Introduction

Natural peatlands, covered by Histosols (FAO/ISRIC/ISSS 1998), are important sinks for carbon (C) and nitrogen (N) due to prevailing permanent anaerobic conditions and resultant low rates of decomposition (Moore et al. 2004). They represent only about 3% (3.88–4.38 * 106 km2) of the world’s area but comprise up to 30% of their C and N reserves (Martikainen et al. 1993). Thus, peatlands play an important role in the global C and N cycle and they may also contribute significantly to the production and consumption of the greenhouse gases nitrous oxide (N2O; Martikainen et al. 1993; Regina et al. 1996; Regina et al. 2004) and methane (CH4; Augustin et al. 1996; Khalil 1999) but have not been considered in respective global estimates (Brumme et al. 2005) due to large uncertainties resulting from the diversity of peatlands.

The global warming potentials (GWP) of nitrous oxide (N2O) and methane (CH4) are 310 and 21 times, respectively, than that of CO2 during a 100 year time scale (IPCC 2007). N2O is mainly produced in soils due to autotrophic nitrification and heterotrophic denitrification (Davidson 1991). Virgin peatlands with natural high water table levels show low N2O fluxes or consumption of N2O due to limitation of oxygen and hence absence of nitrification resulting in low NO3 − concentrations (Martikainen et al. 1993; Augustin et al. 1996; Regina et al. 1996; Aerts 1997; Brumme et al. 1999).

The greenhouse gas CH4 is produced during the anaerobic decomposition processes in anoxic parts of peats by methanogenic bacteria whereas it is oxidized in the presence of free oxygen by methanotrophic bacteria (Joabsson et al. 1999). Pristine peatlands are a significant source of CH4 to the atmosphere. Three emission pathways have been reported for CH4: molecular diffusion of dissolved CH4, ebullition (bubbling) of CH4 and transport of CH4 through plant roots and shoots (Tokida et al. 2005).

Previous studies have pointed out that both groundwater table levels and temperature mostly affected the fluxes of trace gases in peatlands (Augustin et al. 1996; Koponen and Martikainen 2004). The few studies of greenhouse gas emissions from virgin peatlands have been conducted in classical bogs and fens without permanent water inflow (e. g. Martikainen et al. 1993; Regina et al. 1996; Heikkinen et al. 2002). Reports of N2O and CH4 fluxes from slope mires, a peatland type widespread in mountain forests are lacking. Globally slope mires represent 1% of the world’s peatlands comprising an area of about 4 million hectare (Joosten, 2007, personal communication). Solely the different movement of soil water in slope mires should change controlling factors of gas formation and resulting greenhouse gas fluxes. We hypothesize that slope mires release more N2O because of increased influx of O2 which inhibits N2O reductase and concurrently restricts N2 formation. The influx of O2 might adversely affect CH4 formation and might favour CH4 oxidation resulting in lower CH4 emission from slope mires than that of pristine bogs and fens.

To test these hypotheses we conducted a field experiment in a pristine slope mire at the Harz Mountains with the following objectives:

  1. 1.

    To quantify and to compare fluxes of N2O and CH4 in mire zones at a central non-forested and a peripheral forested area representing different nutrient status, water table levels and oxygen input caused by slope water inflow

  2. 2.

    To monitor N contents and other relevant features of mire pore water in order to identify relationships to the production and consumption of N2O and CH4

  3. 3.

    To find out seasonal dynamics of N2O and CH4 fluxes

Materials and methods

Study site

The study was conducted in the Harz mountains, central Germany (federal state Saxony-Anhalt), close to the highest peak of the Harz mountain (Mt Brocken, 1,142 m a.s.l., 51°48′N, 10°37′E) at an elevation of 843–860 m a.s.l. Springs of the Ilse stream characterize features of the studied area (Böhlmann et al. 2005). Average annual precipitation (1951–1980) amounts to 1,609 mm and mean annual temperature is +2.8°C (Wegener and Kison 2002). The annual bulk N deposition at the mire was about 51 kg ha−1 year−1 in 2003 of which 25 kg ha−1 year−1 was deposited as nitrate (Böhlmann 2004).

The investigated mire comprising Fibric Histosols covers a slope of 0.015 km2 which is part of a catchment of 0.13 km2. The mire was formed about 1,000 years ago (14C analysis, Böhlmann 2004) on granite debris of the Mt Brocken. The thickness of the peat layer is up to 1.4 m. The largest part of the slope mire is a swamp overgrown with spruce plants (0.01 km2) and a Calamagrostio villosae-Piceetum-community. Vegetation of the central part of the mire is dominated by a Eriophorum angustifolium-community and is mostly free of spruce trees (Böhlmann et al. 2005).

Field sampling and analysis

Fluxes of N2O and CH4 were measured biweekly with the closed chamber method during the study period that lasted from 19 June 2002 to 9 July 2003. Measurements have not been made at days with high snow cover during winter from November to March. For flux measurements, 10 chambers consisting of PVC rings (30 cm height, 30 cm diameter) were installed into the peat and remained throughout the entire study period. Four chambers were set up at the peripheral forested site and six chambers at the central non-forested site of the slope mire. Gas samples were taken at 0, 20 and 40 min after closing the chambers with a lid using evacuated glass flasks (100 ml). Before sampling, air pressure of the glass flasks was checked by pressure sensor (Loftfield et al. 1997). Air temperature was measured at a height of 2 m and soil temperatures were measured in 2.5, 5, 10 cm soil depth during gas sampling periods. The gas samples were analysed within few days by an automated gas chromatographic system consisting of a computer-controlled gas chromatograph (Carlo Erba) equipped with a flame ionization detector (FID) for CH4, an electron capture detector (ECD) for N2O measurements, and an autosampler for 64 sample flasks (Loftfield et al. 1997).

Peat samples were taken at the beginning of measurements in June 2002 next to the chambers from a depth of 0–20 cm. Air-dried peat samples were analysed for total C (Ct) and N (Nt) with an elemental analyzer (Hanau, Germany). The pH of the peat samples was determined after shaking with 0.01 M CaCl2. NH4 +and NO3 − of the peat samples have been extracted with a 2 M KCl-solution and analysed with a continuous-flow-analyser (Skalar, Netherlands) using the indophenol-blue method.

Groundwater piezometers were placed in the peat layer adjacent to the chambers. The groundwater piezometers were used to measure water table levels (WT), redox potentials (Eh, WTW, Germany), and oxygen levels (O2, WTW, Germany) at the time of gas measurements. Measurements were performed at a depth of 30 dcm. Furthermore, water samples were collected and analysed for ammonium, nitrate, pH (WTW, Germany) and electric conductivity (EC) (WTW, Germany).

Statistical analyses

All statistical analyses were performed using a SAS software package [SAS Institute (1999–2001), Inc., Cary, USA, Release 8.2] and Statistica for Windows v. 5.1 (StatSoft, Inc. 1996, Tulsa). Pearson correlation coefficients were calculated to elucidate relationships between gas fluxes, soil temperature and variables of mire pore water. One-way analysis of variance (ANOVA) has been used to test significance of differences between gas fluxes at the forested and non-forested part of the mire. Data were tested for normality and equality of variance before statistical analyses.

Results

Soil temperature, water table, peat soil, mire pore water

Thickness of peat differed significantly (p < 0.05, n = 4, Table 1) between the central and peripheral site of the mire being 105 and 56 cm respectively. Top layer (0–20 cm) of peat at both locations showed similar pH, bulk density, total N and C:N ratio while nitrate content was higher in the forested area of the mire (p < 0.05, n = 4, Table 1).

Table 1 Characteristics of the peat soil (Fibric Histosol) at the forested peripheral and the non-forested central area of the slope mire (0–20 cm depth, n = 4)
Full size table

Not being different between soil depths soil temperatures ranged from −0.1 to 19.6°C for the dates monitored (Fig. 1a). Water table levels differed significantly between the forested and non-forested areas (p < 0.05, n = 16, Fig. 2a). At the peripheral forested area, water table levels varied between −1 and −10 cm below the mire surface and between −4 and −27 cm at the central non-forested part of the mire. The highest water table levels were found during snow melting in April 2003, the lowest in July 2002.

Fig. 1
figure 1

a–c: Seasonal pattern of soil temperatures (°C) (a) and gas fluxes of N2O (b) and CH4 (c) (µg m−2 h−1) at the forested and non-forested monitoring sites of the slope mire (means with standard deviations)

Full size image
Fig. 2
figure 2

a–c: Seasonal patterns of water table levels (WT, cm) (a), nitrate concentrations (mg L−1) (b) and oxygen concentrations (mg L−1) of the mire pore water at the forested and non-forested monitoring sites of the slope mire (means with standard deviations)

Full size image

Pattern of pore water characteristics followed the temporal variation of the water level. Values of O2 being low during low water table levels in summer increased with water table in winter and peaked in April. On average the redox potential was 150 mV and 14 mV and O2 content averaged 2.6 mg L−1 and 1.4 mg L−1 at the forested and non-forested area, respectively. Resulting from a considerable spatial variation of O2 and Eh in pore water (Table 2) these values did not significantly differ. Values of pH, ammonium, total N and EC remained relatively constant during the study (not shown) while nitrate showed some variations with lowest values in May 2003 (Fig. 2b). In contrast to pH and values of ammonium and nitrate concentrations the conductivity of pore water differed significantly (p < 0.05, n = 16) between the forested and non-forested areas of the mire averaging 53 and 34 µS cm−1 respectively.

Table 2 Hydro-chemical variables of mire pore water at the forested peripheral and the non-forested central area of the slope mire during the study period (30 cm depth, n = 16)
Full size table

N2O

The N2O emissions ranged between −0.9 and +37.1 µg m−2 h−1 with peak emissions at the 19th June 2002 and 6th May 2003 (Fig. 1b).

No relationships were found between N2O release of the mire and soil temperature, water table level, Eh and ammonium concentration. Negative correlations were observed between N2O release and O2 contents at the central part of the mire and between N2O release and nitrate for the one measuring point only at the peripheral part of the mire (Table 3). Mean annual N2O emissions did not differ significantly between the sites and amounted to 0.4 ± 0.5 kg N ha−1 year−1 at the forested and 0.2 ± 0.4 at the non-forested site.

Table 3 Pearson correlation coefficients between gas fluxes (µg m−2 h−1), soil temperature (°C), water table levels and hydro-chemical variables of mire pore water at the forested peripheral (P) and the non-forested central area of the slope mire (C)
Full size table

CH4

The mire was found to be a CH4 sink at almost all sampling dates (Fig. 1c). The annual CH4 uptake of the mire adds up to −0.02 ± 0.01 kg ha−1 CH4-C. The highest uptake of −88.7 µg m−2 h−1 occurred during low level of water table at the 19th June 2002. At all other observation dates CH4 fluxes of the mire varied around zero with CH4 rates of −1.8 to 0.005 µg m−2 h−1. The exceptional singular high peak uptake does cause the significant correlation between CH4 uptake and soil temperature (Table 3). None of the other variables revealed a significant correlation to CH4 fluxes. CH4 fluxes amounted to −0.01 ± 0.01 kg ha−1 year−1 at the forested and −0.02 ± 0.01 kg ha−1 year−1 at the non-forested areas and did not differ significantly.

Discussion

Temporal pattern of N2O emission

Two N2O emission peaks observed in this study could be classified as event-related emissions according to Brumme et al. (1999). One N2O peak of 30 µg m−2 h−1 was measured at the central part of the mire on 19th June 2002 during low water table level. The second high N2O release on 06th May 2003 followed a freezing–thawing periods. Increased N2O emissions during the freezing–thawing cycles have been reported by several other studies (Teepe et al. 2000; Koponen and Martikainen 2004; Koponen et al. 2006). Soil thawing has been suggested to favour denitrification processes due to high water saturation and limited O2 diffusion (Teepe et al. 2000). Furthermore freezing of soils destroys microbial cells providing substrates for denitrifying microbes (Papen and Butterbach-Bahl 1999). Inhibition of N2O reductase at low temperatures was discussed also as a reason for high N2O release during freezing–thawing cycles (Melin and Nommik 1983). Several authors reported that increasing nitrate concentrations in liquid water during frost enhance N2O production (Stähli and Stadler 1997).

With the exception of the event-related emissions N2O fluxes remained low at both sites despite differences in water table levels. N2O release of peatlands has been reported to be mainly affected by water table and nutrition status (Martikainen et al. 1993; Augustin et al. 1996; Regina et al. 1996; Brumme et al. 1999). Negative relationships between emitted N2O amounts and water table levels have been established in various studies (Maljanen et al. 2003; Drösler 2005) due to increases of O2 availability promoting nitrification in drier peat that either directly increased N2O formation or produce NO3 − as precursor of denitrification (Seitzinger 1994; Regina et al. 1996). Very high N2O emissions between 25 and 72 kg N2O-N ha−1 within 46 weeks were observed in single chambers at a drained alder forest in Bornhöved, Germany (Brumme et al. 1999). N2O emissions up to 27 kg N2O-N ha−1 year−1 have been monitored at newly drained peatlands with a decreasing water table level down to −60 cm (Augustin et al. 1998). Similar drainage caused high N2O releases up to 79 µg m−2 h−1 for a short time (Dorwick et al. 1999). Several authors observed annual N2O emission rates of >1 kg N ha−1 at mean water table levels of −23 to −50 cm (Regina et al. 1996; Augustin et al. 1998; Drösler 2005). Commonly found higher water table levels in our study might be the reason for the generally observed low N2O emission.

Annual N2O fluxes

Annual mean N2O fluxes amounting to 0.25 kg N2O-N ha−1 did not statistically differ between the monitoring sites at the mire. These fluxes are in line with those reported for various studies of pristine bogs and fens with high water table levels (Martikainen et al. 1993; Augustin et al. 1996; Regina et al. 1996; Drösler 2005). Hence our hypothesis that raising O2 input through flowing water increases N2O release from mire water below the water table level by decreasing N2 formation could not be confirmed. Likewise assumptions that observed high O2 concentrations in the pore water are assumed to restrict N2 formation when nitrate is not limiting denitrification processes due to high atmospheric input (Davidson 1991; Granli and Bøckman 1994, Scholefield et al. 1997) have to be questioned. Denitrification processes resulting in N2 formation in water and sediments require O2 contents of <0.1 to a maximum of 0.8 mg L−1 (Granli and Bøckman 1994; Mühlherr and Hiscock 1998; Venterink et al. 2003). Accordingly nitrate should not have been reduced to N2 in large amounts in our slope mire during the study period since the oxygen contents mostly exceeded these values. In addition low pH of the mire pore water should have restricted N2 formation. Acidity inhibits N2O reductase and thus increases the N2O:N2 ratio (Granli and Bøckman 1994; Aerts 1997). At the prevailing hydro-chemical conditions of our slope mire formation of N2O should be favoured compared to N2. Various authors reported that groundwater was oversaturated with N2O which emitted to the atmosphere when the water reached the surface (Mühlherr and Hiscock 1998; Hefting et al. 2006). N2O captured in groundwaters below our slope mire is likely to be emitted when those groundwaters are rising to the surface in the spring which obviously requires further investigation.

Temporal pattern of CH4 emissions

CH4 fluxes of the mire didn’t show any seasonal pattern. Hydro-chemical variables of the mire pore water did not explain fluxes except probably the high CH4 uptake rates of 88 µg m−2 h−1 at the central part of the mire on the day of 19th June 2002. The high uptake occurred during low water table level of −27 cm and was assumed to be the result of CH4 oxidation by aerobic methanotrophic bacteria when aerobic conditions in peat could be enhanced. Water table has been identified to be the major controlling factor of CH4 emission rates in peatlands. With the water table up to 62% of the variability of CH4 fluxes have been explained (MacDonald et al. 1998; Drösler 2005). Different water table levels have been reported at which peatlands may change from CH4 release to CH4 uptake. Some authors reported a critical water table level of >−10 cm (Christensen et al. 2003; Drösler 2005). In contrast Augustin et al. (1996) observed still small CH4 releases of 0.6–3.5 kg C ha−1 year−1 in drained mires with water table levels down to −60 cm. Beside water table levels plant species composition and cover of vascular plants are known to affect CH4 release from peatlands. Plant species with aerenchymous tissue are able to transport CH4 through the root surface without being oxidized by methanotrophic bacteria (Joabsson and Christensen 2001). Even though these CH4 fluxes would be included in our measurements the mire showed no significant CH4 release. Thus plant mediated CH4 fluxes seems not to be important at the study site although our slope mire show a high cover of vascular plants. Similarly positive impact of available nutrients and labile carbon on the CH4 release (Joabsson and Christensen 2001; Yavitt et al. 2005) could not be confirmed in our study.

Annual CH4 fluxes

Annual CH4 flux revealed an uptake of −0.02kg CH4-C ha−1 not differing between the forested and non forested parts of the slope mire. The small annual consumption of CH4 is in contrast to reported CH4 releases from peatlands with similar high water table levels (Augustin et al. 1996; Drösler 2005). Restricted oxygen diffusion causing strictly anaerobic conditions of Eh values lower than −250 mV and high thickness of peat enhancing anaerobic peat layer capable of CH4 production were discussed to be reasons for high methane fluxes in peatlands with high water table levels (Mitsch and Gosselink 2000). The uptake of CH4 at our slope mire might be the result of a restricted CH4 production due to O2 input through groundwater movement causing redox potential far above the necessary value for methane production. A weak positive correlation between water inflow and oxygen content in pore water (R = 0.52, p < 0.05, n = 16) does indicate the importance of flowing groundwater for methane oxidation in slope mires. An additional explanation for low CH4 emission rates might be the thin peat layer of our slope mire with low total production of methane not exceeding methane oxidation in the aerobic part of the slope mire.

Conclusions

In contrast to peatlands without permanent water inflow the studied slope mire showed oxygen influx through inflowing pore water restricting reduction processes. Compared to pristine peatlands the influx of oxygen is assumed to restrict CH4 production of the slope mire explaining its weak CH4 consumption. N2O release from the slope mire didn’t differ from those of reported peatlands with high water table levels and low oxygen availability. Nevertheless slope mires have a high potential for the N2O formation due to available NO3 −. Our assumption that areas of slope mires with springs may be hot spots for N2O releases requires further research.