Diversity of prokaryotic microorganisms in alkaline saline soil of the Qarhan Salt Lake area in the Qinghai–Tibet Plateau

The composition of microbial communities varies considerably across ecological environments, particularly in extreme environments, where unique microorganisms are typically used as the indicators of environmental conditions. However, the ecological reasons for the differences in microbial communities remain largely unknown. Herein, we analyzed taxonomic and functional community profiles via high-throughput sequencing to determine the alkaline saline soil bacterial and archaeal communities in the Qarhan Salt Lake area in the Qinghai–Tibet Plateau. The results showed that Betaproteobacteria (Proteobacteria) and Halobacteria (Euryarchaeota) were the most abundant in the soils of this area, which are common in high salinity environments. Accordingly, microbes that can adapt to local extremes typically have unique metabolic pathways and functions, such as chemoheterotrophy, aerobic chemoheterotrophy, nitrogen fixation, ureolysis, nitrate reduction, fermentation, dark hydrogen oxidation, and methanogenesis. Methanogenesis pathways include hydrogenotrophic methanogenesis, CO2 reduction with H2, and formate methanogenesis. Thus, prokaryotic microorganisms in high salinity environments are indispensable in nitrogen and carbon cycling via particular metabolic pathways.

www.nature.com/scientificreports/ The Qarhan Salt Lake, located in the northeastern Qinghai-Tibet Plateau, is the largest in China, consisting of ten modern salt lakes 38 , with a total area of 5856 km 2 . The Qarhan Salt Lake area has an extremely arid desert climate; the mean annual temperature is 5.33 °C, mean annual precipitation is approximately 24 mm, annual evaporation is approximately 3564 mm, average wind speed is 4.3 m/s, and relative moisture is 27.7% 39 . Qarhan Salt Lake is also the largest large-scale inland comprehensive salt deposit in China with industrial exploitation value of quaternary stone salt, potassium salt, magnesium salt, and high concentrations of boron, lithium, rubidium, cesium, bromine, iodine, and other valuable chemical elements. The salt lake is primarily composed of potassium and magnesium brine ores that coexist with solid and liquid. Approximately 90% of the sodium chloride has been deposited into stable solid mineral layers. In contrast, the remainder of the potassium, magnesium, lithium, boron, rubidium, cesium, and other minerals are primarily found in brine 39 . The Qarhan Salt Lake is an important resource for both industry and agriculture, and it has been studied for decades. Researchers are also interested in the microbes in this area because they are representative of an extreme high-salt environment. Zhu et al. 40 studied the core bacterial communities associated with hypersaline environments in lake water and sediments from the Qaidam Basin. Liu et al. 41 investigated Gammaproteobacterial diversity and carbon utilization in lakes on the Qinghai-Tibet Plateau in response to salinity, while Zhong et al. 42 studied the prokaryotic community structure influenced by salinity and ionic concentrations in plateau lakes of the Tibetan Plateau. However, there have been few studies on soil microorganisms in bare land and plant-covered saline-alkali land around salt lakes in this area, which merits further investigations.
Herein, we present a study of the prokaryotic community of hypersaline soil in the Qarhan Salt Lake area using high-throughput sequencing and the ecological function of prokaryotes in this area. This study aims to (1) improve our current understanding of the prokaryotic community in a previously uncharacterized inland hypersaline environment and (2) provide clues about how microbes adapt to the extreme environments of high salinity at high altitudes.

Methods
Sample collection. The sampling site is near the Qarhan Salt Lake (36°36′57″N, 95°11′24″E; altitude 2651 m) in the state of Qinghai-Tibet Plateau, China. Soil samples were collected during the summer, on July 14, 2020, at a temperature of 22 °C. Field experiment photographs are shown in Supplementary Fig. S1.
Five soil samples were collected; one from bare land (QSB) and the other four from the grassland (QSG1, QSG2, QSG3, and QSG4), with a distance of 100 m between each sample. Five sub-samples (100 g) were collected with a hand spade from the 0 to 10 cm layer, pooled, homogeneously mixed into one 500 g sample, and transported to the laboratory. The samples were sieved with a 5 mm test sieve (WSTYLER, USA) under aseptic conditions. A portion of the soil (250 g) was used to characterize soil properties, and the remainder (250 g) was stored at − 80 °C for sequencing. The contents of various elements in the soil samples were determined using the ZSX Primus IV X-ray fluorescence spectrometer (Rigaku, Japan) according to the manufacturer's instructions, and the results are summarized in Supplementary Table S1. DNA extraction, PCR, and sequencing. Total genomic DNA (gDNA) from soil samples (~ 500 mg) was extracted using the E.Z.N.A™ Mag-Bind Soil DNA Kit according to the manufacturer's instructions (OMEGA Bio-Tek, USA). The DNA yield was quantified with the Qubit3.0 DNA Test Kit (Life Technologies, USA). Purified DNA was used as the template for the amplification of 16S rDNA genes via polymerase chain reaction (PCR). Approximately 10-20 ng of gDNA was used as a PCR template for amplification.
For bacteria, the primers Nobar_341F (5′-CCT ACG GGNGGC WGC AG-3′) and Nobar_805R (5′-GAC TAC HVGGG TAT CTA ATC C-3′) 43 were used in the PCR, including a barcode on the forward primer. The PCR reactions were performed in 30 µL reactions for denaturation at 94 °C for 3 min, followed by 5 cycles consisting of Subsequently, Illumina bridge PCR compatible primers were introduced, and PCR was performed in 30 µL reactions containing 20-30 ng of PCR product of bacteria or archaea, which was used as template DNA, the primer F 1 µL, Index-PCR Primer R 1 µL, and 2 × Hieff® Robust PCR Master Mix (Yeasen) 15 µL. The PCR reactions included denaturation at 95 °C for 3 min, followed by 5 cycles of denaturation at 94 °C for 20 s, annealing at 55 °C for 20 s, extension at 72 °C for 30 s, and a final elongation step at 72 °C for 5 min.
PCR products were assessed via agarose gel electrophoresis. To obtain a uniform long cluster effect and highquality sequencing data, the library concentration was determined using a Qubit 3.0 fluorometer (Invitrogen, USA). Subsequently, the amplicons were loaded onto an Illumina HiSeq platform (Illumina, Inc. San Diego, CA, USA), according to the manufacturer's guidelines.
Bioinformatics and statistical analyses. Sequences were analyzed using a combination of USEARCH 11.0.667 and QIIME v1.8.0 44 . The sequencing primer connector of the Read 3' -end was removed from Cutadapt www.nature.com/scientificreports/ 1.18 45 . PEAR 0.9.8 was used to merge the pairs of reads into a sequence according to the overlapping relationship between paired-end reads (PE reads) 46 . Barcodes were removed from the multiplexed FASTQ files using the USEARCH python command script fastq_strip_barcode_relabel2.py. PRINSEQ 0.20.4 was used to remove the bases with a tail mass value below 20 reads, and a window of 10 bp was set. If the average mass value in the window was lower than 20, the back-end bases were cut off to filter the N-containing sequences and short sequences after quality control, and the low-complexity sequences were finally filtered out 47 . The FASTA files were de-replicated, abundance sorted, and singleton sequences were removed. The operational taxonomic units (OTUs) were clustered de novo using USEARCH 11.0.667 48 . The OTUs were then mapped back to the original reads, and an OTU table was produced. Taxonomy was assigned to OTUs using the BLAST method in QIIME and against the RDP 16 S database 2.12: http:// rdp. cme. msu. edu/ misc/ resou rces. jsp. Mothur 1.43.0 was used to determine the alpha diversity index 49 . Principal component analysis was used to reflect the differences and distances between samples using the vegan R package (v. 2.5-6).The relative abundances of bacterial taxa were summarized using the Venn diagram package (v. 1.6.20) for R 50 . OTU co-occurrence network analysis was conducted using the R graph package (v. 2.0.0) based on the Bray-Curtis distance metric. Redundancy analysis (RDA) was conducted to evaluate the association between community composition and environmental parameters using the RDA function of the vegan package for R (v.2.5-6) 51 . Correlation heat maps were used evaluate the correlation between microbial classification and environmental variables using R (v.3.3.0).

Results
Microbial community structure in soils around salt lakes. Microbial community composition was investigated via high-throughput Illumina sequencing. The number of bacterial and archaeal sequences in the five samples were 205,563 and 283,308, respectively. A total of 643 OTUs were recovered, comprising 611 and 32 bacterial and archaeal OTUs, respectively. The rarefaction curves of all samples were flat, indicating that the amount of sequencing data was sufficient (See Supplementary Fig. S2).
Alpha diversity analysis revealed that bacterial and archaeal community richness (Chao1), diversity (Shannon and Simpson), and evenness (Shannoneven) varied widely among the samples (Table 1). In particular, the lowest bacterial richness, diversity, and evenness were samples from QSG4, and the highest richness was QSB, with the highest diversity and evenness being QSG1. For archaea, the lowest richness and diversity were samples from QSG1, and the highest were samples from QSG2; the lowest evenness was QSB, and the highest was QSG1.
Alkaline saline soil prokaryotic β-diversity. Unweighted UniFrac distance metrics were used to estimate bacterial and archaeal β-diversity and identify dissimilarities between the samples. The principal coordinate analysis (PCoA) plot illustrated the dissimilarity of OTU composition; the first two principal components explained 79.18% (PCoA 1 + PCoA 2; bacteria) and 79.18% (PCoA 1 + PCoA 2; archaea) of the total variations (Fig. 2). For the analysis of multivariate homogeneity among groups, the analysis of similarities (ANOSIM) test was performed, and the results showed that there were no significant differences between the bare land and the grassland (p > 0.05). www.nature.com/scientificreports/ Bacteria from bare land and grassland shared 187 OTUs (Fig. 3A), and unique OTUs (102) were recovered from QSG3, a number that exceeded the unique OTUs found in bare land QSB (96) (Fig. 3B). For archaea, bare land and grassland shared seven OTUs (Fig. 3C), more unique OTUs (15) were recovered from QSG2, a number that also exceeded the unique OTUs found in bare land QSB (3) (Fig. 3D).
Potential correlations between microbial communities and soil variables. RDA was performed to reveal the relationship between microbial community structures and the soil variables. The first two RDA axes Spearman's rank correlation test was performed to clarify the relationship between environmental factors and prokaryotic composition (relative abundance at the genus level) (Fig. 4). For bacteria, Ralstonia and Cupriavidus were positively correlated with Mg 2+ and K + , whereas Mesorhizobium, Escherichia, Shigella, and Bradyrhizobium were negatively correlated with Mg 2+ and K + ; Burkholderia was negatively correlated with Na, whereas Chitinophaga, Phenylobacterium, and Mesorhizobium were positively correlated with the Na, and Phenylobacterium and Mesorhizobium were negatively correlated with P (Fig. 4A). For archaea, Halovivax was positively correlated with Mg 2+ and K + ; Halomicrobium and Methanobrevibacter were negatively correlated with Na but positively correlated with P; Methanomicrobium was positively correlated with Na (Fig. 4B). These findings suggest that soil variables are important contributing factors for the regulation of soil prokaryotes.
The co-occurrence network effectively reveals the relationship between individual group members and the entire ecosystem 55,56 . The co-occurrence network clusters suggest that core bacterial and archaeal taxa in alkaline saline soil are likely to collaborate and play a role in key metabolic steps in response to environmental changes. (Fig. 5). Thus, the study of physiological and metabolic characteristics belonging to these key species can help us understand the mechanisms of microbial adaptation to the environment.   Table 2 presents the number of sequence reads of the predicted genes involved in adaptation to a high-salt environment. The OTUs detected in all samples were compared with the FAPROTAX annotation rule in an automated manner; however, most OTUs could not be assigned to any functional group. Thus, only those OTUs that were successfully annotated were analyzed. Chemoheterotrophy, aerobic chemoheterotrophy, nitrogen fixation, ureolysis, nitrate reduction, fermentation, predatory, and exoparasitic were the most abundant bacterial functional groups (Fig. 6A). Methanogenesis, hydrogenotrophic methanogenesis, methanogenesis by CO 2 reduction with H 2 , chemoheterotrophy, methanogenesis using formate, dark hydrogen oxidation, nitrate reduction, and aerobic chemoheterotrophy were the most abundance archaeal functional groups (Fig. 6B). These functional groups provide directions for understanding the mechanisms of the adaptation of prokaryotes to high salinity environments.
The metabolic pathways of microbial consortia predicted by PICRUSt were further analyzed. Metabolic pathways were identified at three levels. The functions of bacteria and archaea related to the high-salt environment in level 1 include cellular processes (4.19-4.31%, 1.78-3.99%), environmental information processing (15.87-17.12%, 10.74-12.55%), genetic information processing (13.44-14.32%, 17.18-18.99%), and metabolism (49.29-49.62%, 46.69-52.02%). The distribution of bacterial and archaeal functions at level 2 was investigated further. For bacteria, the relative abundances of membrane transport, amino acid metabolism, carbohydrate metabolism, and replication and repair were enriched in the alkaline saline soil, with little difference between the samples (Fig. 7A). However, the relative abundances of amino acid metabolism, carbohydrate metabolism, membrane transport, energy metabolism, and translation for archaea were enriched in alkaline saline soil. There was a great deal of variations among samples (Fig. 7B). It is reasonable that bacteria and archaea may adopt  www.nature.com/scientificreports/ different strategies when coping with extreme environments, and the bacterial community is relatively stable, while the archaea community is reasonably different.

Discussion
Metagenomic technology is a powerful tool to explore microorganisms in extreme habitats and their environmental adaptation mechanisms 57 . Using this technique, we found that the predominant phyla within the bacterial communities were Proteobacteria (85.08%), followed by Bacteroidetes (10.37%), Firmicutes (2.99%), and Actinobacteria (0.34%), and Proteobacteria were ubiquitous across all samples in the soil of the Qarhan Salt Lake area (Fig. 1A). Numerous studies have revealed that the bacterial communities are dominated by Proteobacteria, followed by Firmicutes, Bacteroidetes, Cyanobacteria, Actinobacteria, and Verrucomicrobia 11,40,58 . Among the Proteobacteria, Alpha-(16.01%), Beta-(66.65%), Gamma-(2.18%), and Delta-Proteobacteria (0.18%) were detected in all samples (Fig. 1B). These taxa have previously been confirmed in other hypersaline environments 9,42,59,60 , which is consistent with the present results. Betaproteobacteria were dominant in the salt water and sediments from Chott El Jerid Lake (75%-95%) 11 , and other studies have revealed that Gammaproteobacteria 10,40,41 and Alphaproteobacteria were the dominant classes 61 . Bacteroidetes was the second most abundant phylum; it has been linked to nutrient conversion in lake sediments 62,63 . Its relative abundance in inland lakes is strongly correlated with the salinity gradient 42,64-66 . The network revealed two keystone OTUs assigned to the phyla Bacteroidetes, genus Prevotella (OTU19 and OTU13). Figure 5, combined with the abundance and widespread distribution, demonstrates its ecological significance in the alkaline saline soil.
The phyla Firmicutes and Actinobacteria were also found in hypersaline environments, which is consistent with our findings 11,67 . Actinobacteria can decompose cellulose and other organic materials in hypersaline environments 68 . Thus, the ecological role of Actinobacteria is particularly important in vegetation-covered salinealkali land.
The top ten bacterial genera (> 1% of all sequences) accounted for 88.98% of the microbial community. Burkholderia was the most abundant, followed by Phenylobacterium and Ralstonia (Fig. 1E). Consistent with other studies 11 , Burkholderia predominated in our samples, and it has previously been reported to degrade aromatic hydrocarbons 69 . Ralstonia was also a common taxon in hypersaline environments 11,70,71 .
Archaea play an important role in the carbon and nitrogen 72 . The results showed that archaea in the soil near Qarhan Salt Lake were dominated by Halobacteria (Fig. 1F). Previous research has also revealed that Halobacteria live in salt lakes and salterns and propagate in salt crystals 9 .
The dominant family in these samples was Halobacteriaceae (51.30%) (Fig. 1H), which is consistent with studies of archaea from Ebinur Lake Wetland 73 , heavy metal-contaminated soils 74 , salt pans around Bhavnagar Coast 75 , inland saltern ecosystems in the Alto Vinalopó Valley 76 , and Lake Gasikule of the Tibetan Plateau 42 . These results showed that Halobacteriaceae was the dominant family in the majority of terrestrial high-salt environments. Halobacteriaceae can accumulate large quantities of inorganic ions (K + , Na + , and Cl − ). Their intracellular proteins and macromolecules are not damaged by high intracellular salt concentrations 77 , ensuring their survival and dominance in high salt environments.
The genus-level composition of archaea varied greatly between samples (Fig. 1I). In particular, Methanomicrobium predominated in samples QSG1 and QSG3, whereas Methanobrevibacter predominated in sample QSB, which is uncommon in other related studies. Nevertheless, the core genus is significantly different from other www.nature.com/scientificreports/ high salt environments and represents a relatively unique archaea community. However, the percentage of unannotated archaea (56.96%) is remarkable. Network interactions between taxa can capture ecological community habitat preference and taxa interactions 78,79 . Statistically, in our prokaryotic microorganisms in the alkaline saline soil of the Qarhan Salt Lake area, several keystone OTUs with high degrees were identified (Fig. 5), indicating that these OTUs could make a crucial difference in the soil microbial ecosystem. The metabolism of these keystone taxa is likely to be critical for the overall stability of the ecosystem, maintaining a fragile ecological balance in high-altitude and high-salt environments. Thus, the dynamics of any identified keystone OTUs may have a significant impact on this ecosystem.
The majority of the bacterial and archaeal species in the microbial community in the Qarhan Salt Lake area's alkaline saline soil had genes involved in synthesizing halo-adaptation compounds such as ectoine, glycine betaine, glutamate, trehalose, and choline (Table 2). This result, similar to a study of bacterial communities in Lake Tuz, indicates that halophilic microbes' unique cellular enzymatic machinery enables them to effectively use hydrocarbons as their sole source of both carbon and energy 80 . www.nature.com/scientificreports/ Figure 6. Functional community heatmap. Predict gene families based on prokaryotic metagenomes by modeling genes from 16S rRNA data derived from the generated OTUs and its reference genome database using FAPROTAX (A-bacteria and B-archaea). Red colors correspond to higher relative abundances.