The legacy effect of biochar application on soil nitrous oxide emissions

Existing studies suggest that biochar application can reduce soil nitrous oxide (N2O) emissions, mainly based on short‐term results. However, it remains unclear what the effects (i.e., legacy effects) and underlying mechanisms are on N2O emissions after many years of a single application of biochar. Here, we collected intact soil columns from plots without and with biochar application in a subtropical tea plantation 7 years ago for an incubation experiment. We used the N2O isotopocule analysis combined with ammonia oxidizer‐specific inhibitors and molecular biology approaches to investigate how the legacy effect of biochar affected soil N2O emissions. Results showed that the soil in the presence of biochar had lower N2O emissions than the control albeit statistically insignificant. The legacy effect of biochar in decreasing N2O emissions may be attributed to the reduced effectiveness of the soil substrate, nitrification and denitrification activities, and the promotion of the further reduction of N2O. The legacy effect of biochar reduced the relative contribution of nitrifier denitrification/bacterial denitrification, nitrification‐related N2O production, and the relative abundance of several microorganisms involved in the nitrogen cycle. Our global meta‐analysis also showed that the reduction of N2O by biochar increased with increasing application rate but diminished and possibly even reversed with increasing experimental time. In conclusion, our findings suggest that the abatement capacity of biochar on soil N2O emissions may weaken over time after biochar application, but this remains under further investigation.


| INTRODUCTION
Nitrous oxide (N 2 O) is a potent greenhouse gas that can cause a greenhouse effect and destroy the stratospheric ozone layer (Ravishankara et al., 2009;Stocker et al., 2013). Soil N 2 O is produced primarily through biological processes, such as heterotrophic bacterial denitrification, autotrophic nitrification, nitrifier denitrification, fungal denitrification, and heterotrophic nitrification (Butterbach-Bahl et al., 2013). These processes can be modulated by various biotic and environmental factors (Braker & Conrad, 2011;Hu et al., 2015). The current increase in atmospheric N 2 O concentration is mainly driven by anthropogenic emissions, predominantly released from agricultural activities (Tian et al., 2020). Among these, direct soil emissions from nitrogen (N) fertilizer application were considered to be the majority source of agricultural N 2 O emissions (Reay et al., 2012).
As an effective carbon sequestration and emissions reduction measure, biochar is increasingly becoming a priority in combating climate change (Lehmann et al., 2021;Wang et al., 2016). Over the past decade, numerous field experiments have been conducted globally to examine the impact of biochar application on soil N 2 O emissions. Several meta-analysis studies have shown that compared to controls without biochar amendment, biochar can reduce N 2 O emissions by 28%-38% (Borchard et al., 2019;Cayuela et al., 2015), while in acidic agricultural soils this reduction would be reduced to 17% . Contrary to the majority of studies reporting that biochar can decrease N 2 O emissions from the soil, some individual studies have found that biochar exhibited either a promoted or no effect on N 2 O emissions (Clough et al., 2010;Lan et al., 2019;Lin et al., 2017;Xiang et al., 2014). Different hypotheses have been proposed to explain why biochar addition may have different effects on N 2 O emissions. For example, biochar addition inhibits soil N 2 O emissions mainly by increasing soil aeration and pH, reducing substrate effectiveness, absorbing N 2 O, or by its toxic effects on nitrifying and denitrifying microbial communities (see reviews by Cornelissen et al., 2013;Spokas et al., 2010;Van Zwieten et al., 2010). By contrast, increased N 2 O emissions from biochar application may be attributed to (i) the priming effect on soil organic matter or release of biochar embodied-N, (ii) increased soil moisture and improved denitrification conditions, and (iii) provision of inorganic N or carbon substrate to microorganisms (Clough et al., 2013). In addition to climate, soil, and plants, the effect of biochar on N 2 O emissions can be strongly influenced by the properties of biochar (Cayuela et al., 2014). Biochar applied to the soil is exposed to various biotic and abiotic factors over the years, and its physicochemical properties are altered (Dong et al., 2017;Wang et al., 2021), which may affect the nitrogen transformation processes and thus N 2 O emissions in the soil.
It remains unclear how biochar affects soil N 2 O emissions years after its application. The existing findings of meta-analyses do not reflect this legacy effect of biochar on soil N 2 O emissions since most field studies in the dataset are short term (Borchard et al., 2019;Zhang et al., 2021). Moreover, the long-term effects of biochar application are inconsistent in these meta-analyses, showing negligible effects on N 2 O emissions from the soil with biochar application after 1 year (Borchard et al., 2019) or increase N 2 O emissions as the experiment was prolonged (Zhang et al., 2021). While several independent studies have focused on the effects of biochar on soil N 2 O emissions after years of being subjected to biotic and various environmental factors in the field, these findings are also mixed, with results of stimulation , inhibition (Hagemann et al., 2017;Liao, Müller, et al., 2021), and no effect Mukherjee et al., 2014;Spokas, 2013) were reported. Thus, the lack of understanding of the legacy effects of biochar on N 2 O emissions will make the use of biochar as a promising technology for negative emissions from the agricultural sector uncertain, as the possible increase in N 2 O emissions will offset its carbon sequestration contribution (Guenet et al., 2021;Lehmann et al., 2021). These contradictory results call for a more comprehensive assessment of the mitigation effects of biochar on soil N 2 O emissions and its underlying mechanisms from a longterm perspective.
Acidic soils are a global hotspot for N 2 O emissions, with higher emission intensities than alkaline soils for the same N fertilizer inputs (Hergoualc'h et al., 2021;Wang et al., 2022). Tea plantation soil is one of the typical acidic soils in the world. The high input of N fertilizer leads to increased acidification of tea plantation soils, thus making tea plantations one of the global hotspots of agricultural soil N 2 O emissions . Therefore, any targeted emission reduction measures implemented on this soil type will contribute to agricultural non-CO 2 emission reduction. Previous studies indicated that biochar applied annually in a tea plantation could sustainably reduce N 2 O emissions (Ji et al., 2020). By contrast, biochar application decreased N 2 O emissions from a tea plantation only in the first year, but the effect was insignificant in the second year . Despite this, the legacy effects of biochar application on N 2 O emissions from acidic tea plantation soils are still poorly understood.
The objective of this study was to examine the legacy effect of biochar application on soil N 2 O emissions and the mechanisms behind this effect. To address these questions, we collected intact soil columns from the plots without and with biochar applied 7 years ago in a subtropical tea plantation. We analyzed N 2 O fluxes, N 2 O isotope signatures, the abundance, composition, and community structure of functional microbes associated with N-cycling, and the relative contributions of different N 2 O production pathways. Emerging evidence suggests that the N 2 O reduction effect of biochar may persist throughout its life cycle (Lehmann et al., 2021). Therefore, we hypothesized that biochar application to tea plantation soils would still reduce N 2 O emissions 7 years later. As a complementary analysis, we also conducted a global meta-analysis to explore changes in the effect of biochar on N 2 O emissions over time and the response of N-cycling genes to biochar addition.

| Field experiment description and soil collection
The field experiment was established in October 2014 on a commercial organic tea plantation in Yixing city, eastern China (31°14′N, 119°46′E). The tea trees were planted in 2002. This area has a humid subtropical climate (Cfa) according to a modified Köppen climate classification, with mean annual temperature and precipitation of 17°C and 1208 mm, respectively. The soil is classified as an Alisol (FAO, 1995) and contains 48.3% clay, 13.2% silt, and 38.5% sand.
The field experiment consisted of two treatments with three replications: N fertilizer (F) and N fertilizer plus biochar (F + BC). The application rate was 300 kg N ha −1 year −1 of N fertilizer in the form of compound fertilizer (N: P 2 O 5 : K 2 O = 16:16:16). Fertilizer was applied in two splits, a quarter as a basal fertilizer in mid-October, and the remainder as a top dressing around March. In the F + BC plots, biochar was incorporated into the soil at a rate of 20 t ha −1 along with basal fertilizers in 2014 and 2015. Following the local farming practices, fertilizer and biochar were added as band application in the soil between rows with widths of 20 cm and then incorporated into soils at a depth of 10 cm. The biochar was derived from wheat straw at a pyrolysis temperature of 350-500°C. The biochar had a total organic carbon of 46.7%, a total N of 0.6%, and a pH of 10.9. The field observation experiment ended in October 2016. From October 2016 to the time of soil sample collection used for the following soil column experiment, fertilization management was maintained consistent with the conventional management of the tea plantation. Rapeseed cake and urea were applied at 1.5 t ha −1 year −1 and 150 kg N ha −1 year −1 , respectively. The tea leaves were harvested and trimmed in early April and late May each year, respectively. Other practices were consistent with local tea plantation management.
To investigate the legacy effect of biochar application on soil N 2 O emissions, we collected soils from F and F + BC treatments in May 2021 for laboratory experiments and analysis. In the plot between rows, we collected an intact soil column using a PVC cylinder (inner diameter: 8 cm, height: 15 cm) and the surrounding topsoil (0-15 cm) using a small shovel. The soil samples were immediately transported back to the laboratory to conduct subsequent gas emission observations and analysis of soil physicochemical and biological process measurements. The soil samples were sieved with a 2 mm mesh to remove gravel and plant fragments and homogenized them thoroughly. The homogenized sieved soil was then separated into three parts: one was air-dried for soil physicochemical analysis, one was stored at 4°C for subsequent analysis, and the third was frozen at −80°C for soil DNA extraction.

| Soil column experiment
All intact soil columns were preincubated in the laboratory for 5 days to stabilize soil microbial activity. Then, we added urea at 100 kg N ha −1 to each soil column to simulate fertilization. After 1 week, we found noticeable soil moisture loss affected by fluctuations in room temperature (26.0-29.5°C). We added 60 ml of deionized water to each soil column. In addition, the soil columns were covered with plastic films with pinholes to maintain aerobic conditions and minimize water evaporation. Water losses were replenished by periodically weighing and adding deionized water. The laboratory experiment was conducted for 45 days. The average temperature throughout the incubation period was 27.7°C, comparable to the average temperature of tea plants in the field during the growing season.

| Gas flux measurements in field and soil column experiments
Gas samples for the field experiment were collected using the static chamber method for 2 years (2014)(2015)(2016). Briefly, we installed PVC collars at a depth of 15 cm between the rows of the tea plantation. A PVC cylinder (diameter: 30 cm, height: 80 cm) was placed on the base collar during gas sample collection, and five gas samples were collected at 5-min intervals. The gas samples were stored in 1.5-L pre-evacuated gas bags (Delin Gas Packing Co., LTD, Dalian, China) and transported back to the laboratory for analysis. Gas sample collection was carried out between 8:00 am and 10:00 am local time. The frequency of gas sample collection was weekly throughout the experimental period, except for intensive sampling after fertilization events (three times per week).
For the soil column experiment, gas samples were taken at 1, 3, 5, 7, 9, 11, 13, 16, 20, 26, 33, and 45 days. Soil columns were placed in 2-L Lock & Lock containers, and then gas samples were collected using a lid with a rubber stopper after sealing. Within 2 h of lid closure, three 10-ml gas samples were collected from the top of the container using a 25-ml syringe. On day 16 of the incubation, we collected an additional 150 ml of gas and stored it in a 0.2-L pre-evacuated gas bag (Delin Gas Packing Co., LTD) after gas sample collection, while air samples were collected from the laboratory at that time for the determination of N 2 O isotopic signature values.

| Aerobic inhibitor incubation and slurry experiments
To distinguish the contribution of autotrophic nitrifiers and other microorganisms to N 2 O emissions, we conducted an aerobic incubation experiment with or without specific inhibitors of autotrophic ammonia oxidation (acetylene and 1-octyne) (Taylor et al., 2013). Specifically, we set up the following treatments for field F and F + BC soils: control (i.e., no inhibitor), acetylene (inhibit AOA and AOB), and 1-octyne (inhibit AOB). The microcosm incubation experiment was established by weighing the fresh soils (10 g dry weight equivalent) into 120-ml Wheaton bottles. Each bottle was uniformly spiked with 1 ml of NH 4 Cl (final concentration was 100 mg N kg −1 ). Then we injected 1-octyne (2.6 ml, 4 μm) or acetylene (288 μl, 6 μm) in the corresponding treated bottles (Mushinski et al., 2019). The microcosm incubation was performed at 25°C for 20 days. We extracted gas samples from each bottle for analysis on incubation days 1, 3, 7, 12, and 20. The gas samples were collected by sealing the serum bottles with aluminum caps with butyl stoppers after collecting headspace samples (5 ml) from the bottles at 0 and 2 h, respectively, using a syringe connected to a three-way valve.
We used a modified shaking slurry method (Hart et al., 1994) to determine the potential for N 2 O production from autotrophic and heterotrophic nitrification. Briefly, 5 g (equivalent dry weight) of F and F + BC treated fresh soils was placed into 120-ml Wheaton bottles, and 50 ml of 0.72 mm NH 4 Cl was added to maximize nitrification. We used low concentrations of acetylene as an inhibitor to distinguish the relative contribution of autotrophic and heterotrophic nitrification processes to N 2 O production, which was achieved by the presence or absence of acetylene (288 μl, 6 μm) in the bottle (Mushinski et al., 2019). The bottles were incubated at 200 rpm for 24 h at 25°C. We collected headspace samples from the bottles at 2 and 24 h to determine the cumulative N 2 O production.

| Measurement of N 2 O, isotope analysis, and N 2 O source partitioning
The N 2 O concentrations of gas samples were analyzed by gas chromatograph (Agilent 7890B, Santa Clara) equipped with an electron capture detector (ECD) at 330°C. The N 2 O detection conditions were as described by Han, Wang, Xu, Sun, et al. (2021). A standard N 2 O gas with a concentration of 600 ppb was used to calibrate the N 2 O concentration of the sample.
Gas samples from day 16 of the soil column experiment and laboratory air samples were used for N 2 O isotope signature analysis. N 2 O isotope analyses of natural abundances are often used to estimate the proportion of possible contributions from specific pathways (Toyoda et al., 2017;Yu et al., 2020). The δ 15 N bulk , δ 15 N α , and δ 18 O isotope signatures of N 2 O were determined by analysis on an isotope ratio mass spectrometer (Isoprime100, Cheadle) at the Institute of Environment and Sustainable Development in Agriculture, Chinese Academy of Agricultural Sciences. The measured δ 18 O and δ 15 N isotope signatures were expressed regarding Vienna Standard Mean Ocean Water (VSMOW) and atmospheric air-N 2 , respectively. The details of correction and calibration are described previously (Heil et al., 2015). The δ 15 N α and δ 15 N bulk are the central position and average δ 15 N on the N 2 O molecules, respectively. The values of δ 15 N β and δ 15 N SP are calculated as follows: The soil-emitted N 2 O isotope signatures, including δ 15 N, δ 18 O, or δ 15 N SP values, were corrected following the mass conservation (Well et al., 2006): where C and δ represent the isotope signatures and concentrations of N 2 O pools in ambient air and gas samples, expressed as subscripts. The N 2 O concentration in ambient laboratory air was 309.9 ± 0.29 ppb.
Nitrification (Ni), fungal denitrification (fD), bacterial denitrification (bD), and nitrifier denitrification (nD) are four major microbial processes of soil N 2 O emissions. According to the difference between SP and δ 18 O, the four microbial processes can be split into two major parts to quantify the contribution of microbiological processes and the degree of N 2 O reduction (Buchen et al., 2018). That is, Ni and fD with similar and high values of δ 15 N SP and δ 18 O are assumed to be Ni/fD, while nD and bD with low and similar δ 15 N SP and δ 18 O values are considered as nD/bD. To assess the contribution of soil-emitted N 2 O originating from Ni/fD and nD/bD while also considering the occurrence of N 2 O reduction, we have relied on the following equation (Buchen et al., 2018): where X represents the fraction of Ni/fD to total N 2 O release. ηr the net isotope effect associated with N 2 O reduction. Fr the extent of N 2 O reduction (i.e., expresses the N 2 / (N 2 O + N 2 )). δ Ni/fD and δ nD/bD the isotopic signature values (δ 15 N SP and δ 18 O) from Ni/fD and nD/bD, respectively. 1-Fr is the unreduced N 2 O fraction (r N 2 O , i.e., expresses the N 2 O/ (N 2 O + N 2 )). The δ 15 N SP and δ 18 O of N 2 O in the four pathways are summarized, and the net isotope effect of N 2 O reduction to N 2 , η 15 N SP red , and η 18 O red

| Analysis of soil characteristics
Soil total carbon (TC) and nitrogen (TN) were measured with a EuroEA 3000 elemental analyzer (EuroVector). Soil pH and EC were determined using 1:2.5 soil-to-water ratios. We determined the gravimetric moisture content of fresh soil by drying it at 105°C until it reached a constant mass. Soil bulk density at a depth of 5 cm was determined by the soil core method, obtained by calculating the ratio of soil mass to total volume after oven drying to a constant weight at 105°C. The water-filled pore space (WFPS) was calculated using dividing volumetric water content by total porosity (Linn & Doran, 1984). Soil mineral N (NH 4 + , NO 3 − , and NO 2 − ) was extracted by 2 m KCl in a 1:5 (w/v) soil-to-water solution and measured using the colorimetric methods on a microplate reader (BioTek). Soil dissolved organic carbon (DOC) extracted from ultrapure water was analyzed by a Shimadzu TOC analyzer (TOC-L, Shimadzu). To assess net N mineralization and nitrification rates, soil samples were incubated aerobically at 25°C in the dark (Hart et al., 1994). Soil potential nitrification and denitrification rates were determined using the shaken-slurry method (Hart et al., 1994) and the modified acetylene inhibition technique (Patra et al., 2005), respectively.

| Quantitative PCR and highthroughput sequencing of soil microbial communities and functional genes
For the soil DNA extraction, 0.25 g of fresh soil was used with the DNeasy PowerSoil Kit (Qiagen Inc.) according to the manufacturer's instructions. The concentration was detected with a Nanodrop ND-100 spectrophotometer (Thermo Scientific). Real-time quantitative PCR was performed on each DNA sample with three analytical replicates to determine the copy number of bacterial 16 S rRNA, fungal ITS (internal transcribed spacer) region, nitrification genes (archaeal and bacterial amoA), denitrification genes (nirS, nirK, fungal nirK, nosZI, and nosZII). Reactions were carried out in 96-well plates using the StepOnePlus™ Real-Time PCR System (ABI). A total of 20 μl of qPCR mixture contained 2 μl template DNA, 6.8 μl sterile water, 0.4 μl ROX reference dye (50×), 0.4 μl each forward and reverse primers, and 10 μl SYBR @ Premix Ex Taq. Melting curve analysis was performed at the end of the assay to confirm the specificity of the PCR product. The amplification efficiencies were 90%-95%, and standard curves R 2 were 0.990-0.998. Table S1 presents gene-specific primers and thermal conditions.
To examine the legacy effects of biochar application on soil microbial communities, DNA samples were subjected to high-throughput sequencing analysis for AOA, AOB, nirS, nirK, nosZI, and nosZII genes. The primers and PCR conditions for these genes were consistent with those of the qPCR described above. The target PCR products were then purified and sequenced on an Illumina MiSeq sequencer at Shanghai Biozeron Biological Technology Co. Ltd. Sequencing reads were assigned according to unique barcodes and processed with the QIIME open-source bioinformatics pipeline (Caporaso et al., 2010). Filtering of noisy sequences, chimera checking, and sequences was clustered into operational taxonomic units (OTUs) at a 97% similarity level and picking out representative sequences from each OTU using USEARCH software (Edgar, 2013). Unfortunately, the nosZII amplification was not eligible for sequencing, and no sequence data were available. The raw sequence data of these functional genes were deposited in the NCBI SRA database under the accession number PRJNA846981.

| Meta-analysis and statistical analysis
To investigate the effects of biochar application on soil N 2 O emissions and functional genes associated with N-cycling, we compiled and updated the datasets of several existing meta-analysis studies (Borchard et al., 2019;Liu et al., 2019;Xiao et al., 2019). The literature collection cutoff date is March 2022. Our database contained only field and pot Fr) studies and excluded laboratory incubation experiments because of its underrepresentation of in situ environments. In addition, we referred to the search terms and screening criteria used in previous meta-analyses when updating the literature. In total, we obtained 747 pairs of observations from 96 articles, of which 398 were for N 2 O emissions, 166 were for genes related to ammonia oxidation processes, and 183 were for genes related to denitrification processes (Table S2).
To evaluate the effects of biochar addition on soil N 2 O emissions and functional genes, we used a natural logarithm of the response ratio (lnR) (Hedges et al., 1999): where X t and X c are the means of N 2 O emissions or Ncycling gene abundances in the biochar-amended treatment and control, respectively. Consistent with previous studies (Van Groenigen et al., 2011;Wang & Zou, 2020), we performed weighting using the number of replicates: where N c and N t represent the numbers of replicates in the control and biochar-amended treatment, respectively. The overall impact of biochar application on N-cycling gene abundances is expressed as (e lnR -1) × 100%. If the 95% confidence interval (CI) did not include zero, it was considered significant.
Normality and equal variance tests were performed on all data. For parameters with unequal variances and non-normal distributions, logarithmically or square-root transformation is required before analysis. Differences in soil characteristics, N 2 O emissions, αdiversity, and abundances of investigated genes between treatments were analyzed using the nonparametric Kruskal-Wallis test. A linear mixed-effects model with a study as a random factor was used to examine the overall effects of biochar application on N 2 O emissions and N-cycling gene abundances. We used a linear mixed-effects model to investigate whether biochar application rate, experimental duration, and their interactions affect the response ratio of N 2 O emissions to biochar application. Missing values were excluded from analyses, and statistical significance was recognized if p ≤ 0.05. All statistical analyses were carried out using R version 4.1.2 (R Core Team, 2021).

| Soil properties
Overall, there was no statistically significant difference in soil physicochemical properties of tea plantations between the F and F + BC treatments after 7 years of biochar application (Table 1). There was a higher content of soil carbon in the F + BC treatment compared to the F treatment. However, biochar addition significantly lowered soil bulk density (p = 0.049). While there was no difference in mineral N content, net N mineralization, and nitrification rates of soils were lower in the F + BC treatment than in the F treatment. Similarly, biochar addition reduced soil nitrification and denitrification potentials.

| N 2 O emissions from the field and soil column experiments
The field observations showed that soil N 2 O emissions exhibited marked temporal variation, but different treatments had similar emission patterns (Figure 1a) (reported by Ji et al., 2020). Specifically, higher N 2 O emissions occurred mainly during the period of high temperature and high humidity after fertilizer application. For cumulative N 2 O emissions, there was an obvious difference between F + BC and F treatments of 41.6 and 54.7 kg N 2 O-N ha −1 , respectively (p = 0.049). Similar to the field observations, the temporal variability of soil N 2 O emissions was consistent in the F and F + BC treatments (Figure 1b). We observed a significant increase in smooth N 2 O emissions after urea was added, followed by a sharp decrease in emission intensity to the lowest level due to significant water evaporation. After adding deionized water to adjust soil moisture, N 2 O emissions rose again to peak and then gradually decreased toward the end of the incubation. Across the observation period, there was a lower total N 2 O emissions in the F + BC treatment (33.3 g N 2 O-N ha −1 ) compared to the F treatment (60.3 g N 2 O-N ha −1 ), albeit not significantly (p = 0.82).

| N 2 O isotope signatures and production pathways
In the soil column experiment, δ 15 N SP and δ 18 O of N 2 O during the peak emission period were marginally higher in the F + BC treatment compared with the F treatment ( Figure 2a). As shown in the dual-isotope plots, the position of N 2 O isotope signature values for the F + BC treatment was closer to nitrification/fungal denitrification, while that for the F treatment was closer to the other end, that is, nitrifier denitrification/bacterial denitrification. This suggests that the legacy effect of biochar may have changed the contribution of different microbiological processes to N 2 O production. The source partitioning results based on the two-source mixing model suggested that nitrifier denitrification/bacterial denitrification may be the dominant pathway for N 2 O emissions from tea plantation soils. The contribution of nitrifier denitrification/bacterial denitrification was lower in the F + BC treatment than in the F treatment (Figure 2b). Correspondingly, the proportion of N 2 O reduction (Fr) increased, and the proportion of residual ((r N 2 O ) decreased in the treatment with biochar compared to the control.
The short-term aerobic incubation showed that N 2 O emissions from the F + BC treatment were lower than those from the F treatment (Figure 2c). In both treatments, the N 2 O emissions of soil with specific inhibitors were lower than that without inhibitors. N 2 O emissions were comparable between the treatments with 1-octyne and acetylene addition, indicating that the ammonia oxidation process is predominantly AOB-dominated in the soils. Heterotrophic nitrification and other processes also contributed comparably or more than AOB to N 2 O emissions (Table S3). In contrast, the proportion of N 2 O produced by AOB increased slightly in the treatment with biochar application. The shaking slurry assay under aerobic conditions showed that the nitrification process had less N 2 O in the F + BC treatment than in the F treatment, albeit not statistically significant either. The contribution of N 2 O emissions from heterotrophic nitrification was slightly higher in both treatments relative to autotrophic nitrification (Table S3). Relative to the F treatment, the F + BC treatment increased the contribution of heterotrophic nitrification to N 2 O from nitrification.

| Microbial gene abundance and community composition
We observed differences in soil microbial abundance between soils with and without biochar (Figure 3). Compared to fertilizer alone, the F + BC treatment tended to increase soil bacterial 16 S rRNA, AOB, and nosZII gene abundances but decrease fungal ITS, nirK, nirS, nosZI, and nosZ (nosZI + nosZII), albeit the absence of statistically significant changes. However, the abundance of AOA and fungal nirK was remarkably lower in the F + BC treatment compared to the F treatment (p = 0.049). These changes resulted in a higher AOB/AOA ratio and nosZ/ (AOA + AOB + fungal nirK + nirK + nirS) in the F + BC treatment than in the F treatment.
The αdiversity of nitrifying and denitrifying communities differed in the two treatments (Table S4). Compared to the nitrogen fertilizer alone treatment, the F + BC treatment tended to reduce the richness and Shannon index of AOA, AOB, and nosZI. The Shannon index of nirK was obviously higher in the F + BC treatment compared with the F treatment, but there was no noticeable alteration in its richness (p = 0.13).
We compared the variation in N-cycling microbial community composition among treatments at the genus level (Figure 4). The AOA community composition was dominated by Nitrososphaera, accounting for about 70%. The abundance of Nitrososphaera and Candidatus_Nitrosopelagicus was lower, and Candidatus_Nitrosotalea was higher in the F + BC treatment compared to the F treatment. Nitrosospira constituted the majority of the AOB community, ranging from 56% to 95%, with lower abundance in the F + BC treatment than in the F treatment. The most abundant genus was Bradyrhizobium, and this taxon accounted for more than 15% of the nirK community. The presence of F I G U R E 2 Soil-emitted N 2 O isotope data in δ 15 N SP /δ 18 O isotope mapping approach (SP/O map) (a). The relative contribution of N 2 O production from two endmembers nitrification/ fungal denitrification and nitrifier denitrification/bacterial denitrification (b). Changes in cumulative N 2 O emission from the soils with or without ammonia oxidizer-specific inhibitors over a 20-day aerobic incubation (c). The dashed red line indicates the line of reduction in theory (slope η 15 N SP red /η 18 O red ratios). The ranges of values for endmember (standard deviation of means) are presented as boxes. F, fertilization without biochar addition; F + BC, fertilization with biochar addition. Values represent mean ± SEM (n = 3). biochar reduced Bradyrhizobium, albeit not significantly. Candidatus_Competibacter (13%-42%) and Lautropia (40%-82%) were the main components of the nirS community. Similarly, the proportion of Candidatus_ Competibacter decreased in soils where biochar was present. Methylocella was the dominant species in the nosZI community ranging from 41% to 67%. In addition, we found that Candidatus_Contendobacter and Planctomyces were occupied in the treatment with added biochar. However, NMDS analysis showed no structural change in the microbial communities associated with soil N-cycling (data not shown). Overall, the legacy effect of biochar changed the composition but not the structure of soil N-cycling microbial communities.

F I G U R E 3
Comparison of the abundance and ratios of bacterial, fungal, and N-cycling genes in soil. F, fertilization without biochar addition; F + BC, fertilization with biochar addition. Values represent mean ± SEM (n = 3).

F I G U R E 4
Relative abundances of nitrifying and denitrifying microbial community composition at the genus level. F, fertilization without biochar addition; F + BC, fertilization with biochar addition.

| Global analysis of soil N 2 O emissions and N-cycling genes under biochar addition
The updated dataset shows that existing studies on biochar application's effects on N 2 O emissions and its associated functional microorganisms are located mainly in China and Europe, followed by Australia, with sporadic distribution elsewhere in the world (Figure 5a). Overall, biochar application reduced N 2 O emissions by 28.2% (95% CI: −37.6% to −17.2%; Figure 5c). We found that this reduction effect of biochar on N 2 O decreased with the experimental duration (slope = 0.157, p = 0.002) and increased with its application rate (slope = −0.014, p < 0.001; Figure 5b). Biochar addition on N 2 O reduction effect was influenced by the experimental duration. This was manifested by a decrease in N 2 O emissions with increasing biochar application in short-term experiments (i.e., within 1 year), while this effect of biochar may be reversed in longterm experiments. Due to the relatively limited number of long-term experiments, differences between subgroups of different experimental durations were not significant. For functional genes related to N-cycling, we found that the biochar application remarkably increased the abundance of AOA (28.4%, 95% CI: 5.1%-56.8%), AOB (39.9%, 95% CI: 6.7%-83.4%), and nosZ (21.1%, 95% CI: 0.0%-46.6%) genes, and similarly increased the abundance of nir-type denitrifiers (nirK: 20.9%, 95% CI: −10.6% to 63.5%, nirS: 17.7%, 95% CI: −5.5% to 46.6%) but not statistically significant.

| DISCUSSION
Our results suggest that the legacy effect of biochar applied to acidic tea plantation soils can contribute to reduce N 2 O emissions, which is in line with our hypothesis. We compiled data from previous studies for comparison ( Table 2), most of which showed results consistent with the present study. However, contrary to our conclusion, the legacy effect of biochar applied in tea plantation ecosystems increased soil N 2 O emissions . This may be because biochar ultimately increased soil pH, while the present study reduced soil pH. In addition, the results of our meta-analysis suggest that the reasons for the inconsistent F I G U R E 5 The global distribution of study sites of inclusion in this meta-analysis (a). The main effect of biochar application rate, experimental duration, and experimental duration-dependent changes of the response ratio (lnR) of N 2 O emissions (b). Boxplot and density curve displayed the lnR distribution of N 2 O emission and N-cycling genes to biochar addition (c). Values are means and 95% confidence intervals of bootstrap. The red triangle represents the biochar application rate slope of all observations. The line of vertical dashes is drawn at the zero point.
results of the above studies may be related to the amount of biochar applied and the experimental duration, and have been reported in a similar study . It is noteworthy that N 2 O emissions in the field were not continuously observed, so we could not determine whether the reduction in soil N 2 O emissions of tea plantations persisted for 7 years following biochar application.
The legacy effect of biochar amendment on N 2 O reduction may be related to changes in soil physicochemical properties and functional microbial activities associated with N-cycling. We found that NH 4 + concentration in the soil with biochar was high (Table 1), which is in agreement with the previous finding that higher NH 4 + concentration in the soil after a single application of biochar for 4 years than in the soil without biochar (Liao et al., 2020). This finding could be attributed to the extraction of NH 4 + adsorbed on biochar by the KCl extract (Clough et al., 2013;Wang et al., 2015). The bioavailability of NH 4 + is usually reduced in biochar-amended soil, thus affecting the growth of ammonia-oxidizing microorganisms. Indeed, we found a reduced abundance of autotrophic ammonia-oxidizers in the soil with biochar, supported by the reduced NH 4 + available to microorganisms under field conditions. In addition, we found lower N 2 O emissions driven by autotrophic nitrification and other microbial processes in the F + BC compared to the F plots (Table S3). This is consistent with the finding that aging biochar reduced N 2 O emissions by reducing autotrophic nitrification via reduced bioavailability of NH 4 + (Liao, Müller, et al., 2021). Therefore, we conclude that the reduction in substrate effectiveness and microbial functional gene abundance may have affected soil nitrification, thereby suppressing its driven N 2 O emissions (Liao, Müller, et al., 2021;Yao et al., 2022). In addition, our results showed that the legacy effect of biochar altered the nitrification process in soils. In acidic soils, the proportion of heterotrophic nitrification in total nitrification may be higher than that of autotrophic nitrification (Stange et al., 2013;Zhang et al., 2014;. Consistent with this perception, heterotrophic nitrification also contributed more N 2 O to nitrification production than autotrophic nitrification, with a higher contribution in the F + BC than in the F treatment, which may be evidenced by the higher C/N and lower pH (Table 1) (Seredych et al., 2011;. AOA is the dominant ammoniaoxidizing community in acidic soils due to its ability to adapt to more extreme environments such as acidic soils than AOB (Figure 4). However, autotrophic nitrification is dominated by AOB in our acidic tea plantation soils. This is consistent with the previous study that higher numbers of AOA in soils do not necessarily mean that they are the major nitrifiers (Nicol et al., 2008). Furthermore, we found that the contribution of AOB to autotrophic nitrification in the F + BC treatment was higher than in the F treatment (Table S3), which is supported by the increased AOB/AOA ratio. The observed divergence in the effect of biochar on the abundance of AOA and AOB genes may be attributed to their different ecological niches (Hink et al., 2018).   Wu et al. (2021) This is inconsistent with the results of the meta-analysis, where the application of biochar increased the abundance of ammonia-oxidizers ( Figure 5). This inconsistency may be mainly attributed to the fact that the present study is a legacy effect of biochar application, whereas most of the studies in the meta-analysis are short-term experiments. Denitrification is mainly regulated by heterotrophic microorganisms whose growth and activity are influenced by available carbon sources (Moser et al., 2018;Wu et al., 2017). DOC is the primary factor controlling denitrification and explaining the abundance of denitrifiers in the topsoil, as it is the main source of energy for heterotrophic denitrifiers and acts as an electron donor during denitrification (Kandeler et al., 2006;McCarty & Bremner, 1992;Moser et al., 2018). The lower DOC content of the soil with biochar indicated a reduction in available organic C for heterotrophic denitrifiers, which resulted in a lower abundance of denitrification genes. This is contrary to the results of our meta-analysis showing that biochar addition increased the denitrifying gene abundances ( Figure 5), probably due to the meta-analysis database mainly being derived from short-term experiments. Lower DOC content inhibited soil denitrification, which led to a decrease in N 2 O emissions, an intermediate product of the denitrification process (Liao et al., 2020). This is also supported by the lower potential denitrification rate and relative contribution of nD/bD to N 2 O production, as well as higher NO 3 − concentration. Biochar may also inhibit nitrite and nitrate reductase activities through sorption, thereby reducing N 2 O production driven by denitrification (Spokas et al., 2010;Spokas & Reicosky, 2009). Furthermore, WFPS affects the relative importance of the denitrification process on N 2 O emissions (Loick et al., 2021), and the lower contribution of nD/bD to N 2 O production in the F + BC treatment may be due to the lower WFPS. In general, biochar has a neutral-alkaline pH and may increase soil pH. Interestingly, we found that soil pH was lower in the F + BC than in the F treatment, which agrees with earlier studies (He et al., 2018;Liao et al., 2020). This may be attributed to the increase in carbonyl, carboxyl, and phenolic functional groups on the surface of biochar along with aging, thereby leading to a decrease in soil pH (Seredych et al., 2011;Yao et al., 2011). The lower soil pH in the F + BC treatment further supports the lower nosZ abundance, as low pH conditions are unfavorable for N 2 O-consuming microorganisms (carrying the nosZ gene) (Hallin et al., 2018). However, the higher nosZ/(AOA + AOB + fungal nirK + nirK + nirS) ratio supports the finding that the F + BC treatment tended to have higher Fr and lower r N2O than the F treatment based on N 2 O isotopocule analysis (Figure 2). These results indicate that a higher proportion of N 2 O produced in biochar-added soils was further reduced to N 2 , leading to lower N 2 O emissions. Thus, the legacy effect of biochar application also affects N 2 O emissions by influencing the reduction of N 2 O in the soil. Biochar applied to soil affects microbial communities associated with N-cycling (Harter et al., 2014). Our results indicate that biochar added 7 years ago affected functional microorganisms associated with soil N-cycling and had inconsistent effects on their αdiversity (Table S4). This may be due to the different ecological niches of these microorganisms (Hallin et al., 2018;Hink et al., 2018;Sun et al., 2021), making them inconsistently respond to biochar Huang et al., 2019;Krause et al., 2018;. In addition, recent studies demonstrated the important role of microbial community composition in soil N 2 O emissions (Kits et al., 2019). Previous studies have suggested that changes in the relative abundance of some microorganisms at the level of N-cycling microbial genera are associated with N 2 O emissions . We observed that the presence of biochar altered the relative abundance of N-cycling functional microbial communities at the genus level. Specifically, soils with biochar reduced the abundance of genera involved in N 2 O emissions, Nitrososphaera and Nitrosospira hydrolyzed urea with urease and produced ammonia (Burton & Prosser, 2001;Tourna et al., 2011), and Candidatus_ Competibacter was responsible for the reduction of nitrate to nitrite (Rubio-Rincón et al., 2017). This may support, to some extent, the reduction of N 2 O emissions in the treatment with biochar added. Furthermore, we found that biochar application did not change the functional microbial community structure of N-cycling after 7 years. Note that it cannot be ruled out that soil heterogeneity might have masked the effect of biochar. It has been reported that the reduction of N 2 O emissions by biochar may result from altered soil properties (e.g., pH, available N) rather than microbial communities . Therefore, these findings suggest that changes in N 2 O emissions may not always accompany changes in community structure.

| CONCLUSIONS
In summary, we found a somewhat reduction in N 2 O emissions from soils with biochar compared to the control after 7 years. This is somewhat aligned with the results obtained from the meta-analysis that found an overall significant reduction in N 2 O emissions in the season of biochar application but that it becomes smaller or even reversed over time. Our results suggest that the legacy effect of biochar is to decrease N 2 O emissions by reducing soil substrate effectiveness, nitrification and denitrification activities, and promoting N 2 O reduction. Nevertheless, our results also imply that the legacy effects on soil N 2 O emissions and associated N-cycling processes and microorganisms of biochar application remain to be investigated, owing mainly to the lack of long-term observational data. Considering that biochar is increasingly viewed as one of the promising tools to address climate change, future studies are needed to clarify further the legacy effects on N 2 O emissions from agricultural soils across diverse climate and soil conditions with single or multiple applications of biochar.