1 Introduction

The transformation and fate of nitrate are of particular importance in paddy fields, given that they are related to nitrogen (N) loss, nitrous oxide (N2O) production, and N pollution (Kögel-Knabner et al. 2010), as well as N use efficiency (Lassaletta et al. 2016). The frequent dry–wet cycles in paddy fields enable the supply of nitrate by favoring the nitrate producing process (i.e., nitrification), and the extended periods of waterlogging during the entire growth stage of rice create a unique condition for dissimilatory nitrate reduction processes [e.g., denitrification, anaerobic ammonium oxidation (anammox), and dissimilatory nitrate reduction to ammonium (DNRA)] (Kraft et al. 2014; Kuypers et al. 2018; Wang et al. 2019). By converting nitrate into inert N2, denitrification and anammox play significant roles in removing nitrate and contribute to 57.36‒99.23% and 0.15‒40.78% of the total nitrate removal in paddy soils, respectively (Li et al. 2020; Shan et al. 2016; Yang et al. 2015b). However, DNRA transforms loss-prone nitrate into soil-retainable ammonium, alleviating N loss by competing with denitrification and anammox in paddy soils (Pandey et al. 2019; Putz et al. 2018). The release of N2O, a potent greenhouse gas and ozone depleter, is also intimately related to nitrate reduction processes since both denitrification and DNRA are capable of producing N2O (Giles et al. 2012; Kraft et al. 2011).

The nitrate reduction processes and the partitioning of nitrate among denitrification, anammox, and DNRA are susceptible to agricultural practices such as fertilization and straw returning (Abbas et al. 2019; Pandey et al. 2019; Zhang et al. 2021). It has been reported that the rates of denitrification, anammox, and DNRA in paddy soils exhibit high variation and are mainly determined by organic carbon (OC) availability, N contents, and soil pH (Li et al. 2020; Shan et al. 2016; Zhang et al. 2021). The ratio of OC to nitrate is the key controlling factor for nitrate partitioning between denitrification and DNRA, in which a high OC/NO3− ratio generally favors nitrate partitioning toward DNRA process (Hardison et al. 2015; Kraft et al. 2014; Rahman et al. 2019a). The share of DNRA among nitrate reduction processes is commonly enhanced following elevated pH, due to DNRA exhibiting optimum activities at pH 7.0‒8.0, while the optimal pH for denitrification was around 6.0‒7.0 (Kim et al. 2017). Nevertheless, information on the partitioning of nitrate among different reduction processes as affected by agricultural practices is still limited in paddy soils, which hinders a better understanding of N regulation in paddy fields.

Biochar, as a carbonaceous material produced from the pyrolysis of different feedstocks (e.g., crop residues, woody biomass, and animal wastes), has been proposed as a useful soil amendment in terms of mitigating greenhouse gas emissions (Kuppusamy et al. 2016; Lehmann et al. 2021; Lévesque et al. 2020; Wu et al. 2021a) and improving soil fertility (Novak et al. 2009) in agroecosystems. In most contexts of biochar research, biochar amendment can significantly reduce N2O emissions (Dong et al. 2020; Wu et al. 2021b; Zhang et al. 2013). In upland soils, several mechanisms have been proposed for the reduced N2O emissions caused by biochar, showing that biochar may sorb N2O (Cornelissen et al. 2013; Van Zwieten et al. 2014), limit soil nitrification by binding NH4+ (Yang et al. 2015a), inhibit soil denitrification by enhancing soil aeration (Van Zwieten et al. 2010), increase soil pH or act as an “electron shuttle’’ facilitating the last step of denitrification (Cayuela et al. 2013; Yanai et al. 2007), as well as modify nitrifier and denitrifier communities (Dai et al. 2021; Xu et al. 2014). Similarly, the effect of biochar on reducing N2O emissions has also been reported in paddy soils (Awad et al. 2018; Borchard et al. 2019); however, paddy soils are different from upland soils in terms of drying-wetting alternation and flooding conditions during the rice growth stage, which can facilitate complete denitrification. Additionally, few studies have directly evaluated the effect of biochar on the reduction of N2O to N2 (Cayuela et al. 2013; Fungo et al. 2019; Yuan et al. 2019). Thus, we hypothesized that the effect of biochar on reducing N2O emissions in paddy fields is more profound by facilitating the last step of denitrification (i.e., the reduction of N2O to N2).

Previous studies related to the effects of biochar on nitrate reduction processes mainly focused on denitrification; however, competition among denitrification, anammox, and DNRA can be expected in paddy fields (Li et al. 2020; Shan et al. 2016). Thus, it is necessary to simultaneously investigate the three nitrate reduction processes in paddy soils that received biochar. Indeed, increases in pH, C availability, and N mineralization have been extensively reported in biochar-amended soils (Awad et al. 2018; Nguyen et al. 2017; Oladele 2019; Yin et al. 2021), which  imply that denitrification, anammox, and DNRA processes could be altered in biochar-amended paddy soils. Two studies have evaluated the effect of biochar on anammox in paddy soils, suggesting that biochar amendment significantly stimulated anammox rates by acting as an electron shuttle (Pan et al. 2017; Zhou et al. 2016). Until now, no publication can be found about the influence of biochar on the DNRA process in paddy soils, and the role of biochar amendment in nitrate partitioning among denitrification, anammox, and DNRA in paddy fields is still poorly understood. On the other hand, the influences of biochar amendment on either abundances or community compositions of denitrifiers (i.e., nirK-, nirS- or nosZ-encoding microbes) have been well documented (Zhang et al. 2021; Wang et al. 2020; Shi et al. 2019); whereas, information about effects of biochar amendment on DNRA bacteria was limited. Moreover, it was found that nirK- and nirS-encoding denitrifiers may work cooperatively in soils under organic management (Han et al. 2020), and the community of DNRA bacteria could be strongly shaped by long-term N loading and straw return regimes (Zhang et al. 2021). Consequently, we hypothesized that the abundances and community compositions of denitrifying and DNRA bacteria, as well as their interaction patterns, may be altered by biochar amendment in paddy soils.

In the present study, using membrane inlet mass spectrometry (MIMS) and the N2/Ar technique integrated with the 15N tracer method, we simultaneously investigated the potential rates of denitrification, anammox, and DNRA, and determined N2O emission rates in a paddy soil receiving varying doses of biochar (0%, 0.03%, 0.1%, 0.5%, and 1.0%; w/w) over six consecutive years. To compare the difference between the effects of straw and its biochar at the same dose (0.1%, w/w) on the aforementioned processes, the soil consecutively amended with straw was also collected. Additionally, the abundance and microbial community of nitrate reducers were also investigated by qPCR and high-throughput sequencing. The objectives were to (1) investigate the effects of biochar amendment on rates of denitrification, anammox, and DNRA and their relative contributions to the total nitrate reduction, as well as the N2O emission rates; (2) explore the influence of biochar amendment on the abundances and community compositions of nitrate reducer; (3) elucidate the critical factors and microbial mechanisms behind biochar amendment effects.

2 Materials and methods

2.1 Soil sampling

The soil samples were collected from a long-term rice–wheat field trial following consecutive biochar or straw amendment, which was performed at the Yixing Base of Changshu National Agro-Ecosystem Observation and Research Station of the Chinese Academy of Sciences in Jiangsu Province (31° 37′ N, 120° 03′ E), China. The collected soil belongs to gleyic hydragric Anthrosol soil, which developed from a silty loam deposit. Specifically, six treatments were arranged in a fully randomized design as follows: (1) non-biochar or straw amendment (Control); (2) 0.75 Mg ha−1 biochar amendment [equivalent to a biochar amendment rate of 0.03% to soil (calculated by the weight of the 0–20 cm plough layer, w/w), 0.03%BC]; (3) 2.25 Mg ha−1 biochar amendment (0.1%BC); (4) 11.3 Mg ha−1 biochar amendment (0.5%BC); (5) 22.5 Mg ha−1 biochar amendment (1.0%BC); and (6) 2.25 Mg ha−1 crop-straw amendment [equivalent to a straw amendment rate of 0.1% to soil (w/w), 0.1%ST]. Biochar used in the present study was pyrolysed in a muffle furnace at the temperature of 500 °C (with a heating rate of 5 °C min–1) for 8 h under oxygen-limited conditions. Detailed information on the biochar properties was listed in Additional file 1: Table S1. Based on the local fertilization regime, each plot of the above treatments annually received approximately 450 kg urea-N ha−1, with 250 and 200 kg N applied for the rice and wheat seasons, respectively. All soil samples (0–20 cm) were composites of five cores randomly collected from each plot (i.e., three replicates of six treatments) during the maturity period of rice, in which the fields were characterized by the typical flooding conditions and represented the main stage of rice growth. Briefly, all soil samples were then placed in sterile plastic bags with ice packs surrounded, and immediately transported to the laboratory. The soil samples were divided into three subsamples: one subsample was air-dried and sieved (< 2 mm) for the analysis of physicochemical properties, the second subsample was briefly stored at 4 °C and then processed for soil slurry incubation experiments, and the third subsample was stored at − 80 °C for DNA extraction and the following molecular analysis. The detailed methods for the analysis of the characteristics of the soil samples were described in the Supplementary Material.

2.2 Denitrification, anammox, and DNRA rates

The potential rates of denitrification, anammox, and DNRA were determined by the 15N tracer technique combined with MIMS (Bay Instruments, Easton, MD, USA) measurement of 29N2 and 30N2 in the soil slurry (Shan et al. 2016). Briefly, soil slurries were prepared by thoroughly mixing the fresh soil with helium (He)-purged water at a ratio of 1:5 (soil/water) in 12 ml glass transparent vials (Exetainer; Labco, UK). To eliminate the residual nitrate and dissolved oxygen in the soil slurries, the vials were preincubated for one week in a vertical shaker at 25 °C. Thereafter, the vials were spiked with 100 μL 15NO3− He-purged stock solution via a microsyringe, resulting in a targeted concentration of 100 μM 15 N in each vial. Subsequently, all the vials were divided evenly into two batches, with one batch for determining the potential rates of denitrification and anammox and another batch for determining the potential rates of DNRA in the soil slurry (See detailed information in the Supplementary Material).

2.3 N2O emission rates

Briefly, soil slurries were prepared in 25 ml glass transparent vials without adding 15NO3− solution as described above. Then, all the vials were incubated at 25 ℃, and the incubation was stopped by the addition of 200 μL of saturated HgCl2 solution at 1, 3, and 8 h after incubation. Subsequently, all the vials were centrifuged, and 5 mL of the supernatants were transferred into 18.5 mL evacuated serum vials to measure the dissolved N2O concentration in soil slurries via gas chromatography (Agilent 7890B, Santa Clara, CA, USA) equipped with an electron capture detector. The emission rates of N2O over the time course of the incubations were estimated based on the linear regression of the dissolved N2O concentration during the incubation using the following equation:

$$R_{{N_{2} O}} = \frac{{K \times V}}{W}$$

where RN2O (nmol N g−1 h−1) is the N2O emission rate, K is the slope calculated from the dissolved N2O concentration versus incubation time (R2 > 0.80), V (L) is the volume of the vial, and W (g) is the dry weight of the soil.

Notably, net N2 emission rates were calculated by the reported relationship between total N2 emission rates (i.e., sum of denitrification and anammox rates determined by the laboratory slurry-based 15 N tracer technique) and net N2 flux (directly determined by the soil core incubation-based N2/Ar technique) in typical paddy soils (Shan et al. 2016). Accordingly, the N2O emission rates and net N2 emission rates were used to calculate the product stoichiometry of denitrification [i.e., N2O/(N2O + N2) ratio] in this study.

2.4 DNA extraction and quantitative PCR assay

Total DNA was extracted from 0.5 g freeze-dried soils using a FastDNA Spin Kit for Soil (MP Biomedicals, Cleveland, OH, USA) according to the manufacturer’s instructions. The quantity and purity of the extracted DNA were determined with a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). To determine the abundances of nitrate reduction-related genes (narG, nirK, nirS, nosZ Clade I, napA, and nrfA), real-time quantitative PCR was performed with the extracted DNA. The primers for the nitrate reduction-related genes were 1060F/2050m2R for narG, cd3aF/R3cd for nirS, F1aCuF/R3CuR for nirK, 2F/2R for nosZ Clade I, 3F/3R for napA, and F2aw/R1 for nrfA. The gene copy numbers of these genes were determined using a CFX96™ Real-Time System (Bio-Rad). The detailed primer information and the thermal cycling conditions for the aforementioned genes were described in Additional file 1: Table S2. The standard curves for these functional genes were generated using a tenfold dilution series (102–109 copies) of standard plasmid DNA. All qPCR reactions were performed in triplicate for each sample, and the amplification efficiencies were 96.4‒110.3%.

2.5 High-throughput sequencing and network analysis

High-throughput sequencing libraries targeting the nirK, nirS (denitrification related), and nrfA (DNRA related) genes were built using DNA Library Prep kits (NEB Next, Ultra, USA). The high-throughput sequencing analysis of the abovementioned genes was conducted on an Illumina MiSeq platform (Illumina, San Diego, CA, USA) according to the standard protocols of Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). The raw fastq files were quality-filtered using Trimmomatic and merged by FLASH. The sequencing data were then filtered based on their qualities using Quantitative Insights into Microbial Ecology (QIIME) (Caporaso et al. 2010). The operational taxonomic units (OTUs) were clustered at 97% similarity using UPARSE version 7.1 (Edgar 2013), and the chimeric sequences were identified and removed. The identification and selection of the representative sequence for each OTU were conducted according to the published nirK, nirS, and nrfA sequences in the Fungene database (http://fungene.cme.msu.edu/). The raw sequencing reads of the nirK, nirS, and nrfA genes were deposited in the NCBI Sequence Read Archive (SRA) database under accession numbers PRJNA805174, PRJNA805183, and PRJNA805297, respectively. The relative abundance of taxa was calculated from the normalized OTU table after rarefying all samples to the same sequencing depth. Specifically, the relative abundance of specific orders encoded by nirK, nirS, or nrfA genes was calculated based on the composition of nitrate reducer communities (i.e., the numbers of OTUs for the respective orders) in each treatment.

Co-occurrence network analysis was conducted based on SparCC (Friedman and Alm 2012) using the ‘sparccboot’ function from the SpiecEasi package in R (Kurtz et al. 2015). The OTU table for each gene (nirK, nirS, and nrfA) was filtered with two constraints to improve the reliability of the network: only OTUs that appeared in at least > 50% of all replicates for the corresponding treatment and only OTUs with average relative abundances > 0.01% were kept. Only those OTUs with significant correlations (both positive and negative, with P < 0.01, correlation <  − 0.60 or > 0.60) by bootstrapping (100 times) were included in the networks. Prior to constructing the network, the reliability of the association was tested (Additional file 1: Fig. S3) using the method described by Cougoul et al. (2019). We created a separate network for each treatment (Control, 0.03%BC, 0.1%BC, 0.5%BC, 1.0%BC, and 0.1%ST), and the properties and images of the network were calculated and generated using the igraph package in R (Csárdi and Nepusz 2006) and Gephi (http://gephi.github.io/). To calculate the P values for the network statistics, we performed 10,000 permutations of the networks.

The role of each node within the co-occurrence network was estimated by two topological parameters: within-module connectivity (Zi), which quantified to what extent a node was connected to other nodes in its own module, and among-module connectivity (Pi), which quantified how well the node was connected to different modules (Guimerà and Nunes Amaral 2005). The nodes with either a high Zi or Pi value were defined as keystone taxa, including module hubs (Zi ≥ 2.5, Pi < 0.62), connectors (Zi < 2.5, Pi ≥ 0.62), and network hubs (Zi ≥ 2.5, Pi ≥ 0.62) (Olesen et al. 2007).

2.6 Data analysis

One-way analysis of variance (ANOVA) followed by the Duncan test was performed to evaluate the differences in N transformation rates (denitrification, anammox, DNRA, and N2O emission) among different treatments (at P < 0.05). Pearson correlations were conducted using R software (version 4.0.3; R Development Core Team 2008) to evaluate the correlations among the N transformation rates, soil physicochemical properties, and the abundances and community compositions of nitrate reducers or the relative abundance of keystone orders in the respective networks. The SEM analysis was performed with the lavaan package in R (Rosseel 2012). The Chi-square test (χ2), comparative fit index (CFI), comparative fit index (CFI), standardized root mean square residual (SRMR), and root mean square error of approximation (RMSEA) were used to assess the model fit.

3 Results

3.1 Soil characteristics

Biochar amendment over six years significantly altered the soil properties. The biochar amendment significantly (P < 0.05) increased the soil pH by 4.27–24.36%, as suggested by a range of 5.8‒7.0 in the biochar-amended soils compared to those in non-amended (5.6 ± 0.02, Control treatment) and straw amended (6.0 ± 0.02, 0.1%ST treatment) soils (Table 1). The ammonium (NH4+) and nitrate (NO3−) contents in the studied soils varied from 5.2 to 5.9 and 7.7 to 15.5 mg N kg−1 dry soil, respectively. Corresponding to the change in soil pH, a significant decrease (P < 0.05) in the soil NH4+ content was observed among the biochar-amended treatments except for the 1.0%BC treatment. The total N (TN) content ranged from 1.4 to 2.5 g kg−1 among different treatments, which resulted in total carbon/total nitrogen (TC/TN) ratios ranging from 8.2% to 16.7%. The soil organic carbon (SOC) content ranged from 11.3 to 43.5 g C kg−1 among different treatments, and the EOC content varied from 126.9 to 183.3 mg C kg−1, both of which significantly (P < 0.05) increased with the increasing amendment rates of biochar. Similarly, straw amendment in the 0.1%ST treatment also caused a significant (P < 0.05) increase in the SOC and EOC contents as compared to the non-amended Control treatment (Table 1).

Table 1 Physicochemical characteristics of soil in the Control, 0.03%BC, 0.5%BC, 1.0%BC, and 0.1%ST treatments
Full size table

3.2 N transformation rates

Among the biochar-amended soils,  the denitrification rate significantly (P < 0.05) increased by 46.03%, 46.73%, and 10.19%, respectively in soils that received low amendment rates of biochar (0.03%BC, 0.1%BC, and 0.5%BC treatments) relative to that in non-amended soil (25.40 ± 1.28 nmol N g−1 h−1, Control treatment) (Fig. 1a). However, a significant suppression effect on denitrification rate was observed when the amendment rate of biochar reached 1.0% (Fig. 1a). In addition, the denitrification rate in the 0.1%ST treatment was 44.69 ± 0.19 nmol N g−1 h−1, which was significantly (P < 0.05) higher than that in the 0.1%BC treatment (36.51 ± 1.91 nmol N g−1 h−1). The anammox rate was 0.06 ± 0.03 nmol N g−1 h−1 in the Control treatment, which was significantly (P < 0.05) increased by 2.90, 5.14, and 15.63 times when the biochar amendment rates were 0.03%, 0.5%, and 1.0%, respectively (Fig. 1a). Notably, anammox rates in the 0.1%BC and 0.1%ST treatments showed negative values in the present study.

Fig. 1
figure 1

Denitrification and anammox rates (a); DNRA rates (b); DNRA/(Den + DNRA) ratios (c); and N2O emissions rates (d) in the Control, 0.03%BC, 0.1%BC, 0.5%BC, 1.0%BC, and 0.1%ST treatments. Error bars show the standard deviation of each treatment (n = 3). Means denoted by a different letter differ significantly among different treatments according to one-way ANOVA (P < 0.05). “n.v” means that the anammox rates in the 0.1%BC and 0.1%ST treatments were negative values

Full size image

The DNRA rates among the different treatments varied from 0.42 to 3.26 nmol N g−1 h−1, with the highest value (3.26 ± 0.30 nmol N g−1 h−1) observed in the 1.0%BC treatment (Fig. 1b). With the increase of amendment rate, biochar amendment significantly (P < 0.05) increased DNRA rates (Fig. 1b). The DNRA rate in the straw-amended soil (0.1%ST treatment) was 2.76 ± 0.30 nmol N g−1 h−1, which was 5.65 times significantly (P < 0.05) higher than that of the Control treatment. In addition, the DNRA rate in the 0.1%ST treatment was also significantly (P < 0.05) higher than that in the 0.1%BC treatment (2.12 ± 0.12 nmol N g−1 h−1) (Fig. 1b).

Based on rates of nitrate reduction processes, the DNRA/(denitrification + DNRA) [DNRA/(Den + DNRA)] ratios were also calculated in the present study (Fig. 1c), showing that DNRA accounted for 2.85‒13.18% of nitrate partitioning between denitrification and DNRA in the biochar-amended treatments. Biochar amendment significantly increased the share of DNRA between denitrification and DNRA (Fig. 1c). Overall, the DNRA/(Den + DNRA) ratio in the 0.03%BC, 0.1%BC, 0.5%BC, 1.0%BC, and 0.1%ST treatments was 0.77, 2.41, 4.18, 7.54, and 2.62 times higher than that in the Control treatment, respectively (Fig. 1c).

Along with the altered nitrate reduction processes, N2O emissions were also significantly affected by the amendment of biochar. The N2O emission rate was 0.52 ± 0.04 nmol N g−1 h−1 in the Control treatment, while it was significantly (P < 0.05) suppressed by 97.65%, 80.52%, 61.86%, and 89.46% in the 0.03%BC, 0.1%BC, 0.5%BC, and 1.0%BC treatments, respectively (Fig. 1d). Relative to the 0.1%ST treatment (0.13 ± 0.03 nmol N g−1 h−1), biochar amendment (0.1%BC treatment) decreased N2O emission rate by 19.83%. Based on the calculated net N2 emission rates (Additional file 1: Fig. S1a), the N2O/(N2O + N2) ratios were estimated, showing that the N2O/(N2O + N2) ratios ranged from 0.01% to 0.61% among biochar-amended treatments (Additional file 1: Fig. S1b). Similar to the N2O emission rates, the N2O/(N2O + N2) ratios were significantly (P < 0.05) decreased by 65.29‒98.39% following biochar amendment, relative to the Control treatment (Additional file 1: Fig. S1b).

3.3 Abundances of nitrate reduction-related genes

The abundances of narG, nirS, nirK, nosZ Clade I, napA, and nrfA genes ranged from 8.99 \(\times\) 106 to 2.11 \(\times\) 107, 1.24 \(\times\) 107 to 2.31 \(\times\) 107, 8.50 \(\times\) 106 to 1.25 \(\times\) 107, 1.17 \(\times\) 106 to 2.94 \(\times\) 106, 7.40 \(\times\) 106 to 1.89 \(\times\) 107, and 6.04 \(\times\) 106 to 1.63 \(\times\) 107 copies g−1 dry soil in all treatments, respectively (Fig. 2). There was a clear increase in the abundances of the nirS, nrfA, and napA genes in the biochar-amended treatments compared to those in the Control treatment; however, significant changes were only detected in the 1.0%BC treatment (Fig. 2b and d), in which the abundances of the napA and nrfA genes were 0.52‒1.70- and 0‒1.53-fold higher, respectively than those in the Control treatment. Interestingly, in comparison with the Control treatment, straw amendment significantly (P < 0.05) increased the nosZ Clade I gene abundance by 146.05%, whereas no significant changes were detected among the biochar-amended treatments. Additionally, compared with gene abundances in the 0.1%ST treatment, all nitrate reduction-related genes except the nirK gene were decreased by 0.37‒44.35% in the 0.1%BC treatment (Fig. 2).

Fig. 2
figure 2

Nitrate reduction-related gene [narG (a), napA (b), nrfA (c), nirS (d), nirK (e), and nosZ Clade I (f)] abundance in the Control, 0.03%BC, 0.1%BC, 0.5%BC, 1.0%BC, and 0.1%ST treatments. Error bars show the standard deviation of each treatment (n = 3). Different letters indicate significant differences at P < 0.05

Full size image

3.4 Diversity and composition of nitrate reducer community

Unconstrained principal coordinates analysis (PCoA) revealed that the composition of microbiota involved in nitrate reduction (i.e., nrfA-, nirK-, and nirS-encoding bacteria) in the biochar-amended treatments and the Control treatment differed significantly (P < 0.01) in the first dimension (Fig. 3a, c, and e), being more pronounced in the 1.0%BC treatment. Significant differences were also observed between the 0.1%ST and 0.1%BC treatments. Permutational multivariate analysis of variance (PERMANOVA) further verified the statistical significance of the differences in microbial community composition among all treatments (P = 0.001). The diversity index of nrfA-encoding DNRA bacteria and nirK- and nirS-encoding denitrifying bacteria was different among treatments. For nrfA, the Shannon index was significantly (P < 0.05) increased in the 0.5%, 1%BC, and 0.1%ST treatments relative to that in the Control treatment. In the case of nirK, the Shannon index in the biochar- or straw-amended treatments was significantly (P < 0.05) higher than that in the Control treatment. In the case of nirS, except for the 0.1%BC treatment, the Shannon index was significantly (P < 0.05) decreased compared with that in the Control treatment (Additional file 1: Fig. S2a). By calculating the Bray–Curtis dissimilarity of microbial communities within each treatment, the beta-diversity of nirS-encoding denitrifier communities in the biochar-amended treatments was significantly (P < 0.05) lower than that in the Control treatment (Additional file 1: Fig. S2b), indicating a decrease in the heterogeneity of nirS-encoding denitrifier habitat in the soil matrix amended with biochar. In the case of nrfA, only in the 0.03%BC treatment was the beta diversity significantly (P < 0.05) increased compared to that in the Control treatment. For nirK, the beta-diversity of the corresponding microbial communities was significantly (P < 0.05) increased in the 0.01%, 0.5%, and 1%BC treatments compared to that in the Control treatment (Additional file 1: Fig. S2b). The main order of DNRA bacteria detected in all the treatments included Myxococcales (53.47–58.78%), Opitutales (8.81–18.34%), and Pirellulales (3.28–7.82%, Fig. 3b). In terms of nirK-encoding denitrifying bacteria, the dominant orders were Rhizobiales (39.58–68.00%) and Nirtrosomonadales (3.03–47.04%) (Fig. 3d), while the orders unclassified_p_Proteobacteria (35.68–57.50%) and unclassified_k_norank_d__Bacteria (32.76–51.30%) dominated in nirS-encoding denitrifying bacteria (Fig. 3f).

Fig. 3
figure 3

Nitrate reducers in the Control, 0.03%BC, 0.1%BC, 0.5%BC, 1.0%BC, and 0.1%ST treatments. Unconstrained principal coordinate analysis (PCoA) shows that soils amended with different amounts of biochar (0%, 0.03%, 0.1%, 0.5%, and 1.0% rates) or straw at 0.1% rate have significantly distinct microbiota, as suggested by permutational multivariate analysis of variance (PERMANOVA): nitrate reducers encoded by the nrfA (a), nirK (c), and nirS (e) genes. Circos plots show the composition of nrfA (b), nirK (d), and nirS- (f) encoding nitrate reducer communities at the order level in each treatment

Full size image

3.5 Co-occurrence networks of nitrate reducer community

The interaction patterns across DNRA (nrfA-encoding) and denitrifying (nirK- or nirS-encoding) bacteria in the Control, 0.03%BC, 0.1%BC, 0.5%BC, 1.0%BC, and 0.1%ST treatments were explored through network analysis based on strong significant correlations (Fig. 4a–f). The co-occurrence networks for the different treatments varied in the number of nodes and edges (Additional file 1: Table S3); there were more nodes while relatively fewer edges in the biochar-amended treatments than in the Control treatment, resulting in a decrease in the connectance of the respective networks. All networks exhibited scale-free characteristics (Power-law model: R2 = 0.975, 0.923, 0.926, 0.946, 0.920, and 0.902 for the Control, 0.03%BC, 0.1%BC, 0.5%BC, 1.0%BC, and 0.1%ST treatments, respectively, Additional file 1: Fig. S3), which suggested that the organization of ecological networks in all treatments was not random. The relationships among nitrate reducers tended to shift from cooccurrence to coexclusion after long-term biochar amendment, as indicated by the increase in the ratios of negative and positive correlations in the biochar-amended treatments (Fig. 4a–f and Additional file 1: Table S3). By comparing the interactions and niches of different nitrate reducers (DNRA bacteria and denitrifiers) in each network, we found that DNRA bacteria (nrfA- encoding) exhibited more interactions with denitrifying bacteria (nirK- or nirS- encoding) in the biochar-amended treatments than in the Control treatments, characterized by increased negative correlations between them (Additional file 1: Table S4). According to the Pi and Zi values of the co-occurrence network analysis, some nodes belonging to connectors, network hubs, and module hubs among nrfA, nirK, and nirS genes were found in the Control, 0.03%BC, 0.1%BC, 0.5%BC, 1.0%BC, and 0.1%ST treatments (Additional file 1: Fig. S4), which were mainly from the order Deinococcales (nrfA), Desulfovibrionales (nrfA), Myxococcales (nrfA), Opitutales (nrfA), Pirellulales (nrfA), Selenomonadales (nrfA), Vicinamibacterales (nrfA), Nitrosomonadales (nirK), Rhizobiales (nirK), unclassied_k__norank_d__Bacteria (nirK), norank_p__environmental_samples (nirK), Xanthomonadales (nirS), unclassified_c__Betaproteobacteria (nirS), unclassied_k__norank_d__Bacteria (nirS), and unclassified_p__Proteobacteria (nirS).

Fig. 4
figure 4

Co-occurrence network analysis between nirK-, nirS-, and nrfA-encoding nitrate reducer communities (OTU level) in the Control, 0.03%BC, 0.1%BC, 0.5%BC, 1.0%BC, and 0.1%ST treatments. In microbial networks, nodes with different colors represent individual OTUs whose size is positively correlated with the node degree, with red, yellow, and dark green nodes representing nrfA OTUs, nirK OTUs, and nirS OTUs, respectively; edges represent significant Sparcc correlations (|correlation|> 0.60 and P < 0.01), of which the color indicates the interactions of positive (gray) and negative (red) between the OTUs (a‒f). Pearson correlations between the relative abundance of keystone orders encoded by nrfA, nirK, and nirS genes (nodes within modular hubs, network hubs, or connectors) and N transformation rates [Denitrification, DNRA, N2O emission, and DNRA/(Den + DNRA) ratio], as well as properties of the network among biochar amendment treatments (g). Coexclusion relationship (Total): ratios of negative and positive correlations among DNRA bacteria and denitrifiers [nrfA-nirK(S), nrfA-nrfA, and nirK(S)-nirK(S)]; Coexclusion relationship (DNRA-Den): ratios of negative and positive correlations between DNRA bacteria and denitrifiers [nrfA-nirK(S)]

Full size image

4 Discussion

4.1 Regulation of biochar amendment on nitrate reduction processes

Denitrification is considered to be the predominant pathway for nitrate reduction processes, particularly in waterlogged paddy soils (Ishii et al. 2011; Shan et al. 2016). Our results showed that the denitrification in the fertilized paddy soil (Control treatment) contributed 98.39% to the total nitrate reduction, which was consistent with the reported proportions of total nitrate reduced by denitrification (57.36‒99.23%) (Li et al. 2020; Shan et al. 2016; Yang et al. 2015b). Compared with paddy soils receiving no N fertilizer and low N input (25 kg urea-N ha−1), denitrification becomes more dominant over other nitrate reduction processes (i.e., anammox and DNRA) in relatively high N input (150‒157 kg urea-N ha−1) paddy soils, where the availability of C is limited (Pandey et al. 2019). In terms of the effects of biochar on N transformation processes, a recent meta-analysis showed that biochar amendment could modify organic N mineralization, N immobilization, and nitrification, and trigger more N2O reduction to N2 (Liu et al. 2018). Our results further demonstrated that biochar amendment significantly (P < 0.05) altered denitrification rates in paddy soils, and the effects were dose-dependent (Fig. 1a). When the amendment rate was below 1.0% (w/w), biochar significantly (P < 0.05) enhanced the denitrification rates, whereas the suppression effect was found once the amendment rate reached 1.0% (Fig. 1a). In the 1.0%BC treatment, the SOC content and the TC/TN ratio were 3.85-fold and 2.04-fold higher,  respectively,  than those in the control treatment (Table 1). Under the C-rich condition, the supply of nitrate may be insufficient to support denitrification (Heo et al. 2020), which may partly explain why denitrification rate was significantly inhibited once biochar amendment rate reached 1.0%. Besides, DNRA rate was the highest in the 1.0%BC treatment, which may in turn limit the nitrate supply for denitrifiers. Similarly, by quantifying net N2 flux, previous studies also showed that the effects of biochar amendment on denitrification rate depended on the biochar dose (He et al. 2019; Obia et al. 2015). In this study, straw amendment at a rate of 0.1% could trigger a higher denitrification rate than straw-derived biochar at the same amendment rate (0.1%BC treatment) (Fig. 1a) in paddy soils, which was attributed to the higher labile C content of the straw (Chèneby et al. 2010; Miller et al. 2008).

Biochar may act as mediators for microbial electron transfer or provide favorable habitat for anammox bacteria, resulting in the enhancement of anammox activities (Xu et al. 2020). The rate of anammox and its contribution to the total N2 production significantly (P < 0.05) increased and were positively correlated with the biochar amendment rates (Fig. 1a), which was in line with the reported promoting effect of biochar on anammox (Pan et al. 2017; Xu et al. 2020). It has been reported that anammox activities are often inhibited in OC-rich environments because OC is adverse to chemolithoautotrophic anammox bacteria, and the abundances and metabolic rates of anammox bacteria are significantly lower than denitrifiers under C-rich conditions (Jin et al. 2012; Shan et al. 2018). The negative anammox rates in the 0.1%BC treatments reflected the competitive disadvantage of anammox process under C-rich environments, while the methodology noise can not be ruled out since the absolute values of anammox rate were low. Indeed, our study also clearly showed negative anammox rates in paddy soils amended with straw (0.1%ST treatment) containing significant labile C. The negative correlation between denitrification and anammox rates (P < 0.01) further supported the competitive disadvantage of anammox bacteria relative to denitrifiers (Additional file 1: Table S5).

Unlike denitrification, the role of DNRA in conserving N is generally ignored in agricultural soils, especially in paddy soils receiving high annual N fertilization (Pandey et al. 2018, 2020). High levels of soil NO3− caused by N loading could inhibit the DNRA rate by suppressing the synthesis of cytochrome c nitrite reductase encoded by the nrfA gene (Schmidt et al. 2011). Furthermore, under high NO3− conditions, denitrifiers can outcompete DNRA bacteria, benefiting from their higher affinity for nitrate/nitrite (Heo et al. 2020; Kraft et al. 2014). In our results, DNRA rate in the fertilized paddy soil (Control treatment) was extremely low and contributed only 1.61% to the total nitrate reduced by denitrification and DNRA (Fig. 1b and c). However, consecutive biochar amendment significantly (P < 0.05) increased the DNRA rates and DNRA/(Den + DNRA) ratios (Fig. 1b and c). This was mainly attributed to the increased C availability in the biochar amendment treatments, as pure bacterial culture experiments have shown that DNRA bacteria could achieve higher growth yields than denitrifiers in a relatively higher OC (or lower nitrate) environment (van den Berg et al. 2015). To our knowledge, our study is the first to report the stimulating effect of biochar on the DNRA rate. Previously, only one study indirectly evaluated the effects of biochar on DNRA in soil by the process-based N cycle model (Xie et al. 2020), showing that biochar exerted a positive effect on DNRA process. In addition, DNRA rates gradually increased with the elevated amendment rates of biochar, resulting in more nitrate being reduced through DNRA process. Consequently, the DNRA/(Den + DNRA) ratio in the 1.0%BC treatment was approximately 7.2-fold higher than that in the Control treatment (Fig. 1c). These results demonstrated that biochar amendment significantly improved N retention by increasing the DNRA rates and favoring more nitrate partitioning toward DNRA process in the paddy field. There was no significant difference in nitrate partitioning to DNRA process between straw (0.1%ST treatment) and straw-derived biochar (0.1%BC treatment) amended treatments, albeit the N loss derived from denitrification was larger than the N conservation derived from DNRA in the 0.1%ST treatment (Fig. 1a and b). This result implied that straw-derived biochar was superior to straw itself in alleviating N loss in the paddy field.

4.2 Mitigation of biochar amendment on N2O emissions

Regardless of the amendment rate, the N2O emission rate was significantly (P < 0.05) decreased following biochar amendment, relative to that in the Control treatment (Fig. 1d). Such an inhibitory effect has been well documented in previous studies (Borchard et al. 2019; Cayuela et al. 2014), which was attributed to either the suppressing effect of biochar on N2O production or the promoting effect of biochar on the reduction of N2O to N2 in upland agricultural soils (Dong et al. 2020; Yuan et al. 2019). Several studies have reported that the decrease in apparent N2O emissions may be caused by either a direct decrease in the activities of nitrifiers and denitrifiers (Dong et al. 2020; Song et al. 2019) or an indirect decline in the availability of NH4+-N and NO3−-N substrates (for producing N2O) triggered by the entrapment of biochar (Hagemann et al. 2017; Nguyen et al. 2017). While in our study, experiments were performed under waterlogged conditions, where N2O emission is predominately produced by denitrification (Ishii et al. 2011). As a result, the inhibition of N2O emission by biochar should be closely related to its effects on denitrification. Although the mitigation of N2O emissions by biochar was not proportional to the biochar amendment rates (Fig. 1d), our results showed a significant decline in the N2O/(N2O + N2) ratios in all biochar-amended treatments compared to the Control treatment (Additional file 1: Fig. S1b), indicating the promotion of the complete reduction of nitrate to N2 in soils with biochar amendment. Furthermore, the N2O emission rates in all biochar-amended treatments were negatively (P < 0.05) correlated with the soil EOC content therein (Additional file 1: Table S5), suggesting that more N2O was reduced to N2 following the enhanced C availability (Giles et al. 2012). Although the decreased N2O emissions in the 1.0%BC treatment may be attributed to the inhibited denitrification, the significant decrease in the N2O/(N2O + N2) ratio indicated the promotion of N2O reduction therein (Additional file 1: Fig. S1). These results supported our research hypothesis that biochar amendment reduces N2O emissions in paddy soils by favoring the last step of denitrification.

Additionally, it has been reported that N2O can also be released as a byproduct of DNRA (Kraft et al. 2011). Although the N2O produced by DNRA may be increased as DNRA process was enhanced by biochar amendment, DNRA may only play a minor role in determining N2O emissions given that the share of DNRA among total nitrate reduction was less than 13.18% (Fig. 1C). It has been reported that the genome of DNRA bacteria also contains the atypical N2O reductase genes (nosZ Clade II) (Sanford et al. 2012), implying that DNRA bacteria could not only convert nitrate to ammonium but also be effectively involved in the reduction of N2O to N2. Nonetheless, the detailed mechanism for the coupling of increased DNRA rate and reduced N2O emission in the biochar amended paddy soils needs to be further investigated. Collectively, our results indicated that the impact of biochar on N2O mitigation was attributed to not only the enhancement of complete denitrification but also the shift in nitrate partitioning toward DNRA process.

4.3 Mechanisms of biochar amendment effects

The rate of nitrate reduction processes and the partitioning of nitrate among denitrification, anammox, and DNRA are generally governed by several interrelated environmental factors, including pH, redox potential, C/N ratio, quality and availability of C (Hardison et al. 2015; Kim et al. 2017; Kraft et al. 2014; Li et al. 2020). In our study, soil pH, TC, and SOC contents were positively (P < 0.001) correlated with DNRA rates, in contrast to denitrification rates (Additional file 1: Table S5). In addition, significant positive relationships were also observed between DNRA rates and soil TN and EOC contents (Additional file 1: Table S5). The abundances of nitrate reduction-related genes (i.e., narG, nirK, nirS, napA, and nrfA) were also influenced by soil pH, TC, TN, SOC, and EOC in the biochar-amended treatments (Fig. 5), which ultimately regulated the nitrate reduction processes. The effect of pH on nitrate reduction processes has been well documented in previous studies (Rutting et al. 2011; Simek and Cooper 2002) and is related to changes in the reassembly of the microbial community and the synthesis of nitrate reduction-related reductase (Saggar et al. 2013). In our study, the diversity of nitrate reducers was significantly (P < 0.05) correlated with soil pH (Fig. 5), and a significant shift in the community of nrfA-, nirK-, and nirS-encoding bacteria in response to biochar amendment was observed (Fig. 3a, c, and d). The SEM analysis also suggested that soil pH exhibited a direct and significant effect (P < 0.001) on the community of DNRA bacteria (Fig. 7).

Fig. 5
figure 5

Pearson correlations between abundance, alpha and beta diversity of nitrate reducers, and N transformation rates [Denitrification, DNRA, N2O emission, and DNRA/(Den + DNRA) ratio], and soil properties in all biochar-amended treatments (Control, 0.03%BC, 0.1%BC, 0.5%BC, and 1.0%BC treatments). The color gradient and square denote Pearson’s correlation coefficients. Asterisks indicate statistical significance with significance levels of *P ≤ 0.05, **P ≤ 0.005, and ***P ≤ 0.001 for P values. The abundance and diversity of related microbes were related to each physicochemical property and N transformation rate by Pearson’s correlation test. Red, green, and gray edges indicate significant at P < 0.01, P < 0.05, and nonsignificant relationships (P ≥ 0.05), respectively. Solid and dashed lines indicate positive and negative relationships, and the thickness of the line denotes Pearson’s correlation coefficients. The alpha and beta diversity of related microbes encoded by different genes were represented by the corresponding Shannon index and Bray–Curtis dissimilarity, respectively

Full size image

Biochar amendment can not only change soil physicochemical properties but also influence the availability of C and N (Bruun et al. 2012; Case et al. 2015). Characterized by the significant increase in the ratios of TC/TN and SOC/NO3− (Table 1), C availability was enhanced in soils consecutively amended with biochar. Our results demonstrated that DNRA rates positively responded to the increased TC/TN and SOC/NO3− (r = 0.93, P = 0.01; r = 0.95, P < 0.01; Fig. 6c and d), whereas denitrification rates showed negative responses (r = − 0.64, P = 0.01; r = − 0.55, P < 0.01; Fig. 6a and b) in the biochar-amended treatments. These results were consistent with previous studies (Pandey et al. 2019; Shan et al. 2016), where positive correlations between DNRA rates and TC/TN or SOC/NO3− were reported in paddy soils. Conventionally, denitrification rate may show a positive response to TC/TN ratio; however, under the C-rich environment such as in the biochar amended paddy soils, denitrification is mainly determined by the supply of nitrate, resulting in its negative correlation with TC/TN. And under nitrate limited while C-rich environments, DNRA can out-compete denitrification since more electrons are transferred per mole of nitrate through DNRA process (Rutting et al., 2011). Furthermore, it has been reported that the ratio of OC to nitrate (electron donors vs. accepters) also affects nitrate partitioning between DNRA and denitrification in soils (Pandey et al. 2019; Rahman et al. 2019b). This was further confirmed in our study, in which a positive correlation between TC/TN (or SOC/NO3−) ratios and DNRA/(Den + DNRA) ratios was observed (r = 0.98, P < 0.01; Fig. 6e and f). Our SEM analysis also revealed that biochar amendment indirectly regulated nitrate partitioning to DNRA process by increasing soil C availability (Fig. 7).

Fig. 6
figure 6

Relationship of denitrification rates (a, b), DNRA rates (c, d), and DNRA/(Den + DNRA) ratios (e, f) with TC/TN or SOC/NO3− ratios among all biochar-amended treatments (Control, 0.03%BC, 0.1%BC, 0.5%BC, and 1.0%BC treatments)

Full size image
Fig. 7
figure 7

Structural equation model based on the effects of abiotic (soil pH and TC/TN ratio) and biotic factors (community and competitiveness of nrfA-encoding DNRA bacteria) on the DNRA/(Den + DNRA) ratio. The community of nrfA-encoding DNRA bacteria was represented by the corresponding alpha diversity (i.e., Shannon index). Competitiveness means the coexclusion relationship that is represented by ratios of negative and positive correlations between DNRA bacteria and denitrifiers [nrfA-nirK(S))]. Blue and red solid arrows indicate significant positive and negative relationships, respectively. The thickness of the arrows is directly proportional to the standardized path coefficient on the single arrow. R2 represents the variance of factors explained by Model (a). Standardized effects (direct and indirect effects) derived from the structural equation model used (b)

Full size image

Crop straw incorporation has been widely accepted as an effective way to regulate N2O emission and the associated N2O/(N2O + N2) ratio in soils (Baggs et al. 2000; Wei et al. 2020). Our results demonstrated that straw amendment significantly (P < 0.05) decreased N2O emission rates and N2O/(N2O + N2) ratios in comparison with the Control treatment (Fig. 1d). Interestingly, higher denitrification and DNRA rates were detected in the 0.1%ST treatment than in the 0.1%BC treatment (Fig. 1a and b), which may result from the higher availability of OC in the straw than in its biochar (Baldock and Smernik 2002). This could also be evidenced by the significant increase in the EOC content in the 0.1%ST treatment relative to the 0.1%BC treatment, albeit no significant change in the SOC/NO3− ratio was observed (Table 1). On the other hand, straw amendment would create a more anaerobic environment favoring the nitrate reduction processes because its decomposition can lower the oxygen level at microsites and provide labile C for soil microorganisms (Chèneby et al. 2010; Zhu et al. 2022).

Network analysis revealed that with biochar amendment, the significance and competitiveness of DNRA bacteria (encoded by nrfA) were enhanced in the overall nitrate reducer community in soils (Fig. 4a‒f). This was evidenced by higher nrfA-OTUs involvement in bacterial interactions (Fig. 4) and the increased ratios of negative and positive interactions between DNRA bacteria (encoded by nrfA) and denitrifiers (encoded by nirK or nirS) (Additional file 1: Table S3). Furthermore, such enhanced ratios of interactions significantly (P < 0.001) exhibited a positive correlation with DNRA rates and DNRA/(Den + DNRA) ratios (Fig. 4g). In addition, consistent with the findings of Han et al. (2020) that nirK- and nirS-encoding denitrifying bacteria may work cooperatively in soils under organic management, more interactions between nirK- and nirS-encoding denitrifiers were detected in either biochar- or straw-amended treatments (Additional file 1: Table S4). More importantly, the compositions and relative abundances of key nodes (keystone order of denitrifiers and DNRA bacteria) exhibited significant changes in response to biochar amendment (Additional file 1: Fig. S5). Opitutales has been widely reported to be involved in performing DNRA process in aquatic ecosystems (Bu et al. 2017). In our results, the relative abundance of the keystone order Opitutales significantly (P < 0.01) increased in the biochar-amended treatments in comparison with the Control treatment (Additional file 1: Fig. S5a and b) and positively (P < 0.001) correlated with DNRA rates and DNRA/(Den + DNRA) ratios (Fig. 4g). Meanwhile, the enrichment of denitrifying bacteria was also detected in the biochar-amended soils, such as the orders Rhizobiales and Rhodocyclales (Additional file 1: Fig. S5a and b). These orders are responsible for nitrate reduction processes (Hesselsoe et al. 2009), which may trigger higher N2 or N2O emissions in paddy soils.

5 Conclusion

Our results showed that biochar amendment significantly (P < 0.05) increased DNRA rates by 1.63‒6.84 times. Meanwhile, denitrification rates were also enhanced by 10.19‒46.03% following biochar amendment except for the highest amendment rate (1.0%). Consequently, biochar amendment shifted more nitrate partitioning toward DNRA process, which was mainly attributed to the increased soil C availability, the enhanced dominance and interactions of DNRA bacteria in overall nitrate reducers, and the enriched DNRA bacteria in the biochar-amended soils. Conversely, N2O emission rates following biochar amendment were significantly (P < 0.05) decreased by 61.86‒97.65%, owing to the enhancement of complete denitrification. At the same amendment rate (0.1%, w/w), straw-derived biochar was superior to straw itself in terms of alleviating N loss since biochar caused less denitrification-driven N loss. Overall, our study suggests that biochar amendment favors nitrate partitioning toward DNRA process, which is beneficial for alleviating reactive N loss in the paddy field. Our study also provides direct evidence for the stimulating effects of long-term biochar amendment on the DNRA rate, improving our understanding of how biochar mediates N cycling in the paddy field.