Soil biochemical characteristics
There were significant differences (P < 0.01) in pH, SOM, TS, SO42-, Cl-, UA, APA, and DHA in soil samples (Table 1). TS contents ranged from 3.81% to 24.35%, salt concentration exceeds certain thresholds (0.2%) (Dion and Nautiyal, 2008), indicating a high hypersaline environment in soil samples. DS.ST showed the highest SOM (14.56 g/kg), TS (24.35%). DS.CST had the highest SO42-, Cl-, DHA and the lowest APA. OS.CT had the highest UA and the lowest SOM and DHA. The lowest TS (3.81%) and UA (0.05 mg/g) were found in MS.SCT. SO42- was lowest for OS.ST. MS.ST had the highest APA and the lowest Cl-.
Alpha diversity patterns
Total effective sequences of soil samples ranged from 47,376 to 64,521 (Table 2). Coverage of all samples was more than 97%, demonstrating that the sequence was sufficient. Analysis of the Chao and ACE indexes and Shannon and Simpson indexes showed that OS.SCT had the lowest species richness and diversity.
Change in bacterial community compositions in soil samples
Venn diagrams revealed the total observed OTUs in soil samples (Fig. 2). The numbers of unique OTUs for MS.CT, MS.CST, MS.SCT, and MS. ST was 620, 634, 720 and 780, respectively. The number of shared OTUs for 4 soil genera was 1270, accounting for 17.52% of all observed OTUs (Fig. 2A). The number of specific OTUs in OS.CT, OS.CST, OS.SCT, and OS.ST were 979, 542, 642 and 1151, respectively. The total number of OTUs for 4 soil genera were 741(Fig. 2B), accounting for 10.18% of the total OTUs. The number of specific OTUs in DS.CT, DS.CST, and DS.SCT was 612, 785 and 1166, respectively. The total number of OTUs for 3 soil genera was 1706 (Fig. 2C), accounting for 28.13% of the total OTUs. The number of specific OTUs in BS.CT and AS.MgS was 2318 and 1596, respectively. The total OTUs for bacteria was 2073 (Fig. 2D), accounting for 34.63% of the total OTUs.
The main bacterial phyla included Euryarchaeota, Proteobacteria, Bacteroidetes, Actinobacteria, Firmicutes, Gemmatimonadetes, Acidobacteria, TM7, Tenericutes, and Verrucomicrobia (Fig. 3). The relative abundances of these phyla together made up an average of 93.70% for all bacteria. ANOVA revealed that except Firmicutes and Tenericutes, other bacterial phyla showed significant differences in all samples (P < 0.01 or P < 0.05). Except MS.ST, OS.ST, OS.CT, and BS.CT, Euryarchaeota was the most abundant in 9 soil genera, whereas Proteobacteria had the greatest abundance in other 4 soil genera.
Average relative abundance of the top 10 microbial dominant classes accounted for 68.44% of all bacterial classes (Fig. 4). Halobacteria (s.e.m=0.156**), Gammaproteobacteria (s.e.m=0.035**), Clostridia (s.e.m=0.037**), Cytophagia (s.e.m=0.021*), Alphaproteobacteria (s.e.m=0.029**), Rhodothermi (s.e.m=0.025**), Gemm-4 (s.e.m=0.014**), Acidimicrobiia (s.e.m=0.013**), and MJK10 (s.e.m=0.009**) except Bacilli varied significantly in soil samples.
Relationship between bacterial community compositions and edaphic factors
SOM, pH, TS, SO42-, Cl-, UA, DHA, and APA were closely correlated with the abundance of the dominant bacterial phyla (Table 3). The abundance of Proteobacteria (r=-0.319, p=0.047) was significantly negatively correlated with SOM. The abundance of Proteobacteria (r=0.386, p=0.015), Bacteroidetes (r=0.456, p<0.01), Actinobacteria (r=0.492, p<0.01), Firmicutes (r=0.332, p=0.039), and Gemmatimonadetes (r=0.499, p<0.01) were significantly positively correlated with pH, while Euryarchaeota (r=-0.622, p<0.01) was significantly negatively correlated with pH. The abundance of Euryarchaeota had a markedly positively relationship with TS (r=0.512, p<0.01) and the Proteobacteria (r=-0.343, p=0.032), Bacteroidetes (r=-0.407, p=0.01), Actinobacteria (r=-0.325, p=0.043), and Gemmatimonadetes (r=-0.561, p<0.01) exhibited a highly significant negative correlation with TS. The abundance of Euryarchaeota (r=0.425, p<0.01) and Tenericutes (r=0.317, p=0.049) were significantly positively correlated with SO42-, whereas Proteobacteria (r=-0.560, p<0.01), Gemmatimonadetes (r=-0.451, p<0.01), and Acidobacteria (r=-0.495, p<0.01) were significantly negatively correlated with SO42-, then Gemmatimonadetes (r=-0.387, p=0.015) was significantly negatively correlated with Cl-, while Tenericutes (r=0.346, p=0.031) was significantly positively correlated with Cl-. Additionally, there was a significant positive correlation between Firmicutes (r=0.483, p<0.01) and UA, Acidobacteria (r=0.354, p=0.027) and DHA, Bacteroidetes (r=0.616, p<0.01) and APA.
The two redundancy analysis (RDA) axes explained 61.5% of the variation between the soil bacterial communities (Fig. 5). The distinctions of bacterial community structure among 5 subtypes and 13 genera of soil were also supported by the redundancy analysis (RDA). Soil pH value, TS, SO42-, and UA had significant effects on bacterial community compositions (P = 0.001 by the Monte Carlo permutation test) (Table 4).