By assessing the influence of rhizocompartment types (i.e. root, rhizosphere soil, root zone soil, and inter-shrub bulk soil) on the diversity of soil microbial communities under desert leguminous plant shrubs, and the influence of, and variations in, soil physicochemical factors in interactions among leguminous plants, soil, and microbes. Both 16S rRNA high-throughput genome sequencing and conventional soil physicochemical index determination were used to characterise the bacterial diversity and soil physicochemical properties in the rhizocompartments of two Hedysarum spp. (Hedysarum mongolicum and Hedysarum scoparium) in the Mu Us Desert. We found that all the nutrient indices (except TP and AP), values in rhizosphere soil were uniformly higher than those in root zone soil and inter-shrub bulk soil (P < 0.05). The bacterial community diversity in the root, under-shrub (rhizosphere, root zone) and inter-shrub bulk soil also have significant differences (P < 0.05). Desert leguminous plants had significant effects on hierarchical filtration and enrichment of specific soil bacterial microbiomes (P < 0.05). Root endophyte and rhizosphere soil microbiomes were mainly influenced by soil nutrients, while the bacterial communities in root zones soil and inter-shrub bulk soil were mainly influenced by soil pH and NH₄⁺-N. The rhizocompartment types of desert leguminous plants have a significant influence on the diversity of soil microbial communities. According to our findings, nitrogen-fixing rhizobia can co-exist with non-symbiotic endophytes in the roots of desert leguminous plants, and plants have a hierarchical filtering and enriching effect on beneficial microbes in soil via rhizocompartments. Soil physicochemical factors have a significant influence on the structure and composition of microbial communities in various rhizocompartments, and this influence is derived from the interactions among leguminous plants, soil, and microbes.

After excavating to the roots, we used tweezers to pick up the soil particles adhering to the roots as rhizosphere soil only for determine soil physicochemical properties. After that, the roots were dug out and slightly shaken to remove the attached large soil aggregates, and gloves were worn to collect the large soil aggregates as root zone soil, which were then put into an aseptic bag. Root samples were obtained from the secondary or tertiary branches of plant roots, and healthy and intact roots with an even thickness (5-8 cm) were removed and stored in sterile sample bags. Inter-shrub bulk soil was collected at the same sampling depth as root zone soil, 10-40 cm below inter-shrub bare soil. All the samples collected were stored in preservation boxes with dry ice, and brought back to the laboratory for subsequent treatment.

DNA was extracted from fresh 0.5 g samples using the EZNA Soil DNA Kit (Omega Bio-Tek, Norcross, GA, USA) following the manufacturer’s instructions, and stored at -80℃ for later use. After thawing on ice, extracted DNA samples were centrifuged separately and fully mixed; sample quality was determined using NanoDrop, and 30 ng DNA was used for PCR amplification. PCR amplification was performed in 25-μL reaction volumes containing 10× PCR buffer, 0.5 μL dNTPs, 1 μL of each primer, 3 μL bovine serum albumin (2 ng/μL), 12.5 μL 2× Taq Plus Master Mix, ultrapure H₂O, and 30 ng template DNA. The PCR amplification program included initial denaturation at 94°C for 5 min, followed by 28 cycles of denaturing at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 60 s, followed by completion of the PCR amplification program at 4°C. The forward primer F799 (5′-AACMGGATTAGATACCCKG-3′) and reverse primer R1193 (5′-ACGTCATCCCCACCTTCC-3′) were used to target the V5–V7 regions of 16S rRNA.  An agarose gel DNA purification kit (Axygen Biosciences, Union City, CA, USA) was used for purifying and combining PCR amplicons. After purification, PCR amplicons were mixed at an equimolar concentration, followed by paired-end sequencing using the Illumina MiSeq sequencing system (Illumina, San Diego, CA, USA) according to a standardised process.

After sequencing using the Illumina MiSeq system (Illumina), the results were stored in Fastq format. The quantitative insights into microbial ecology (QIIME) software (version 1.8; http://qiime.org/) was used to analyse original Fastq files and conduct quality control according to the following criteria:

(i) base sequences with a quality score < 20 at read tails were removed, and the window was set at 50 bp; if the mean quality score in the window was < 20, the posterior-end base sequences were discarded from the window, and reads shorter than 50 bp were removed after quality control;

(ii) paired reads were assembled into one sequence according to the overlapping relationship between paired-end reads (minimum overlapping length: 10 bp);

(iii) the maximum allowable mismatch ratio of the overlapping areas of assembled sequences was set at 0.1, and sequences failing to meet this criterion in pairs were removed;

(iv) samples were distinguished according to barcodes and primers at the head and tail ends of sequences, and the sequence directions were adjusted based on the number of mismatches allowed by barcodes;

(v) different reference databases were selected according to the type of sequencing data, using Usearch (version 8.1.1861; http://www.drive5.com/usearch/) to remove chimeras, and smaller-length tags were discarded using Mothur to acquire clean tags of high-quality sequences.

The sequences were clustered into operational taxonomic units (OTUs) using UPARSE (version 7.1; http://drive5.com/uparse/) based on a 97% sequence identity cut-off (excluding single sequences), and representative sequences and an OTU table were obtained.

After sequencing using the Illumina MiSeq system (Illumina), the results were stored in Fastq format (fastq1 and fastq2).