The research samples were collected from Zhaodong (Heilongjiang Province, China), located in the middle of Songnen plain with a temperate continental monsoon climate, and the weather is characterized by high temperature-rainy summers and dry cold winters. The terrain is flat, but the ground is threatened by salinization. Seasonal precipitation results in the alternation of soil desalting and salt accumulation in rainy and dry seasons, respectively, and the soil is salinized severely. The annual average temperature is 3.6–4.4°C, and the temperature exceeds 0°C for more than 210 days annually (Su et al., 2020). S. glauca, P. tenuiflora, and L. chinensis communities are the main vegetation types of Songnen saline-alkali grassland, among which the third stage of succession (L. chinensis communities) is the most abundant in this land. Besides, there are also a large number of alkali spots without vegetation growth before the first succession stage. Five large quadrats (50 m × 50 m) separated by roads were set in this area at a constant distance from each other. Three small sites (15 m × 15 m) within a quadrat were selected randomly. S. glauca, P. tenuiflora, and L. chinensis communities grew simultaneously in every small site, with no cross growth; therefore, each site possessed a complete succession sequence (Fig. 1). The information on the five large quadrats is shown in Table 1.

Ten clumps of S. glauca, P. tenuiflora, and L. chinensis with different growth statuses and ages were selected within each small site according to the “multi-point parallel sampling method” and “five-point sampling method” on May 24, 2021 (Meng, 1996). The selected plants were shaken vigorously to remove excess soil. The rhizosphere soil still stuck to the roots, was removed using a sterile brush (Jin et al., 2021). Subsequently, the rhizosphere soils (about 1 kg) with the number of absorptive roots from depths of 0–30 cm were harvested from every small site. After this, 0.5 kg rhizosphere soil was retained by the quartering method, and three soil specimens of every succession stage collected from three small sites within a large quadrat were mixed into one sample so that five samples were collected from each of the three succession stages. Fifteen rhizosphere soil samples were obtained, which were respectively packed into sterilized sealed bags, numbered, and latitude and longitude were marked. The retrieved rhizosphere soils were sieved through 0.85 mm mesh after being air-dried and then divided into two parts. One portion was stored at 4°C for the determination of soil properties and enzyme activities, and the other was immediately submitted for DNA extraction and high-throughput sequencing of AM fungal communities.

Soil physicochemical properties were determined according to Bao (2000), and the soil enzyme activities were measured as described by Guan (1986). Soil alkalinity was explained in terms of the pH value of soil-water immersion liquid (1:5), and it was measured by a PHS-3C pH meter (Shanghai Lei Ci Scientific Instrument Factory). Soil salinity was illustrated by the conductivity of soil-water (1:4) saturated extract and was measured by a DDS-11A conductivity meter (Shanghai Precision Instrument Factory). The OC and OM were measured using the potassium chromate volumetric analysis method, carbonate content of soil-water immersion liquid (1:5) was estimated by the phenolphthalein-neutralization titration, and Total N was estimated by Kjeldahl’s semi-micro method after the soil samples were digested in the H₂SO₄ and accelerator at 410°C. Total P content was determined using the molybdenum antimony colorimetric method after the soil samples covered with NaOH were liquated at 450°C. Sucrase activity was determined by 3, 5-dinitrosalicylic acid colorimetry after 5 g soil and 15 ml 8% sucrose solution were cultured at 37°C for 24 h. Cellulase activity was determined by dinitrosalicylic acid colorimetry after 10 g soil and 20 ml 1% carboxymethyl cellulose solution was incubated at 37°C for 72 h. Five grams of soil and 10 ml of 10% urea were incubated at 37°C for 24 h. Then the NaClO-sodium phenolate colorimetry was used to measure urease activities. Catalase activity was determined by potassium permanganate volumetric analysis. All the above-mentioned indices were measured at least three times.

Five grams of rhizosphere soil was accurately weighed from each of the 15 samples in aseptic conditions, placed into 10 ml sterilizing centrifuge tubes, and submitted to Personal Biotechnology Co., Ltd., Shanghai, China, for subsequent processing and sample sequencing. The Omega Mag-bind soil DNA kit (Omega M5635-02) was used for the soil-sample DNA extraction, and fragments were amplified using the AM fungal-specific primer AMV4.5NF (5’-AAGCTCGTAGTTGAATTTCG-3’) and AMDGR (5’-CCCAACTATCCCTATTAATTAT-3’) (fragment of about 280 bp length). Amplification system (25 μl) contained 5 × reaction buffer (5 μl), 5×GC buffer (5 μl), dNTP (2.5 mM, 2 μl), forward primer (10 µM, 1 μl), reverse primer (10 µM, 1 μl), DNA template (2 μl), ddH₂O 8.75 μl, and Q5 DNA Polymerase (0.25 μl). The amplification conditions were set as follows: initial denaturation at 98°C for 2 min, then 30 cycles of denaturation at 98°C for 15 s, annealing at 55°C for 30 s, extension at 72°C for 30 s, and a final extension at 72°C for 5 min. The sequencing platform was Illumina (NovaSeq-PE250), and more than 80000 sequences were generated for each sample. QIIME2 was used to denoise the obtained sequences; the main steps were as follows: “qiime cutadapt trim-paired” was called to excise sequences and discard these with unmatched primers; then, DADA2 (“qiime dada2 denoise-paired”) was applied for qualitative control, denoising, splicing, and chimera removal. Subsequently, the clustering was performed at the level of 100% similarity (Callahan et al., 2016). After that, representative sequences of ASVs and their tables were merged. The “classify-sklearn algorithm” of QIIME2 (Bokulich et al., 2018) was used to align the characteristic sequence of each ASV to the MaarjAM database for species annotation.

SPSS version 25 was used to test the soil properties and enzyme activities among the three stages by one-way ANOVA. Pearson correlation analysis was performed between the above indices and alpha diversity of AM fungal community, and the significance of differences was marked according to Waller Duncan’s test results. The “qiime diversity alpha-rarefaction” function in QIIME2 was used to construct the Rarefaction Curve to predict the total species number and the relative abundance of each sample at a given sequence depth (Heck et al., 1975; Kemp and Aller, 2004). The Anosim function in the R4.1.1 Vegan package was used to test the significance of the difference in community composition between different stages. The decorana function was used to test the axis lengths of community data, and then the canonical correlation analysis (CCA) was constructed to explain the influence of soil characteristics on AM fungal community. QIIME2 was used to analyze the alpha diversity (Shannon’s and Simpson’s indices, Faith’s PD, Pielou’s evenness, Chao1 index, Observed species, and Good’s coverage indicated the richness, diversity, evolution-based diversity, evenness, and coverage, respectively, of AM fungal community) and then the boxplot was plotted using R4.1.1 ggplot2 package (Chao, 1984; Faith, 1992; Good, 1953; Pielou, 1966; Simpson, 1949; Shannon, 1948). The alpha diversity indices were tested by Dunn’s post-hoc test. The ggraph and ggplot2 packages were used to build taxonomic rank tree plots to illustrate the species composition of AM fungal community at different stages (Carrión et al., 2019). Vegan, ape, and ggtree packages were used for hierarchical clustering analysis to reveal the Beta diversity of AM fungal communities in the three stages. MetagenomeSeq package was used for the MetagenomeSeq analysis of AM fungal species that changed significantly among three stages (Zgadzaj et al., 2016). Metabolic pathway difference analysis, which was applied to identify the significantly enriched metabolic pathways among AM fungal communities, was performed with PICRUSt2 and MetagenomeSeq package in R4.1.1. Among these, PICRUSt2 predicted the functional pathways based on the abundance of pathways genes in sequencing data, as per the process described in https://github.com/picrust/picrust2/wiki. To test the stability of AM fungal community in changing environments, the igraph package was used to analyze the degree of distribution. Further, gephi software was used to elucidate interspecific relationships within AM fungal communities at the same succession stage (Bastian et al., 2009).

Soil physicochemical properties and enzyme activities differed among different succession stages (Table 2). The pH, EC, and carbonate were the maximum at the stage of S. glauca, while those at the L. chinensis were the lowest, indicating that the highest salinization appeared in the S. glauca stage, and the soil salinization gradually decreased as the succession progressed. Contrary to the above properties, Total P, sucrase activity, and catalase activity were the highest in the L. chinensis stage, followed by P. tenuiflora and S. glauca, contrary to the succession sequence in the Songnen saline-alkali grassland. The OC, OM, and Total N contents showed the same trend, and decreased gradually in the order of L. chinensis, S. glauca, and P. tenuiflora stages. C/N was the lowest in the L. chinensis stage, and was significantly different from that in the other two stages (p < 0.05), while there was no significant difference between L. chinensis and S. glauca (p > 0.05) in terms of cellulase activity; the value was significantly lower than that in the P. tenuiflora stage (p < 0.05). Furthermore, the difference in urease activities between L. chinensis and P. tenuiflora stages was not significant (p > 0.05), and that of the S. glauca stage was the lowest (p < 0.05).

The variation in AM fungal species diversity and total abundance with the increase in the number of sequences could be reflected by the dilution curve. Fig. 2 shows the relationship between Chao1 as well as Observed_species and sequencing depth. At the time when the depth reached 50000 sequences, the dilution curves constructed for 15 samples based on different indices flattened and reached 99.9% coverage, indicating that the sequencing data in this experiment could comprehensively reflect the AM fungal community composition. A total of 1024090 valid sequences were obtained after the unqualified bands were removed from 1389470 original sequences. In this study, 3916 ASVs were clustered, among which 2362 were from the rhizosphere soils of L. chinensis, 1063 were from that of P. tenuiflora, and 1078 ASVs were detected at the S. glauca stage. Ninety-eight ASVs were presented simultaneously in three phases, and the number of endemic ASVs in three stages was 2180, 584, and 609, respectively.

AM fungal alpha diversity analysis (Fig. 3) for three succession stages showed a significantly higher species richness of the third stage (L. chinensis) than in other stages (p < 0.05), and its value of Chao1 and Observed_species at five sampling quadrats were 606.119–883.19 and 598.3–873.9, respectively, while these indices did not differ significantly between P. tenuiflora and S. glauca stages (p = 0.52). As indices reflecting the diversity of AM fungal community, large Shannon and Simpson indices indicated a greater species diversity within the community. The Shannon and Simpson indices varied significantly only between the L. chinensis and P. tenuiflora stages (p = 0.0034, p = 0.0056), and that of L. chinensis showed the highest average values of 7.14 and 0.98, respectively. That is, the AM fungal diversity was the highest in the succession stage of L. chinensis. Pielou_e removed the richness effect of the Shannon index and emphasized the uniformity of community; the large values indicated even community composition and significant differences between L. chinensis and P. tenuiflora stages (p = 0.0071); the mean Pielou_e value of the three succession stages (L. chinensis, P. tenuiflora, and S. glauca) were 0.76, 0.58, and 0.65, respectively. The phylogenetic diversity of the community was represented by Faith_PD, which evaluated the genetic diversity by calculating the full length of the clade that represented the ASV sequence in the constructed tree. There was no significant difference in the three succession stages in terms of Faith_PD (p > 0.05), indicating some similarities in the genetic diversity of different AM fungal communities.

The significant difference in AM fungal community composition among three succession stages could be confirmed after the Anosim test (r = 0.6756, p = 0.002). Taxonomic hierarchy tree analysis (Fig. 4A) showed a larger number of unique species in the L. chinensis stage, while the AM species were more common in the adjacent succession stages (P. tenuiflora and S. glauca stages). In addition, in the top 100 ASVs, the most abundant AM fungal species appeared in the P. tenuiflora and S. glauca stages. Hierarchical clustering analysis (Fig. 4B) performed at the classification level of the family showed that Glomeraceae was the most dominant family with the largest proportion in all the succession stages, followed by Paraglomeraceae, which was distributed in the three succession stages and found in all 15 samples, and was thus the subdominant family. The distribution of Claroideoglomeraceae, Diversisporaceae, Ambisporaceae, Gigasporaceae, and Archaeosporaceae varied greatly among succession stages and samples; these AM fungal species might be more sensitive to the changing environmental conditions. Some quadrats of P. tenuiflora and S. glauca with a similar composition clustered together, consistent with the results of alpha diversity analysis, indicating that the species composition of adjacent succession stages was closely related.

Based on the results of the MetagenomeSeq analysis, the Manhattan plot was constructed, and the following figures (Fig. 5) showed the ASVs enriched in different succession stages. The results showed that four ASVs were enriched significantly in S. glauca compared with P. tenuiflora stage (Fig. 5A), among which one belonged to Paraglomus, two belonged to Glomus, and another one belonged to Glomeromycota, although the information about its family and genus was not apparent after species annotation. Compared with the L. chinensis stage (Fig. 5B), 90 ASVs were enriched significantly at the S. glauca stage, including 28 in Glomus and one in the Ambispora genus. The most enriched species was affiliated with Glomus. Compared with S. glauca, 129 ASVs were significantly enriched in the L. chinensis stage (Fig. 5C), and the enrichment effect of 32 ASVs was extremely significant. The significantly upregulated species were mainly distributed in the Glomus, Claroideoglomus, and Paraglomus (85, two, and five, respectively). One-hundred and fifteen ASVs were significantly enriched in the L. chinensis stage compared with that in the P. tenuiflora (Fig. 5D) stage, of which 107 coincided with the enriched ASVs of L. chinensis relative to those in the S. glauca stage. Significantly up-regulated species were also mainly distributed in Glomus, Claroideoglomus, and Paraglomus. Interestingly, MetagenomeSeq analysis of P. tenuiflora samples and its comparison to those of S. glauca showed that no ASV was enriched significantly (Fig. 5E), which further verified the similarity of AM fungal community composition in the first two stages of saline-alkali grassland succession. Sixty-three ASVs were enriched in the P. tenuiflora stage compared to the L. chinensis (Fig. 5F) stage, among which 17 belonged to Glomus, two belonged to Paraglomus, and 29 were the overlapping species that were significantly enriched in the S. glauca stage, in contrast to the L. chinensis stage. These significantly enriched species were considered the markers of different succession stages, and they might play a great impact on the response of the community to grassland succession.

The results of association networks analysis (Fig. 6A) of AM fungal community in different succession stages showed the closest association presented in the rhizosphere of L. chinensis, with a dense associated network having 38 nodes and 40 connections. The AM fungal community surrounding P. tenuiflora had the lowest association with only eight nodes, while the AM fungal community around S. glauca roots had 19 nodes and 18 lines. A high correlation suggested a close relationship between AM fungal species within the community. This also indicates that the community had low evolutionary diversity and was more vulnerable to dramatically changing conditions. Both the AM fungal composition in succession stages of S. glauca and P. tenuiflora had a lower correlation, indicating that their composition was diverse and the communities were stable. Moreover, the degree distribution analysis (Fig. 6B) of S. glauca and P. tenuiflora stages showed that the empirical network and random network possessed similar morphology and were presented as a regular bell shape, in line with the characteristics of the small-world network, equipped with high stability compared to the scale-free network that showed both random fault robustness and vulnerability to a targeted attack.

AM fungal communities often play ecological functions, which may be closely related to their metabolic pathways. Functional predictions (Fig. 7) were performed based on the pathway gene abundance of sequencing data to explore the functional differences in AM fungal communities in the three succession stages. No significant difference was observed between metabolic pathways of S. glauca and P. tenuiflora, which shared similarities in the community composition analysis. Seventy-eight differentially expressed metabolic pathways were observed in the L. chinensis stage compared to those in the S. glauca stage, 51 of which were significantly different (Fig. 7A). The metabolic pathways with the largest values of up-regulation and down-regulation were HSERMETANA-PWY and PWY-5754, respectively, and the metabolic processes involved were L-methionine biosynthesis III and 4-hydroxybenzoate biosynthesis I. The most significantly regulated metabolic process was palmitate biosynthesis I with the pathway number PWY-5994. A total of 77 differentially expressed metabolic pathways were observed in the L. chinensis stage compared with the P. tenuiflora stage, among which 46 were significantly different (Fig. 7B). The most significant difference was observed in the PWY-7007 pathway, whose primary function is methyl ketone biosynthesis. The largest up-regulation and down-regulation were observed in pathways PWY-6837 and PWY-5754, with respective functions of fatty acid beta-oxidation V (unsaturated, odd number, di-isomerase-dependent) and 4-hydroxybenzoate biosynthesis I (eukaryotes), respectively.

The axis length of community data was 7.178 (> 4) so that CCA between AM fungal communities and soil properties were conducted to understand the response mechanism of AM fungal community to the succession of Songnen saline-alkali grassland (Fig. 8) and the characteristic values of CCA1 and CCA2 were 0.924 and 0.587, respectively. AM fungal communities in S. glauca and P. tenuiflora stages showed similarity, and that manifested as the relatively close distance in the figure. Cellulase activity exerted a slight and insignificant effect on AM fungal community, while the pH played the most significant effect (r = 0.977, p = 0.001). In addition, Total N content and sucrase activity also played a prominent impact on AM fungal community composition (r = 0.954, r = 0.94, p = 0.001). The directions of ASV_7, ASV_15, and ASV_17 were relatively consistent with Total N, Total P, OC, OM, and the activities of urease, sucrase, and catalase, showed a positive correlation, while ASV_1, ASV_2, ASV_3, ASV_4, ASV_5, ASV_6, and ASV_8 were negatively correlated with these indices but presented a positive correlation with cellulase activity, EC, carbonate, C/N, and pH. The arrows of these ASVs were lengthy and had a close projection distance with most of the quadrats, indicating that these ASVs played a great influence on the composition of AM fungal communities in different succession stages.

The correlation analysis (Table 3) between soil physicochemical properties and various alpha diversity indices showed the significant correlation of pH, OC, OM, and sucrase activity with all alpha diversity indices; among these, pH was negatively correlated with these indices. These properties not only affected the distribution of AM fungal species in the rhizosphere but also influenced the diversity within the communities. All the indicators except cellulase activity possessed significant or extremely significant associations with Chao1 and Observed_sp of AM fungal communities. The indices of eleven soil properties and enzyme activities jointly impacted the richness of AM fungal communities and the distribution of some species. Furthermore, various properties except for catalase activity and urease activity were significantly or extremely significantly associated with the Shannon index; among these, pH, OC, OM, Total N, and sucrase activity were also correlated with the Simpson index. These soil properties were important to drive the variation of communities. Pielou_e correlated with all the indices except for EC, catalase activity, and urease activity; the others exerted effects on the uniformity of AM fungal communities, among which the pH, carbonate, C/N, and cellulase activity were negatively associated with Pielou_e. The results also indicated that the pH, carbonate, C/N, and cellulase activity suppressed the homogeneous distribution of ASVs within the communities and exerted a selective role in the appearance of species.

The salinization in Songnen saline-alkali grassland has aggravated in recent years compared with that reported in previous research (Chen, 2017; Yue, 2015). In the first succession stage after the alkali patch, the pH, EC, and CO content were the highest, confirming the highest degree of salinization. Furthermore, the soil quality improved, and the saline-alkali habitat was restored in the process of positive-going succession. The low-Total P content was observed at every stage, and it was related to the fact that the Songnen saline-alkaline soil has a high content of carbonates (Yan et al., 2015). The presence of abundant free calcium carbonate promotes the conversion of phosphorus to calcium phosphate (Bao, 2000). The gradual increase in the Total P content as the succession occurred, was in accordance with the variation in carbonate concentration in the three succession stages. The variation tendencies of OC, OM, and Total N were consistent, and the highest values were observed in the third stage (L. chinensis). Relevant studies have shown the C and N of different grassland types and natural zones were discrepant significantly (Wang et al., 2014). Moreover, these indicators were closely related to microbial activities. Interestingly, this investigation showed disparities in functional pathways and composition of AM fungal communities among three succession stages, which would affect the secretion and release of some substances. The comprehensive effect of the above factors may be the reason for the differences mentioned in OC, OM, and Total N. Furthermore, the changes in OC, OM, and Total N were also closely related to the activities of sucrase, cellulase, and urease. Sucrase improved the bioavailability of OM and OC (Xia et al., 2018), urease improved soil nitrogen supplement (Xie et al., 2017), and cellulase played an important role in the decomposition of plant residues to soil carbohydrates. The decomposition of hydrogen peroxide by catalase in the soil prevented its toxicity to organisms (Guan, 1986) so that catalase activity could act as an index to evaluate soil oxidation ability, and it also could be considered to associate with soil OM and microbes (Yang and Lu, 2022). Urease might manifest higher activity in response to the lacking Total N in the P. tenuiflora stage, ensuring N supply. Meanwhile, the salinity of Songnen saline-alkaline grassland showed dynamic variation among seasons (Su et al., 2020). As the intermediate stage of succession, P. tenuiflora might be more sensitive to environmental changes. The degradation or succession of grassland caused by the variation in salinity was brought about the residues decomposition of P. tenuiflora; subsequently, significantly high values of cellulase activity were presented at this stage. The variation in activities of catalase and sucrase showed a tendency that coincided with the succession sequence of Songnen saline-alkaline, indicating a possibly high correlation with salinity.

The dominant AM fungal genus in all the three succession stages was Glomus. This was consistent with the previous research and the view that Glomus is a broad-spectrum symbiotic system (Araujo et al., 2021; Haug et al., 2021; Zhang et al., 1994). Furthermore, Paraglomus, Claroideoglomus, Diversispora, Ambispora, and Archaeospora were annotated in this study and accounted for some proportions. Due to the small size of the spores in these genera, their microspores produce a large number of spores in a short time and are easy to spread (Hepper, 1984). Community analysis showed that the alpha diversity, community composition, association network, and functional pathways of the L. chinensis stage differed significantly from the previous two stages. The consequence occurred due to environmental conditions, and the perspective that the composition of AM fungal community is highly influenced by environmental factors has been discussed in previous research (Davison et al., 2015; Dumbrell et al., 2010). The lowest salinization with higher environmental specialization presented in the L. chinensis stage, and its rhizosphere environment was more suitable than that of S. glauca and P. tenuiflora for the subsistence of AM fungi (Edwards et al., 2015). As a result, the AM fungi in the L. chinensis stage were highly diverse but with non-significant genetic diversity among the three stages. In the first two stages of succession (S. glauca and P. tenuiflora), while the soil properties and enzyme activities differed, the salinization was relatively serious in both, which might inhibit the colonization of AM fungi, resulting in insignificant differences in AM fungal community composition. Moreover, the succession in Songnen saline-alkali grassland was greatly affected by the rainy season (Yang et al., 2019); thus, the first two stages were unstable and prone to degradation, which resulted in the selection of AM fungi species. Therefore, less ASV was significantly enriched in S. glauca and P. tenuiflora stages compared to the L. chinensis stage. Community composition and functional pathways were often closely related to the function of AM fungi. The marker species and differential pathways at three stages might be related to the survival of AM fungal communities and hosts in special habitats. These fungi might play an important role in promoting saline-alkali succession, and the associated, distinct functional pathways should be further investigated. However, community analysis based on high-throughput sequencing technology lacked a complete database for species annotation, and some ASVs failed to be specifically classified. Classical morphological identification should be used to identify the AM fungi at different succession stages and to screen AM fungal species with high application value in later studies. The results of diversity analysis based on different sequencing targets were disparate (Justine et al., 2020). Internal transcribed spacer (ITS) possessed a 72% success rate in distinguishing fungi (Schoch et al., 2012), large subunit (LSU) worked better in the species annotation of some communities, and small subunit (SSU) was also commonly used in diversity analysis. It is more scientific and combines the sequencing results of different targets to analyze community diversity.

We observed a significant negative correlation between pH and alpha diversity, consistent with Adenan et al. (2020), and it verified the claim that pH directly affects the occurrence and population distribution of AM fungi (Adenan et al., 2020; Carvalho et al., 2003). The EC and carbonate content were also negatively correlated with multiple diversity indicators because saline stress limited the spore germination and mycelium growth (Medina et al., 2015). Besides, cellulase activity negatively correlated with AM fungal diversity because it participated in the decomposition of plant root residues (Wu et al., 2022) and could reduce the colonization sites of AM fungi; subsequently, some AM fungi died and the community diversity decreased. OC, OM, and sucrase activity are indicators related to soil nutrition, and they were associated with fungal community diversity in the rhizosphere of S. glauca, P. tenuiflora, and L. chinensis. Identical results were obtained in previous research (Ren et al., 2021); thus it could be interpreted that the areas with a richness of C sources and OM contribute to the growth and colonization of microorganisms (Li, 2012). Catalase activity was positively correlated with fungal diversity, and the possible reason was that the sample sites with high catalase activity cleared the hydrogen peroxide in the soil on time to avoid damage to plants and microorganisms. Conclusions for the impact of soil N content on AM fungal community diversity are confusing (Emery et al., 2022; Justine et al., 2020). Exogenous addition experiments showed that N application could increase the AM fungal richness and diversity in the condition of P deficiency (Louise et al., 2007), demonstrating a positive correlation between N and AM fungal diversity, consistent with the results of this study. Urease activity was associated with N supply and thus showed consistent results with N. It was noteworthy that Total P content was positively correlated with AM fungal diversity, contrary to the previous understanding (Jerbi et al., 2021; Chen et al., 2014); we hypothesize that the low P content in the experimental quadrats significantly inhibited the growth of plants. The P content increased within a certain range and guaranteed the nutrient supply of plants, thus ensuring the stability of symbiosis and increasing the diversity of AM fungi. These results are of great significance for the realization of the response of AM fungal community composition to the succession of saline-alkali land and lay a foundation for further research on the functional characteristics of AM fungal communities in different succession stages. These response mechanisms should be unraveled at the molecular level in future studies to thus provide technical support for artificial regulation of community composition and function.