Samples were taken from four locations (Table 1): the Yangbajing Basin, which is rich in geothermal resources; the core experimental area of the Everest National Nature Reserve; and Zarinanmucuo and Zabuyecuo, which are located in Qiangtang National Nature Reserve in the northern Tibetan Plateau. This scientific investigation has been reported to and approved by the Tibet Autonomous Region Forestry Department and the competent departments of nature reserves.

This research started in August 2017, and O. glacialis was sampled during the peak period. There were a total of four sampling points (Table 1). Each sample point was set with three repeats, and the distance between the three repeats was above 100 m (Dombrowski et al., 2017). A total of 12 samples were collected. In particular, soil of root system, also called root-zone soil, refers to soil that is loosely attached to or close to the roots, while bulk soil is soil outside the root zone, far from the plant’s roots (Derr et al., 2016). For the sampling, the whole plant of O. glacialis was dug up, the bulk soil was shaken into a sterile bag for the determination of the soil physical and chemical factors, and then the soil of root system of the plants was rinsed into 95% ethanol and loaded into sterile tubes. Then, the samples were stored at −20°C in a car refrigerator (Meigu Mobicool, CF-50), brought back to the laboratory, and placed in a −80°C ultralow temperature refrigerator (Jiangsu Shenglan, DW86L-158) for cryopreservation. Finally, the samples were used for total DNA extraction of soil fungal microbiome.

The environmental factors to be measured included pH value, soil moisture (SM), electrical conductivity (EC), organic matter (OM), available K (AK), available P (AP), and ammonium N (AN), for a total of seven indicators. The pH was measured by the potentiometer method (HANA, HI98103), and the ratio of water and soil was 1:1. The soil water content was measured by the high-temperature drying method. Five grams of fresh soil was weighed and placed in a constant-temperature drying box (Shanghai Jinghong, DK-420S) and measured at 105 ± 2°C for 48 h. The calculation formula was SM (%) = (5-dry weight)/5 × 100%. The conductivity was determined by the conductivity meter method (HANA, HI98304), and the water-soil ratio was 3:1. OM, available potassium, available phosphorus, and ammonium nitrogen were measured by a portable soil composition analyzer (Beijing Youputongyong, UPA–B506); OM measurement was based on the potassium complicate method, available potassium was based on ammonium acetate extraction-atomic absorption spectrophotometry, available phosphorus was based on sodium bicarbonate extraction-molybdenum antimony anti-spectrophotometry, and ammonium nitrogen is based on Nessler reagent colorimetry (Cao et al., 2019).

The total DNA of the soil fungal of root system was extracted by the sodium dodecyl sulfate (SDS) plus enzyme method (Liu et al., 2015). The primers were ITS1F (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS2R (5′-GCTGCGTTCTTCATCGATGC-3′) for PCR amplification of the fungal ITS1-ITS2 region (Adams et al., 2013). PCR used a 20 μl reaction system: 10X buffer, 2 μl; 2.5 mM dNTPs, 2 μl; bidirectional primers (5 μM), 0.8 μl; rTaq polymerase, 0.2 μl; BSA, 0.2 μl; template, DNA 10 ng; and ddH2O supplement to 20 μl. The PCR parameters were 95°C for 3 min, followed by 25 cycles of 95°C for 30 s, 56°C for 30 s, and 72°C for 45 s, then 72°C for 10 min, and 10°C until halted by user. NanoDrop One was used to detect the purity and concentration of PCR products, and the samples were sent to Shanghai Majorbio Biopharm Technology Co., Ltd. for sequencing using the Illumina HiSeq platform.

Using Illumina HiSeq sequencing to obtain paired-end double-end sequence splicing with FLAS software (FLAS 1.2.11), the raw tags obtained by Trimmomatic software were subjected to sequence defiltering to obtain high-quality clean tag data. Usearch software was used to perform sequence classification and annotation on clean tag data in the Silva database base, and clustering and species classification statistical analysis were performed based on the operational classification unit OTU (operational taxonomic unit) data of 97% similarity pairs (Nouioui et al., 2018). Analysis of similarities (ANOSIM) similarity analysis based on the minimum number of samples was used to determine whether the sample grouping setting is feasible. Mothur software (version v.1.30.1) was used to analyze the alpha diversity, including the Shannon, Sobs, Simpson, Ace, Chao, and coverage indexes, and a t-test was used to test the difference between the diversity groups. Beta diversity was analyzed through non-metric multidimensional scaling (NMDS) non-metric multidimensional scaling analysis. The bar graph and Venn diagram reflect the structure and function of fungal community. Different species and functional genes of different fungal community, using One_way ANOVA single factor analysis of variance, Kruskal-Wallis H test multiple test correction Fdr, Post hoc test Tukey-kramer level value is 0.95, and passed Linear discriminant analysis Effect Size (LEfSe) Discriminant analysis of multilevel species differences determined the species and genes with significantly different enrichment. In terms of fungal structure association and model prediction, random forest analysis was used to determine the ecological evolution process of the fungal community, and the network co-occurrence network and the correlation network were used to determine the core flora and key species. Using MEGA software, the selected SW-producing fungal sequences were combined with known related fungal sequences to construct an NJ phylogenetic tree. Variance expansion factor (VIF) was used to screen environmental factors and select environmental factors that have a greater effect on the flora, analyze the impact of environmental factors on the flora through canonical correspondence analysis (CCA) and use the correlation heatmap based on the Spearman coefficient to analyze the diversity of environmental factors on the flora and the specific impact of sexuality and structural function. The original pictures of related SVGs were grouped and processed by PS (Adobe Photoshop CS6) and AI (Adobe Illustrator CS6).

The raw sequencing read dataset was deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive with accession No. PRJNA630993.

Analysis of the species richness and diversity indexes of the four groups of samples included the six indexes of Ace, Chao, Coverage, Shannon, Simpson, and Sobs. Table 2 shows that the species richness and diversity of the soil fungal in the roots of O. glacialis in the four regions were ranked as YBJ > CQ > ZF > ZBY. The richness index of Ace, Chao, and Sobs was the highest for YBJ and the lowest for ZBY. The Shannon diversity index is the highest for YBJ and the lowest for ZBY. The Simpson diversity index is the highest for ZBY and the lowest for YBJ; these findings are consistent with the performance of the Shannon index. The Coverage index table is between 99.85 and 99.97%; the ZBY has the highest coverage, while the YBJ has the lowest coverage.

At the genus level, one-way analysis of variance (one-way ANOVA) was used to test the significance of the difference between the four groups of differential species. Then, the post hoc test was performed for the species with differences, resulting in the top 15 different species based on relative average abundance (Figure 2D), and four samples have significant differences (P < 0.01) in Phoma, Endocarpon, Cyphellophora, Fusarium, Chaetomium, Ulocladium, Acremonium, Bradymyces, Pezizales, Sporormiaceae, Cryptococcus, Articulospora, Periconia, Dactylella, and Eurotiomyce (P < 0.01).

To further determine the flora with significant differences in different ecological types (Figure 3A), LEfSe multilevel discriminant analysis (LDA ≥ 2) was used to count from the phylum to genus levels (Figure 3B). For YBJ, approximately eight types of fungal were significantly enriched: Tremelellas to Cryptococcu, Coniochaetales, Amphisphaeriaceae, Phoma, Phaeosphaeriaceae, Dothideales to Selenophoma, Cladophialophora, and Chaetothyriales. Nine types of fungal were significantly enriched for ZF, including Hirsutella, Paraphoma, Lentitheciaceae to Darksidea, Leptosphaeriaceae, Helotiales to Tetracladium, Bradymyces, Leotiomycetes, Myxotrichaceae, and Chaetothyriaceae to Cyphellophora. Five types of fungal were significantly enriched for CQ, including Chaetomiaceae to Chaetomium, Thielavia, Fusarium, Hypocreales to Stachybotry, and Sporomiaceae, and six types of fungal were significantly enriched for ZBY, including Spizellomyces, Myrothecium, Pezizomycetes to Pezizales, Sordariales, Pleosporaceae to Curvularia, and Ulocladium. At the same time, the main groups with a large impact on the difference effect were selected from the four ecological types through LDA linear regression analysis, resulting in YBJ, ZF, CQ, and ZBY containing 16, 14, 8, and 8 groups, respectively. CQ includes Hypocreales, Fusarium, Chaetomiaceae, Chaetomium, Sporormiaceae, Sporormiaceae, Thielavia, Stachybotrys. YBJ includes Phoma, Tremellales, Cryptococcus, Tremellales, Dothioraceae, Phaeosphaeriaceae, Dothideales, Coniochaetales, Coniochaetales Coniochaetales, Chaetothyriales, Selenophoma, Chaetothyriales, Amphisphaeriaceae, Amphisphaeriaceae, Cladophialophora. ZBY includes Sordariales, Pezizales, Pleosporaceae, Pezizomycetes, Ulocladium, Spizellomyces, Myrothecium, Curvularia. ZF includes Leptosphaeriaceae, Chaetothyriaceae, Cyphellophora, Bradymyces, Helotiales, Tetracladium, Myxotrichaceae, Leotiomycetes, Paraphoma, Leotiomycetes, Leotiomycetes, Hirsutella, Lentitheciaceae, Darksidea.

The physical and chemical factors of the root soil of different samples are shown in Table 3. According to the analysis of VIF, OM, AK, AP, AN, pH, and SM were found to be significantly related to the structure of fungal community after screening the soil physical and chemical factors, which can reflect the impact on fungal community to the greatest extent.

Through principal coordinate analysis (PCoA) (Figure 5A), the first sorting axis PC1 explained 20.16%, and the second sorting axis PC2 explained 19.25%; these findings explained 39.41% of the species composition changes. The four samples showed similarities in sex and differences. While the fungal were clustered together, the CQ and ZF samples were relatively similar, and the ZBY sample was discrete from the YBJ, CQ, and ZF samples, which showed a certain degree of heterogeneity.

The CCA analysis results show that the interpretation degree of the first sorting axis (CCA1) is 16.19% (Figure 5A), and the correlation coefficients with ammonium nitrogen, available phosphorus, pH and water content are −0.9261, −0.6635, 0.8143, and −0.8943, respectively. The second sorting axis (CCA2) has an explanation degree of 11.92%, and the correlation coefficients with ammonium nitrogen, available phosphorus, pH, and water content are 0.3774, −0.7482, 0.5805, and −0.4474, respectively. In addition, the two axes explain 28.11% of the composition change in species. The fungal community was significantly correlated with pH, water content and ammonium nitrogen (P < 0.01) and significantly correlated with available phosphorus (P < 0.05). The angle between available phosphorus (Figure 5B), ammonium nitrogen and water content was acute and positively correlated, and the angle between pH was obtuse and negatively correlated. Ammonium nitrogen, available phosphorus, pH and water content determine the distribution of community species. The coefficients (R² values) were 0.8099, 0.5544, 0.8664, and 0.8292, and the order of influence on the flora was pH > pH > SM > AP > AN.

By calculating the Spearman rank correlation coefficient between the soil physical and chemical factors and the flora, the results of the heatmap diagram (Figure 5C) show that the soil physical and chemical factors have a significant impact on the fungal community, and different fungal are affected by the soil physical and chemical factors. The study revealed that soil organic matter, available phosphorus, ammonium nitrogen, and water content were positively correlated with the fungal community, while available potassium was mainly negatively correlated with the fungal microflora. We found that 12 genera were significantly positively correlated with the soil physical and chemical factors and that 13 genera were significantly negatively correlated with the soil physical and chemical factors. Additionally, soil available potassium, pH, and conductivity were significantly positively correlated with the genera of fungal belonging to Sordariomycetes and Pezizomycetes and negatively correlated with the genera of fungal belonging to Zygomycota, Chytridiomycota, and Ascomycota. Soil ammonium nitrogen, water content, available phosphorus, and OM were significantly positively correlated with the genera of fungal in Zygomycota, Chytridiomycota, and Ascomycota and negatively correlated with the genera of fungal belonging to Sordariomycetes and Pezizomycetes. pH and Spizellomycetaceae were extremely positively correlated. Spizellomycetaceae can promote the growth of plant roots, promote the accumulation and absorption of nutrients through root tissues, and increase the nutrient and organic contents in the root soil. The water content is significantly negatively correlated with Ulocladium, which can assist plants in coping with drought resistance stress. OM and Phaeosphaeria are highly positively correlated (P ≤ 0.001), and the intracellular and extracellular metabolites of Phaeosphaeri have the function of resisting pathogenic fungal in plants through spectral analysis.

There were 158 unannotated fungal in the Ascomycota, Dothideomycetes, Pleosporales, Pleosporaceae, and Alternaria species, where the SW fungal were located. We constructed phylogenetic treesfrom a total of six fungal sequences (Figure 6), including SW-producing fungal and known SW-producing fungal, and found that OTU7 (66.61%), OTU592 (13.05%), OTU253 (9.46%), OTU567 (7.78%), OTU557 (2.09%), and OTU271 (1.00%) had the potential to produce swainonine. Among them, OTU576 was similar to Embellisia grisea; OTU557, OTU7, OTU592, OTU271, and OTU253 were similar to E. Oxytropis; and OTU253 and OTU271 were only distributed in YBJ.