Increased CO2 fluxes from a sandy Cambisol under agricultural use in the Wendland region, Northern Germany, three years after biochar substrates application

In recent years, biochar has been discussed as an opportunity for carbon sequestration in arable soils. Field experiments under realistic conditions investigating the CO2 emission from soil after biochar combined with fertilizer additions are scarce. Therefore, we investigated the CO2 emission and its 13C signature after addition of compost, biogas digestate (originating from C4 feedstock) and mineral fertilizer with and without biochar (0, 3, 10, 40 Mg biochar/ha) to a sandy Cambisol in Northern Germany. Biomass residues were pyrolized at ~650°C to obtain biochar with C3 signature. Gas samples were taken biweekly during the growing season using static chambers three years after biochar substrate addition. The CO2 concentration and its δ13C isotope signature were measured using a gas chromatograph coupled to an isotope ratio mass spectrometer. Results showed increased CO2 emission (30%–60%) when high biochar amount (40 Mg/ha) was applied three years ago together with mineral fertilizer and biogas digestate. On average, 59% of the emitted CO2 had a C3 signature (thus, deriving from biochar and/or soil organic matter), independent of the amount of biochar added. In addition, our results clearly demonstrated that only a small amount of released CO2 derived from biochar. The results of this field experiment suggest that biochar most likely stimulates microbial activity in soil leading to increased CO2 emissions derived from soil organic matter and fertilizers mineralization rather than from biochar. Nevertheless, compared to the amount of carbon added by biochar, additional CO2 emission is marginal corroborating the C sequestration potential of biochar.

CO 2 concentration increases (Glaser & Stoknes, 2014;Lal, 2004). Aside from reduced soil C stocks caused by human activities, some soils have low soil C stocks due to natural limitations. Examples are sandy soils in the North German Plain, formed during the late Quaternary by glacial and periglacial deposits, being poor in soil organic matter and nutrients. Only high amounts of mineral fertilizer allow agricultural use of these disfavored regions although being ineffective.
In recent years, the idea of transferring atmospheric CO 2 into stable soil C pools was triggered by studies about charcoal-rich Anthrosols such as Terra Preta de Indio in the humid tropics of Brazil (Glaser, 2007;Glaser & Birk, 2012;Glaser, Haumaier, Guggenberger, & Zech, 2001). Soils, similar to Terra Preta are also described in other climate regions, which leave no doubt that charcoal may last in soil over millennia Wiedner, Schneeweiß, Dippold, & Glaser, 2015). However, stabilization mechanisms of pyrogenic carbon are still poorly understood but most likely a combination of chemical recalcitrance and physical protection takes place (Glaser, Balashov, Haumaier, Guggenberger, & Zech, 2000;Kuzyakov, Bogomolova, & Glaser, 2014;Marschner et al., 2008). Furthermore, type of feedstock and charring conditions (e.g., temperature and duration) heavily influence physical and chemical properties of charred particles (biochar) and thus their fate in soil environment (Schimmelpfennig & Glaser, 2012;Wiedner et al., 2013). Nevertheless, the presence of oxygen-containing functional groups on surfaces of ancient biochar indicates degradation processes, which are probably affected by the complex interaction of abiotic and biotic factors such as soil moisture, soil temperature, soil microbial community structure, and land use practices (Wang, Xiong, & Kuzyakov, 2016;Wiedner, Fischer, et al., 2015).
Most of the studies dealing with biochar stability in soil environment are performed in laboratory experiments or extrapolations based on investigations of fire-affected soils (He et al., 2017). For instance, half-life of biochar exposed in temperate rainforests are estimated up to 6623 years (Preston & Schmidt, 2006). Long-term mean residence time (MRT) of biochar in two different savanna regions were estimated up to 1300 and 2600 years (Lehmann et al., 2008). In contrast, Hammes, Torn, Lapenas, and Schmidt (2008) calculated a turnover time of biochar in a Russian steppe soil of only 293 years and Bird, Moyo, Veenendaal, Lloyd, & Frost, (1999) report a half-life of charcoal in savannah soil of <100 years. A meta-analysis by Wang et al., (2016) revealed that the decomposed amount of biochar increased with experimental duration, but the decomposition rate decreased with time and MRT of biochar was estimated to 556 AE 483 years. The examples listed show clearly that biochar decomposition differs between different climates, soil types, experiment duration, feedstock, and biochar production conditions (Lorenz & Lal, 2014;Wang et al., 2016).
A further complication is the big knowledge gap on biochar interactions and stability combined with other fertilizers under realistic field conditions. It is mandatory to understand and characterize the behavior of biochar in arable soils to estimate its potential of long-term carbon sequestration. Thus, we performed a large-scale field experiment under both practical agronomic conditions and scientific requirements to investigate long-term C sequestration potential of complex biochar fertilizers in a sandy soil under temperate climate conditions. For this purpose, different amounts of biochar (0, 3, 10, 40 Mg/ha) were co-applied with compost, biogas digestate and mineral fertilizer to determine the release of CO 2 after three years of realistic agronomic land use. Furthermore, the emission source will be identified using stable carbon isotope signature (d 13 C). The study addresses the following three research questions: 1. Does biochar increase CO 2 emissions when combined with mineral fertilizer, biogas digestate, or compost? 2. Are CO 2 emissions different when low or high amounts of biochar (3 or 40 Mg/ha) were applied to mineral fertilizer and biogas digestate? 3. Does the emitted CO 2 derive from biochar, soil organic matter, and/or applied fertilizers? 2 | MATERIALS AND METHODS

| Study site and agricultural management
The large-scale field experiment is located near Gartow at 53°1.154 0 N and 11°29.834 0 E, 19 m above sea level in the eastern part of Lower Saxony, Germany. The experiment was established and run under agronomic practice since the end of May 2012 (Glaser, Wiedner, Seelig, Schmidt, & Gerber, 2015). Gartow has a mean annual temperature and precipitation of 8.8°C and 575 mm, respectively . During the growing season 2014 (April to August 2014), mean temperature and precipitation were 16.2°C and 271 mm, respectively (DWD Climate Data Center, 2015). The sandy soil was classified as Stagnic Cambisol resulting from Quaternary dynamics  with a d 13 C isotope signature of À26.7 mUr and a soil organic carbon content of 0.6%.
The field experiment was designed as a Latin rectangle consisting of 50 plots in total, divided into ten different treatments with five replicates . Treatments comprised mineral fertilizer, biogas digestate, fermented biogas digestate (inoculated with indigenous microorganisms, which were extracted from neighboring forest soils), and compost produced from local biomass residues. In addition, biochar was added to these pure fertilizers either annually in low amount (1 Mg biochar/ha) or once in high amount (10 Mg biochar/ha for compost and 40 Mg biochar/ha for the other treatments). Before application, all fertilizers were adjusted to common practice nitrogen levels depending on the cultivated crop. All field management activities were carried out according to agronomic practice.
Biochar was made out of nutrient-poor biomass residues produced by pyrolysis in a PYREG reactor at~650°C (PYREG GmbH, D€ orth, Germany) with the following properties: pH CaCl2 8.6, electrical conductivity 1000 lS/cm, ash content 12.6%, total carbon content 71.3%, total nitrogen content 1.0% (Wiedner, Fischer, et al., 2015). Further information about the preparation and properties of the fertilizers are published by Glaser et al. (2015).
From May 30, 2012 to September 24, 2012, silage hybrid maize (Zea mays, variety KALVIN, Syngenta Agro GmbH, Maintal, Germany) was cultivated. Hybrid winter rye (Secale cereale, variety BRASETTO, KWS SAAT SE, Einbeck, Germany) was sown after maize harvest at the mid of October 2012. On April 11, 2013, a treatment-specific fertilization on a fixed nitrogen level of 120 kg/ha was performed. Three weeks after the rye harvest on July 24, 2013, a mixture of 13 catch crops was sown. Before the third fertilization on March 23, 2014, the catch crops were chopped by a rotary hoe and subsequently incorporated by a shallow plow. Due to cultivation of the legume narrow-leafed lupine (Lupinus angustifolius, variety BORE-GINE, Saatzucht Steinach GmbH & Co. KG, Steinach, Germany) in 2014, nitrogen application was reduced to 36 kg/ha for each treatment. For this purpose, each plot was fertilized uniformly with 252 L biogas digestate (Raiffeisen Warengenossenschaft Jameln eG, Jameln, Germany) mixed with 0.37 kg elemental sulfur (90% S) except the plots treated with mineral fertilizer. The mineral fertilizer mixture for each of the conventional farming treatments consisted of 2.0 kg KALISOP (50% K 2 O and 45% SO 3 ) from K+S Kali GmbH, 0.35 kg MgSO 4 , 0.40 kg quicklime, and 14.6 L Organic Plant Feed (9% N, 2% P 2 O 5 and 2% K 2 O) from Plant Health Cure BV. As in the previous two years of the field experiment, the treatments with annual biochar application received 77 kg biochar (47% DM), corresponding to 1 Mg biochar/ha (Biochar (3) digestate and biochar (3) mineral fertilizer). Immediately after the application of fertilizers, a shallow plow was used for fertilizer incorporation and narrow-leafed lupine was sown with a spring-tooth harrow. In 2014, no irrigation or application of herbicide or fungicide was performed during the whole growing season. A tined weeder was used once to control weeds mechanically during the growing season. The lupine harvest was on August 4, 2014.
2.2 | Gas sampling CO 2 released from the sandy soil was sampled biweekly from April 1, 2014 to August 5, 2014, using static PVC-U chambers (Gebr. Ostendorf Kunstoffe GmbH, Vechta, Germany). To avoid boundary effects, for example, of neighboring plots, each chamber was placed in the plot center ( Figure 1a). The chambers were 0.55 m long with 0.11 m inner diameter ( Figure 1b). Each chamber was inserted 0.15 m deep into the soil (Figure 1b). Therefore, the chamber covering an area of 0.01 m². Together with the remaining 0.40 m length above the soil surface, each chamber had an aboveground cylinder volume of 3.8 dm 3 .
The gas samples were transferred through multisample needles (Ø 0.8 9 38 mm) (Vacutest Kima S.r.l., Piove di Sacco, Italy) to sealed and pre-evacuated 0.2 dm 3 headspace glass vials, which were closed by aluminum crimp caps combined with butyl hollow stopper (IVA Analysentechnik GmbH & Co. KG, Meerbusch, Germany). Gas samples were taken immediately after chamber closure as well as after 10, 20 and 30 min. To avoid measurement of root respiration, chambers were kept free of vegetation. According to Parkin and Kaspar (2003), gas sampling was carried out during 11 am to 1 pm to get representative results reflecting the daily mean emissions.

| Instrumentation and calculations
CO 2 measurements were performed using a Trace GC Ultra gas chromatograph (Thermo Fisher Scientific, Milan, Italy) additionally equipped with a CombiPAL autosampler (CTC Analytics, Zwingen, Switzerland). Carbon dioxide concentration and stable C isotope composition was detected by a Finnigan Delta V Advantage Isotope Ratio Mass Spectrometer (Thermo Fisher Scientific, Bremen, Germany), coupled via a ConFlo IV universal interface (Thermo Fisher Scientific, Bremen, Germany) to the Trace GC Ultra gas chromatograph.
Injections were carried out using a 2.5 cm 3 gas-tight headspace syringe SYRC HS2.5-23-5 (CTC Analytics, Zwingen, Switzerland) in split mode (split ratio of 5 and split flow of 13 cm 3 /min) through a glass inlet liner (TQ CE 5 mm inner diameter, SGE Europe Ltd., Milton Keynes, UK). The injector block was heated at 100°C. To remove potential isobaric interferences (e.g., N 2 O), a chromatographic separation of CO 2 was carried out using a Carboxen 1010 PLOT column (30 m, 0.32 mm internal diameter, Supelco, Bellefonte, PA, USA) with a constant helium flow (99.9997%, pure) of 2.5 cm 3 /min. The temperature program started at 40°C (1.00 min holding time), heated up to 100°C at 30°C/min (holding time 2.50 min) ended with 230°C at 100°C/min (holding time 2.00 min).
ISODAT 3.0 (Thermo Fisher Scientific) and EXCEL 2013 (Microsoft, Redmond, WA, USA) were used for data processing. An external standard (IAEA-CO-8, À5.764 mUr, Vienna Pee Dee Belemnite (VPDB), International Atomic Energy Agency (IAEA), Vienna, Austria) was analyzed with each measurement and used for carbon concentration and isotope calibration. To produce CO 2 from these inorganic solid materials, orthophosphoric acid (85%, Baker analyzed, Mallinckrodt Baker, Deventer, the Netherlands) was added to the argon (99.999%, pure) purged vial, which contained weighted portion of the reference standard. This procedure was performed for three different amounts of the standard (30, 60, and 90 lg). d 13 C signature was calculated according to Craig (1957) and all isotope signatures are expressed by the SI-compliant unit Urey (Ur) as recommended by Brand and Coplen (2012) where R is the 13 C/ 12 C ratio of the standard or the sample and d describes the relative isotope ratio of the sample relative to the standard IAEA-CO-8 in mUr. d 13 C measurements were drift-and amount dependencecorrected according to Zech & Glaser, 2008. Calculation of the CO 2 -C fluxes was performed using Equation (2) (Flessa, Wild, Klemisch, & Pfadenhauer, 1998; modify by Beetz et al. 2013): where F CO 2 -C is the flux rate of CO 2 -C (mg CO 2 -C m À2 hr À1 ), k CO 2 (0.536 kg C/m 3 ) serve as conversion factor for the ideal gas at 273.15 K, T is the daily mean air temperature (K), V is the volume of the chamber (m 3 ), A is the basal area within the chamber and DC Dt À1 is the concentration change of CO 2 -C over time (ppm v per hr) in the headspace of the chamber obtained by linear regression. For calculating the cumulative CO 2 -C fluxes, no linear interpolation between the sampling days was used, because of the strong variations under field conditions and unavailable data of needed parameters (e.g., soil temperature, soil moisture). Therefore, an interpolation would overestimate the actually emitted carbon dioxide.
In general, C3 and C4 vegetation shows average values of À27.7 mUr or À13.5 mUr (Troughton, Card, & Hendry, 1974). The isotope composition of CO 2 -C makes it possible to determine different sources of CO 2 -C emissions using a two source-mixing model (Fry, 2006) as shown in Equation (3): (3) where the C3-derived fraction (including soil organic matter and biochar) or the C4-derived one (containing biogas digestate) in percent is calculated by isotope signatures of the sample and the two emission sources. We are aware that isotope fractionation during CO 2 production through soil organic matter decomposition occurs. Isotope fractionation is highly complex and strongly influenced by several variables such as soil organic matter quality, water content, or the carbon-nitrogen ratio of soils (Wang, Jia, & Li, 2015). In our study, we decided against a correction factor, because the fractionation intensities were not determinable. For removing the atmospheric isotope background, Miller-Tans plots were used (Miller & Tans, 2003). For this purpose, the products of the individual CO 2 -C concentrations and the d 13 C values of sampled CO 2 were plotted against their corresponding CO 2 -C concentrations. The resulting slope of the linear regression shows the isotope signature of the source without background. This technique allows to identify the isotope signature of the soil respired CO 2 , especially if the air background concentrations vary over time and are not stable as required for the Keeling plot (Miller & Tans, 2003). Feedstock and bulk soil samples were air-dried and ground before d 13 C analysis. Total carbon and nitrogen content and the isotope signature of the amendments were detected by an Euro EA Elemental Analyser (Eurovector, Milan, Italy), which was coupled via a Finnigan ConFlo III (Thermo Fisher Scientific) universal interface to a Finnigan Delta V Advantage Isotope Ratio Mass Spectrometer (Thermo Fisher Scientific).

| Statistical analyses
Statistical analysis and graphical design were carried out using R 3.3.1 (R Core Team, 2016) and EXCEL 2013 (Microsoft Corporation, Redmond, WA, USA). Box plots (not shown) were used to identify outliers. To do this, all values beyond the limit of 1.5 times of the interquartile range were eliminated from further analysis. Afterward, arithmetic means of the remaining replicates (n = 3-5) were calculated. Differences in hourly-emitted carbon dioxide-carbon (CO 2 -C) during the growing season between treatments and corresponding pure fertilizer without biochar means were statistically evaluated using the two-sided unpaired two-sample t test. Prior test assumption of normally distributed data was examined using Shapiro-Wilk test. In case of non-normal distributed data (treatment mineral fertilizer), Brown-Forsythe test was used for checking the homogeneity of variances. In all other cases, Bartlett's test was used.

| CO 2 emissions
Temporal variation of the CO 2 fluxes and weather conditions for comparison are shown in Figure 2. From April to the beginning of May 2014, CO 2 fluxes within treatments were more or less stable but varied among treatments from approximately 50 mg CO 2 -C m À2 hr À1 (pure fertilizer) to 150 mg CO 2 -C m À2 hr À1 (high biochar addition). At the beginning of May 2014, CO 2 emissions increased continuously until June 10, 2014. On June 25, 2014, all treatments showed a reduced CO 2 emission. Thereafter, a strong increase in CO 2 flux was observed for all treatments.
The CO 2 flux generally increased in the order mineral fertilizer < digestate (pure and fermented) < compost. The yearly re-application of 1 Mg biochar/ha mixed with digestate did not increase the CO 2 fluxes substantially compared to the corresponding fertilizers. Furthermore, the differences between high biochar applications (40 Mg/ha) and the pure fertilizers were obviously higher using mineral fertilizer and digestate compared to compost and fermented digestate (Figure 2).
The mean CO 2 fluxes per hour (Table 1) varied from 86 to 151 mg CO 2 -C m À2 hr À1 with significantly increased emissions when biochar was applied in high application amounts. For instance, CO 2 emission of mineral fertilizer or biogas digestate combined with 40 Mg/ha biochar increased up to 64% and 53% compared to the pure fertilizers, respectively (Table 1). Fermented digestate including 40 Mg biochar/ha emitted 30% more CO 2 -C than the corresponding pure fermented digestate alone. Low amounts of biochar (3 Mg/ha) combined with biogas digestate did not increase the CO 2 flux, which is in contrast to mineral fertilizer where CO 2 -C emission increased by 26% (Table 1). Pure compost showed higher CO 2 -C fluxes than mineral fertilizer, fermented and nonfermented biogas digestate and biochar addition did not significantly increase CO 2 -C emission of the compost treatment (Table 1). Table 2 shows the 13 C isotope signature and total carbon content of the applied fertilizers within the three years of the field experiment. As expected, biochar addition to the fertilizers increased total carbon content multiple times. With exception of mineral fertilizer (2012), biochar addition increased 13 C isotope signature of fertilizers. Furthermore, 13 C isotope signature of pure fertilizers was more or less equal in each year with exception of the mineral fertilizer applied 2014, which was considerably higher compared to 2012 and 2013.

| Carbon isotope signature (d 13 C)
Maize biogas digestate showed a typical C4 plant 13 C isotope signature of approximately À12.7 mUr. Due to the C3 signature of biomass residues biochar used in this study (~À28.0 mUr), sources of emitted CO 2 could be determined. Yakir and Sternberg (2000) classified the 13 C isotope composition of À25.0 mUr for soils containing C3 soil organic matter. During the growing season 2014, the CO 2 emissions showed mostly an intermediate 13 C signature, indicating mixed C4 and C3 sources (Figure 3). However, short-term negative or positive offsets could be observed during the measurement (Figure 3). Especially digestate and mineral fertilizer treatments, including biochar-amended treatments, revealed a strong variation and a daily fluctuation of the emission source. In contrast, CO 2 emissions of compost and fermented digestate showed smaller variations during the growing season. Compared with the seasonal course of the CO 2 fluxes (Figure 2) almost all treatments showed a rapid change to approximately À25 mUr on June 9, 2014. Only controls such as fermented digestate and compost as well as biochar (40) mineral fertilizer do not follow this pattern.

| Cumulative emissions and emission sources
The cumulative CO 2 emissions of all sampling dates confirm the results of increased CO 2 releases after biochar addition ( Figure 4). CO 2 emissions of biochar-treated plots (40 Mg biochar/ha) increased up to 57% compared to corresponding fertilizers without biochar. CO 2 -C emissions of compost and biochar (10) compost were more or less equal. There is only a slight increase in the soil-derived CO 2 -C by biochar addition to the mineral fertilizer treatment. On the other hand, the fertilizer-derived carbon dioxide output rises. Both digestate treatments (untreated and fermented) showed an increased level of soil-derived T A B L E 1 Hourly-emitted carbon dioxide-carbon (CO 2 -C) of the sampling dates during the growing season of narrow-leafed lupine (Weighted mean AE standard error of the weighted mean). Relative changes refer to the comparison of the treatment and pure fertilizers without biochar carbon emissions after the addition of biochar. The mixture of 40 Mg biochar/ha and digestate increases also the emission from fertilizer. Although the total emissions always increased after the addition of high biochar amounts (40 Mg/ha), the relative contribution of soil-derived CO 2 -C mostly did not raise for all treatments (Figure 4). Carbon dioxide emissions with a C3 signature of biochar (40) mineral fertilizer and biochar (40) digestate showed a reduction of the relative contribution. On the other hand, fermented digestate mixed with 40 Mg biochar/ha did not follow this trend and showed minor enhancement C3-derived CO 2 emissions. It is conspicuous that the biochar (3) digestate does not follow this trend. For biochar (3) digestate, the cumulative emissions stagnate, but the relative contribution shifts to more C3-derived CO 2 emissions. In comparison with all other treatments, the addition of 10 Mg biochar/ha to compost effects only minor changes of the relative contribution of the CO 2 emission sources.

| Heterotrophic respiration as function of fertilizer type
Against the observations of several laboratory experiments (e.g., Marstorp, 1996;Stumpe et al., 2012), no enhanced CO 2 production immediately after fertilizer application was observed (Figure 2). The reason could be the low soil temperature of approximately 11°C, dry soil conditions, which restrict the microbial activity, or the lagged gas sampling performed one week and a half after the fertilizer incorporation. Therefore, the initial microbial stimulation by the fertilizer ceased until we started gas sampling.
In general, it can be expected that the application of easily available carbon (e.g., in the form of liquid organic fertilizer or biogas digestate) enhances CO 2 emission of well-aerated soils such as sandy soils. The fast decomposition of easily degradable organic compounds is responsible for additional emissions (Bol, Moering, Kuzyakov, & Amelung, 2003;Joergensen, Meyer, Roden, & Wittke, 1996;Stumpe et al., 2012). In our case, CO 2 emission increased after organic fertilizers application compared to mineral fertilizer (all without biochar). Both average and total emissions of the whole growing season are increased by 13% for biogas digestate, 19% for inoculated biogas digestate, and 42% for compost (Table 1; Figure 4). The process of fermentation led to a reduced amount of easily degradable organic compounds. Therefore, the CO 2 emissions should be decreased on plots treated with fermented digestate. But our experiment shows contrary results. Additional fermentation of already fermented biogas digestate did not further reduce CO 2 emission. Another explanation of enhanced degradation of organic fertilizers is that nitrogen fertilization stimulates microbial activity (Lu et al., 2014). The use of organic fertilizers provides high nitrogen amounts and labile carbon structures, which stimulates heterotrophic respiration.

| Heterotrophic respiration as influenced by biochar application
Our experiment showed increased CO 2 release from the sandy Cambisol when high amount of biochar (40 Mg/ha) T A B L E 2 Total carbon concentration and carbon isotope signature (d 13 C) of the used amendments, which were applied during 2012 and was applied, independent from the fertilizer used in addition. These results are compliant with a meta-analysis by He et al. (2017), who found an increased CO 2 release from soils after biochar addition. Similar results were reported by Lanza, Wirth, Gessler, and Kern (2015) for a short-term dynamic incubation experiment, which showed higher CO 2 effluxes of amendments with fermented biochar. However, as we did not measure CO 2 release during fermentation, total carbon balance remains unclear. But it could be shown that total carbon balance does not differ between composting and fermentation of biochar (Fischer & Glaser, 2012). Within this study, we could not clarify the responsible processes, which led to increased CO 2 emissions. However, in our opinion, the most probable explanation is that due to high porosity of biochar, microorganisms are protected against predators leading to higher microbial biomass (Lehmann et al., 2011;Thies & Rillig, 2009). In the presence of easily available carbon (e.g., in the form of liquid organic fertilizer or biogas digestate), microbial degradation F I G U R E 3 Isotope composition (d 13 C) of the CO 2 emissions obtained from the Miller-Tans mixing model (Miller & Tans, 2003). of these products is increased, which leads to higher CO 2 emissions of well-aerated soils. As described in several studies (Kuzyakov et al., 2009(Kuzyakov et al., , 2014Wang et al., 2016), the decomposition of biochar mainly takes place in a cometabolic way.
Furthermore, biochar significantly improved the plantavailable water holding capacity of sandy soils Liu et al., 2012). Consequently, it also ensures water supply to soil microorganisms during dryer periods. On the other hand, a waterlogging situation on June 25, 2014, may explain the reduced CO 2 emission from all treatments.
Another explanation for increased CO 2 release upon high biochar application is a missing protection against co-metabolic degradation due to missing organic-mineral interactions between biochar and minerals caused by the sandy texture of our soil with a minor clay fraction (<5%). Also, Wang et al. (2016) showed in a meta-analysis that clay-poor soils (<10%) could have up to 20% higher carbon losses compared to clay-rich soils when high amounts of biochar were applied. Brodowski, Amelung, Haumaier, Abetz, and Zech (2005) and Glaser et al. (2000) demonstrated that oxidized zones on biochar surface are more susceptible to organic-mineral interactions, which are a main driver of long-term stability of biochar in addition to its intrinsic chemical recalcitrance. These interactions effect a physicochemical protection through aggregation, which prevent the ongoing microbial or chemical oxidation of the charcoal surface (Glaser et al., 2000). A sandy-textured soil is characterized by a bigger size of its primary particles and inert particle surfaces, compared to clay or silt, which reduces the formation of mineral aggregates containing biochar. Also, direct organic-mineral interactions between biochar and mineral surfaces, caused by electrostatic cation bridges, hydrophobic interactions or H-bondings, are inhibited because of the sandy texture (von L€ utzow et al., 2006). Therefore, physical protection against degradation can mostly be neglected. A further study by Brodowski, John, Flessa, and Amelung (2006) demonstrated that macroaggregates are not effective to enclose biochar in sandy soils. This missing physical protection could also enhance the vulnerability of biochar for (co-metabolic) degradation, which leads to higher carbon dioxide emissions.

| Carbon sources of heterotrophic respiration
The shifting carbon isotope ratio of the mineral fertilizer represents the change from conventionally produced mineral fertilizer, which possess d 13 C values of approximately À40 mUr (typical signature of fossil methane, which is a basic commodity of the urea production, Vitoria, Otero, Soler, & Canals, 2004), to an immediately available organic-based fertilizer, which was used in 2014. The carbon isotope signature of À17.4 mUr shows a mixture of C3 and C4 feedstocks. Both biogas digestates showed C4type d 13 C values well-reflecting maize used as feedstock (Table 2). Biochar exhibited a d 13 C value indicating C3 origin of biomass used as feedstock, while compost showed a slightly more positive d 13 C value still indicating C3 origin but also 13 C enrichment due to intensive microbial degradation via composting process (Table 2).
Generally, d 13 C value of heterotrophic respiration became more negative during the growing season due to the mineralization shift from fertilizer to soil organic matter, which is based on the short-term availability of the applied fertilizers (Figure 3). This effect was independent from the applied fertilizer (Figure 3). The shifting d 13 C F I G U R E 4 Quantitative carbon dioxide-carbon (CO 2 -C) emissions, their origin and the biochar effect during the narrow-leafed lupine growing season 2014. This effect indicates additional emissions through biochar-induced priming of soil organic matter or mineralization of biochar, which are both C3-derived values during the growing season suggest that directly after the fertilizer application, fertilizer-derived organic material was degraded by microorganisms. During the growing season 2014, the decrease of d 13 C values indicates an increasing microbial degradation of soil-and biochar-derived organic matter. However, fluctuation of CO 2 d 13 C values during the experimental period was rather high (Figure 3). Therefore, all d 13 C values were integrated into an isotope mass balance (Figure 4). Figure 4 clearly shows that the more biochar was added, the more CO 2 was released up to 60%. There was only a small increase of soil-and/or biochar-derived organic matter, indicated by a small increase of C3-derived CO 2 (Figure 4). Due to the same isotope composition of soil organic matter and the applied biochar, it is not possible to differentiate between biochar-derived and soil organic matter-derived emitted CO 2 . Additional C3-derived CO 2 of the biochar treatments compared to the corresponding pure fertilizers without biochar is either due to biochar degradation or caused by a positive priming of soil organic matter evoked by biochar. Most probably it is a mixture of both processes. Comparing isotope measurements of the treatments with and without biochar clearly indicates a minor contribution (positive priming or biochar degradation) to total CO 2 emissions when combined with mineral fertilizer or compost (Figure 4). On the other hand, more biochar or soil organic matter was decomposed, when biochar was combined with biogas digestate (fermented or nonfermented; Figure 4). However, compared to the high amount of biochar added (40 Mg/ha) additional CO 2 release was negligible (about 0.1 Mg/ha).

| Biochar as a passive influencer
Our study clearly shows increased CO 2 emissions at high biochar application amounts (40 Mg/ha) from a sandy Cambisol under practical agronomic conditions in Northern Germany, especially when combined with mineral fertilizer and digestate. Stable isotope (d 13 C) measurements show the main source of enhanced CO 2 emissions being fertilizer-derived organic carbon when biochar was combined with mineral fertilizer (containing organic carbon) and compost, while the positive priming potential of biochar or the co-metabolic decomposition of biochar was a substantial source of enhanced CO 2 emissions, when combined with biogas digestate even in case this was fermented beforehand.
In view of the fact that biochar should be able to sequester large amounts of carbon due to long-term stability, the increased CO 2 emissions after the treatment with 40 Mg biochar/ha are negligible (about 0.1 Mg/ha). However, further long-term measurements of CO 2 emissions under field conditions are necessary to get a clear picture of the carbon sequestration potential of biochar.
In summary, increased carbon losses from biochar-treated agriculturally used soils under temperate conditions result from different factors, which are stimulated by the presence of biochar. The most important emission driver might be the enlargement of the microbial biomass ( Figure 5) because of the suitable habitat and the protection against predators provided by the microporous surface of biochar. Another main driver of the microbial-derived soil emissions is the soil temperature and humidity, which are both modified by biochar. The decomposition of organic plant residues, soil organic matter, and biochar is strongly influenced by the amended fertilizers.