The experimental site is located in Xuji village (34°04′N, 116°93′E) in Huaibei Municipality of Anhui Province, China, lying in Huang-Huai-Hai Plain. The experiment was conducted in June of 2017. In this area, a wheat-soybean rotation was being implemented as the traditional cropping system. During the last 10 years, the area has been witnessing four distinct seasons and a typical temperate monsoon climate. The annual average temperature of the region was 14.8°C, the annual mean precipitation was 830 mm, and the annual frost-free period was 202 days. Field management was carried out by following local cultivars and conventional crop practices, including fertilization, insecticide application and weed control. The soil texture was sandy loam, derived from fluvio lacustrine sediments and classified as Vertisol. The soil had a pH (H₂O) of 8.05, soil organic carbon (SOC) of 11.35 g⋅kg^–1, total N (TN) of 0.98 g⋅kg^–1, available P (AP) and K (AK) of 14.61 and 120.29 mg⋅kg^–1, respectively.

The experiment was composed of five treatments: control (CK) with chemical fertilizer, wheat crop straw addition plus chemical fertilizer (CS), low biochar addition at 7 t⋅ha^–1 plus chemical fertilizer (LB), high biochar addition at 20 t⋅ha^–1 plus chemical fertilizer (HB) and BCF. Each treatment was replicated three times on a single plot of 20 m² (4 m × 5 m) in area, set out in a randomized block design. The biochar was evenly spread on the soil surface and tilled into the soil of approximately 20 cm in June 2017. For the HB treatment, biochar was applied once at 20 t⋅ha^–1, while the biochar in LB treatment was added at 3 t⋅ha^–1 for soybean and 4 t⋅ha^–1 for wheat season. The crop straw was chopped into small pieces and then spread on the soil surface. Biochar used in this experiment was purchased from Nanjing Qinfeng Crop Straw Technology Company, China. The rice straw was firstly chopped into small pieces and then pyrolyzed in a rotational kiln at 550–650°C under oxygen-deficient conditions. After cooling, biochar samples were ground into powder, sieved (< 2 mm) and measured their physiochemical properties. The biochar has a pH (H₂O) of 10.07, organic C of 470.70 g⋅kg^–1, total N of 9.19 g⋅kg^–1, total P of 6.01 g kg^–1, total K 19.12 g kg^–1, ash of 37.4%, electric conductivity of 139.75 μs cm^–1. The BCF contained 15% of N, 15% of P₂O₅, 10% of K₂O and 40% of C contents. In the experiment, all treatments had the same amounts of total N, P and K applied. For the treatments with chemical fertilizer (CK, CS, LB, and HB), N fertilizer was applied at 112.5 kg N ha^–1 as a basal fertilizer, P and K fertilizers were applied at 112.5 kg P₂O₅ ha^–1 and 112.5 kg K₂O ha^–1, respectively. For the BCF treatment, BCF was applied at 750 kg⋅ha^–1. The management practices of applying pesticide and herbicide were consistent across all treatments.

The soil samples were collected 1 week before soybean harvest in October of 2019. For the rhizosphere soil, 6–8 individual crops were randomly selected from each plot. After gentle shaking off the loosely adhered root soil, the remaining soil (about 1–2 cm thickness) was homogenized to form one composite sample. For the bulk soil, ten cores (5 cm in diameter) adjacent to the crops were randomly collected from the plow layer soil (0–15 cm) in each plot with a drill, and then mixed uniformly to form one composite sample. The 30 composite soil samples, including 15 samples of bulk and rhizosphere, respectively, were transferred to sterile plastic bags, sealed and placed on ice to be transported to the laboratory. The soil samples were sieved through 2-mm mesh and plant residues and gravels were removed. A part of the soil samples was air-dried at room temperature to soil physicochemical analyses, the other part sample was stored at −20°C to DNA extraction within 1 week.

Soil pH was determined at a soil-to-water ratio of 1:2.5 using a pH meter (Mettler Toledo FE20, China). Soil NH₄⁺ and NO₃^– were extracted with 2 M KCl and their concentrations in the extracts were determined at 25°C by a Continuous Flow Chemical Analytical System (TRAACA-2000, Germany). SOC and TN were determined by wet digestion with K₂Cr₂O₇ oxidation and semi-Kjeldahl method, respectively. AP and AK were measured by colorimetric assay and flame photometer method, respectively.

Total DNA was extracted from 0.5 g bulk and rhizosphere soil using a PowerSoil^TM DNA Isolation Kit (MoBio Laboratories Inc., CA, United States), according to the manufacturer’s instructions. The DNA was purified through SanPrep Column PCR Product Purification Kit (Sangon Biotech Co., China). The concentration and quality of the extracted DNA were assessed using a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies, United States).

The qPCR was performed in a Roche LightCycler 480 system (Roche Diagnostics, Germany) via fluorometric monitoring with SYBR Green 1 dye. The primer pairs 338F/518R (Fierer et al., 2005) and 5.8s/ITS1F (Yao et al., 2017b) were used to quantify bacterial 16S rRNA and fungal ITS genes in all samples, respectively. Each reaction was performed in a 20 μL volume mixture containing 10 μL of 2 × Universal SYBR Green Fast qPCR Mix (ABclonal Technology, United States), 0.5 μL each of 10 μM forward and reverse primers, 1.0 μL of standard or extracted soil DNA, and 8.0 μL sterilized water. The plasmid DNA preparation was obtained from the clone with the correct insert using a Miniprep kit (Qiagen, Germantown, United States). The plasmid concentrations were measured using a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies, United States) and then 10-fold serially diluted for the standard curves. Melting curve analysis of qPCR products was conducted following each assay to confirm that specific amplification was not from primer-dimers or other impurities. Amplification efficiencies of 95–104% were obtained with R²-values over 0.98.

The community compositions of both bacteria and fungi in each plot were identified by the Centre for Genetic and Genomic Analysis, GENESKY Biotechnologies Inc. (Shanghai, China). The DNA of each soil sample was used as a template for PCR amplification. Moreover, the bacterial 16S rRNA V4-V5 and fungal ITS1 were amplified using primer pairs 515F/907R and ITSI/ITS2, respectively. Each sample was amplified in triplicates. At the same time, standard genomic DNA of the bacteria and fungi was used as a positive control. The PCR products were checked in a 1.5% agarose gel electrophoresis to assess their specificity. These were then purified using Agencourt AMpure XP PCR Purification Beads (Sangon Biotech Co., China). Purified amplicons were pooled in equimolar and paired-end sequenced (2 × 250 bp) using Illumina Miseq amplicon sequencing.

The raw sequencing data were processed and trimmed using Quantitative Insights Into Microbial Ecology (QIIME) software to remove the low-quality sequences (quality score < 20), primers, barcodes, adaptors (Caporaso et al., 2010), and chimeras were detected and removed using the UCHIME algorithm (Edgar et al., 2011). The remaining high-quality sequences were clustered into Operational Taxonomic Units (OTUs), with a 97% similarity cutoff. The representative sequences for each OTU were aligned using the Python Nearest Alignment Space Termination (PyNAST) (DeSantis et al., 2006; Yao et al., 2017a). The taxonomy of each depty phylotype was assigned using a BLAST comparison against sequences within the GenBank database. Rarefaction curves and alpha diversity indices were calculated using QIIME software based on the obtained OTUs. All sequences have been deposited in the National Center for Biotechnology Information (NCBI) Short Reads Archive database SRP307320.

Statistical analysis was performed using the SPSS Version 20.0 for Windows. The data are checked for normality and homogeneity of variance (Duncan’s test) before analysis of variance (ANOVA) and were subsequently transformed to meet the assumptions of the ANOVA analyses if necessary. Significant differences in soil properties, gene copy number, alpha diversity and relative abundances among the treatments in the bulk and rhizosphere were identified separately by One-way ANOVA followed by Duncan’s-test (p < 0.05). Pearson’s correlation was used to determine the relationships between gene copy number and soil properties. OTUs were analyzed for alpha and beta diversity for bacteria and fungi. Non-metric multidimensional scaling (NMDS) plots based on Bray-Curtis distance were generated to represent the differences in microbial community between different treatments. Redundancy analysis (RDA) was used to assay the relationships between microbial community and soil properties. The NMDS and RDA were performed using the “vegan” package in R (R Development Core Team, 2015).

Soil physicochemical properties of the treatments are shown in Table 1. The soil was alkaline with pH ranging from 8.15 to 8.20 in the bulk soil, while between 8.13 and 8.18 in the rhizosphere. In general, biochar application induces significant changes in most soil parameters (p < 0.05) except in soil pH, NH₄⁺ and NO₃^–. The amendment of biochar or BCF greatly increased the content of SOC and AP in the bulk soil as compared to CK. The concentration of TN was significantly increased with HB treatment, while there was no significant difference with other treatments compared to CK. The level of AK was higher in HB than the control, while no remarkable changes were observed with LB and BCF when compared to CK. A similar trend was observed in the rhizosphere.

The abundance of soil bacterial 16S rRNA genes in all treatments ranged from 2.19 × 10¹⁰ to 3.15 × 10¹⁰ copies g^–1 dry soil, were 19.40–31.96% higher in the rhizosphere than in bulk soil (Figure 1A). With the amendment of biochar or BCF the abundance was slightly higher than in the CK treatment, while the difference was not statistically significant (p > 0.05) regardless of the bulk or rhizosphere soil. The copy number of fungal ITS genes ranged from 3.84 × 10⁸ to 6.43 × 10⁸ copies g^–1 dry soil, were 8.71–23.56% higher in the rhizosphere than in bulk soil (Figure 1B). In the bulk soil, the amendment of biochar (LB and HB) significantly increased the fungal abundance compared to CK. In addition, LB also greatly increased rhizosphere fungal abundance. Moreover, the Pearson’s correlation analysis showed that fungal gene abundance a significant positive correlated with SOC (p < 0.01) in bulk and rhizosphere soil, while the same correlation was observed with TN (p < 0.05) in the bulk soil (Table 2).

After sequence quality control, a total of 3,298,552 bacterial sequences and 4,656,942 fungal sequences were obtained from the 30 soil samples. The number of OTUs ranged from 4,428 to 4,764 of bacteria and 479–585 OTUs of fungi at a 97% similarity (Table 3). When compared to CK, straw and biochar treatments had no significant effects on the alpha diversity, including Chao1, ACE and Shannon (p > 0.05). The dominant bacterial phyla (relative abundance > 5%) across all soil samples were Proteobacteria, Actinobacteria, Acidobacteria, Planctomycetes, Chloroflexi and Bacteroidetes, with relative abundances ranging from 26.80 to 33.95%, 11.25 to 21.43%, 9.01 to 14.58%, 8.55 to 11.56%, 6.66 to 8.95% and 4.29 to 7.63%, respectively (Figure 2A). In addition, the phyla of Firmicutes, Thaumarchaeota, Gemmatimonadetes, and Nitrospirae were less abundant (relative abundance > 0.1% but < 5%), but were still detected across all samples (data not shown). The application of biochar (LB and HB) increased the relative abundance of Actinobacteria in both bulk and rhizosphere when compared with CS. And LB treatment significantly decreased the relative abundance of Bacteroidetes compared in the bulk soil, while other main phyla did not show significant change with any treatment used. Further taxonomical classification at the family level revealed that 11 family (with relative abundance > 1%) were detected in all soil samples. Among them, the family Planctomycetaceae, Rhizobiaceae, Nitrososphaeraceae, Nocardioidaceae, Gaiellaceae, Anaerolineaceae, and Chitinophagaceae were abundant (relative abundance > 2%) in the soybean bulk and rhizosphere soil (Supplementary Figure 2). LB significantly increased the relative abundance of Rhizobiaceae but decreased the relative abundance of Chitinophagaceae in the rhizosphere.

The dominant fungal phyla (relative abundance > 5%) across the soil samples were Ascomycota, Mortierellomycota and Basidiomycota, with relative abundances ranging from 30.91 to 49.74%, 29.98 to 50.53% and 7.79 to 15.68%, respectively (Figure 2B). Compared with CK, LB treatment significantly decreased the relative abundance of Ascomycota but increased the relative abundance of Mortierellomycota in both bulk and rhizosphere soil. At the family level of soil fungi, Mortierellaceae, Nectriaceae, Ceratobasidiaceae, Bionectriaceae, and Trichocomaceae were abundant (relative abundance > 1%) in all soil samples (Supplementary Figure 3). When compared with CK, the relative abundance of Mortierellaceae was significantly increased in LB, HB and BCF for the bulk and increased in CS and LB for the rhizosphere. Moreover, LB treatment significantly increased the Chao1, ACE and Shannon indices of fungi (Table 3).

The effects of soil properties on the microbial communities were analyzed using redundancy analysis (RDA) (Figure 3). These soil variables explained 41.2% of the variety in bacterial communities, with the first two axes explained 28.6% and 12.6% of the total variation, respectively (Figure 3A). For fungal communities, axis1 and axis2 respectively explained 31.8 and 17.5% of the total variation (Figure 3B). Furthermore, soil SOC and TN contents significantly (p < 0.05) affected both soil bacterial and fungal community structure. In addition, the effect of soil AK on the fungal community structure was also significant (p < 0.05). Similar to RDA results, NMDS showed that bacterial and fungal community compositions separated clearly between biochar and chemical fertilizer regimes (Supplementary Figure 1). For the bacterial community, CK and CS clustered together and separated from biochar treatments (LB, HB, and BCF). For fungal community, CK, CS, and BCF clustered together and separated from LB and HB. This implied that bacterial and fungal community structures were significantly affected by biochar amendment. Furthermore, the community structure of fungi between the bulk soil and rhizosphere were well separated along the NMDS2.

Plenty of previous studies have investigated the effects of biochar application through pot experiments or short-term field studies while only a few reported long-term field experiments (Ameloot et al., 2013; Gul et al., 2015; Yao et al., 2017a). Most of these studies reported that biochar amendment improved soil physicochemical properties, such as pH, organic carbon and available nutrients (Atkinson et al., 2010; Lehmann et al., 2011). In a study by Yao et al. (2017a), the application of biochar was observed to significantly increase the soil pH and TN content, especially at a higher application rate (8%). It was similar to the results by Zheng et al. (2016), who reported an increase in pH, SOC and TN after a 4-year field trial with biochar application at a higher rate of 40 t⋅ha^–1. In this study, we found that the treatment with LB, HB and BCF greatly increased the content of SOC, whereas soil pH showed no significant change. This was due to an initial high pH (8.05) to which biochar amendment brought a slight increase (8.13–8.20) but had no statistical significance (p > 0.05). It was also reported that soil pH showed no significant change when biochar was applied in an alkaline field (Luo et al., 2017). As observed in this study and previous studies (Egamberdieva et al., 2016), the labile organic carbon in biochar could be directly utilized by microbes through root exudation. In addition, a significant increase in microbial carbon usage efficiency was observed in biochar-amended soil, thus, promoting the soil SOC retention (Zheng et al., 2016). The observations from current work show that TN and AK content greatly increased with HB treatment. The concentration of AP was also significantly increased with the amendment of biochar and BCF. This suggested that the sorption of N and P by biochar could be a crucial process (Yu et al., 2020). In other studies, these increases were regulated by the available nutrients of biochar and interaction of biochar with fertilizer (Ibrahim et al., 2020). Additionally, Gao et al. (2019) also indicated that biochar when applied alone or in combination with chemical fertilizers could remarkably promote the soil P availability.

Several studies have documented that biochar amendment altered the soil microbial abundance (Chen et al., 2015; Gul et al., 2015). It was also reported that the factors such as soil pH and microbial adhesion, influenced by biochar amendment, indirectly effects the microbial abundance (Lehmann et al., 2011; Cole et al., 2019). Another study also reported an increase in bacterial abundance by biochar addition, especially at a high rate (Yao et al., 2017b). Likewise, Chen et al. (2015) also suggested that bacterial 16S rRNA gene copy numbers increased by 37–60% under biochar amendment at a rate of 40 t⋅ha^–1, observed with three sites. It should be noted that, various studies showed that soil pH is a dominant factor for the change in bacterial abundance (Yao et al., 2017b; Liu et al., 2019; Zheng et al., 2019; Sun et al., 2020). However, test soil in this study was alkaline, and biochar application resulted no changes in soil pH and bacterial abundance. Similarly, Luo et al. (2017) found that biochar amendment had no significant effect on soil pH in alkaline agricultural soil. In a 2-year field experiment, Soinne et al. (2020) found that soil pH did not differ from the control after the forest residue biochar amendment and there were no differences in bacterial 16S rRNA gene copy numbers between treatments. In this study, the fungal abundance was 8.72–23.56% higher in the rhizosphere than in the bulk soil under all treatments (Figure 1B). This could be due to the divergent properties of rhizosphere, such as root exudations, microbial activity and plant absorption (Philippot et al., 2013). In addition, biochar amendment significantly increased fungal abundance compared with the control. These observations were consistent with the previous studies that showed that biochar application could promote the soil fungal growth (Steinbeiss et al., 2009; Jones et al., 2012; Luo et al., 2017). A similar result was reported by Yao et al. (2017a), who speculated that the possible explanation could be the change in soil pH and nutrient content caused by biochar amendment. In another semiarid farmland trial, a rate of 50 t⋅ha^–1 (C50) biochar addition remarkably increased the absolute and proportional abundance of fungi, which was ascribed to the increase in SOC (Luo et al., 2017). The current study also reported that biochar amendment greatly increases SOC content and the pairwise correlation analysis showed a significant positive correlation of fungal abundance and SOC (p < 0.01).

In a 3-year field research, Yao et al. (2017b) revealed that the bacterial diversity was positively correlated with biochar amendment rate when the soil pH was around 6.0. A similar result was reported in a short-term biochar amendment experiment by Hu et al. (2014) where the original soil pH was 3.7. In a number of previous studies, soil pH was frequently reported as a dominant factor effecting bacterial diversity in acidic and neutral soils (Gul et al., 2015; Yao et al., 2017b). However, the soil used in this study was alkaline (pH = 8.19), and biochar amendment resulted no change in soil pH. A similar result was observed by Luo et al. (2017) who found that biochar amended to an alkaline maize field had no impact on soil pH. Moreover, the LB treatment significantly decreased the relative abundance of Bacteroidetes compared with the control in the bulk soil, which only accounted for 4.29–7.63% of the total abundance. Bacteroidetes species exhibit copiotrophic attributes and were abundant in soil with high C availability (Fierer et al., 2007). Zheng et al. (2016) found that the relative abundances of Bacteroidetes decreased by 27–50% in poplar plantation soil compared with the cropland, which may be attributable to the reduction in the organic carbon bioavailability in the poplar plantation soil. In this study, we found that the relative abundance of Chitinophagaceae family affiliated with Bacteroidetes were significantly decreased in LB treatment compared with the control (Supplementary Figure 2). Chitinophagaceae plays an important role in organic carbon decomposition, and decrease in the relative abundance of Chitinophagaceae could be attributed to low availability of organic C in biochar amended soils (Zheng et al., 2017a). Though we observed that the concentrations of SOC were increased in the biochar amended treatments, it was likely that the available C rather than the total organic C may play a more important influence on these species, as the dissolved organic C was extremely lower in biochar amended soils compared with the control (Zheng et al., 2017a). Besides, other dominant phyla showed no significant changes with biochar amendment. Thus, the richness and diversity of bacterial community had no significant response to biochar application.

The composition and structure of soil microbial communities is complex and biochar application has been frequently reported to affect the soil microbial community (Atkinson et al., 2010; Xu et al., 2014; Zhang et al., 2017). Studies have demonstrated that a shift in soil properties after biochar amendment, such as soil pH, availability of C, N and other nutrients resulted in the change of microbial community (Lehmann et al., 2011). In this study, NMDS analysis showed that bacterial community in CS treatment clustered together with the CK, which separated from low (LB) and high (HB) biochar treatments, suggesting that biochar application significantly altered the community structure of soil bacteria (Supplementary Figure 1A). In addition, RDA analysis indicated that SOC and TN were two most important factors resulting in a change in bacterial community (Figure 3A). This result was consistent with many previous studies (Luo et al., 2017; Yao et al., 2017b). Likewise, Xu et al. (2014) indicated that TN was the key contributor for shaping the bacterial community structure. Zheng et al. (2016) also revealed that bacterial community in the biochar amended soil was positively correlated with SOC and TN. These studies implied that SOC and TN could alter the bacterial community compositions at the phylum level. Compared to CK, the relative abundance of Actinobacteria in the rhizosphere was increased by 37.35 and 33.69% under LB and HB treatments, respectively (Figure 2). It is known that Actinobacteria is a group of growing readily on the carbon-rich material and related to decompose the organic materials or substrates in biochar-amended soil (O’Neill et al., 2009; Chen et al., 2015). The increase of Actinobacteria in this study could be attributed to the biochar addition which not only adds nutrients like carbon source into the soil, but also stimulates the growth and multiplication of bacteria (Anderson et al., 2011; Lehmann et al., 2011).

In contrast, the LB treatment greatly increased the OTUs and fungal alpha diversity indices, including Chao1, ACE, and Shannon, regardless of the bulk soil or rhizosphere. This suggested that the response of different microbial groups to biochar amendment was varied, which was also consistent with previous research (Yao et al., 2017a, b). In a 133 days incubation experiment, hydrochar addition distinctly increased fungal diversity compared to the control, especially at a rate of 30 t⋅ha^–1 (Sun et al., 2020). Moreover, Zheng et al. (2016) and Cheng et al. (2019) revealed that the porous structure of biochar could provide a more suitable refuge for fungi to grow and protect them from predators. Biochar contains lower concentrations of N, P, K and other available nutrients compared to other organic matter, but could stabilize these elements in substrate ascribed to its porous structure and strong adsorption (Liu et al., 2016). On the other hand, it was suggested that fungi might have more readily available nutrients by absorbing to biochar (Meng et al., 2019).

Similar to bacteria, fungal community structure was significantly affected by biochar application (Supplementary Figure 1A). The fungal community structure was found to be distinct between the bulk soil and rhizosphere. Rhizosphere microbes could be affected by root exudations and plant absorption (Philippot et al., 2013; Wang et al., 2018). On the other hand, fungi are more sensitive to rhizosphere than bacteria (Hu et al., 2014; Yao et al., 2017a). Previous short-term experiments have shown that biochar amendment altered the fungal relative abundances at the phylum level (Chen et al., 2013; Hu et al., 2014). In this study, the dominant fungal phyla across all soil samples were Ascomycota, Mortierellomycota, and Basidiomycota. The LB treatment significantly decreased the relative abundance of Ascomycota phyla but increased the relative abundance of Mortierellaceae family when compared to the control. Moreover, it was also found that Ascomycota significantly decreased by 11% after 40 t⋅ha^–1 biochar amendment and this may be attributed to co-variations with soil pH, SOC, and C/N (Zheng et al., 2016). Their study also discovered that Mortierella was greatly enhanced by biochar amendment especially at a high application rate and was positively correlated with SOC. It was consistent with the results from our study where RDA analysis indicated that the three most important contributors to fungal community variations were SOC, AK, and TN (Figure 3B). Changes in the availability of soil nutrients caused by biochar addition could alter the soil enzyme activities (Marschner, 2003). These reasons ultimately resulted in the changes in the abundance and community composition of soil fungi.