Carbon loss after forest drainage of three peatlands in southern Sweden

Abstract: Increased amounts of carbon dioxide in the atmosphere influence the global climate. It is therefore important to understand the effects of forestry measures on the distribution of carbon between soil, plants and atmosphere. In the present study, net changes in the peat carbon stores of mires, as a result of forestry drainage, were investigated. Drainage of a mire generally increases the decomposition of the peat, which leads to an increased flux of carbon dioxide to the atmosphere. On the other hand, drainage increases forest growth, which leads to an accumulation of carbon into the biomass of the trees and thereby also an addition of carbon to the peat by an increased production of litter. Three mire areas in southern Sweden were included in the study: the Siksjöbäcken area, where two drained fens, Letjärn and Särkalampi, and one undrained fen, Hamptjärn, were investigated; the Torvbråten area, which consists of one undrained and one drained bog, Myggbotjärn and Torvbråten; and Gillermossen, which is a drained fen. The differences between the areas made it possible to study how changes in the carbon stores after drainage depend on site factors. The two fen areas (the Siksjöbäcken area and Gillermossen), the Torvbråten bog plane and the Torvbråten marginal slope were regarded as four different sites. Drainage was carried out at Gillermossen in 1990, at Torvbråten in 1982 and at Letjärn and Särkalampi in 1981. In connection to the drainage of the mires, a hydrological study was carried out. Groundwater tubes were then installed at some stations at each mire. Using these tubes, the changes in the peat surface level after drainage could be measured for the present study. Volumetric peat samples were also collected in connection to the drainage, at the same stations. The change in the peat surface level, together with volume weights calculated for the early volumetric samples and for new volumetric samples, were used to calculate the total carbon content in the peat, down to a certain level, at a point in time close to the drainage and in the year 2000, i.e. 10-20 years later. The net change in the peat carbon store was determined by comparing the total carbon content of the peat at these two occasions, for each station. For the Siksjöbäcken area, the earlier values of volume weight could not be used and the carbon content calculated for the drained mires for the year 2000, in a certain layer, was instead compared to the carbon content in the corresponding layer in the undrained reference mire, Hamptjärn, in the same year. The main purpose of investigating both drained and undrained mires was that the undrained mires could serve as reference mires for the drained mires in the same area. The changes in the carbon stores of the undrained mires were subtracted from the changes at the drained mires and the resulting values were assumed to represent only the changes in the carbon stores that were actually caused by drainage. No correction could be made for Gillermossen, since there was no reference mire. For Letjärn and Särkalampi, no separate correction was needed, due to the different method used to calculate the carbon loss. The possible effects of the fertilization carried out on some of the mires, in connection to the drainage, were included in the "effects of drainage", since it is common practice to fertilize a mire after it has been drained. The peat was analyzed with respect to pH, C-content and N-content, both in connection to the drainage and in the year 2000. The degree of decomposition was determined for some stations in connection to the drainage and for all stations in the year 2000. The early values of pH, C/N ratio and degree of decomposition were used to analyze why the changes in the carbon stores differed between sites. Another site factor considered was the tree cover. If the different stations could be regarded as having a tree cover or not was decided in the year 2000. The changes in the carbon stores after drainage were also assumed to be related to how much the groundwater level had been lowered. At each station, the average groundwater level, during the frost-free part of the year, was compared for two years with similar amounts of precipitation, one before drainage and one some years after drainage, when the water table was assumed to have stabilized. The difference was interpreted as the change in the groundwater table that was caused by drainage. The water-level drawdown was then 30.4 cm at Letjärn and Särkalampi, 12.9 cm at the Torvbråten marginal slope, 28.8 cm at the Torvbråten bog plane and 50.9 cm at Gillermossen. The peat surface subsided at all stations during the period studied, which was unexpected for the undrained mires. The mean subsidence was 6.9 cm at Hamptjärn, 12.5 cm at Letjärn and Särkalampi, 3.7 cm at Myggbotjärn, 9.7 cm at the Torvbråten marginal slope, 39.4 cm at the Torvbråten bog plane and 11.0 cm at Gillermossen. The preliminary calculations of changes in the carbon stores, before the corrections using the reference mires, also showed a loss of carbon from all stations, even from those that were not drained. This could be an indication of real losses of carbon from the undrained mires, caused by dry periods or possibly by deposition of nutrients from anthropogenic sources. However, it could also be a result of systematic errors in the measurements. Explaining the subsidence and the relatively high values of carbon loss at the undrained mires requires additional information and is beyond the scope of the present study. However, it should be an interesting subject of further investigations. After the correction of the carbon loss values, by the use of the reference mires, the mean loss of carbon was 76 g C m-2a-1 for Letjärn and Särkalampi, 813 g C m-2a-1 for Gillermossen, 8 g C m-2a-1 for the marginal slope at Torvbråten and 635 g C m-2a-1 for the bog plane at Torvbråten. These values can be compared to the average accumulation of carbon for an undrained northern peatland, which is 21 g C m-2a-1 (Clymo et al., 1998). The uncertainties in the determinations of the peat surface level were at many stations relatively large compared to the change in the peat surface level that was caused by drainage. These uncertainties were considered to be the largest source of errors in the carbon loss values. The maximum errors in the carbon loss values were estimated, in two different ways. First, the errors that could arise from the uncertainties regarding the determinations of the peat surface level, at different stages in the investigation, were estimated and added together. Secondly, the errors in the final carbon loss values, that could still be left after the reference mire correction, were estimated by the calculation of confidence intervals, for the mires where the reference mires were used, i.e. for Torvbråten and for Letjärn and Särkalampi. The uncertainties were found to be quite large, but they were regarded to be small enough to allow some conclusions to be drawn about how the amounts of carbon lost may be related to site factors. The drainage impact, the tree cover and perhaps the initial degree of decomposition seemed to be the most important factors influencing the changes in the carbon stores, while the C/N ratio and the pH seemed to be less important. As an example, the losses of carbon from the Torvbråten bog plane were large despite a low pH and a high C/N ratio. Since the change in the groundwater table was virtually the same for the bog plane as for the fens in the Siksjöbäcken area (Letjärn and Särkalampi), 28.8 cm and 30.4 cm respectively, the large difference in carbon loss between these sites was assumed to be caused either by the larger share of tree-covered stations in the fen area or by the initially higher degree of decomposition of the fen peat, or by both. The large difference in carbon loss between the two fen areas, Gillermossen on the one hand and Letjärn and Särkalampi on the other hand, was probably due to a combination of different factors: a difference in drainage impact, a difference in C/N ratio, the lack of a reference mire correction for Gillermossen and a shorter period of study for Gillermossen.