In spite of numerous studies on biochar as a strategy to mitigate N2O emissions from soil1,2,3,4,5,6,7,8,9,10,11,12,13,14,15, the factors and mechanisms involved remain elusive. This is the case in part because N2O can be formed through several distinct but often connected processes in soil16, which poses challenges to its quantification due to high spatial and temporal variability17,18.

Although abiotic denitrification has been reported19, research has shown that most N2O emitted from soil is produced by three main biotic processes: denitrification17, nitrification20 and nitrifier denitrification21. These mechanisms may occur simultaneously at different micro-sites of the same soil, but it is generally assumed that most N2O emitted from agricultural lands is produced through denitrification16,17,22,23.

Two principally different pathways can lead to lower denitrification N2O in soil16: (i) a decline in the total N denitrified (with less N2O emitted from soil in the intermediate reaction) or (ii) an enhancement of its further reduction to N2. The second pathway does not minimize total N losses (N2 + N2O), but the ratio N2O/(N2 + N2O) decreases and the environmental consequences of N2O emissions decrease. There are many mechanisms by which biochar might affect these two pathways. Biochar can modify the microbial activity in soil24, the concentration of available NO3 and organic C25,26, pH11,27 and soil aeration28, which are all important factors known to change both the N2O/(N2 + N2O) ratio and the total N denitrified29. At present there is no consensus about what makes a biochar able to mitigate N2O emissions. Most studies found that, in general, slow pyrolysis high-temperature biochars lead to the greatest N2O reductions8,9,10,11,12,13,14. However, there are no studies distinguishing between different N2O production mechanisms or quantifying total denitrification, which makes interpretation and generalization difficult. Therefore, we investigated the causes and the magnitude of denitrification with particular attention to the climate-relevant N2O after biochar addition to agricultural soils.


Does biochar promote or inhibit abiotic denitrification?

We found no N2O emitted from soil under abiotic conditions both in the presence and absence of biochar (Experiment 1; Table S1).

Does biochar reduce N2O emissions during denitrification and by which pathway?

In an incubation study (Experiment 2) with brush biochar and 15 agricultural soils (Tables S2, S3) we found significantly (P < 0.001) lower N2O emissions when biochar had been added (Fig. S1, Table 1). The intensity of mitigation ranged from 10 to 90%. In 10 out of the 15 measured soils, biochar did not only decrease the ratio N2O/(N2 + N2O) but also the total N denitrified. However, this result was less consistent, with five soils increasing the total N denitrified between 4 and 232%.

Table 1 Total N2O emissions from 15 agricultural soils un-amended (control) or mixed with 2% biochar (dry weight basis) under denitrification conditions (90% WFPS and 30°C). Soils were spiked with KNO3 (15N 99% enrichment), which allowed the determination of the total N denitrified (N2 + N2O) and the ratio N2O/(N2 + N2O) by the 15N gas-flux method

In order to discern which soil characteristics influenced the ability of biochar additions to decrease N2O emissions, we performed a multivariate correlation (redundancy analysis) with the soil characteristics (Table S3) as predictor variables and the changes induced by biochar as response variables (Fig. 1). The first two ordination axes explained 49% of the variance. The first axis alone explained a significant part of the variation of the response variables (F = 21.9; P ≤ 0.002). High predictor-response correlations (first canonical axis: 0.915; second canonical axis: 0.654) revealed a strong relationship between soil characteristics and biochar effectiveness to reduce N2O emission. Biochar decreased the ratio of N2O/(N2 + N2O) predominantly in fine-textured soils. However, the ability of biochar to decrease total N2O emissions was independent of soil texture but highly correlated with initial soil NO3 concentrations and dissolved organic C.

Figure 1
figure 1

Correlation triplot based on a redundancy analysis (RDA) depicting the relationship between the main physico-chemical characteristics of the soils (predictor variables) and the differences induced by biochar applications (response in soil) (according to Lepš and Šmilauer49).

Blue arrows point to maximum shifts produced by the biochar amendment, i.e. a decrease in the total cumulative N2O, the N2O/(N2 + N2O) ratio and the flux of total N denitrified (N2 + N2O). Eigenvalues of the first two axes are 0.343 and 0.161, the sum of all canonical axes is 0.555. “Cumulative N2O” represents the difference (control-biochar) in total N2O emitted during the entire incubation period; “ratio” and “Total N denitrified” represent the differences (control-biochar) at the day selected for isotopic gas analysis (see Fig. S1). Tsilt, Tclay and Tsand represent the percentages of soil silt, clay and sand. DOC: dissolved organic C in soil.

Why does biochar reduce N2O emissions?

In Experiment 3 we measured N2O emitted after soil amendment using nine biochars with different C/N ratios (Table S2; Fig. S2) after their pH had been adjusted to the same pH as the soil (pH: 5.6) (Fig. 2.A). Then, we repeated the incubation with the same biochars added without adjusting the pH (Fig. 2.B). The difference between Fig. 2.A and Fig. 2.B corresponds to the effect of biochar pH on total emissions. We found that this difference strongly correlates (r2 = 0.809; P < 0.01) with biochar buffer capacity (i.e. mmol H3O+ per gram of biochar necessary to adjust its pH to the same pH of the soil), but not with pH alone (r2 = 0.615; P > 0.05).

Figure 2
figure 2

Total N2O emissions after 30 days of incubation of a muck soil (Elba) amended with different biochars (2% weight) under denitrification conditions (90% WFPS, 30°C).

The dashed line represents emissions from the control soil (unamended). Fig. 2A shows N2O emissions from soil amended with biochars for which the pH had been adjusted to the pH of the soil (5.6). Fig. 2.B shows N2O emissions from soil amended with biochars at their actual pH. Biochars are arranged from high to low C/N ratios. Error bars represent standard errors of the mean (n = 4).

Given the known effect of pH on N2O production, we performed a parallel incubation (Experiment 4) to monitor the shift in soil pH induced by the different biochar additions. However, only dairy manure biochar led to a statistically significant yet low increase of 0.1 pH units (Table S4). We then analyzed N2O emissions from the same soil where the pH had been experimentally increased using additions of CaCO3 (Fig. S3) and observed no N2O mitigation (total N2O emissions after 13 days: 38.7 ± 4.4 for control (pH 5.60), 44.5 ± 3.9 for pH 5.79 and 45.5 ± 5.5 for pH 6.10, respectively).

Is it a short-term effect?

After 30 days of incubation (Experiment 3), N2O fluxes had leveled off in all treatments (Fig. S2) pointing to a depletion of NO3 or available organic C in soil. At day 34 we added 100 mg of NO3-N and 1.0 g of glucose-C per kg of soil and production of N2O increased immediately (Experiment 5; Fig. S4). When pH was adjusted, only two of the biochars (pine and oak) significantly reduced N2O emissions (Fig. 3A). When pH was not adjusted, however, most biochars decrease the total amount of N2O emitted (between 41 and 72%), although only additions of biochars made from bamboo and oak had significantly lower emissions compared to the control at P < 0.05 (Fig. 3B). It is important to note that the different treatments had undergone 30 days under denitrification conditions and their initial NO3 concentration (before spiking them with NO3 and glucose) might significantly differ.

Figure 3
figure 3

Total cumulative N2O emissions produced after 7 days of incubation of a muck soil (Elba) spiked with 100 mg NO3 -N and 1 g of glucose-C per kg of soil.

The soil had been incubated with different biochars (2% weight) under denitrification conditions (90% WFPS, 30°C) during 1 month prior to N and C addition. The dashed line represents emissions from the control soil (without biochar). Fig. 1A shows N2O emissions from soil amended with biochars for which the pH had been previously adjusted to the pH of the soil (5.6). Fig. 1.B shows N2O emissions from soil amended with biochars at their actual pH. Biochars are arranged from high to low C/N ratios. Error bars represent standard errors of the mean (n = 4).

In order to more closely investigate the mechanism of the observed medium-term mitigation we performed a second experiment where three soils (Secanos, Tioga and Elba) were pre-incubated (with and without biochar) for 14 days under denitrification conditions (Experiment 6) and subsequently spiked with a solution of K15NO3. After the pre-incubation period, the soils with added biochar showed far more NO3 compared to the soils without added biochar (Table 2). Even after the 15NO3 addition, the alkaline soil (Secanos) hardly showed any denitrification. The reason may be a lack of available C after the pre-incubation period, since this soil had a markedly lower total organic carbon concentration (8 g C kg−1 soil) compared to Tioga (29 g C kg−1 soil) or Elba soils (495 g C kg−1 soil). Both Tioga and Elba soils emitted N2O, but without any detectable N2. Recognizing the different initial NO3 concentrations of soils with and without biochar additions, we calculated the N2O emitted (as a result of the 15NO3 spike) per unit of NO3 in soil and observed a comparable mitigation effect for the Tioga soil and an even larger one for the Elba soil (Table 2) compared to Experiment 2 (Table 1), where soils were not pre-incubated.

Table 2 Influence of biochar (made from brush at 500°C) on N2O emissions from three different soils after a preincubation period (2 weeks)


Although biological processes dominate N2O production in most environments, chemodenitrification, an abiotic process wherein inorganic N species are reduced to gaseous species has been reported in soils with high concentrations of Fe (II) or humic acid extracts19. We did not find N2O emitted under abiotic conditions in the two soils most susceptible to chemodenitrification used in this study (Elba and Guarapuava), which implies that the N2O emitted from the soils was entirely produced through biotically mediated mechanisms. It also shows that biochar did in our study not abiotically induce N2O formation in soil through previously proposed catalytic reactions with hydroquinones30,31, metal ions or radicals19,32, which are abundant in biochars31,33.

Previous studies on N2O emissions and biochar are limited to one or two soil types3,4,7,9,10,12,13,34,35. Given that these experiments have been conducted under varying environmental conditions and with different biochars, comparison between soil properties is difficult, which enormously limits generalization. Our study revealed that soil texture was closely related to the ability of biochar to decrease the ratio of N2O/(N2 + N2O) at the time of maximum N2O emissions (Fig. 1). The fact that biochar promoted the last step of denitrification predominantly in fine-textured soils (e.g. Lentiscosa, Madalin, Costa) indicates that the mechanism of reduction is not linked to an increase in soil aeration (if that were the case, biochar would decrease total denitrified N in these soils, instead of promoting the last step of denitrification to N2). This confirms recent findings6, which noticed soil aeration to be a negligible factor for N2O mitigation in soils containing biochar.

The strong correlation between NO3 concentration in the soil and total N2O mitigation by biochar additions (Fig. 1) suggested that biochar might reduce NO3 availability. A reduction in NO3 availability would indeed decrease the total N denitrified and it would favor the last step of denitrification (decreasing the ratio N2O/(N2 + N2O))36. On the other hand, N2O mitigation was also highly correlated with soil DOC and less strongly with soil pH, which are known to control the denitrification capacity in soil37.

In Experiments 3 and 4 we further investigated these possible reasons for the observed reductions in N2O emissions. We postulated that by using a soil where the addition of biochar does not significantly influence its water filled pore space, the effect of biochar on denitrification N2O would mostly depend on biochar pH and its C/N ratio. Most biochars used in this study may immobilize NO3, since their C/N ratios are greater than that of the soil (C/N = 18.7). Nitrogen immobilization in soil has often been found to decrease denitrification N2O38. Despite their high C/N ratios, we did not find a correlation between N2O mitigation and biochar C/N (range of C/N ratios: 11–859) irrespective of whether or not the pH of the biochars was adjusted. The high recalcitrance of high-temperature biochars to microbial degradation1,8,27 and the lack of correlation between biochar C/N and N2O mitigation found in our experiment do not suggest microbial immobilization of NO3 as the driving mechanism for the observed N2O reductions.

When the pH was adjusted, additions of high-N biochars led to more emitted N2O (Fig. 2B), which is probably due to an increase of NO3 concentrations in the soil. Although some N from the biochar may be mineralized, this does normally not exceed 10–20% of its total N content15. A salting-out effect through an increase of the ionic strength in the soil solution, which has shown to reduce the solubility of N2O (Setchenow effect) favoring its emission22, might have played a role for the differences among biochars. The correlation between ash content and N2O emissions was highly significant (r2 = 0.809; P = 0.002). However, the persistence of the reduction in denitrification found in Experiment 6 sheds doubts on this explanation.

A mere shift in soil pH was not the reason for the mitigation of N2O emissions as demonstrated in Experiment 4 (Fig. S3). Our finding is in agreement with those by Yanai et al.2, who after increasing the soil pH with ash applications (instead of biochar) did not observe any reductions in N2O emissions. This suggests that the change in soil pH does not by itself induce the N2O reductions, but rather other properties of biochar intrinsically connected to pH.

Despite its refractory nature, biochar contains abundant redox-reactive organic and inorganic compounds31. For example, quinone groups may be reduced during suboxic conditions to hydroquinone or phenols that can subsequently act as electron donors. Such redox systems readily changing from one steady-state to another31 have been described for other forms of organic matter in soil, such as humic substances extracts39 or plant and microbial exudates40 and are known as “electron shuttles”. The presence of free radicals in some biochars may even increase their reactivity31.

Under suboxic conditions, facultative anaerobic microorganisms can utilize NO3, Mn (IV) and Fe (III) compounds substituting O2 as electron acceptor41. Biochar contains both Mn and Fe at varying concentrations depending on the feedstock and pyrolysis temperature27. For example Enders et al.27 reported Mn and Fe concentrations of up to 9% (d.w.) in biochars produced at 500°C. Mn (IV) and Fe (III) are known to readily function as electron acceptors at circumneutral pH42. Biochar might act as a reducing agent itself and additionally as an electron shuttle, facilitating the electron transfer to microorganisms by acting as an electrical conduit. Thus, biochar might compete with NO3 as an electron sink, which could explain the lower total N denitrified in many soils. On the other hand, its function as an electron shuttle connected to its liming effect might facilitate the activity of N2O reductase and therefore promote the last step of denitrification29, both hypotheses that need further testing.

Another controversial aspect of N2O mitigation with biochar is the debate about its transitory effect. In a recent study Spokas43 found that three years ageing negated the N2O mitigation that was originally observed from the fresh biochar in laboratory incubations. Ageing is known to substantially alter biochar surface chemistry and reactivity31, which may impact the ability of biochars to function as an electron acceptor or shuttle. Experiment 5 showed that N2O mitigation is effective one month after biochar application and it is still strongly related to pH, albeit not a result of pH changes as indicated above. One of the hypotheses proposed for temporary N2O mitigation is that, immediately after application, biochar might have a short-term inhibitory effect, decreasing the microbial activity in soil. For example, Spokas et al.35 found a correlation between ethylene concentration and N2O production in soil amended with biochar and proposed the inhibitory impact of ethylene on microbial biomass as a mechanism that could significantly contribute to N2O mitigation. This mechanism may be significant for biochars produced at low temperatures, with a high concentration of tars, which might temporarily reduce the activity of denitrifier communities in soil. Nonetheless, for biochars produced at 500°C, the contribution of this mechanism appears to be minor. Looking more closely at the mechanism of medium-term mitigation in Experiment 6, the soils pre-incubated with biochar showed far more NO3 compared to the control soils (Table 2). These results stand in stark contrast to the hypothesis that decreasing N2O emissions were caused by the ability of biochar to adsorb NO3 in soil6,13. Even though we investigated medium-term effects after several weeks separately from effects observed immediately after addition of fresh biochars to soil, aging of biochars over periods of months and years may affect the results43. Additional long-term studies on N2O mitigation with biochar are required to quantify the duration of this effect31,43.

In summary, under optimum denitrification conditions, biochar consistently reduced N2O emissions in the investigated agricultural soils. It decreased the ratio of N2O/(N2 + N2O) and in most cases also the total N denitrified. Biochars obtained at 500°C by slow pyrolysis, independently of their original feedstock, were able to decrease N2O fluxes produced by denitrification. In light of our results, we discard biochar toxicity, NO3 immobilization or NO3 sorption as relevant mechanisms for N2O mitigation with high-temperature slow-pyrolysis biochar. The complexity of biotic and abiotic biochar-soil interactions points at several mitigation mechanisms occurring simultaneously. Biochar buffer capacity appears to be fundamental to decreasing N2O emissions during denitrification, not because of a change of soil pH in itself, but because the mechanism of mitigation is intrinsically connected to pH. We have demonstrated that biochars promote the last step of denitrification and in two out of three cases it also decreased total N denitrified. We propose as a plausible explanation that biochar is able to facilitate the transfer of electrons to denitrifying microorganisms in soil, thus acting as an electron shuttle. This together with its liming effect and high surface area would promote the reduction of N2O to N2. Our results are based on short-term laboratory assays that did not take into account other important aspects occurring under field conditions, including biochar ageing or soil-plant-microbe interactions. The results of our study allowed us to formulate a new hypothesis that had not been considered before and that might play an important role explaining reductions in N2O formation when denitrification is the dominant pathway. This hypothesis requires further experimentation to prove its magnitude and eventually its practical significance.



Nine biochars produced at 500°C by slow pyrolysis as described in Enders et al.27 were selected with the aim of obtaining a wide range of C/N ratios from 11 to 859 and pH values between 6.4 and 10.7 (Table S2). Biochars were ground and sieved to a particle size between 200–500 μm before soil application. Brush biochar was used in all experiments and selected for those involving different soils (Experiments 1, 2 and 6) because of its neutral pH and also because it can be considered as representative of a generic biochar widely utilized and available worldwide.


Fifteen agricultural soils from 3 different geographical areas (USA, Spain and Brazil) were selected comprising a wide range of textures and pH values (Table S3). The soils were sampled from a depth of 0–0.25 m of agricultural fields, air-dried and sieved (<2 mm). Agricultural organic soils have been reported to emit high fluxes of N2O44. In experiments involving different biochars (Experiments 3 and 5) we therefore selected a muck soil (drained cultivated Histosol) in order to obtain high N2O emissions that allowed us to detect the effects of different biochars more sensitively. If an effect (positive or negative) was not observed at high emissions, biochar would not play a role. In addition, this organic soil has a high porosity, which allowed us to assume that biochar additions would not modify its water filled pore space. For practical reasons the collected muck soil was maintained in its field moist condition, sieved (<2 mm) and stored at 4°C until the beginning of the experiment.

Chemical-physical analyses of biochars and soils

Biochars: Proximate analysis was conducted using ASTM D1762-84 Chemical Analysis of Wood Charcoal. Total C and N were determined by Dumas combustion using a PDZ Europa ANCA-GSL elemental analyzer connected to a PDZ Europa 20–20 isotope ratio mass spectrometer (Sercon Ltd., Cheshire, UK). pH and electrical conductivity (EC) were determined in 1:10 (w:v) water-soluble extracts. NH4+ and NO3 were extracted with 2.0 M KCl at 1:10 (w:v) and quantified colorimetrically using a continuous flow analyzer (Bran and Luebbe Autoanalyzer, SPX, Charlotte, NC). Soils: Soil texture was determined using the pipette method according to Kettler et al.45. pH was determined in 1:10 (w:v) water extracts. Dissolved organic C (DOC) and total dissolved N (TDN) were determined in 1:10 (w:v) water extracts (shaken for 2 h, centrifuged for 10 min at 1500×g and filtered (GF/F Whatman glass filters) with a TOC analyzer (Shimadzu Total Organic Carbon-Visionary Series; TOC-VCSH).

Incubation studies

The incubation experiments were performed with units consisting of 100 g dry soil (control) or 98 g dry soil and 2 g biochar (treatments) in 250 ml (or 500 ml in the case of the muck soil) glass jars at optimum conditions for denitrification: 30°C and a moisture of 90% of the water filled pore space (WFPS). The biochars were thoroughly mixed with the dry soil to obtain a completely homogeneous mixture. Subsequently, water (or a solution containing the appropriate concentration of NO3) was added to attain the required moisture. The jars were covered with a polyethylene sheet that allows gas exchange but minimizes evaporation. Moisture was kept constant by adjusting the water content every other day. The experiments were laid out as a randomized block design with four replicates.

Experiment 1

This experiment was aimed at quantifying any abiotic contribution to N2O formation (chemodenitrification) from the two soils most susceptible to chemodenitrification in this study (Elba and Guarapuava; total organic C: 495 and 43.5 g kg−1, respectively). 50 g of Elba or 100 g Gurapuava soils were either untreated (controls) or treated with 2% biochar (made from brush), subsequently placed in 250-ml jars at 90% WFPS and doubly autoclaved for sterilisation (103.4 kPa and 121°C for 1 h, incubated 24 h and autoclaved for an additional 1 h). The samples were cooled down to 30°C overnight and N2O concentrations were measured by means of gas chromatography equipped with an electron capture detector (ECD) 12 h after sterilization. Since no N2O was emitted from any of the samples, a sterile solution of KNO3 was added (100 mg N kg−1 soil) and the N2O accumulated in the headspace during 40 min was immediately measured (Table S1).

Experiment 2

This study was designed to determine the magnitude of N2O mitigation by biochar in different soils as well as investigating the mechanism involved. Brush biochar was applied at an equal rate (2% dry weight) to 15 different agricultural soils. Moisture was adjusted to 90% WFPS in each jar by adding the required volume of a solution containing KNO3 (>99% 15N enrichment) at the appropriate concentration to obtain 90% WFPS and exactly 5 mg of NO3-15N-per jar (50 mg of NO3-15N kg−1 soil). Rewetting the soils in this fashion guaranteed a homogenous 15N-NO3 pool, which is essential to correctly apply the equations derived by Mulvaney and Boast46 to calculate total N denitrified. We measured N2O fluxes from the 15 unamended soils compared to those amended with 2% biochar (Fig. S1).

Experiment 3

The organic soil (Elba) was sampled from a depth of 0–0.25 m of a horticultural field in Genesse County, NY. We assumed that the addition of biochar (at 2%, w:w) to this soil would not modify its aeration at 90% WFPS. With the aim of verifying this assumption we tested its water retention curves (Fig. S6) and observed no significant differences among treatments.

To differentiate the effects of pH values from C/N ratios we carried out two incubations. In the first one the pH of the biochars was adjusted to the same pH of the soil (5.6) prior to application. The pH adjustment was done by shaking each biochar with water (1:20, w:v) for 12 hours; pH was measured and adjusted to 5.6 first with a 2 M HCl solution and later with a 0.1 M HCl solution (close to the end point). The volume of HCl solution necessary to adjust the pH was registered in order to calculate the biochars' buffer capacity (i.e. mmol H3O+ per gram of biochar necessary to adjust its pH to 5.6). Once the pH had been adjusted, the biochars were shaken for 12 hours and the pH was tested again. Biochars were then dried for 2 days at 80°C and homogenized before their addition to soil. We also checked that the pH adjustment did not modify the amount of water-extractable NO3 in biochars. In the second incubation the biochars were applied without any pre-treatment. In both cases biochars were applied at 2% (w:w) and thoroughly mixed with the soil, which was re-packed to the average field bulk density (0.65 g l−1) inside the jars. De-ionized water was added to adjust the moisture to 90% WFPS, equivalent to 2.07 g water per g dry soil. N2O fluxes were analyzed during one month of incubation (Fig. S2).

Experiment 4

In this experiment we investigated how the manipulation of soil pH influences N2O emissions. First we evaluated the magnitude of pH changes induced by the biochars in Experiment 2. We carried out a parallel incubation (2 replicates per treatment) under identical environmental conditions where we monitored the pH after 1, 3 and 14 days of incubation (Table S4). Second, we performed an incubation study with the same soil where we modified its pH in the same order of magnitude as was induced by biochars measured in the preceding experiment. We included a control (Elba soil at its natural pH of 5.6) and two treatments where we increased the pH of the soil to 5.79 and 6.10 by adjusting to 90% WFPS with water solutions containing 0.010 and 0.020 g of CaCO3 per 100 g of dry soil, respectively. We compared the N2O emissions over two weeks (Fig. S3).

Experiment 5

This experiment was an extension of Experiment 3 aimed at studying the temporal dynamics of biochar effects. After 1 month of incubation N2O fluxes were low in all samples (Fig. S2). At day 34 we applied 5 ml of a solution containing KNO3 (2 g N L−1) and glucose (20 g C L−1), which is equivalent to 100 mg N and 1 g C kg−1 soil and measured the N2O fluxes resulting from this addition (Fig. S4).

Experiment 6

Two mineral soils with a loamy texture and contrasting pH (Secanos, Tioga) and the organic soil (Elba) were selected for this experiment. Each control soil and its corresponding biochar treatment (2%, w:w) was pre-incubated for 14 days under denitrification conditions (90% WFPS, 30°C). From day 13 to day 16 the samples were left to dry to ca. 50% WFPS, which was verified gravimetrically. At day 16, moisture was re-adjusted to 90% WFPS in each jar by adding the required volume of a solution containing KNO3 (>99%15N enrichment) at the necessary concentration to obtain 5 mg of NO3-15N per jar (similarly to Experiment 2) (Fig. S5).

N2O sampling and measurements

Gas sampling was conducted by sealing each unit with screw caps for 40 min. 10 ml of the headspace gas was sampled with 25 ml gastight polypropylene syringes and measured within 12 hours by gas chromatography (Shimadzu GC-14A GC equipped with ECD (Ni63) detector (Kyoto, Japan)). Measurements were done daily during the first three days; decreasing subsequently to every other day, three times per week, etc. (see Fig. S1, Fig. S2, Fig. S3 etc. in the supporting information section).

N2O fluxes were calculated assuming a linear increase during the accumulation (closing) period, a fact that was checked prior to the experiments (every 15 minutes for 1 hour). Cumulative N2O was calculated assuming linear changes in fluxes between the two closest measurement points.

In experiments with 15NO3 (Experiments 2 and 6) the 15N gas-flux method46,47,48 was used to quantify N2O and N2 emissions. The gas sampling for isotopic analysis was made each day preceding the gas sampling for GC-ECD analysis and within an independent accumulation period. Two gas samples were collected using a 12-ml syringe and needle: one immediately after the lid was fitted to the jar (t = 0) and the second after 40 min (t = 40). The gas samples were transferred to 12-ml vials (Labco) previously purged with He and evacuated. A posteriori, gas samples were selected at time points where the difference in N2O fluxes (measured by GC-ECD) between biochar and control soils were the greatest, which normally corresponded with the peak in N2O emissions. Selected samples (a total of 320) were analyzed for their isotope ratios of N2 (29/28 (29R) and 30/28 (30R)) and N2O (45/44 (45R) and 46/44 (46R)) by automated isotope ratio mass spectroscopy (ThermoFinnigan GasBench & PreCon trace gas concentration system interfaced to a ThermoScientific Delta V Plus isotope-ratio mass spectrometer (Bremen, Germany)).

Data calculations and statistics

The molar fraction of 15N-NO3 (15XN) in the soil pool was calculated from the Δ45R and Δ46R according to Stevens and Laughlin48. The flux of N2 and N2O was then calculated by the equations given by Mulvaney and Boast46.

Since data were not normally distributed, they were ln-transformed prior to univariate analysis of variance with SPSS 19.0. The correlation between soil properties and biochar effectiveness in reducing emissions was determined by redundancy analysis (RDA) using CANOCO 4.5 for Windows. The characteristics of the soil (silt, clay, sand, pH, DOC, NO3, NH4+) were included as predictor variables and the (i) total N2O emissions, (ii) flux of total N denitrified and (iii) N2O/(N2 + N2O) ratios between controls and biochar treatments as the dependent variables. Data were centered and standardized. Significance of the ordination axes was calculated by the Monte-Carlo permutation test; n = 499.