Cultivation and Characterization of the Bacterial Assemblage of Epsomic Basque Lake, BC

Abstract

Athalassohaline waters that are rich in divalent ions are good analogues for the chemical environments of Mars and the ocean worlds. Sulfate salts, along with chlorides, are important in Mars regolith with Ca, Fe, Mg, and Na counterions. Certain lakes in the Pacific Northwest are saturated with MgSO₄ as epsomite. Here we report on the microbial community of Basque Lake, BC, a group of playas that is saturated with MgSO₄. More than 60 bacterial isolates were obtained from Basque Lake soils by enrichment culture and repetitive streak-plating using media containing 10% (~1.7 M) NaCl or 50% (~2 M) MgSO₄. Most of the isolates (~75%) were Gram-positive, motile, and produced endospores. Isolates related to Marinococcus halophilus and Virgibacillus marismortui dominated the collection. Halomonas and Salinivibrio were Gram-negative genera found at Basque Lake. Nearly all of the Basque Lake isolates grew at 50% MgSO₄, with 65% growing at 60% MgSO₄. Several isolates could grow in saturated (67%) MgSO₄ (a_w = 0.90). All of the isolates grew at 10% NaCl with 70% growing at 20% salinity (~3.5 M NaCl; a_w = 0.82). Basque Lake isolates grew better at basic pH than acidic pH, with 80% growing at pH 9 and 30% growing at pH 10. Only 20% of the isolates grew at pH 5. Numerical taxonomy dendrograms based on 44 phenetic characteristics showed a strong correspondence to phylogenetic trees constructed from 16S rRNA gene sequences. Pyrosequencing of 16S rRNA gene sequences from direct DNA extracts of Basque Lake soils recovered predominantly Proteobacteria (60%), Firmicutes (11%), and unclassified bacteria (27%). Microbes capable of growth under the extreme chemical conditions of Mars are a particular concern for forward planetary protection should they contaminate a spacecraft.

Introduction

Athalassohaline lakes of the Canadian Plains are rich in sulfate salts including Glauber’s salt and epsomite (Hammer, 1978, 1986; Hammer and Haynes, 1978; Last and Slezak 1988; Nesbitt 2004; Last and Ginn 2005). The vast majority of the lakes in this region (53 of 60) are dominated by sulfate anions, with most of these exceeding 80 eq. %. Epsomic lakes are relatively rare natural environments that are hypersaline (high in salts), but not hyperhaline (high in NaCl). Microbiologists have mainly worked in environments rich in NaCl and few studies have investigated epsomic environments. Early observational work at Hot Lake, WA, an area permanently saturated with MgSO₄, reported green sulfidic mats, consistent with Chlorobium, Oscillatoria, and Plectonema (Handy 1916; Anderson 1958; Hammer 1978, 1986). Microbes able to grow at high concentrations of MgSO₄, those that are epsotolerant, have been reported from soils and constructed environments (Markovitz 1961; Markovitz and Sylvan 1962; Laiz et al. 2000; Mandrioli and Saiz-Jimenez 2002).

Our community analysis at Hot Lake (Kilmer et al. 2014) was preceded by a study of the epsotolerance of a diverse collection of halotolerant bacteria from the Great Salt Plains (GSP) of Oklahoma, a hypersaline NaCl environment (Crisler et al. 2012). While selected for their halotolerance, the GSP isolates showed remarkable epsotolerance. Strong growth was observed in the presence of 50% MgSO₄ (~2 M as epsomite) by the vast majority of GSP bacterial isolates, including Bacillus and Halomonas. The isolate collection from Hot Lake soils was rich in Gram-negative bacteria, including Halomonas, Idiomarina, and Marinobacter (Kilmer et al. 2014). Bacillus, Marinococcus, Planococcus, and Virgibacillus were observed among the Gram-positive bacteria, as were the Actinomycetes Nesterenkonia and Nocardiopsis. No archaea were isolated, which may not be surprising given that halophilic archaea require higher NaCl concentrations than present in Hot Lake (Schneegurt 2012). Hot Lake bacteria were tolerant to 50% MgSO₄ and many grew at higher concentrations, including saturated MgSO₄ (Kilmer et al. 2014; Wilks et al. 2019). None of the isolates were epsophilic, requiring high MgSO₄ concentrations for growth. Random 16S rRNA gene clone libraries constructed from direct DNA extracts of Hot Lake soils recovered Gram-positive bacteria (e.g. Bacillus, Clostridia) and many clones related to uncultured Actinomycetes. The Gram-negative clones included Acidovorax, Coxiella, Legionella, and Deltaproteobacteria. Nearly one-third of the clones from Hot Lake were from uncultured candidate divisions. Deep sequencing of Hot Lake mat communities showed a predominance of cyanobacteria (Tychonema), Chlorobia, Chromatia (Halochromatium and Thiohalocapsa), Halomonads, and Verrucomicrobia (Lindemann et al. 2013).

Basque Lake is similar to Hot Lake in several respects, particularly in being an environment saturated with MgSO₄. It is a series of epsomic playas, with small pools interspersed by muddy soils. Due to the similarity of its composition to potential brines on Mars, it was the subject of a study using IR spectroscopy to detect the signatures of life (Hyde et al. 2007; Foster et al. 2010). While enrichment cultures for halotolerant microbes were positive, no isolates or microbiological analyses were reported. A study on the survival of microbes in evaporating brines of relevance to Mars, included inocula from Basque Lake (Fox-Powell et al. 2016). The initial community in sulfate-rich medium had moderate diversity, with Arthrobacter, Bacillus, and Virgibacillus. No microbes from Basque Lake formed colonies after the evaporation of sulfate-rich brines. A recent study used molecular methods to describe the microbial community of Spotted Lake, an epsomic lake in the same region as Basque Lake (Pontefract et al. 2017).

The microbial populations of Basque Lake were characterized in the current report using cultivation and molecular techniques. Sixty-two salinotolerant bacterial isolates were purified and then identified through phylogenetic analysis of 16S rRNA gene sequences. These isolates were further characterized by measuring their epsotolerance, halotolerance, and metabolic capabilities. Deep sequencing of rRNA genes from soil extracts revealed additional microbial diversity. The microbial community at Basque Lake is compared to those found at epsomic Hot Lake and the hyperhaline GSP. Our findings are placed in the astrobiology context of planetary protection and life detection.

Materials and Methods

Site Description and Sample Collection

Water and lake margin soils (~100 g from top 4 cm) were collected during summer using sterile tools and containers from Basque Lake, BC at three locations: S1 (50° 36' 02.5" N 121° 21' 32.2' W), S2 (50° 35' 59.8” N 121° 21' 27.6" W) and S3 (50° 35' 58.9" N 121° 21' 27.2" W) (Fig. 1). Composite samples were collected from five spots within a 1-m² areas of the raised mud that separates the epsomic playa pools and not from sediments or algal mats. Samples were mixed and divided into an aliquot for molecular work that was frozen in the field and kept frozen during transport on dry ice and an aliquot for live culture work that remained fresh at ambient temperature. The soils were predominantly a grey and black fine clay and silt, often with a strong sulfidic odor, and saturated with salts. The geochemistry and hydrology of Basque Lake has been reported previously (Hammer, 1978, 1986; Hammer and Haynes, 1978; Last and Slezak 1988; Nesbitt 2004; Last and Ginn 2005).

Figure 1.

Contextual map of Basque Lake showing sampling sites.

Microbial Enrichment and Isolation

Direct plating, liquid enrichment, and dilution plating were used to isolate halotolerant and epsotolerant microbes from Hot Lake waters and soils. The media used were based on SP medium (Caton et al. 2004), a nutrient-rich and moderately saline (10%, ~1.7 M) medium containing per liter: NaCl, 98 g; KCl, 2.0 g; MgSO₄·7H₂O, 1.0 g; CaCl₂·2H₂O, 0.36 g; NaHCO₃, 0.06 g; NaBr, 0.23 g; FeCl₃·6H₂O, 1.0 mg; trace minerals, 0.5 ml; tryptone, 5.0 g; yeast extract, 10.0 g; glucose, 1.0 g; final pH 7.0. This was prepared with either 10% NaCl, 10% (~0.4 M) MgSO₄, or 50% MgSO₄. While these media were useful for liquid enrichments, agar plates with 50% MgSO₄ would not gel sufficiently, so isolates from liquid enrichments in this medium were transferred to plates with 10% NaCl medium. Enrichment cultures (100 ml) were maintained as shake-flasks on a rotary shaking platform (2.5-cm stroke dia) at 150 rpm and incubated at 7, 25, or 37 °C, before aliquots (100 µl) were plated at 24 or 48 h after inoculation. For dilution plating, samples (10 g) were diluted 10-fold with appropriate media and serially diluted prior to plating. In some cases, soil aliquots (approximately 1 g) were spread directly onto the surface of plates. Culture plates were maintained in moist boxes and colonies were collected after several days or more.

Colonies arising on the plates were selected for isolation based on gross colony morphological and physiological features, including pigmentation, size, margin, or rate of growth. Colonies were transferred to fresh SP agar plates with 10% NaCl or 10% MgSO₄ and isolated using the streak plate method. Each isolate was subjected to at least five successive streak platings to ensure clonal purity. The isolates were archived as 50% glycerol stocks at –80 °C and as agar slants.

Characterization of Isolates

SP medium composition was modified with different concentrations of NaCl or MgSO₄ to measure salinity tolerances in liquid shake-tubes. Similarly, the pH of SP medium was modified to determine growth tolerances. Results are reported as the maximum and minimum conditions for growth. A threshold of 0.05 OD units was used for positive growth and inconclusive tests were repeated. Pearson correlation coefficients were calculated using 2-tailed tests within SPSS.

Characterization of the bacterial isolates followed protocols outlined previously (Caton et al. 2004; Litzner et al. 2006). All isolates were Gram-stained using the Protocol Gram-staining kit (Fisher Diagnostics) following the manufacturer’s instructions. Acid-fast staining used carbolfuchsin and endospores were stained with malachite green. Motility was assessed by examining wet mounts of 24-h cultures at 1000X and by stab inoculation of Sulfur-Indole-Motility medium (SIM; BBL). The addition of 3% hydrogen peroxide solution to confluent plates or smears of confluent culture on slides was used to detect catalase. Oxidase testing was performed using the BBL DrySlide system according to the manufacturer’s instructions. Amylase was determined after a 5-d incubation on Starch agar (Difco), followed by the flooding of plates with 25% stabilized Gram’s iodine solution. Hydrogen sulfide production was assayed using SIM medium; Kovac’s reagent was added after incubation to test for the production of indole. Urease activity was determined at 5 d using urea broth with phenol red indicator. Production of acid and gas from carbohydrates was tested using 0.5% (w/v) glucose, lactose, or sucrose in culture tubes containing inverted Durham tubes, using a 10% NaCl solution supplemented with tryptone (10 g l⁻¹) and phenol red (0.018 g l^–1; pH 7.3).

Phenetic trees of relatedness were generated using the numerical taxonomy package NTSYSpc 2.1 (Applied Biostatistics, Inc., NY), as previously described (Litzner et al. 2006). Each characteristic (44 total) was introduced as binary data to determine Jaccard similarity coefficients. The similarity matrix was converted into phenetic dendrograms using a UPGMA method (SAHN).

PCR, DNA Sequencing, and Phylogenetic Analyses

Crude DNA extracts from each isolate were prepared using a freeze-thaw technique as described in Caton et al. (2004). Genomic DNA in the supernatant was the target for PCR amplification of nearly complete 16S rRNA gene fragments using primers specific for the domain Bacteria (EUBPA: 5'–AGAGTTTGATCCTGGCTCAG–3' and EUBPH: 5'–AAGGAGGTGATCCAGCCGCA–3') (Edwards et al. 1989). PCR was performed in a thermal cycler (Eppendorf Mastercycler) as 25-µL reactions containing 0.2 μM of each primer, 1 U of ExTaq DNA polymerase and associated master mix (Takara), and 5 µL of cell extract. 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. PCR amplicons were single-pass sequenced by a commercial vendor (Eurofins Genomics, Louisville, KY) using the EUBPA primer. Raw sequences were trimmed to remove remaining vector regions. All sequences appear in GenBank with accession numbers MG461380 to MG461441.

Sequences were aligned using SILVA and maximum likelihood analysis was used to build trees in MEGA v7.0 (Kumar et al. 2016). Known contextual 16S rRNA gene sequences from type organisms were identified in GenBank using BLAST. The trees were rooted using a functional outgroup.

Pyrosequencing was performed using 454 Life Sciences technology by a commercial vendor. Direct DNA extracts were obtained from composite soil samples using a bead-beating and freeze-thaw method as previously described and amplified with 28F (5'–GAGTTTGATCNTGGCTCAG–3') and 519R (5'–GTNTTACNGCGGCKGCTG–3') bacterial primers. Over 19,000 reads were obtained and analyzed. Sequences were processed using MOTHUR v.1.39.3 according to the 454 SOP (https://www.mothur.org/wiki/454_SOP [accessed 5/9/2018]) with the following modifications (Schloss et al. 2009; Schloss and Westcott 2011). Denoised sequences were screened to permit a maximum homopolymer of 9 and a minimum length of 250 nts. Sequences were aligned to the MOTHUR-formatted version of the SILVA seed alignment, v. 132 (Quast et al. 2013). Aligned sequences were screened for a start position of 1046 within the alignment and optimized in total length to retain 80% of the sequences. Chimera checking was performed using the VSEARCH algorithm implemented in MOTHUR. Sequences were classified using the MOTHUR-formatted Ribosomal Database Project training set (version 16). Sequences classified within domain Eukarya, as mitochondrial sequences, or with unknown classification at the domain level, were removed from further processing. Ecological α-diversity metrics were calculated using the nseqs, coverage, sobs, invsimpson, simpsoneven, Chao, and Shannon calculators as implemented in MOTHUR.

Results

Collection and Characterization of Bacterial Isolates

All three of the composite soil samples from Basque Lake produced bacterial isolates through enrichment cultures. Sampling site and enrichment conditions are indicated within the phylogenetic trees (Figs. S1 and S2). Nearly all of the isolates were obtained from enrichments at 25 °C and 10% NaCl. Seven isolates were obtained at 37 °C and one at 30 °C. Eight isolates were collected from enrichments at 50% MgSO₄ (~2 M). While Marinococcus isolates produced characteristically orange colonies, none of the colonies arising from even eutrophic high-salt media enrichments at 37 °C were red, a trait often associated with halophiles, particularly haloarchaea that contain the photopigment bacteriorhodopsin.

Biochemical and physiological characteristics of the Basque Lake isolates are shown on Figs. 2–3 and Tables S1–S6. The vast majority of Basque Lake isolates (77%) stained Gram-positive and no isolates assigned to Gram-negative phyla by 16S rRNA sequencing stained Gram-positive (Fig. 2a, Table S1). All of the Virgibacillus isolates, and no others, produced endospores. Nearly all of the isolates were motile, with only two Gram-positive isolates (Marinococcus sp. str. BL17 and Staphylococcus sp. str. BL61) being nonmotile. While all but one (Staphylococcus sp. str. BL61) of the Gram-positive isolates were oxidase-positive, only 29% of the Gram-negative isolates were oxidase-positive. All isolates were catalase-positive. No isolates produced indole from tryptophan and only three expressed amylase (Halomonas sp. str. BL1, Salinivibrio sp. str. BL46, and Staphylococcus sp. str. BL61). The Marinococcus and Staphylococcus isolates expressed urease. Sulfide production was nearly absent for Gram-positive isolates (except Virgibacillus sp. str. BL14, Salinivibrio sp. str. BL46, and Staphylococcus sp. str. BL61), but was present for ~80% of Gram-negative isolates. Fermentation was widespread among the Basque Lake isolate collection, with the vast majority of isolates fermenting glucose, lactose, and sucrose (Fig. 2b, Table S2). Marinococcus isolates showed no fermentation, five Halomonas strains (BL 30, 48, 50, 58, and 59) only fermented glucose, and Salinivibrio sp. str. BL46 did not ferment sucrose. Only two isolates produced gas during fermentation, Salinivibrio sp. str. BL46 with glucose and Staphylococcus sp. str. BL 61 with lactose.

Figure 2.

Biochemical and physiological assays of Basque Lake bacterial isolates. The value within each segment of the column is the percentage of positive organisms within the Gram-positive (solid) and Gram-negative (open) isolates. The value above each column is the percentage of positive organisms among all of the isolates. a: assays of amylase, catalase, endospore formation, Gram reaction, motility, oxidase, sulfide production, and urease. b: assays of fermentation of glucose, lactose, and sucrose.

Figure 3.

Tolerances of Basque Lake isolates to environmental conditions as determined by growth in liquid culture. BL46 is a Salinivibrio and BL61 is a Staphylococcus. The threshold for positive growth was 0.05 OD. a: pH tolerance; none of the isolates grew at pH 4. b: epsotolerance. c: halotolerance.

Basque Lake bacterial isolates generally grew better at basic pH than at acidic pH (Fig. 3a, Table S3). All of the isolates grew at pH 8. Nearly 80% of the isolates grew at pH 9 and about one-third (31%) grew at pH 10. Only three isolates, Halomonas sp. str. BL 48, 58, and 59, grew at pH 11, the highest pH tested. All three of these, along with BL31, were Halomonas that did not grow below pH 7. Twelve isolates (19%) grew at pH 5 and none of the isolates grew at pH 4. Cold tolerance has particular relevance to astrobiology, given the physical conditions of Mars and the icy worlds. All isolates, except Staphylococcus sp. str. BL61, grew at 10 °C (Table S6). Of the 62 Basque Lake isolates, 55 (89%) grew at 7 °C and 45 (73%) grew at 4 °C.

Epsotolerance and Halotolerance in Bacterial Isolates

The Basque Lake isolate collection was screened for growth tolerance to high NaCl and MgSO₄ concentrations. Salinity tolerance ranges of bacterial isolates from Basque Lake are shown in Figs. 3b–3c and Tables S4–S5. Nearly all (85%) of the bacterial isolates grew at ≥50% MgSO₄ and about two-thirds (65%) grew at 60% MgSO₄ (Fig. 3b). Five isolates (Marinococcus sp. str. BL 10, 37, 38, and 57 and Virgibacillus sp. str. BL13) grew at the saturation point (67%) for MgSO₄ at room temperature. One Virgibacillus isolate (BL42) could not grow above 30% MgSO₄. Two isolates (Marinococcus sp. str. BL 12 and 43) either did not grow or grew particularly poorly at ≤1% MgSO₄.

The Basque Lake isolates showed remarkable growth tolerance to high NaCl concentrations, although this saline environment is not dominated by NaCl or other monovalent cations. All isolates grew at 10% NaCl and 71% of the isolates grew at 20% NaCl (~3.5 M NaCl) (Fig. 3c). Note that the water activity of 20% NaCl (0.82) is substantially lower than that of saturated MgSO₄ (0.90). This is due in part to limited dissociation of MgSO₄ in solution. Some of the isolates (19%) showed growth at 30% NaCl, near saturation. Two isolates (Marinococcus sp. str. BL 12 and 43) either did not grow or grew poorly at ≤1% NaCl.

Halotolerance and epsotolerance were positively correlated (r = 0.609, p = 0.010) for the 62 Basque isolates. However, there was not always a direct correspondence for individual isolates. For instance, Virgibacillus sp. str. BL6 grows at 60% MgSO₄, but only at 10% NaCl (degree of saturation of 90 and 31%, respectively). The correlations between tolerances to these two salts were greater than those seen previously for bacterial growth tolerances to high salts and sugars (Fredsgaard et al. 2017a). The conclusion there was that specific solute effects, and not simply water activity, were working to set degrees of tolerance.

Numerical Taxonomy of Isolates

Characterization of the Basque Lake isolate collection included 44 phenotypic parameters that were used to construct dendrograms (Fig. 4). Six phenoms can be distinguished at a coefficient of 0.40. With few exceptions bacterial isolates clustered on the numerical taxonomy dendrograms within the phylogenetic groups determined by 16S rRNA gene sequencing. The lone Gram-negative Salinivibrio isolate (BL46) was a singleton on the numerical taxonomy dendrogram and clustered within the Gram-positive clade. Two Marinococcus isolates (BL 12 and 43) formed a cluster distinct from the other Marinococcus and Virgibacillus isolates. These two isolates grew poorly (or not at all) at ≤1% salt (Fig. 3c).

Figure 4.

Dendrograms showing the relatedness of Basque Lake bacterial isolates based on 44 phenetic characteristics. Taxonomic assignments were determined by phylogenetic analysis of 16S rRNA gene sequences.

Phylogenetic Groups Recovered

Representatives of the most common colony types from each Basque Lake sample were collected, although this increased the degree of duplication in the overall collection. Phylogenetic analysis was performed on 16S rRNA gene sequences to identify the taxonomic placement of each isolate (Figs. 5 and S1–S2). Gram-positive bacteria dominate the Basque Lake culture collection, however, representatives of only three clades were recovered (Figs. 5a and S1). Isolates related to Marinococcus halophilus and Virgibacillus marismortui accounted for all but one of the Gram-positive isolates. The Marinococcus isolates produced BLAST correspondences to bacteria found in hypersaline environments (Tang et al. 2011; Kilmer et al. 2014; Li and Yu 2015) and to our previous isolates from epsomic Hot Lake (e.g., HL 6–9, 29, 60, 72, and 88). The Virgibacillus isolates had BLAST similarities to isolates obtained from hypersaline environments and to our previous isolates from the Great Salt Plains (e.g., GSP17) (Caton et al. 2004; Caton and Schneegurt 2012; Kim et al., 2012; Li and Yu 2015). It is interesting to note that many of the bacterial isolates had sequences nearly identical to those obtained from Death Valley soils or from Yuncheng Lake in China, which is rich in SO₄ and Na, with elevated Mg and reduced Cl.

Figure 5.

Phylogenetic trees of bacteria isolated from Basque Lake based on most likelihood analysis of aligned 16S rRNA gene sequences. Expanded trees with collateral data and GenBank accession numbers are given in Figs. S1 and S2. a: Gram-positive bacterial isolates. b: Gram-negative bacterial isolates.

The Gram-negative isolates from Basque Lake (Figs. 5b and ^S2) include a Salinivibrio and species of Halomonas, which are widely found in hypersaline environments (Caton et al. 2004; Tang et al. 2011; Kilmer et al. 2014; Li and Yu 2015). The Halomonas group that includes BL 48, 58, and 59 showed BLAST correspondence to isolates from haloalkaline environments, some rich in sulfates (Montoya et al. 2013; Kilmer et al. 2014). The isolates in this clade did not grow below pH 7 and were the only isolates to grow at pH 11 (Fig. 3a). The closest BLAST matches for Salinivibrio sp. str. BL46 are again from Yuncheng Lake, which is rich in Na₂SO₄ (Li and Yu 2015).

Culture-Independent Community Analysis

Pyrosequencing of PCR-amplified 16S rRNA genes in direct DNA extracts was used to examine the bacterial community in Basque Lake soils. This represents a snapshot of the bacterial community at a particular location and time. There is likely considerable variability among different areas of Basque Lake and across temporal and seasonal scales. Samples were taken from muddy soils rising between epsomic playa pools. The distribution of amplicons into phyla indicated that Proteobacteria were the predominant bacterial group, encompassing 60% of the sequences recovered (Figs. 6 and S7). The abundance of Betaproteobacteria (27%) was greater than the abundance of Alphaproteobacteria (19%) and more than twice that of the Gammaproteobacteria (12%). Betaproteobacteria were mainly within Burkholderiales. Rhizobiales, Rhodobacteriales and Sphingomonadales were the main contributors to the Alphaproteobacteria. Gammaproteobacteria were predominantly Chromatiales and Oceanospirillales. Other Gram-negative bacterial groups were minor components of the community. Sulfate-reducing Deltaproteobacteria were observed. Gram-positive bacteria represented 11% of the library, with 95% clustering within Firmicutes, mainly Bacillales and Clostridiales. Bacteroidetes were minor populations (1.3%) within this community. Unclassified bacterial sequences represented 27% of the library.

Figure 6.

Relative abundance of major bacterial phyla recovered through pyrosequencing of 16S rRNA gene sequences in direct DNA extracts from a composite sample of Basque Lake soil. Numerical data is presented in Table S7.

At the 97% identity level for 16S rRNA gene sequences, 955 species were observed. Chao estimators predict that the bacterial community contains 4354 species (95% CI of 3573 – 5369) and Good’s coverage was 0.79. Bacterial diversity appears high for the extreme environment of Basque Lake based on a Shannon index of 5.07 (95% CI of 5.00 – 5.14) and an inverse Simpson index of 32.37 (95% CI 29.62 – 35.68). Simpson’s E value was low (0.034) indicating that species are not evenly distributed within the community.

Discussion

Environments rich in NaCl are abundant on Earth, from the oceans to solar salterns to hyperhaline lakes and salt flats. The communities and growth tolerances of microbes from hyperhaline environments have seen extensive study for over a century (Grant 2004; Schneegurt 2012). Natural environments dominated by salts besides NaCl are relatively rare and consequently have been investigated far less frequently (Handy 1916; Anderson 1958; Markovitz 1961; Markovitz and Sylvan 1962; Boring et al. 1963; Hammer 1978, 1986; Laiz et al. 2000; Mandrioli and Saiz-Jimenez 2002; Nesbitt 2004; Hyde et al. 2007; Foster et al. 2010; Crisler et al. 2012; Lindemann et al. 2013; Kilmer et al. 2014; Pontefract et al. 2017). Epsomic lakes and mud flats rich in MgSO₄ present a different challenge to microbes, since the ions in these environments are predominantly divalent. It is not clear what impact divalent ions have on microbes, how microbes respond and adapt to high concentrations of divalent salts, and how microbial responses compare to those observed with high NaCl.

A fundamental difference between solutions rich in NaCl and those rich in MgSO₄ is the result of differing levels of dissociation among these salts. Water activity is much more greatly reduced by NaCl than MgSO₄ at the same concentration (or degree of saturation), since MgSO₄ does not fully dissociate in water (Crisler et al. 2012; Kilmer et al. 2014). Previous work has demonstrated that water activity is not the sole physical parameter limiting microbial growth in media with very high solute concentrations, whether these are salts or sugars (Fredsgaard et al. 2017a). Our studies of bacteria isolated from epsomic Hot Lake have revealed broad tolerances to a wide range of MgSO₄ concentrations, with strong growth for most isolates to at least 2 M salt (Kilmer et al. 2014). Basque Lake bacteria have exhibited broad tolerances to MgSO₄ exposure in the same way. Broad tolerances may indicate that the cells are exposed to variable salt concentrations in their environment, such as might occur with snowmelt that dilutes briny playa pools.

Basque Lake may have the highest concentrations and greatest dominance of divalent ions of any surface waters on Earth (Hammer 1978, 1986; Haynes and Hammer 1978; Last and Slezak 1988; Last and Ginn 2005). Both Basque Lake and Hot Lake present environments that are saturated with respect to epsomite. Our earlier work demonstrated growth of GSP and Hot Lake isolates in 60% MgSO₄ (Crisler et al. 2012; Kilmer et al. 2014). Here we have extended this range to demonstrate microbial growth in media essentially saturated with MgSO₄ (~67%). MgSO₄ is the only salt besides NaCl where microbial growth has been observed in a saturated salt solution and this is the highest concentration of Mg (~2.7 M) for which growth has been demonstrated (Schneegurt 2012). The astrobiology implications of this observation for deliquescing salts on Mars are discussed below. While Basque Lake is an environment rich in divalent ions, the isolates from Basque Lake grew well in high concentrations of NaCl. Certain isolates were able to grow in NaCl near saturation, as well as saturated MgSO₄ (~67%). Overall, there was a positive statistical correlation between halotolerance and epsotolerance for the Basque Lake isolates. However, this was not the case for every individual isolate.

The isolate collection from Basque Lake was not as diverse as a similar collection from Hot Lake (Kilmer et al. 2014). This could reflect true differences in these communities or biases in the enrichment and selection of colonies for purification. Gram-negative Proteobacteria seem to be particularly undersampled. However, the predominant genera captured at Basque Lake also were abundant members of the Hot Lake community, as well as found in the GSP community (Caton et al. 2004; Kilmer et al. 2014). Halomonas, Marinococcus, and Virgibacillus (and other Bacilli) dominate the culturable community of these hypersaline sites. Basque Lake and GSP did not yield the Actinomycetes recovered from Hot Lake. Initial deep sequencing of Basque Lake amplicons yielded 16S rRNA sequences from cyanobacteria, Clostridia, and sulfate-reducing bacteria that were not targeted in the current enrichment scheme. Similar analysis of epsomic Spotted Lake found a predominance of Proteobacteria, Firmicutes, and Bacteroidetes, with similarities to the Basque community (Pontefract et al. 2017). Community analysis performed through cultivation and culture-independent analyses do not match in the current study given their very different criteria. The cultivated isolates have corresponding representatives in the deep sequence libraries, but many of the taxa observed were not captured in the cultivated collection. Identification of isolates by phenetic characteristics was not pursued the necessary characteristics for taxonomic assignment of microbial isolates were not all assayed.

While archaea, as haloarchaea, are often in high abundance in hyperhaline environments, this does not seem to be the case for all epsomic environments. No archaea were recovered from either Hot Lake or Basque Lake, despite using enrichment conditions known to select for archaea (increased Mg, high amino acid content, 37 °C; Schneegurt 2012). Archaea are not entirely absent from these locations since archaeal sequences can be PCR-amplified using primers specific for the domain Archaea. Initial studies indicated that these were nearly all within the methanogens, and not the haloarchaea (Vaishampayan et al. 2013; Chen et al. 2016). Archaea were reported from epsomic Spotted Lake (Pontefract et al. 2017). Haloarchaea typically require at least 1.5 M NaCl for growth and the NaCl cannot be replaced by another salt (Mohr and Larsen 1963; Mullakhanbhai and Larsen 1975; Onishi et al. 1980; Vreeland and Martin 1980). Hot Lake and Basque Lake do not contain this level of NaCl.

Salts in the regolith of Mars appear to be dominated by divalent ions, with sulfates being more abundant than chlorides, and generally the sulfate has Ca, Fe, and Mg counterions (Clark and van Hart 1981; Clark 1993; Wänke et al. 2001; Vaniman et al. 2004; Clark et al. 2005; Gendrin et al. 2005; Altheide et al. 2009). Liquid water is a key requirement for life and given the cold environment of Mars, ephemeral liquid water is expected to be hypersaline, whether generated from melting permafrost or by deliquescing salts (Cull et al. 2010; Lanza et al. 2010; McEwen et al. 2011; Möhlmann and Thomsen 2011; McKay et al. 2013). The subsurface oceans of Ceres, Enceladus, Europa, and the other icy worlds also appear to be salty and perhaps rich in MgSO₄ (Kargel et al. 2000; Marion et al. 2003; Hussmann et al. 2006; Mottl et al. 2007; Postberg et al. 2011; Nimmo and Pappalardo 2016).

Heavy brines on Mars formed through deliquescence may exist as eutectic solutions with very low freezing points (Chevrier and Altheide 2008; Möhlmann and Thomsen 2011; Clark and Kounaves 2016; Rivera-Valentin et al. 2018). The eutectic conditions for epsomite are 43% (w/w) at –4 °C. Initial studies with psychrotolerant isolates obtained from Basque Lake by enrichment cultures at 4 °C and 50% MgSO₄, showed slow growth under eutectic conditions (Wilks et al. 2018). The Curiosity rover has found that atmospheric humidity is high enough for (per)chlorate salts to deliquesce and these may be the most likely sources of liquid water on Mars (Chevrier et al. 2009; Martín-Torres et al. 2015; Ojha et al. 2015). The eutectic points for these salts are ~40% and –40 °C (or lower), conditions unlikely to support microbial growth. However, initial studies have demonstrated bacterial growth at the potassium chlorate eutectic (3% at –4 °C) and room temperature growth was observed at >25% (per)chlorates (Al Soudi et al. 2017; Wilks et al. 2018). Our findings will inform planetary protection efforts that seek to reduce contamination on robotic spacecraft, especially for life detection missions (Kargel et al. 2000; Marion et al. 2003; Hussmann et al. 2006; Mottl et al. 2007; Tosca et al. 2008; Fox-Powell et al. 2016). On this note, it also appears that epsotolerant microbes are found in common soils and spacecraft assembly facilities (Fredsgaard et al. 2017b; Eberl and Schneegurt, unpublished observations).

Supplementary Material

10482_2019_1244_MOESM1_ESM

Figure S1. Phylogenetic tree for Gram-positive bacteria from Basque Lake based on 16S rRNA gene sequences. GenBank accession numbers, sampling site, enrichment temperature, and enrichment medium (10% NaCl or 50% MgSO₄) are given for each isolate.