Future scenarios of N2O and NO emissions from European forest soils

Authors


Abstract

[1] In this study we investigated possible feedbacks of predicted future climate change on forest soil NO and N2O emissions in Europe. For this we used two climate scenarios, one representing a 10-year period of present-day climate (1991–2000) and a 9-year period for future climate conditions (2031–2039). The climate scenarios were used to drive the GIS-coupled biogeochemical model Photosynthesis-Evapotranspiration-Model–Denitrification-Decomposition-Model (PnET-N-DNDC), which has currently been tested for its predicting capability for soil N trace gas emissions for various sites across Europe. The model results show a complex, spatially differentiated pattern of changes in future N2O and NO emissions from the forest soils across Europe, which were driven by the combined effect of changes in precipitation and temperature. Overall, the model predicted that N2O emissions from the European forest soils will on average decrease by 6%. This decrease was mainly due to the shift in N2O:N2 ratio driven by enhanced denitrification. NO emissions were found to increase by 9%. The increases in NO emissions were mainly due to increases in temperature. Only for the regions where soil moisture was predicted to markedly increase or suffer from water stress during the vegetation period, a reduction of NO emissions was simulated. The simulations show the possibility and feasibility for assessing climate change feedbacks on biogenic N trace gas emissions from soils at a regional scale.

1. Introduction

[2] Nitrous oxide (N2O) and nitric oxide (NO) are atmospheric trace gases, which are of importance for atmospheric chemistry and global warming [e.g., Cicerone, 1987]. The main sink for N2O is its stratospheric destruction by photolysis to NO. For that reason N2O is also involved in stratospheric ozone chemistry [Crutzen, 1976, 1995]. N2O is also an important greenhouse gas, and its globally averaged surface abundance was 314 ppb in 1998 [Intergovernmental Panel on Climate Change (IPCC), 2001]. In view of its atmospheric increase of about 0.3% per year and of its atmospheric life time of about 150 years [Khalil and Rasmussen, 1992] it can be expected that the contribution of N2O to global warming will further increase in the future. Approximately two thirds of the sources of atmospheric N2O and approximately one third of tropospheric NO are linked to the biogenic processes of nitrification and denitrification [IPCC, 2001; Davidson and Kingerlee, 1997], two key processes of global nitrogen turnover driven by microbes. Here soils play a dominant role as habitats for these microbes. Besides agricultural soils, the source strength of which is estimated to be approximately 3.3 Tg N yr−1 or 20% of total atmospheric N2O sources [IPCC, 2001], forest soils have been identified as significant sources for atmospheric N2O and NO [Potter et al., 1996; IPCC, 2001], and NO [e.g., Potter et al., 1996; Pilegaard et al., 1999]. Kesik et al. [2005] estimated that forest soils in Europe emit approximately 82 kilotonnes (kt) N2O per year (94 kt NO yr−1), which is approximately 15% of the N2O source strength of agricultural soils in Europe [Boeckx and Van Cleemput, 2001]. This relatively high contribution of forest soils, which cover approximately 22% of the simulated land area of Europe, to total N2O emissions is at least partly a result of chronically high rates of atmospheric N deposition to formerly N limited forest ecosystems in Europe [e.g., Brumme and Beese, 1992; Butterbach-Bahl et al., 2002]. Recent research has shown that the increased availability of N in temperate forest ecosystems has stimulated soil N2O and NO emissions [e.g., Butterbach-Bahl et al., 1997; Van Dijk and Duyzer, 1999; Zechmeister-Boltenstern et al., 2002].

[3] Like all biogenic processes, the microbial processes of nitrification and denitrification vary highly in magnitude with variations of environmental conditions. Soil temperature and soil moisture have been identified as major drivers on hourly to interannual timescales for observed temporal changes of N2O and NO emissions from forest soils [e.g., Firestone and Davidson, 1989; Gasche and Papen, 1999; Butterbach-Bahl et al., 2004a]. The effect of temperature for soil NO and N2O emissions is mostly direct. With increase in temperature, enzyme kinetics and thus metabolic turnover rates of nitrifiers and denitrifiers will increase up to an optimum at approximately 30°–35°C [Granli and Bøckmann, 1994]. In contrast to temperature the effect of soil moisture is mostly indirect. Soil moisture largely effects the rate of O2 diffusion into the soil profile and thus also determines if a soil is predominantly aerobic or anaerobic [Smith, 1980]. The oxygen status of the soil determines whether the aerobic process of nitrification or the anaerobic process of denitrification predominates. Although both processes can produce NO as well as N2O, production of N2O by denitrification can be more pronounced, since loss rates of N2O during the denitrification process are higher than during the nitrification process. Nitrification was identified as the dominating microbial production process for NO production [Conrad, 2002]. For the given reasons the optimum soil moisture for NO emissions is lower (approximately 50% water filled pore space) than for N2O emissions (approximately 60–65% WFPS). At soil moisture values >70–80%, N2O emissions will be strongly reduced, since N2O in the soil profile will be further reduced to N2 [Davidson et al., 2000]. However, under certain environmental conditions (e.g., high soil nitrate concentration), N2O instead of N2 can be the end product of denitrification [e.g., Firestone and Davidson, 1989; Conrad, 1996, 2002].

[4] Owing to the sensitivity of microbial N2O and NO production and consumption processes to changes in environmental conditions, one can assume that future changes in climate will strongly affect the magnitude of N2O and NO exchange between forest soils and the atmosphere. Owing to the complexity of these feedback processes, they cannot be adequately addressed in field studies from a limited number of sites. However, recently, a process-oriented biogeochemical model capable of simulating C and N cycling and the associated soil-atmosphere N trace gas exchange processes has been recently developed [Li et al., 2000; Butterbach-Bahl et al., 2001]. This model, the so-called PnET-N-DNDC model, has been successfully validated for its capability of simulating NO and N2O emissions for a wide variety of forest soils across Europe [Kesik et al., 2005]. Recent model applications show that the model can reliably simulate the seasonal and interannual variability of N trace gas fluxes due to changes in environmental conditions [Stange et al., 2000; Butterbach-Bahl et al., 2001; Kesik et al., 2005]. Thus the model can serve as an appropriate tool to evaluate possible feedbacks of forest soil N trace gas emissions to predicted climate changes. In this study we focused on the question of how predicted future changes in climate for Europe will affect the source strength and regional distribution of N2O and NO emissions from forest soils as compared to present-day conditions.

2. Materials and Methods

2.1. PnET-N-DNDC Model

[5] The biogeochemical model PnET-N-DNDC was used for the calculation of present and future soil N2O and NO emissions from European forest soils. The model has been applied for regional NO and N2O emission inventories for European temperate, Mediterranean and boreal forest ecosystems [Butterbach-Bahl et al., 2001, 2004b; Kesik et al., 2005], and recently also for the calculation of a N2O emission inventory for tropical rain forests in Australia [Kiese et al., 2005]. The PnET-N-DNDC model was developed to predict soil carbon and nitrogen biogeochemistry in temperate forest ecosystems and to simulate the emissions of N2O and NO from forest soils [Li et al., 2000; Stange et al., 2000]. The model is mainly based on the PnET model (Photosynthesis-Evapotranspiration-Model), the DNDC model (Denitrification-Decomposition-Model), and a nitrification module. For further details on the PnET-N-DNDC model, we refer to Li et al. [1992, 1996, 2000], Li [2000], Stange et al. [2000], Butterbach-Bahl et al. [2001, 2004b], Kiese et al. [2005], and Kesik et al. [2005].

[6] To simulate N trace gas emissions for a specific site in daily resolution the PnET-N-DNDC requires the following input parameters: daily climate data (precipitation, minimum and maximum air temperature, optional: solar radiation), soil properties (texture, clay content, pH, soil organic carbon content, stone content, humus type), and information on forest properties (forest type and age, aboveground and below-ground biomass, plant physiology parameters). The PnET-N-DNDC is currently parameterized for 12 tree species/genera, i.e., pine, spruce, hemlock, fir, hardwoods, oak, birch, beech, slash pine, larch, cypress and evergreen oak. If there are no site-specific forest data available except for the forest type and age, the model will use default values. For each forest type these default values are taken from an internal database, which is based on available literature data on tree physiological (e.g., maximum photosynthesis rate, water use efficiency) or forest stand properties (e.g., forest biomass) [Li et al., 2000]. Furthermore, the model needs information about inorganic N concentrations in rainfall, which are used to calculate throughfall values for N, a surrogate of wet and dry deposition. The amount of throughfall is dependent on forest type and N concentration [Li et al., 2000].

[7] The capability of the PnET-N-DNDC model to simulate N trace gas emissions from forest soils for Europe was tested by comparing model results with results from field measurements at 19 different field sites across Europe and one in the USA [Kesik et al., 2005]. The testing sites were located in different climatic regions, namely boreal, temperate oceanic, temperate continental, Mediterranean, in order to assure a wide applicability of the model. The results of the study by Kesik et al. [2005] show that the PnET-N-DNDC model was capable of capturing observed differences between high and low emitting sites, on the basis of general information on soil and vegetation properties and by considering the local climatic conditions. The linear regression between simulated and observed annual mean emission rates was r2 = 0.68 for N2O and r2 = 0.78 for NO [Kesik et al., 2005]. For both N trace gases the model tended to nonsystematically underestimate emissions at the test sites by approximately 24% for N2O and by approximately 27% for NO. For further details on the model evaluation and performance, see Kesik et al. [2005], and the earlier work by Stange et al. [2000] and Butterbach-Bahl et al. [2001, 2004b]. Differences in N2O and NO emissions, precipitation and air surface temperature for present (1991–2000) and future (2031–2039) climate conditions were analyzed using analysis of variance (ANOVA, α = 0.05). The LSD test was used for multiple comparisons between means. All statistical analyses were performed with SPSS 8.0 (SPSS Inc., Chicago) and Microcal Origin 4.0 (Microcal Software, Northampton).

2.2. GIS Database Construction

[8] For the regionalization of N trace gas emissions by use of the PnET-N-DNDC model a detailed GIS database covering all EU states plus Romania, Bulgaria, Switzerland and Norway was created. This GIS database contained all relevant initialization and driving parameters and variables. Spatially resolved information included soil, forest and climate properties.

[9] Climate information for the years 1991–2000 and 2031–2039 was obtained from calculations by the coupled climate-chemistry model MCCM (Multiscale Climate Chemistry Model [Grell et al., 2000]) using ECHAM4 simulations for current and future climate as boundary conditions. The ECHAM4 model runs by the Max-Planck Institute for Meteorology in Hamburg, Germany, are based on historic atmospheric CO2 concentrations for the years 1860–1990 and on the IPCC IS92a scenario [IPCC, 1992] for the years 1990–2100. From this simulation the results, representing present (1991–2000) and future climate conditions (2031–2039), have been used for calculating a regional climate simulation for Europe (except the part of Scandinavia, north of 66°N and Iceland, which were out of the domain of the regional climate model) with MCCM. The MCCM calculates simultaneously the meteorology and chemistry for a defined region in high spatial resolution. For the climate simulation the MCCM model has been directly nested into the global ECHAM4 model with a resolution of 60 km × 60 km. By this nesting procedure, the climate information of the global ECHAM4 model with a resolution of only 200–300 km (T42) was regionalized [Forkel and Knoche, 2006; Knoche et al., 2003]. The climate data, resulting from the MCCM simulation runs, were used as climate input for the PnET-N-DNDC model, i.e., daily precipitation, daily minimum and maximum air temperature and daily mean solar radiation.

[10] Within the NOFRETETE project (http://195.127.136.75/nofretete/), information on atmospheric N deposition for the year 2000 was provided by the Norwegian Meteorological Institute (DNMI), from the inputs of the EMEP MSC-W photo-oxidant model [Simpson et al., 2003]. In our model runs we assumed that N deposition remains at the year 2000 level, which is in line with conclusions by Galloway et al. [2004] that the magnitude of total inorganic nitrogen in western Europe will not significantly increase until 2050. Data on soil properties (e.g., pH, SOC, texture, forest type distribution) were provided by the Joint Research Centre (JRC) in Ispra, Italy. Since data were delivered in different formats and projections, transformations into the ArcGIS by ESRI format with the Lambert-Azimuthal projection on the basis of the EMEP (European Monitoring and Evaluation Program) raster were necessary. The used EMEP raster is a polar stereographic projected grid with a resolution of 50 km × 50 km at 60° North [Simpson et al., 2003] (http://www.emep.int). For more details on forest and soil information, we refer to Kesik et al. [2005].

[11] The forest, soil, and climate information was aggregated and linked to the EMEP raster. An individual identification number was assigned to each of the 2386 grid cells of the EMEP raster covering the simulated area of Europe.

3. Results and Discussion

3.1. Simulated Changes in European Climate

[12] The regionalized climate scenarios show that for the period 2031–2039 mean annual surface temperatures in Europe will be approximately 1.8°C higher (from 7.6°C to 9.4°C, Table 1) compared to the period of 1991–2000 (Table 1). The highest average increases in surface temperatures are predicted for the winter and spring season across Europe with 1.8°C and 2.2°C, respectively. The increases in average summer and autumn temperatures across Europe are 1.7°C and 1.5°C, respectively. However, the model runs show a distinct regional pattern of temperature changes in Europe. Figure 1 shows an increase in the average annual temperature in Scandinavia, the Baltic States, and parts of Ukraine and Russia for the period 2031–2039 of 2°C or more. For other areas such as the western part of the Alps and some maritime areas of Portugal, Spain and France the predicted temperature increase is in the range of 1.5°–2.0°C. For all other parts of Europe, i.e., for most of the Mediterranean area, central Europe, and the UK and Ireland, the predicted mean annual temperature change is less pronounced, and in the range of 0.75°–1.5°C. Both in the Mediterranean region and in Southern Norway and Sweden future summer temperatures may strongly increase by up to 2.5°C. This locally fixed temperature increase in southern Norway and Sweden can be due to an overestimation of the temperature for the simulated future climate (2031–2039), as such a persistent trend of temperature increase for this region can not be found for the simulation years 1991–2000. For the rest of Europe the temperature increase is about 1.5°–2°C.

Figure 1.

Absolute changes in (a) mean annual temperatures (°C), (c) spring temperatures, and (e) summer temperatures between future (2031–2039) and present (1991–2000) climate conditions. Relative changes in the (b) annual sum of precipitation (%) or (d) summer period precipitation or (f) summer period precipitation between future (2031–2039) and present (1991–2000) climate conditions.

Table 1. Seasonal Variations in N2O and NO Emissions From Forest Soils, Precipitation, Air Surface Temperature, and Water-Filled Pore Space (WFPS) for Present (1991–2000) and Future (2031–2039) Climate Conditionsa
SeasonPresent/FutureN2ONOPrecipitation, mmTemperature, degWFPS, %
kg N ha−1kt Nkg N ha−1kt N
  • a

    Different letters after values indicate significant seasonal differences for the individual parameters.

Dec–FebPresent0.08 a10.3 a0.08 a9.7 a246 a1.1 a0.61 a
Future0.08 a10.3 a0.06 a8.3 a257 a2.9 b0.66 b
March–MayPresent0.17 b21.4 b0.09 ab12.4 ab181 b5.0 c0.56 ac
Future0.13 c17.2 c0.12 bc14.8 bc167 b7.2 d0.59 a
June-AugPresent0.18 d23.7 d0.22 d28.7 d110 c15.8 e0.42 d
Future0.18 bd23.0 bd0.26 d32.8 d104 c17.5 f0.40 d
Sep–NovPresent0.10 e12.8 e0.13 c17.1 c245 a8.4 g0.51 ce
Future0.11 e13.7 e0.14 c18.4 c245 a9.9 h0.50 e
YearPresent0.5368.30.5268.07827.60.53
Future0.5064.30.5874.37739.40.54

[13] Total average annual precipitation is predicted to be nearly unchanged across Europe for the period 2031–2039 as compared to the period 1991–2000 (Table 1). However, regional and seasonal changes in precipitation do occur. Figure 1 shows that in some parts of the Mediterranean region and parts of the west coast of France and Scotland annual precipitation is predicted to increase up to 30%. In all other European regions, except for parts of Norway and Finland, annual precipitation is predicted to decrease in the range of 0–20%. Considering the seasonality of rainfall, the MCCM model predicts slightly higher values for winter precipitation, but lower precipitation for spring and summer, in most areas. However, none of the seasonal changes in precipitation between the present and the future climate are statistically significant (Table 1). Especially pronounced are the seasonal changes in precipitation for the Mediterranean area, where summer precipitation is predicted to decrease regionally by up to 80%. Also for other regions such as central Europe (except northern Germany), Sweden and Finland, the east and southeast of Europe, decreasing values of summer precipitation are predicted for the period 2031–2039. However, along the east coast of the North Sea, mainly for Norway, Denmark, and northern Germany, parts of Scotland and Northern Ireland, and northern Portugal, precipitation is predicted to increase up to 30% during the summer months.

[14] Future changes in precipitation and temperature will also feedback on the soil moisture. Figure 2 shows that under future climate conditions mean annual soil moisture across Europe, as calculated by the PnET-N-DNDC model, will not be changed significantly (1.6%). Also no obvious relationship between predicted soil moisture changes and soil texture can be demonstrated (Figure 2). Significant changes in soil moisture could only be found on a seasonal basis (Figure 3), with a marked increase in soil moisture in spring (+22%) and slight decreases in summer (−2.4%) and autumn (−1.6%). However, on a regional basis these changes were more pronounced and mirrored changes in precipitation (Figure 3).

Figure 2.

Relative changes in soil moisture (given as percent water filled pore space, WFPS) for European forest soils under future climate conditions as compared to present day as a function of classified percent clay content. Gray bars indicate the range of SD from the mean (straight line).

Figure 3.

Relative changes in soil moisture (as percent change in WFPS with WFPS: percent water filled pore space) for European forest soils under future climate conditions as compared to present day for winter (DJF) spring (MAM), summer (JJA), and autumn (SON).

[15] The calculated changes in future climate as calculated by the MCCM model are supported by another recent study. Räisänen et al. [2004] calculated the climate change from 1961–1990 to 2071–2100 with the regional model RCAO driven by the global models HadAM3H and ECHAM4/OPYC3. These authors used the two SRES scenarios A2 and B2. All four simulations by Räisänen et al. [2004] indicate for northern Europe a larger warming in winter than in summer. In southern and central Europe a very large increase in summer temperatures was predicted, which was most pronounced for the southwestern parts of Europe and most notably in the ECHAM4/OPYC3-based A2 scenario. For this scenario the warming locally exceeds 10°C (e.g., in France). These results by Räisänen et al. [2004] are consistent with results obtained by the MCCM simulation, even though predicted changes in summer temperatures for France by MCCM are less pronounced. Also with regard to precipitation changes the regional climate simulation used in this study is in agreement with results of Räisänen et al. [2004]. In the four scenario simulations these authors found a general increase in winter precipitation in northern and central Europe and a decrease in summer precipitation in central and southern Europe. The latter finding was also supported by MCCM regional climate predictions, i.e., a precipitation decrease in the summer months for central and south Europe. The range as well as the regional pattern of predicted changes in precipitation and temperature used in our study are also consistent with a series of recent studies on future climate change in Europe such as Maracchi et al. [2005], Benestad [2005], Giorgi et al. [2004], and Kjellstrom [2004].

[16] It needs to be noted that in this study only the output of regionalized ECHAM4 data for the IPCC 92a scenario were used as drivers for the biogeochemical model. Outputs of other global climate models (GCM) for future climate have been shown to be highly variable with regard to regional precipitation patterns [Dubrovsky et al., 2005; Wang, 2005; Hanssen-Bauer et al., 2003]. However, for Europe the majority of GCMs is highly consistent in predicting future precipitation responses: a substantial increase in precipitation for December, January and February for middle and high latitudes, but a decrease for the Mediterranean. Also for the summer period June, July and August, increased drought periods are consistently predicted by GCMs for the Mediterranean, but not for northern Europe [Wang, 2005]. Predicted changes in soil moisture are even more variable, since the global and regional pattern of the direction of soil moisture changes differs significantly from model to model mostly owing to differences in land surface parameterization [Wang, 2005]. However, in our study we calculated soil moisture values with our biogeochemical model based on predicted meteorological data and soil and forest stand properties. The variability in the predictions of future changes in precipitation among different GCMs would directly feedback on the magnitude of N trace gas emissions from forest soils. For entire Europe, Kesik et al. [2005] showed that forest soil N2O and NO emissions can vary under present climate conditions by 15% for different meteorological years. We do need to assume that the use of other GCMs' climate predictions would have resulted at least in a comparable uncertainty range with regard to forest soil N trace gas emissions.

3.2. Future Changes in Regional N2O Emissions

[17] For exploring possible feedbacks of future climate change on N trace gas emissions from European forest soils the PnET-N-DNDC model was run for the periods 1991–2000 and 2031–2039 with the climate scenarios described above. In our simulations, mean annual N2O emissions from forest soils across Europe were found to slightly decrease by 4 kt N2O-N (−6%) for the period 2031–2039 as compared to the period 1991–2000 (68.3 kt N2O-N) (Tables 13). This is mainly due to the significant decrease in emissions during spring from 21.4 kt N2O-N under present climate conditions to 17.2 kt N2O-N under future climate conditions (Tables 1 and 2). However, the model runs with PnET-N-DNDC also revealed a pronounced interannual variability in N2O (but also NO) emissions (Table 2), which are due to changes in the meteorological conditions from year to year. This reflects the sensitivity of the PnET-N-DNDC model toward the meteorological drivers. Furthermore, the model runs also revealed a pronounced regional pattern of future changes in N2O emissions due to climate change (Table 3 and Figure 4). In Scandinavia, and in some regions of Poland, Estonia, Latvia, Lithuania, Romania, Slovakia, and Bulgaria future N2O emissions may decrease by up to 33%. Also, for the Alpine region the model predicted a decrease of N2O emissions from forest soils by up to 30% in the future. In most of these regions light texture soils predominate, i.e., soils with clay contents mostly below 15% (see, e.g., soil texture map of Kesik et al. [2005]). For such soils, stable or decreasing precipitation and increasing temperature under future climatic conditions will result in decreased soil moisture values especially for the summer and autumn season (see, e.g., Figures 3b and 3c). Owing to the improved aeration the denitrification activity in such soils, and thus, N2O production by denitrification, decreases. On the other hand NO production by nitrification increases in these regions during these periods (see also below). For winter and spring seasons a significant increase in soil moisture can be observed for Scandinavia and the Baltic States. This also feedbacks on forest soil N2O emissions, since under such conditions even in light textured soils oxygen diffusion into the soil profile is strongly reduced. As a consequence, predominantly anaerobic soil conditions are predicted and the end product of denitrification is N2 rather than N2O (see also below). However, in central Europe and in the Mediterranean area our model predicted an increase of forest soil N2O emissions by approximately 15%. In northern Germany, the Benelux states and the UK with Ireland, N2O emissions may even increase by up to 20%.

Figure 4.

Relative changes in (a) mean annual N2O emissions (%), (c) mean spring N20 emissions, and (e) mean summer N2O emissions between future (2031–2039) and present (1991–2000) climate conditions. Relative changes in mean (b) annual or (d) spring or (f) summer NO emissions between future (2031–2039) and the present (1991–2000) climate conditions for NO.

Table 2. Interannual Variations in Mean Annual N2O and NO Emissions From Forest Soils, Air Surface Temperature and Mean Annual Sum of Precipitation for Present (1991–2000) and Future (2031–2039) Climate Conditionsa
 Temperature, °CPrecipitation, mmN2O, kg N ha−1NO, kg N ha−1
  • a

    CV represents the coefficient of variation.

Present
19917.538240.510.49
19926.717510.550.24
19937.408120.530.81
19946.847330.590.45
19957.817300.540.54
19968.008030.530.47
19978.018160.480.51
19987.327110.560.49
19997.658560.550.52
20008.907870.480.55
Mean7.627820.530.51
CV, %8.256.166.7027.29
 
Future
20319.058010.520.58
20329.587590.480.59
20338.747530.590.58
20348.827090.520.56
203510.487690.470.62
20369.408090.480.59
20378.597820.520.53
20389.558260.490.59
203910.497490.450.59
Mean9.417730.500.58
CV, %7.484.628.164.48
Table 3. Average and Total Simulated N2O and NO Emissions From Forest Soils for Present and Future Climatic Conditions for Individual European Countries
 N2ONO 
 1991–20002031–2039 1991–20002031–2039 
CountryForest Area, km2kg N ha−1 yr−1kt N yr−1kg N ha−1 yr−1kt N yr−1Change in Emission, %kg N ha−1 yr−1kt N yr−1kg N ha−1 yr−1kt N yr−1Change in Emission, %
Andorra2320.163.7 × 10−30.173.9 × 10−360.214.9 × 10−30.122.8 × 10−3−44
Austria24,0320.461.110.431.03−70.441.070.411.00−6
Belgium76990.770.590.870.67141.531.181.731.3313
Bulgaria28,4940.521.500.511.45−30.320.900.300.86−4
Croatia12,5740.460.570.490.6280.460.580.460.581
Czech.20,4060.430.880.440.9130.661.360.681.392
Republic18,6080.601.120.721.34210.891.661.051.9518
Denmark18,3410.601.100.400.73−330.500.910.611.1223
Estonia116,1260.829.540.627.21−240.485.590.596.8022
Finland132,3950.445.810.486.3490.557.230.557.321
Germany117,8490.566.570.637.43130.9511.151.0412.2410
Gibraltar0.430.582.5 × 10−50.632.7 × 10−590.062.7 × 10−60.072.9 × 10−67
Greece30,6760.501.540.531.6470.280.870.290.881
Hungary21,1810.631.320.621.3200.350.730.340.72−2
Irish Republic55230.200.110.240.13200.500.280.490.27−3
Italy59,8340.482.890.492.9630.402.390.402.421
Latvia28,2290.661.860.461.29−300.571.600.671.8918
Liechtenstein890.564.9 × 10−30.363.2 × 10−3−350.423.8 × 10−30.302.6 × 10−3−30
Lithuania18,8430.460.870.360.68−230.400.760.410.782
Luxembourg10320.450.050.520.05150.810.080.820.082
Monaco0.210.193.9 × 10−60.214.4 × 10−6120.112.4 × 10−60.081.7 × 10−6−28
Netherlands82711.170.971.361.13162.532.093.012.4919
Norway108,9870.181.910.131.45−240.161.740.060.67−62
Poland78,3580.564.260.503.82−100.765.840.866.5813
Portugal32,7130.411.350.441.4470.140.470.120.40−15
Romania41,2840.692.850.622.56−100.411.700.421.732
San Marino0.350.238.2 × 10−60.258.9 × 10−680.248.4 × 10−60.207.0 × 10−6−17
Slovakia91620.680.620.600.55−120.470.430.480.442
Slovenia78810.480.30.490.3920.490.390.450.35−8
Spain128,4840.547.490.588.0880.253.490.243.27−6
Sweden163,7820.609.910.487.93−200.7111.700.9215.1229
Switzerland12,4070.400.490.350.43−120.420.530.340.42−21
United Kingdom22,4810.270.620.330.73190.561.270.551.23−3
Sum1,283,978 68.27 64.32  67.96 74.32 
Average 0.53 0.50 −60.53 0.58 9

[18] The magnitude of forest soil N2O emissions for present-day conditions in this study is well in agreement with results from previous work by Kesik et al. [2005]. On the basis of an extensive model testing for various sites across Europe, Kesik et al. [2005] used the GIS coupled PnET-N-DNDC model to estimate present-day N trace gas emissions from forest soils using daily meteorological data for the years 1990, 1995 and 2000 as derived from the EMEP MSC-W model [Sandnes-Lenschow and Tsyro, 2000; Simpson et al., 2003]. Mean N2O emissions in the study by Kesik et al. [2005] using EMEP climate data were on average slightly higher (0.58 kg N2O-N ha−1 yr−1) than those found in this study (0.53 kg N2O-N ha−1 yr−1), owing to differences in the simulation area (in the current study we excluded parts of Scandinavia), length of the simulation period (10 years versus 3 years), and differences in the climate data sets (MCCM versus EMEP).

[19] The simulation results show that even in view of a general increase in temperature under future climate conditions, N2O emissions are projected to slightly decrease. Most field and laboratory studies show that temperature has a stimulating effect on N2O emissions [e.g., Granli and Bøckmann, 1994; Brumme, 1995; Papen and Butterbach-Bahl, 1999]. Reported Q10 values for soil N2O emissions are in the range of 0.9 to 23 [see, e.g., Smith, 1997]. For this reason, one would expect that under future climate conditions N2O emissions should generally increase with a predicted increase in temperature across Europe. The much more differentiated picture on future changes in forest soil N2O emissions as obtained in our model study is a result of the complexity of the processes involved in production and consumption of this trace gas. If mean annual soil moisture increases by more than 10% (relative change in water-filled pore space (WFPS); for mean absolute values of WFPS across Europe, see Table 1) and mean annual soil temperature by more than 2.0°C under future climate conditions, simulated N2O emissions are predicted to decrease by at least 10% (Figure 5). Only under the conditions with average annual soil moisture remaining unchanged or only slightly decreasing in comparison with the decrease in present-day conditions, N2O emissions are predicted to increase. The main reason for the simulated decrease in N2O emissions is a shift in the relative proportions of N2O versus N2 emitted from the soils under the future climate conditions (Figure 6). In our simulations the emitted N2O:N2 ratio is mostly predicted to decrease. This means that under future climate conditions in regions for which higher annual average soil moisture values as well as higher soil temperature are predicted, a tendency can be found that more nitrogen will be emitted as N2 rather than N2O. These predictions are in line with laboratory studies, which have shown that the N2O:N2 ratio decreases with increasing temperature [Nommik, 1956; Keeney et al., 1979]. Moreover, it is well known that if WFPS increases above a certain threshold of approximately 70–80% [Davidson, 1991], N2 rather than N2O will be the main product of denitrification.

Figure 5.

Contour plot showing the effect of average annual relative changes in soil water filled pore space (%) and absolute changes in soil temperature (°C) on relative changes in forest soil N2O emissions (%). For this analysis, simulation results for present and future climate predictions for all grid cells across Europe were used. Prior to the calculation of contour lines with SigmaPlot2000 (SPSS, Inc.) data were smoothed using a second-order polynom (r2 of the underlying polynom = 0.68).

Figure 6.

Contour plot showing the differences in the average annual N2O:N2 ratio between future and present climate conditions as influenced by relative changes in soil water filled pore space (%) and the difference in soil temperature (°C) between future and present climate conditions. Positive values of the N2O:N2 ratio imply an increase in N2O emissions and negative values imply an increase in N2 emissions. For this analysis, simulation results for present and future climate predictions for all grid cells across Europe were used. Prior to the calculation of contour lines with SigmaPlot2000 (SPSS, Inc.) data were smoothed using a second-order polynom (r2 of the underlying polynom = 0.25).

[20] The provided insights in possible consequences of future climate changes on forest soil N trace gas emissions can only be obtained if biogeochemical models such as PnET-N-DNDC are used, which do not only rely on empirical relationships between, for example, temperature, moisture and N trace gas fluxes, but which are based on the underlying main microbial and physico-chemical processes involved in the production, consumption and emission of N trace gases [Li et al., 2000]. For example, denitrification is described in the model as a series of sequential reductions driven by microorganisms using N oxides as electron acceptors under anaerobic conditions [e.g., Kuenen and Robertson, 1987]. As intermediates of the processes, NO and N2O are tightly controlled by the kinetics of each step in the sequential reactions [Li et al., 2000]. As a soil becomes increasingly anaerobic, for example, owing to high WFPS or to stimulated soil respiration at higher soil temperatures, the model will simulate that more N2O, either derived from nitrification or denitrification, will be reduced to N2 as compared to conditions when the soil is well aerated. This is in accordance also with laboratory studies on soil N2O and N2 emissions [e.g., Nommik, 1956; Keeney et al., 1979].

3.3. Future Changes in Regional NO Emissions

[21] In contrast to the average decrease of forest soil N2O emissions under predicted future climate changes, forest soil NO emissions are predicted to increase in the future by 9% on average (from 68.0 to 74.3 kt NO-N) (Tables 13). The increase in NO emissions is most pronounced in summer with a predicted increase from 28.7 kt NO-N under present-day climate conditions to 32.8 kt NO-N in the future. The model-simulated distinct regional differences in NO emissions in Europe under predicted climate changes. Only for a few regions such as the Alpine region, the UK, Ireland, Norway, Spain, Portugal and some countries in east Europe (Bulgaria, Hungary, Slovenia) a decrease of average annual NO emissions under future climate conditions was predicted (down to 66%). However, in all other regions the model predicted an increase of NO emissions for the period 2031–2039 as compared to the period 1991–2000. Especially for the Baltic Sea region, NO emissions from forest soils may increase by up to 29% (Table 3 and Figure 4). As already outlined above, this region is characterized by prevailing light textured soils. The regional NO emission changes (Figure 4) in summer are not significantly different from that of annual average NO emission changes under future climate conditions. However, for central Europe the increase of forest soil NO emissions are more pronounced in the summer months than throughout the year.

[22] NO emissions from soils are mainly driven by the activity of nitrifying organisms in the soil [e.g., Conrad, 1996, 2002]. NO production in soils and NO emissions at the soil-atmosphere interface have been found to be positively correlated with temperature [Valente and Thornton, 1993; Gasche and Papen, 1999; Ludwig et al., 2001], mostly owing to the stimulating effect of temperature on NO production via nitrification [Conrad, 2002]. The optimal soil moisture range for NO emissions is less than for N2O emissions, since at higher soil moisture values the O2 availability in the soil profile decreases owing to diffusion restrictions. Besides soil moisture the soil texture has a main influence on O2 availability in the soil profile. For that reason, regions with predominating light textured soils, which do facilitate O2 diffusion into the soil profile, as, for example, in the area of the Baltic Sea, tend to have elevated NO emissions. Laboratory and field studies showed that at WFPS values above 60% NO will be further reduced to N2O or N2 [e.g., Davidson, 1991; Bollmann and Conrad, 1998]. Most of the NO is already consumed in the soil before it will be released to the atmosphere. Comparable consumption and production mechanisms are also implemented in the PnET-N-DNDC model. In the model we use the concept of an anaerobic balloon to describe the prevalence of aerobic or anaerobic zones in dependency of the aeration status of the soil [Li et al., 2000], i.e., on the basis of a gas diffusion model which takes into account soil texture and moisture as well as considering heterotrophic and autotrophic O2 consumption the availability of O2 is calculated for the different soil layers. On the basis of the O2 availability the soil layer is partitioned into anaerobic and aerobic soil zones. If a soil becomes predominantly anaerobic, for example, owing to diffusion restrictions at high WFPS values, NO produced by nitrification (or denitrification) can be taken up and further reduced by denitrifiers (the model does not allow aerobic NO consumption at present). In our simulations, mean annual NO emissions were predominantly stimulated in those regions which showed the highest increase in mean annual temperature and an increase in WFPS of less than 18% (Figure 7). If mean annual WFPS was further increased under future climate conditions NO emissions would strongly be reduced to N2O or N2, owing to enhanced denitrification under anaerobic conditions. The decrease in summer NO emissions for some Mediterranean regions in our model simulations were due to the effect of water stress on microbial processes. This prediction is in accordance with results from field and laboratory studies, which reveal a strong decrease in NO emissions if soils experience a longer dry period [e.g., Davidson, 1991; Ludwig et al., 2001], even though large pulses of NO emissions may occurring following rewetting [Davidson et al., 1993; Butterbach-Bahl et al., 2004a].

Figure 7.

Contour plot showing the effect of relative changes in soil water filled pore space (%) and absolute changes in soil temperature (°C) on relative changes in forest soil NO emissions (%). For this analysis simulation results for present and future climate predictions for all grid cells across Europe were used. Prior to the calculation of contour lines with SigmaPlot2000 (SPSS, Inc.) data were smoothed using a second-order polynom (r2 of the underlying polynom < 0.1).

4. Conclusion

[23] In this study we have explored the consequences of predicted future climate changes on forest soil NO and N2O emissions using a GIS-coupled biogeochemical model. The results show that future changes in N trace gas emissions from forest soils cannot be predicted by changes in one factor (e.g., temperature) alone. In our study, changes in precipitation and thus, in soil moisture, exerted a greater effect on N2O emissions than changes in temperature. For NO, temperature was the main driving factor. Overall, mean N2O emissions across Europe were predicted to decrease by 6%, whereas NO emissions were predicted to increase by 9%. The decrease in N2O emissions was mainly due to the decrease in the N2O:N2 ratio, i.e., under future climate conditions the model simulated an increase in N2 emission by denitrification at the expense of N2O emission via denitrification. However, our knowledge about the combined effect of changes in moisture and temperature on annual N2O, NO and N2 emissions from soils is still rather limited. This is especially true with regard to N2 emissions due to denitrification in soils. Furthermore, our study shows a complex, regionally differentiated response of N trace gas emissions to future climate change, and underlines that more experimental and modeling studies are required to better understand the consequences of climate changes for biosphere-atmosphere exchange processes. Especially, multifactorial manipulation experiments in which the combined effects of changes in temperature, precipitation and atmospheric CO2 concentrations on biosphere-atmosphere greenhouse gas exchange are investigated are urgently needed for further model validation.

Acknowledgments

[24] The work was funded by the European Commission in the NOFRETETE project (EVK2-CT2001-00106) of the fifth framework program and the NATAIR project (contract 513699) of the sixth framework programme. The work of David Simpson was also supported by the Co-operative Programme for monitoring and Evaluation of the Long-range Transmission of Air Pollutants in Europe (EMEP) under UNECE.

Ancillary