The experiment sample site was located in a Korean pine plantation in Liangshui National Nature Reserve, Heilongjiang Province, China (128°53′20″E, 47°10′50″N); natural N deposition is 12.93 kg N ha⁻¹ year⁻¹. The area has a distinct temperate continental monsoon climate with an average annual temperature of –0.3°C, of which the average annual precipitation of 676 mm, a frost-free period of 100–120 d, and an annual snowpack period of 130–150 d. The Korean pine plantation in this test site was established in 1954 after a clear-cutting of the mixed broadleaved-Korean pine forest. The dominant plant species are Korean pine and undergrowth-associated herbs (Oxalis corniculate and Mitella nuda).

The simulated N deposition experiment commenced in 2014, and the sample plots were treated with N application by understory manual spraying from June to September (four times per year). Each sample plot was 5 m × 30 m, bounded by a 10 m wide buffer strip. The required NH₄NO₃ was dissolved in 20 L water when it was needed. There were four treatments as follows: CK (0 kg N ha⁻¹ year⁻¹), LN (20 kg N ha⁻¹ year⁻¹), MN (40 kg N ha⁻¹ year⁻¹), and HN (80 kg N ha⁻¹ year⁻¹), which were approximately twice, four times, and eight times the level of bulk N deposition in the region, respectively (Song et al., 2019). There were three replicates in each treatment with a completely randomized design.

Soil samples were collected in July 2020. After further removal of the top layer of plastic material, three small random sample squares (1 m×1 m) were set in each sample plot, and five random samples were collected from each plot and mixed into a composite sample (15 cm³, a total of 12 composite samples). Soil samples were sieved and divided into three parts: one part was stored at –80°C for soil microbial DNA extraction, another part was air-dried and used for soil chemical property determination, and the remaining part was used to determine the composition of soil aggregates and measures the content of GRSP.

The air-dried soil samples were sieved through soil sieves with apertures of 2 mm, 1 mm, and 0.25 mm to obtain four groups of soil samples (>2 mm, 1–2 mm, 0.25–1 mm, <0.25 mm) (Muruganandam et al., 2009). Each group of soil samples was weighed. Soil structural stability was evaluated by the mean weight diameter (MWD) of water-stable aggregates (Zhao et al., 2017). The contents of easily extractable glomalin-related soil protein (EE-GRSP) and total glomalin-related soil protein (T-GRSP) were determined (Janos et al., 2008). EE-GRSP was extracted from 0.5 g of soil with 4 ml of 20 mmol L⁻¹ sodium citrate (pH=7.0) and autoclaved at 121°C for 30 min. T-GRSP was extracted from 0.5 g of soil with 4 ml of 50 mmol L⁻¹ sodium citrate (pH=8.0) and autoclaved at 121°C for 60 min. The supernatant was collected at 4°C, 10,000 rpm for 10 min. T-GRSP extraction was performed six times until the solution turned straw-colored (Gao et al., 2017). The GRSP content was determined using a spectrophotometer (595 nm) with bovine serum albumin as the standard.

The soil pH of the air-dried soil samples was determined using a pH meter (FiveEasy FE20) on a mixture with a water-to-soil ratio of 2.5:1. TC (total carbon), TOC (total organic carbon), and total N (TN) were determined by an analyzer (multi N/C 2100 TOC/TN Analyze, Analytik Jena, Germany). Soil ammonium N (NH₄ ⁺-N) and nitrate N (NO₃ ^–N) were measured using a continuous flow analyzer (BRAN+LUEBBEAA3, Germany). Soil total phosphorus (TP) and available phosphorus (AP) were determined using the molybdenum antimony colorimetric method determined by the molybdenum antimony anti-colorimetric method.

We explored herbaceous plant diversity in July 2020. Three small random sample squares (1 m×1 m) were set in each sample plot, and the species of herbaceous plants, the number of individuals of the species (counted as a single plant in the above-ground part), height, and cover were investigated. The fine roots (<2 mm in diameter) of herbaceous plants were selected, and the root biomass was measured (Li et al., 2021). We determined the AM fungal colonization rate of roots from herbs by a modified aniline blue staining method (Zubek et al., 2011). In addition, we randomly selected 50 mixed herbaceous roots from each sample site for the infestation rate experiment.

One-way ANOVA and Duncan’s multiple tests (P < 0.05) were performed using IBM SPSS Statistics 26 (IBM Corporation, New York) software to compare the means of plant, AM fungi, and soil physicochemical property data. The correlation between soil physicochemical properties and herbaceous plant diversity was calculated by Pearson’s test (two-tailed) at two significant levels, P < 0.05 and P < 0.01. RDA analysis was used to reveal the effect of soil properties on the community composition of AM fungi. Structural equation modeling (SEM) was used to predict causal pathways of N addition, key soil physicochemical indicators, and AM fungi on herbaceous plant diversity. The fit of models that had been previously built was evaluated using chi-squared test, root mean square error of approximation (RMSEA), and goodness-of-fit index (GFI).

RDA analysis showed that six soil physicochemical factors explained 91.12% of the variation in the AM fungal community, as shown in Figure 5 . Soil pH and NO₃ ^–N, AP, NH₄ ⁺-N, and TN content significantly influenced the distribution of AM fungal species. In addition, Glomus was negatively correlated with pH and positively correlated with NH₄ ⁺-N and NO₃ ^–N content. Paraglomus was positively correlated with AP content and pH value.

Long-term N addition reduced the diversity of understory herbaceous plants. The correlation was analyzed between α diversity index of herbs and biochemical factors by Pearson correlation. The diversity of understory herbaceous plants was significantly and positively correlated with the diversity of AM fungi (P < 0.05, Figure 6 ) and the colonization rate of AM fungi (R² = 0.66). The diversity of herbs was significantly and positively correlated with total N content (P < 0.05), significantly and negatively correlated with NH₄ ⁺-N content (P < 0.01), and significantly and positively correlated with pH (R² = 0.67). Additionally, herbaceous plant fine root biomass was significantly and positively correlated with their diversity (P < 0.01).

After correlation analysis of soil physical and chemical indicators, herbaceous plants, and AM fungi, we built SEM models for important indicators and obtained the following models after multiple fitting ( Figure 7 ). The P-value of χ² was higher than 0.05, indicating that the model structure was reasonable. The SEM throughput analysis largely explained the effects of biotic as well as abiotic factors on herb diversity (R² > 90%). N addition indirectly affected AM fungal diversity and plant diversity through soil pH value, ammonium N, and AP content. Among these, AM fungal diversity was the strongest pathway affecting plant diversity.

Soil acidification is caused by N deposition, and soil pH decreases linearly with increasing N application (Tian and Niu, 2015). In this study, long-term HN treatment (80 kg N ha⁻¹ year⁻¹) reduced soil pH, while LN and MN treatments had no significant effect on soil pH. The reason may be that soils in terrestrial ecosystems are more sensitive to N inputs, while forest ecosystems are more stable than other types of ecosystems (Roth et al., 2021).

MN treatment had a significant effect on soil TP content, while AP content decreased to varying degrees with N addition. This may be because soil acidification activates aluminum ions and iron ions in the soil, causing more phosphorus to be absorbed by the soil and resulting in a decrease in AP content. When plants and microorganisms absorb and utilize AP, they get a corresponding supplement from the phosphorus element of other components, which leads to the reduction of AP content.

Soil TN content was stable, while ammonium content increased linearly with N application. The results of Qin et al. (2022) showed that long-term N application can change the soil ammonium and nitrate content. In addition, total N increased significantly with the increase of N addition. The stability of the whole N content in this experiment may be because the forest ecosystem is more stable and can resist a certain concentration of N addition.

There was an increased tendency for EE-GRSP content as the N addition concentration was increased, and the T-GRSP content was significantly higher in the HN treatment than in the CK treatment, which was probably due to the significant acidification of the soil (Agnihotri et al., 2022). On the contrary, short-term N addition reduced the GRSP content in the soil (Jia et al., 2021; Huang et al., 2022). The reason might be that short-term N additions could not significantly change soil acidity and alkalinity, but lead to a decrease in GRSP by changing soil dissolved organic carbon and N/P ratio. The proportion of large soil aggregates and GRSP content were significantly elevated in the HN treatment. Pearson correlation analysis of in-situ soil GRSP content and MWD showed that both EE-GRSP and T-GRSP were significantly positively correlated with MWD values ( Figure 3 ).

N addition and local soil conditions affect plants and soil microorganisms (Cao et al., 2020; Zeng et al., 2021). In this study, both high-throughput sequencing and traditional spore morphological identification were used to investigate the community structure and diversity of woodland AM fungi. Morphological identification is inconsistent, limited, and subjective, but can be improved by means such as increasing the sample size. High-throughput sequencing technology has been widely used by researchers but still has some limitations. Due to the variety of microorganisms, the current NCBI database cannot cover all the microorganisms, or lacks accurate species information, resulting in the omission of some low-abundance species. Thus, we combined the two methods to accurately reflect the diversity and structural characteristics of AM fungi (Chaudhary et al., 2020). Both high-throughput sequencing and morphological identification results showed that N addition altered AM fungi community composition and reduced AM fungal alpha diversity.

RDA analysis showed that the community composition of AM fungi could be significantly affected by pH in the Korean pine plantation ( Figure 5 ). Studies have reported that soil pH is one of the most important abiotic factors affecting the ecological distribution of AM fungi (Davison et al., 2021). As the N concentration increased, the abundance of Glomus gradually increased and was significantly higher in the HN treatment ( Figure 4 ). The main reason for this phenomenon might be that Glomus showed better ecological advantages in acidic soils. Studies have shown that AM fungi alleviate the negative effect of nitrogen deposition on ecosystem functions (Kang et al., 2020). In the future, attention should be paid to the ecological functions of Glomus under the background of nitrogen deposition. Short-term HN treatment (80 kg N ha⁻¹ year⁻¹) also reduced fungal diversity and changed community composition (Cao et al., 2020). This phenomenon is mainly due to the changes in phosphorus elements in the soil caused by N addition, causing plants to select AM fungi with high phosphorus uptake ability.

Cases of reduced plant diversity caused by soil acidification induced by N deposition have been reported (Kimmel et al., 2020). In this study, we found that the diversity of understory herbaceous plants in forests also decreased with lower soil pH. Additionally, excess N can also create nutrient imbalances. Large inputs of exogenous N will break the original stable stoichiometry, causing nutrient imbalances in soil and plants, and leading to the loss of certain plant species. N deposition indirectly affects plant species diversity by changing the composition of soil microbial communities (Bobbink et al., 2010). This study also found that changes in understory plant diversity were linearly positively correlated with AM fungal diversity and AM fungal colonization rate ( Figure 6 ), which also verified the role of AM fungi in maintaining the stability of understory plants. In addition, long-term N application will significantly reduce plant fine root biomass, while a small amount of N application will increase plant root biomass (Frew, 2022). SEM clearly showed that AM fungal diversity was the strongest influencing pathway affecting herb diversity under N addition conditions ( Figure 7 ), which further proved the ability of AM fungi to maintain understory herb diversity.