: Monique M. P. D. Heijmans, Nature Conservation and Plant Ecology Group, Wageningen University, Bornsesteeg 69, NL-6708 PD Wageningen, The Netherlands (tel.: +31 317483528, fax: +31 317484845, e-mail:email@example.com).
1 We studied the effects of elevated atmospheric CO2 and increased N deposition on the plant species composition of a Sphagnum-dominated bog ecosystem in the Netherlands. Large peat monoliths (surface area 1 m2, depth 0.6 m) with intact bog vegetation were kept outdoors in large containers and were exposed to elevated CO2 or increased N deposition for three growing seasons. Elevated CO2 conditions (target concentration 560 µmol CO2 mol−1) were created using MiniFACE technology. In a separate experiment, N deposition was increased by 5 g N m−2 year−1 by adding dissolved NH4NO3 at 3 week intervals during the growing season.
2 Elevated atmospheric CO2 increased height growth of Sphagnum magellanicum, the dominant Sphagnum species, in the second and third growing seasons. Vascular plant biomass was not significantly affected by elevated CO2, but growth of species growing close to the moss surface was influenced negatively by the increased Sphagnum height growth. Elevated CO2 did not change allocation to below-ground plant parts.
3 Adding N increased above-ground vascular plant biomass. The shallow-rooted species Vaccinium oxycoccus responded most to the increased N deposition. Sphagnum growth was significantly reduced in the third growing season. This reduction was likely the result of the increased vascular plant cover, given the observed negative relation between vascular plant cover and Sphagnum growth.
4 The observed shifts in species composition as a result of species-specific responses to treatments, and interactions between peat mosses and vascular plants will have important consequences for the sequestration of carbon in the bog ecosystem.
There has been a great deal of research on the effects of elevated atmospheric CO2 concentrations on growth of individual plants, but relatively little on the effects on natural vegetation. However, plant level responses to elevated CO2 cannot be directly extrapolated to the ecosystem level, because interspecific interactions and resource limitation of growth come into play (Körner 1996). Clearly, elevated CO2 affects ecosystems, but multiyear studies with natural vegetation are needed to characterize the long-term responses or the mechanisms (Koch & Mooney 1996; Körner 1996).
The impact of elevated CO2 on ecosystems has repercussions on CO2 in the atmosphere. Increased sequestration by terrestrial ecosystems will slow the accumulation of CO2 in the atmosphere and thus the rate of climate change (Schimel 1995; Walker et al. 1997). Peat-forming ecosystems, such as bogs, constitute a long-term sink for atmospheric CO2, such that about one-third of the world’s pool of soil organic carbon (455 × 1015 g) has accumulated in northern peatlands (tundra and boreal forest regions) over thousands of years (Gorham 1991). Sphagnum-dominated systems have the greatest peat accumulation potential due to their extremely low rates of decomposition (Thormann et al. 1999) and climate-induced or other changes could have important consequences for the global carbon cycle.
The amount of carbon sequestered in peatlands depends on the balance between the production and decomposition of plant material, either of which may change with species composition. In bog vegetation, Sphagnum species decompose slowly (Clymo & Hayward 1982) and their recalcitrant litter forms the bulk of bog peat (Coulson & Butterfield 1978). The other (vascular) plants contribute less to peat accumulation and enhance methane emission by supplying labile organic carbon through root exudation. In addition, some vascular plant species (Eriophorum) transport methane through aerenchymae from the anaerobic root zone to the atmosphere, thus diminishing the fraction of methane that oxidizes in the upper, aerobic, peat layer (Frenzel & Rudolph 1998; Saarnio et al. 1998; Joabsson et al. 1999). Carbon sequestration in bogs is therefore expected to increase when Sphagnum gains a competitive advantage over vascular plants.
Elevated atmospheric CO2 and increased N deposition, both associated with global change, could alter this competitive balance. Elevated CO2 generally stimulates growth of individual plants, but net production of a whole plant community is not necessarily increased. One of the few consistent findings in CO2-enrichment studies in multispecies systems is that the CO2 response is species specific, with some species declining and other species gaining in abundance (Körner 1996; Warwick et al. 1998; Leadley et al. 1999).
Data for bog vegetation are restricted to measurements on individual Sphagnum species (Silvola 1985; Jauhiainen et al. 1994, 1998). Vegetation responses, however, often depend on interactions with other environmental factors, so that in nutrient-poor ecosystems, such as arctic tundra, nutrient availability and/or uptake must increase before plants are able to benefit from CO2 fertilization (Oechel & Vourlitis 1994). The only sources of mineral nutrients in ombrotrophic bogs are atmospheric deposition and the mineralization of dead organic matter. Sphagnum is very effective in absorbing nutrients from atmospheric deposition (Woodin & Lee 1987; Williams et al. 1999), thereby reducing this supply of nutrients to vascular plants. Vascular plants depend largely on the nutrients mineralized from dead organic matter, and even this supply is reduced by the slow rate of decomposition of Sphagnum litter.
We hypothesize that elevated CO2 will change the competitive balance between mosses and vascular plants to the benefit of the less nutrient-limited Sphagnum (Sveinbjörnsson & Oechel 1992; Jauhiainen et al. 1998). Ambient N deposition in the Netherlands is high and effective interception by Sphagnum may well lead to N limitation of vascular plants, but not of moss growth (Aerts et al. 1992). When N deposition is increased, Sphagnum may no longer capture all nutrients (Woodin & Lee 1987) and N reaching the rhizosphere may increase growth rates of vascular plants (Malmer et al. 1994). Increased vascular plant cover may depress Sphagnum growth by shading (Clymo 1973; Hayward & Clymo 1983), thus reversing the competitive balance. Although many studies have addressed the short-term response of Sphagnum to increased N deposition, experimental evidence obtained from intact bog vegetation is still scarce.
To test these hypotheses, we kept large peat monoliths with intact bog vegetation in large containers outside and exposed them to elevated CO2 or increased N deposition during three growing seasons. We followed Sphagnum growth (height increment) and the abundance of vascular plant species and determined Sphagnum biomass growth, vascular plant biomass (above and below ground) and nutrient concentrations in plant tissues at final harvest.
Materials and methods
Peat monoliths were taken from a small mire in the State Forest of Dwingeloo (52°49′ N, 6°25′ E) in the north of the Netherlands. This formerly extensive heathland with active sand dunes has been forested with pine and oak since the beginning of the 20th century. Our site is one of a series of heathland pools (0.5–1.5 m deep) in a former glacial valley that has become filled with peat. The mire measures 50 m by 150 m and is surrounded by forest. The mean annual rainfall is 840 mm and the mean annual temperature 9 °C. A surface layer of relatively undecomposed Sphagnum litter between 20 cm and 40 cm thick overlies wet, highly decomposed peat and, at 5–15 cm depth has a bulk density of 30–50 g dm−3; N concentration of 11 mg N g−1; P concentration of 0.4 mg P g−1; and K concentration of 0.6 mg K g−1. The pool bed consists of loamy sand.
Irregular peat cutting in the past has resulted in a mosaic of secondary succession stages. The vegetation of the monoliths consisted of Sphagnum magellanicum Brid. lawns with Vaccinium oxycoccus L., Erica tetralix L. and Eriophorum angustifolium Honck. as dominant vascular plant species. Other species present were Sphagnum papillosum Lindb., Sphagnum fallax (Klinggr.) Klinggr., Aulacomnium palustre (Hedw.) Schwägr., Calliergon stramineum (Brid.) Kindb., Drosera rotundifolia L., Calluna vulgaris (L.) Hull, Empetrum nigrum L., Rhynchospora alba (L.) Vahl, Eriophorum vaginatum L., Andromeda polifolia L. and Pinus sylvestris L.
Experiments were carried out in a grassland near Wageningen (51°99′ N, 5°70′ E), fenced to exclude rabbits, over a 3 year period during which the weather varied considerably (Fig. 1). The first year (1996) was very dry with the driest spring of this century, while 1998 was exceptionally wet. Weather in 1997 was generally warm and sunny. The winter of 1997/1998 was very mild. In 1998, spring and summer were cloudy.
Wet N deposition (NH4 + NO3) has been measured in Witteveen (15 km from field site) and in Wageningen (1.5 km from experimental site) and ranged from 11 to 15 kg N ha−1 year−1 (driest vs. wettest year) at both sites (Boschloo & Stolk 1999a,b,c). Total N deposition is, however, mainly (50–60%) composed of dry NHx compounds, which is higher in the region around Wageningen than in the north of the Netherlands. Dry NHx deposition has been calculated from detailed emission data and comparison of measured and calculated NH3 concentrations in the atmosphere. The resulting total N deposition in 1997 amounted to 52 and 37 kg N ha−1 year−1 in the area of experimental and field sites, respectively (RIVM 1999).
Twenty peat monoliths (1.1 m diameter, c. 25 cm deep), including the intact bog vegetation, were cut in March 1996 when the surface layer of peat was frozen, and transferred together with the underlying wet highly decomposed peat to the experimental site in Wageningen. The loose peat was put in plastic containers (1.1 m diameter, 60 cm deep) that were buried to a depth of 50 cm and tap water was added to replace drained water, before the intact frozen monolith was placed on top. We expected this set-up to give realistic results, because bog ecosystems are generally isolated from the mineral substrate, with external input only from atmospheric deposition.
Monoliths were exposed to ambient or elevated (560 µmol mol−1) CO2 and a second set to ambient or increased (5 g N m−2 year−1 added) N deposition. Unfortunately, financial restrictions made it impossible to include combined CO2 and N treatments. The 20 peat monoliths were randomly assigned to one of the four treatments (i.e. five replicates). The elevated CO2 plots were located at least 6 m from the nearest control plot to prevent CO2 enrichment (Miglietta et al. 2001). To avoid edge effects, no measurements were taken in the outer 15 cm of each plot.
The CO2 treatments were established using MiniFACE technology. Each MiniFACE consisted of a ring (1.1 m in diameter) made of 5 cm diameter polyethylene tubing, fitted with 72 venting pipes, each 18 cm tall. Each pipe had two holes, at 6 and 12 cm above the moss surface. Ambient air was supplied by blowers located next to the MiniFACE rings at 50 cm above the surface. The CO2 concentration of the ambient air fluctuated diurnally with a mean of 360 µmol mol−1 in daytime. In the elevated treatment, pure CO2 was added to the airflow from the blowers and the supply rate adjusted automatically by a PC connected to mass flow controllers, based on measured wind speed and continuous monitoring by IR-gasanalyser of CO2 concentration 7.5 cm above the moss surface in the middle of the ring (see Miglietta et al. 2001). The CO2 concentration was within 20% of the preset target concentration of 560 µmol mol−1 for 97.5% of the time during 1996 and 1997 (Miglietta et al. 2001).
Nitrogen was added in the form of NH4NO3 dissolved in demineralized water. A total amount of 5 g N m−2 (corresponding with 50 kg N ha−1 year−1) was added each growing season to the high N plots in six applications (c. once every 3 weeks) by watering with 2 L of N solution (demineralized water in ambient treatment), simulating a rain event of 2 mm. Whenever possible, the N treatments were applied during rainy weather; otherwise a further 2 L of artificial rainwater (see below) was applied after N addition.
Water levels in the plots were allowed to fluctuate between 5 and 20 cm below moss surface. A hole in the container at 5 cm below moss surface allowed overflow, and artificial rainwater was added to raise the level by 10 cm whenever it fell to the minimum permitted. Natural levels of fluctuation were therefore allowed, but extreme events, such as flooding and drying out of the peat, were prevented. The artificial rainwater was an 8000-fold dilution of Garrels & Christ’s (1965) seawater solution, and is equivalent to very clean rainwater without N or P. Electrical conductivity was c. 11 µS cm−1 and pH was c. 5.8.
The CO2 treatments started in May 1996 and the first N was added in June 1996. In the winter months (December, January and February) the MiniFACE system was turned off because of the low rates of biological activity. The experiment ended with a final harvest in the beginning of September 1998, immediately after peak biomass for vascular plants, and just before major senescence and reallocation of nutrients.
Plant species composition and abundance were measured non-destructively using the point-quadrat method (Jonasson 1988) in a permanently marked subplot in each monolith (pq subplot), chosen to be representative of the type of vegetation. Plots were 25 cm × 37.5 cm with a 2.5 cm grid and each species hit by a sharpened knitting needle lowered to the top of the moss surface at each of 150 points was noted. The abundance of the moss species was measured at the beginning of the experiment, at the end of their growing season (October 1996 and 1997) and prior to final harvest. Vascular plant abundance was measured in June 1996, and at peak biomass (late August–early September) in all 3 years.
Sphagnum growth was measured non-destructively during the experiment by measuring height increment using the cranked wire method (Clymo 1970). In each plot, four stainless steel wires were placed c. 8 cm deep and anchored by plastic bristles. The length of the wire extending above the moss surface was measured monthly, except in winter. The mosses grew faster than expected and cranked wires were replaced by new ones when they became overgrown.
At the end of the experiment Sphagnum bulk density was determined so that height growth could be converted into biomass growth. A column, 8 cm in diameter and 10 cm deep, was cut with a sharp knife around each cranked wire. The upper 3 cm of this cylinder was then separated, vascular plant parts were removed and the capitulum (defined as top 1 cm) of each Sphagnum shoot was cut-off with scissors. The capitula were counted before all sections were dried at 70 °C for at least 48 h and weighed. Sphagnum net production in 1998 (g m−2 year−1) was calculated as height increment (mm) times bulk density of the 0–3 cm layer (g dm−3).
Vascular plant biomass
The total above-ground vegetation in the pq subplots was harvested in early September 1998, 1 week after the last point-quadrat recordings. All vascular plants were clipped off at the moss surface. The litter on top of the moss surface was also collected. The harvested species were sorted into current-year parts, other living parts and dead parts; the living parts were sorted into leaves, stems, flowering stems, flowers and berries. All plant parts were dried at 70 °C for at least 48 h and weighed. It was possible to separate current-year parts from older parts for all ericaceous species and thus to estimate annual above-ground production, since secondary stem growth in these species is very small (Backéus 1985). Drosera rotundifolia, E. angustifolium and other graminoid species had no older living parts, so the above-ground biomass was equal to the current-year above-ground production.
We used regressions between the final point-quadrat data and the above-ground vascular plant biomass at the final harvest for each species to reconstruct the development of the above-ground vascular plant biomass during the experiment. These regressions were highly significant (P < 0.001, R2 = 0.60–0.95, n = 13–20). Species were treated separately because the biomass per hit differed strongly between the species. Those with horizontal leaves, such as V. oxycoccus and D. rotundifolia, were ‘easily’ hit and had a lower biomass per hit (101 mg hit−1 and 32 mg hit−1, respectively) than a tall erect species such as E. angustifolium (324 mg hit−1).
The below-ground vascular plant biomass at final harvest was determined in three peat columns of 11 cm diameter and 30 cm deep. The peat columns were cut into 5 cm thick sections and all below-ground stems, rhizomes and roots were extracted and partitioned into five groups: (i) V. oxycoccus, (ii) other ericaceous species, (iii) E. angustifolium, (iv) other graminoid species, and (v) D. rotundifolia. All fractions were dried at 70 °C for at least 48 h and weighed.
Total C, N, P and K concentrations were measured in all current-year plant parts for S. magellanicum, V. oxycoccus, E. tetralix and E. angustifolium. P and K concentrations were measured in dry (70 °C), milled samples digested with sulphuric acid, salicylic acid, hydrogen peroxide and selenium and analysed by colourimetry for P concentrations and by flame atomic emission spectroscopy for K concentrations. C and N concentrations were determined in dry, ball milled samples by a CN analyser (Fisons Instruments, Rodano, MI, Italy). To determine water and ash content, subsamples were subsequently dried at 105 °C and ignited at 550 °C. The total C and N concentrations were expressed as a proportion of the organic matter content.
Data were tested for normality and equality of variance. Biomass data for separate species and root:shoot ratios were ln-transformed prior to analysis, as these data deviated from normality. CO2 and N effects were analysed separately using t-tests. Pearson’s correlation coefficients and other statistics were calculated using SPSS for Windows (8.0) (SPSS, Chicago, USA).
The moss layer was always dominated by S. magellanicum, which on average covered 95% of the moss surface. There were no significant changes in the cover of the moss species during the experiment (data not shown). Sphagnum height increment did not respond to either treatment in the first growing season, but was significantly increased at elevated CO2 in the second and third growing seasons (Fig. 2). Sphagnum height increment over three growing seasons ranged from 70 mm to 124 mm in elevated CO2 compared with 53–100 mm for ambient CO2. In contrast, increased N deposition resulted in a significant reduction in Sphagnum height increment but only in the third growing season. For three treatments, the mean height increment was highest in 1998, despite the fact that the final measurement was made in August, well before the end of the growing season, probably because the wet weather offered favourable conditions for Sphagnum growth.
Differences in height increment in Sphagnum between treatments were partly compensated by changes in bulk density (Table 1). We found a significant negative linear correlation between Sphagnum bulk density 0–3 cm (g dm−3) and height increment (r = −0.69, P < 0.01, n = 20). There was also a significant negative correlation between dry weight per capitulum and the number of capitula (r = −0.71, P < 0.01, n = 20). Thus, as numbers of capitula (per unit area) increased, their size decreased. It seems likely that the differences in bulk density were caused by differences in morphology: since we also observed that Sphagnum shoots had fewer branches per unit length of stem under elevated CO2 whereas the shoots in the high-N treatment were very compact.
Table 1. Growth characteristics of Sphagnum, determined after three growing seasons of CO2 or N treatments in 8 cm-diameter cores. Data are means ± SE, n = 5 plots, four cores per plot
Elevated CO2 resulted in a 17% higher biomass production in 1998 (Table 1), but, because of the relatively low bulk density, this effect was less than the increase of height increment and was not significant (P = 0.16). However, the increased height increment itself has important consequences for the ecosystem, as will be discussed later. Increased N deposition resulted in a significantly lower (−32%) Sphagnum biomass production, caused mainly by the reduced height increment.
Vascular plant biomass and species composition
Total vascular plant biomass
Above-ground vascular plant biomass increased in most plots during the experiment, particularly in the wet year (1998). Elevated atmospheric CO2 did not significantly affect above-ground biomass (Fig. 3, Table 2), and although N addition resulted in 54 and 41% larger above-ground peak-biomass in 1997 and 1998, respectively (Fig. 3), large between-plot variability made this effect only marginally significant (Table 2).
Table 2. Above-ground biomass, litter mass (including standing dead) and production (= current-year biomass) of vascular plants after three growing seasons of CO2 or N treatments. Data are means ± SE, n = 5 plots. Other species are: Calluna vulgaris, Empetrum nigrum, Rhynchospora alba, Eriophorum vaginatum, Andromeda polifolia, Molinia caerulea and Pinus sylvestris
The proportion of current-year parts in above-ground biomass was large (76%) and was greatest in the elevated CO2 treatment (86%). In V. oxycoccus, that grows at the moss surface, much of the older parts became overgrown by Sphagnum, particularly when moss growth was increased in the elevated CO2 treatment. There was a significant negative correlation between Sphagnum height increment and the above-ground biomass of older plant parts (r = −0.77, P < 0.01, n = 20). Because of the high proportion of current-year biomass, the 36% (P = 0.26) increase in above-ground vascular plant production (Table 2) brought about by elevated CO2 was greater than the increase in biomass (that also includes older parts). Increased N deposition boosted above-ground production by 44%, which was marginally statistically significant (P = 0.08).
Below-ground biomass, that accounted for 69% of the total, consisted of roots (26%), together with subterranean stems of ericaceous species and subterranean stems and rhizomes of graminoid species (Table 3). Neither treatment had significant effects (Table 3), although N increased total below-ground and root biomass (29% and 18% increases, respectively). This may be due to the large variation between replicates, as well as a lack of response by non-growing parts, such as the overgrown stems of ericaceous species.
Table 3. Below-ground biomass (below-ground stems, rhizomes and roots) and root biomass (g m−2) of vascular plants after three growing seasons of CO2 or N treatments. Data are means ± SE, n = 5 plots, three cores per plot. There are no significant treatment effects
Ambient N deposition
Increased N deposition
Below-ground biomass (g m−2)
607 ± 119
680 ± 114
526 ± 95
676 ± 115
271 ± 84
318 ± 78
308 ± 58
319 ± 65
Other ericoid species
248 ± 62
256 ± 56
172 ± 52
290 ± 68
86 ± 26
120 ± 15
48 ± 21
65 ± 11
Roots (g m−2)
145 ± 28
168 ± 29
147 ± 24
174 ± 27
81 ± 23
97 ± 25
102 ± 16
104 ± 21
Other ericoid species
46 ± 18
42 ± 8
37 ± 12
50 ± 10
18 ± 6
27 ± 5
12 ± 4
19 ± 3
There were no indications that plant species allocated more biomass to roots under elevated CO2 (Fig. 4). Only V. oxycoccus had a slightly higher root:shoot ratio under elevated CO2, but this may be because a greater proportion of the above-ground biomass had been overgrown by Sphagnum, resulting in a relatively small above-ground biomass. Increased N deposition resulted in lower root:shoot ratios for all three species groups, in accordance with the usual response of plants to N fertilization, but this effect was marginally significant only in V. oxycoccus.
Eleven species of vascular plants were recorded in the pq subplots, including V. oxycoccus, E. tetralix, E. angustifolium and D. rotundifolia, which were always present. Point-quadrat measurements did not show major changes in species composition in the CO2 experiment (Fig. 5). However, D. rotundifolia did not increase in cover in the elevated CO2 plots, although it increased in all other treatments. It seems likely that the low abundance of D. rotundifolia in the elevated CO2 plots (Table 2) was caused by the great height increment of Sphagnum. There was a significant negative correlation between Sphagnum height increment during the whole experiment and D. rotundifolia above-ground biomass (r = −0.57, P = 0.01, n = 20).
Species abundance responded more strongly to the N treatment (Fig. 5). The abundance of all species except E. angustifolium increased more in the high than in the ambient N treatment. For V. oxycoccus, the N treatment effect was significant (P = 0.01). Although cover of E. angustifolium increased during the first two growing seasons, its cover declined in 1998 (data not shown), resulting in no net change from the beginning (Fig. 5). The large increase in litter abundance in the high N treatment (Fig. 5) is the result of the great abundance of E. angustifolium in the previous growing season. Litter mass (including standing dead) was almost significantly increased by the N addition treatment (Table 2). In all treatments, litter was mainly from E. angustifolium. The point-quadrat recordings revealed that it was mainly E. angustifolium that caused the increased above-ground vascular plant biomass in the high N treatment in 1997, while V. oxycoccus responded most to N addition in 1998. Above-ground biomass of V. oxycoccus was increased by 46% at the final harvest in response to N fertilization (P = 0.075).
Effects of vascular plants on Sphagnum
The increased abundance of vascular plants under increased N deposition probably affected Sphagnum growth (Fig. 6). We found a significant negative correlation between vascular plant cover (expressed as total number of hits) and Sphagnum biomass growth in 1998 (r = −0.63, P < 0.01, n = 20). There was also a relation with Sphagnum height increment, though it was less significant (r = −0.43, P = 0.06, n = 20; not shown). Increased N deposition had a very significant effect on vascular plant cover (P < 0.01), more significant than on (living) biomass because litter also has a shading effect and is included in plant cover and because V. oxycoccus had a high cover at relatively little biomass. Both the cover of litter and V. oxycoccus increased significantly under high N deposition (Fig. 5).
Nutrient concentrations in sphagnum and vascular plants
Table 4 lists nutrient concentrations (N, P and K) and C:N ratios in the capitula (Sphagnum) and green leaves (vascular plants). In all four dominant species, N concentrations had fallen after three seasons’ growth under elevated CO2. As a consequence, the C:N ratio increased. Elevated CO2 reduced the N concentration by 11% on average and, over all species, this CO2 effect was highly significant (P = 0.01; anova: CO2 and species as factors, species as random factor). The same holds for the C:N ratio. The reduction in N concentration in plant tissues is one of the few consistent responses of plants to CO2 enrichment (Koch & Mooney 1996; Körner 1996; Cotrufo et al. 1998). The reduction in N concentration in all dominant species combined with the increases in biomass, although not significant may indicate that they all benefited from elevated CO2. N addition raised the N concentrations and reduced the C:N ratio in all dominant species except E. angustifolium. In S. magellanicum the N concentration increased by 44%. The N effect was highly significant (P < 0.01) for S. magellanicum and E. tetralix and marginally significant (P = 0.07) for V. oxycoccus. There were no CO2 or N treatment effects on P and K concentrations.
Table 4. Nutrient concentrations and C:N ratios in capitula (Sphagnum) and current-year leaves (vascular plant species) after three growing seasons of CO2 or N treatments. Data are means ± SE, n = 5 plots
There was no evidence that transferring the peat monoliths had had any irreversible effect on the vegetation. We observed that the species composition in the control plots hardly changed, except that V. oxycoccus increased in abundance, mainly in 1998, possibly due to the wet conditions in that year or the higher ambient N deposition in Wageningen than at the field site. Sphagnum growth might be expected to be greater in our experiment than in the field site because we prevented severe drought, but we did note that S. magellanicum hardly grew in height during the warmest months when water tables tended to be lower (data not shown). Height increment of S. magellanicum in the control plots was remarkably similar to that measured in another comparable small mire near our field site, in 1998 and 1999 (J. Limpens, unpublished data).
Three growing seasons should be sufficient to measure the effects of changed interactions between species in response to sudden changes in CO2 and N supply. The lack of response during the first growing season (Figs 2 & 3) agrees with findings from other CO2 experiments in low productivity systems (Warwick et al. 1998; Leadley et al. 1999). We suggest that slow-growing species that recycle a large proportion of earlier assimilated nutrients and carbon are less responsive to sudden changes in the external environment, because in the first year of the treatment the responses are buffered by the large stocks these species have built up in previous years.
The observed effect on Sphagnum growth is probably a direct response to elevated CO2 rather than an indirect response brought about by increased moisture availability, as observed in other ecosystems (Drake et al. 1996; Owensby et al. 1999). Water level relative to the moss surface (measured twice a week, unpublished data) was no higher in the elevated CO2 plots and extremely low water levels were prevented. However, we did find reduced water use by the bog vegetation under elevated CO2 (Heijmans 2000), that might be beneficial for Sphagnum growth under real field conditions, as Sphagnum species are very sensitive to changes in water content (Clymo & Hayward 1982; Schipperges & Rydin 1998).
We are not aware of other CO2-enrichment studies in Sphagnum-dominated mire vegetation. In glasshouse experiments, Jauhiainen et al. (1994, 1998) found CO2-induced increases in height increment and production for S.angustifolium, but not for S. fuscum. The dominant Sphagnum in our study, S. magellanicum, may have a greater diffusive resistance to CO2 uptake within its leaves than other Sphagnum species, because its photosynthetic cells are surrounded by water-filled hyaline cells (Proctor 1982), and, if CO2-limited, might react to CO2 enrichment.
The failure to find any statistically significant increase in above-ground vascular plant production, despite an increase of 36%, can be partly attributed to large between-plot variability. Furthermore, differences in Sphagnum height growth complicated the measurement of above-ground biomass. At the final harvest we observed that even parts of current-year stems, particularly of V. oxycoccus, were overgrown by Sphagnum. Once covered by the moss, leaves soon fall off and it is almost impossible to distinguish these current-year stems from older stems. Therefore we conclude that the above-ground vascular plant production, determined from above-ground current-year biomass, was underestimated, particularly in the elevated CO2 treatment.
The only CO2 response found in a long-term tundra experiment was an increased tiller density of Eriophorum vaginatum (Tissue & Oechel 1987; Oechel & Vourlitis 1996). Eriophorum vaginatum is also a common species in bog ecosystems, but was present in only a few plots in our experiment. Of all multiyear, CO2-enrichment studies we know, the tundra ecosystem study is probably most similar to our bog study. The lack of response at the ecosystem level in the tundra study was attributed to nutrient limitation. The growing conditions in our study were probably much more favourable than in the tundra, because of the milder Dutch climate, longer growing season and much higher ambient N deposition.
Despite the high ambient N deposition at our experimental site (c. 50 kg N ha−1 year−1) the vegetation still responded to the added N. Sphagnum growth was obviously not limited by ambient levels of N, given the lack of a stimulating effect of N addition on growth and the high N:P ratio (24 ± 1) in S. magellanicum capitula, consistent with findings from Aerts et al. (1992) who showed that growth of S. magellanicum from south Sweden with high N:P ratios (> 14) was P-limited. The mosses were nevertheless still able to capture a large portion of deposited N: under ambient N, c. 60% of an aerially deposited 15N label was found in the living Sphagnum layer 15 months after addition (Heijmans 2000). However, significantly less 15N was found in Sphagnum and significantly more in the vascular plants when N deposition was increased. In addition, inorganic N concentrations in soil water, sampled at 10 cm, 30 cm and 60 cm depth at the end of the experiment, were significantly increased in the high N treatment (unpublished data).
It is clear that Sphagnum was less able to retain the additionally deposited N, resulting in increased N availability for vascular plants. The shallow-rooted V. oxycoccus significantly increased cover accordingly (Fig. 5) and the lack of response in terms of biomass or cover of other species might be attributed to the P-limiting conditions, suggested by the high N:P ratios (27 ± 2 and 22 ± 1 for E. tetralix and E. angustifolium, respectively, in the N control treatment). According to Koerselman & Meuleman (1996) N:P ratios in above-ground plant material larger than 16 indicate P limitation. The N:P ratio in V. oxycoccus was significantly increased in the high N treatment (26 ± 2 vs. 19 ± 2 in the N control treatment), suggesting that any further expansion of the cover of V. oxycoccus at increased N deposition might be restricted by P limitation.
Our results agree with those of Lütke Twenhöven (1992a,b), who manipulated N deposition in similar bog vegetation in Germany and reported a similar significant increase in the standing crop of V. oxycoccus after 2 years of ammonium addition. In his study, E. angustifolium did not respond and S. magellanicum was unaffected or negatively affected by the N addition.
Indirect effects: interactions between sphagnum and vascular plants
The increased height increment of Sphagnum under elevated CO2 affected two species (D. rotundifolia and V. oxycoccus) that grow close to the moss surface. The abundance of D. rotundifolia increased in all treatments, but not in the elevated CO2 plots (Fig. 5) where its final biomass dropped to very low values (Table 2). It seems that the small D. rotundifolia plants could not keep pace with the rate of Sphagnum growth, which is confirmed by the inverse relation between D. rotundifolia biomass and Sphagnum height increment. Svensson (1995) observed that height growth of D. rotundifolia paralleled that of S. fuscum, but Svensson observed much less height growth of Sphagnum than in our study. The negative correlation between the above-ground biomass of old stems of V. oxycoccus and Sphagnum height increment means that V. oxycoccus became overgrown faster and had to invest more in above-ground growth to stay on top of the moss surface. Within the experimental period, however, this did not lead to a reduced above-ground biomass or production of this species.
Previous studies have suggested that shading by vascular plants may reduce Sphagnum growth (Clymo 1973; Hayward & Clymo 1983). The increased vascular plant cover (including litter) under increased N deposition would therefore be expected to reduce the Sphagnum growth by the third season (Fig. 6). It is also possible, however, that added N affected Sphagnum growth directly but it may have taken two growing seasons to saturate the Sphagnum layer with N, and thus to reduce Sphagnum growth in the third growing season. Laboratory experiments with monocultures of Sphagnum species, Press et al. (1986) and Jauhiainen et al. (1994, 1998) indeed showed a negative response to high N supply, and in a Canadian bog S. fuscum production decreased after the addition of 15 g N m−2 in one growing season whereas vascular plant production increased only slightly (Thormann & Bayley 1997).
In bog ecosystems, recalcitrant Sphagnum litter sequesters more carbon than the more-easily-decomposable litter of vascular plants (Clymo & Hayward 1982) whereas vascular plants contribute more to methane emission (Joabsson et al. 1999). Thus, changes in relative contributions of both species groups will have consequences for the exchange of greenhouse gases between bog and atmosphere. Species-specific responses to treatments, as well as interactions between peat mosses and vascular plants, can lead to such changes, as we observed even in a relatively short period of three growing seasons.
Little is known about the impact of CO2 itself, but our results show that elevated CO2 gives a competitive advantage to Sphagnum, resulting in increased peat accumulation (thus C sequestration) in the long term, that feeds back to the CO2 concentration in the atmosphere. However, we must be careful when drawing conclusions on carbon sequestration in the absence of information on the effects of elevated CO2 on decomposition, an important component of the carbon balance.
Increased N deposition may result in an opposite chain of effects. We observed reduced Sphagnum and increased vascular plant production in our experiment, that corresponds with results from other studies (Press et al. 1986; Lütke Twenhöven 1992a; Thormann & Bayley 1997). This change in the competitive balance between peat mosses and vascular plants to the benefit of vascular plants is expected to decrease sequestration of CO2. Furthermore, plant tissues containing more N might be decomposed faster (Coulson & Butterfield 1978). The N addition in our experiment resulted in higher N concentrations, not only in live tissues (Table 4), but also in brown Sphagnum, 5–10 cm below moss surface (unpublished data). We therefore believe that the increased abundance of vascular plants and the reduced Sphagnum growth brought about by high N deposition will reduce carbon sequestration and increase methane emission from bog ecosystems.
This study was part of the European Bog Ecosystem Research Initiative (BERI) project, financed by the European Union, Environment and Climate Research Programme (contract number ENV4-CT95-0028). BERI was approved by the GCTE (Global Change and Terrestrial Ecosystems) as a core activity. Staatsbosbeheer, the National Forest Service in the Netherlands, is thanked for permission to take large peat monoliths from their terrain. We thank Jan van Walsem for laboratory assistance, and Joyce Burrough, Lindsay Haddon and two anonymous referees for improving the text with helpful comments.
Received 20 June 2000 revision accepted 23 October 2000