Sample collection and dataset information
Details on soil sampling, vegetation survey, measurement of soil properties, sequencing and bioinformatics were described in our previous study (Yang et al. 2017; Shi et al. 2019). Briefly, in the growing season of 2011, 180 soil samples were collected across the north-eastern and central Tibetan Plateau.
Within each of the sites, three of the ten small plots - each 1×1m in size - were randomly chosen on the diagonals of the 1-ha site at least 40 m apart (Fig S1). In each 1×1m plot, seven randomly located soil cores of the top 5 cm of topsoil (under any litter layer) were collected and homogenized as a composite soil sample. All vascular plant species and the number of individuals of each were recorded, at each of the 180 plots, and summarized at the site level (60 sites), and plant species richness was calculated.
All soil properties, including pH, soil moisture (SM), total carbon, total nitrogen, soil total phosphorus (STP), soil organic carbon (SOC), were measured at the individual plot level following (Jing et al.2015), and then averaged at the site level. As three small plots were merged from each site, this gave 60 sites samples in total.
Soil DNA from each plot was extracted and amplified in the Illumina MiSeq platform PE250 (Illumina Inc., San Diego, CA, USA). The primers ITS3 (50-GCATCGATGAAGAACGCAGC)/ITS4 (50-TCCTCCGCTTATTGATATGC) were used to target the fungal internal transcribed spacer 2 (ITS2) region. The bioinformatic pipeline on ITS2 sequencing data was identical to the description of Teng et al (Yang et al. 2017). The primer set of F515: 5’-GTGCCAGCMGCCGCGG, R907: 5’-CCGTCAATTCMTTTRAGTTT (Lane et al. 1985) were used to amplify the V4-V5 hyper-variable regions of bacterial 16S rRNA gene. The high-throughput sequencing process and analysis of soil fungi and bacteria has been described already by Teng et al (Yang et al. 2017) and Shi et al (Shi et al. 2019). In order to be compatible with plant and soil data, we merged the OTU table at the site level. Finally, 11,576,489 fungal sequences (min=123,753 sequences per site) and 1,031,092 bacterial sequences (min=9,037 sequences per site) were obtained in the 60 sites. In order to analyze the microbial data at the same sequencing depth, the OTU matrices were rarefied to 123,753 sequences per sample for fungi. But for bacteria, most sites were rarefied to 12850 sequences, only two sites far less than 12850 sequences were rarefied to 9037 reads, to keep as many reads as possible without compromising the bacterial alpha diversity. The richness of bacteria and fungi species were then calculated by “countif” function in Excel.