A hybrid No. 22 wheat variety was selected, and it was bred by the Crop Research Institute of Shandong Academy of Agricultural Sciences. Normal field soil (salt content of 0.06%) from Taian, China, and raw saline-alkali soil (salt content of approximately 0.8%, organic matter content of approximately 0.65%, and pH 7.9) from Dongying, China, were selected for this study.

Ten grams of raw saline-alkali soil was added to 100 mL of deionized water and shaken at 37°C and 180 rpm for 30 min; then, 100 µL of the diluted liquid was spread onto LB culture medium containing 8% NaCl for culture at 37°C for 48–72 h. Single colonies with different sizes and shapes were isolated and purified via three-zone streaking. The purified strains were cultured in an LB liquid medium, and then, 50 µL of the corresponding liquid obtained from the initial screen was spread onto LB solid media with different concentrations of NaCl, ranging from 0% to 17%, or with different acidic and alkaline pH values for incubation and observations of the growth capacity. Gram staining of bacterial cells was conducted to observe the cell morphology, and bacterial microbiochemical identification tubes (Haibo Biotechnology Co., Ltd., China) were used to determine some physiological and biochemical characteristics of the strains (Zhao et al., 2022). The 16S rDNA (amplified using 27F/1492R primers) and gyrB gene (amplified using the primers gyrB-F “GAT​CAA​CTA​ACA​GCA​AAG​GCC​TTC​ACC​A” and gyrB-R “AGA​CCT​TTA​TAA​CGC​TGG​AGG​CCG​GGC​T”) of bacteria were obtained for sequencing (Shengong Biotechnology Co., Ltd., China) (Ye et al., 2022). Combining the morphological characteristics of cells, physiological and biochemical indicators, gyrB gene sequence comparison, and phylogenetic tree analysis of 16S rDNA sequences (Constructed using MEGA 11.0), the bacterial species were identified (Zhao et al., 2022; Rahma et al., 2020).

The normal field soil was first sieved through an 18-mesh sieve and then mixed with vermiculite at a volume ratio of 2:1, and the resulting soil was used for the pot experiment comprising simulated saline-alkali soil. For the pot experiment using raw saline-alkali soil, the raw saline-alkali soil from Dongying was mixed with vermiculite at a 1:1 volume ratio and then mixed with 2% chicken organic manure. For the wheat pot experiment, 0.5 kg of the soil mixture was added to each pot.

Under dark cultivation conditions for 8 h, wheat seeds were soaked in sterile water until budding, and 3–5 wheat seeds with the same budding characteristics were selected and planted in a pot. After covering them with a thin coat of soil, as in the pot experiment using simulated saline-alkali soil, 100 mL of salt water with a mixture of 0.655 mM NaCl, 5.85 mM Na₂SO₄, 5.85 mM NaHCO₃, and 0.65 mM Na₂CO₃ (pH = 8.3) was added, and the same amount of water was added in the control group. The wheat seedlings were cultivated at 20°C under 40,000 lx light and 0 lx darkness conditions. After 48 h of cultivation, wheat seedlings with small growth differences were retained in the pots. Strains were activated twice using an LB liquid medium. Each pot was watered with 100 mL of a bacterial suspension with approximately 10⁶ CFU mL⁻¹ (single strain or mixed strains), and the same amount of water was added to the control group. During the wheat seedling growth process, 30 mL of water was used to water each pot every 2 days, and the physiological plant height was measured every 7 days.

During the pot experiment using simulated saline-alkali soil, the data of the conventional agronomic traits of wheat seedlings were recorded after growth for 14 days, as reported (Wang et al., 2024). During the pot experiment using raw saline-alkali soil, the data of conventional agronomic traits, dry and fresh weights, leaf and root enzyme activities, root system, soil enzymatic activities, and malondialdehyde (MDA) concentrations in wheat seedlings were recorded after growth for 21 days as reported (Wang et al., 2024). The nitroblue tetrazolium photochemical reduction method was used to determine superoxide dismutase (SOD) enzyme activity (Begum et al., 2022), the pyrogallol autoxidation method was used to determine peroxidase activity (Wang J. et al., 2022), catalase (CAT) activity was measured based on the reaction comprising the catalase-mediated degradation of hydrogen peroxide (Li et al., 2019), and the MDA concentration was determined based on its reaction with thiobarbituric acid (Pervaiz et al., 2021). Different enzyme activity assay kits (Suzhou Grice Biotechnology Co., Ltd., China) were used to measure the activities of neutral phosphatase, alkaline phosphatase, acid phosphatase, catalase, urease, and sucrase in wheat rhizosphere soil or non-rhizosphere soil.

Furthermore, the effects of the strains on the germination of the wheat seeds were also tested under saline-alkali conditions with a mixture of 0.655 mM NaCl, 5.85 mM Na₂SO₄, 5.85 mM NaHCO₃, and 0.65 mM Na₂CO₃ (pH = 8.3). The procedures were performed as reported (Wang K. et al., 2022).

The raw saline-alkali soil was mixed with vermiculite at a 1:1 volume ratio and then sterilized. Then, 0.5 kg of the soil mixture was added to each pot. The wheat seedling sowing and cultivation methods were the same as those mentioned previously herein. Strains with good growth-promoting effects were selected for the wheat seedling self-selection experiment. Each strain was cultivated twice in an LB liquid medium. Then, equal amounts of bacteria were mixed at a 1:1 ratio. The bacterial suspension was diluted 100 times and then applied to the potted wheat seedlings. After 21 days, root zone soil samples were collected for metagenomic sequencing to determine the colonization ratio of different bacteria.

The complete DNA of the root zone soil was extracted using the OMEGA Mag-Bind soil DNA Kit (Omega Bio-Tek, United States) and then sequenced at Shanghai Personalbio Technology Co., Ltd. (Personalbio, China) for the metagenomic analysis. Three biological samples (SSP_1, SSP_2, and SSP_3) were used. The different insert fragment libraries were constructed using the whole-genome shotgun strategy. Then, next-generation sequencing technology, based on the Illumina NovaSeq sequencing platform, and third-generation single-molecule sequencing technology, based on the Oxford Nanopore ONT sequencing platform, were used to sequence these libraries (Xiao et al., 2021). The sequencing adapters were removed from reads using Cutadapt (v1.2.1) (Martin, 2011). The original metagenomic sequencing data can contain some low-quality sequences, the presence of which could interfere with subsequent analyses. To ensure the reliability of the subsequent analysis, fastp (v0.23.2) software was used to screen and filter the raw data obtained from sequencing (Chen et al., 2018). Once quality-filtered reads were obtained, taxonomical classifications of metagenomics sequencing reads from each sample were performed using the method Kaiju (v1.9.0) against an NCBI-nt-derived database (Menzel et al., 2016). A tool specifically designed to predict genes in metagenomic sequences was used for gene prediction (Zhang X. et al., 2022). Megahit (v1.1.2) was used for assembly of reads in each sample using the meta-large presetted parameters. In MEGAN (Metagenome Analyzer, http://ab.inf.uni-tuebingen.de/software/megan5/) software, the lowest common ancestor algorithm was used for the species classification annotation information of the target sequence (Huson et al., 2016). R software was used to generate bar charts of the compositions of dominant species at various taxonomic levels in each sample (Deng et al., 2020).

In total, 151 strains were first isolated from the raw saline-alkali soil, and then, five highly saline-alkali-tolerant strains, namely, S-3, BJYX, G51-1, YL-10, and G63-1, were finally screened. The NaCl tolerance limits of strains S-3, BJYX, G51-1, YL-10, and G63-1 on plates were 16, 11, 12, 12, and 12%, respectively, and their alkali tolerance limits were pH 14, 13, 10, 8, and 11, respectively. Strains S-3, BJYX, G51-1, YL-10, and G63-1 were all found to be Gram-positive bacteria. The colonies of strain S-3 were milky yellow, opaque, and circular in shape; the colonies of strain BJYX were milky white, semi-transparent, and circular in shape; the colonies of strain G51-1 were beige, opaque, and irregular in shape; the colonies of strain YL-10 were gray-white, opaque, and circular in shape; and the colonies of strain G63-1 were milky white, semi-transparent, and irregular in shape, and they were all able to produce spores (Figure 1). The 16S rDNA gene sequences of the strains were analyzed through the generation of phylogenetic trees to search for closely related species (Figure 1), and the results were further verified based on the gyrB gene sequence of each strain. Finally, it was determined that strain S-3 is Rossellomorea aquimaris, strain BJYX is Bacillus subtilis, strain G51-1 is Bacillus velezensis, strain YL-10 is Bacillus vallismortis, and strain G63-1 is Bacillus zhangzhouensis.

All newly screened saline-alkali-tolerant PGPR and the previously preserved strains Kocuria dechangensis 5–33 (Wang et al., 2015) and Bacillus atrophaeus CNY01 (Sun et al., 2024) were tested for their capacity to affect wheat growth using pot experiments under simulated saline-alkali conditions of a moderate environment. After 7 days (Figure 2A) and 14 days (Figure 2B) of wheat seedling growth, the addition of strains 5-33, BJYX, G51-1, S-3, CNY01, YL-10, and G63-1 had obvious growth-promoting effects on wheat seedlings, as follows: for 7 days growth, physiological plant heights were increased by 44.95%, 84.44% (p < 0.01), 74.89% (p < 0.01), 88.81% (p < 0.01), 77.29% (p < 0.01), 71.49% (p < 0.01), and 48.94% (p < 0.01), respectively (Figure 2A); for 14 days growth, the physiological plant heights were increased by 21.55%, 35.36% (p < 0.05), 34.07% (p < 0.05), 32.66% (p < 0.05), 37.54% (p < 0.05), 23.18%, and 27.89% (p < 0.05), respectively, compared with those in the control group (Figure 2B). All strains promoted the growth of wheat seedlings in the simulated saline-alkali soil, and almost all of them resulted in a wheat growth trend similar to that in normal soil.

Equal amounts of strains S-3, BJYX, G51-1, YL-10, G63-1, 5-33, and CNY01 were added to the rhizosphere microenvironment of wheat seedlings. After cultivation for 21 days, the root-zone soil of wheat seedlings was collected for metagenomic sequencing to determine the colonization capacity and ratio of the different bacteria added. During the metagenomic sequencing process, the numbers of raw reads generated from three samples were 101,824,340, 102,932,646, and 93,404,390, respectively. Meanwhile, the numbers of quality-filtered reads generated from three samples were 99,170,304, 99,887,882, and 91,054,674, respectively. The minimum read length before sticking paired-end reads was 150 bp. After the concatenation was complete, contigs with a length of not less than 300 bp were retained. For the metagenomic sequencing data, the proportion of valid sequences was greater than 96%, the proportion of valid sequence base numbers was also greater than 96%, and the slopes of the rarefaction curves became smaller and tended to flatten at last (Supplementary Figure S1), indicating that samples could be further analyzed.

In terms of the community composition of the rhizosphere soil of wheat seedlings, the top 10 phyla of bacteria with the highest proportions were determined, and the levels ranged from high to low and included Proteobacteria, Bacteroidetes, Verrucomicrobia, Actinobacteria, Myxococcota, Planctomycetota, Chloroflexota, Myxococcota A, Gemmatimonadetes, and Fibrobacterota (Figure 3A). Among them, Proteobacteria was the dominant phylum in the rhizosphere soil of wheat in all treated samples, with an average relative abundance of 66.5%. The relative abundances of the other phyla Bacteroidetes, Verrucomicrobia, Actinobacteria, Myxococcota, Planctomycetota, Chloroflexota, Myxococcota A, Gemmatimonadetes, and Fibrobacterota were 14.70, 5.75, 5.20, 3.14, 0.97, 0.97, 0.64, 0.58, and 0.31%, respectively.

In terms of the community composition of the rhizosphere soil of wheat seedlings, the top 10 genera with the highest proportions were determined, and the levels ranged from high to low, including Halomonas, Saccharospirillum, Marinobacter, Thalassospira, Marinimicrobium, Pelagicoccus, Alcanivorax, Devosia, Idiomarina, and Muricauda (Figure 3B). Among them, Halomonas was the dominant genus in the wheat rhizosphere soil of all treatment samples, with an average relative abundance of 9.36%, whereas the average proportions of the other genera Saccharospirillum, Marinobacter, Thalassospira, Marinimicrobium, Pelagicoccus, Alcanivorax, Devosia, Idiomarina, and Muricauda were 5.05, 4.62, 3.79, 3.36, 3.26, 2.40, 2.38, 2.29, and 1.75%, respectively.

In terms of the community composition of the rhizosphere soil of wheat, the top 10 species with the highest proportions were determined, and the levels ranged from high to low, including M. agarilyticum, H. cupida, A. sediminum, S. salsuginis, H. ventosae, Confluentibacter sp018883545, Thalassospira xianhensis, Cellvibrio sp002162135, Thalassospira lucentensis A, and M. koreense (Figure 3C). Among them, Marinimicrobium agarilyticum was the dominant species in the rhizosphere soil of wheat seedlings in all treatment samples, with an average relative abundance of 3.28% in the three parallel samples, whereas the average proportions of the other species Halomonas cupida, Aquamicrobium sediminum, Saccharospirillum salsuginis, Halomonas ventosae, Confluentibacter sp018883545, T. xianhensis, Cellvibrio sp002162135, T. lucentensis A, and Marinimicrobium koreense were 2.79, 1.98, 1.92, 1.62, 1.60, 1.58, 1.49, 1.22, and 1.20%, respectively.

Bacterial numbers for the added species were also determined (Figure 3D). The species K. dechangensis 5–33, R. aquimaris S-3, B. subtilis BJYX, and B. velezensis G51-1 ultimately colonized the rhizosphere soil of wheat seedlings at a ratio of 275:63:5:1. This suggests that these four species have a high affinity for the wheat rhizosphere microenvironment.

To verify that the self-selected microflora in the wheat rhizosphere microenvironment has better saline-alkali resistance and growth-promoting effects on wheat seedlings, pot experiments using raw saline-alkali soil were conducted, with a control group and six treatment groups. The six treatment groups comprised four single-strain treatments (strains 5-33, S-3, BJYX, and G51-1), a mixed treatment with an equal proportion of the four selected strains (MBE group), and a mixed treatment with a self-selected proportion (selected by wheat seedling) of the four selected strains (MB group).

The physiological plant heights of wheat seedlings cultivated for 7 and 21 days after the strains were added are shown in Figure 4A. For all six treatment groups, the physiological plant heights were all greater than those of the control group. Compared with those of the control group, the physiological plant heights of the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were increased by 21.01, 24.23, 19.89, 23.70, 17.28, and 19.45%, respectively, at 7 days (Figure 4A). After the wheat seedlings were cultivated for 21 days, compared to those in the control group, the physiological heights in the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were increased by 8.86, 13.72, 14.49, 11.95, 27.39 (p < 0.01), and 15.92%, respectively (Figure 4B); the aboveground fresh weights of the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were increased by 125.06% (p < 0.01), 119.12% (p < 0.01), 136.76% (p < 0.01), 110.82% (p < 0.05), 147.33% (p < 0.01), and 109.91% (p < 0.05), respectively (Figure 4C); the aboveground dry weights of the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were increased by 162.09% (p < 0.01), 155.52% (p < 0.01), 173.02% (p < 0.01), 123.01% (p < 0.05), 194.86% (p < 0.01), and 146.61%(p < 0.05), respectively (Figure 4D); the underground fresh weights of the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were increased by 203.12% (p < 0.01), 198.31% (p < 0.01), 213.19% (p < 0.01), 159.27% (p < 0.05), 282.98% (p < 0.01), and 112.92%, respectively (Figure 4E); and the underground dry weights of the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were increased by 144.77%, 140.69%, 131.39%, 91.28%, 218.60% (p < 0.01), and 161.63% (p < 0.05), respectively (Figure 4F).

An observation of wheat seedlings with different treatments (Figure 4G) and root scanning (Figure 4H) revealed that wheat seedlings treated with the inoculants of the strains showed better growth compared to the control, and various root indicators were increased by different degrees compared to those in the control group (Table 1). Especially in the MB treatment group, compared to those in the control group, all agronomic trait data were optimal. Therefore, the in situ self-selected strain treatment had a better promoting effect on wheat in response to saline-alkali stress than the single-strain treatment and equal-proportion treatment.

The enzyme activities of the leaves and roots of wheat seedlings were further determined (Figure 5). In the wheat leaves, compared with those in the control group, the activities of SOD in the BJYX, G51–1, MB, and MBE treatment groups were increased by 16.99, 27.85, 68.38, and 23.34%, respectively (Figure 5A). Peroxidase (POD) activities in the leaves of the 5-33, BJYX, G51-1, S-3, MB, and MBE groups were increased by 7.87, 11.96, 19.73, 9.61, 55.42, and 66.14%, respectively (Figure 5B). The concentrations of MDA in the leaves of the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were reduced by 12.48, 3.01, 5.83, 6.43, 1.74, and 5.85%, respectively (Figure 5C). In the root systems of wheat, compared with those in the control group, the activities of SOD in the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were decreased by 55.02% (p < 0.05), 37.85%, 53.03% (p < 0.05), 36.31%, 43.03%, and 49.95% (p < 0.05), respectively (Figure 5D), whereas POD activities were increased by 38.95%, 86.05%, 53.30%, 119.25% (p < 0.05), 95.79% (p < 0.05), and 99.32% (p < 0.05), respectively (Figure 5E). The concentrations of MDA in the roots of the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were decreased by 15.60, 8.51, 1.42, 12.05, 9.22, and 4.96%, respectively (Figure 5F). Moreover, root peroxidase activities in the BJYX, G51-1, S-3, MB, and MBE treatment groups were increased by 107.14, 29.59, 46.05, 111.51, 120.84% (p < 0.05), respectively (Figure 5G).

The activities of acid phosphatase, neutral phosphatase, and alkaline phosphatase in the rhizosphere and non-rhizosphere soils of different groups were also determined (Figure 6). The rhizosphere soils generally had higher levels of all three phosphatase enzymes than the non-rhizosphere soils. Compared with those in the control group, the activities of acid phosphatase in the rhizosphere soil of 5–33-, BJYX-, G51-1-, and S-3-treated groups were decreased by 59.43, 56.37, 12.90, and 22.10%, respectively (Figure 6A). Meanwhile, the rhizospheric acid phosphatase activities in the MB and MBE treatment groups were increased by 61.39% and 25.96%, respectively, compared with those in the control group (Figure 6A). The neutral phosphatase activities in the rhizosphere soils of the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were increased by 233.42%, 176.13%, 138.19%, 79.57%, 393.90% (p < 0.1), and 142.97%, respectively (Figure 6B). However, the root alkaline phosphatase activities in the BJYX, G51-1, S-3, and MBE groups were decreased by 14.12, 34.15, 49.94, and 8.55%, respectively (Figure 6C). Meanwhile, those in the MB and 5-33 treatment groups were increased by 13.24% and 11.15%, respectively.

In the saline-alkali stress environment, the effects of microbial agents on the activities of soil catalase, sucrase, and urease were simultaneously determined (Figure 6). The rhizosphere soil had higher levels of catalase, sucrase, and urease activities than the non-rhizosphere soil. The catalase activities in the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were all decreased, by 11.06%, 28.27% (p < 0.01), 33.71% (p < 0.01), 31.82% (p < 0.01), 17.15% (p < 0.05), and 23.89% (p < 0.01), respectively (Figure 6D). Meanwhile, cellulase activities in the 5-33, BJYX, G51-1, S-3, MB, and MBE treatment groups were decreased by 26.83% (p < 0.01), 23.88% (p < 0.01), 24.30% (p < 0.01), 27.36% (p < 0.01), 27.56% (p < 0.01), and 26.85% (p < 0.01), respectively (Figure 6E). Further, the 5-33, G51-1, S-3, and MB treatment groups exhibited increased urease activities by 16.91, 4.54, 7.05, and 12.66%, respectively. Compared to those in the control group, urease activities in the BJYX and MBE treatment groups were decreased by 52.49% and 3.92%, respectively (Figure 6F).

Besides, the application of microbial agents could improve the germination indicators of wheat seeds under saline-alkali stress to varying degrees (Supplementary Table S1). The germination potentials (GP) in the treatment groups of 5–33, G51-1, S-3, MB, and MBE were increased by 6.66, 4.44, 6.66, 10.00, and 10.00%, respectively. The increase of the germination index indicated that the application of microbial agents significantly alleviated the inhibitory effect of saline-alkali stress on the germination of wheat seeds.