Temporal pattern of GHG fluxes

Compared to natural bogs in Canada, the range of Reco at the study sites are greater.

According to Waddington & Warner (2001), Reco ranged from 0.41 to 5.21, 0.63 to 3.68 and 0.35 to 6.3 g CO2-C m^-²d^-1 in natural hummocks, natural lawns and mined sites, respectively.

At all sites the seasonal pattern of the Reco followed basically the course of the temperature.

The deviations from this course were due to the phenology of the vegetation. In spring, when temperatures are raised already, the vegetation is not fully developed, while in late summer and autumn, senescence occurs, although temperatures are still quite high. Thus, only heterotrophic respiration contributes to Reco, while autotrophic respiration is low. In July, temperatures are highest and the vegetation is fully developed, thus highest Reco occurs during July, which is consistent with Beetz et al. (2013). Another driving factor for Reco is the wl (Silvola et al. 1996, Flessa et al. 1997, Waddington et al. 2002). However, an effect was not observable because it was overshadowed by other influencing factors. Thus, beside temperature, vegetation is probably the main driving force for the seasonal variation in Reco.

159 4.4.2.2 Net ecosystem exchange

The results at the Molinia site were in accordance with a Dutch bog, which takes up CO2

during June, July and August, while the rest of the year, from September until May, the bog emits CO2 (Nieveen et al. 1998). Mean summer NEE (June to Sept.) varied between -1,010 and -1,810 kg CO2–C ha^-1 in Eriophorum dominated rewetted former peat cut sites in Finland (south boreal zone) (Kivimäki et al. 2008). This complies with the summer flux rates of the Molinia site and Eriophorum site. Lafleur et al. (2003) and Drösler (2005) found seasonal flux rates similar to the S. cuspidatum site and the S. papillosum site in natural bogs. However, they found lower maximum net release and maximum net uptake rates.

Important factors for GPP, beside PAR, are phenology and type of the vegetation (Lafleur et al. 1997, Tuittila et al. 1999). According to Buchmann & Schulze (1999) and Wilson et al.

(2007), GPP is related to leaf area index (LAI) and vascular green area (VGA). Daily GPP of the Molinia site and Eriophorum site on the one hand, and the S. cuspidatum site and S.

papillosum site on the other hand were similar, due to similar types of vegetation, respectively. The delayed course of the photosynthesis in spring (compared to the course of the PAR) was due to the smaller LAI or VGA, while the decrease of photosynthetic activity in late summer was caused by the onset of senescence (Wilson et al. 2007). In comparison to the other sites, at the Eriophorum site GPP increased early in spring and decreased late in autumn. Eriophorum has a high potential for photosynthesis early in the season and throughout most of the season (Tuittila et al. 1999).

Another important driver of the gas exchange is the wl (Titus et al. 1983, Schipperges &

Rydin 1998, Tuittila et al. 1999, Waddington & Warner 2001, Lafleur et al. 2003, Wilson et al. 2007). The strong decrease of GPP in July 2010 at the S. cuspidatum site was caused by the extreme decline of the wl in connection with warm and dry weather, leading to a dry-out of the Sphagnum. Sphagnum mosses are very exposed to wl changes, and photosynthesis decreases with decreasing tissue water content (Titus et al. 1983, Schipperges & Rydin 1998).

The moss capitula might even not recover from drought if the plants are dried above their water compensation point (Schipperges & Rydin 1998, Lafleur et al. 2003). However, during the following measurement campaign in August 2010, the Sphagnum had recovered from drought. Prior to this measurement campaign, there was more precipitation, less global radiation and consequently a higher wl. In contrast, the other sites were not affected by the dry period in July 2010, because the Molinia site and Eriophorum site were not dominated by

160 Sphagnum, but by species like Eriophorum which are less vulnerable for wl changes. These plants are able to keep their stomata open during dry periods because the roots go deep into the ground (Tuittila et al. 1999). At the S. papillosum site the wl was maintained high all year round, thus also during the summer. Consequently, the Sphagnum mosses did not dry out. The clearly visible effect of the dry period at the S. cuspidatum site proves the ability of the models to account for such influencing parameters.

4.4.2.3 Methane

Methane emissions of bogs range between less than 1 to more than 500 mg CH4-C m^-2 d^-1 (Crill et al. 1988, Moore & Knowles 1989, Freeman et al. 1993, Waddington & Price 2000, Sommer & Fiedler 2002). At the study sites the results were well below 500 mg CH4-C m^-2 d

-1. The maximum CH4 uptake and maximum CH4 release of the study sites are in agreement with the results of Drösler (2005), who found a maximum methane uptake of -0.10 mg CH4-C m^-2 h^-1 and a maximum methane release of 10.30 mg CH4-C m^-2 h^-1 in rewetted bogs in South Germany as well as a maximum CH4 uptake of 0 mg CH4-C m^-2 h^-1 and a maximum CH4

release of 18.54 mg CH4-C m^-2 h^-1 in natural bogs in South Germany.

The main driving forces for CH4 emissions are wl and soil temperature. Methane is built by methanogenic bacteria in the saturated (anaerobic) zone, and is oxidized to CO2 in the aerobic layer by methanotrophic bacteria (Munk 2001). While Roulet et al. (1993) declared that the zone of maximal potential for CH4 production is located in the uppermost part of the saturated zone and the maximum of CH4 oxidation is situated directly above the saturated zone, Kettunen et al. (1999), established that the maximal CH4 production is on average about 20 cm below the wl and the maximal CH4 oxidation about 10 cm below wl. However, only a few dm of the aerobic layer is sufficient for complete oxidization of the produced methane (Roulet et al. 1993, Meyer et al. 2001).

The models demonstrate that fluxes increase with rising wl and rising soil temperatures and that there is a threshold value of the wl above which the fluxes increase sharply (Fig.4.14).

This confirms the findings of Christensen et al. (2003) and Drösler (2005). A significant proportion of the variance can be explained by the models. However, the regressions are not satisfactory. Firstly, not even half of the variance could be explained. Secondly, the residuals fluxes are not normally distributed, but are skewed to the right (Fig.4.15). Thirdly, the residuals show a decreasing trend plotted against wl (Fig.4.15). The skewed distribution may

161 not be a major issue and is because soils can emit high amounts of CH4, but take up only small amounts, thus most values are close to 0, while a few values show high positive values.

A reason for the trend of the residuals is the bad quality of the raw data. In several cases, when gas-concentrations of the gas samples were low, and thus the slope was flat, no significant fluxes could be determined because the analysis of the samples in the gas chromatograph revealed too imprecise gas-concentrations.

Including more variables might improve the models. Granberg et al. (1997) used wl and soil temperature as important variables in their methane-model, but included also substrate effects in the model. Updegraff et al. (1998) and Kettunen et al. (1999) observed hysteresis effects.

Moreover, Granberg et al. (1997) used temperature values of the anoxic and oxic parts of the profile in order to model methane production and methane consumption, whereas in this study the soil temperatures were measured at a fixed depth (2 cm and 5 cm), with the consequence that sometimes the temperature was measured in the oxic zone and sometimes in the anoxic zone. According to Le Mer & Roger (2001), higher temperatures promote methanogenesis, whereas methanotrophy is less temperature-dependent. However, Granberg et al. (1997) discovered that the temperatures from the oxic and the anoxic zones explained 21 % of the variance, while the temperatures at a fixed depth explained only 5 %. The final models of Granberg et al. (1997) revealed coefficients of determination between R² = 0.49 and R² = 0.75.

According to the model, methane emissions are expected mainly in spring and autumn. This is in line with Beetz et al. (2013), who measured highest emissions in spring and autumn. If the wl is kept at a high level in summer, the highest methane fluxes are expectable in summer.

However, at the Eriophorum site and the S. cuspidatum site occasional measurement dates showed comparatively high wl and high temperatures, but CH4 emissions were low. Hence, there might be other factors, e.g. air pressure, which exert influence (Tokida et al. 2007).

The results are important for the management of peatlands. In order to keep CH4 emissions low, the wl should be kept below ground surface.

4.4.2.4 Nitrous oxide

The erratic pattern with high temporal variability in the course of the year was also described by Drösler (2005) and Beetz et al. (2013). Periodical wetness probably induced the N2O peaks (Flessa et al. 1998, Meyer 1999). Drösler (2005) determined at rewetted sites a maximum

162 release of 0.028 mg N2O-N m h and a maximum uptake of -0.022 mg N2O-N m h . Natural sites exhibit lower emissions (maximum release: 0.016 mg N2O-N m^-2 h^-1, maximum uptake: -0.030 mg N2O-N m^-2 h^-1). The values at the study sites were higher.