Study site
The experimental plot was in Xiaolingwei, Xuanwu District, Nanjing, China (32°02′37′′–32°02′39″ N, 118°49′41″–118°49′43″ E; 37 m a.s.l.). The site was comprised of flat terrain and a northern subtropical monsoon climate with distinct seasons. According to the historical records of Zhongshan Cemetery in Nanjing, the area previously contained buildings that were demolished, after which the region was covered with plantation forestry in 50–60 cm of soil. We analysed 15-year-old pure stands of Ligustrum lucidum W.T. Aiton (broad-leaf privet; family: Oleaceae) with tree spacing > 2 m and canopy density approximately 85%. The average tree height was 7.5 m, average crown was 2.5 m, and average diameter at breast height was 10.9 cm. The basic physical and chemical properties of the soil are shown in Table 1.
Table 1
Physical and chemical properties of the experimental soil used in the study
Soil layer | pH | Total carbon /g·kg− 1 | Total nitrogen /g·kg− 1 | Ammonium /mg·kg− 1 | Nitrate /mg·kg− 1 | Total phosphorus /g·kg− 1 | Total potassium /g·kg− 1 | Available phosphorus /mg·kg− 1 | Available potassium /g·kg− 1 |
0–20 cm | 7.29 | 13.12 | 1.58 | 2.50 | 1.71 | 0.40 | 1.08 | 29.02 | 0.14 |
20–40 cm | 7.31 | 11.20 | 1.42 | 1.43 | 1.69 | 0.32 | 0.95 | 23.12 | 0.13 |
Experimental design
Four adjacent trees were randomly selected as an experimental plot with 32 experimental plots (128 trees) established. According to the ‘Technical specification for the application of organic mulch on urban and rural greening’ of Shanghai, China and related literature (Dietrich et al. 2019; Zhang et al. 2020), we applied four treatments: 0, 35, 70, and 140 kg of mulch/tree in each plot, which were randomly allocated. The mulch was carefully mounded around each tree as uniformly as possible to a height of 0, 5, 10, or 20 cm above the ground (codes OM0, OM5, OM10, or OM20, respectively). The mulch extended 80 cm away from each trunk, allowing a > 0.5 m buffer between trees. The treatments were applied in November 2017. The organic mulch consisted of composted municipal green waste derived from urban gardens and was produced by Shanghai Moqi Garden Co., Ltd. (Shanghai, China). The basic physical and chemical properties of the mulch are presented in Table 2.
Table 2
Physical and chemical properties of the organic mulch used in the study
pH | Electrical conductivity /mS·cm− 1 | Organic matter /g·kg− 1 | Dry density /g·cm− 3 | Wet density /g·cm− 3 | Porosity /m3.m− 3 | Total nitrogen /g·kg− 1 | Total phosphorus /g·kg− 1 | Total potassium /g·kg− 1 |
6.40 | 1.35 | 902.00 | 0.14 | 0.79 | 318.00 | 23.80 | 4.30 | 19.50 |
Field sampling
The soil was sampled twice, after 6 and 12 months of organic mulching. During each sampling, soil was recovered from three randomly selected experimental plots (i.e., n = 3 per treatment, 12 trees) with each experimental plot used only once. The soil profiles were sampled 50 cm away from the tree trunk, and each profile was divided into two layers (0–20 cm and 20–40 cm below the mulch layer). Soil blocks of 20 × 20 × 20 cm3 were recovered. The fine roots were removed by hand. Rhizosphere soil was collected by gently shaking off the soil adhered to the roots. All fine roots and soil samples were placed in self-sealing bags and immediately transported to the laboratory for analysis. The soil samples were sieved (2 mm) and stored at 4°C until physicochemical analysis. In addition, 5–10 g of rhizosphere soil was collected from each sample, and after removing impurities such as plant roots and animal remains, the soil was placed in sterile centrifuge tubes, which were then placed in an ice box and transported to the laboratory where they were stored at -80°C for subsequent microbial sequencing.
Laboratory analysis
The physiochemical properties and enzyme activities of the rhizosphere soil were determined as described in our previous study (Sun et al. 2021a, b). These properties included the water content, pH, SOC, dissolved organic C (DOC), total N (TN), dissolved N (DN), ammonium, nitrate, microbial biomass C (MBC), microbial biomass N (MBN), total P (TP), enzyme (invertase, urease, peroxidase, and peroxidase) activity, and fine-root traits, namely, specific root length (SRL), specific surface area (SSA), root tissue density (RTD), fine root biomass (FRB), and fine root C and N concentrations (FRC and FRN, respectively). Additionally, the total fine root P (FRP) was detected calorimetrically following digestion (Campbell et al. 1991).
Genomic DNA was extracted from rhizosphere samples using the FastDNA Spin Kit for Soil (MP Biomedicals, Santa Ana, CA, USA). The DNA purity and concentration were detected and monitored using a Nanodrop ND-2000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). Next, 1% agarose gel electrophoresis was performed to assess the DNA quality, and the qualified DNA was stored at − 80°C for subsequent polymerase chain reaction (PCR) analysis. The V4–V5 hypervariable region fragments of the bacterial 16S ribosomal RNA gene were amplified with primers 515F (5′-GTGCCAGCMGCCGCGG-3′) and 907R (5′-CCGTCAATTCMTTTRAGTTT-3′) using a thermocycler PCR system (GeneAmp 9700; Applied Biosystems, Foster City, CA, USA) (Mohd Yusoff et al. 2013); for fungi, the primers ITSF (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS2R (5′-GCTGCGTTCTTCATCGATGC-3′) were used for amplification (Adams et al. 2013). PCR was performed in triplicate in 20-µL mixtures containing 4 µL of 5× FastPfu Buffer, 2 µL of 2.5 mM dNTPs, 0.8 µL of each primer (5 µM), 0.4 µL of FastPfu polymerase, and 10 ng of template DNA under the following conditions: 3 min of denaturation at 95°C, followed by 27 cycles (bacteria) or 30 cycles (fungi), for 30 s at 95°C, 30 s of annealing at 55°C, 45 s of elongation at 72°C, and a final extension at 72°C for 10 min. The resulting PCR products were extracted from the 2% agarose gel and further purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) and quantified using QuantiFluor™-ST (Promega, Madison, WI, USA) according to the manufacturer’s protocol. The purified amplicons were pooled at equimolar levels, and then paired-end sequenced on an Illumina MiSeq platform (San Diego, CA, USA) according to standard protocols by Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China).
Statistical analysis
Obtained raw FASTQ files were processed using Trimmomatic software for sequence quality control and filtering. FLASH software was used for stitching according to the overlap relation. After the samples were differentiated, UPARSE software (Edgar 2013) was used for operational taxonomic unit (OTU) clustering according to a 97% similarity level. The species classification annotation determined using the Silva database was compared with the RDP classifier (Pruesse et al. 2007). The diversity index was calculated using Mothur software (Schloss et al. 2009).
All statistical analyses were performed using R v.3.5.3 software (core Team 2018), and the corresponding figures were created using the ‘ggplot2’ software package in R. Linear mixed effects models were calculated using the R package ‘lme4’ to evaluate differences in rhizosphere soil bacterial and fungal diversity (Shannon index), rhizosphere soil properties, and fine-root traits among treatments, soil layer, time after organic mulching, and their interactions (Bates et al. 2015). Treatment (four levels), soil layer (two levels), time after organic mulching (two levels), and their interactions (three-way) were treated as fixed factors, whereas the sampling plots were treated as random factors. Bacterial and fungal diversity (Shannon index) between treatments were compared using one-way analyses of variance and Tukey’s pair-wise comparison tests.
We used the adonis function in the ‘vegan’ package (Oksanen et al. 2019) to perform non-parametric multivariate analysis of variance (PerMANOVA) (Anderson 2001) to determine statistical differences in the microbial community membership and abundance among soil properties and fine-root traits. Non-metric multidimensional scaling (NMDS) of the rhizosphere bacterial and fungal community composition and its influencing factors after 6 and 12 months of mulching were analysed relative to the PERMANOVA results. Species showing significant differences in bacterial and fungal community abundance among different treatments in different soil layers were analysed at the order level using a non-parametric factorial Kruskal–Wallis sum-rank test. We also performed principal component analysis based on Bray–Curtis dissimilarities in bacterial and fungal communities using the cmdscale function (Fig S4). Principal component analysis scores were used as proxies for the community composition in subsequent structural equation modelling (SEM) (de Vries et al. 2018).
SEM was used to analyse the relationships by which rhizosphere properties and fine-root traits collectively affect the microbial community diversity and composition. To simplify the modelling, we used composite variables to account for the collective effects of rhizosphere properties (temperature, water content, pH, SOC, DOC, MBC, TN, ammonium, nitrate, MBN, and TP) and fine-root traits (SRL, SSA, RTD, FRB, FRC, FRN, and FRP) according to previous studies (Grace and Bollen 2008; Xiao et al. 2019). Observed indicators for each composite were selected based on multiple regression analyses for microbial community diversity and composition and Akaike’s Information Criterion (AIC) (Grace et al. 2010). To verify the SEM normalisation and reduce normal deviation, all numerical variables in the model were standardised according to the same standard deviation (Grace et al. 2016). All possible interaction paths were pre-validated. When the overall model failed to fit, it was improved by removing meaningless direct or indirect paths. The model was evaluated and reduced based on the goodness of fit, whereas the AIC was applied to ensure optimal selection among different models. That is, the model with the lowest AIC value was selected as the final model. We implemented SEM using the piecewiseSEM package with ‘plot’ as the random effect to account for autocorrelation among split plots (Lefcheck 2016). All variables were initially tested for a normal distribution. P < 0.05 was considered as statistically significant.