The study site was located in Zhongtiao Mountain (Latitude 35°24′00″ to 35°40′00″ N, Longitude 111°56′12″ to 112°14′00″ E). Zhongtiao Mountain spans the cities of Linfen, Jincheng and Yuncheng, belongs to the tributaries of Taihang Mountain, and is located in the warm temperate semi-humid continental monsoon climate zone. The average annual temperature is 10.3°C, the lowest temperature is -8.2°C, the highest temperature is 28.8°C (Ning et al., 2023). The annual sunshine are 2,679.8 hours and the frost-free period is 197 days. The average annual precipitation is 600–720 mm, mainly in June to September. Soil types range from meadow soil to brown forest soil, drenched brown soil, and brown soil. The dominant tree species including P. tabuliformis, Quercus mongolica, Betula platyphylla, Populus davidiana, Carpinus turczaninovii and Toxicodendron vernicifluum.

In July 2024, based on the proportion of P. tabuliformis (PT) in the forest, four sample plots with different percentages of P. tabuliformis planting were established in the Xiachuan Management Area of Zhongcun Forest Field under the similar standing conditions (10% P. tabuliformis planting, PT10%; 20% P. tabuliformis planting, PT20%; 60% P. tabuliformis planting, PT60%; 100% P. tabuliformis planting, PT100%), with four replicates for each sample plot. Each plot measured 20 m × 20 m. The location of the survey sample plots was recorded ( Figure 1 ; Supplementary Figure S1 ). Tree species, diameter at breast height (DBH) and height of all trees with DBH ≥ 5 cm were recorded. Five 5 m × 5 m subplots were established within each 20 m × 20 m plot following the five-point sampling method to assess understory shrubs. A total of 80 subplots (5 × 16) were surveyed. At the center of each 5 m × 5 m subplot, a 1 m × 1 m subplot was used to investigate understory herbs. Shrub and herb species, ground diameter and height in the plots were recorded, and the top three dominant species in each plot were filtered by calculating their importance values ( Table 1 ; Supplementary Table S1 ).

The importance values (IV) of shrubs and herbs were calculated using the following formulas (Zhang et al., 2024b):

An additional 1 m × 1 m plot was set up in the center of each 5 m × 5 m subplots set up within each 20 m × 20 m plot, and soil samples were collected from the 1 m × 1 m plots. Three bags of 0–20 cm topsoil were collected from each sampling site, with 15 samples from each plot. In the laboratory, the 15 samples from each plot were combined, large stones and rhizomes were removed, and the mixture was sieved through a 0.9 mm sieve. The sieved samples were stored at -80°C for further analysis.

Statistical analyses were conducted in R (v4.4.1). Data were assessed for normality and homoscedasticity using normal distribution and mean square tests. One-way ANOVA followed by Least Significant Difference (LSD) multiple comparisons was used to analyze significant differences in soil properties between forests planted with different percentages of P. tabuliformis.

β-diversity of soil microbiological was analyzed using non-metric multidimensional scaling (NMDS) based on Bray-Curtis dissimilarity (Mei et al., 2025). PERMANOVA was applied to assess the influence of the proportion of planting of P. tabuliformis on the composition of soil microbiological, while analysis of similarity (ANOSIM) was used to evaluate differences between forests with different proportions of planting of P. tabuliformis (n = 999 permutations). The first axis of NMDS explains the greatest variation in microbiological community composition. Spearman’s correlation analysis was conducted to investigate symbiosis patterns among microbiological community composition in forests with four proportions of P. tabuliformis plantations, focusing on microorganisms with a relative abundance > 0.5% (Barberan et al., 2012). The Spearman’s correlation coefficient (r) > 0.6 with corresponding p-values< 0.01 were considered statistically robust and were used to generate the network. Results of the symbiotic network analysis were visualized using Gephi 0.10.1 (Bastian et al., 2009), which also identified dominant phyla. The colors of the nodes represent the dominant phyla. Network edges represented nodes that were pairwise related, with pink lines representing positive connections and green lines representing negative connections.

Spearman correlation analysis was used to study the relationship between soil properties and microbiological relative abundance of dominant phyla. Relationships between soil properties (pH, AK, TN, MBN, MBC and SOC) and the first NMDS axis (NMDS1) of soil microbiological diversity were explored using regression analysis (Liu et al., 2022). In order to test whether there are differences in the response of soil microorganisms to soil properties, the drivers of soil bacterial and fungal diversity were analyzed using the GLMMS. We fitted a model with soil microbiological diversity as the response variable. Sample plots with the same percentage of P. tabuliformis planting were considered as random effects. To ensure a uniform dimension, we standardized and centered the independent variables. To avoid collinearity, we grouped the independent variables (fixed effects) according to a hierarchical structure through hierarchical partitioning and further decomposed the degree of contribution of each independent variable to the dependent variable. Among the independent variables were divided into four groups, namely soil C (MBC and SOC), soil N (MBN and TN), AK and pH (Gao et al., 2024).

Partial Least Squares Passage Modeling (PLS-PM) (Tenenhaus et al., 2005) was conducted to explore the relationships among soil microbiological diversity (scores of the first NMDS axes for soil bacteria and fungi) (Cheng et al., 2023), percentage of P. tabuliformis planted in forests (percentage of PT), soil C (SOC, MBC), soil N (MBN, TN), AK and pH. PLS-PM is suitable for small sample sizes or not normally distributed data and contains both external and internal models. PLS-PM visualizes the results of the path analysis in the form of path coefficients, which are used to test the direction and strength of the causal relationships between latent variables. The fit of PLS-PM is represented by the coefficient of determination (R²) and the goodness of fit (GoF). The fit is weak when R² is above 0.19, moderate when R² is above 0.33, and strong when R² is above 0.67. The goodness of fit is weak when GoF is above 0.1, moderate when GoF is above 0.25, and strong when GoF is above 0.36. This approach analyzed potential direct and indirect effects among these factors and the soil microbiological diversity.

With the exception of MBC and SOC, all of the soil properties showed a decrease with increasing percentage of plantings of P. tabuliformis. Soil pH (p = 0.0024) in PT10% was significantly higher than in the other forests and was closest to neutral. Soil AK (p = 0.0013) was significantly higher in PT10% than in PT60% and PT100%, there were no significant differences between PT20% and PT60%, and soil AK content in PT20% was significantly higher than in PT100%. Soil TN (p = 0.0187) was significantly higher in PT10% than in PT100%. Soil MBN (p = 0.0027) was significantly higher in PT10% than in PT60% and PT100%. Soil MBC (p = 0.0008) and SOC (p = 0.0003) were significantly lower in PT100% than in the other three forests ( Table 2 ).

Unique bacterial and fungal species were more prevalent than shared species across four different forests. Unique bacteria accounted for 91%, and unique fungi accounted for 84% of total taxa ( Figures 2A, B ). Proteobacteria, Acidobacteriota and Actinobacteriota were the dominant bacterial taxa, representing over 70% of sequences ( Figure 2C ). Dominant fungal taxa included Ascomycota, Basidiomycota, and Mortierellomycota, comprising more than 85% of sequences ( Figure 2D ). The relative abundance of Ascomycota in the soil decreased gradually with increasing percentage of P. tabuliformis planted. Soil microbiological β-diversity, assessed at the OTU level via NMDS based on Bray-Curtis distances, revealed significant compositional differences in soil bacteria (Stress = 0.070) ( Figure 2E ) and fungi (Stress= 0.128) ( Figure 2F ) among four different forests.

Co-occurrence network topologies revealed that bacterial and fungal interactions varied by four forests ( Figure 3 ). Overall, the complexity of the bacterial co-occurrence network was greater than that of the fungal. For bacteria, PT60% has the most complex network, PT20% the next most complex, and PT60% has the most nodes and edges. For fungi, PT20% has the most complex network, followed by PT10%, and PT20% has the most nodes and edges.

Spearman’s correlation analysis showed that the relationship between soil properties and the relative abundance of soil microbes. For bacteria, the relative abundance of Actinobacteriota showed a significant positive correlation (p< 0.05) with AK, MBC and SOC; the relative abundance of Bacteroidota showed a significant positive correlation (p< 0.05) with pH and MBN; the relative abundance of Verrucomicrobiota showed a significant negative correlation (p< 0.05) with pH, AK, MBN, MBC and SOC; the relative abundance of Gemmatimonadota showed a significant negative correlation (p< 0.05) with TN and MBN ( Figure 4A ). For fungi, the relative abundance of Ascomycota showed a significant positive correlation (p< 0.05) with pH, AK, TN, and MBC; the relative abundance of Mortierellomycota showed a significant negative correlation (p< 0.05) with pH, TN, MBC and SOC; the relative abundance of Chytridiomycota showed a significant positive correlation (p< 0.05) with pH, AK, TN, MBN and MBC; the relative abundance of Olpidiomycota showed a significant positive correlation (p< 0.05) with SOC; the relative abundance of Basidiobolomycota showed a significant positive correlation (p< 0.05) with TN ( Figure 4B ).

Soil microbiological β-diversity (NMDS1) was affected by pH ( Figures 5A, G ), AK ( Figures 5B, H ), TN ( Figures 5C, I ), MBN ( Figures 5D, J ), MBC ( Figures 5E, K ) and SOC ( Figures 5F, L ). Between different forests, soil bacterial β-diversity was most affected by pH (R² = 0.8566, p< 0.001), in contrast to soil fungal β-diversity (NMDS1), which was least affected by pH (R² = 0.4667, p< 0.01). Soil bacterial β-diversity (R² = 0.6045, p< 0.001) was more affected by MBN than soil fungal β-diversity (R² = 0.4728, p< 0.01), except that soil bacterial β-diversity was less affected by AK, TN, MBC and SOC than soil fungal β-diversity.

Detailed assessment of the separate GLMM of microorganisms showed that the combination of soil C, soil N, AK and pH explained the variation in bacterial and fungal β-diversity well (Marginal R² > 0.700 and Conditional R² > 0.900). pH was the dominant driver of the variation in bacterial β-diversity and accounted for 44.29% of the fixed effect of bacterial β-diversity ( Figure 6A ). Soil C was the dominant driver of fungal β-diversity change, accounting for 36.09% of the fungal β-diversity fixed effect ( Figure 6B ).

PLS-PM revealed complex interactions between percentage of PT, soil C, soil N, AK, pH and microbiological diversity. For bacteria (GoF = 0.81) ( Figures 7A, B ), percentage of PT had a significant negative effect on soil C (SOC, MBC) (r = -0.82, p< 0.001) and AK (r = -0.55, p< 0.05) but a positive effect on soil bacterial diversity (r = 0.41, p< 0.05). AK positively influenced bacterial diversity (r = 0.36, p< 0.05). pH negatively influenced bacterial diversity (r = -0.77, p< 0.001). For fungal (GoF = 0.79) ( Figures 7C, D ), percentage of PT had a significant negative effect on soil C (SOC, MBC) (r = -0.82, p< 0.001) but a positive effect on soil fungal diversity (r = 0.81, p< 0.01). Soil N positively influenced bacterial diversity (r = 0.79, p< 0.001).

Soil properties are sensitive to changes in the proportion of forests planted with P. tabuliformis. Among the four forests, soil contents of TN, MBN, AK and SOC were highest in PT10% and lowest in PT100% ( Table 2 ). PLS-PM modelling showed that the proportion of forests planted with P. tabuliformis had effects on soil C and AK. SOC content was significantly higher in forests with less than 60% P. tabuliformis planting than in PT100%, suggesting that low numbers of P. tabuliformis plantings can significantly enhance the carbon sequestration capacity of soil. This may be due to the fact that, as the proportion of P. tabuliformis planted increases, the proportion of other tree species planted in the forest decreases, the composition of tree species in the forest gradually becomes homogeneous, the understory vegetation is sparse, and the root system and apoplastic material are reduced accordingly, so that the content of SOC that is transformed through decomposition is reduced (Lange et al., 2015; Zarafshar et al., 2020). The accumulation of soil C (MBC, SOC) in the soil provided a more diverse nutrient pool for the growth and development of microorganisms (Hou et al., 2021; Li et al., 2025), thereby increasing the abundance of dominant microbiological communities in the soil (Ji et al., 2014; Chen et al., 2024). The significant positive correlation between the soil C (MBC and SOC) contents and the relative abundance of Actinobacteriota in soil was attributed to the important contribution of Actinobacteriota in increasing soil nutrients and degrading difficult-to-degrade organic matter (Li et al., 2021), which can be involved in carbon sequestration and organic matter cycling (Chen et al., 2024). The results of the study showed that soil AK content decreased with an increase in the proportion of forests planted with P. tabuliformis. Coniferous leaves are rich in lignin, and decay of apoplastic material in the forest can degrade lignin in apoplastic coniferous leaves in the soil, thereby releasing nutrients and organic matter (Chen et al., 2017). PT100% is very homogeneous, its apoplastic composition is also relatively simple, and a large amount of lignin in the soil is not degraded, and the soil organic matter content is low. Basidiomycota mainly decomposes lignocellulose (Lundell et al., 2010). Basidiomycota did not show a significant correlation with soil properties, even though it is also a common dominant phylum in all four forest types, a phenomenon that suggests that Basidiomycota is well adapted to different forest community compositions (Wu et al., 2021). Ascomycota is significantly and positively correlated with pH, AK, TN, and MBC. The relative abundance of Ascomycota was close to 50% of the total relative abundance of all fungi in the soil in each of the four forests with different percentages of P. tabuliformis plantings. Ascomycota is key decomposers in soil ecosystems (Ma et al., 2013). The results of the study showed that soil TN content and the relative abundance of Ascomycetes decreased gradually with the percentage of P. tabuliformis plantings. This trend that indicates the positive impact of Ascomycota on host plants under conditions of high nutrient availability as reported (Leff et al., 2015). At the same time, Ascomycota is well adapted to poor nutrient environments (Sterkenburg et al., 2015), it is effective in degrading organic residues in soil (Hartshornea et al., 2009), and breaks down waste into nutrients that can be taken up by microorganisms (Purahong et al., 2016). Although Ascomycota is better suited to survive in near-neutral soils, the growth and development of Ascomycota was relatively inhibited in near-neutral soils due to the more moderate soil pH and the successive colonization of fungi other than Ascomycota.

NMDS analyses revealed significant differences in the β-diversity of soil bacteria and fungi among the four forests, indicating that changes in the proportion of forests planted with P. tabuliformis had a significant effect on the β-diversity of soil microorganisms ( Figure 2E, F ). The microbiological co-occurrence network highlighted intricate interactions among soil microbiological communities. All networks displayed modular structures (Modularity > 0.4) ( Figure 3 ) (Xue et al., 2017). Although the dominant phyla of soil bacteria and fungi were consistent across four forests, their relative abundance varied significantly. For bacteria, total relative abundance of dominant microbiological phyla was lowest in PT100% soils. pH was key predictor of bacterial diversity ( Figure 6A ), the significant negative correlation between pH and the relative abundance of Verrucomicrobiota suggests that the relative abundance of Verrucomicrobiota increases at low pH, which is due to the relatively narrow growth tolerance of most microbiological taxa (Xue et al., 2017). PLS-PM analyses integrated soil C, soil N, AK, pH and soil microbiological diversity, and elucidated the interrelationships among these components ( Figure 7 ) (Zhong et al., 2020; Li et al., 2021). In this study, percentage of PT was associated with changes in soil bacterial diversity. Percentage of PT was found to influence the AK, and changes in AK was an important factor (Zarafshar et al., 2020) affecting the soil bacterial diversity (Lange et al., 2015). pH indirectly affects soil bacterial diversity. The relative abundance of Acidobacteriota was lower in PT100% than in other three forests, and the pH of the PT100% was the most acidic, suggesting that excessive planting of P. tabuliformis in forests tends to lead to soil acidification and promotes the growth and development of Acidobacteriota. For fungi, AK and SOC were key predictors of fungi diversity ( Figure 6B ). PLS-PM analyses showed that the percentage of PT was associated with changes in soil fungal diversity. Changes in the proportion of P. tabuliformis plantings had a smaller effect on bacterial diversity than fungal diversity, due to the fact that bacteria can metabolize a wider range of compounds, and are therefore more adaptable to environmental change and have a more stable community structure than fungal communities (Uroz et al., 2016; Wang et al., 2019; Cheng et al., 2023).