Simultaneous high C fixation and high C emissions in Sphagnum mires

Abstract. Peatlands play an important role in the global carbon (C) cycle due to their large C storage potential. Their C sequestration rates, however, highly vary depending on climatic and geohydrological conditions. Transitional mires are often characterised by floating peat with infiltration of buffered groundwater or surface water. Sphagnum mosses grow on top, producing recalcitrant organic matter and fuelling large C stocks. As Sphagnum species strongly differ in their tolerance to the higher pH in these mires, their species composition can be expected to influence C dynamics in transitional mires. We therefore experimentally determined growth and net C sequestration rates for four different Sphagnum species (Sphagnum squarrosum, S. palustre, S. fallax and S. magellanicum) in aquaria, with floating peat influenced by the infiltration of buffered water. Surprisingly, even though the first three species increased their biomass, the moss-covered peat still showed a net efflux of CO2 that was up to 3 times higher than that of bare peat. This species-dependent C release could be explained by Sphagnum's active lowering of the pH, which triggers the chemical release of CO2 from bicarbonate. Our results clearly illustrate that high Sphagnum biomass production may still coincide with high C emission. These counterintuitive C dynamics in mire succession seem to be the result of both species- and biomass-dependent acidification and buffered water infiltration. Together, these processes can explain part of the large variation in C fluxes (ranging from C sequestration to C release) reported for pristine mires in the literature.


Introduction
Since peatlands store approximately one-third of all terrestrial carbon (C), they are important in the global C cycle (Gorham, 1991), and their C dynamics have been studied throughout the world (Gorham et al., 2003;Bortoluzzi et al., 2006;Golovatskaya and Dyukarev, 2009;Rowson et al., 2010). Although it is well known that degraded and drained peatlands generally are net C sources due to increased decomposition rates (Alm et al., 1999;Waddington et al., 2001;Moore, 2002) -with net emissions ranging from +80 to +880 g C m −2 yr −1 (Lamers et al., 2015; for all presented values of C fluxes, positive values represent net C losses to the atmosphere, whereas negative values represent net storage of C in growing peat throughout the manuscript) -pristine, growing peatlands (mires) accumulate C and are therefore considered to be C sinks (Belyea and Malmer, 2004). The full greenhouse gas budget is, however, more complex. First, almost all peatlands are sources of methane (CH 4 ) (Moore and Roulet, 1995;Saarnio et al., 2007), and second, not all pristine peatlands appear to be sinks of carbon dioxide (CO 2 ) (Waddington and Roulet, 2000;Riutta et al., 2007). For groundwater-or surface-waterfed (minerotrophic) fens, CO 2 fluxes have been reported to range from −208 to +190 g C m −2 yr −1 (Martikainen et al., 1995;Carroll and Crill, 1997;Bubier et al., 2003), whereas for transitional mires, fluxes of −124 to +58 g C m −2 yr −1 have been reported (Moore and Knowles, 1987;Koch et al., 2008;Salm et al., 2009 (Wheeler and Proctor, 2000;Sjörs and Gunnarsson, 2002). Other examples include edges of bog systems (lagg zones) influenced by surrounding surface water and local patches influenced by percolating water (Giller and Wheeler, 1988). Transitional mires often consist of floating peat infiltrated by moderately base-rich water, which determines species composition and stimulates buoyancy, through its effect on decomposition and subsequent gas production (Lamers et al., 1999;Smolders et al., 2002). Since they increase habitat heterogeneity at various scales, these intermediate peatland systems often form hotspots of biodiversity (Verberk et al., 2010). Transitional, floating mires are mainly characterised by Cyperaceae and a moss layer of different Sphagnum species, whose dominance strongly increases during succession (Du Rietz, 1954;Vitt and Chee, 1990;Wheeler and Proctor, 2000). Sphagnum growth in transitional mires is, however, not as straightforward as in bogs, since most Sphagnum species are sensitive to both high pH and increased concentrations of calcium (Ca) and bicarbonate (HCO − 3 ) in pore water and surface water (Clymo, 1973). As Sphagnum spp. lack stomata, water conducting tissue and roots, their growth, nutrition and vitality depend on the chemical composition of the surrounding water (Robroek et al., 2009). Despite Ca-and HCO − 3 -rich conditions, floating rafts in transitional mires may, however, still form suitable habitats for Sphagnum species, since they are always watersaturated and are fed by rainwater, which accumulates in the top (moss) layer and dilutes the buffered surface water (Lamers et al., 1999;Smolders et al., 2003) Sphagnum spp. strongly influence their environment and are thus important ecosystem engineers in peatlands (Van Breemen, 1995). They are capable of actively acidifying their habitat by exchanging cations for protons (Clymo, 1963;Hajek and Adamec, 2009) and releasing organic acids (Van Breemen, 1995). Furthermore, Sphagnum spp. keep their environment moist due to the high water holding capacity of their hyaline cells (Clymo, 1973) and compact growth structure. By increasing the acidity and moisture content of their habitat, Sphagnum spp. also slow down decomposition rates, thus providing optimal conditions for the accumulation of organic material. Moreover, the high concentration of phenolic compounds in their tissues, including antibiotics (Verhoeven and Toth, 1995), further decreases decomposition rates (Yavitt et al., 2000;Freeman et al., 2001). This combination of traits results in a strong contribution of Sphagnum mosses to C sequestration and peat formation worldwide (Coulson and Butterfield, 1978;Limpens and Berendse, 2003).
Due to differences in habitat preference among Sphagnum species, they inhabit different successional stages in peatlands (Vitt and Chee, 1990). Since biomass production (Gerdol, 1995), acidification rates (Kooijman and Bakker, 1994), decomposition rates (Rochefort et al., 1990;Limpens and Berendse, 2003) and drought-tolerance (Nijp et al., 2014) are species-specific, the species composition of the Sphagnum layer in turn may strongly influence the biogeochem-istry and C balance of their habitat. This means that the C sequestration potential of the different successional stages of peatlands may strongly depend on which Sphagnum species is dominant at a particular stage. In transitional mires, the species composition will strongly depend on pH, buffering components and water content. How the Sphagnum species composition influences the biogeochemistry and C balance in transitional mires, however, remains largely unknown.
Although a vast number of studies have presented field measurements of C dynamics in all types of peatland systems, including transitional mires, establishing the origin of the huge variation reported for both CH 4 and CO 2 fluxes in these field studies is challenging. Studies on both C dynamics and the influence of Sphagnum mosses using a controlled laboratory approach, however, have not yet been performed to our knowledge. The goal of this study was therefore twofold: first, to investigate the growth of different Sphagnum species under controlled environmental conditions characteristic for transitional mires, and second, to study C fluxes and their underlying mechanisms in these systems. Four different Sphagnum species, S. squarrosum, S. palustre, S. fallax and S. magellanicum, were grown on peat floating on Caand HCO − 3 -rich water. Besides growth parameters of these mosses, we studied their contribution to the net C fluxes in these potentially peat forming systems. We hypothesised that Ca-and HCO − 3 -rich conditions would lead to considerable differences in performance between the four Sphagnum species, based on differences in their tolerance to these buffering components and in their growth rates. Furthermore, we expected more tolerant Sphagnum species to strongly determine the C sequestration of these systems.

Experimental set-up
Intact floating peat monoliths (25 × 25 cm; height 21.85 ± 2.08 cm; n = 8) were cut from a floating mire in the southern Netherlands (51 • 24 6.1 N, 6 • 11 10.5 E) in late March 2012. This floating mire was dominated by helophytes species Typha latifolia and Calla palustris, whereas the moss layer consisted mainly of Sphagnum fallax. After cutting, all vegetation was removed and the bare peat was transferred to glass aquaria (25 × 25 × 30 cm; length × width × height) in the field to minimise damage to the peat structure. The peat had an organic matter content of 92.7 ± 0.4 % (determined by loss on ignition; 3 h at 550 • C) and contained 3.6 ± 0.4 mmol kg −1 fresh weight (FW) of Ca (determined by digestion of 200 mg of dry soil with 4 mL of HNO 3 and 1 mL of H 2 O 2 using a microwave oven (MLS 1200 Mega, Milestone Inc., Sorisole, Italy), after which diluted digestates were analysed by inductively coupled plasma spectrometry (ICP-OES iCAP 6000; Thermo Fisher Scientific)). In the laboratory, 6.25 L of Ca-and HCO − 3 -rich treatment water was added to each aquarium (Table 1), on which the peat floated. The underlying water layer was subsequently refreshed with treatment water at a rate of 5 L week −1 using peristaltic pumps (Masterflex L/S, Cole-Parmer, Vernon Hills, IL, USA). All floating peat monoliths received artificial rainwater (Table 1) five times a week, at a rate corresponding to the Dutch annual rainfall of 800 mm. During the experiment, the aquaria were kept in a water bath maintained at 18 • C (up to a maximum of 23 • C at the end of the day) using a cryostat (NESLAB, Thermoflex 1400, Breda, the Netherlands). Furthermore, a light regime of 200 µmol m −2 s −1 (PAR; 16 h light/8 h dark) was maintained (Master Son-T Pia Plus, Philips, Eindhoven, the Netherlands). This regime of temperature and light was chosen to mimic summer conditions.
On four floating peat monoliths, four different species of Sphagnum (Sphagnum. squarrosum, S. fallax, S. palustre and S. magellanicum) were planted together. S. squarrosum is a species of moderately rich fens and occurs in environments with pH values up to 7 (Clymo, 1973). S. fallax, on the other hand, can be quite sensitive to high pH or drought, but is also known for its high potential growth rate under minerotrophic conditions (Buttler et al., 1998). S. palustre is a widespread species found in habitats that are neither highly calcareous nor highly acidic (Daniels and Eddy, 1990). S. magellanicum is a species associated with poor fens and bogs, and it is restricted to a more acidic habitat (Vitt and Chee, 1990;Hajek et al., 2006). The first three species were collected in a peatland area in the north-western part of the Netherlands (Ilperveld; 52 • 26 42.5 N, 4 • 55 45.1 E), while the latter species was collected in an area in the south of the Netherlands (Maasduinen; 51 • 34 56.3 N, 6 • 6 13.5 E). For all species, a patch of 50 ± 10 g fresh material (1.6 ± 0.8 g DW; moss length 3 cm) was applied randomly to one of the corners of the aquarium. Mosses were put upright in a patch of approximately 50 cm 2 . The remaining four floating peat monoliths were kept as non-vegetated controls.
Since soils were floating and not inundated, the "surface water" will be called infiltrating water throughout this paper. This infiltrating water was sampled underneath the peat monolith, while pore water was extracted using 10 cm soil moisture samplers (SMS Rhizons, Eijkelkamp, Giesbeek, the Netherlands), which were inserted vertically into the soil. For each peat monolith, two SMS rhizons were installed, and samples were taken by attaching vacuum bottles. Analyses were performed on pooled samples to reduce the effect of variation within the soil.

Chemical analyses
During the 12 weeks of the experiment, pH and total inorganic carbon (TIC) concentration of infiltrating water and pore water were measured every 2 weeks (seven times in total). pH was measured with a standard Ag/AgCl electrode (Orion, Thermo Fisher Scientific, Waltham, MA, USA) combined with a pH meter (Tim840 titration manager; Radiometer Analytical, Lyon, France). TIC was measured by injecting 0.2 mL of sample into a compartment with 1 mL phosphoric acid (0.4 M) in an infrared gas analyser (IRGA; ABB Analytical, Frankfurt, Germany), after which concentrations of HCO − 3 and CO 2 were calculated based on the pH equilibrium. Concentrations of PO 3− 4 , NO − 3 and NH + 4 were measured colourimetrically on an AutoAnalyser 3 System (Bran&Lubbe, Norderstedt, Germany) using ammonium molybdate (Henriksen, 1965), hydrazine sulfate (Kamphake et al., 1967) and salicylate (Grasshof and Johannse, 1972), respectively. Concentrations of Ca, Fe, K, Mg, total P and SO 4 were analysed by inductively coupled plasma spectrometry (ICP-OES iCAP 6000; Thermo Fisher Scientific).

Plant data
To preserve bare control soils and monocultures of the Sphagnum species, all aboveground biomass of non-Sphagnum species was carefully removed every 2 weeks. This vegetation consisted mainly of Typha latifolia and Juncus effusus seedlings. Every 2 weeks, growth and expansion of the mosses were recorded. Mosses were allowed to grow outside of their designated quarters to include the effects of competition between species. After 8 weeks of moss growth, pH was measured within the Sphagnum vegetation at 0.5-1 cm above soil level, using a pH meter (HQ 40d, Hach, Loveland, CO, USA) and an Ag/AgCl pH electrode (Orion 9156BNPW, Thermo Fisher Scientific, Waltham, MA, USA).
After 12 weeks, all moss biomass was harvested and the number of capitula (top 8-10 mm of the photosynthetically active tissue of the mosses) was counted for each plot. Length of the moss fragments was measured before living plant parts and dead parts were separated and weighed. Biomass was dried for 48 h at 70 • C to determine dry weight (DW). C and N contents (%) of dried moss material were determined using an elemental analyser (Carlo Erba NA1500, Thermo Fisher Scientific, Waltham, MA, USA).

Carbon fluxes
C fluxes were determined after 6 weeks of experimental treatments. Since the cover of S. magellanicum had declined severely by this time, the remaining patch was too small to cover with a closed chamber and the species was excluded from these measurements. C fluxes from soils covered with S. squarrosum, S. palustre or S. fallax and from bare control soils were measured in transparent and closed chambers (length, width and height of 10, 10 and 12 cm) for light and dark conditions, respectively. Dark measurements started at the end of the 8 h dark period and lights remained off during measurements, so that mosses remained dark-adapted. Samples were taken from the headspace immediately after placing the chambers on the aquaria, and subsequently after 2 and 4 h using 1 mL syringes, which were first flushed with headspace. They were analysed for CO 2 using an IRGA (ABB Analytical, Frankfurt, Germany) and for CH 4 using a gas chromatograph (5890 GC, Hewlett Packard, Wilmington, DE, USA). The slopes of the linear increases in both gasses were used to determine areal net C fluxes for each Sphagnum species and for bare peat. Measurements on CO 2 and CH 4 fluxes carried out under light and dark conditions were combined to calculate daily C fluxes. Under natural conditions, Sphagnum spp. tend to grow vertically, whereas in our experiment -due to the absence of supporting neighbouring mosses -elongated moss fragments fell over, causing an apparent horizontal growth. Our areal C fluxes measured with the closed chambers covering only part of the elongated fragments are therefore underestimates. We corrected for this by multiplying the areal C fluxes with the ratio of the area covered by elongated Sphagnum fragments and the area of the chamber.

Statistical analyses
All data were checked for normality of residuals and homogeneity of variance using the Shapiro-Wilk test for normality and the Levene test of equality of error variances, respectively. Differences in the chemical composition of surface water and pore water of soils with and without moss cover were analysed over time using linear mixed models. Differences between growth parameters of Sphagnum mosses (Figs. 1 and 2, Table 3) and C fluxes were tested using oneway ANOVAs with Tukey post hoc tests. In all tables and figures, averages are presented with standard error of the mean (SEM). All statistical analyses were carried out using SPSS for Mac (V21, IBM Statistics). 3 Results

Chemical composition of infiltrating water and pore water
Growth of Sphagnum significantly lowered the pH of the infiltrating water (P < 0.001), whereas pore water pH increased (P = 0.007) when mosses were present (Table 2). Concentrations of NO − 3 (P = 0.002) and Fe (P = 0.018) were significantly higher in the infiltrating water of moss-covered soils, whereas concentrations of Ca (P = 0.014) and P (P = 0.012) were higher in pore water of bare soils (Table 2). Still, concentrations of NO − 3 were low since N was predominantly present as NH + 4 in both infiltrating water and pore water (Table 2). Overall, NH + 4 concentrations in the infiltrating water were high compared to similar systems (e.g. Kooijman and Bakker, 1994), with values around 40 µmol L −1 (Table 2).

Sphagnum growth and acidification
S. squarrosum, S. fallax and S. palustre increased their biomass during the experiment (Fig. 1). The initial biomass of these species increased by 340, 250 and 600 %, respectively (P < 0.001). While length of the green part of the mosses (living moss) remained approximately the same, the total length of the moss fragments increased by 5-6 cm (Table 3). In S. fallax and S. palustre, elongation appeared to be the main factor responsible for the increased biomass, since the number of capitula decreased in these species by 35 and 19 % (Fig. 2). S. squarrosum was the only species that increased both in biomass (P < 0.001) and in number of capitula (P = 0.004). This species also increased its coverage in the aquaria (P = 0.001; Table 3), thereby invading the quarters of other species and proving to be a strong competi- Table 3. Characteristics of the different Sphagnum moss patches, including pH within the vegetation (0.5-1 cm above soil level; measured after 6 weeks of growth), and final data (after 12 weeks) on C : N ratio, length of the moss fragments and cover of the patches. For all variables, P values are given and significant differences between species are noted with different superscript letters. ns is not significant.
tor in this experimental set-up. In contrast, S. magellanicum was strongly affected by the HCO − 3 -rich water and showed a strong decline in both biomass (Fig. 1) and number of capitula (Fig. 2) during the experiment. S. squarrosum acidified its environment the most, with pH values down to 4.5 (P = 0.027; Table 3). S. fallax and S. palustre both had pH values of around 5.2-5.4, whereas S. magellanicum had the highest pH with values around pH 6 ( Table 3). The acidification rate was linearly correlated (P = 0.005; R 2 = 0.43) to the net increase in biomass shown by the three species, with S. magellanicum showing both the lowest biomass increase and the highest pH, and S. squarrosum showing the lowest pH and highest growth (Supplement Fig. S1).

Carbon exchange
Bare floating peat had a C emission of 0.3 ± 0.1 g C m −2 d −1 (Fig. 3). Surprisingly, when Sphagnum species were grown on these soils, net C emission appeared to be similar to or even higher than that of bare peat (Fig. 3). Soils covered with S. squarrosum showed the highest net emissions of 1.1 ± 0.2 g C m −2 d −1 (P < 0.001), despite being the species with the highest biomass increase. S. fallax and S. palustre had net C effluxes similar to those of bare peat, with 0.5 ± 0.1 and 0.2 ± 0.1 g C m −2 d −1 , respectively, even though significant increases in biomass were observed for these two species. C fluxes consisted almost exclusively of CO 2 , as methane (CH 4 ) contributed less than 2.5 % to net C emission and ranged from −0.8 to 43 mg C m −2 d −1 (data not shown). Still, due to the higher global warming potential of CH 4 compared with CO 2 , CH 4 contributed 10 to 30 % to the total greenhouse gas emissions expressed in CO 2 equivalents. These greenhouse gas emissions were higher for plots covered by S. squarrosum (1750 ± 350 g CO 2 -eq m −2 yr −1 ) than for plots covered by other species (580 ± 106 g CO 2 -eq m −2 yr −1 ) or unvegetated (550 ± 180 g CO 2 -eq m −2 yr −1 ) plots due to the higher CO 2 emission from these plots (P = 0.002; data not shown).

Discussion
For Sphagnum species growing on top of floating peat monoliths, the influence of groundwater and surface water infiltra- tion, buffered by Ca 2+ -HCO − 3 , into the peat was shown to be reduced, and even moderately sensitive species were capable of growing under these conditions. It was remarkable, however, to discover that while some of these species strongly increased their biomass, Sphagnum-covered patches simultaneously showed a net C efflux.

Tolerance of Sphagnum species to buffered conditions
Transitional mires are Sphagnum-rich systems characterised by the influence of calcareous and alkaline surface water or groundwater in the subsoil and are thus partly buffered systems. These environmental conditions are, however, not limited to transitional mires and occur more widely, since local spots with higher influence of groundwater or edges in contact with calcareous surface water occur in many peatlands. The acid neutralising capacity (ANC) of peatlands is mainly based on the presence of HCO − 3 and Ca 2+ . When protons are released into a system, they are initially buffered by the HCO − 3 -CO 2 buffering system (Sherlock et al., 1995;Lamers et al., 2015). Once most of the HCO − 3 has been consumed, protons (H + ) can be buffered by the cation-exchange capacity (CEC) of the peat, where base cations bound to soil particles are exchanged for H + (Lamers et al., 2015). Ca 2+ is usually the main component of the CEC, since it is the dominant divalent cation in many peatlands (Bache, 1984;Rippy and Nelson, 2007).
Several studies have indicated that Sphagnum can be sensitive to calcareous groundwater and surface water due to Ca and HCO − 3 toxicity (Clymo, 1973;Andrus, 1986;Lamers et al., 1999;Hajek et al., 2006). This sensitivity to one or both of the major buffering components of groundwater and surface water is species-specific and it strongly affected the performance of the Sphagnum species in our study. S. magellanicum appeared to be the most sensitive to the tested conditions as this species decreased both in number of capitula and biomass. S. fallax and S. palustre, on the other hand, increased in biomass, even though they were obviously outcompeted by the better-adapted S. squarrosum. S. squarrosum was able to increase both horizontally, in number of capitula, and vertically, by stem elongation. S. squarrosum is one of the few Sphagnum species that is still vital in systems with a higher influence of calcareous and therefore HCO − 3 -rich water and is even able to tolerate (temporary) immersion in these systems (Clymo, 1973;Vitt and Chee, 1990). Other species, including S. magellanicum, are known to be highly sensitive to increased pH and buffered conditions in their habitat (Clymo, 1973;Granath et al., 2010), which explains why S. magellanicum showed a strong decrease in our study. The typical habitats of S. fallax, S. palustre and S. magellanicum are all characterised by low pH (4.5-4.8) and low Ca 2+ content (35-40 µmol L −1 ) (Vitt and Chee, 1990;Hajek et al., 2006), although S. fallax and S. palustre can tolerate a wider range of environmental conditions in terms of acidity and trophic level than S. magellanicum (Daniels and Eddy, 1990). S. squarrosum, on the other hand, often occurs in rich to moderately rich fens (Vitt and Chee, 1990;Hajek et al., 2006), which are characterised by pH values of 5.1 to 6.7 and Ca 2+ concentrations of 270-500 µmol L −1 (Vitt and Chee, 1990;Kooijman and Bakker, 1994).

Succession of Sphagnum species
The transition of mineral-rich fens to acidic "poor fens" to oligotrophic bogs is believed to be initiated by the acidification of pioneer Sphagnum species (Wilcox and Andrus, 1987;Rydin and Jeglum, 2006;Granath et al., 2010). These pioneer species are expected to tolerate mineral-rich conditions, have a high growth rate and a high acidification capacity under more buffered conditions, which will allow them to change a mineral-rich fen into an acidic poor fen within a few decades (Granath et al., 2010). S. squarrosum may act as such a pioneer species and is often responsible for rapid succession in fens (Giller and Wheeler, 1988;Haraguchi et al., 2003), especially under nutrient-rich conditions (Kooijman and Bakker, 1995).
Our data confirm that S. squarrosum potentially acts as a foundation species for other Sphagnum spp. This species simultaneously increased its biomass considerably and acidified its environment most effectively, lowering pH to values around 4.5 despite continuous infiltration of surface water with an alkalinity of 3 meq L −1 , while the other three species could not lower pH below 5.2. Sphagnum species show differences in acidification rate, based on differences in their cation-exchange capacity (Rippy and Nelson, 2007). Additionally, however, Sphagnum acidification rates depend on their species-specific performance under certain environmental conditions. High growth rates combined with low decomposition rates (5-35 % mass loss yr −1 ; Clymo, 1965;Coulson and Butterfield, 1978;Verhoeven and Toth, 1995;Limpens and Berendse, 2003) result in a fast build-up of the peat layer and succession in species composition, which, in floating transitional mires, will slowly reduce the influence of the underlying calcareous water.

Carbon dynamics
Increase of the thickness of the peat layer due to Sphagnum growth shows that these species can sequester a significant amount of C. Sphagnum biomass can increase by approximately 70 to 600 g DW m −2 yr −1 (Gerdol, 1995;Graf and Rochefort, 2009;Hajek, 2009;Samaritani et al., 2011), which corresponds to a CO 2 fixation rate of approximately 28 to 240 g C m −2 yr −1 . If we extrapolate the daily CO 2 fixation rates of the three growing species in our experiment, S. squarrosum, S. fallax and S. palustre, to calculate yearly production rates, based on a growing season of 8 months, we find high CO 2 fixation rates of approximately 100-450 g C m −2 yr −1 . These values, however, overestimate actual field growth of these mosses, since the experiment was carried out indoors under summer conditions only. Still, even with these high CO 2 fixation rates, we found net C emissions from both bare peat and from peat covered with growing Sphagnum mosses.
Bare peat showed C emission rates of around 0.3 g C m −2 d −1 (Fig. 4), which consisted of 98 % CO 2 and 2 % CH 4 . Both bare peat and vegetated plots were a small source of CH 4 , with average emission rates of 2 to 20 mg C m −2 d −1 , which fall within the range of 4 to 500 mg C m −2 d −1 usually reported for saturated peatlands (e.g. Bartlett and Harris, 1993;Byrne et al., 2004;Saarnio et al., 2007;Salm et al., 2009). Still, the contribution of CH 4 to the greenhouse gas emission is much higher in terms of CO 2 equivalents, since the global warming potential of CH 4 is 34 times that of CO 2 (IPCC, 2013). The higher greenhouse gas emissions (as CO 2 equivalents) from the plots vegetated by S. squarrosum were, however, not due to differences in CH 4 emissions, but resulted from the much higher emissions of CO 2 from these plots.
When plots were vegetated by growing Sphagnum spp., CO 2 emissions increased, despite the accumulation of biomass by all three species (Fig. 4), which indicates that the source of this CO 2 could not solely be the decomposition of Sphagnum litter. The only likely explanation for this remaining net CO 2 efflux is therefore the chemical reaction R (1) that occurs when HCO − 3 -rich water comes into contact with the acidifying mosses (Fig. 4). The transition of HCO − 3 to CO 2 is the first step in the ANC of aquatic systems and will occur much faster than other buffering mechanisms, such as 3 -rich water, there is high heterogeneity within these systems. Part of the floating raft is shown in more detail below. Here, peat soils are covered with different Sphagnum species. Rates of C fixation in peat (downward arrow) and C emission to the atmosphere (upward arrows) are both derived from C flux measurements and presented in g C m −2 d −1 . As the mosses showed differences in final biomass, higher or lower amounts of biomass are depicted in the figure. Furthermore, the mosses differ in acidification rate, with significantly higher amounts of acids produced by Sphagnum squarrosum (left) than the other species. Since Sphagnum magellanicum declined severely in biomass due to its sensitivity to the calcareous water, its C fluxes could not be measured and the species was excluded from this figure.
cation exchange of Ca 2+ (Lamers et al., 2015). Active acidification was mainly observed in S. squarrosum, while S. fallax and S. palustre did not significantly lower pH more than Table 4. Origin and rates of C fluxes (in g C m −2 d −1 ) of peat covered with different species of Sphagnum. Net C fluxes, gross C fixation rates and gross C emissions are based on closed chamber measurements carried out under light and dark conditions. Other fluxes are calculated using Eq. (1), with the fraction of autotrophic respiration based on the maximum value found for Sphagnum respiration in the literature (31 %, Laine et al., 2011).

Net C flux
Gross C fixation Gross C emission Bare peat Autotrophic respiration Additional HCO − the dying S. magellanicum.
To further disentangle the different CO 2 sources responsible for the net CO 2 emission from plots vegetated with different species, we used a mass approach (Eq. 1; Table 4). Net CO 2 fixation was estimated based on the difference between light and dark CO 2 fluxes, whereas CO 2 emission was estimated based on dark fluxes. This CO 2 emission can be further divided into separate contributors, as is shown in Eq. (1).

Net C flux to atmosphere
Here, B represents the CO 2 flux from bare peat to the atmosphere, R is the dark plant respiration, C represents the flux of chemically produced CO 2 according to Reaction (R1) and F is the gross CO 2 fixation, calculated as the light CO 2 flux minus the dark CO 2 flux. Bare peat respiration was derived from dark fluxes of non-vegetated plots. For the Sphagnum respiration factor R, we have used the maximum value (31 %) from the range reported in the literature (12 to 31 % of photosynthetic C fixation, Haraguchi et al., 2003;Laine et al., 2011;Kangas et al., 2014). As a result, we obtain a conservative estimate of the C flux emitted through the chemical Reaction (R1), driven by the acid production of the Sphagnum mosses (C). Although the likely higher respiration rates during the light period and the use of estimated Sphagnum respiration impede an exact quantification of factor C, the fact that we find CO 2 emissions during the light period in growing Sphagnum patches clearly points out that there is a considerable chemical CO 2 source. Table 4 shows the different sources of the C fluxes as presented in Eq. (1). Furthermore, the implications that these values have on the landscape scale are depicted in a schematic overview of a floating transitional mire that is being fed by HCO − 3 -rich water (Fig. 4). Here, we show simultaneous C fixation and C emission of the three growing Sphagnum species from our experiment, with their different growth and acidification rates.
The production of HCO − 3 -derived CO 2 will occur in any situation where HCO − 3 -rich water comes into contact with an acidic environment, such as in the highly acidic lower layers of floating bog systems influenced by HCO − 3 -rich water (Lamers et al., 1999;Smolders et al., 2003). Therefore, CO 2 effluxes measured from the slightly acidic bare peat in our experiment are likely at least partially derived from acid-driven CO 2 production from HCO − 3 , as is illustrated in Fig. 4. Our finding that the most strongly acidifying and fastest growing mosses such as Sphagnum squarrosum show the highest C effluxes strongly suggests that active acidification enhances the production of HCO − 3 -derived CO 2. This leads to the apparent contradiction that while growth of Sphagnum will lead to accumulation of organic matter and thus contributes to the build-up of a peat layer, it is accompanied by a large net efflux of CO 2 ranging from 0.2 to 1.1 g C m −2 d −1 (Table 4, Fig. 4). While we show this phenomenon here in a controlled laboratory setting, net CO 2 effluxes have indeed been reported for transitional mires, with rates ranging from −0.34 to +0.16 g C m −2 d −1 (Moore and Knowles, 1987;Koch et al., 2008;Salm et al., 2009). As mentioned before, however, this phenomenon may not be limited to transitional mires. For example, bogs typically show an outflow of acidic water (H + and organic acids) and therefore Sphagnum-produced acids may also cause chemical CO 2 production outside the peatland system, thereby counteracting at least a part of the C sequestration realised by peat growth.

Conclusion
To provide insight into the processes driving the highly variable C fluxes measured in Sphagnum-dominated mires, we used an experimental approach, which revealed a novel, overlooked mechanism, explaining part of the variation in CO 2 fluxes. Our results clearly show that high biomass production in mires can coincide with a net emission of carbon, due to a combination of biological and chemical processes. We feel that the acidification-driven CO 2 production is an underestimated factor that plays a significant role in C fluxes in transitional mires and other systems where calcareous (Caand HCO − 3 -rich) groundwater or surface water comes into contact with growing and acidifying Sphagnum mosses. Our results suggest that, under these conditions, for every gram of C that is fixed by Sphagnum, there is an emission of 0.8-1.4 g C through chemical processes, depending on Sphagnum acidification potential. We hypothesise that this phenomenon can specifically play an important role in early succession from minerotrophic to ombrotrophic conditions, when the influence of calcareous water in combination with Sphagnum growth is greatest. Due to the continuous build-up in these systems, the thickness of floating rafts will increase during succession and the lateral influence of the calcareous water will decline, leading to strong spatial and temporal variation in C fluxes in these systems. This finding may therefore help explain part of the strong variation in C balances measured in seemingly similar peatland systems.
The Supplement related to this article is available online at doi:10.5194/bg-12-4739-2015-supplement.
Author contributions. The experiment presented in this paper was designed by S. F. Harpenslager, G. van Dijk and L. P. M. Lamers and carried out by S. F Harpenslager and G. van Dijk. Figure 4 was designed by G. van Dijk, based on input provided by S. F. Harpenslager. The manuscript was prepared by S. F. Harpenslager, with contributions of all co-authors.