The wheat variety used in this study was “Chuanmai 104.” The field experiment was conducted from October 29, 2023, to May 6, 2024, at a farmland site along the Urban Greenway on Datian Road, Pidu District, Chengdu, Sichuan, China (30°77′56″N, 103°97′31″E) (Figure 1a). The site was previously designated as green space, characterized by low soil moisture, a dry texture, fine particles, and an abundance of stones and bricks (Figure 1b). According to the USDA classification system, the soil type was identified as silty, with 12.14% clay, 85.12% silt, and 2.74% sand.

The baseline soil nutrient content included: organic matter (12.6 g/kg), total nitrogen (210 mg/kg), nitrate nitrogen (24.84 mg/kg), ammonium nitrogen (1.78 mg/kg), available nitrogen (21 mg/kg), available phosphorus (25.31 mg/kg), and exchangeable potassium (59.95 mg/kg). The soil pH was 6.83. The γ-PGA used in this study was purified from Bacillus subtilis SCP010-1 fermentation broth, with a molecular weight of approximately 1.1 million Da (Hong et al., 2024). Two experimental groups (B1 and B2) were established, with each treatment group consisting of four replicates. Each replicate corresponds to one plot covering an area of 83.33 m², resulting in a total of eight plots, which were randomly arranged. Control group (B1): Fertilizer only. Experimental group (B2): Fertilizer combined with 0.4% γ-PGA powder (mixed with the fertilizer).

The field was manually cleared of stones, bricks, and plastic waste. A rotary tiller was used to incorporate either the base fertilizer or γ-PGA powder to a depth of 40 cm, ensure even distribution of fertilizer in the root zone. Wheat seeds were sown with a row spacing of 20 cm at a density of 3.5 million plants per hectare. Nitrogen fertilizer was applied in three stages using urea (46% nitrogen): 60% as base fertilizer, 20% at the tillering stage, and 20% at the jointing stage. Phosphorus and potassium were applied using calcium superphosphate (52% P₂O₅) and potassium sulfate (50% K₂O) as base fertilizers.

During jointing and flowering stages, accurately weigh γ- PGA powder, completely dissolve it in water, and the γ- PGA solution is evenly sprayed on the soil within 10 cm of the wheat roots in B2 group (300 g/mu each time) with a sprayer. Soil samples were collected at four key growth stages of wheat: tillering (T), jointing (J), flowering (F), and maturity (M).

When more than 50% of the wheat plants reached the tillering, jointing, flowering, and maturity stages, samples were collected following standard protocols (Miller et al., 2003). The wheat development timeline was as follows: sowing: Oct 29, tillering: Nov 25–Dec 28, jointing: Feb 27–Mar 26, flowering: Apr 2–Apr 15, maturity: May 6 (Figure 1c).

Soil samples were collected on Dec 19, Mar 11, Apr 12, and May 6. For each group, three biological replicates were randomly selected for analysis. Intact wheat plants were carefully excavated. After gently shaking off the loosely attached soil from the roots, the tightly bound rhizosphere soil adhering to the root surface was collected using a sterile spoon to prepare a sample for subsequent analysis. Samples were sealed in plastic containers, transported on dry ice to the laboratory, and divided into two parts: one part was stored at −80°C for microbial and soil metabolomics analysis, the other part was air-dried, cleaned of impurities, ground, and sieved through 20-mesh and 100-mesh screens for physicochemical analysis.

The particle size distribution was measured using a Malvern Mastersizer 3000 particle size analyzer (UK), following Liu et al. (2013). Soil infiltration properties were assessed using vertical soil columns (Guo et al., 2021): soil column height: 30 cm; inner diameter: 5 cm, mariotte bottle height: 50 cm; inner diameter: 5 cm, soil height: 24 cm; constant water head: 2 cm.

Soil physicochemical properties were analyzed following the procedures in Soil Agrochemical Analysis (Rukun, 1999). Bulk density was determined using the ring knife method. Soil pH was measured with a pH meter after passing air-dried soil through a 2 mm sieve. Organic matter was analyzed using the high-temperature external heating potassium dichromate method. Total nitrogen was measured via the Kjeldahl method with potassium permanganate and ferrous sulfate modification. Available nitrogen, nitrate nitrogen, and available potassium were determined using colorimetric, alkaline diffusion, and flame photometry methods, respectively. Enzymatic activities were measured using Solarbio reagent kits (China): urease and sucrase activities were determined directly, while alkaline phosphatase was assessed via micro-assay kits.

LC–MS/MS analysis was conducted with five biological replicates per group by Majorbio Biotech Co., Ltd. (Shanghai, China), following Hou et al. (2021) with slight modifications. Rhizosphere metabolites stored at −80°C were ultrasonically extracted in 1 mL of methanol (4:1, v/v) containing internal standards at 5°C for 30 min. After centrifugation (13,000 g, 4°C, 15 min), the supernatant was concentrated under nitrogen gas. The samples were then dissolved in 120 μL of acetonitrile (1:1, v/v) and subjected to LC–MS/MS analysis using a Thermo UHPLC-Q Exactive HF-X system with an ACQUITY BEH C18 column (100 mm × 2.1 mm, 1.7 μm; Waters, United States).

Data processing was performed with Progenesis QI 2.3 software (Nonlinear Dynamics, Waters, USA) for peak detection and alignment. Identified metabolites were annotated using the KEGG database.¹ Supervised clustering and PLS-DA were employed to assess group differences. Differential metabolites (DEMs) were defined by VIP > 1.0 and p < 0.05 (t-test) and visualized using volcano plots and bubble charts for enriched pathways.

Microbial DNA was extracted from homogenized soil samples using the E.Z.N.A.^® Soil DNA Kit (Omega Bio-tek, United States). DNA concentration was measured using QuantiFluor™-ST (Promega, Beijing). The V3-V4 region of the 16S rRNA gene was amplified with primers 338F and 806R. Sequencing libraries were prepared using the TruSeq^TM DNA Sample Prep Kit, and sequencing was performed on the Illumina MiSeq PE300 platform. The original sequencing sequence has been stored in NCBI (BioProject ID: PRJNA1176352).

OTU-based methods were used for bioinformatic analysis, and alpha diversity metrics (coverage, Chao1, and Shannon indices) were calculated. Beta diversity was analyzed using Bray-Curtis distances with PCoA and NNMDS plots. ANOVA followed by post-hoc tests identified significant differences in taxa abundance at the genus level.

The correlations between microbial communities and metabolite expression were analyzed using Pearson correlation, visualized with the corrplot R package. The relationships between bacterial genus-level diversity and DEMs were explored to identify potential interactions.

Data were processed using Excel 2010 and SPSS 17.0. One-way ANOVA followed by least significant difference (LSD) tests was employed to assess statistical significance among groups.

To investigate the effects of γ-PGA on soil infiltration, we conducted a one-dimensional column infiltration experiment, tracking the wetting front, cumulative infiltration, and infiltration rate over time (Figure 3). As shown in Figure 3a, with a wetting front of 24 cm, the infiltration rate in the B1 group was consistently faster than in the B2 group across all four growth stages. Specifically, the B1 group reached the wetting front 60 min earlier during the tillering stage, 90 min earlier during the jointing stage, and 120 min earlier during both the flowering and maturity stages.

As shown in Figure 3b, upon reaching a 24 cm wetting front in the B1 group, cumulative infiltration in the B2 group measured 8.9 cm (tillering), 7.7 cm (jointing), 7.42 cm (flowering), and 7.39 cm (maturity), representing reductions of 11.2, 3.9, 10.5, and 5.6%, respectively. As shown in Figure 3c, the infiltration rate stabilized after 80 min, with the B2 group demonstrating consistently lower rates across all stages: declines of 4.27% (tillering), 7.45% (jointing), 10.33% (flowering), and 5.8% (maturity).

Overall, γ-PGA application slowed the wetting front movement and decreased both the infiltration rate and cumulative infiltration, suggesting that γ-PGA improved soil water retention by reducing permeability. These findings align with Gao et al. (2024), who also observed enhanced water retention in shallow soils treated with γ-PGA.

Soil samples were collected at four stages of wheat growth to compare the control group (B1) with the γ-PGA-treated group (B2), each with three biological replicates. As shown in Table 1, the bulk density of the B2 group was consistently lower than that of B1, with the most significant reduction of 14.17% observed during the maturity stage (MB2). Additionally, γ-PGA application increased soil pH, with the flowering stage (FB2) showing the largest increase of 2.62%. Organic matter content was also higher in the B2 group, with increases of 14.6, 18.85, 13.00, and 12.48% across the four growth stages.

Regarding nitrogen content, γ-PGA notably increased total nitrogen, nitrate nitrogen, and alkali-hydrolyzable nitrogen levels. Enzyme activity also rose substantially, with urease activity increasing by 11.93, 7.82, 16.06, and 13.12%, and invertase activity rising by 8.74, 18.6, 56.56, and 61.6%. Alkaline phosphatase activity increased by 14.4, 67.44, 67.61, and 83.35% at each respective stage. Available potassium content was also significantly higher in the B2 group, with the greatest increase of 19.1% observed during the flowering stage (FB2).

These results indicate that γ-PGA application enhanced soil pH, organic matter, and nitrogen content while reducing bulk density. Fan et al. (2022) emphasized the importance of evaluating multiple indicators, including pH, bulk density, organic matter, and enzyme activity, to assess soil conditioner effectiveness. Our findings confirm that γ-PGA enhances soil fertility by improving mechanical composition and enzyme activity. Yin et al. (2018) also demonstrated that γ-PGA fermentation broth improved soil physicochemical properties in pot experiments, consistent with our observations. Recent studies further suggest that γ-PGA significantly reduces soil bulk density (Garbowski et al., 2023), promoting macroaggregate formation and improving aeration and water retention. These improvements foster a healthier soil environment, support plant growth, and contribute to sustainable development goals.

During the wheat jointing stage, soil samples were collected and analyzed using untargeted metabolomics to assess metabolite level changes between the B1 and B2 groups. A total of 995 metabolites were identified across 10 soil samples, categorized into 14 types (Figure 7a). These metabolites included lipids and lipid-like molecules (285, 30.55%), organic heterocyclic compounds (162, 17.36%), organic acids and derivatives (112, 12%), benzenoids (103, 11.04%), organic oxides (84, 9%), phenylpropanoids and polyketides (70, 7.50%), nucleotides and analogs (30, 3.22%), organic nitrogen compounds (30, 3.22%), unidentified metabolites (24, 2.57%), alkaloids and derivatives (22, 2.36%), hydrocarbons (6, 0.64%), lignans and related compounds (2, 0.21%), organic sulfur compounds (2, 0.21%), and organic phosphates (1, 0.11%).

Based on VIP values above 1.0 and p-values below 0.05, 958 metabolites were detected in the γ-PGA-treated group, with 90 upregulated and 21 downregulated, indicating an overall upward trend (Figure 7b). Visualization of the top 20 differential metabolites showed consistent upward trends, though the degree of change varied (Figure 7c).

Seven significantly enriched metabolic pathways were identified using the KEGG database: carbohydrate digestion and absorption, starch and sucrose metabolism, nucleotide metabolism, pyrimidine metabolism, the phosphotransferase system, galactose metabolism, and the ABC transport system (Figure 7d). Furthermore, 11 upregulated differential metabolites participated in these enriched pathways, including sucrose, deoxycytidine, cytidine, N-acetyl-D-galactosamine, lactose, cellobiose, inositol galactoside, raffinose, deoxyuridine, trehalose, and stachyose. γ-PGA application significantly elevated soil metabolite levels, creating a new metabolic profile (Figure 8). These metabolite changes are likely to impact the composition of rhizosphere microorganisms, as they shape the microbial food web, regulate soil chemical properties, influence microbial gene expression, and serve as essential chemical signals in microbial interactions (Hu et al., 2018).

Our analysis of soil metabolites at the wheat jointing stage showed that lipids, lipid-like molecules, organic heterocyclic compounds, organic acids, and benzenoids accounted for approximately 71% of the metabolites. These metabolites provide essential nutrients for plant and soil microbial growth and are highly enriched in carbohydrate and nucleotide metabolism pathways. Additionally, eight significantly upregulated metabolites were enriched in the ABC transporter metabolic pathway, which is crucial for microbial nutrient uptake and utilization. ABC transporters are membrane-associated, energy-dependent transport proteins that move nutrients across cell membranes via ATP binding and hydrolysis (Greene et al., 2018). In high-carbohydrate environments, nutrients tend to form complexes with hydrophobic hydrocarbons, limiting microbial utilization (Obayori et al., 2015). γ-PGA treatment promotes microbial metabolic activity by enhancing nutrient transport.

Six differential metabolites were significantly enriched in the galactose metabolism pathway. For example, Vitamin C (AsA) not only provides plants with protection against drought, ozone, and UV radiation (Ali et al., 2019) but is also essential for microbial growth. AsA synthesis mainly depends on the L-galactose/D-mannose pathway, which is crucial for the health of both plant and soil microbial communities (Tao et al., 2020). Carbohydrate metabolic pathways, such as sucrose metabolism and the phosphotransferase system (PTS), play a vital role in enhancing soil fertility. PTS regulates complex carbon and nitrogen metabolism processes (Deutscher et al., 2006), thereby increasing soil fertility. Raffinose and stachyose are galactosyl sucrose derivatives belonging to the RFO family of oligosaccharides (Salvi et al., 2021), essential for transporting photosynthetic products in plants, especially within the Verbenaceae, Cucurbitaceae, and Lamiaceae families (Gangola and Ramadoss, 2018).

Trehalose, a disaccharide composed of two glucose molecules linked by an α-1-1 bond, has significant osmoprotective properties, helping maintain membrane lipid stability (Saddhe et al., 2021). Additionally, trehalose preserves protein structure and scavenges ROS (Zulfiqar et al., 2019), enhancing plant tolerance to abiotic stresses. The levels of sucrose, lactose, and trehalose were significantly elevated in γ-PGA-treated soil, potentially contributing to improved plant stress resistance.

In conclusion, γ-PGA application significantly altered the soil’s metabolic landscape by modulating the types and concentrations of various chemical metabolites. This not only improved the soil’s physical and chemical properties, creating a more favorable environment for plant growth, but also provided essential nutrients for microbial activity. By promoting microbial growth and reproduction, γ-PGA treatment enhances the diversity and functionality of soil microbial communities, further increasing soil fertility and plant resilience to stress.