Emissions of NO, NO2, and N2O to the atmosphere were measured with a fully automated laboratory system from undisturbed soil columns obtained from five different temperate and one boreal forest sites. The soils were chosen to cover a transect through Europe, sandy and loamy textures, and different atmospheric nitrogen deposition rates. In a two-factorial experimental design, soil cores were kept under varying conditions with respect to temperature (range 5–20°C) and soil moisture (range 0–300 kPa). The combination of soil temperature and soil moisture could explain a better part of variations in NO (up to 74%) and N2O (up to 86%) emissions for individual soils, but average emissions differed significantly between various forest soils. Generally, NO and N2O were emitted from all soils except from the boreal pine forest soil, where NO was consumed. NO emissions from the German spruce forest receiving highest yearly nitrogen inputs of >35 kg ha−1 yr−1 ranged from 1.3 to 608.9 μg NO-N m−2 h−1 and largely exceeded emissions from other soils. Average N2O emissions from this soil tended also to be highest (171.7 ± 42.2 μg N2O-N m−2 h−1), but did not differ significantly from other soils. NO2 deposition occurred in all soils and strongly correlated to NO emissions. NO and N2O emissions showed a positive exponential relationship to soil temperature. With activation energies between 57 and 133 kJ mol−1, N2O emissions from the various soils responded more uniformely to temperature than NO emissions with 41 and 199 kJ mol−1. The two Austrian beech forest soils showed exceptionally high activation energies for NO emissions, which might be attributed to chemodenitrification. N2O emissions increased with increasing water filled pore space (WFPS) or decreasing water tension, respectively. Maximal N2O emissions were measured between 80 and 95% WFPS or 0 kPa water tension. Optimal moisture for NO emission differed significantly between the soils, and ranged between 15% WFPS in sandy Italian floodplain soil and 65% in loamy Austrian beech forest soils. These differences may be related to the specific adaptation of the microbial communities to draught conditions.
 In contrast to NO, nitrous oxide (N2O) is a direct greenhouse gas with an atmospheric lifetime of 120 years. Its 100-year global warming potential is about 300 times as high as that of carbon dioxide, and the rate of concentration change is +0.8 ppb yr−1 [IPCC, 2001]. The major sink is the stratosphere, where N2O is involved in the destruction of stratospheric ozone. Estimates on the contribution of temperate forest soils to atmospheric nitrous oxide have a high degree of uncertainty and range between 0.1 and 2 Tg N yr−1 [Kroeze et al., 1999]. Hence, more information on environmental variables controlling the emission of NO and N2O in temperate and boreal forests is needed considering that these ecosystems make up for 40% of the total forest land on Earth.
 Since microbial processes such as nitrification and denitrification predominantly cause production of NO and N2O in soils [Firestone and Davidson, 1989], soil temperature is a key variable influencing emission rates. Emissions of both NO [Slemr and Seiler, 1984] and N2O [Skiba et al., 1998; Smith et al., 1998] increase with rising soil temperature due to the fact that rates of enzymatic processes generally increase with temperature as long as other factors (e.g., substrate or moisture) have no limiting effect. Despite its function as a transport medium for NO3− and NH4+, soil water influences the rate of O2 supply and thereby controls whether aerobic processes such as nitrification or anaerobic processes such as denitrification dominate within the soil. While N2O emissions are accepted to increase at higher water contents through greater losses from denitrification [Wolf and Russow, 2000], the relationship between NO flux and soil water is more complex. Through limited substrate diffusion at very low water content and limited gas diffusion at high water content, maximum nitric oxide emissions are suspected to occur at low to medium soil water content. Ludwig et al.  suggested an optimum of NO emissions at approximately 20% water filled pore space and a strong decrease towards extreme moisture conditions. However, other studies reported maximum NO fluxes between 43% and 85% [Ormeci et al., 1999; Van Dijk and Duyzer, 1999].
 This study is part of the NOFRETETE project, an EU project with the overall objective of improving the knowledge about and to enable a better prediction of the quantity of emissions of nitrogen oxides from European forest ecosystems. In order (1) to assess the influence of soil temperature and soil moisture content on the emission, (2) to compare N-trace gas emission from different European forest soils, a laboratory incubation experiment with different soils from Italy, Austria, Germany and Finland was carried out. With the aim to prevent an overlap of temperature and moisture effects as it may be found in field measurements, undisturbed soil cores were drained to definite soil water tension levels and the NO, NO2 and N2O emissions were measured under different temperatures.
2. Materials and Methods
2.1. Study Sites
 The southernmost forest chosen for soil core collection is located in the Ticino natural park in Italy. The mixed deciduous forest stand (Quercus robur, Populus alba x tremula, Fraxinus spp., Salix spp. Corylus avellana, Prunus padus, Crataegus monogyna) is situated in close proximity to the river Ticino and is periodically flooded. The sandy soils are mainly populated with Convallaria maialis and Anemone nemorosa as undergrowth vegetation. For a detailed site comparison, see Table 1.
Table 1. Site Characteristics of the Different Forests Chosen for Soil Core Sampling
 In Austria, three different sites were selected for soil sampling. The easternmost site, the so-called “Schottenwald” (SW), is located on a hill in the Vienna Woods approximately 10 km west of the city at 370 m a.s.l. The plot itself is found on a SE-exposed slope supporting a uniform 140-year old beech stand. The soil of the plot was classified as dystric cambisol over sandstone with a silty loam texture.
 The second site in the Vienna Woods was a 60-year old uniform beech stand (510 m a.s.l.) near Klausen–Leopoldsdorf (KL), a village located about 40 km west of Vienna. The plot lies on a NE-exposed slope with a gradient of 20%. The soil is a cambisol over sandstone with up to 20% rock content in the upper mineral soil layers. In KL, different wind conditions and the greater distance from the city lead to lower nitrogen deposition.
 The only limestone site was the so-called “Mühleggerköpfl” (895 m a.s.l.) near Achenkirch (AK) in the north Tyrolean limestone Alps. The 125-year old warm, central montane spruce-fir-beech forest stand is dominated by spruce. Pine and beech occur isolated or in groups. The soil types are chromic cambisols as well as rendzic leptosols, both with a high small-scale variability of the skeleton content.
 In Germany, soil cores were collected at the Höglwald (HW) experimental site, where annual cycles of NO, NO2 and N2O have been investigated since 1993. The 96-year old Norway spruce stand (450 m a.s.l.) is situated 50 km west of Munich in the hilly peralpine zone. The soil is a very acid hapludalf derived from loess [Kreutzer, 1995]. The stand is characterised by N-saturation due to long-term heavy nitrogen inputs (at least 35 kg ha−1 yr−1) from the atmosphere [Butterbach-Bahl et al., 1997].
 Boreal forests were represented by the SMEAR II (Station for Measuring Forest Ecosystem-Atmosphere Relations) field measurement station in Hyytiälä (HY), Southern Finland (181 m a.s.l.). The uniform 40-year old stand is dominated by Scotch pine. Other tree species like birch, grey alder and common aspen occur in quantities of less than 1%. The parent material of the soil is coarse silty glacial till and the soil is classified as a haplic podzol [Vesala et al., 1998].
2.2. Experimental Layout
 For the NO, NO2, and N2O measurements, 4 undisturbed soil cores from each collection site were incubated at 4 different water tension levels (300, 100, 30 and 0 kPa) and 4 different temperatures (5, 10, 15 and 20°C) per tension level. At each field site, 8 undisturbed soil cores (mineral soil + litter) were collected randomly using stainless steel cylinders (diameter: 7.2 cm, height: 5 cm). The cylinders were closed with stainless steel lids immediately after sampling. To provide similar preconditions, soil cores were collected in spring, when individual soils reached 8°C in 5 cm soil depth. Four cores were stored at 5°C until measuring, the other 4 cores were oven-dried, and the percent volumetric water content was determined. We assumed the same water content for the stored soil cores, and calculated the soil core weights for the defined water potential levels (30, 100 and 300 kPa) from already known volumetric water content values for this tension levels. If no volumetric water content was available, the missing values were generated using ceramic pressure plates.
 We did not disturb the soil through sieving or other treatments in order to keep conditions as natural as possible and to avoid emission losses [Brumme et al., 1999]. Therefore the calculated weights of the soil cores could only be estimates of the water content per tension level, because of the unknown rock content in the cylinders and small-scale differences in soil texture and moisture. The actual volumetric water content and the water filled pore space (WFPS) of the soil cores were determined after the gas flux measurements by oven-drying at 105°C and differed slightly (0.3–4%) from the estimates. Nevertheless, we opt for this method as a good way to compare soils with different textures.
 At the beginning of the measuring cycle, the soil cores were dried or watered till the calculated weight for the highest (300 kPa) tension level was reached. The cores were incubated at 5°C for 6 hours (N2O determination) followed by at least 15 hours of NOx measuring. Subsequently, the temperature was increased to the next level (10°C), and the weights of the cores were controlled and if necessary readjusted by moistening. An adjustment period of at least 10 hours until the next measuring provided for adaptation of the soils to the new temperature conditions. When all measurements at the highest tension level were completed, the soil cores were watered until the next level (100 kPa) was reached and cooled down to 5°C. In experiments preceding the study, soils showed a strong increase of NO emissions in response to wetting. The response decreased within several hours, and emissions were stable 2 days after wetting at the latest. Therefore a 3-day “period of adaptation” to the new water tension level had to pass before starting the measurements again. The 0 kPa water potential level was reached by complete water saturation of the soil cores. Because of the limited number of test chambers (12) available, the Austrian soil cores (SW, KL and AK) were measured first, followed by the Italian (TP), German (HW) and Finnish (HY) soils.
2.3. Gas Flux Measurements
2.3.1. Description of the Measuring System for the Determination of NO and NO2
 In accordance with a field measuring system [Holtermann, 1996], a fully automated laboratory measuring system was designed and constructed. Thirteen adapted Kilner jars (Figure 1) serving as test chambers were placed in a temperature-controlled incubator/refrigerator (I/R) and connected to a NOx-analyzer via PTFE tubing (inner diameter: 4 mm; length: <1.5 m) and magnetic valves fitted with stainless steel fittings (Figure 2). All components of the system except the I/R were made of chemically inert materials.
 The Kilner jars were fitted with glass inlets (inner diameter: 7 mm) and stainless steel fittings as outlets. Twelve test chambers were filled with soil cores, and one empty chamber served as reference chamber. The sequence and the opening time of the valves as well as data collection were controlled by a PC running a specially developed computer program. The valves were closed in normal position and electrically connected in parallel to exits 1 to 13 of the PC-interface. To link the valves with the analyzer, the outlets of the 13 valves were pooled in a Teflon collector, to whose exit the NOx-analyzer was connected. When an exit of the interface was shut down during the run of the computer program, the belonging magnetic valve opened and established the connection between the belonging test chamber and the NOx-analyzer.
 For NO and NO2 flux rate determination, air from inside the I/R was sucked through the chambers to the NOx-analyzer. NO and NO2 were measured with a HORIBA APNA-360 chemo luminescence nitrogen oxide analyzer (detection limit: 0.5 ppb). The instrument was calibrated weekly using zero air and standard gas (Umweltbundesamt, 153 ppb NO in N2). The constant flow rate of 1.5 l min−1 by the pump of the analyzer, combined with the small headspace volume (450 ml) of the chamber, ensured a low residence time and good mixture of the air inside the test chamber. To obtain stable air pressure inside the I/R, a connection tube to the ambient air outside was placed into the wall. In addition to this inflow tube, a fan mixed the incoming air with the air inside (Figure 2).
 A complete measurement cycle during which one NO and NO2 flux rate for each of the 12 test chambers could be determined lasted for 128 min. During one complete cycle the reference chamber was measured after every three sample chambers.
 Preliminary tests showed that it took about 7 min to reach steady state in a test chamber because of the accumulation of NO during the period without airflow. Therefore the measuring time of each chamber was 8 min, and only the values obtained during the last 10 s were used for flux rate determination. During this period, data for NO and NO2 concentrations were collected at a frequency of 1 Hz, and a 0–10 s average was subsequently calculated. The flux rates of NO (and in analogy the NO2 flux rates) were calculated from the following equation:
where F is the net flux in μg N m−2 h−1, M is the atomic weight of the element (N = 14.008 g mol−1), Vm is the standard gaseous molar volume (24.055 · 10−3 m3 mol−1), Ceq is the mixing ratio (ppbv = 10−9 m3 m−3) of the gas when the chamber under investigation has reached steady state, C0 is the mixing ratio of the gas in the reference chamber, Q is the mass flow rate of air through the chamber (≈0.0015 m3 min−1), and A is the soil surface area of the soil core (0.0041 m2).
 Incubations took place in the dark, and no corrections were made for photolysis of NO2 (producing NO). Standards that were passed through the system showed no detectable loss of NO. Therefore the loss of NO or NO2 due to reactions with or absorption by the components of the measuring system were considered to be of minor importance. The reaction of NO with O3 of the ambient air would have led to an underestimation of the NO fluxes and to higher NO2 flux rates. In our case, NO2 was generally deposited to the soil or NO2 fluxes were slow. Therefore the reaction of NO with O3 in the chambers was not considered in this study.
2.3.2. N2O Measurements
 N2O fluxes were determined by headspace concentration changes in sealed test chambers. For that purpose, the glass inlets and outlets were sealed with removable septa. Each soil core was enclosed in the gas-tight test chamber and a 15-ml gas samples was taken at 0, 3 and 6 h with a gas-tight syringe and transferred into evacuated 10 ml glass vials. The vials were sealed with silicone grease and stored under water until analysis. Gas samples were analyzed by gas chromatography using a 63Ni electron capture detector (injector 50°C, detector 330°C, oven 80°C, carrier gas N2). Calibration was performed using 5 μl 1−1 N2O (Linde Gas) standard gas. The N2O flux rates were calculated from linear concentration increase in the closed chambers.
2.4. Statistical Analyses
 Statistical analyses were carried out with SPSS (Version 10.0) and partly with SIGMA PLOT (Version 2001). Analyses of significant differences in trace gas fluxes between various soils were carried out using a nonparametric statistical test (Mann-Whitney U-test). Multivariate nonlinear regression analyses were done using SIGMA PLOT.
3.1. General Results
 Mean NO fluxes differed significantly between most sites, with highest emissions at HW and average NO uptake at HY (Table 2). The various soils showed decreasing NO emissions in the following order: HW > SW ≥ TP ≥ KL > AK > HY. In contrast to NO fluxes, NO2 fluxes were generally negative, i.e., the soils functioned as a net sink for atmospheric NO2. A net NOx emission was observed at TP, SW, KL and HW, since NO emission exceeded NO2 deposition. AK showed balanced fluxes while HY functioned as a net sink of NOx. Mean N2O fluxes were positive in all cases and varied in a narrower range than NO fluxes. Mean N2O emissions followed the sequence: HW ≥ SW ≥ KL ≥ AK ≥ TP > HY.
Table 2. Mean (±S.E.) Flux Rates of NO, NO2, and N2O From the First 5 cm Top Soil as Observed Over All Temperature and Moisture Conditionsa
Ticino Park (TP)
Units are in μg N m−2 h−1. The same letters within one row indicate no significant difference at p < 0.05.
18.5 ± 5.8 a
25.5 ± 7.5 a
10.2 ± 3.4 a
2.8 ± 1.4 b
213.4 ± 49.6 c
−8.4 ± 1.3 d
−5.8 ± 0.9 a
−0.1 ± 0.7 b
−2.2 ± 0.7 c
−2.7 ± 0.8 c
−33.4 ± 6.9 d
−2.5 ± 0.3 c
26.5 ± 10.0 a
129.9 ± 31.3 b
128.2 ± 33.8 b
55.2 ± 12.9 a b
171.7 ± 42.2 b
4.8 ± 0.7 c
3.2. Effects of Soil Temperature and Soil Moisture
3.2.1. NO/NO2 Fluxes
Figure 3 shows the NO and NO2 fluxes of all soils at the 4 water tension levels defined. NO emission increased exponentially with increasing soil temperature at the 300, 100 and 30 kPa levels. At water-saturated conditions (0 kPa), no or extremely low NO fluxes were measured. The activation energies of the NO emission rates were calculated from the Arrhenius equation
where R is the gas constant (8.314 J K−1 mol−1), Ts is the soil temperature (K), A, the pre- exponential factor is a constant, and Ea is the activation energy which was calculated from the slope of the regression line of the inverse temperature versus the ln NO flux data. The differing slope of the regression lines in Figure 4 indicates high variability of activation energy among the different sites. Values of activation energy are presented in Table 3 and range between 41.2 kJ mol−1 at AK and 198.6 kJ mol−1 at KL.
Table 3. Exponential Regression Analysis Between Soil Temperature and Mean NO Flux Rates at Defined Water Tension Levels (N = 4)a
Ticino Park (TP)
Klausen Leopoldsdorf (KL)
Here r2, coefficient of determination; correlation is positive for r2, except values in parentheses which have negative correlation; Ea, activation energy; n.s., not significant.
 Results obtained from exponential regression analysis between NO emission and soil temperature at different water tension levels are presented in Table 3. NO emissions from SW, KL, AK and HW showed high correlation at the 300, 100 and 30 kPa tension levels. Coefficients of determination indicate that 99.9% (SW), 99.0% (KL), 91.1% (AK) and 99.9% (HW) of variations in NO emission rates per water tension level could be explained by variations in soil temperature. NO emissions from TP were correlated to soil temperature under dryer conditions (300 and 100 kPa) only. All sites showed no correlation between NO emission and soil temperature under water saturation (0 kPa). HY was the only site where NO emission correlated negatively (r2 = 0.871, p < 0.05) with soil temperature. At HW and TP, NO2 deposition increased exponentially with increasing temperature at the 300, 100 and 30 kPa water tension levels with a maximum NO2 deposition rate of 88.7 ± 9.8 μg N m−2 h−1 (HW, 100 kPa, 20°C), while other sites showed no clear trend and varying deposition between 0 and 5 μg N m−2 h−1. Under water saturation, NO2 flux rates at all sites showed no temperature dependence. NO2 emission was measured at SW under dry conditions (300 kPa) and temperatures up to 15°C (5.2 ± 1.0 μg N m−2 h−1) exclusively. NO2 and NO fluxes did not correlate between SW, KL and AK. At HY, a low negative correlation (r2 = 0.143; p < 0.001) could be observed, while sites with a clear trend in NO2 fluxes like HW and TP correlated negatively with NO emission (r2 = 0.987 and r2 = 0.850 respectively; p < 0.001).
 Most soils reached highest NO emissions at the 100 kPa water tension level, with the maximum emission of HW (608.9 ± 71.3 μg N m−2 h−1). Only SW emitted most (95.2 ± 10.5 μg N m−2 h−1) under high water tension (300 kPa), while KL reached maximum fluxes (44.7 ± 7.8 μg N m−2 h−1) at 30 kPa water tension. Under water saturation (0 kPa) and below 20°C soil temperature, NO fluxes at all sites were around zero. At 20°C, emission could be measured at SW (11.1 ± 2.3 μg N m−2 h−1) and KL (4.4 ± 0.8 μg N m−2 h−1), while at all other sites consumption occurred (AK: 4.2 ± 0.8 μg N m−2 h−1).
Figure 5 shows NO flux rates plotted against soil temperature and water filled pore space (WFPS) at HW. The grid results from multiple regression analysis in the form of
where z is the NO flux rate in μg NO-N m−2 h−1, x is the soil temperature in °C, y is the WFPS in % and x0 (representing the optimum temperature for the investigated temperature range), y0 (representing the optimum moisture), a (maximum value), b (temperature effect reduction factor), c (moisture effect reduction factor) are the regression coefficients. Results of the multiple regression analysis are presented in Table 4. Up to 74% (TP) of variations in NO emission could be explained by the combination of WFPS and soil temperature. Maximum NO flux rates appeared generally at low WFPS but varied with sites. TP had highest NO fluxes around 15% WFPS, while SW, AK and HW reached highest fluxes between 30 and 45% WFPS. KL showed maximum fluxes at 65% WFPS.
Table 4. Multiple Regression Results for NO Emissions Per Equation (3) Relating NO Emission to Soil Temperature and WFPSa
Here r2, coefficient of determination; correlation is positive for r2; N, number of cases.
 Mean N2O flux rates of the various soils are shown in Figure 6. All sites emitted no or little N2O at the 300 and 100 kPa water tension levels. Emission increased dramatically under conditions of higher water saturation (30 kPa) and under water saturation (note the different scaling of the y-axis at the 0 kPa level). Sites differed in their reaction to soil moisture increase. HY showed low emission at all water tension levels, while at SW and KL emission increased continuously with decreasing water tension and rising soil temperature. AK reached the highest N2O fluxes of all sites at the 30 kPa level but showed no further increase at the 0 kPa level. Emission at HW and TP was low at the 300, 100 and 30 kPa tension levels and increased exponentially with water saturation. HW showed even high emission (321.1 ± 146.3 μg N m−2 h−1) at 5°C soil temperature and reached maximum emission at 15°C (1012.7 ± 145.5 μg N m−2 h−1), while emission decreased again at 20°C. Similar patterns could be observed at TP, but N2O fluxes were low compared to HW.
 Results of multiple regression analysis are expressed in the form of
where z is the N2O flux rate at soil temperature a, and the WFPS b with z0, a, b as regression coefficients, are presented in Table 5. The correlation was relatively low at TP, but for all other sites a high percentage of variation in N2O fluxes could be explained by variations in soil temperature and WFPS, with soil temperature being less important than WFPS. Activation energies ranged between 57.6 kJ mol−1 (HW, 30 kPa) and 133.2 kJ mol−1 (KL, 100 kPa). Activation energies at KL and SW (102.1 kJ mol−1) exceeded activation energies at all other sites, but not to the extent observed for NO fluxes.
Table 5. Results of Multiple Regression Analyses Between N2O Flux Rates and Soil Temperature and WFPSa
Total r2 represents the coefficient of determination when all parameters tested are included; correlation is positive; n.s., not significant; all r2 are significant on the level p < 0.001.
4.1. Soil Temperature Effects
 For all soils except for the Finnish sample (HY), NO emission increased exponentially with soil temperature. This is in line with results of Gasche and Papen , who found an exponentional relationship between NO emission and soil temperature at the German (HW) field site, and with others, who found exponential relationships for agricultural [Williams et al., 1988; Roelle et al., 2001] and forest soils [Van Dijk and Meixner, 2001]. With regard to the intensity of NO emission response to an increase in soil temperature, the soils could be divided into two groups. (1) Soils with activation energies that correspond well with values reported in other studies, ranging from 40 to 108 kJ mol−1 [Yamulki et al., 1995; Ormeci et al., 1999; Van Dijk and Duyzer, 1999] like the Italian (TP), German (HW) and Austrian limestone soils (AK), and (2) soils with very high activation energies like the two Austrian beech forest soils (SW and KL). Activation energies of 125 to 199 kJ mol−1, as observed in SW and KL, indicate a fivefold to ninefold increase in NO emission for each 10°C temperature increase and have, to our knowledge, not been reported so far. This strong temperature response surmounts the two- to threefold increase in activity for each 10°C temperature increase, commonly observed [Focht and Verstraete, 1977] for microbial nitrification and denitrification. An explanation might be NO production due to abiotic processes like HNO2 self decomposition and other chemical reactions summarized as chemodenitrification. Van Cleemput and Baert  found activation energies for abiotic nitrite decomposition in the range of 25–236 kJ mol−1. Activation energies were highest at low (4–5) pH. A soil pH of 4.2 and 5.1 at SW and KL respectively supports this explanation, but the lowest soil pH (3.2) was found in the German spruce forest (HW), where activation energies were low compared to SW and KL. In addition to these nonbiological reactions, a strong response of nitrifying bacteria to the temperature increase could also be responsible, as activation energies of 28–166 for nitrification were reported by Focht and Verstraete . NO was generally consumed by the Finnish (HY) soil, with increasing consumption at increasing soil temperature, but NO consumption was also positively correlated (r2 = 0.80) to the NO concentration in the incubator/refrigerator. Therefore the existence of a NO compensation point (mixing ratio in the ambient air, where net flux between the soil and the atmosphere is zero) was probable. The NO compensation point was found at 0.2 ppbv and, in contrast to the findings of Slemr and Seiler , was independent of temperature. The finding of almost no NO emission from the Finnish soil is in line with first results of field measurements in HY [Pilegaard et al., 2003]. They assumed a combination of low temperature and low nitrogen deposition to be responsible for low fluxes, but soil temperature may be excluded, since in our incubation experiment, even at 20°C soil temperature, no NO emission could be detected (nor at the beginning of incubation, when ambient NO concentration was <1 ppbv).
 However, further improvement in understanding the response of NO emissions to soil temperature may be gathered using a more kinetically orientated approach [e.g., McKenney et al., 1984; Gödde and Conrad, 1999; Van Dijk et al., 2002], and by differing between responses of NO production and consumption. Even NO production through biological (nitrification, denitrification) and chemical processes (chemodenitrification) may be sensitive to temperature to varying degrees. Activation energies for NO production were so far especially published for chemodenitrification. McKenney et al.  reported 70–79 kJ mol−1 for clay and loam respectively, whereas Venterea and Rolston  calculated 67 kJ mol−1 for fertilized agricultural soil. In our study, differentiation between activation energies of biological and chemical processes was not possible because we tried to keep conditions in the soil cores as natural as possible and avoided chemical treatments of the soil cores.
 NO2 deposition increased exponentially with increasing soil temperature at HW and TP and strongly correlated to NO emissions from both soils. The observed deposition rates (up to 88.7 ± 9.8 μg N m−2 h−1) from the HW soil cores were in the range of observations from field studies at the HW site [Gasche and Papen, 1999]. Furthermore, most NO2 deposition rates observed were in the range of deposition velocities of NO2 to leaf area surfaces which range approximately from 0 to 1000 pmol NO2 m−2 leaf area s−1 [Hanson and Lindberg, 1991; Weber and Rennenberg, 1996; Sparks et al., 2001]. Only NO2 deposition from HW exceeded 1000 pmol under optimal temperature and moisture conditions (note: 1000 pmol NO2 m−2 s−1 = 50.6 μg NO2 m−2 h−1). The high NO2 deposition at HW is likely to be caused by the thick, lightly structured, very acid organic layer on top of the HW soil cores, which provided a huge surface for NO2 deposition.
 At most sites, nitrous oxide emissions increased exponentially with soil temperature. Only at HY, where generally little N2O was emitted, no significant relationship between soil temperature and N2O emission could be observed. Increasing N2O emissions with increasing soil temperatures were reported in several field and laboratory studies [Smith et al., 1998; Dobbie and Smith, 2001] and are in line with our findings. Activation energies were also highest at KL and SW but were clearly lower than activation energies for NO fluxes. The decline of N2O emissions at 20°C soil temperature and water saturation (0 kPa) at HW and TP was probably caused by shortage of substrate throughout the last part of the incubation period. Additional measurements 4 days later supported this assumption with a further decrease in N2O emission (results not shown).
4.2. Soil Moisture Effects
 The effect of soil moisture on NO emission has been investigated by several researchers [e.g., Cardenas et al., 1993; Ormeci et al., 1999; Roelle et al., 2001]. Typically, each soil will have a specific soil moisture which optimizes NO flux. For example, Van Dijk and Meixner  reported an optimum of 27% WFPS for tropical forest soil. Moisture conditions above optimum will restrict gas transport [Skopp et al., 1990] and thereby lead to anaerobic conditions. Moisture conditions below optimum will limit NO fluxes due to restricted substrate diffusion and water stress of soil microbes. The moisture optima in this study ranged from 15 to 65% WFPS. A direct comparison to field measurements for the German soil (HW) showed an identical moisture optimum of around 45% WFPS in the field [Gasche and Papen, 1999] and in our laboratory incubation experiment. The lowest optimum (around 15% WFPS) was observed in the sandy floodplain soil in Italy (TP), and the highest optimum of around 65% WFPS in one of the Austrian beech forests soils (KL). Moisture optima for NO emission were neither correlated to soil texture (clay or sand content) nor to bulk density. Hence different moister optima may be explained by microbial communities specially adapted to the typical water regime of the various soils. While the sandy Italian soil periodically dries up, the loamy soil in KL is only moderately drained throughout the year. Van Dijk and Duyzer  also explained their finding of an uncommonly high (up to 85% WFPS) moisture optimum for NO emission of a forest soil in the Netherlands by the adaptation of soil microbes to generally moist soil conditions.
 NO emissions exceeded N2O emissions by a factor of 10 under aerobic conditions, while N2O emissions were highest at water saturation, where almost no NO emission occurred. In water-saturated soils, nitrous oxide is mainly formed by denitrification of nitrate [Russow et al., 2000]. Increasing water saturation increases the number of anaerobic microsites and therefore promotes denitrification. Maximum N2O emissions at water saturation (0 kPa or 85 to 95% WFPS of the various soil cores in our experiment), as observed from all soils except Austrian limestone soil (AK), are in line with Dobbie and Smith , who also found highest N2O emissions of intact soil cores near complete water saturation. Smith et al.  reported an optimum of N2O emission at −5 kPa water potential, corresponding to 97% WFPS; at even higher water contents (−2.5 kPa), they observed a decrease in N2O emissions, probably due to increased N2O reduction [Focht and Verstraete, 1977]. A resolution according to water potential level as high as that achieved by Smith et al.  in their experiment with large soil monoliths was not possible in this study. However, at the Austrian limestone site (AK), we did observe a lack of N2O emission increase when soil moisture was increased from 30 to 0 kPa tension level, suggesting increased N2O reduction. AK seems to have a lower emission maximum between 60 and 90% WFPS, as suggested by Dobbie et al. . This relatively low emission maximum compared to the other soils is in line with laboratory measurements of N2 emissions conducted by K. Butterbach–Bahl and S. Zechmeister-Boltenstern (personal communication), where high N2 emissions were measured from the AK soil at low water tensions. Generally, it should be taken into account that all gas emissions in this study originated from the first 5 cm topsoil, and potential N2O from deeper soil layers might also influence the actual optimal water content for N2O emission at field conditions.
4.3. Other Controlling Factors
 Since the forest soils in our study showed significantly different NO as well as N2O emissions at similar soil moisture and temperature conditions, NO and N2O emissions must have been controlled by other factors as well. Rates of nitrification and denitrification strongly depend on the amount of nitrogen available in the soil solution. Therefore forest soils at or near nitrogen saturation combined with high nitrogen input, such as HW (>35 kg N ha−1 yr−1) and SW (>30 kg N ha−1 yr−1), showed highest NO and N2O emissions. This finding is in line with several studies [Butterbach-Bahl et al., 1997; Skiba et al., 1998; Venterea et al., 2003] that reported increased nitrogen oxide emissions at sites with high nitrogen input. Contrary to this, the boreal forest soil (HY), where little atmospheric N deposition occurs, showed no NO and lowest N2O emissions. In addition, wide C/N ratios in the organic layer as well as in the mineral soil (Table 1) indicate that nitrogen is hardly accessible for microorganisms in boreal forest soil. Mean NO emissions from the different soils varied to a higher extent and seem to react stronger to additional nitrogen input than N2O emissions.
 Another important factor influencing biotic and abiotic processes is soil pH. In the German spruce forest soil, a pH of 3.2 could also be partly responsible for high NO emissions. Li et al.  and Venterea et al.  suggested that soil acidity may be an important factor in promoting N losses from forests influenced by atmospheric nitrogen deposition, since chemodenitrification requires the protonation of NO2−. Highest NO emissions at low soil pH have also been reported by Ormeci et al.  for agricultural soils and Yamulki et al.  for grasslands. A soil pH of 6.7–7.1 at the only limestone site (AK) restricts self-decomposition of HNO2 [Van Cleemput and Samater, 1996] and may therefore be responsible for low NO fluxes. Another explanation of high fluxes from HW may be a stressed microflora in sites with high N deposition and low pH. Results from phospholipid fatty acid (PLFA) analysis (S. Zechmeister-Boltenstern, manuscript in preparation, 2004) showed that microflora in HW is in a stressed state whereas microflora in AK with neutral pH and high microbial biomass [Härtel et al., 2002] seems to be in a sound state. Changes in the structure of cell membranes of stressed microorganisms [Guckert et al., 1986] may cause losses of intermediary products in the nitrification or denitrification pathway and thereby lead to higher NO and N2O emissions. This assumption is supported by the finding of high N2 emissions and relatively low NO and N2O emissions in AK where denitrification seems to go the whole pathway from NO3− to N2.
 High N2O emissions from sites with low pH as observed in HW are in line with observations by Simek et al. , who found highest denitrifying enzyme activities at the prevailing natural soil pH even at a relatively low pH of 4.7, indicating that water-saturated soils are also capable of emitting high quantities of N2O at low soil pH.
 The relation of NO/N2O fluxes in our experiment also depended on the forest type. A ratio of 1.9 for the German spruce forest at HW is placed in contrast to ratios of 0.3 and 0.1 for the two Austrian beech forests at SW and KL, respectively. The mixed deciduous floodplain forest at TP showed a balanced ratio of 1.1. Others [Butterbach-Bahl et al., 1997; Van Dijk and Duyzer, 1999] also reported that NO emissions dominate in spruce forests while N2O emissions dominate in beech forests. The differences were suggested to arise from the structure and thickness of the humus layer and the density of the adjacent mineral soil. A laminar structure like a bed of deciduous leaves compared with that formed by needles may induce a barrier of gas diffusion and lead to longer residence time of NO within the soil, enhancing the possibility of NO consumption by denitrifying bacteria or chemical reactions. In our experiment, the HW spruce soil had a thick (3–4 cm) but lightly structured organic layer on top of mineral soil with low bulk density, compared to the beech soils with an organic layer of <1 cm and more dense mineral soil layers. N2O emissions from beech forests were found to exceed N2O emissions from spruce forests [Butterbach-Bahl et al., 1997; Brüggemann et al., 2003]. In this study, the absolute N2O emission of the HW spruce forest exceeded the emissions of SW and KL, but the differences in NO/N2O ratios may suggest generally higher N2O emissions in beech forests.
 The combination of soil temperature and soil moisture could explain a better part of variations in NO (up to 74%) and N2O (up to 86%) emissions for individual soils, but average emissions differed significantly between various forest soils. The finding of huge NO emissions from soil cores taken from the HW site, which are confirmed by previous results of field measurements, is most likely due to the combined effect of high nitrogen availability (due to high rates of atmospheric N deposition) and low soil pH values. No other site investigated in this study showed this combination. However, many forests in central Europe are growing on acid forest soils and are receiving atmospheric nitrogen loads in a comparable range as found for the HW site (>35 kg N ha−1 yr−1). Therefore it is very likely that high rates of NO emissions are not only restricted to the HW site, but can be found elsewhere across Europe were forest are growing on acidic soil and under chronically high rates of atmospheric N deposition. Preliminary results for a Sitka spruce forest in the Netherlands, where NO as well as N2O emissions even exceed emission rates as found at the HW site (J. Duyzer, personal communication), confirm this hypothesis.
 Furthermore, the results of this study show that the emissions of N trace gases from different forest sites are highly variable, and that temperature and moisture effects are prominent factors controlling the volume of emissions at a given site. For the description of site differences, more complex models are needed which best describe in detail the biogeochemical N and C turnover processes in forest ecosystems as well as the processes involved in trace gas production, consumption and emission. Such models are already available, e.g., the PnET-N-DNDC model [Li et al., 2000], and will be used within the NOFRETETE project (http://126.96.36.199/nofretete/) to calculate inventories of N trace gas emissions from forest soils across Europe.
 We would like to thank Christian Holtermann for technical support. The study is a part of the NOFRETETE project EVK2-CT2001-00106 funded by the European Commission DG Research – Vth Framework Programme.