Spatial and temporal variability of microbial communities and salt distributions across an aridity gradient before and after heavy rains in the Atacama Desert

Over the past 150 million years, the hyperarid core of the Atacama Desert has been transformed by geologic and atmospheric conditions into one of the most unique and inhospitable landscapes on the planet. This makes it an ideal Mars analog that has been explored for decades as preliminary studies on the space life discovery. However, two heavy rainfalls that occurred in the Atacama in 2015 and 2017 provide a unique opportunity to study the response of resident extremophiles to rapid environmental change associated with excessive water and salt shock. Here we combine geochemical analyses with molecular biology to study the variations in salts and microbial communities along an aridity gradient, and to examine the reshuffling of hyperarid microbiomes before and after the two rainfall events. Analysis of microbial community composition revealed that soils within the southern desert were consistently dominated by Actinobacteria, Proteobacteria, Acidobacteria, Planctomycetes, Chloroflexi, Bacteroidetes, Gemmatimonadetes, and Verrucomicrobia; soils within the hyperarid sites were dominated by Aquificae and Deinococcus-Thermus before heavy rainfalls, while these organisms almost totally diminished after rainfall, and the hyperarid microbial consortia and metabolisms transformed to a more southern desert pattern along with increased biodiversity. Salts at the shallow subsurface were dissolved and leached down to a deeper layer, both benefitting and challenging indigenous microorganisms with the excessive input of water and ions. Microbial viability was found to change with aridity and rainfall events but correlated with elevation, pH, conductivity, chloride, nitrate, sulfate, and soil organic matters (SOM). Metagenomic functional pathways related to stressor responses also increased in post-rainfall hyperarid soils. Our findings contribute to the primary goal of this study sheds light on the structure of xerophilic, halophilic, and radioresistant microbiomes in hyperarid environments, and their response to changes in water availability. supernatant filtrate. Cell pellets resuspended for propidium iodine and SYTO® 9 fluorochromes LIVE/DEAD® BacLight™ Bacterial Viability The total microbial concentrations and viability were an Accuri™ C6 Flow Cytometer Biosciences) The abundance and viability of AT-17 microbial communities were analyzed via duplicate trypan blue staining assay within one month of samples collection, and cultivation methods within one month and replicate within half a year after sample collection. Duplicate AT-17 microbiological soils were suspended, and 10× serially diluted two to four times to clear out any sand particles. 0.4% Trypan blue was added to 9× of the dilute solution. Viable (non-colored) and non-viable (blue) microbial cells were mounted on a Hirschmann Instruments™ Counting Chamber and counted using oil immersion light microscopy (AmScope). Detailed microbial cultivations of AT-17 samples were elucidated AT-17 samples each homogenized

desiccation may not be the sole or even the primary factor influencing microbial life in desert environments (Hullar et al. 1996;Navarro-Gonzalez et al. 2003;Quinn et al. 2005;Reynolds et al. 2004;Schwinning et al. 2004;Shen et al. 2019;Shirey et al. 2012). Counter-intuitively, an abrupt increase in water availability in hyperarid soils is extremely harmful to xerophilic microorganisms because these cells are induced to transform from the defensive or dormant state to the metabolically active state while unexpectedly being exposed to attack from extreme temperature and UV radiation (Cockell et al. 2017). In addition, excessive water causes high osmotic shock to the microbial semipermeable membrane and disturbs their survival strategies adapted for limited moisture (Azua-Bustos et al. 2018;Stevenson et al. 2015). After the heavy rainfall in 2017 at the core region of the Atacama Desert, 75-87% of pre-rainfall species disappeared, and no viable archaea or eukaryotes were detected in undrained brines (Azua-Bustos et al. 2018). Although these rainfall events can severely damage the extremotolerant microbial communities, previous studies demonstrated that the community structure can recover as soon as one month after, or up to more than one year (Armstrong et al. 2016;Uritskiy et al. 2019), using a variety of biochemical mechanisms and osmoregulatory systems (Wood 2015).
Instantaneously after rainfall, microorganisms start producing proteins and metabolites that are crucial in fundamental biosynthetic pathways, energy supplements, desiccation resistance, radiation protection, and oxidation defense for the preparation of the upcoming hyperarid period .
Mars is likely to have been a much wetter planet around 4 billion years ago. Between then and the drier planet we recognize today, a transition dry period with occasional moisture may have occurred (Ramirez and Craddock 2018). Detection results by the Curiosity rover show evidence of temporary subsurface liquid during night-time on equatorial Mars and possibly beyond (Martin-Torres et al. 2015).
Therefore, the rare heavy rains in the hyperarid Atacama act as an analog to the transient availability of liquid water on Mars. In this study, we compared the differences in microbiome and soil salt compositions along a precipitation gradient of the Atacama Desert before and after two heavy rainfall events. Environmental factors were identified to investigate their impacts on local microbial communities. We used these data to understand the natural microbial compositions, networks, metabolisms, and adaptations to water perturbation in terms of both spatial and temporal variability.
Our results also highlight the environmental conditions that most impact microbial communities in hyperarid and less arid transition sites of the Atacama Desert, and the effects those conditions exert on the microbial communities within the two regions. Between November 30 and December 6, 2017, seven field sites at a longitudinal transect from 22°S to 29°S ( Figure 1) were selected for sampling as described in Shen et al. (2019). Briefly, samples were obtained from three hyperarid sites (MES, PONR, and Yungay) and four transition sites (TZ-0, TZ-4, . Within each site, three pits were sampled in the same manner as AT-12 samples for geochemical analyses and DNA extractions, and the third sampling pit was additionally sampled for microbial cultivations. Similarly, soils obtained in 2017 are collectively referred as AT-17. The geographic locations, physicochemical properties, and local daily precipitation of each AT-17 study site during the 2017 heavy rainfall event was presented in Shen et al. (2019).

Soil characterizations
In order to characterize AT-12 samples for soil pH and conductivity measurements, triplicate soil slurries were prepared by adding 1 g of soil to an appropriate volume of deionized water. The pH and electrical conductivity were measured with a YSI Model 30 Handheld Salinity, Conductivity and Temperature System (YSI Inc., Yellow Springs, Ohio, USA) after 15 min agitation at 220 rpm and settling for 5 min. For carbon (C) and nitrogen (N) quantification, triplicate pre-dried (100°C overnight) AT-12 geochemical soil samples were ground in mortar and pestle. Soils were stored in 20 mL scintillation vials. Subsamples of 0.1 mg were quantified for total soil C and N content in a Carlo Erba EA1108 (Isomass Scientific Inc., Calgary, Canada). Lastly, soil organic matter was determined as the loss of mass via ignition at 500°C after carbonates were removed using a 24 hour incubation at room temperature with 0.5 M HCl.
General soil properties (pH, electrical conductivity, major sediment elements by X-ray fluorescence (XRF) methods, and the concentrations of total organic carbon (TOC), total organic nitrogen (TON), carbonate, and nitrate) of AT-17 geochemical samples were reported in Shen et al. (2019). SOM in AT-17 samples were determined as the difference between the mass loss of ignition measured by XRF and the carbon dioxide loss calculated based on carbonate. Both AT-12 and AT-17 geochemical samples were sieved through 1.4 mm prior to ion chromatography for soluble anion determinations as described in Shen et al. (2019). Quantitative X-ray diffraction (XRD) analysis for AT-12 was reported in Harris et al. (2016). AT-17 soil samples for XRD were crushed using a Planetary Micro Mill (PULVERISETTE 6, FRITSCH) and sieved through a 355-μm sieve. The mineralogy of these crushed and sieved samples was analyzed on a Philips X-Ray Diffractometer (PW1830 generator, PW1050/80 goniometer, PW1710 diffractometer) using Co Kα radiation. Generator settings for the measurement were 30 kV, 30 mA, 3-70° range 2θ scan, 0.01° step, 1 s/step. Mineral phases were identified using the EVA 2 software from SOCABIM with the ICDD PDF-2 database.

DNA extractions and sequencing
AT-12 microbiological samples were stored at room temperature and extracted for DNA sequencing within 1 month. Triplicate 1 g of AT-12 microbiological soil samples from each site for DNA extractions were desalted with 1 mL sterile deionized water, followed by a wash with 10% sodium acetate and twice with deionized water. Samples were centrifuged at 3,000 × g for 10 min and the supernatants were decanted after each wash. Total soil DNA was extracted from desalted samples using the MP Biomedicals™ FastDNA™ SPIN Kit for Soil following a modified manufacturer's protocol: the cell lysis step was extended to an overnight incubation at 50°C with gentle agitation. Following extraction, the triplicate DNA extracts from each location were pooled to a final volume of 100 µl. DNA extracts were amplified for barcoded 454 pyrosequencing using Platinum® PCR Supermix (Invitrogen, USA) and 16S rRNA primer pair 515F (5'-GTGCCAGCMGCCGCGGTAA-3') and 806R (5'-GGACTACHVGGGTWTCTAAT-3') (Caporaso et al. 2012

Salt and water amendments
To inspect the effects of a variety of excessive dissolved salts on the viable microbial community, soils from each site were amended with salt solutions. Solutions of 10% Sodium chloride, 10% sodium sulfate, 10% sodium carbonate, 10% sodium acetate, and 10% sodium L-lactate were prepared in ultrapure water and autoclaved at 121°C for 30 min. After cooling to room temperature, 10 g of AT-12 soil was combined with the salt solution for 4 days at 21°C (Shen et al. 2019). Duplicate salt-amended soils were suspended in appropriate volume of sterile ultrapure water and spread on tryptic soy agar, LB agar, and plate count agar plates. Plates were sealed with Bemis™ Parafilm™ M Laboratory Wrapping Film and incubated for 20 days at 21°C prior to cell counting (Bagaley 2006). Colony forming units (CFUs) were determined by the multiplication of the number of colonies, dilution factor, and 1.45 to account for the addition of 4.5 mL solution to 10 g soil.
In addition, 4.5 mL, 3 mL, and 1.5 mL of sterile ultrapure water was added to each 10 g of AT-17 microbiological samples, which covers about the 1/3, 2/3, and full of soil area, respectively. Since the bottom diameter of Petri dishes used for culturing experiments is 90 mm, these volumes of water are equivalent to approximately 0.24, 0.47, and 0.71 mm daily precipitation, respectively. CFUs were cultured and determined in the same manner as salt amendments. CFUs from cultivation experiments were converted via common logarithmic transformation. CFUs on salt amendments were standardized by subtracting their respective 4.5-mL water amendment groups; and CFUs on water amendments were standardized by subtracting their corresponding groups without any amendments.

Data analyses
Multiplex 454 sequence reads of AT-12 bacterial sequencing results were pre-processed with the open source package Quantitative Insights into Microbial Ecology (QIIME) (Caporaso et al. 2010b). Reads were filtered by removing those of low quality, with ambiguous characters and missing barcoded primers, of a length less than 150 base pairs, and with quality scores lower than 25. The remaining reads were aligned using Python Nearest Alignment Space Determination (PyNAST) (Caporaso et al. 2010a).
Illunima-sequenced AT-17 bacterial reads were combined and demultiplexed into Casava 1.8 pairedend format with QIIME 2 (Bolyen et al. 2019). Reads of low quality were denoised with DADA2 (Callahan et al. 2016). Reads were filtered by trimming the first 13 base pairs and removing those of low quality, with ambiguous characters and missing barcoded primers, of a length less than 300 base pairs.
AT-12 and AT-17 sequences were then dereplicated, merged, and further analyzed together in QIIME 2. Merged AT-12/ 17 sequences were clustered into operational taxonomic units (OTUs) at a 99% similarity cutoff. The quality of total clustered sequences was double confirmed via noise and chimera pattern checking by VSEARCH (Rognes et al. 2016) and UCHIME (Edgar et al. 2011).
The representative taxonomic identities were aligned at 99% full-length sequence homology using pretrained naïve Bayes classifier based on SILVA 132 marker gene reference database : the primer pair used to extract reference reads are 341F 5'-CCTACGGGNGGCWGCAG-3' and the combinative reverse primer 5'-GACTACHVGGGTWTCTAAT-3' as a conjunction of the 16S rRNA gene regions used for AT-12 and AT-17 sequencing. The reverse primers -785R, 806R, and combinative reverse primer -were compared in silico for differences in bacteria identification using the probe match webtool at the Ribosomal Database Project (https://rdp.cme.msu.edu/probematch) (Cole et al. 2014).
These three reverse primers matched 11,958, 11,857, and 11,970 type species of 30 identical bacterial phyla, respectively. Therefore, bias in the results due to the two different primer pairs was insignificant.
OTU tables were rarefied to remove sampling depth heterogeneity at 1,186 even sampling depth. Alpha diversity indices (binary logarithmic Shannon diversity, Faith's phylogenetic diversity, Pielou's Evenness, and observed OTUs) were computed with the alpha-phylogenetic package (Faith 1992) of QIIME 2. Alpha rarefaction curves were generated at the maximum 5,000 sequencing depth.
Characteristic bacterial phyla associated with the sampling year and aridity were predicted and clustered with the q2-sample-classifier package (Bokulich et al. 2018) showing the 100 most representative sequences. Analysis of composition of microbiomes (ANCOM) was employed to determine the genera that were significantly different among sampling year and aridity (Mandal et al. 2015). Hierarchical clustering and principal coordinate analysis (PCoA) of each individual site using Bray-Curtis distance matrix were performed in Past 3.26 (Hammer et al. 2001). Finally, since microbial community network closely links with their ecosystem function (Finlay et al. 1997), we conducted network analysis on Atacama microbiota based on OTU numbers. AT-12 and AT-hyperarid sites (AT-12 KM40, AT-12 Yungay, and 3 pits of AT-17 MES) and 17 transition sites (AT-12 TZ-4, AT-12 TZ-5, AT-12 TZ-6, AT-12 TZ-7, AT-12 TZ-8, AT-12 TZ-9, 2 pits of AT-17 TZ-0, 3 pits of AT-17 TZ-4, 3 pits of AT-17 TZ-5, and 3 pits of AT-17 TZ-6). OTUs with abundance of less than 0.001% of the total number of OTUs were removed, resulting in a final subset of 988 OTUs from the hyperarid core and of 8,636 OTUs from the transition zone. Using individual OTUs and intercorrelation as nodes and connecting edges respectively, the co-occurrence pattern of OTUs that represents microbial network was defined based on their Spearman correlation coefficients calculated by the Hmisc package (Harrell 2019) of R (https://www.r-project.org/). Co-occurrence was regarded as reliable when the Spearman correlation coefficient was both greater than 0.6 and statistically significant at p < 0.01 (Barberan et al. 2012;Junker and Schreiber 2008). Network diagram was drawn and network statistics (i.e., edge number, cumulative degree distribution of nodes, average path length, eigenvector centrality, centralization of graph according to node betweenness, and modularity) were computed using the igraph package (Csardi and Nepusz 2006) of R.

Soil geochemical features
To investigate Atacama soil physicochemical context, samples were analyzed by total C, N, and soil organic matter (SOM) quantifications, X-ray diffraction (XRD), X-ray fluorescence (XRF), and ion chromatography (IC). In AT-12 soils before heavy rainfalls, the pH ranged from 8.0 to 9.5, with the lowest pH in KM40 and the highest in TZ-5. Electrical conductivity of AT-12 ranged from 0.03 to 22.30 mS/cm, with the highest conductivity in KM40. Soil carbon content ranged from 0.1% in KM40 and TZ-7 to 1.3% in TZ-5 and TZ-6. Soil nitrogen content ranged from 0.06% in TZ-7 to 0.32% in KM40. SOM ranged from 0.57% in KM40 to 4.90% in TZ-8 (Table 1). X-ray diffraction (XRD) results showed significant amounts of quartz, albite, microcline, gypsum, anhydrite, and chloride among all AT-12 sample sites ( Figure 2a) (Harris et al. 2016). AT-17 sites were similarly characterized by quartz, albite, and microcline. However hydrated minerals, brushite and gypsum, took over a significant percentage in the two post-rainfall hyperarid sites -PONR and Yungay. (Figure 2b & Figure S1). Chloride minerals, primarily halite, were ~3.5% in AT-12 sites but became undetectable in all AT-17 sites. The AT-17 soluble salts (i.e., chloride, nitrate, and sulfate) remained at about the same level as AT-12 in the sediments of the transition zone. However, soluble salts in AT-17 reduced remarkably in the hyperarid sites after heavy rainfalls compared to AT-12 samples ( Figure 2 & Table S1).  The areas of pie charts range from 17 ppm to 270,000 ppm.

Bacterial abundance and viability
To assess the abundance and viability of Atacama bacterial communities, flow cytometry, trypan blue staining assay, and cell culture experiments were employed to Atacama soil samples.  Table S2). Although no significant difference in total cell counts was found between the hyperarid and southern transition sites either within AT-12 or Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 28 February 2020 doi:10.20944/preprints202002.0433.v1 AT-17 samples (Figure 4 & Table S2), the active viable bacterial colonies increased significantly toward the transition zone with higher annual precipitation ( Figure 5). The viable : non-viable ratio in hyperarid soils also increased sharply from 2012 to 2017. Despite a large variation from 0.6 to 6.1 in the viable : non-viable ratios, the number of viable cells and total cells in the transition zone stayed stable ( Figure 4 & Table S2). Soil microbial communities show different preferences to the water volume on different agar plates (Bagaley 2006;Balestra and Misaghi 1997;Drees et al. 2006;Knief et al. 2019;Navarro-Gonzalez et al. 2003;Sieuwerts et al. 2008). Water amendment cultures (1.5, 3 and 4.5 mL) demonstrate that transition sites were slightly less water-limited than hyperarid sites. In general, bacteria from all Atacama sampling sites benefit from increased water, but growth rate did not always increase with further water addition, especially the 1.5 mL and 3 mL amendments (Figure 7 & Table S3 & Table S4). On average, excessive sodium chloride and sodium carbonate amendments decreased cell counts; sodium sulfate and sodium acetate had no effect; and sodium L-lactate increased cell counts. Sites with higher annual precipitation generally had a higher tolerance or higher preference to the amendments with these salts, except for sodium carbonate (Figure 7 & Table S5).

Ordination analyses of study sites
Hierarchical clustering indicated that the least diverse AT-12 transition site, TZ-6, had similar microbial compositions to the most humid transition sites, TZ-8 and TZ-9. AT-12 TZ-4, TZ-5, and TZ-7 were closely related to each other, and most similar to AT-17 TZ-6 ( Figure 10a). Three pits within each individual AT-17 site generally had similarity more than 85%, except for the 60% similarity in site TZ-  Table 2.

Pathway and network analyses
Metabolism structure of hyperarid AT-17 microbiomes was overall more similar to transition sites than hyperarid AT-12. Compared to transition sites, hyperarid AT-17 had less proportions of amine and polyamine biosynthesis, aminoacyl-tRNA charging, detoxification, glycan biosynthesis and degradation, nucleoside and nucleotide degradation, polymeric compound degradation, and secondary metabolite degradation. Hyperarid AT-12 microbiomes were generally the lowest in all metabolisms, but the highest in C1 compound utilization and assimilation, cell structure biosynthesis, metabolic regulator biosynthesis, nucleic acid processing, nucleoside and nucleotide biosynthesis, pentose phosphate pathways, photosynthesis, and secondary metabolite biosynthesis ( Figure 11).
Various adaptation pathways coupled with environmental stress reactions and salt/ small organic consumptions were predicted based on feature sequences ( Figure 12). Transition AT-12 and AT-17 sites had similar proportions of these pathways. Hyperarid AT-17 sites had slightly lower proportions of pathways coupled with stress reactions, but higher proportions of nitrate and sulfate utilization pathways. Hyperarid AT-12 sites had the lowest proportions of all these targeted pathways (Table 3).
Here, the co-occurrences of bacterial OTUs that are statistically significant in Spearman's rank correlation matrix are used to construct the edges of network (Barberan et al. 2012). The network analysis results revealed that the microbial OTUs of transition sites had more edges, higher edge to node ratios, and higher degrees (edges connected to a node) than hyperarid sites ( Figure 13 & Table 4), indicating a more densely connected community in the transition zone. OTUs from the hyperarid core showed a higher modularity (>0.4 indicates a modular structure) and an unconnected "island effect" (Neilson et al. 2017;Newman 2003). However within each hyperarid "island", the network was tighter, according to its shorter average path length and lower betweenness centralization score, than the whole transition community (Table 4).

Effects of water and salt regulations on microbial growth
In the Atacama Desert, gypsum, halite, anhydrite, atmospherically derived nitrate, and other inorganic salts accumulate in the soil (Ericksen 1983), leading to soil oxidation, variable electrical conductivity, and abiotic chemical decomposition of soil organic matters (SOM) (Ewing et al. 2007;Gómez-Silva et al. 2008;Quinn et al. 2007). The accretion of inorganic salts is a rare phenomenon that occurs only under extremely arid conditions and makes the Atacama Desert an ideal environment to examine the influence of extreme aridity on subsurface terrestrial or even Martian microbial "life" We found that the viable and total cell counts were not significantly different between hyperarid and transition sites in terms of the order of magnitude (Figure 4 & Table S2). However, microbial viability in Atacama soils was variable, with the lowest in the northern hyperarid soils in 2012 ( Figure 4 & Table   S2). Across the AT-12 sampling transect, the percentage of viable cells ranged from 31% (KM40) to 75% (TZ-4), much lower than typical soils with viable cell counts which usually surpass 90% (Janssen et al. 2002;Parinkina 1973), yet within the range of viability in soils associated with increased selective pressures (e.g., desiccated mineral soils, high UV radiation, low nutrient, and water availability) (Hansen et al. 2007;Saul-Tcherkas and Steinberger 2011;Shi et al. 1997). The harsh environmental conditions associated with hyperaridity, a reduced environment with low water and nutrient availability, represent the most likely cause of this difference in microbial viability between the two regions. However, in AT-17 sampling sites, the microbial viability decreased from northern hyperarid soils, up to 89%, to southern transition soils, down to 39% (Figure 10c & Table S2). This increased viability in the hyperarid region could be a consequence of the loss of non-viable cells as an additional implication of the rainfall leaching effects (Artz et al. 2005;Pruett et al. 1980).

The microbial cultivation experiments (without any amendments) demonstrated that the CFUs from
AT-17 samples increase up to 4 orders of magnitude from the hyperarid sites to transition sites ( Figure   5 & Table S3). The reasons for the difference between viable cell counts and culturable microbial colony numbers might be that 1) more unculturable species live in the hyperarid core (Ma et al. 2009;Warren-Rhodes et al. 2019), and that 2) most of the hyperarid microbiomes are metabolically inactive (Barros et al. 2008;Jones 1971;Navarro-Gonzalez et al. 2003;Schulze-Makuch et al. 2018;Zahran 1997). Since the Actinobacteria-dominated microbial community structure does not alter much between the hyperarid AT-17 and transition AT-17 sites (Figure 9 & Figure 10a, b), the latter hypothesis might play a more important role in the culturable CFUs.
The agar cultured bacteria from surface samples collected within the Atacama Desert are limited in diversity, and the majority are members of Actinobacteria and Firmicutes. A small amount of Proteobacteria and Bacteroidetes has also been recovered (Bagaley 2006;Navarro-Gonzalez et al. 2003).
More specifically, previously identified culturable bacteria belong to Geodermatophilaceae, Sphingomonas, Bacillus, Arthrobacter, Brevibacillus, Kocuria, Cellulomonas, Hymenobacter, Asticcacaulis, Mesorhizobium, Bradyrhizobium, Afipia, Alphaproteobacteria, and Betaproteobacteria (Gómez-Silva et al. 2008;Lester et al. 2007). On our agar plates, these bacterial taxa act as representatives of the whole microbial community from their sampled sites. Thus, the manner of growth and the change in CFUs of our cultivation experiments primarily reflect the preferences of bacteria within these taxa.
The water amendment experiments suggest that active culturable microorganisms are not impaired by rainfall that is less than 1 mm per day, as the volumes of water we added to these agar plates. Although the proliferation of some of these microbes manifested a decreasing trend after the abrupt water wash, more microbes benefited from the addition of water (Figure 7 & Table S4). However, when precipitation reaches as high as the two unprecedented rainfall events (38.6 mm and 19.6 mm, respectively) in a few days, massive water input can dissolve soluble salts and concentrate them down to more 20 cm depth (Davis et al. 2010;Marion et al. 2008;Shen et al. 2019).  Table S5) (Bagaley 2006). However, TOC content in the hyperarid core is positively associated with water soluble chloride (Figure 6c). This inconsistency of the relationships between microbial biomass and chloride in different forms might be explained as that: some chloride minerals such as halite are highly hygroscopic; the deliquescence process within chloride minerals provide moisture to microbial communities that inhabit inside (Davila et al. 2008;Davila et al. 2013;Finstad et al. 2017;Pfeiffer et al. 2019;Wierzchos et al. 2012); the soluble chloride concentration might indicate the portion of chloride that deliquesces. The positive correlation between TOC and soluble chloride might indirectly reflect the beneficial effect of salt deliquescence on microbial biomass, but not a benefit directly from chloride ions.
Furthermore, water soluble sulfate also contributes to microbial biomass as a long-term outcome (Figure 6d), which again does not agree with salt amendment experiments (Figure 7 & Table S5). Since sulfate in the Atacama Desert is usually formed in hydrated forms (e.g., gypsum, Figure 2), more sulfate indicates the larger proportions of gypsum that can potentially provide crystallization water to nearby microorganisms (Palacio et al. 2014). Additionally, these microorganisms can reduce sulfate for energy production and sulfur assimilation ( Figure 12 & Table 3). Therefore, higher sulfate concentrations can both elevate the moisture in a microhabitat and supply nutrients for microbial life.

Salt distributions and microbial communities along the aridity gradient
The degree of influence of water on life in desert environments is not entirely clear, and the paradigm defining the relationship between precipitation and ecosystem processes in arid environments has recently been questioned. In particular, this debate has proposed revisions to the "pulse-reserve" model that depicts rainfall as the dominant limiting factor for life in arid ecosystems (Reynolds et al. 2004;Schwinning et al. 2004). Our results indicate that microbial communities in Atacama soils were influenced by a variety of environmental factors including elevation, pH, conductivity, SOM, and various salts that change, for the most part, with latitude and aridity ( Figure 10c) (Bossio et al. 1998;Doran et al. 1998;Drenovsky et al. 2004;Gallardo and Schlesinger 1992;Phelps et al. 1994;Zeglin et al. 2009). At southern latitudes, the decrease in aridity and salt concentrations yielded soils that were relatively higher in microbial abundance and diversity (Figure 4 & Figure 5 & Table 2 & Table S3), while northern soils were less amenable to life, producing low diversity microbial communities with decreased activity and abundance (Jones et al. 2018;Schulze-Makuch et al. 2018).
Although the diversity of unknown phyla is high (Contador et al. 2019), the relative abundance of these species were not outstanding at all in our study sites (Figure 9). Soils of the southern arid desert were heterogenous in microbial composition and dominated by Actinobacteria, Proteobacteria, Chloroflexi, Firmicutes, Acidobacteria, Planctomycetes, Bacteroidetes, Gemmatimonadetes, and Verrucomicrobia ( Figure 9), phyla commonly detected in arid desert environments (Connon et al. 2007;Drees et al. 2006;Fernandez-Martinez et al. 2019;Holmes et al. 2000;Mandakovic et al. 2018;Meslier et al. 2018;Neilson et al. 2012;Smith et al. 2006). Hyperarid soils were shown to be homogenously composed of two phyla -Deinococcus-Thermus and Aquificae, which when combined represented ~94% of the soil microorganisms detected there. The other characteristic bacterial phyla in hyperarid AT-12 include Acetothermia, Armatimonadetes, Hydrothermae, and Thermotogae, bacteria within which generally adapt to nutrient-depleted and high temperature conditions (Hao et al. 2018;Jungbluth et al. 2017;Lee et al. 2014). Actinobacteria survived in the pre-rainfall hyperarid core of the Atacama Desert in 2012 were only belong to 9 species -uncultured Ilumatobacteraceae, Corynebacterium sp., Micrococcus sp., Lawsonella sp., uncultured Geodermatophilus sp., uncultured Rothia sp., uncultured MB-A2-108, uncultured RBG-16-55-12, and unclassified Acidimicrobiia. All species in classes Nitriliruptoria, Rubrobacteria, and Thermoleophilia, and most of species in classes Actinobacteria and Acidimicrobiia have disappeared after decades or centuries of the extreme hyperaridity, similar to the microbiota structure of the salt flat ecosystems in the Atacama Desert Farias et al. 2013;Ley et al. 2006;Mobberley et al. 2012;Rasuk et al. 2016;Rasuk et al. 2014;Sahl et al. 2008). Besides the long-term hyperaridity, the deficiency of extractable Actinobacteria in hyperarid AT-12 may be also caused by incomplete extraction or sequencing because of the limit of technology and the novelty of extremophilic actinobacterial species (Goodfellow et al. 2018;Idris et al. 2017). Additionally, some Actinobacteria such as its classes Actinobacteria, Acidimicrobiia, and Rubrobacteria can form spores that are difficult to break down during DNA extraction to counterattack ultraviolet radiation and dehydration (Barka et al. 2016). Analogously, the decreasing proportions of Firmicutes (majorly Bacilli, and minorly Clostridia, Erysipelotrichia, and Limnochordia) in the hyperarid core may be a result of their sporulation (Rose et al. 2011).
Deinococcus-Thermus bacteria possess a variety of adaptations to persist extreme soil conditions that are hypersaline, extremely dry, highly oxidizing and irradiative (Battistuzzi and Hedges 2009;Neilson et al. 2012;Paulino-Lima et al. 2013;Paulino-Lima et al. 2016;Perfumo et al. 2011). After heavy rains, within the phylum of Deinococcus-Thermus, dominance shifted from of the order Thermales, who are mostly thermoresistant, to a dominance of the order Deinococcales, who are both thermoresistant and radioresistant (Battista et al. 1999;Omelchenko et al. 2005). This shift might be triggered by the negative facet of water-driven microbial metabolic activation: since water woke dormant microbes up; these microbes disarmed sporulation and confronted the harsh ambient conditions, especially ionizing irradiation (Armstrong et al. 2016;Stevenson et al. 2015). Therefore, the Deinococcus-Thermus population shifted to a more Deinococcales-dominant structure.
Excitingly, this is the first discovery of bacteria belonging to the phylum Aquificae in the Atacama Desert, and especially of such a high abundance in any desert soils (Huang et al. 2011;Prieto-Barajas et al. 2018). Three uncultured Thermocrinis species in family Aquificaceae and one unspecified species in family Hydrogenothermaceae. In general, members of the phylum Aquificae are non-spore-forming and strictly thermophilic with an optimal growth temperature above 65°C (Griffiths and Gupta 2006;Horiike et al. 2009). Although conditions in subsurface soils of the Atacama are not within the optimal growth range, Aquificae bacteria have evolved enzyme-mediated adaptative mechanisms to counter oxidative stresses (Deckert et al. 1998), and they can autotrophically perform carbon fixation, sulphite oxidation, as well as hydrogen oxidation to water (Gupta and Lali 2013;Huber et al. 1992;L'Haridon et al. 2006). Species of Aquificaceae and Hydrogenothermaceae were branched early in the phylogenetical history (Acca et al. 1994;Coenye and Vandamme 2004;Hugler et al. 2007).
After  Figure S2). However, the structure of microbial communities in the hyperarid region completely shifted toward a pattern that is more common in the southern transition zone, to one dominated by Actinobacteria, Proteobacteria, and Chloroflexi ( Figure 9 & Figure 10a, b). This shift indicates the occurrence of a water-driven microbial community perturbation and reorganization in the hyperarid core (Dion 2008;Edwards 1990;Fernandez-Martinez et al. 2019). The intra-site similarity of microbial composition were ~80% (Figure 10a), and the similarity of a site within the transition zone between the 2012 and 2017 samplings still remained at high percentages, 60%-80%. The intra-site similarity of microbial composition were ~80% (Figure 10a), and the similarity of a site within the transition zone between the 2012 and 2017 samplings still remained at high percentages, 60%-80%.
These similarity results imply that the precipitation in the transition zone caused little change in the microbial communities during the 5 years between 2012 and 2017.

Microbial functions and metabolic responses
Desert soils are exhausted in many of the essential requirements that nurture life, including water, carbon, and nutrients. As a result, desert soil microbial communities are primarily composed of microbes adapted to withstand harsh environmental conditions, and adept at exploiting the scarce Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 28 February 2020 doi:10.20944/preprints202002.0433.v1 nutritional resources necessary for growth. In recent years, these microbiomes have also survived occasional rainfalls which can lead to disastrous destruction of indigenous microbial communities (Armstrong et al. 2016;Azua-Bustos et al. 2018;Uritskiy et al. 2019). In this study, both the hyperarid and transition soils were dominated primarily by bacterial phyla capable of withstanding harsh desert conditions and water osmotic stress both before and after the rains. Potentially existent Martian organisms should evolve similar mechanisms over their history of resistance to the capricious Martian environments.
In 2012 when massive rainfall had not disturbed hyperarid soils for at least several decades (Hartley et al. 2005;Jordan et al. 2019;McKay et al. 2003),  Table 3) and consistently fostered more frequently interacted microbial communities (Neilson et al. 2017) with more complete recycling of substrates between individual organisms ( Figure 13). Notably, these communities can coevolve to establish a more resilient biosphere for survival under extreme circumstances.
Recent studies have revealed that the hyperarid Antarctic Dry Valley soils similarly support microbial communities of greater complexity and diversity than previously described (Cowan et al. 2002;Pointing et al. 2009) Table S4) (Stevenson et al. 2015). The cycle of desiccation and rewetting decomposes dead microbial materials to release intracellular organics and uncovers concealed SOM by smashing soil aggregates (Armstrong et al. 2016), which furthermore offers nutrients to alive microorganisms that are evoked by liquid water from dormancy. However, most microorganisms are vanished when drown in brine (Azua-Bustos et al. 2018). Thus, our findings suggest feasibly detectable Martian "life" may concentrate at the regions of temperate temperature, i.e., around the equatorial regions, and during the day time after the evaporation of the film of liquid brine.

Conclusions
Subjected to millions of years of extreme aridity, soils of the Atacama Desert have become transformed into an inhospitable Mars-analogous environment that is unique even among other extremely arid environments. The Atacama microbial communities are one of the least diverse biological communities on Earth based on all alpha diversity indices Maier et al. 2004;Navarro-Gonzalez et al. 2003). Our results indicate a large disparity between hyperarid and arid soils in soil microbial viability, diversity, activity, metabolism, network, and community composition. As soils become drier toward the north, biological activity decreases, with microbial viability, diversity, and network significantly reduced under hyperarid conditions. Thus, our data support the hypothesis that the dry limit of microbial life is reached in the hyperarid soils of the Atacama (Azua-Bustos et al. 2012b; Barros et al. 2008;Navarro-Gonzalez et al. 2003). We have additionally shown that a variety of factors influence soil microbial communities, including environmental and physiological conditions. Increased precipitation and nutrient availability in the southern transition sites correlated with more diverse and more closely allied soil microbial communities, while soils of the hyperarid sites consisted of more metabolically inactive cells exhibiting low phylogenetic diversity. Confronting the challenges brought by the water and salt shocks as the aftermath of two recent massive precipitations, microbial communities in the hyperarid region shifted to a structure that is common in the southern transition zone. Although the microbial compositions and metabolic functions did not recover to the pre-rainfall level (Armstrong et al. 2016;Fernandez-Martinez et al. 2019;Uritskiy et al. 2019), the diversity of microbiomes and metabolisms was slightly increased within 6 months of these extreme rainfall events.
We therefore can infer similar events that may have happened on the wetter early Mars, as well as during the drying-rewetting cycle on the present Mars.

Availability of supporting data
The raw sequencing data for microbiome analyses of AT-12 and AT-17 in this study are available from the National Center for Biotechnology Information (NCBI) under BioProject ID PRJNA595727 and An OTU definition of sequence homology at or above 99%. Error bars of rarefaction analysis were shown.
Abbreviations as in Table 2.   Table S5. Results of cell cultures on tryptic soy agar, LB agar, and plate count agar plates with salt amendments, recording the change in the order of magnitude of CFUs with salt amendments (all numbers are scaled by logarithmic transformation and normalized as the difference from the plates with 4.5-mL water amendments).
Negative effects of different salts on microbial growth are underlined, and those caused the decrease in CFUs more than 1, 2, 3, and 4 orders of magnitude are labeled with *, **, ***, and ****, respectively.