Fig. 1. Location of sampling points (Fig. A). Each grassland type includes a paired no-grazing and grazing site. Fig B shows a temperate desert sample site; Fig C shows a temperate grassland sample site; Fig D shows a montane meadow sample site. Fig E shows the sample layout.
2.2. Experimental design and field sampling
In July 2021, three areas (Fukang City, Mubi County and Qitai County) on the eastern section of the northern slope of Tianshan were selected, each area corresponds to one grassland type: temperate desert, temperate steppe and mountain meadow. The test sites are all located within the national fixed monitoring sites and have been fenced since 2012. The fenced areas are surrounded by spring and autumn grazing areas for sheep, and the grazing intensity is medium (0.6-1.0 sheep/hm2). At each site, paired plots were sampled, which were long-term grazing plots and nearby grazing exclusion plots. All paired plots had the same soil type and similar geographical conditions including slope, elevation and topography. Three typical sample lines were set up at 50 m intervals in each of the grazing exclusion and grazing areas, three 1 m x 1 m herbaceous samples were laid out at approximately 50 m intervals in each sample line (Total of 54 small samples). If there are shrubs in the plot, 5 additional shrub plots (10m×10m) are measured. Field vegetation collections were carried out to record the species present in each square and to determine cover, height, density and biomass by species. Plant coverage was measured by the projection method; natural heights were measured with the help of a ruler, and individual quantities (densities) of each specie were recorded with the help of statistical methods. Above ground biomass of each specie (only the green plant parts) were figured out by clipping the whole plant from the soil surface using scissors in each sampling plot and brought back to the laboratory for treatment(dried at 80°C for 24 h to constant weight). The specific plant data have all been obtained (Supplementary Table 1)
Soil samples were taken in layers of 0-5 cm and 5-10 cm depth, and each sample line was mixed separately and placed in sealed bags. Some of the samples were stored in a refrigerator at 4°C, while the remainder was dried indoors by picking out plant roots, gravel and other debris, and then ground and mixed, and stored in 1 mm and 0.25 mm sieves for indoor analysis. Soil microbial samples were collected from 0-5 and 5-10 cm soil layers in the sample plots, and each sample line was evenly mixed in a sealed bag and taken back to the laboratory in a vehicle refrigerator (-20°C).
2.3. Soil property analysis
The soil pH was determined using a ratio of 2.5:1 water to soil with a standard pH meter (Cao et al., 2017); soil water content(SWC) was determined gravimetrically by drying the soil samples (105◦C, 24 h); and bulk density(BD) was measured gravimetrically after oven-drying (105 ℃, 24 h). The soil organic carbon(SOC) was determined using the dichromate oxidation method (Walkley and Black, 1934). Total phosphorus (TP) was measured with Mo-Sb colorimetricmethod using a spectrophotometer (Lambda25 UV-visspectrometer, United States). Total nitrogen(TN) was determined using the Kjeldahl method.
2.4. DNA extraction and high-throughput sequencing
High-throughput sequencing was performed using the Illumina MiSeq PE 300 platform to analyse the soil fungal community. Total DNA was extracted from the samples using the macrogenomic DNA extraction kit; the variable region sequence of the ITS1-1F region of the fungal ITS rRNA gene was used as the target, and the fungal primers were ITS1-1F-F and ITS1-1F-R. PCR amplification was performed to obtain the PCR products; the PCR products were quantified and the library was constructed to obtain the fungal variable region base sequence information. The microbial sequencing analyses in this study were all done on the biochemical cloud platform of Shenzhen Microcomputer Technology Group Co.(https://bioincloud.tech/pipelines)
2.5. Statistics and analysis
All data are shown as mean and standard error. Data on soil physicochemical properties and microbial diversity indices were subjected to two-way variance comparisons and Pearson correlations using SPSS 26.0. Bar graphs were generated in OriginPro 2021 (originlab Corporation, USA). Principal Coordinate Analysis (PCoAs) from BrayCurtis distances were used to visualise the effects of grazing exclusion and grassland type on fungal community composition. PERMANOVA was performed in the ’vegan’ package using the ANOSIM function to test for significance of differences in community composition and plotted using the ’ggplot2’ package in R 4.2.1. Finally, we implemented structural equation modelling (SEM) based on the lavaan package (Rosseel., 2011) to assess the effects of grazing exclusion and grassland type on fungal diversity through changes in plant and soil abiotic variables, the statistical analyses of which were carried out using R version 4.2.1.
Results
3.1. Soil physicochemical properties as affected by grazing exclusion and grassland type
In the 0-5 cm soil layer (Table 1), grazing exclusion significantly increased 139.34% of SOC and 36.36% of TP in temperate desert (P < 0.05), while significantly increasing temperate steppe SWC、 SOC and mountain meadow TN (P < 0.05), but grazing exclusion significantly reduced 16.67% of BD in temperate steppe (P < 0.05). Before the grazing exclusion, soil pH and BD were significantly higher in temperate desert than in temperate steppe and montane meadow, while the differences between temperate steppe and montane meadow were not significant. C:N did not differ significantly between the three grassland types. All other soil physicochemical properties were significantly higher in temperate steppe and mountain meadow than in temperate desert. After exclusion, grazing exclusion significantly increased the differences in SWC, SOC, TN and N:P between the three grassland types, while the other soil chemistry properties were consistent with the results before exclusion. The interaction of grazing exclusion and grassland type significantly altered TN and N:P (P < 0.05; Supplementary Table 2). Soil physicochemical properties were all highly significantly influenced by different grassland types in general compared to grazing exclusion, while grazing exclusion, grassland type and the interaction between the two did not significantly alter C:N.
In the 5-10 cm soil layer, grazing exclusion significantly increased TP in temperate deserts by 34.1%, while significantly decreasing BD by 9.8% and N:P by 47.1%. In temperate steppes, none of the effects of grazing exclusion on the measured soil physicochemical properties were significant. In mountain meadows, grazing exclusion resulted in a significant increase in SWC and SOC. BD and SOC were significantly different between the three grassland types before the grazing exclusion, while C:N was not significantly different. All other soil physicochemical properties were significantly different between temperate desert and temperate grassland、 mountain meadow, while the differences between temperate steppe and mountain meadow were not significant. After grazing exclusion, C:N was significantly higher in temperate deserts than in temperate steppe and mountain meadows, while SOC and BD were not significantly different between temperate steppe and mountain meadows. The interaction of grazing exclusion and grassland type significantly altered BD, SOC, TP and C:N (P < 0.05; Supplementary Table 3).
Table 1 Soil physicochemical properties as affected by grazing exclusion and grassland type.