Interactive comment on “ Annual emissions of CH 4 and N 2 O , and ecosystem respiration , from eight organic soils in Western Denmark managed by agriculture

S. O. Petersen, C. C. Hoffmann, C.-M. Schäfer, G. Blicher-Mathiesen, L. Elsgaard, K. Kristensen, S. E. Larsen, S. B. Torp, and M. H. Greve The paper reports on annual emissions of CH4 and N2O from organic soils managed by agriculture in Denmark. Overall the manuscript is well written and carefully justified. However, a few shortcomings should be addressed: the first is the fact that only one year was used. Tha authors highlighted this problem and presented the average temperature and precipitation of the reported year compared to the 20 years average in Table 1. Some of the precipita-


Introduction
On a global scale, organic soils (Histosols) represent a carbon stock that is equivalent to nearly 50 % of atmospheric CO 2 (Drösler et al., 2008).In Europe organic soils cover approximately 5 % of the total land area (European Soil Bureau Network, 2005), and this proportion is similar in Denmark, where organic soils are widely used by agriculture as pastures or for crop production (Maljanen et al., 2010).With an organic matter content of 20 % or more in the top soil (FAO, 1998) the C and N turnover and gaseous exchanges of Histosols are significant for the greenhouse gas balance of Danish agriculture, even if their total area is relatively small (Gyldenkaerne et al., 2005).
The greenhouse gas balance of organic soils may include contributions from CO 2 , CH 4 and N 2 O.The net flux of CO 2 is determined by the balance between total ecosystem respiration (R eco ) and photosynthesis, and R eco can itself be separated into soil and plant respiration (Lambers et al., 1998).Jacobs et al. (2007), quantifying annual fluxes of CO 2 from several Dutch grasslands, found a net release of CO 2 from those on organic soil, but a net sequestration on mineral soils, Published by Copernicus Publications on behalf of the European Geosciences Union.
indicating that soil organic matter decomposition is critical for the carbon balance.
Methane production is expected to mainly occur below the groundwater table, but even here it can vary by several orders of magnitude (Segers, 1998).Controlling factors include anoxia, availability of substrates, the presence of microbial consortia capable of processing these substrates to CH 4 , and competition from other processes such as sulfate reduction (Yavitt and Lang, 1990;Segers, 1998).Drainage will limit the production of CH 4 , but also increase the potential for CH 4 oxidation during passage through the unsaturated zone to the atmosphere.As a result CH 4 fluxes from drained organic soils are consistently low or slightly negative (Langeveld et al., 1997;Drösler et al., 2008;Maljanen et al., 2010).Methane oxidation potentials appear to be highest near the oxic/anoxic interface.Hornibrook et al. (2009) found that CH 4 dissolved in the pore water of four Welsh peatland soils was nearly always zero at the groundwater table and concluded that emissions observed were predominantly mediated by vascular plants.The ability of plants with aerenchymous tissue to transport CH 4 to the atmosphere when CH 4 concentrations build up around the roots is well established (Laanbroek, 2010), and typically occurs when the soil is near saturation (Strack et al., 2006).
Degradation of soil organic matter as a result of drainage and cultivation will stimulate net N mineralization and N transformations via nitrification and denitrification which can then lead to N 2 O production (Freibauer et al., 2004;Goldberg et al., 2010).Maljanen et al. (2010), reviewing GHG monitoring studies from the Nordic countries, reported that N 2 O emissions from organic soils in agricultural use were on average four times higher than those from mineral soils, indicating that N 2 O derived from soil organic matter decomposition dominate overall fluxes.According to Maljanen et al. (2010) annual N 2 O emissions from managed organic soils range from 0.2 to 5.5 g m −2 , with an average of 1.6 g m −2 , but no studies from Denmark were available.
Denmark recently adopted Art.3.4 of the Kyoto protocol concerning carbon stock changes within agriculture and forestry.For organic soils management such as drainage and cultivation will influence C turnover and losses, and consequently fluxes of CH 4 and N 2 O derived from soil organic matter, and the total GHG balance of managed organic soils must therefore be accounted for.The field monitoring study reported here estimated annual fluxes of CH 4 and N 2 O, as well as ecosystem respiration, from eight organic soils managed by agriculture.Protocols and instrumentation used at all monitoring sites were identical, as were procedures for sample analysis and data processing.

Selection of monitoring sites
In the selection of locations for monitoring information about geology and geochemistry, as well as climate variables (insolation, precipitation, temperature) and land use were considered.Denmark has been sub-divided into landscape types, or geo-regions, based on age and genesis (Madsen et al., 1992).The three landscape types with the largest recorded areas of organic soil were: The outwash plains (including subglacial stream trenches) and hill islands of Western Jutland (total area 81 150 ha; region W ), the raised sea bottom of Northern Jutland (total area 21 199 ha; region N), and the younger moraine landscape of Eastern Denmark, including kettle holes and lateral moraine (total area 34 335 ha; region E).The moraine deposits from the last (Weichselian) glaciation, which cover the eastern part of Denmark, have a high calcium content and thus differ geochemically from the deposits of northern and central Jutland.The latter regions, in contrast, have areas with high levels of pyrite (FeS 2 ).
Denmark is characterized by minor gradients in temperature and insolation, and a more significant gradient in precipitation, which ranges from around 500 to 900 mm yr −1 .Table 1 presents selected information about annual mean air temperatures and precipitation measured at the monitoring sites, together with the corresponding information for the period 88/89 to 08/09 based on data from a nation-wide 10 × 10 km 2 (precipitation) or 20 × 20 km 2 grid (temperature) of the Danish Meteorological Institute.Included as online supplementary information are graphical presentations of mean monthly temperatures and precipitation for the monitoring period and the preceding 20-yr period.
The predominant land uses for organic soils managed by agriculture were identified using the General Danish Agricultural Register (GLR).The land use categories arable crops in rotation (AR), permanent grassland (PG) and rotational grass (RG) together account for almost all of the area with organic soils managed by agriculture.Locations of monitoring sites were decided after a number of field trips to visit areas selected on the basis of existing maps of organic soil.Most sites inspected were discarded for reasons such as: the peat layer had disappeared; reluctance of the farmer to give access; or distance incompatible with logistical constraints which made it important to find different land uses near each other.Land use classes AR, PG and RG were identified in regions N and E, whereas only AR and PG were represented in region W .The monitoring sites had the following geographical coordinates (decimal degrees): region W -55.94  The plant cover of the arable sites was dominated by the crop, i.e. spring barley (Hordeum vulgare L.) or potato (Solanum tuberosum L.); during fallow periods some weeds occurred.Grasslands in rotation were dominated by ryegrass (Lolium perenne L.) and white clover (Trifolium repens L.).The latter two species also dominated the permanent grasslands in regions N and E. Region N featured the most diverse permanent grassland with approximately 15 commonly encountered grassland species.In region E, L. perenne was present in most of the area; however, dry parts of E-PG were dominated by Agrostis capillaries L., Poa pratensis L. and Rumex acetosella L.. The relatively small permanent grassland of region W contained a mixture of typical meadow grass species, as well as weeds from the surrounding cropland.Juncus effusus (soft rush) was present at sites N-PG and E-PG.
Information about N inputs in 2006-2008 was obtained from the farmers (Table 2).There were no additional inputs during the monitoring period, but as measurements started in August the management of 2008 had followed the normal practice at all sites, including fertilization and grazing.Management (cuts, harvest and soil cultivation) of the fenced-in monitoring sites followed the practice adopted by the farmers for each field.Cut plant material was collected to determine botanical composition and dry weight.

Experimental design and supporting data
At each monitoring site six 55 cm × 55 cm sampling points for gas flux measurements were organized in three pairs at 5-10 m intervals.The pair-wise distribution was chosen to cover 1 to 10 m-scale variability, and each pair then served as a block in the statistical design.Boardwalks (1 m × 1.5 m) were placed in front of the sampling points during measurements only to minimize disturbances; in permanent grasslands the boardwalks rested on poles installed to >1 m depth.
Piezometers, i.e.PEH tubes (Rotek A/S, Sdr.Felding, Denmark) were installed near each pair of sampling points, i.e., three per site.The depth of the 10-cm screen var-ied between 60 and 130 cm depending on the groundwater level (GWL) in July-August 2008.A separate 10-cm diam.PVC tube with the screen at full length was also installed for continuous recording of GWL with a pressure transducer (H.F. Jensen, Copenhagen, Denmark); data were logged using a Micrologger ver.3.0-3 (Campbell Scientific, UK).All piezometers were surveyed for inter-calibration of all GWL measurements to a common reference point.
Temperatures at 200 cm height and at 5, 10, 30 and 50 cm soil depth were monitored continuously using SKTS thermistors (Skye Instruments, Powys, UK).The average of 2-min readings were logged every hour with a Datahog 2 data logger (Skye Instr.).Soil humidity at 5-10 cm and 15-20 cm depth were monitored in the same way using SKT600 tensiometers.Insolation was determined with a SKP215 PAR quantum sensor (Skye Instr.) using 1-min readings and logging of 30-min averages.Precipitation was monitored at 150 cm height with a rain gauge having an orifice of 200 cm 2 (Rain-o-matic pro; Pronamic, Silkeborg, Denmark).These data were logged with an event logger (Event 101, Madgetech, USA).In periods where climate data were lost due to technical problems, gaps were filled using data from the nearest monitoring site within the same region (distance ∼100 m) or, in a few cases, with relevant grid data obtained from the Danish Meteorological Institute.Missing PAR data were taken from the nearest monitoring station and used directly, whereas soil temperatures were derived from air temperature using site-specific correlations between air and soil temperature.
A mast was installed at each site to support a mobile weather station (Kestrel 4500; Nielsen-Kellerman, Boothwyn, PA, USA), which was installed during measurement campaigns to record wind speed and direction, air temperature (backup), humidity and pressure.
Starting December 2008 concentrations of N 2 O in the upper part of the saturated zone were monitored at each site using an equilibrium method.Gas was sampled from a piece of silicone tubing (length 14 cm, i.d. 2 cm) sealed at both ends with brass caps, which were held apart by three stainless steel rods.The diffusion cell was placed inside a piezometer tube with thick rubber washers to prevent entry of atmospheric air.The upper brass cap was connected via a 1/8" stainless steel tube to a three-way valve in an above-ground sampling unit.
One port was connected to a 50-mL syringe for sampling of gas from the diffusion cell, and the other port was equipped with a hypodermic needle where an evacuated vial (12 mL Exetainer; Labco, High Wycombe, UK) could be mounted for sample collection.A 20-mL gas sample was collected, and then 20 mL helium was injected to re-expand the silicone cell (Jacinthe and Groffman, 2001).Concentrations of N 2 O in the groundwater were calculated as described by Jacinthe and Groffman (2001).
Soil at 0-20 cm depth was sampled for analysis of mineral N on three occasions, in December 2008, April 2009and September 2009.In each case six 20-mm diam.soil cores (0-20 cm depth) were pooled and 10 g fresh wt.soil extracted in 40 mL 1 M KCl.By the end of the monitoring program 50-mm diam.soil cores were taken to the lower boundary of the organic horizon, or a maximum of 132 cm.One core was taken near each pair of sampling points in 34cm subsections; all sub-sections were analyzed for organic dry matter, SOC, total N and pH.

Flux chamber design
The two-part static chambers used in this study were constructed from 4-mm white PVC, largely following the design of Drösler (2005).The supports were 55 cm × 55 cm, 12 cm high and with a sharpened edge at the bottom.A 4-cm wide flange extended outwards 2 cm from the top, giving a maximum insertion depth of 10 cm.The support was fixed to the ground by four 40-cm pegs installed at an angle.The distance to the soil surface inside the supporting frame was determined in a 10 cm × 10 cm grid for correction of total enclosure volume during measurement.
Dimensions of the chamber unit were 60 cm × 60 cm × 41 cm including a closed-cell rubber profile (Emka Type 1011-34; Megatrade, Hvidovre, Denmark) at the bottom.Inter-sections of the same dimensions were used when required due to plant height.Inside the chamber was a 92 mm × 92 mm 12V fan (RS Components, Copenhagen, Denmark) for headspace mixing connected to an outside battery (Yuasa Battery Inc.; Laureldale, PA, USA).A vent tube, designed in accordance with the recommendations of Hutchinson and Mosier (1981), was included with outlet near the ground to minimize effects of wind (Conen and Smith, 1998).A temperature sensor (Conrad Electronic SE; Hirschau, Germany), extending 20 cm below the top, was connected to a digital display (Conrad Electronic SE).A butyl rubber septum was included for gas sampling.Finally, two handles were attached to the top which were also used for straps fixing the chamber firmly against the support.

Sampling protocol
Upon arrival at the field site weather conditions were recorded and the mobile weather station mounted.GWL was then determined at the continuous monitoring station, and in each piezometer at the pair-wise sampling points which were subsequently emptied with a 12V pump and left to refill while gas fluxes were measured.
Gas fluxes were determined using a 60-min enclosure period.Gas samples (20 mL) were taken with a syringe and hypodermic needle immediately after positioning of the chamber and attachment of straps, and then after 15, 30, 45 and 60 min.Gas samples were collected in 12-mL pre-evacuated Exetainer vials that were typically analyzed within 48 h, and always within a week from sampling.A preliminary test had shown that concentrations of CH 4 , CO 2 and N 2 O in these vials were stable during at least two months of storage (data not shown).
Soil temperatures at 5, 10 and 30 cm depth were recorded manually during chamber deployment.These measurements were made with a high precision thermometer (GMH3710, Omega Newport, Deckenpfronn, Germany) between the two chamber units of each block.
Following gas sampling, fresh groundwater was sampled from each piezometer except in a few cases where too little water had accumulated.Approximately 100 mL water was collected, part of which was used to immediately determine groundwater temperature, pH and electrical conductivity using a Cyberscan PC300 (Eutech Instruments; Singapore).The rest of the sample was filtered (0.45 µm nylon membrane SNY 4525, Frisenette, Denmark) directly into 10-mL test tubes that were transported back to the laboratory in a cooler for analysis of SO 2− 4 , Cl − , NH + 4 and NO − 3 .Measurement campaigns were conducted at 3-week intervals between August 2008 and October 2009; in the case of site W-PG frames had to be relocated in October 2008 due to flooding, and therefore the first sampling at this site was on 3 November 2008.Hence, 19 or 17 field campaigns were conducted per site, or in total 51, 49 and 38 campaigns for land use categories AR, PG and RG, respectively.One or two campaigns were included per field trip.Most gas flux measurements were initiated between 9:00 and 12:00, and in a few cases between 12:00 and 13:00.

Soil
Soil dry wt.was determined after drying to constant weight at 100 • C. Soil NH + 4 and NO − 3 in KCl extracts were analyzed by autoanalyzer using standard colorimetric methods (Keeney and Nelson, 1982).Total and organic C and total N were determined on representative subsamples of soil dried at 100 • C according to ISO 10694 and ISO 13878, respectively.

Gas analyses
Nitrous oxide, CH 4 and CO 2 were determined on a dual-inlet Agilent 7890 GC system interfaced with a CTC CombiPal autosampler (Agilent, Naerum, Denmark).The configuration was developed to enable determination of CH 4 , N 2 O and CO 2 with a single 2000-µL injection which is split between www.biogeosciences.net/9/403/2012/Biogeosciences, 9, 403-422, 2012 the two inlets.Two identical instruments were configured, as field sampling and gas analyses were shared between two laboratories.Channel 1 of each system consisted of a 2-m backflushed pre-column with Hayesep P connected to a 2-m main column with Poropak Q via a 6-port valve.The other end of the main column was connected to a four-port valve diverting the gas stream either to an electron capture detector (ECD) for N 2 O analysis, or to a flame ionization detector (FID) for CH 4 analysis.The carrier was N 2 at a flow rate of 45 mL min −1 .For the ECD, Ar-CH 4 (95 %/5 %) at 40 mL min −1 was used as make-up gas.The FID was supplied with 45 mL min −1 H 2 , 450 mL min −1 air and 20 mL min −1 N 2 .Temperatures of injection port, columns, ECD and FID were 80, 80, 325 and 200 • C, respectively.Channel 2 was equipped with a 3-m Poropak Q column and used helium at 42 mL min −1 both as carrier and for the reference cell of the thermocouple detector (TCD).Temperature of inlet, column and TCD were 80, 80 and 250 • C, respectively, and with He at 7 mL min −1 as make-up gas.
The chromatograms were processed using EZ Chrom Elite software.A calibration mixture with approximately 10 µL L −1 CH 4 , 2 µL L −1 N 2 O and 2000 µL L −1 CO 2 (Linde Gas; Copenhagen, Denmark) was used to prepare a five-point dilution series for determination of unknown samples.Standards were included after every ten samples to ascertain the stability of detector responses.
Analyses of N 2 O, CH 4 and CO 2 were carried out in two different laboratories using identical instruments and analytical methods as specified above.Method detection limits were determined to be 0.042, 0.20 and 61 µL L −1 or better for N 2 O, CH 4 and CO 2 , respectively.An inter-calibration between the two instruments used for sample analysis was carried out using samples from two randomly selected sampling days, a batch being analyzed in lab 1 and then in lab 2 and vice versa.The regression slope of 2nd analysis against 1st analysis (n = 42) was 1.01, 1.04 and 1.03 for N 2 O, CH 4 and CO 2 , respectively, when analyzed first in lab 1, and 0.98, 1.01 and 0.96 for N 2 O, CH 4 and CO 2 for the dataset analyzed first in lab 2. Small systematic errors could derive from concentration inaccuracies in the separate calibration mixtures used in the two laboratories.

Flux calculations
Non-linear concentration change over time was frequently observed, and the HMR procedure recently described by Pedersen et al. ( 2010) was therefore adopted.This procedure analyzes non-linear concentration-time series with a regression-based extension of the model described by Hutchinson and Mosier (1981), and linear data by linear regression; statistical information is provided for both categories of flux estimates.A flux model was selected for each individual concentration-time series on the basis of a scatter plot.HMR is available in an add-on package in the free programming software R (http://www.r-project.org).
Annual fluxes of CH 4 , N 2 O and R eco were calculated for the period 21 September 2008 to 20 September 2009 for each site and block; for site W-PG the daily flux prior to the relocation was fixed at the value of 3 November.Daily fluxes were calculated by linear interpolation between sampling days for each of the three blocks (with paired sampling points) of a given site.Annual fluxes were calculated for each block by summation over the period 21 September 2008 to 20 September 2009, the standard errors representing the variation between annual estimates for each block (i.e.within-block variation on individual sampling days was not included).

Effect of sampling time
Flux measurements should ideally be conducted under conditions representing the 24-h period of the sampling day.At 5 cm soil depth the temperature at 10:00 deviated between −2 and +1 • C from daily mean temperature during the period from 21 September 2008 to 20 September 2009 (data not shown), indicating that mid-morning is a suitable time of day for short-term measurements.In practice most measurement campaigns started between 9:00 and 12:00, a few even between 12:00 and 13:00.To evaluate the error associated with the variable time of measurement, annual estimates of N 2 O fluxes and R eco were corrected to daily mean temperature.
Q 10 values for R eco were estimated for individual sites by fitting the van't Hoff equation (Lloyd and Taylor, 1994) to scatter plots of R eco vs. soil temperature at 5 cm depth using SigmaPlot 2000 (SPSS Inc.).This gave site specific apparent Q 10 values ranging from 2 to 3.5.For N 2 O a relationship with soil temperature could not be identified, and a fixed Q 10 of 2 was used (Vicca et al., 2009).R eco and N 2 O flux data were corrected to daily mean temperature using the modified Van't Hoff equation (Davidson et al., 2006): where R is the flux at temperature T (here the daily mean temperature) and R m is the measured flux at T m , the soil temperature at the start of the measurement.Annual fluxes were calculated with corrected data as for measurement data.
No attempt was made to correct CH 4 fluxes, which depend on CH 4 production in saturated parts of the soil profile, and CH 4 oxidation that may also primarily occur near the groundwater table (Hornibrook et al., 2009).At this depth diurnal temperature variations are small, i.e., any error associated with sampling time would be small.

Effects of region and land use
Systematic effects of region (W,N,E), land use (AR, PG, RG), and the interaction between region and land use, on annual fluxes of N 2 O and CH 4 were determined by an approximate test by calculating F-values for each of these effects using the mean square of variations between blocks within each combination as the denominator.All mean squares were calculated using a linear model.To control variance heterogeneity CH 4 fluxes were square-root transformed and N 2 O fluxes ln transformed prior to analysis.The Ftests were only approximate, because the distances between blocks were considered to be too small to represent the true variation within each of the region-land use combinations.

Effects of soil properties and climate
Empirical relationships between soil and environmental conditions and, respectively, fluxes of CH 4 and N 2 O were explored by region using a linear mixed model.Flux rates were transformed as described above prior to testing.Based on a preliminary investigation of collinearity between independent variables the following main effects and interactions were included: Land use, soil temperature at 5 cm depth, GWL, soil moisture at 15-20 cm depth, precipitation during 48 h prior to sampling, peat depth, C:N ratio (0-30 cm), pH (0-30 cm), GW-EC (electrical conductivity), GW-pH, GW-NH + 4 and GW-NO − 3 , as well as the interactions land use x soil temperature, land use x soil moisture, and soil moisture x precipitation.Groundwater SO 2− 4 concentrations were strongly correlated with EC and therefore not included as independent variable.

Emission factors
Emission factors were calculated for rotational crops (AR and RG) and permanent grassland (PG), respectively.Mean fluxes and 95 % confidence limits of CH 4 and N 2 O fluxes were calculated by back-transformation of square-root transformed and ln transformed data, respectively.

Site characteristics
Most peat soils in Denmark are shallow following decades of drainage and cultivation.Peat depth at the eight monitoring sites included in this study was in most cases 50 to 100 cm except for site W-AR, where the peat at one end of the plot extended to 310 cm depth (Fig. 1a).Some characteristics of the peat materials are shown in Table 3.The C:N ratios ranged from 17 to 52 and were consistently lower in region W than in the other two regions.All soils were acidic, with pH values in the range 3.5 to 5.8.Minimum pH was generally observed at 34-64 cm depth in region W and N.At 0-30 cm depth there was a strong positive relationship between soil organic carbon concentration of the soil dry matter and C:N ratio (P < 0.001) (Fig. 1b).Soil mineral N data are shown in Table 4. Concentrations of NH + 4 and NO − 3 were generally higher in the soils of region N than in region W or E. The most notable observation was an accumulation of NO − 3 in the soil of site N-AR, i.e., the potato field, to 100 and 166 mg N kg −1 dry wt.soil in April and September, respectively.

Climatic conditions
The annual mean temperatures recorded in region W , N and E were 9.5, 8.8 and 9.1 • C, respectively.Winter was colder and the early spring warmer than average, see on-line supplementary information.Annual precipitation in region W , N and E was 913, 702 and 579 mm, which was higher than average for region W , and slightly lower than average for region E. With respect to monthly means there were for the monitoring period only minor deviations from 20-yr average www.biogeosciences.net/9/403/2012/Biogeosciences, 9, 403-422, 2012  Figure 2 shows air temperature and soil temperatures at 5 cm depth at the time of measurement campaigns, the data shown were extracted from the database with continuous data.No sub-zero soil temperatures were observed below 5 cm depth.

Groundwater
Figure 3 presents mean GWL of the piezometers.Regions W and N were characterized by significant seasonal fluctuations in GWL, where the unsaturated zone varied from −10 to −50 cm during winter and spring and dropped to between −80 and −100 cm during summer and early autumn.In contrast region E showed a high degree of stability in GWL, but GWL at the arable site, E-AR, was much lower than at sites E-PG and E-RG.At sites N-PG and E-PG the gradients in GWL were due to the topography at these sites.

Concentrations of SO 2−
4 are presented in Fig. 4 (not all samples were analyzed due to resource limitations).The SO 2− 4 level varied little across the year, but significantly between sites.In region N the land uses AR and RG, located side-by-side but with different crops, had very similar levels of SO 2− 4 at the first sampling, 120 and 130 mg L −1 in N-RG and N-AR, respectively, but diverged steadily with concentrations decreasing to 92 mg L −1 at site N-RG, but increasing to 155 mg L −1 at site N-AR (potato).Highest overall groundwater SO 2− 4 concentrations were observed at site W-AR, in particular where the deepest peat was recorded (Fig. 1a).During autumn the concentrations were nearly twice as high, up to 600 mg SO 2− 4 L −1 , at block 3 as at the other two blocks.Groundwater NO − 3 concentrations were low, <3 mg L −1 (Fig. 5).At site E-PG NO − 3 occurred only in block 1 where N 2 O emission was consistently recorded (cf.Sect.3.4.1).In autumn 2009 NO − 3 was only recorded at two sites used for arable crops (W-AR and N-AR).Here the accumulation of NO − 3 coincided with a drop in GWL to 100 cm depth.The dynamics of groundwater NH + 4 were modest, but characterized by large differences between sites.The concentration of groundwater NH + 4 was always low with land uses AR and RG except in region W where 10 to 20 mg NH + 4 -N L −1 was found during autumn 2008, with the highest concentrations at block 3 with the deep peat.
Groundwater pH in regions N and E were largely constant, but varied among land use categories, i.e., between pH 5 and 6 in region N and between 5.8 and 7 in region E (data not shown).At site W-AR groundwater pH remained at pH 5 to 6 during autumn 2008 and the following winter, but then dropped around 1 pH unit following spring during a period with an 80-cm lowering of the groundwater table.At the last site, W-PG, groundwater pH were low, between 3 and 6, but highly variable between samplings and blocks (data not shown).Groundwater EC levels varied between 400 and 700 µS cm −1 except at site W-AR, where EC averaged 1000 µS cm −1 during autumn 2008 and then declined to a constant level of 800 µS cm −1 .Chloride concentrations varied little within site, with concentrations in the range 20 to 40 mg L −1 except at site N-AR, where it was 50 to 60 mg L −1 (data not shown).
Figure 6 shows dissolved N 2 O in the saturated zone as calculated from N 2 O concentrations in a silicone diffusion cell.Jacinthe and Groffman (2001) reported that at 11 • C the concentration at equilibrium with atmospheric air is 0.44 µg N 2 O L −1 , and hence dissolved N 2 O was mostly below or slightly above atmospheric concentrations, but with a notable exception.At site N-AR, the potato field, extremely high levels of N 2 O were consistently recorded, the actual values reported being highly uncertain because they were outside the range of the calibration curve.In contrast, the groundwater N 2 O concentration at site N-RG, located at <20 m distance from site N-AR, never exceeded 2 µg N 2 O L −1 (Fig. 6).

Seasonal dynamics and annual fluxes of N 2 O, R eco and CH 4
The following sections will present seasonal dynamics of N 2 O, R eco and CH 4 , as well as annual fluxes representing the period 21 September 2008 to 20 September 2009.Fluxes are by definition positive when referring to emissions, and negative when referring to net uptake by the soil.

Nitrous oxide
Fluxes of N 2 O from permanent grasslands (PG) were always low, 0 to 500 µg N 2 O m −2 h −1 (Fig. 7).At site E-PG N 2 O emissions only occurred from block 1 located on a ridge where GWL was at −60 to −80 cm, as opposed to −20 to −40 cm at blocks 2 and 3. Also, peat depth was greater at block 1 (Fig. 1a).In region E low N 2 O fluxes were also observed with land use AR and RG.At site W-AR some cases with headspace concentrations of N 2 O in excess of 10 µL L −1 were observed at the two first samplings in autumn 2008 which were outside the range that could be quantified.Attempts to reanalyze after dilution failed, and a total of six individual observations were excluded from further analysis for this reason.A similar pattern was not seen in early autumn 2009.Subsequent peaks in N 2 O flux at site W-AR were observed in late winter, after spring cultivation, and following harvest in late July 2009.Elevated rates of N 2 O emission, though less dramatic, were also recorded in late summer 2008 at site N-AR, followed by emission maxima during winter, spring and early autumn 2009 that coincided with accumulation of NO − 3 in the soil (Table 4).Site N-RG showed a consistent increase in N 2 O emissions during early spring despite the absence of any fertilization or cultivation.The GWL declined from −30 to −100 cm during this period, and presumably this N 2 O was derived from a pool of mineralizable organic N released upon drainage of the top soil.
Annual fluxes of N 2 O are shown in Table 5.The highest rates, recorded at the arable sites W-AR and N-AR, corresponded to 38 and 61 g N 2 O-N kg ha −1 ; some extreme rates at site W-AR during autumn were probably missed as explained above.The remaining sites ranged from 0.5 to 3.7 g N 2 O m −2 , corresponding to annual losses of 3 to 24 kg N 2 O-N ha −1 .

Ecosystem respiration
Ecosystem respiration (R eco ) showed a similar pattern at all monitoring sites, following seasonal trends in temperature (Fig. 8).There was, however, also a reciprocal relationship (P<0.05) with GWL (Fig. 3) at all but one site (E-PG), and thus CO 2 fluxes could also be partly controlled by drainage.
Annual values of R eco ranged from 5.3 kg CO 2 m −2 (1.5 kg C m −2 ) at site N-AR (potato) to a maximum of 12.2 kg CO 2 m −2 (3.3 kg C m −2 ) at site N-RG, where R eco was significant also during winter (Table 5).Hence, the lowest and the highest R eco values occurred from neighbouring fields, the main difference being that the soil of land use AR had no vegetation during an extended period from October 2008 until May 2009, whereas RG was covered by a grass sward throughout the monitoring period.

Methane
Except for sampling points which included J. effusus, fluxes of CH 4 were small, mostly fluctuating within the range −100 to 200 µg CH 4 m −2 h −1 (Fig. 9).Fluxes of CH 4 from site N-PG and E-PG are presented with a separate curve for two (site N-PG) or one sampling point (site E-PG) with tussocks of J. effusus, which all showed consistent net emissions of CH 4 throughout the year.The dry matter of J. effusus in cuts taken during the monitoring period in the two blocks with J. effusus was positively related to the annual mean flux of CH 4 from these sampling points (Fig. 10).

Effect of sampling time
The representativeness of fluxes recorded between midmorning and noon was evaluated for R eco and N 2 O fluxes by temperature adjustment using site-specific Q 10 (R eco ) or a fixed Q 10 (N 2 O).For R eco the deviations from daily mean temperature in the actual sampling program would give relative errors in estimated annual fluxes ranging from 1 to 25 % (Fig. 11).The range and magnitude of relative errors estimated for N 2 O fluxes were smaller, i.e., 2 to 7 %.

Effects of geo-region and land use
In the test of system effects (region, land use) on CH 4 fluxes the two blocks with J. effusus were omitted, as it was not possible to avoid extreme variance heterogeneity despite data transformation.The variation between regions was much higher than within-site variability, whereas there was no effect of land use, and no interaction between region and land use (Table 6).Hence, CH 4 fluxes differed between landscape types (geo-regions) but not land use, but as mentioned above fluxes were generally low.Hence, in permanent grasslands emission hotspots such as those associated with J. effusus in the present study may dominate the CH 4 budget.The variation in N 2 O flux between regions was not significantly higher than the variation between blocks within a site, but there was a significant effect of land use, and a strong tendency (P = 0.0653) that the effect of land use differed between regions (Table 6).These differences were caused mainly by emissions from arable sites (see Discussion).

Effects of soil and climatic conditions
Empirical relationships between, respectively, N 2 O and CH 4 fluxes and several independent variables were examined by region using a linear mixed model.Significant effects were identified in all cases (Table 7), but no consistent patterns that would indicate general controls of N 2 O or CH 4 fluxes were found.Nitrous oxide fluxes were related to mineral N in the groundwater in region E only.In Region E there were significant effects of top soil C:N ratio and pH on CH 4 fluxes.Gradients in soil and groundwater properties at the block level probably contributed to these effects (data not shown).Temperature and water (GWL, soil moisture or precipitation) was involved in most significant effects, either as main effect or interaction.In view of the importance of temperature and O 2 availability for biological processes, the effects of seasonal changes in climate and GWL are not surprising.

Emission factors
Emission factors with 95 % confidence intervals are presented in Table 8.Emission factors were calculated for soils in rotation (land use categories AR and RG) and permanent grasslands (land use category PG), respectively.Land use categories AR and RG were combined for derivation of  emission factors because rotational grass will alternate with other crops of the rotation at regular intervals.Although the system analysis showed an effect of region, rather than ** a T −5 cm -Soil temperature at 5 cm depth; GWL -groundwater level; GW X -groundwater concentration of nitrate or ammonium, or groundwater electrical conductivity, or pH; Soil H 2 O -soil water tension at 5-10 cm depth; Prec 48 h -precipitation during 48 h prior to Sampling; Soil C:N -soil C:N ratio at 0-30 cm depth; Soil pH -soil pH at 0-30 cm depth.* , P < 0.05; * * , P < 0.01; * * * , P < 0.001.land use, on CH 4 fluxes when excluding experimental blocks with J. effusus (Table 6), the overall flux was dominated by these emission hotspots in permanent grasslands.The CH 4 emission factor for permanent grasslands (0.47 g m −2 ) was significantly (P < 0.001) higher than from crops in rotation (0.011 g m −2 ), although the confidence limits of both land use categories included net uptake.The N 2 O emission factors for permanent grasslands and crops in rotation were 0.5 and 2.5 g m −2 , respectively, but the difference was not statistically significant (0.05<P<0.10).

Gas sampling and analysis
The monitoring study was planned to cover major landscape and land use categories.Spatial variability was addressed by the use of a stratified sampling scheme with selection of representative landscape types and land use categories, in accordance with the recommendations of the Intergovernmental Panel on Climate Change (IPCC, 2006).The distribution of sampling points at each site was also stratified to account for 1-to 10-m scale variability, and the experimental blocks were the basic unit used for calculation of annual fluxes.
A sampling strategy based on manually operated chambers was selected.While enclosure-based methods rely on proper replication to cover spatial variability, they also allow fluxes to be linked to soil conditions at the individual sampling points.Between 38 and 51 measurement campaigns were conducted for a given land use category (across all regions), and with a 3-week sampling frequency on average at the individual site.Hence, sampling intensity at the individual site was low, but the ability of a given sampling strategy to precisely estimate cumulative fluxes also depends on the temporal variability (Parkin, 2008).In the present study fluxes were driven by soil organic matter turnover and therefore mainly depended on climatic conditions and soil properties, i.e., there were no short-term dynamics due to fertilizer inputs or excretal returns.
Chamber measurements were initiated between 9:00 and 13:00 in this study.In an attempt to evaluate the representativeness of sampling time we estimated the effect of deviations from daily mean temperature for N 2 O emissions and R eco .Fluxes of N 2 O were less sensitive to variations in sampling time than R eco (Fig. 11), and deviations could not be explained by the use of a variable Q 10 for R eco (data not shown).Diurnal variation in soil temperature is dampened with depth (Fang and Moncrieff, 1998;Bahn et al., 2008), and the difference between R eco and N 2 O could be due to a different vertical distribution of CO 2 producing and N 2 O producing processes.Soil temperature at 5 cm depth was used here, but a representative temperature does not exist for either process (Vicca et al., 2009).Also, the apparent Q 10 values of up to 3.5 observed probably included indirect effects of temperature, for example on substrate availability (Davidson et al., 2006).
Ecosystem respiration and fluxes of CH 4 and N 2 O were determined using a chamber deployment time of 60 min.Trace gas accumulation was often not linear, which would lead to under-estimation of fluxes if linear regression is used (Davidson et al., 2002).Shortening the deployment time can improve the accuracy of flux estimates, but precision may be lost (Venterea et al., 2009).In the present study a non-linear method, HMR (Pedersen et al., 2010), was used for flux estimation in 17, 41 and 73 % of all cases with CH 4 , N 2 O and R eco , respectively.HMR is an exponential model with certain restrictions on parameter estimates.The theoretical analysis of Pedersen et al. (2010), as well as subsequent experimental evidence comparing 1-and 2-h deployment (unpublished results), has indicated that the pre-deployment fluxes estimated by HMR are robust against declining headspace accumulation during chamber deployment.

Nitrous oxide fluxes
This study was planned to quantify fluxes associated with decomposition of soil organic matter in representative His-tosols used by agriculture.According to the IPCC methodology, estimates of N 2 O emissions from managed organic soils should distinguish between emissions derived from external N inputs and this background emission (de Klein et al., 2006).The partitioning of N 2 O emissions is hampered by a complex microbial response to N amendments, and by a significant effect of groundwater table on background emissions (Clough et al., 1996;Augustin et al., 1998;van Beek et al., 2010van Beek et al., , 2011)).In the present study monitoring started in autumn 2008, following a growth season in accordance with current practices for each respective land use, including fertilization and grazing, and continued throughout the following year.Hence, long-term, but not short-term (<3 months) effects, of N amendments were accounted for in the emissions recorded.The system analysis (Table 6) suggested a significant effect of land use on N 2 O fluxes, and a nearly significant region x land use interaction (P < 0.065).This was mainly due to high annual N 2 O emissions at the two arable sites W-AR and N-AR, and low emissions from the arable soil in region E (Fig. 7).
The potential for N 2 O emission is influenced by a number of factors such as electron donor availability, mineral N concentrations, oxygen status and soil pH (Groffman et al., 2000;Flessa et al., 1998;Baggs and Philippot, 2011).Nitrate accumulated during 2009 in the top soil at site N-AR (Table 4), where the highest N 2 O emissions were generally observed (Fig. 7).Extremely high N 2 O concentrations were also dissolved in the groundwater at site N-AR (Fig. 6), but the importance of subsoil N 2 O production for atmospheric emissons remains to be shown.At site W-AR the highest emissions of N 2 O were observed during autumn 2008, where unfortunately no information is available about top soil N status.Groundwater NO − 3 concentrations were <5 mg N L −1 , whereas GW NH + 4 concentrations were consistently at 10-20 mg N L −1 (Fig. 5).At site W-AR the soil pH at 34-64 and 68-98 cm depth was very low at <4 (Table 3) and potentially inhibitory to nitrifiers.Regina et al. (1996) found potential nitrification (accumulation of NO − 2 + NO − 3 ) in drained minerotrophic peat soils at pH 4 despite low counts of ammonia oxidizing bacteria and speculated that heterotrophic organisms could be involved.In our study a supply of NH + 4 in early autumn, derived from mineralization activity or the rising groundwater, may have triggered N 2 O production via nitrification and/or denitrification (Regina et al., 1996;Martikainen and de Boer, 1993).Goldberg et al. (2010) reported that high N 2 O emissions during 15 d following an increase in the water table of a minerotrophic fen accounted for 20-40 % of total emissions during a >500 d monitoring period.Such a pattern would be in accordance with the high N 2 O fluxes observed at sites W-AR and N-AR where GWL increased during August and September 2008 (Fig. 7).A similar response was not seen at sites W-PG, N-RG or N-PG despite similar increases in GWL, nor at site E-AR despite a high potential for net N mineralization in the cultivated soil.The high N 2 O emissions observed in arable soil may thus have been caused by the interaction between rising groundwater level and a high potential for N mineralization following tillage and fertilization the previous spring (Table 2).In accordance with this, van Beek et al. ( 2004) concluded that a substantial part of an estimated N 2 O production of 126 kg N ha −1 yr −1 from an intensively managed grassland on peat was produced at depth in the soil, and that N 2 O emissions were regulated by NO − 3 availability and groundwater level.
A second difference between sites W-AR and N-AR relative to E-AR was the relatively high concentrations of groundwater SO 2− 4 .Microbial sulfate reduction may lead to formation of pyrite (FeS 2 ), although this will depend on pH as FeS 2 is unstable at pH<4 (Shotyk, 1988).In regions W and N the groundwater table dropped by up to 100 cm during spring and summer (cf.Fig. 3), allowing for acidifying processes such as nitrification and possibly Fe(II) oxidation in the drained peat (Lüdecke et al., 2010;McLaughlin and Webster, 2010).Following a rise in water table dissolution of FeS 2 in the acid soil could stimulate N 2 O emissions as Fe(II) may serve as electron donor in denitrification if NO − 3 is present (Madsen and Jensen, 1988;Jørgensen et al., 2009).Also, the relative importance of N 2 O as product of denitrification is higher at low pH (Cuhel et al., 2010).Groundwater SO 2− 4 was high at the sites W-AR, N-AR and N-RG with significant N 2 O emissions, but also at site W-PG with no significant N 2 O emission (Fig. 4).However, in contrast to site W-PG the sites in rotation had a prehistory of N fertilization (cf.Table 2) and, consequently, a high potential for net N mineralization and nitrification activity.

Methane fluxes
Currently CH 4 fluxes are considered to be insignificant for the GHG balance of cultivated organic soils in Denmark (Gyldenkaerne et al., 2005), which is supported by the present study.Four of the eight monitoring sites were neutral with respect to CH 4 fluxes, one site (W-PG) was a small sink at −0.16 k CH 4 m −2 yr −1 , and the two other permanent grasslands were consistent sources at 2.8 and 4.7 k CH 4 m −2 yr −1 , though only as a result of high emissions from a few samplings points; within these the mean flux of CH 4 was positively related to the biomass of soft rush (J.effusus) in plant cuts (Fig. 10).Drösler (2005), using a non-destructive approach, also found a linear relationship between aerenchymous leaf tissue and CH 4 emissions.Soft rush often invades pastures on poorly drained soil (Agnew, 1961), and its capacity to mediate transport of CH 4 from saturated soil to the atmosphere is well known (Garnet et al., 2005).Surprisingly, however, CH 4 fluxes at these emission hotspots were largely constant around the year despite fluctuating soil temperature (Fig. 2) and GWL (Fig. 3, region N only).Apparently CH 4 transport to the atmosphere was severely delayed except around soft rush tussocks at these two grazed pastures; soil porosity could be higher around tussocks due to reduced trampling, but CH 4 emissions at site N-PG proceeded during a period with standing water at these sampling points, which points to J. effusus as the main route of CH 4 transport.Kechavarzia et al. (2010) characterized physical and hydraulic properties of drained fen peat soils used by agriculture, and they found that both horizontal and vertical hydraulic conductivity was strongly reduced in the amorphous peat at 0-15 cm depth, which also had a lower porosity and drained more slowly than less decomposed peat.The loss of peat structure is a possible cause of preferential gas exchange At the permanent grassland W-PG there were no indications of net CH 4 emission at any time around the year.Watson and Nedwell (1998) found that CH 4 production was inversely related to SO 2− 4 availability, probably as a result of competitive inhibition by sulfate reducing bacteria.Groundwater SO 2− 4 (Fig. 4) was high at this site in accordance with this explanation.On the other hand, occasional emissions of CH 4 were observed at sites W-AR where groundwater SO 2− 4 was even higher.

GHG balance of the eight monitoring sites
The two permanent grasslands with soft rush were minor sources of CH 4 , with average net emissions equivalent to 0.7 and 1.2 t CO 2 eq ha −1 yr −1 , respectively.Nitrous oxide fluxes were equivalent to 1-11 t CO 2 eq ha −1 yr −1 at six of the sites, while at the arable sites in region W and N they were 18 and 29 t CO 2 eq ha −1 yr −1 ; the particular soil conditions at these two sites were discussed above.R eco constituted 53-122 t CO 2 eq ha −1 yr −1 in this study, of which plant respiration may account for 35-45 % during periods of active growth (Silvola et al., 1996).A parallel study on net ecosystem exchange of CO 2 determined net fluxes of CO 2 at the eight monitoring sites to range from 13 to 50 t CO 2 eq ha −1 yr −1 (L.Elsgaard, personal communication, 2011).Hence, CO 2 emissions dominated the GHG balance of these Danish organic soils except at the two arable sites with high N 2 O fluxes, i.e. site W-AR and N-AR.Kasimir-Klemedtsson et al. (1997) presented GHG balances for selected managed organic soils in Northern Europe.They reported total emissions from drained organic soils under different management to be 11-70 t CO 2 eq ha −1 yr −1 , the C loss always dominating the GHG balance.A similar conclusion was reached by Jungkunst and Fielder (2007) in a compilation of data from peatlands across three climate zones.
Only one full year was represented in the monitoring program, and clearly more robust emission factors would be achieved with a longer monitoring period to better represent the full range of climatic conditions encountered at the sites (Table 1).High rainfall in August 2008, shortly before the monitoring was initiated, and in region W again in October 2008, as well as higher than average (2 • C) temperatures in April and slightly higher precipitation in early summer may have influenced fluxes, but overall the deviations between actual climatic conditions and 20-yr average values for temperature and precipitation were not dramatic.
In the absence of an extended national database it may be useful to consider emission factors for organic soils in other Nordic countries.Maljanen et al. (2010), in a review of GHG balances for organic soils in the Nordic countries, reported emissions from perennial grasslands to be 0.32 ± 0.64 g CH 4 m −2 (n = 11), which is compara-ble to the 0.47 g CH 4 m −2 of this study.Similarly, for N 2 O the average emission observed at grassland sites on peat in previous studies in Nordic countries was 1.50 ± 1.60 g N 2 O m −2 (n = 12), a range that includes the average emission of 0.5 g N 2 O m −2 of the present study (Table 8).For barley an average flux of CH 4 of −0.03 ± 0.18 g m −2 (n = 5) was reported (Maljanen et al., 2010), the range and lower level relative to permanent grasslands was confirmed by the present study.Finally, N 2 O emission factors for barley grown on peat soil were reported to average 1.7 ± 0.5 g m −2 (n = 5), with one study of a potato field in the same range; these values are also comparable to the 2.5 g N 2 O m −2 determined in the present study.In summary, the new emission factors for the Danish organic soils investigated are consistent with existing data from a much wider range of organic soils in the Nordic countries.
The Danish GHG inventory for organic soils until recently used estimates of soil organic C degradation as a proxy for net N mineralization and N 2 O emission by assuming a fixed relationship between soil organic C and C:N ratio (Gyldenkaerne et al., 2005).The previously derived relationship between SOC and C:N ratio agreed closely with the relationship in Fig. 1b, but there was no simple relationship between N 2 O emissions and C:N ratio (Table 7), which questions the approach.For N 2 O the Intergovernmental Panel on Climate Change has proposed a default emission factor of 8 (range 2-24) kg N 2 O-N ha −1 for managed organic soils under temperate climate conditions (IPCC, 2006).Annual fluxes of N 2 O in the present study varied greatly between sites.Six sites were within the 2-24 kg N 2 O-N ha −1 range (Table 5).The two remaining sites, W-AR and N-AR, showed annual fluxes corresponding to 38 and 61 kg N 2 O-N ha −1 , values matching or exceeding most previously reported fluxes from arable peat soils in Europe (Flessa et al., 1998;Drösler et al., 2008;Maljanen et al., 2010).These two sites received >100 kg fertilizer N during 2008 (Table 2), which was probably excessive considering the N mineralization potential of organic soils.However, manure N had also been applied prior to the monitoring study at sites E-AR and E-RG where fluxes were only 5-10 kg N 2 O-N ha −1 .Hence, site-specific conditions with respect to soil properties, groundwater and management can influence the level of N 2 O emissions, and there is a need to consider a differentiation of N 2 O emission factors for managed organic soils.

Conclusions
This study is part of the first assessment of GHG balances for Danish organic soils drained for agriculture.Annual fluxes of CH 4 and N 2 O were consistent with previous studies of cultivated organic soils in the Nordic countries, but more importantly the stratified experimental design with identical procedures and supporting data acquired at all eight sites helped isolate effects of soil conditions and land use.Contrasting seasonal dynamics of N 2 O emissions at the three arable sites suggested that extremely high emissions were a result of interactions between fluctuating water table, net N mineralization and low pH, possibly involving sulfur transformations; this has implications for acid sulfate soils in agricultural use and should be further investigated.As expected CH 4 fluxes from the drained organic soils were low, but in two pastures on degraded peat aerenchymous plants were a consistent source of CH 4 .There appears to be scope for a future disaggregation of the N 2 O emission factor for organic soils, and possibly also for taking CH 4 emissions into account in the GHG balance of grazed pastures on peat with a shallow groundwater table.

Fig. 2 .
Fig. 2. Seasonal variations in soil temperature at 5 cm depth ( ) and air temperature at 200 cm height ( ) at the eight monitoring sites representing three regions (W,N and E) and three land use categories (AR -arable crop; RG -rotational grass; and PG -permanent grass).The data shown were extracted from a database with continuous data to indicate conditions on sampling days.In region E some data were lost due to technical problems.

Fig. 3 .
Fig. 3. Seasonal variations in groundwater level (GWL) at the eight monitoring sites; for information about site ID, see legend to Fig. 2. The data (mean ± standard error) represent GWL in piezometers at the paired sampling points during gas flux measurement campaigns (n = 3).

Fig. 4 .
Fig. 4. Seasonal variations in groundwater sulfate (GW-SO 2− 4 ) at the eight monitoring sites; for information about site ID, see legend to Fig. 2. Fresh groundwater was sampled from piezometers at the paired sampling points during gas flux measurement campaigns (the data shown represent mean ± standard error, n = 3).The data set was incomplete due to resource limitations.

Fig. 5 .
Fig. 5. Seasonal variations in groundwater ammonium (GW-NH + 4 ) and nitrate (GW-NO − 3 ) at the eight monitoring sites; for information about site ID, see legend to Fig. 2. Fresh groundwater was sampled from piezometers at the paired sampling points during gas flux measurement campaigns (the data shown represent mean ± standard error, n = 3).

Fig. 6 .
Fig. 6.Concentrations of N 2 O dissolved in the groundwater were determined in a central location at each of the eight monitoring sites using a silicone diffusion cell (see text) starting December 2008.Gas samples at equilibrium with the surrounding water were collected during gas flux measurement campaigns.

Fig. 7 .
Fig. 7. Seasonal variations in N 2 O fluxes at the eight monitoring sites; for information about site ID, see legend to Fig. 2. One data set from April 2009 at site N-AR was omitted due to unresolved analytical problems.The data represent mean ± standard error (n = 6).

Fig. 8 .
Fig. 8. Seasonal variations in ecosystem respiration (R eco ) at the eight monitoring sites; for information about site ID, see legend to Fig. 2. One data set from April 2009 at site N-AR was omitted due to unresolved analytical problems.The data represent mean ± standard error (n = 6).

Fig. 9 .
Fig. 9. Seasonal variations in CH 4 fluxes at the eight monitoring sites; for information about site ID, see legend to Fig. 2. One data set from April 2009 at site N-AR was omitted due to unresolved analytical problems.At sites N-PG and E-PG there were a few sampling points, with tussocks of J. effusus (soft rush) that showed consistent emissions of CH 4 .Circles represent mean ± standard error (n = 6 at sites without soft rush).Sampling points with soft rush at site N-PG (n = 2) and E-PG (n = 1) are shown separately (triangles).

Fig. 10 .
Fig. 10.Relationship between dry wt. of J. effusus stems in the biomass of plant cuts and the mean flux of CH 4 .J. effusus occurred in four individual sampling points, although in one of them only with a few scattered stems.

Fig. 11 .
Fig. 11.The scatter plots show the effect of correcting ecosystem respiration (R eco ) and N 2 O fluxes at the actual measurement time to a temperature corresponding to the daily mean of each sampling day.The correction was made with a modified Van't Hoff equation and Q 10 values that were site-specific (R eco ) or fixed (N 2 O) (see text).The data represent mean ± standard error (n = 3).
• N, 8.45 • E; region N -57.23 • N, 9.84 • E; region E -56.38 • N, 10.40 • E. A map of Denmark indicating the location of sites is included as on-line supplementary information.The eight sites will be referred to by the unique combination of region and land use, e.g., W-AR.

Table 1 .
Average annual mean temperature, T ann (08/09), and precipitation, P ann (08/09), were calculated for each region for the period 21 September 2008 to 20 September 2009, the period that was used for estimating annual fluxes of CH 4 and N 2 O.The table also shows 20-yr means and range of annual temperature and precipitation in each region.
a Data from national grid of climate stations of the Danish Meteorological Institute.

Table 2 .
Land use and N inputs via mineral fertilizers and manure during the period 2006-2009 at the eight monitoring sites.AR -arable crop; PG -permanent grassland; RG -rotational grass.

Table 3 .
Soil organic carbon, C:N ratio and pH of the peat in the eight organic soils used in the monitoring study.Rotational grass (RG) was not represented in region W. AR -arable crop; PG -permanent grassland; RG -rotational grass.
NA -not applicable.a AR was spring barley in Regions W and E, potato in Region N .b No peat at this depth.

Table 4 .
Soil mineral N (0-20 cm depth) was determined in the experimental areas on three occasions during the study.The data shown represent mean ± standard error (n = 6).AR -arable crop; PG -permanent grassland; RG -rotational grass.

Table 5 .
Annual fluxes (g m −2 ) of CH 4 , N 2 O and ecosystem respiration (R eco ) for the period 21 September 2008 to 20 September 2009.Fluxes of CH 4 and N 2 O are also expressed as CO 2 equivalents (t ha −1 ).The data represent mean ± standard error (n = 6).AR -arable crop; PG -permanent grassland; RG -rotational grass.

Table 6 .
Effects of region and land use were evaluated with an approximative linear mixed model (see text).Methane fluxes were square-root transformed, and N 2 O fluxes ln transformed, prior to testing.
a Blocks with Juncus were not included.

Table 7 .
Relationships between soil and environmental conditions and, respectively, CH 4 and N 2 O fluxes during the monitoring period were investigated with a linear mixed model.Methane fluxes were square-root transformed, and N 2 O fluxes ln transformed, prior to testing.See text for an explanation of the independent variables and interactions included in the model.

Table 8 .
Annual emission factors (g m −2 ) for CH 4 and N 2 O were calculated for soils in rotation (land use categories AR and RG in this study) and permanent grassland (PG), respectively, using experimental block (paired sampling points) as the basic unit.

www.biogeosciences.net/9/403/2012/ Biogeosciences, 9, 403-422, 2012 S. O. Petersen et al.: Annual emissions of CH 4 and N 2 O via
aerenchymous plants, and it suggests that the role of soft rush in mediating CH 4 emissions from low-lying pastures on peat should be further investigated.