The Salar de Atacama, located in the Chilean Central Andes, is the largest Quaternary halite deposit in the world (3064 km² and >900 m thick; Warren, 2010). This Salar is comprised of a porous halide (90%), the interstices of which are permeated with a sodium chloride brine rich in lithium, potassium, magnesium, and boron (Bevacqua, 1992; Risacher and Alonso, 1996). Laguna Tebenquiche, located in the northern part close to the core zone of the Salar (Risacher et al., 2003), is one of the largest water bodies in this system (Demergasso et al., 2008). The lake is fed by groundwater of Tertiary and Quaternary volcanic origin (Risacher and Alonso, 1996). Bacterial and archaeal microorganisms inhabiting this lake are subject to extreme environmental conditions, such as high solar radiation (incl. UV), extreme diel temperature fluctuations, extreme changes in salinity due to net evaporation, and high lithium, boron, and arsenic concentrations (Lara et al., 2012; Farías et al., 2014). “Extreme” environmental conditions are generally conducive of microbial mat development (Rothschild and Mancinelli, 2001; Dupraz and Visscher, 2005) and a variety of photosynthetic organosedimentary ecosystems were found in Laguna Tebenquiche (Farías et al., 2014). These ecosystems included non-lithifying laminated photosynthetic microbial mats characterized by copious amounts of exopolymeric substances similar to other hypersaline mats (Dupraz et al., 2004, 2009) and also endoevaporitic domes made up of gypsum harboring endolithic phototrophs such as those described by Canfield et al. (2004) and Oren et al. (1995).

Previous studies based on culture-dependent methods recovered a large number of bacterial and archaeal strains from Laguna Tebenquiche's water column and sediments. Isolates included moderately halophilic Gram-negative rods (Prado et al., 1991), moderately halophilic Gram-positive cocci (Valderrama et al., 1991), heterotrophic halophilic microorganisms (Prado et al., 1993), and extremely halophilic archaea (Lizama et al., 2001, 2002). Using morphological and physiological properties, the bacterial isolates were assigned to the genera Vibrio, Halomonas, Acinetobacter, Alteromonas, Psychrobacter, and Marinococcus, all of which grouped within the class Gammaproteobacteria. Archaeal strains were assigned to the genera Halorubrum, Haloarcula, Halobacterium, and Haloferax (phylum Euryarchaeota).

Demergasso et al. (2008) analyzed the bacterial community of the water column by DGGE fingerprinting at several locations of Laguna Tebenquiche during the winter and summer season. Their study revealed a heterogeneous community composition of which changed along a salinity gradient in the water column of the lake. The dominant phylum was Bacteroidetes, which in the most saline part of the lake comprised a cluster related to Salinibacter relatives and at intermediate salinities consisted of clusters distantly related to Psychroflexus spp. Within the Gammaproteobacteria, which encompassed the most abundant class, a cluster related to uncultured bacteria from Mono Lake (USA) dominated. In addition, a few clones could be assigned to the candidate division OP1 (currently reclassified within the phylum Parcubacteria), to uncultivated clones CS_B020 and BD1-5 from marine sediments and to sequences of the KB1 group found in sediments from hypersaline brines.

Recently, we reported on the bacterial diversity, mineral composition, and key metabolic activities in two lakes in the Salar de Atacama (Farías et al., 2014), Laguna La Brava, and Laguna Tebenquiche. In both lakes, discrete bacterial communities, and mineral compositions developed along the salinity gradient of the overlying water. In Laguna Tebenquiche, the most abundant bacterial 16S rRNA amplicons in mats and endoevaporites resembled Bacteroidetes, and the second-most abundant amplicons could be assigned to Proteobacteria (Alphaproteobacteria and Deltaproteobacteria). Surprisingly low cyanobacterial diversity was found, which was corroborated by a low abundance of chlorophyll a (Chla). Another recent study, focusing on phototrophic bacteria (Thiel et al., 2010) was motivated by the low Chla concentration and low abundance of cyanobacteria in Laguna Tebenquiche. That investigation found evidence for a new gammaproteobacterial lineage based on pufLM gene analyses and furthermore found that green sulfur bacteria could not be detected with molecular techniques but could be revealed by culture-dependent methods.

Our previous study focused on the bacterial 16S rRNA sequences in two lakes and included a non-lithifying microbial mat and a gypsum endoevaporite in Laguna Tebenquiche (Farías et al., 2014). In a consecutive field campaign documented here, and with the purpose of covering all microbial ecosystems associated to mineral precipitation, we increased the number different benthic microbial ecosystems to include one endoevaporitic domal mat, two microbial mats with different degrees of lithification, and two rhizome-associated concretions. These concretions were included in our study because they present lithified structures similar to microbialites, but in contrast form in association with plants. Also, in the present investigation we deployed primers, which amplified the V4 hypervariable region of both archaeal and bacterial 16S rDNA genes, in order to cover the total microbial diversity in each system with the purpose to investigate the low cyanobacterial presence determined in a previous study, as well as the vertical distribution of key functional groups, we determined the microbial diversity in discrete depth horizons in a non-lithifying microbial mat and an endoevaporitic mat. Our investigation enables a correlation of taxonomic diversity with geochemical gradients (e.g., oxygen and sulfide profiles) on a small (vertical) scale and with physicochemical and geochemical characteristics (e.g., salinity, water depth) on a large (horizontal) scale. The results of this study increase our knowledge of the genetic and metabolic diversity of the benthic microbial ecosystems in Laguna Tebenquiche and provide novel insights into the microbial processes in extreme ecosystems at high altitude.

Contemporary microbial mats, including microbialites, typically thrive in extreme environments such as High Altitude Andean Lakes (HAAL). Several of the HAAL ecosystems have been described (Thiel et al., 2010; Farías et al., 2013; Gomez et al., 2014; Rasuk et al., 2016), but relatively little is known about the microbial diversity of the community in relation to biogeochemistry, especially that of the benthic ecosystems. The current investigation is the first in which the total microbial diversity was evaluated and interpreted in relation to the geochemical and physicochemical characteristics. The comparison of different types of benthic microbial ecosystems (e.g., with varying degrees of lithification) in Laguna Tebenquiche, and also an in-depth analysis of separate depth horizons in two contrasting laminated systems unambiguously revealed that archaea comprise the bulk of the microbial diversity. Previous diversity studies in benthic ecosystems of HAAL using high-throughput genome sequencing focused on the bacterial composition only (Farías et al., 2014). Furthermore, the archaeal and bacterial diversity in the water column of several HAAL including Laguna Tebenquiche was determined using DGGE in combination with specific primers for each Domain (Demergasso et al., 2004). A high-throughput genomic sequencing approach also deploying different archaeal and bacterial primers was used in a lithifying microbial mat from a hypersaline lake in Kiribati (i.e., not a HAAL) and revealed that the bacterial diversity was approximately three times higher than archaeal (Schneider et al., 2013). Thus, to our knowledge, an accurate assessment of the relative contribution of archaea to the total community composition in any hypersaline microbial sedimentary system was lacking until the current study.

A variety of benthic microbial ecosystems (e.g., mats, rhizome-associated concretions, and endoevaporites) showing an increased degree of lithification developed along a salinity gradient in Laguna Tebenquiche. Archaea were dominant in all five sedimentary structures, with Euryarchaeota the single largest contributing phylum and the presence of Crenarchaeota reported for first time in HAAL ecosystems. Euryarchaeal OTUs comprised 62 and 97% of the total diversity in MA2 and EVD, respectively, which represent the sites at with the highest conductivity. The class Halobacteria, which organisms require Na⁺ for growth and therefore can be expected to thrive in hypersaline Laguna Tebenquiche (Grant, 2001; Oren, 2010), was a major contributor to this phylum. The water column of Laguna Tebenquiche contains a high arsenic concentration (0.07 mM; Table S1), which can promote halobacterial growth. A potential role for As cycling was documented in biofilms of another hypersaline, alkaline HAAL, Laguna Diamante, where halobacterial abundance reached 94% of total OTUs (Rascovan et al., 2016). Analysis of the Diamante biofilm metagenome revealed that genes for As(III) oxidation and As(V) reduction were ubiquitous, indicating that As cycling supported energy conservation (Rascovan et al., 2016). Similar to Laguna Diamante, we postulate that Halobacteria inhabiting Laguna Tebenquiche could use arsenic for bioenergetics purposes.

MA1, RAC1, and RAC2 contained euryarchaeal OTUs with gene sequences similar to those of methanogenic archaea from the class Methanobacteria, the family Methanomassiliicoccaceae (class Thermoplasmata) and the candidate division MSBL1. 16S rRNA sequences and fosmids of the uncultured candidate division MSBL1 have been retrieved from hypersaline anoxic environments such as the deep-sea hypersaline anoxic brines in the Mediterranean Sea (van der Wielen et al., 2005; Daffonchio et al., 2006; Borin et al., 2009; Yakimov et al., 2013) and a hypersaline microbial mat of a solar saltern (López-López et al., 2013). The majority of archaeal 16S rRNA sequences recovered from these ecosystems belonged to the uncultured candidate division MSBL1 and combined methane production that was observed suggests that this archaeal lineage is involved in methanogenesis at extreme salinities. Methane production has been observed in MA1 samples (Visscher, unpublished), validating a potential biogeochemical role for MSBL1-like organisms in Laguna Tebenquiche.

The most abundant euryarchaeal OTUs in RAC1, MA1, and MA2 were classified into the deep-sea hydrothermal vent Euryarchaeotic group 1 (DHVEG-1). The uncultured DHVEG-1 members were found in the water column and sediment of anoxic deep-sea hydrothermal vents no additional information is available (Takai and Horikoshi, 1999).

The crenarchaeal OTUs make up 4–17% of the sequences in mats and rhizome-associated concretions. Many of these OTUs could be classified as MBGB, the biogeochemical function of which is highly speculative, but, based on observations in hypersaline microbial mats, could involve sulfate reduction (Robertson et al., 2009).

In contrast to earlier work (Demergasso et al., 2008; Farías et al., 2014) that reported predominance of ubiquitous Bacteroidetes and Proteobacteria in Laguna Tebenquiche, the current study found that these phyla were not dominant. Instead, Plantomycetes, Firmicutes, and Acetotermia were the main contributors to bacterial diversity. This finding could result from primer bias: earlier studies (5,7) focused on the bacterial community, whereas the current investigation analyzed the total microbial community (bacteria and archaea). OTUs belonging to Planctomycetes, which was the most abundant phylum within the bacteria in MA1 and MA2 and the second and third-most abundant phylum in RAC1 and RAC2, respectively, were mainly associated with the class Phycisphaerae. This class was recently described by one cultured and several uncultured representatives retrieved from marine environments and soils (Fukunaga et al., 2009). Diverse organoheterotrophic capabilities allows organisms belonging to this class to colonize of a wide variety of ecosystems, ranging from aquatic to terrestrial habitats including several extreme environments. Phycisphaerae were previously found in microbial mats of hypersaline lakes in the Bahamas (Baumgartner et al., 2009a) and Kiribati (Schneider et al., 2013).

The OTUs assigned to the phylum Firmicutes, which were prevalent in RAC2 (Figure 4), comprised mainly of fermenter halophilic anaerobic members of the order Halanaerobiales. OTUs assigned to the phylum Acetothermia, especially the taxon KB1, were ubiquitous in RAC2 (25% of total sequences) and less in both mats (4–5% of total sequences; Figure 4). Members of this phylum are uncultivated thermophiles (Hugenholtz et al., 1998; Costa et al., 2009; Kim et al., 2012; Németh et al., 2014) and the partial reconstructed genome indicates that the reductive acetyl-CoA pathway is used for CO₂ fixation. It has been suggested that members of this phylum contribute significantly to primary production under anoxic oligocarbophylic conditions (Takami et al., 2012). It should be noted that mats are organic rich, but the bulk of this organic matter is made up by the complex structure of expolymeric substances (EPS) (Decho, 2000; Decho et al., 2005). The majority of this EPS is recalcitrant, notably in deeper layers, resisting microbial degradation (Braissant et al., 2009). As a result, especially the anoxic parts of EPS-rich mats and microbialites could be deprived of readily available organic substrates for respiration, thereby increasing the importance of anaerobic CO₂ fixation and methanogenesis in these systems.

Typically, cyanobacteria are the principle autotrophs in microbial mats (Visscher and van Gemerden, 1993; Visscher and Stolz, 2005). However, given the limited presence of cyanobacterial as well as proteobacterial OTUs in our samples, it is plausible that, in addition to the Halobacteria discussed above, other organisms such as Acetothermia contribute to community CO₂ fixation in Tebenquiche. The scarcity of cyanobacteria in the current study corroborates earlier observations of a low cyanobacterial presence in mats and evaporites from the Salar de Atacama (Farías et al., 2013, 2014; Rasuk et al., 2014, 2016). However, we cannot rule out that the cyanobacterial community comprises a few species with high abundance, and high metabolic activity. Similar observations of low cyanobacterial diversity were made previously by other investigators in a variety of mats and microbialites (McKay et al., 2003; Ley et al., 2006; Baumgartner et al., 2009a,b; Lynch et al., 2012). Alphaproteobacteria and Gammaproteobacteria, prevalent in MA1 and EVD, are typical inhabitants of microbial mats and microbialites (Dupraz et al., 2011) and may significantly contribute to primary production (van Gemerden, 1993). The depth profiles of sulfide (Figure 3) support the notion that sulfide-oxidizing bacteria, including both chemolithoautotrophs and photolithoautrophs, could also contribute to the organic carbon pool.

OTUs belonging to Chlorothixaceae and Anaerolinaeae (Chloroflexi), Rhodovibrio (Alphaproteobacteria), and purple sulfur bacteria classified into the family Chromatiaceae and Ectothiorhodospiraceae (Gammaproteobacteria) were found and most likely carrying out the anoxygenic photosynthesis in Laguna Tebenquiche.

Members of Bacteroidetes were present but scarce in all samples. Bacteroidetes are believed to be among the best-adapted organisms to growth under the wide range of physicochemical conditions found in Atacama Desert (Demergasso et al., 2004). In addition, this phylum was observed in a variety of other hypersaline systems including microbial mats (Sørensen et al., 2005; Ley et al., 2006), water and sediment samples (Demergasso et al., 2004, 2008, 2010; Dorador, 2007), and evaporites (Stivaletta et al., 2011; Farías et al., 2014; Rasuk et al., 2014).

The presence of the phylum Verrucomicrobia was strongly correlated to RAC1. In similar microbialites found under less saline conditions [e.g., stromatolites from Laguna Socompa (Argentina) and a microbialites in Laguna La Brava in Atacama Desert (Chile)], with a conductivity of less than conductivity 115 mS/cm (Farías et al., 2013, 2014), OTUs related to this phylum were also found. It is plausible that members of this phylum are ubiquitous in microbialites found in high altitude whenever the salinity is not excessively high.

To further investigate the impact of the limited cyanobacterial diversity discussed above and unusually high abundance of archaea, we determined the microbial diversity in discrete depth horizons.

Deinococcus sp. has been a model organism for investigating UV resistance (Arrage et al., 1993), and is frequently found in a variety of microbial mats (Skirnisdottir et al., 2000; Pagaling et al., 2012; Abed et al., 2014; Tytgat et al., 2014), including stromatolites of Shark Bay, Australia (Goh et al., 2009). In the non-lithifying mat MA1, a large proportion of 16S rRNA sequences in the surface layer were associated with the phylum Deinococcus-Thermus. Farías et al. (2013) suggested that elevated levels of UV radiation associated with high altitude ecosystems limit microbial abundance and diversity, e.g., in stromatolites found in Laguna Socompa (Argentina). Therefore, the high dominance of the phylum Deinococcus-Thermus in the surface layer was likely due to their UV resistance mechanism. We also found that crenarchaeal and acetothermial OTUs that classified as MBGB and KB1, respectively, both prevalent in deeper layers, were absent in the surface layer of MA1 where the oxygen concentrations peaked (Table S3).

Some metabolic quandaries could be addressed through analysis in individual layers of MA1 and EVD. As outlined above, the cyanobacterial contribution to primary production appeared limited. Anoxygenic phototrophic Alphaproteobacteria and Gammaproteobacteria and Chloroflexi were present in the surface and second layers of MA1 and in the second and third layers of EVD, immediately underneath the surface halite layer. This distribution was further substantiated by microelectrode measurements of oxygen distribution (Figure 3), showing a maximum concentration at 1.5–1.75 mm of depth in MA1 (corresponding to surface and second layer) and between 6 and 8 mm of depth in EVD (coinciding with the second and third layers). Some gammaproteobacterial OTUs in both MA1 and EVD were related to the order Chromatiales (Gammaproteobacteria), which has representatives containing bacteriochlorophyll a and b. Some of these Chromatiales grow under both oxic and anoxic conditions and are capable of photolithoautotrophic and chemolithoautotrophic metabolism (van Gemerden, 1993).

Both aerobic and anaerobic heterotrophic microorganisms were detected in MA1 and EVD. The main heterotrophs in all layers of EVD were aerobic haloarchaea. Euryarchaeal OTUs related to anaerobic and methanogenic archaea were found in MA1 and EVD, with increasing proportions of 16S rRNA gene sequences with increasing depth, where permanently anoxic conditions prevail. A plethora of anaerobic fermenters resided in all layers of MA1 and EVD, however mainly represented by Firmicutes. Sulfate reduction was likely carried out by Deltaproteobacteria in MA1, including in the surface and second layers, both of which are oxic. Studies performed in several microbial mats, including hypersaline mats in Guerrero Negro, Solar Lake, Kiritimati Atoll, Shark Bay, Texel, and the Bahamas revealed high sulfate reduction rates in oxic layers as well (Canfield and Des Marais, 1991; Fründ and Cohen, 1992; Visscher et al., 1992, 1998; Bühring et al., 2009; Pages et al., 2014; Wong et al., 2015).

Most diversity studies in microbial mats and microbialites to date focused only on bacteria and showed important contributions of Cyanobateria, Proteobacteria, and Bacteroidetes to the total diversity in these system. Although primer bias, mentioned above, clearly impact the outcome of molecular investigations, the primacy of bacteria over archaea has been reported in numerous benthic microbial ecosystems, including hypersaline mats of Kribati (Schneider et al., 2013), open marine stromatolites (Baumgartner et al., 2009a), and hypersaline microbialites in the Bahamas (Baumgartner et al., 2009b) and Cuatro Cienegas, Guerrero Negro, and Shark Bay (Burns et al., 2004; Ley et al., 2006; Valeria et al., 2007; Demergasso et al., 2008). The high relative abundance of archaea in the samples from Laguna Tebenquiche warrants a reassessment of the role of archaea in these benthic microbial ecosystems and necessitates a similar approach when investigating other microbial mats.

In conclusion, in Laguna Tebenquiche, Euryarchaeota was one of the most abundant phyla in all samples studied, notably in EVD, in which it represented 97% of the 16S rRNA sequences. Most of these euryarchaeal OTUs were classified within the class Halobacteria or anaerobic and methanogenic archaea. Halobacteria mainly grow by aerobic oxidation of amino acids, carbohydrates or alcohol (i.e., glycerol), but can also photophosphorylate under anoxic conditions (Hartmann et al., 1980). This suggests an important role for (an)aerobic heterotrophy and potentially methanogenesis carried out by Euryarchaeota in these benthic ecosystems. Planctomycetes played also a key role in mats and rhizome-associated concretion samples, notably the class Phycisphaerae, which are aerobic organoheterotrophs. In addition to cyanobacterial primary production, anoxygenic photosynthesis by Chromatiales, possibly Chloroflexi and the candidate family Chlorotrichaceae (in RAC1 and mat samples) could contribute to CO₂ fixation in the mat. Other taxa present that could play a metabolic role include the uncultured candidate division MSBL1, possibly be involved in methanogenesis, and the uncultured MBGB. Some members of Proteobacteria could contribute to sulfate reduction and the uncultured phylum Acetothermia could perform CO₂ fixation through the reductive acetyl-CoA pathway. The importance of understanding the taxonomic and metabolic diversity in Laguna Tebenquiche resides in the extreme conditions of this environment, which may provide new insights into microbial processes in the Earth's early history and potentially on habitable exoplanets.

AF performed data analysis, interpreted data and wrote paper. MR contributed in sampling and data analysis. PV contributed in work proposal, sampling, performed oxygen and sulfide profile analysis and wrote paper. MC, FN obtained funding for the original project idea and performed physicochemical analysis. DP performed mineral analysis. MP contributed in oxygen and sulfide profile analysis. AV contributed in data analysis. MF obtained funding for the original project idea, contributed in work proposal, sampling and wrote paper. All authors read and approved this manuscript.