The Yellow River Delta in northern China is the most prominent newborn wetland (Wang et al., 2012). In this area, riverine sediment deposition in the intertidal zone of the Bohai Sea promotes the rise and desalination of the low tide flats. The middle and high tidal beaches are successively covered by alkali, Tamarix and reed vegetation from low to high tidal flats. There are large areas of Phragmites australis and S. salsa swamps in the Yellow River Delta. And the pH of the soil in Yellow River Deltais between 7.8 and 8.5, and the electrical conductivity (EC) of soil Yellow River Deltais is 0.24 dS/m~ 4.83 dS/m (Li et al., 2023).

In the present study, 18 soil samples were collected from the natural S. salsa-growing area (YDJ), artificial S. salsa restoration area (YDB), and nonrestoration area (BKS). Each sample was collected in a quadrat using the five-point sampling method. At each sampling station, the whole plant of S. salsa was excavated and associated plants were removed. Then, a brush was used to move the soil adhered to the surface of the roots of S. salsa into a sterile sampling bag. Bulk soils were also collected and then thoroughly mixed to obtain one bulk soil sample. In each sample, a 20 g soil sample was weighed in a 250 ml shaker flask with 100 ml water and placed on a reciprocating degree constant temperature oscillator for 30 min. After standing for 30 min, the extracts were filtered or centrifuged and collected in a 100 ml beaker until measured. Then, the EC of soil samples was measured using an electrical conductivity instrument. The pH of the samples was measured using a pH meter.

For bacterial community analysis, the V3–V4 regions of the 16S rRNA gene were amplified using the PCR primers 341F (5′-CCTACGGGNGGCWGCAG-3′) and 805R (5′-GACTACHVGGGTATCTAATCC-3′) (Cai et al., 2017). The primers ITS1F (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS 2R (5′-GCTGCGTTCTTCATCGATGC-3′) were used to amplify the transcribed spacer region ITS1 for fungal community analysis (Saravanakumar et al., 2017). The library was constructed using Illumina MixSeq for sequencing. The sequencing data are available at NCBI (https://www.ncbi.nlm.nih.gov/, accession numbers PRJNA949580 and PRJNA949571).

To explore the similarity of different samples in terms of species diversity, QIIME software (Version 1.9.1) was used to calculate alpha and beta diversity (Bolyen et al., 2019). The graph was drawn using R (Version 3.4.4) (McMurdie & Holmes, 2013). The Chao1 and ACE indices measure species abundance as the number of species. The Shannon and Simpson indices are used estimate species diversity. The variance of soil properties between two groups was determined using one-way analysis of variance (ANOVA) (Kim et al., 2009). Based on the random matrix theory, the OTU data were used to construct molecular ecological network models of soil microorganisms in three different groups. And the interactions among soil microorganisms were analyzed (Xiao et al., 2022).

In the present study, indices, including ACE, Chao1, Simpson, and Shannon, were used to assess the alpha diversity of bacteria ( Table 1 ) and fungi ( Table 2 ). Our results revealed that the ACE, Chao1, Simpson, and Shannon indices were highest in the YDJ group and significantly different from those in the YDB group (P < 0.05), suggesting the great abundance and diversity of soil bacteria in the natural S. salsa-growing area. No significant differences were observed for the four parameters between the YDB and BKS groups (P > 0.05).

For fungi, the ACE (1077.1960) and Chao1 (1077.1667) indices were higher in the YDJ group than in the YDB and BKS groups, suggesting the great abundance of fungi in the natural S. salsa-growing area. However, the differences in the Simpson and Shannon indices were not significant among the three groups, indicating that S. salsa does not strongly affect fungal diversity.

In this study, 26,663 OTUs were found in three groups of soil bacteria. The OTU numbers of YDJ group were higher than in the BKS and YDB groups ( Figure 1 ). In total, 981, 242, and 363 OTUs were shared by the BKS and YDB, YDB and YDJ, and BKS and YDJ groups, respectively. Additionally, 11,619 OTUs were found in three groups of soil fungi. The OTU numbers of YDJ group were higher than in the BKS and YDB. Of these, 976, 994, and 952 OTUs were shared by the BKS and YDB, YDB and YDJ, and BKS and YDJ groups, respectively. The YDJ group had the greatest number of OTUs for both bacteria and fungi.

The microbial community structure at the phylum level in the three groups is presented in Figure 2 . Proteobacteria, Bacteroidota, and Gemmatimonadota were the dominant bacterial groups (accounting for 53.0% of all species) in the BKS group. Proteobacteria, Bacteroidota, and Acidobacteriota dominated in the YDB group (62.9%), and Proteobacteria, Desulfobacterota, and Chloroflexi were the major contributors in the YDJ group (57.5%). It was obvious that Proteobacteria had the greatest abundance in all three groups (30.9%, 42.9%, and 40.3% in the BKS, YDB, and YDJ groups, respectively). Regarding fungi, Ascomycota, Basidiomycota, and Olpidiomycota were dominant in all three groups (accounting for 90.4%, 85.9%, and 85.7% of fungi in the BKS, YDB, and YDJ groups, respectively). It could also be observed that Olpidiomycota had great abundance (13.5%) in the BKS group but low abundance in the YDB and YDJ groups (2.7% and 2.8%, respectively).

At the genus level, the top 30 microbial groups are presented in Figure 3 for bacteria. Except for the unclassified species, the Woeseia (3.1%) and Pelagibius (1.3%) were the dominant bacterial genera in YDJ group; the Limibaculum (2.8%), Halomonas (1.7%), and Lactobacillus (1.3%) were the dominant bacterial genus in YDB groups; the Lactobacillus (2.6%), Ardenticatena (1.8%), Sphingomonas (1.2%) and Rehaibacterium (1.1%) were the dominant bacterial genera in BKS groups.

For fungi, the top 10 microbial groups are presented in Figure 3 . The Olpidium, Thermoascus, and Fusarium dominated in the BKS group (28.3%); Thermoascus, Cladosporium, and Fusarium dominated in the YDB group (18.4%); and Xeromyces, Fusarium, and Thermoascus dominated in the YDJ group (28.9%). Xeromyces was the most abundant genera in the YDJ (15.2%), and the Olpidium was the most abundant genera in BKS groups (13.5%).

PCA enables the classification of multiple samples, further demonstrating the differences in species diversity among the samples. Bacterial PCA plots revealed a contribution of 25.55% for PC1 and 16.12% for PC2. In comparison, fungal PCA plots revealed a contribution of 57.69% for PC1 and 14.56% for PC2 ( Figure 4 ). For bacteria, the YDB samples were more distant and dispersed from those in the other two groups on the axes in the two coordinate plots, indicating that the YDB group had low similarity with the other two groups in terms of bacteria composition. For fungi, the YDJ group was more distant and dispersed from the other two groups on the axes in the two coordinate plots, indicating that the YDJ group had low similarity with the other two groups in terms of fungal composition.

At the bacterial level, the molecular ecological network of the bacterial community and the leading network characteristic parameters were calculated to describe the structure of the bacterial molecular network. The R² of the exponential distribution relationship of the three groups (BKS, YDB and YDJ) were 0.754, 0.821, and 0.697, respectively. The similarity thresholds were 0.920, 0.940, and 0.920, which aligned with the power law ( Table 3 ). At the fungi level, the molecular ecological network of the fungal community and the leading network characteristic parameters were calculated to describe the structure of the fungal molecular network. The R² of the exponential distribution relationship of the three groups (BKS, YDB and YDJ) were 0.743, 0.557, and 0.603, respectively. The similarity thresholds were 0.910, 0.880, and 0.910 in three groups (BKS, YDB and YDJ), which aligned with the power law ( Table 4 ). Our result indicates that the node relationship in the constructed bacterial and fungal molecular network model is reasonable and practical and can be further analyzed.

The total number of nodes and connections in a microbial molecular network can reflect the size of the network and the complexity of the relationships among species. This study’s BKS, YDB, and YDJ bacterial networks comprised 294, 347,319 nodes and 1420, 1265, and 1232 edges, respectively. This study’s BKS, YDB, and YDJ fungal networks contained 166, 200, and 235 nodes and 492, 539, and 688 edges, respectively. The results showed that the number of bacterial connections and total nodes and modules in BKS soil was the highest, the connectivity degree was the highest, and the interaction between species was the strongest ( Figure 5 ). The groups of YDJ soil had the highest number of fungal connections, total number of nodes and modules, the highest connectivity degree, the most vital interaction among species, and the most complex relationship among species ( Figure 6 ).

R² calculated via PERMANOVA indicates the degree of explanation of sample differences by subgroup. A larger R² indicates a higher degree of explanation of differences by subgroup, indicating a more significant difference in subgroups. The R² values of 0.197 (P = 0.001) in PERMANOVA for bacteria and 0.126 (P = 0.001) in PERMANOVA for fungi indicate a high and significant degree of difference among the groups ( Figure 7 ).

In the present study, two soil physiochemical properties (EC and pH) of 18 samples were investigated to analyze the relationships among the groups ( Figure 8 ). Our results found that for bacteria, pH was significantly and negatively correlated with the abundance of Acidobacteriota and Proteobacteria, whereas EC was significantly and positively correlated with the abundance of Firmicutes and negatively correlated with the abundance of Proteobacteria. For fungi, pH and EC were significantly and negatively correlated with the abundance of Chytridiomycota ( Figure 9 ).