (at WP6) and the South Crystal Dig (SCD) site (<1 km south of WP3). The current study did not detect some genera observed earlier, such as Geitlerinema and Phormidium. No heterocystous cyanobacterial genera were observed in the current clone libraries. Planctomycetes and Verrucomicrobium, which cluster with the cyanobacteria, were detected at WP6 (Figs. 4 and S1).

Soils from both sites were rich in Gram-negative Proteobacteria and organisms in the Cytophaga-Flavobacterium-Bacteroides (CFB) complex (Figs. 5, 6, S2 and S3). Previous culture collections from these sites found abundant Halomonas spp. and related Gammaproteobacteria [4]. While the WP6 site did not show cyanobacteria, a dozen clones related to the anaerobic photolithotrophic bacterium Halochromatium were observed [26]. A large number of clones from the Deltaproteobacteria, mainly at WP6, were found, including the single most abundant clone (24% of clones from WP6), an organism that is related to Desulfuromonas palmitatis. This is a sulfur oxidizer that does not reduce sulfate, but can reduce iron and manganese [27]. A denitrifier capable of sulfur oxidation, Thiohalorhabdus [28], also was detected. Although anaerobic Halanaerobium have been isolated from the GSP [29], no organisms from this genus were present in the clone libraries. The CFB cluster was the most populated overall (33% of total clones). Bacteroidetes were common in the current rRNA clone libraries, as they were in cultivation collections made from these same soil samples [4].

Figure 5.

Phylogenetic tree for GSP Gram-negative genera based on 16S rRNA gene sequences. Bootstrap values greater than 50% are shown. A full tree with GenBank accession numbers can be found in Fig. S2.

Figure 6.

Phylogenetic tree for GSP CFB cluster genera based on 16S rRNA gene sequences. Bootstrap values greater than 50% are shown. A full tree with GenBank accession numbers can be found in Fig. S3.

Gram-positive bacteria were predominantly obtained from WP6 soil (Figs. 7 and S4). While Bacillus were common in the cultivation collection [4], they were relatively rare in the clone library. In addition to the low G+C Gram-positive Bacillus, Clostridium, and Spirochaetes, some high G+C Gram-positive actinomycetes related to Propionibacterium were observed. Clones from deeply branching lineages also were predominantly from WP6 soil (Figs. 8 and S5). This included representatives of Chloroflexi, Deinococcus, and Gemmatimonadetes.

Figure 7.

Phylogenetic tree for GSP Gram-positive genera based on 16S rRNA gene sequences. Bootstrap values greater than 50% are shown. A full tree with GenBank accession numbers can be found in Fig. S4.

Discussion

WP3 and WP6 share many physical similarities, but differ in important ways. The WP6 site is close (~100 m) to vegetation and a broad ephemeral creek bed, while WP3 is further from vegetation (~1 km) and streams (>2 km). WP6 is subjected to more frequent and disruptive flooding events that scour the landscape and lower nutrient availability [3]. Thus, WP6 is less stable in salinity and often has lower salinity in pore waters than at WP3, where freshwater is generally obtained from rain and not flooding [30]. There is more groundwater push at WP6, keeping this soil wetter than the soil at WP3, with more saturated soil presumably more anoxic. While the WP3 soil was rich in oxygenic cyanobacteria, the WP6 soil was rich in anaerobic bacteria, many of which are connected to the sulfur cycle. It is unclear whether the community differences between WP3 and WP6 are due to the occurrence and intensity of salinity changes, soil moisture differences due to hydrological considerations, or other abiotic factors. Furthermore, the results summarized here are a temporal snapshot of the bacterial community at discrete locations on the GSP and may not be wholly representative.

Bacterial diversity at the GSP, as measured here by Sanger-based clone libraries and expressed as Shannon indexes, is comparable to those reported for other hypersaline environments. Indexes between 3.5 and 4.5 generally were observed at the 97 or 99% sequence identity level for soils at the GSP, sediments at Wadi An Natrun and Lonar soda lake, and the remains of the saltern at Texcoco [8, 9, 31]. Hypersaline waters at the Salton Sea and a hypersaline lagoon were in a similar range [10, 23]. Furthermore, the range of phyla observed in ephemerally wet playa systems is similar to that found at the GSP. Wadi An Natrun was rich in low G+C Gram-positive bacteria, Alphaproteobacteria, and Bacteroidetes, including many Clostridiales, Firmicutes, and Halanaerobiales [8]. Actinomycetes, Spirochaetes, and errucomicrobia also were detected. In the Salton Sea, Gammaproteobacteria represented 50% of the clones along with high abundances of Alphaproteobacteria and Bacteroidetes [10]. The Deltaproteobacteria were in high abundance in the sediments. It is interesting to note that most of the Gammaproteobacteria were from genera typically associated with fish. At the Texcoco saltern, Alpha- and Gammaproteobacteria were important, along with actinomycetes, Bacteroidetes, Chloroflexi, cyanobacteria, and Firmicutes [9].

Oligohaline soils tend to have greater bacterial diversity than that observed at the GSP. Shannon indexes are generally over 4.0 and many are above 5.0, including arable lands (4.8), agricultural soil (4.4), topsoil (4.1), deglaciated soil (5.1), and arid soil (5.1) [32–37]. While the GSP bacterial community at single sites was not as diverse as those from oligohaline soils, the diversity seems substantial for such a restricted extreme environment. Bacterial diversity at the GSP was higher than archaeal diversity, which gave Shannon indexes of 3.0 or less at the 97 or 99% level of identity [7]. The GSP bacterial cultivation collection [4] also was far more diverse than the archaeal collection of isolates [7].

Functional diversity at the GSP, and for hypersaline soils in general, has not yet been sufficiently examined. Clearly, the WP6 site has an active sulfur cycle, with the most abundant clone being the sulfur-reducer, Desulfuromonas. The phototrophic purple sulfur bacterium Halochromatium, which co-localizes with Desulfuromonas and Thiohalorhabdus at the GSP, has a complementary metabolism, forming deposits of elemental sulfur. While some organic nutrients enter the GSP via flooding, phototrophic microorganisms seem to be widely distributed in GSP areas devoid of plants. Low concentrations of available phosphates have been suggested to limit algal growth overall [3] and their distribution appears to be based on nutrient availability, with a positive correlation to the level of ammonia [30]. Most of the phototrophic activity has been attributed to cyanobacteria, which were abundant in ammonia-rich areas near WP3. Diatoms appear to be more important near the WP6 site and the current study observed abundant purple sulfur bacteria at WP6.

While clones from several biogeochemical functional guilds were observed (e.g., denitrifiers), representatives of other important guilds (e.g., nitrifiers) have not been observed. It is interesting to note that archaeal 16S rRNA clone libraries from the anaerobe-rich WP6 site did not include any methanogen sequences [7]. Substantial nitrogen-fixation activity has been measured at the GSP and nifH genes have been detected at WP3 [29]. Fixation activity was attributed to filamentous cyanobacteria and affected by soil moisture, with activity limited to wet soils during laboratory drying and rewetting cycles. Attempts to isolate nitrifiers or detect amoA gene sequences from SCD soils have been unsuccessful, even though this area is rich in ammonia (Perkins and Ngansop, unpublished). It has been suggested that nitrifiers are unable to generate sufficient energy to grow at high salinity [38]. Therefore, it is not clear whether complete biogeochemical cycles can be supported in hypersaline soils.