2.1 Soil collection
Soil was collected from an agriculture experimental field with an annual mean temperature of 15.4℃ and annual precipitation of 1106 mm (Liu et al. 2019) at Nanjing, Jiangsu Province, China (32.06°N, 118.54°E). Five 0.5×0.5 m sample plots were randomly selected, and the topsoil (0-20 cm) was collected and mixed. All the soil was sieved with 3 mm fine-structure mesh to remove plant materials and stones. Total C and N concentration of litter were determined using an elemental analyzer (Elemental Vario Micro, Germany) (Jia et al. 2015; Tian et al. 2018). Organic matter content was measured after igniting the dry samples for 5 h at 550°C (Täumer et al. 2005). Soil pH was measured using a glass electrode at a ratio of 1:2.5 (soil:water) after shaking for approximately 30 min (Dick et al. 2000). The original physical and chemical properties of soil and microbial populations are shown in Table S1.
2.2 Preparation of A. thaliana litter inoculated with and without endophytic bacteria
The EB strain was used for preprocessing the seedling of A. thaliana as follows: the EB was labeled with the streptomycin sulfate resistance gene (which make the EB strain can grow on plate containing 300 ug/ml streptomycin), incubated in 100 ml beef extract peptone medium (0.3% beef extract, 1% peptone and 0.5% NaCl) in a 250 ml bottle at 28°C with 160 rpm for 48 h. The cells were collected by centrifuging the medium at 5000 rpm for 3 min, then rinsed three times with sterile water. The 10 ml EB suspension in sterile water (OD600 = 0.536, the concentration of bacteria was 3.42 × 108 cfu/ml) was inoculated to rhizosphere of three-week old A. thaliana rhizosphere. A. thaliana inoculated with sterile water served as a control. A. thaliana litters (whole dead plant except seeds) with EB and without EB were collected in the end of the growth period (60 d) and were dried at 40°C to a constant weight for further uses.
2.3 Litterbag incubation experiment
A microcosm experiment was conducted at Nanjing University from November 2018 to May 2019 to examine the regulating effect of EB (B. cereus BCM2) on A. thaliana decomposition. The collected litter samples were placed in 15 cm diameter × 14 cm height plastic pots with 600 g experimental field soil. Nylon litterbags (0.3-mm mesh size, 5 × 10 cm filled with 1 g of litter of A. thaliana) were used for the incubation experiment to prevent litter being chewed by soil fauna (Tian et al. 2018). Litterbags were placed at the depth of 5 cm beneath the soil surface to mimic standard field decomposition condition. A treatment (T) of 60 ml BCM2 suspension with OD600 = 0.060 (the cell concentration was about 5.63×106 cfu/ml) was added to each treatment pot, to ensure a concentration (5.63×105 cfu/g) in the soil, which the concentration was a mean value obtained in the rhizosphere of A. thaliana during the growth for litter preparation (Fig. S1). The control (C) pots were inoculated with 60 ml sterile water. Soil moisture was maintained at 30-40% (v/v) and detected with a moisture meter (Ke shunda technology cot. LTD, Shenzhen, China). The culture condition was 20 ± 2°C. In total, 36 litterbags were prepared and deployed onto the treatment pots. Three litterbags were harvested each month during incubation time. And about 20 g soil samples beneath the litterbags were collected and divided into two parts, one stored in a sealed plastic bag at -20°C for high throughput sequencing and the another for laboratory analyses.
2.4 Determination of litter mass loss, C and N loss
Litter samples were rinsed gently on a 0.5 mm sieve, and oven-dried to a constant weight (40°C for 48 h) to determine mass loss. Total C and N concentrations of dried litter and soil samples were determined using an elemental analyzer, calculated as [(Mi × CNi) ‒ (Mf × CNf)] / (Mi × CNi) × 100 (Osono and Takeda 2002), where Mi and Mf are the initial and final incubated litter dry mass, respectively, and CNi and CNp are the initial and final C or N concentration (% of dry mass). The C and N addition in soil was calculated as (Sf ‒ Si)/Si × 100, where Sf and Si are the final and initial C or N concentration in soil. The positive litter C and N loss indicates net C and N release, the increase of C and N in soil indicates C and N sequestration. Total P in litter was determined by the molybdenum-antimony colorimetric method (Allen 1989; Hu et al. 2017). P content in plants was calculated as follows:
Where p is the concentration of P found in the standard curve (mg·l-1), v is volume of chromogenic liquid, ts is digested liquid constant volume was absorbed in the experiment, 10-4 converts the mg/l to a percentage, and m is the mass of plants.
2.5 Microbial biomass and enzymatic activity measurements
Soil microbial biomass was measured using the substrate-induced respiration (SIR) method (Bailey et al. 2002; Jia et al. 2015). Briefly, three samples of 1 g soil were respective placed in glasses vial (100 ml) and incubated with glucose (10 mg glucose g-1 soil dry weight) at 25°C. CO2 production was detected two times through an infrared gas analyzer (DISLab LW-B802, Shanghai Digital Experiment System R&D Center) after 1 h of glucose addition. The population of EB in the soil surrounding A. thaliana litter incubation sites was assessed with 1 g soil dilution coated on plate containing 300 ug /ml streptomycin at each harvest time. We assayed the potential activities of nine extracellular enzymes involved in litter C and nutrient cycling (Talbot et al. 2012; Tian et al. 2018), including cellobiohydrolase (CBH1 71123), β-1,4-glucosidase (BG), β-1,4-xylosidase (BX), nitrate reductase (NR), urease (URE), acid phosphatase (ACP), alkaline phosphatase (ALP), peroxidase (POD) and phenol oxidase (Phox). Detailed descriptions of the assay methods were referred to Tian et al. (2018), the detail description is available in the supplement file.
2.6 DNA extraction and Illumina MiSeq sequencing of the 16S rRNA and ITS gene
Soil genomic DNA was extracted by the Mobio Po werSoil DNA isolation kit (MoBio Laboratories, Inc., USA). The DNAs extracted from 24 soil samples which from EB and control treatments after decomposition of 1, 2, 4 and 6 months, have been sequenced by Huada Genomics Institute (BGI, Shenzhen, China). Primer set 515F (5’-GTGCCAGCMGCCGCGG-3’) and 806R (5’- GGACTACHVGGGTWTCTAAT-3’) was used to amplify the bacterial V4 region of 16S rRNA genes (Caporaso et al. 2011, Apprill et al. 2015), and primer set ITS1F (5’-CTTGGTCATTTAGAGGAAGTAA-3’) and ITS2R (5’-GCTGCGTTCTTCATCGATGC-3’) was used to amplify the Internal Transcribed Spacer 1 (ITS1) of fungi (Xiong et al., 2017), which were selected for many large-scale microbiomes sequencing. PCR was performed following previously published amplication conditions (Xiong et al., 2017). PCR products were purified by Agencourt AMPure XP magnetic beads (Beckman Coulter, Brea, CA, USA). After assessing the library quality and quantity (Agilent Bioanalyzer 2100 system, Agilent Tech Inc., Santa Clara, USA). The 16S rRNA gene and ITS gene fragments were sequenced using the Illumina Hiseq platform.
2.7 Bioinformatic analyses
The raw sequence data were assigned to each sample according to the unique barcodes, and then removed the barcodes and the primer sequences. The obtained raw of 16S rRNA and ITS sequence data were processed using the Quantitative Insights Into Microbial Ecology (QIIME) pipeline (Caporaso et al. 2010). Briefly, an over 25-bp window size was used to trim the unqualified sequences using BTRIM (Kong 2011), removing joint contamination reads (with at least a 15-bp overlap and < 0.1 mismatches). The obtained sequences were normalized to the minimum number of reads across all samples for the downstream analysis. Bacterial and fungal sequences were then independently clustered into operational taxonomic units (OTUs) at a 97% identity threshold using UPARSE (v7 .0.1090) (Edgar 2013). The chimera produced by PCR amplification was removed using UCHIME (v4.2.40). Taxonomy of bacteria was assigned to each sequence through BLASTing of the RDP (http://rdp.cme.msu.edu/) classifer (v2.2) (Cole et al. 2009) with the confidence threshold is set to 0.8, and Greengene (http://www.greengene.com), UNITE (https://github.com/downloads/qiime/its-reference-otus/its_12_11_otus.tar.gz) were database for fungi (Nilsson et al., 2018).
The α-diversity of soil bacteria and fungi, including the number of observed Chao1 richness and Shannon’s diversity index, was calculated (Dini-Andreote et al. 2014) using alpha_diversity.py script of QIIME (http://qiime.org/scripts/alpha_diversity.html). The bacterial and fungal DNA sequences of 24 soil samples have been deposited in the SRA of the NCBI database under the accession of PRJNA624026.
2.8 Activity of phosphate solubilization and phosphatase production of EB
The original EB solubilizing activity of organophosphorus and inorganic phosphorus was assessed after litter decomposition using the dissolved phosphorus cycle method with modified National Botanical Research Institute's phosphate growth medium (NBRIY) (Nautiyal 1999; Jorquera et al. 2008) containing glucose, 10 g/l; Ca3(PO4)2, 5 g/l; (NH4)2SO4, 0.3 g/l; NaCl, 0.2 g/l; MgSO4·7H2O, 0.3 g/l; KCl, 0.2 g/l; MnSO4·H2O, 0.002 g/l; FeSO4·7H2O, 0.002 g/l and agar, 20 g/l. To detect the ability of organophosphorus dissolution, the EB strain was incubated using basic Pikovskya's (PVK) agar medium with lecithin: glucose, 10 g/l; (NH4)2SO4, 0.3 g/l; lecithin 1 g/l; NaCl, 0.2 g/l; MgSO4·7H2O, 0.3 g/l; KCl, 0.2 g; MnSO4·H2O, 0.002 g/l; FeSO4·7H2O, 0.002 g/l; yeast extract, 0.4 g/l and agar, 15 g/l). Petri plates were incubated at 28°C for 120 h, and the phosphorus hydrolysis circle were recorded.
The alkaline and acid phosphatase production capacity of the EB strain was detected as follows: bacteria were incubated in beef extract peptone medium at 28°C for 48 h. Then, aliquots of each sample were added to 0.48 ml of 0.1 M universal buffer with pH 6.5 or pH 11, respectively, for acid and alkaline phosphatase activity, and 0.12 ml p-nitrophenyl phosphate (pNPP) 0.05 M solution, followed by 1 h incubation at 37°C. A control treatment containing only liquid medium was included in each experiment with pNPP added after incubation. The yellow color was detected at 410 nm (Tabatai and Bremmer 1969; Oliveira et al. 2009). In order to detect phosphatase activity during the litter decomposition, 0.2 g of A. thaliana litter was added into 100 ml de-nutrition beef extract peptone medium (ten times dilution) incubated at 28°C for 48 h. The control added 0.2 g A. thaliana litter not infected with BCM2. Then, alkaline and acid phosphatase production activity was detected as above.
2.9 Available phosphorus in soil during litter decomposition and total P in plants
Available phosphorus (A-P) in harvest soil was determined using the molybdate colorimetric method after ascorbic acid reduction (Allen 1989). A 2 g sample of air-dried soil passed through a 20-mesh sieve (bore diameter 850 μm) was used to detect available phosphorus detection. Then, 50 ml of 0.5 mol/l NaHCO3 solution was added to each sample and agitated, shaken for 30 min, and filtered with phosphorus-free filter paper. We pipetted 10 ml of filtrate, added 35 ml of distilled water to dilute the sample, and then added 5 ml of molybdenum antimony anti-chromogenic agent. The solution was mixed well and allowed to stand for 30 min, and the blue color was measured at 880 nm. Total P was determined by the molybdenum-antimony colorimetric method after the samples were digested with H2SO4 and H2O2 (Allen 1989; Hu et al. 2017). The 5 ml digestion was diluted to 50 ml, first adding 2 drops of 2, 4-dinitrophenol indicator, and then 6 mol NaOH was used to neutralize the digested liquid to just yellow. We then added 1 drop of 2 mol H2SO4 to fade to yellow, and added 5 ml of molybdenum-antimony anti-chromogenic agent. The solution was mixed well and allowed to stand for 30 min, the blue color was measured at 880 nm. The content of available phosphorus in soil was determined using the following equation:
Where is the concentration of P found in the standard curve (μg·ml-1), v is the volume of constant volume during color rendering, ts is the fraction multiple (total volume of extraction extract absorbed in the experiment), m is the mass of air-dried soil, k is conversion coefficient from air-dried soil to drying soil, 103 is to convert the μg to mg, 1000 is to convert the concentration to P per kilogram.
2.10. ‘After-life effect’ on wheat growth by litter with EB decomposition
To study litter decomposition feedbacks to subsequent plant growth in the ecosystem, 5 wheat (Triticum aestivum L.) seedlings were planted into each soil after litter incubation. Soil moisture was maintained at 30-40%, and temperature at 20 ± 2°C. Fresh and dry weight of wheat were measured after each month. The total P content of wheat after one month as also determined by the molybdenum-antimony colorimetric method (Allen 1989; Hu et al. 2017).
2.11 Data analysis
The constant of potential mass loss rate over time was determined using an exponential equation (Olson 1963):
Where xo is the original mass of litter, xt is the amount of litter remaining after time t (month), and k is the litter decomposition coefficient (month-1).
We used independent sample t tests to assess differences between the treatment and the control group in this study. The correlation of EB richness in soil and SIR was based on the Pearson's product-moment correlation. The effect of A-P on growth parameters of wheat was evaluated using semi-parametric permutational multivariate ANOVA (PERMANOVA,) (Anderson 2001) using the ‘vegan’ package in R. The variation in the bacterial and fungal community caused by six-month decomposition and EB treatment was investigated using Principal coordinate analysis (PCoA) based on Bray–Curtis dissimilarities using QIIME (Caporaso et al. 2010). To detect the effect of EB on the soil microbial community, SIR, soil enzyme activity, litter decomposition, A-P in soil and subsequent plant P content, and minimized the confounding interactions among causal factors, a structural equation model (SEM) was implemented to further reveal the possible pathways and interactions between the factors through a priori modeling (Fig. S3), using R package “lavaan” (Rosseel 2012). Each pathway in the model was evaluated for significant contributions to the model. Indices of model fit were the chi-square test (a lower chi-square indicates a better model), with P traditionally > 0.05, the root mean square error of approximation (RMSEA; the model has a good fit when RMSEA <0.05) and the 90% confidence intervals (Oberski et al. 2014).