Experimental response of peatland carbon dynamics to a water table fluctuationChristian Blodau* and Tim R. Moore
Department of Geography and Centre for Climate and Global Change Research, McGill University,805 Sherbrooke Street West, Montreal QC, H3A2K6, Canada
Received: 23 October 2001; revised manuscript accepted: 9 August 2002
Abstract. Water table fluctuations influence the carbonbalance of wetlands, but the effects are difficult to isolateand quantify in field investigations. Thus, we compared Cmineralization in a peatland mesocosm exposed to a wa-ter table fluctuation (from 5 to 67 cm beneath the surface)with that in mesocosms with a stable high water table (2to 6 cm depth) and with production rates obtained fromflask incubations. Net turnover rates were calculatedfrom concentration data by diffusive-advective mass-bal-ances. Under stable, high water table conditions, net pro-duction of CO2 (6.1 mmol m–2 d–1), CH4 (2.1 mmol m–2
d–1) and DOC (15.4 mmol m–2 d–1) were vertically strati-fied and production and fluxes equilibrated. Loweringand raising the water table from 5 to 67 cm resulted incomplex patterns of net CH4 and CO2 production. Re-sponse and equilibration times of processes upon
drainage and flooding ranged from days (aerobic CO2
production, CH4 oxidation, fluxes under unsaturated con-ditions) to months (CH4 production, fluxes under satu-rated conditions). Averaged over the water table fluctua-tion, net production of CH4 decreased to 0.36 mmol m–2
d–1 and that of CO2 increased to 140 mmol m–2 d–1. Phys-ical disturbance through the incubation of peat stronglyincreased production rates of CO2, CH4 and DOC com-pared to in situ, steady state rates. The decoupling of pro-duction and fluxes to the atmosphere under conditionswith variable water table depths potentially explains partof the frequently reported lack of correlation between en-vironmental variables and trace gas fluxes in field inves-tigations, and questions the applicability of predictions ofgas flux based on empirical relationships established un-der stable average conditions.
Northern peatlands play an important role in the globalcarbon (C) cycle, storing about 450 Gt C, about 30% ofthe global soil C (Gorham, 1991). Northern peatlandsfunction as long-term sinks for atmospheric carbon diox-ide (CO2) and sources of atmospheric methane (CH4), andhave significantly lowered atmospheric CO2 and raised
CH4 concentrations since the end of the last glaciation(Fung et al., 1991; Blunier et al., 1995). Carbon dioxideand CH4 exchange rates vary strongly between peatlandsand the atmosphere (Moore et al., 1998). The processesthat produce and consume trace gases in peatlands andthe factors that control the exchange rates are, therefore,currently intensely investigated (Blodau, 2002).
The important controls on organic matter decomposi-tion and C mineralization are soil temperature, plantcommunity structure, position of redox boundaries asso-ciated with the water table, and the chemical compositionof plant tissues and peat (Bubier et al., 1993, 1995; Whit-ing and Chanton, 1993; Yavitt et al., 1997). Empiricalmodels including these controls can explain much of the
* Corresponding author present address: Limnological ResearchStation and Department of Hydrology, University of Bayreuth, D-95440 Bayreuth, Germany; phone: + 49 921 552223; fax: + 49 921 552366; e-mail: [email protected] on Web: March 19, 2003
variation in seasonal CH4 exchange rates and reflect av-erage differences in environmental and ecological vari-ables within and among peatlands (Moore et al., 1998).At smaller scales, much less of the variation of CH4 ex-change rates can be explained by environmental and eco-logical variables (e.g., Moore et al., 1994; Shannon andWhite, 1994; Kettunen et al., 1996; Bellisario et al.,1999). Changes in the frequency and intensity of weather-related events at such scales, e.g., local water table low-ering during droughts, may be a primary consequence ofclimate change. This insight raises concern about the accuracy of scenarios that are based on average changesin environmental and ecological variables, rather thandisturbances (Moore et al., 1998). The response of other processes in the C cycle, such as mineralization oforganic C and release of dissolved organic carbon(DOC), also have not been investigated in studies using experiments with intact soils simulating distur-bance.
In order to understand the impact of short-term dis-turbances, the response of sources, sinks and transport oftrace gases and DOC in peatlands needs to be investi-gated. This study is based on a mass balance approach toestimate fluxes and turnover rates from concentrationdata in mesocosms, comparing steady-state and dynamicexperiments. We compare the response of below-groundC turnover to a water table fluctuation with C turnover atsteady state. The net effect of this fluctuation is quanti-fied, and the rates in intact cores are compared to ratesfrom disturbed samples in flask incubations. Based onthese comparisons, some implications of disturbances onC biogeochemistry and C balance of peatlands are dis-cussed.
Methods
SiteThe Mer Bleue site near Ottawa, eastern Ontario, Canada,is an open, slightly domed, oligotrophic peatland domi-nated by mosses (e.g., Sphagnum capillifolium, Sphag-num angustifolium, Sphagnum magellanicum and Poly-trichum strictum) and shrubs (e.g., Ledum groenland-icum, Chamaedaphne calyculata, Kalmia angustifolia,Vaccinium myrtilloides). The peat was about 4 m deep atthe sampling site. The decomposition degree on the vonPost scale increased from 2 to 3 in the upper 10 cm toabout 6 to 10 at a depth of 70 cm.
Definitions and conventionsThe term mesocosm or core is used for the subsequentlydescribed experimental setting. The term incubation isused with respect to flask experiments. Rates in meso-cosms are referred to as in situ, whereas rates in flask in-
cubations are referred to as potential turnover rates. CO2
is used synonymously with dissolved inorganic carbon(DIC) because dissolved CO2 was, at pH values of 3–5,the predominant carbonate species. The term water tablerefers to the depth at which mainly water was extracted bysuction with a syringe. Losses of CO2 and CH4 from thepeat to the atmosphere or due to drainage have been givena positive sign. All indicators of variability in the text andfigures are standard deviations. All production valuesrepresent net rates. Rates and pools of dissolved CO2,CH4 and DOC are given in molar units. All rates werebased on dry weights or volume of peat.
Mesocosm experimentsPeat cores, 20 cm in diameter and 75 cm long, were col-lected in PVC tubes from hollows in late summer to fall,and had a drainage mesh and cap attached at the bottom.The vegetation, consisting primarily of Sphagnummosses, minor numbers of Polytrichum and small speci-mens of Ledum groenlandicum, Chamaedaphne calycu-lata, Kalmia angustifolia, was left intact. Pore water sam-plers (Bev-Line IV, Cole Parmer, 7 mm outer diameter, 3mm inner diameter, ca. 30 perforations per sampler) werehorizontally inserted at 2-cm intervals. In these cores,hereby denoted “steady-state cores”, the water table wasadjusted with distilled water to 2–6 cm below the mosscover and held constant. The temperature was initially22°C during the day and 8°C during the night. After day50 the temperature was lowered to 12°C in the day and8°C at night to slow down the abundant Sphagnum growthand C mineralization rates. Relative humidity was kept at70%. Light intensity was adjusted to 250 µmol m–2 s–1.Solution was added with a sprinkler for 5 to 6 days aweek, and water manually retrieved at 2 to 3 mm d–1 fromthe base of the column.
The inflowing solute contained two concentration levels of the ions H3O+ (92/358 µmol L–1), SO4
2– (26/104 µmol L–1), NO3
– (40/120 µmol L–1), and NH4+ (40/120
µmol L–1) and one concentration level of the ions Ca2+
(5 µmol L–1), and Cl– (150–265 µmol L–1). These concen-tration levels were a consequence of the mesocosms being part of a N and S deposition study. The analysesshowed that the concentrations of DIC, dissolved CH4 andDOC, and the CO2 and CH4 fluxes from the mesocosmswere unaffected by the different treatments (Blodau,2002), thus data from four mesocosms were averaged forobtaining average steady state turnover and gas exchangedata. Measurements started after an equilibration periodof 60 days, and were subsequently run for about 220 dayson an approximately monthly basis. The total moss bio-mass in the mesocosms was between 200 and 600 g m–2
(dry weight). This value is an upper estimate since theSphagnum were pulled from the mesocosms and all lower
48 Christian Blodau and Tim R. Moore Response of C dynamics to a water table fluctuation
parts of living Sphagnum stems were included in the mea-surement.
To investigate C turnover during water table fluctua-tions, another mesocosm was prepared that allowed thesampling of pore water and pore gas from a multitude oflocations, whose concentration data were averaged foreach depth increment. This core was initially kept withthe water table at a depth of 66 cm for 4 months, thenflooded for 103 days (water table at 3 to 5 cm below plantcover and 1 to 2 cm above first sampling depth), thendrained for 94 days and flooded again for 32 days. At day150, the temperature was raised to 20°C. The duration ofthe first cycle of unsaturated-saturated-unsaturated con-ditions was chosen to represent the adjustment of the peatsoil to a large water table fluctuation as might occur inpeatlands on a seasonal basis. The duration of the secondflooding was chosen to examine whether patterns in CH4
production observed initially after the first flooding werereplicable at the higher temperature.
Under saturated conditions, we only replaced evapo-rative losses in this mesocosm to keep the water tableconstant with the above described solution. CO2 and CH4 concentrations were determined on 23 occasions byextracting pore water or gas with a syringe. We also determined DOC and O2 on 4 and 8 occasions, respec-tively.
Peat core-atmosphere exchange rates of CO2 and CH4
were calculated from linear regression of 4 to 6 head-space concentrations at 8-minute intervals using the sta-tic chamber technique (steady state mesocosms). Diffu-sive fluxes were calculated from concentration gradientsof the uppermost sampling ports and the atmosphere. ThePVC chambers were 20 cm tall and were manually mixedwith a 60 mL syringe before sampling. Regressions withR2 < 0.75 were eliminated, as were obvious artifacts (e.g.,exponential concentration increase) even when R2 > 0.75.Mosses in the mesocosms grew at a rate of ca. 5 mm permonth. The total moss biomass was ca. 200 g m–2. The pH of the pore water was between 3.4 to 3.8 at the surfaceand 4.5 at 66 cm and changed little over time.
Incubation experimentsIncubations with peat were carried out after the end of themesocosm experiments at room temperature (20–22°C).Anaerobic potential production rates of CO2, CH4, andDOC were determined at four depth increments (8–12,22–26, 32–36, and 58–62 cm). Peat cores were extruded,dissected and sampled under N2 in a glove chamber. Thepeat (50–100 g wet weight, 7.2 ± 3.2 g (st. dev.) dryweight) was placed in 125–250 mL, rubber-stoppered Er-lenmeyer flasks and fully immersed in deaerated water.No gas headspace was left in the flask except for a smallbubble used for occasional mixing of the water. Theflasks were not shaken, but occasionally turned over to
avoid chemical stratification within the flasks. The waterwas sampled through suction samplers inserted throughthe rubber stoppers. Within 7 to 9 days, 10 mL of waterwere extracted on 5 to 6 occasions and CO2, CH4, andDOC determined as described below. The water was re-placed with deaerated solutions. Rates were determinedby linear regression of concentration over time correctedfor the replacement of water sampled from the flasks withthe deaerated solution.
In the mesocosm exposed to water table fluctuation,the peat volumes used for incubations had to be smallerbecause samples were taken in more spatial detail. Incu-bations were thus carried out in 12 mL Vacutainers withpeat samples (2–3 g wet weight) from all sampling portsof the core after the end of the experiment, which was 36days after the second flooding. Samples were repeatedlyevacuated and flushed with N2, 2 mL of a solution wereadded and the headspace was adjusted to 500–800 ppmCH4 to determine potential production or consumption ofCH4. In these incubations potential CH4 production rateswere determined from volume-corrected linear regres-sions of gas headspace concentration changes over a pe-riod of 11 days, and CO2 production rates over a period of8 days after a 30-day equilibration period and flushing ofthe Vacutainer headspace with N2. DOC concentrationscould not be determined due to the small sample volume.CH4 loss rate of spiked controls was smaller than the an-alytical error, and experimental variability of CH4 pro-duction rates within treatments was small (172 ± 17 and400 ± 6 nmol g d–1, st. dev., n = 4).
Pore water and pore air analysesDOC was determined after filtration of extracted porewater with a syringe micro-filter (0.45 mm, nylon) on aShimadzu 5050 TOC analyzer. Dissolved inorganic car-bon (DIC), and CH4 was determined on a Shimadzu Mini2 gas chromatograph with methanizer (CO2) in the gasphase of 1.8 mL vials after addition of 20 µL of 4M HCland 0.5 mL sample. Losses of CO2 and CH4 from the GCvials were corrected with exponential loss functions (n =16, R2 = 0.99; concentration of CO2: C = C0 ¥ 10(–0.0264t)
and concentration of CH4: C = C0 ¥ 10(–0.00635 t) with t in h.The original dissolved concentration was reconstructedusing the head space concentrations, the volumes ofheadspace and water phase and Henry’s law (KH = 10–1.5
(mol L–1 atm–1) for CO2 and KH = 2 ¥ 10–3 (mol L–1 atm–1)for CH4). In the unsaturated zone, CO2 and CH4 were di-rectly determined on gas samples extracted from the sam-pling ports with syringes. O2 was determined ampero-metrically with a low-current electrode (Orion) and H2Sand pH potentiometrically (AgS/glass-electrode, Wa-tertest) on 0.5 to 3 mL of sample with a conventional me-ter (Orion). O2 contamination due to the sampling proce-dure was ca. 0.5 mg L–1.
Aquat. Sci. Vol. 65, 2003 Research Article 49
In situ turnover calculationsThe water movement in the peat cores could be suffi-ciently described by advective-diffusive transport (Blo-dau and Moore, 2002a). Variability in concentration pro-files through the use of suction samplers was minimizedby averaging concentration data within 6 cm or 12 cmdepth segments. For each segment, vertical advective-dif-fusive mass balances were calculated using temperature-corrected diffusion coefficients for dissolved CO2, CH4
and O2 (Lerman, 1979). Net turnover is then given byequation (1)
R = DSA/DT – (Din DCA/Dx)in + (Dout DCA/Dx)out
– v (CA, in CA, out ) (1)
in which R is the turnover rate (nmol cm–3 d–1), CA is theconcentration of species A, DSA/DT is the change in stor-age of species A in a segment (nmol cm–3 d–1), D is thewhole peat diffusion coefficient (cm2 d–1), DCA/Dx is theconcentration gradient of species A (nmol cm–4) and v isthe advection rate (cm d–1); in: at inflow; out: at outflow ofsegment.
The effect of bulk density on diffusion coefficientswas taken into account by using separately determinedbulk density profiles that were fitted against a powerfunction (Bd = 0.0107 x0.567 with Bd = bulk density (g cm–3) and x = depth (cm); R2 = 0.79; Blodau andMoore, 2002a).
Based on preliminary calculations with tabulated dif-fusion coefficients (Cornel et al., 1985), diffusion wasneglected for the calculation of DOC turnover. The con-centration profiles in the steady-state cores were treatedas a series of steady states and rates calculated from ad-vective-diffusive mass balances. Diffusive exchange withthe atmosphere was calculated using Fick’s 1st law. Forthis calculation, it was assumed that gaseous concentra-tions 1–2 cm above the water table were in equilibriumwith the dissolved concentration at the water table. In theexperiment with variable water table, turnover rates werecalculated from diffusion rates between sampling pointsand from changes in storage.
To obtain turnover rates under unsaturated conditions,we separately determined water content and bulk densityof the depth segments under identical conditions and fitted the data to linear (water content, R2 = 0.97) andsquare functions (bulk density, R2 = 0.79). Equation (2)(Millington and Quirk, 1961; Jin and Jury, 1996), andgaseous, temperature-corrected diffusion coefficients(Lerman, 1979) were used to estimate unsaturated diffu-sion coefficients in the peat
a (a) = a2 b –2/3 (2)
in which a is the diffusion coefficient correction factor( ), a is the volumetric air content ( ), and b is the soilporosity ( ).
Ambient atmospheric concentrations, diffusive massbalances and changes in storage were used to calculateturnover and diffusive exchange rates of CO2 and CH4
with the atmosphere. A few sampling locations with ele-vated CH4 concentrations, indicating anaerobic pockets,were discarded from the unsaturated rate calculations. Inthese locations water contents were obviously muchlarger than average, and could not be used for the calcu-lation of rates according to equation (2). Since these lo-cations were excluded, the calculated net CH4 consump-tion is an upper estimate.
Production rates were visualized using isoline plots in WinSurf, release 5.00, using linear interpolation. Forthe analysis of the isoline plot data in this study, it should be kept in mind that only spatially broad or temporally consistent trends in the data can be inter-preted.
Results
Constant water tableIn this treatment, the boundary between oxic and anoxicconditions was located 1 to 3 cm below the water table(data not shown). DOC was the predominant form of car-bon in the pore water with concentrations between 2000and 8000 µmol L–1 (Fig. 1). DIC concentrations rangedfrom 100–600 µmol L–1 at the water table to 3000 µmolL–1 at greater depths (Fig. 1). Methane concentrationsgenerally ranged from 10–190 µmol L–1 just below thewater table to 500–800 µmol L–1 at depths of 40 to 70 cm(Fig. 1).
Dissolved C pools, fluxes and turnover changed onlyslowly (Figs. 2A, 3A and 4A). The DIC reservoir rangedfrom 1000 to 2250 mmol m–2, the CH4 pool from 260 to 500 mmol m–2 and the DOC reservoir from 3300 to4800 mmol m–2. Average turnover times were 600 to 700 d, 250 to 490 d, and about 300 d for DIC, CH4 andDOC, respectively. The export of C as CH4, CO2 and DIC,and DOC was, within the errors of the approach, similarto the net production within the peat (Table 1).
Within the cores, turnover rates of DIC, DOC andCH4 were vertically stratified (Figs. 2B, 3B and 4B).Maximum net production of DIC, CH4 , and DOC oc-curred either close to the surface or between a depth of 15 to 30 cm, with weak net production or uptakedeeper. Among depths, in situ CH4 , DIC, and DOC net production rates were correlated (R2 = 0.57–0.72, a < 0.05).
Variable water tableFlooding of the mesocosm resulted in depletion of dis-solved O2 to background levels (about 0.5 mg L–1) withina few days. The pools of DIC (Fig. 5A), CH4 (Fig. 6A)
50 Christian Blodau and Tim R. Moore Response of C dynamics to a water table fluctuation
and DOC rapidly increased. DIC concentrations in-creased throughout the core to 1500 to 2000 µmol L–1
within the first 2 weeks after flooding, and mostly con-tinued increasing at a slower rate. The maximal DIC con-centrations reached 3000–5000 µmol L–1 on day 103 afterthe first flooding, just before drainage (Fig. 1). By then,CH4 concentrations had reached 40 to 160 µmol L–1 (Fig.1). DOC concentrations were fairly uniformly distributedand reached 5000–6800 µmol L–1 just before drainage(Fig. 1).
The diffusive flux of CO2 and CH4 from the core in-creased but mostly remained much smaller than the pro-duction rates (Figs. 5A, 6B). Results for the secondflooding, at 20–22°C, were similar, but the time lag in theresponse of CH4 production was shorter (Fig. 6B). Dur-ing this second flooding (around day 220), the CO2 fluxfrom the mesocosm exceeded production for some time(Fig. 5A). DOC net production was initially large anddropped to an average of 23 nmol cm–3 d–1 between day32 and 103, which was a similar value to that determinedin the steady state cores.
Aquat. Sci. Vol. 65, 2003 Research Article 51
Figure 1. In situ average concentrations of DIC, CH4, and DOC concentrations in steady state cores 32 days and 103 days (just prior todrainage) after flooding. Bars indicate standard deviations among cores (steady state cores, n = 4) and among depth profiles within the coreused for the water table fluctuation experiment (n = 7), respectively.
Table 1. Aggregated net in situ production and potential production (mmol m–2 d–1) and fluxes (mmol m–2 d–1) during steady state and averaged over flooded and drained periods. Mean and standard deviation are displayed.
After drainage (day 103), the calculated in situturnover rates for CO2 initially reached up to 2500 nmolcm–3 d–1 in the surface layers and then decreased to < 1000nmol cm–3 d–1 on day 137 (Fig. 5B). Increasing the tem-perature by 10–12°C resulted in an increase in the maxi-mal rates to >3000 nmol cm–3 d–1. At greater depths(36–66 cm) rates were much slower (<100 nmol cm–3 d–1)than in the upper layers.
52 Christian Blodau and Tim R. Moore Response of C dynamics to a water table fluctuation
Figure 2. DIC/CO2 dynamics in cores with a stable water table. Panel A shows measured pool sizes, exchange rates due to photosynthe-sis, plant respiration and C mineralization, and calculated dissolved export, turnover and diffusive emission to the atmosphere on an areabasis (mmol m–2 d–1). Averages of 4 mesocosms are displayed and error bars represent standard deviations. At day 50 the temperature waschanged from 22/8 °C to 12/8 °C. Panel B shows the corresponding depth-time distribution of turnover rates, on a volumetric basis (nmolcm–3 d–1), in the cores. Shaded areas indicate consumption in this and following figures.
After some equilibration time methane was consumedafter drainage in most of the peat core at 0 to 1.8 nmolcm–3 d–1, but CH4 concentrations were elevated in somelocations at depths of 36 to 60 centimeters indicating con-tinued production. Under unsaturated conditions afterdrainage, dissolved pools were very small, turnover rateswere usually less than a day, and fluxes and productionrates of CO2 and CH4 were in equilibrium (Figs. 5A, 6Aand B).
A
B
Anaerobic potential CO2 production in incubationswith peat from the core with fluctuating water tablesranged from 890 ± 820 nmol g–1 d–1 at 54 cm depth to5600 ± 2000 nmol g–1 d–1 at 18 cm depth. Anaerobic po-tential CH4 production peaked at 30 cm depth at 550 ±210 nmol g–1 d–1 (Fig. 7). Below a depth of 42 cm, rateswere near 0 nmol g–1 d–1 (Fig. 7).
Aquat. Sci. Vol. 65, 2003 Research Article 53
Figure 3. CH4 dynamics in cores with a stable water table. Panel A shows measured pool sizes, exchange rates and calculated dissolvedexport, turnover and diffusive emission to the atmosphere. Averages of 4 mesocosms are displayed and error bars represent standard devi-ations. At day 50 the temperature was changed from 22/8°C to 12/8°C. Panel B shows the corresponding depth-time distribution ofturnover rates (nmol cm–3 d–1) in the cores.
Potential rate measurementsAnaerobic potential CO2 production rates in incubationswith peat from the steady state cores ranged from 2700 ±2400 nmol g–1 d–1 at 60 cm depth to 11300 ± 5800 nmolg–1 d–1 at 10 cm depth (Fig. 7). Potential CH4 productionrates peaked on average at 24 cm and at 720 ± 320 nmolg–1 d–1. Potential DOC production rates ranged from 5300± 1900 nmol g–1 d–1 at 60 cm depth to 23000 ± 13400nmol g–1 d–1 at a depth of 10 cm (Fig. 8).
Discussion
Rates after disturbance and during steady stateIt has been noted earlier that disturbance, either physical,e.g. by stirring in flask incubations, or chemical bychanging the water table levels or redox state, has the po-tential to change CH4 and CO2 production rates in water-logged peat and lake sediments (Kelly and Chynoweth,1980; Hall et al., 1996; Aerts and Ludwig, 1997; Öquistand Sundh, 1998; Kettunen et al., 1999). This potential isof importance in several respects. Methodologically, itaffects the comparability of rates that have been obtainedunder different experimental conditions and questions
whether rates can be extrapolated to different scenarios.This insight is of importance when production rates areextrapolated from the flask to the field scale, used inmathematical or statistical models of trace gas dynamics,or when data from a multitude of studies are synthesized.Also, it is important because it indicates that disturbancewill alter in situ rates in the field compared to conditionsthat are more stable, which would have implications forthe C balance of peatlands. To assess the role of physicaland water table disturbances on CO2 , CH4 and DOC dynamics, comparisons must be made between rates in intact soils under steady state, after water table distur-bance, and in incubations with peat, extracted under
54 Christian Blodau and Tim R. Moore Response of C dynamics to a water table fluctuation
Figure 4. DOC dynamics in cores with a stable water table. Panel A shows measured pool sizes, calculated advective export, and turnoverrates. Averages of 4 mesocosms are displayed and error bars represent standard deviations. At day 50 the temperature was changed from22/8°C to 12/8°C. Panel B shows the corresponding depth-time distribution of turnover rates (nmol cm–3 d–1) in the cores.
steady state conditions and after water table distur-bance.
The results from this study showed that rates of CO2,CH4, and DOC production in peat were strongly elevatedusing incubations, compared to in situ production rates atsteady state. In Figure 9 this result is indicated by pro-duction rate quotients (incubation/steady state) >1.Whether peat was extracted from steady state conditions,or 32 days after the water table disturbance, was of littleimportance for the magnitude of this increase in rates inthe incubations (Fig. 9). The increase in production rateswas, however, larger at greater depths (Fig. 9).
The results thus suggest first of all that in peat, com-mon incubation techniques over-estimate in situ produc-tion rates of CO2, CH4 and DOC at greater depths. This isin agreement with an earlier finding that CO2 productionrates might increase up to a factor 200 when stabilized,usually anaerobic peat from larger depths is incubated inflasks under aerobic conditions (Hogg, 1993). Due to thedepth dependency of this incubation effect the averagerate quotients (incubation/in situ steady state) were expo-nentially related to depth at moderate to large explainedvariance (0.54 < R2 < 0.96, Table 2). Such relationshipsmight thus allow for the estimate of in situ rates from in-
Aquat. Sci. Vol. 65, 2003 Research Article 55
Figure 5. The effect of water table fluctuation on DIC/CO2 dynamics. Panel A shows measured pools sizes, calculated net turnover ratesof CO2 and DIC and exchange rates of CO2 with the atmosphere. Error bars represent standard deviations of seven parallel depth profileswithin one mesocosm. At day 150 the temperature was changed from 12/8°C to 20–22°C. Panel B shows the corresponding depth-timedistribution of turnover rates (nmol cm–3 d–1) in the cores. The full isolines follow each other at intervals of 25 nmol cm–3 d–1, the shortdashed lines at intervals of 500 nmol cm–3 d–1.
cubation measurements when calibrated to a particularset of in situ conditions.
Flooding of the mesocosms had an effect on CO2,CH4, and DOC production rates that was time dependent.Averaged over the first 6 days after flooding, the in situCO2 production rate was similar to the rates in incuba-tions (Figs. 7 and 9). Particularly at larger depths, these
rates were much larger than at in situ steady state. Over-all, flooding an unsaturated peat core increased the aver-age anaerobic CO2 production rate by a factor of 2.5(22/8°C) and 5 (12/8°C, Table 1).
The effects of flooding on DOC production rates wereprobably similar. DOC production rates were higherwhen compared to steady state and increased more at
56 Christian Blodau and Tim R. Moore Response of C dynamics to a water table fluctuation
Figure 6. Effect of water table fluctuation on CH4 dynamics. Panel A shows measured pool sizes, panel B calculated turnover of CH4 andH2S and exchange rates of CH4 with the atmosphere. Averages of 7 depth profiles within one core are displayed. Error bars represent stan-dard deviations. At day 150 the temperature was changed from 12/8°C to 20–22°C. Panel C shows the corresponding depth-time distrib-ution of turnover rates (nmol cm–3 d–1) in the cores. The full isolines follow at intervals of 0.25 nmol cm–3 d–1, and the dashed lines isolinesof –2, –10 and –20 nmol cm–3 d–1.
greater depths (Fig. 9). Methane production was, in con-trast, much slower after flooding and only slowly in-creased later (Figs. 6 and 9). The unsaturated period priorto flooding decreased the CH4 production by ca. 42%, andincreased DOC release roughly by a factor 2, comparedto a core that had been kept flooded continually (Table 2).
Many factors might have contributed to these results.Possible causes include (i) enhanced substrate release dueto redox-induced chemical breakdown (Aller, 1994), (ii)enhanced recycling of biomass after the redox and waterpotential disturbance (Bottner, 1985; Kieft et al., 1987;Clein and Schimel, 1994), and (iii) low concentration lev-
Aquat. Sci. Vol. 65, 2003 Research Article 57
Figure 7. Average CO2 and CH4 production rates in incubations with peat from the steady state cores (“incubation steady state”), and fromthe core with fluctuating water table 32 days after the second flooding (“incubation transient state”). Maximal rates obtained after flood-ing (“in situ transient state”) and during steady state (“in situ, steady state”) are also displayed.
Figure 8. Average DOC production rates in incubations with peat from the steady state cores (“incubation steady state”), and in situ 0–32days and 32–55 days after the first flooding (“in situ transient state”). Rates during steady state (“in situ, steady state”) are also displayed.
els of mineralization products just after flooding and inincubations in comparison to the steady state. Activationof exo-enzymes after short term exposure to O2 also canstrongly increase C mineralization rates (C. Freeman,personal communication). Methanogenesis was only im-paired after long-term aeration, but not in the incuba-tions, when compared to in situ at steady state. This resultis in agreement with previous studies documenting thatprolonged exposure to O2 reduces potential productionrates, albeit not the viability of methanogens (Fetzer et
al., 1993). It has also previously been noted that the initi-ation of CH4 production in soils is retarded when soil isstored under aerated, but moist conditions (Mayer andConrad, 1990).
Drainage resulted in in situ DIC production being 5 to6 times larger under unsaturated and aerobic conditionsthan the average rate under flooded and mostly anaerobicconditions. In this study, this ratio was poorly constrainedbecause of the large errors associated with the calculationof in situ production rates from concentration profiles ofCO2 in unsaturated peat. It is, however, fairly reasonablebased on results from several other incubation and col-umn studies in which aerobic-anaerobic ratios typicallyranged from 1.2 to 6 (Moore and Knowles, 1989; Mooreand Dalva, 1993; Updegraff et al., 1996; Yavitt et al.,1997; Aerts and Ludwig, 1997; Öquist and Sundh, 1998).
Despite the large variation in rates between incuba-tions, transient, and steady state in situ production rates,the incubation potential and in situ rates fell within therange reported in or recalculated from the literature (Fig.10). For DIC, potential rates (incubations) tended to be inthe upper range of literature data, whereas the median ofin situ (mesocosm/columns) rates was lower than the re-ported median of potential and column rates from the lit-erature. Methane emission from the cores was fairly closeto the median of reported CH4 emissions in field studies(Fig. 10). Based on our findings, part of the enormousvariability in reported rates may have been caused by dif-
58 Christian Blodau and Tim R. Moore Response of C dynamics to a water table fluctuation
Figure 9. Quotient between in situ production at steady state and (I) rates after 6 and 55 days after flooding (“in situ transient”), (II) ratesin incubations with peat from the steady state cores (“incubation steady state”), and (III) from the core with fluctuating water table 32 daysafter the second flooding (“incubation transient state”).
Table 2. Regression of rate quotients against depth according tolog (rate/in situ steady state rate) = a x depth (cm) + b, „: level ofsignificance and n: number of rate quotients entering the regres-sion.
Log (rate quotient) a b R2 a n
CO2
Transient, 6 d 0.0199 0.66 0.64 <0.05 10Transient, 55 d 0.0246 –0.65 0.69 <0.05 10Incubation steady state 0.0216 0.63 0.77 0.12 4Incubation transient 0.0181 0.69 0.59 <0.05 10
CH4
Transient, 6 d 0.0248 –1.87 0.52 <0.05 9Transient, 55 d 0.0306 –1.21 0.79 <0.05 8Transient, 103 d 0.198 –0.88 0.44 0.07 8Incubation steady state 0.046 0.41 0.95 <0.05 4Incubation transient 0.041 –0.50 0.55 0.09 8
ferences in the experimental conditions under which rateswere determined, and by the history of the peat prior tothe investigation. Under field conditions in peatlands,CO2 and CH4 concentrations in the peat are mostly rela-tively uniform at levels of 1 to 10 mmol L–1 (CO2) and 0.1to 1 mmol L–1 (CH4) (Nilsson and Bohlin, 1993). The dis-solved concentrations determined in the mesocosms ofthis study were more or less in the middle of this range,with associated production rates being much lower thanin the incubations. In situ field rates of CO2 and CH4 pro-duction in the saturated peat, which are, as in the meso-cosms, reflected by the concentration gradients with theatmosphere, are thus probably more uniform than ratesreported using incubations. They should also be lowerthan the potential rates that have been obtained with com-mon flask incubation techniques.
Response of processesProduction, consumption, storage, and transport of gasesin the peat equilibrated only slowly after the water tableincrease. Since the adjustment of water tables and tem-perature was instantaneous in this study, all differences inprocess rates occurred, within the experimental error,while environmental conditions were held constant. It is
therefore straightforward to identify the time scales nec-essary for the adjustment of processes to changed envi-ronmental conditions. This identification is difficult toaccomplish in field investigations, but important for un-derstanding the relationship between environmental vari-ables and trace gas fluxes.
A number of processes adjusted quickly to thechanged conditions. The depletion of O2 after floodingwas very rapid, attributed to the large C mineralizationrate under oxic conditions, the small reservoir size of O2
(ca. 350 µmol L–1) in equilibrium with the atmosphere,and the large diffusive resistance under saturated condi-tions. With the high initial net CO2 production rates afterflooding (100–300 nmol cm–3 d–1, day 7 of first flooding;day 203–206 during second flooding; Fig. 5B), theturnover time of O2 was only on the order of one to threedays. This restricted oxic C mineralization to the upperfew centimeters of the saturated peat profile within a fewdays after flooding.
Similarly, CH4 oxidation in the unsaturated peatreached a steady state a few days after drainage (Fig. 6C),and was attributed to the large aerobic CH4 oxidation po-tentials that were present, even under anaerobic condi-tions (Basiliko, unpublished data). This finding agreeswith other studies (Moore et al., 1994) and results from
Aquat. Sci. Vol. 65, 2003 Research Article 59
Figure 10. Summary of steady state in situ rates, potential rates, and reported rates of DIC/CO2 and CH4 turnover. The figure displays me-dian, quartiles, 90% quartiles and extreme values. Reported potential rates were recalculated for a 40 cm profile (bulk density 0.08 g cm–3)and molar units. “Wt” stands for water table. Literature values were derived from: Aerts and Ludwig (1997), Aerts and Toet (1997), Bel-lisario et al. (1999), Bubier et al. (1993), Bubier (1995), Freeman et al. (1993), Funk et al. (1994), Hall et al. (1996), Hogg (1993), Libliket al. (1997), Moore and Dalva (1993), Moore and Knowles (1989), Moore et al. (1994), Nykänen et al. (1998), Schimel (1995), Valentineet al. (1994), van den Pol-van Dasselaar and Oenema (1999), Yavitt et al. (1990) and Yavitt et al. (1997).
the capacity of methanotrophs to survive anaerobiosis(Yavitt et al., 1990; Edwards et al., 1998).
DOC production also increased quickly after the firstflooding but fell back sharply afterward (Fig. 8). In theflooding experiment, the increase in the DOC concentra-tions was largely completed after about 20 to 30 days andoccurred very rapidly in the physically disturbed incuba-tion samples. This response is probably partly driven bychemical or hydrological rather than biological factors(Kalbitz et al., 2000). DOC was, for example, producedin deeper sections of the steady state cores (Fig. 8) al-though CO2 was not (Fig. 2), which suggests that theDOC release there was chemically or physically induced.Advective removal of pore water DOC in the steady statecores also allowed for continuous production, whereasthe production in the water table fluctuation experiment,without such removal, seemed to drop off when concen-trations reached more than ca. 50 mg L–1 DOC. This find-ing also suggests a chemical control on DOC release. Inthe short run DOC dynamics might therefore be relativelyindependent of biological processes after flooding.
Adjustments in anaerobic DIC and particularly CH4
production and fluxes after flooding were much slowerand production rates were decoupled from gas exchangerates with the atmosphere (Figs. 5A, 6B). This result ispartly explained by time lags in the development of themicrobial potentials and partly by the slow transport ve-locities in the water-saturated peat column. When satu-rated, significant exchange between the peat and the at-mosphere will occur only after large pools of dissolvedgases have built up in the peat. This results in local trans-port after formation of steep concentration gradients atthe interface to the atmosphere (or to the unsaturatedpeat), or to degassing due to partial CH4 pressures ex-ceeding atmospheric pressures (Fechner-Levy and He-mond, 1996).
At the turnover rates and pool sizes determined, thedevelopment of large dissolved CO2 and CH4 pools tookweeks to months after flooding. During these periods, thefluxes of CO2 and CH4 were disconnected from C miner-alization rates and the controls on them. CH4 productionand emissions were particularly affected by this phenom-enon, as the slower production rate did not allow for thebuildup of a considerable dissolved CH4 pool in compar-ison to the CO2 pool (Figs. 1, 5A and 6A). As the com-parison between production rates and diffusive fluxes forthe flooded period illustrates, only 5 to 8% of the CH4
produced was emitted during flooding, compared to 24 to28% of the DIC produced (Table 1). These results explainearlier observations that CH4 emissions remain low afterthe water table has been raised, whereas CO2 emissionsincrease more quickly (Moore and Dalva, 1993).
Implications for CO2 and CH4 emissions from peatlandsThe results of this study have both implications with re-spect to the investigation of CO2 and CH4 fluxes frompeatlands, and with respect to the role of peatlands in theglobal C cycle. They indirectly imply that there are limitsto the prediction of trace gas exchange from statisticalmodels relating environmental variables, such as watertable and temperature, to trace gas fluxes between peat-lands and the atmosphere. This is a consequence of the in-teraction of production, consumption, storage, and trans-port of gases in the peat, which, at time scales of weeks tomonths will not reach an equilibrium when environmen-tal controls vary. Hence, biogeochemical processes in thepeat are, to a varying degree, decoupled from fluxes to theatmosphere, which explains field observations (e.g.,Moore et al., 1990; Shannon and White, 1994; Schimel,1995). Poor statistical correlation between environmentalvariables and gas exchange rates seems thus inevitable atshort time scales, when changes in dissolved gas poolsare not taken into account. Better results can be expected,and have been obtained, for seasonally cumulated ex-change rates of CH4 (e.g., Moore et al., 1998) that elimi-nate the short-term decoupling of production and ex-change rates.
This methodological constraint also is illustrated bythe decreasing CH4 production rates at day 218–222 ofthe experiment, which did not result in decreased diffu-sive CH4 fluxes (Fig. 6B). The CH4 production patternmight have been caused by a suppressive effect of SO4
2–
reduction (Blodau and Moore, 2002b), since decreasingCH4 production at days 217–222 coincided with a periodof net production of H2S. During transient conditions,biogeochemical effects on the CH4 and CO2 production,such as through sulfate reduction after a sulfate pulse,thus cannot be identified by instantaneous gas flux mea-surements over short periods of time, but must be basedon long-term cumulative fluxes. This is in agreement withDise and Verry’s finding (2001) that sulfate addition infield experiments significantly decreased CH4 emissionrates from a peatland only when fluxes were cumulative.
The results also imply that the response of C cyclingin peatlands will depend not only on average changes inwater table levels, but also on the frequency of water tablefluctuations. This is suggested by the 5-fold increase inanaerobic CO2 production after an unsaturated period(Table 1) and by the decrease in CH4 emission from ca.1.1 ± 0.7 mmol m–2d–1 (including ebullition 2 ± 0.8 mmolm–2d–1) to 0.64 ± 0.12 mmol m–2d–1. The combined effectof environmental variables on the trace gas exchange re-mains, however, uncertain, as has been pointed out inmodeling of the CH4 dynamics in peatlands (Walter andHeimann, 2000). Although the conditions chosen in theexperiments are really not representative for the field, theresults of this study suggest, for example, that unsatu-
60 Christian Blodau and Tim R. Moore Response of C dynamics to a water table fluctuation
rated periods do not necessarily decrease CH4 emissionrates because water table fluctuations potentially increasethe fraction of CH4 that can escape oxidation. CH4 emis-sions decreased from 1.9 mmol m–2 d–1 (the sum of dis-solved export from the mesocosms with the removed wa-ter and diffusive emission, Table 1) under steady state to0.03 mmol m–2 d–1 during flooding after an unsaturatedperiod (Table 1). Rapid degassing due to changing watertables, however, partly offset this decrease, probablyadding most of the accumulated methane pool to the CH4 emission, which would add an estimated 0.54 mmolm–2 d–1 to the average CH4 flux over the flooded-drained-flooded period.
ConclusionsShort-term disturbances, such as water table draw-downs,change patterns and rates of C mineralization in peat. Theresults of the study suggest that periodic water tabledraw-downs will increase aerobic and anaerobic C min-eralization and emission and decrease CH4 productionand emission from peatlands. As well, the results suggestthat the physical disturbance of incubating peat increasesproduction rates compared to in situ at steady state, par-ticularly deeper in the peat. Short-term disturbance can,moreover, result in disparate production rates and tracegas fluxes. In this study the equilibration time of the indi-vidual processes after the disturbance ranged from in-stantaneous (transport), days (aerobic mineralization,CH4 oxidation), weeks (DOC production, anaerobic CO2
production) to months (CH4 production, adjustment offluxes to production). As a consequence, anaerobic pro-duction rates of CO2 and CH4 and exchange rates with theatmosphere are decoupled from each other at short timescales, potentially explaining the frequently reported lackof correlation between environmental variables and tracegas fluxes in field investigations. Degassing of CH4 afterwater table draw-downs might contribute significantly tothe overall CH4 emissions and partly compensate forsmaller CH4 production in peatlands that are affected bywater table fluctuations.
Acknowledgments
We thank J. Bubier for the identification of plant species,C. L. Roehm, N. Basiliko, E. Gorham and two anony-mous reviewers for valuable comments on the manu-script. The work was funded by the Natural Sciences andEngineering Research Council of Canada. The first author was supported by a McGill-McConnell Fellow-ship, a Government of Canada Award by the InternationalCouncil for Canadian Studies, and a stipend by the FCAR Centre for Global and Climate Change Researchat McGill.
References
Aerts, R. and F. Ludwig, 1997. Water-table changes and nutritionalstatus affect trace gas emissions from laboratory columns ofpeatland soils. Soil. Biol. Biochem. 29: 1691–1698.
Aerts, R. and S. Toet, 1997. Nutritional controls on carbon dioxideand methane emission from Carex dominated peat soils. Soil.Biol. Biochem. 29: 1997.
Aller, R. C., 1994. Bioturbation and remineralization of sedimen-tary organic matter: effects of redox oscillation. ChemicalGeol. 114: 331–345.
Bellisario, L. M., J. L. Bubier, T. R. Moore and J. B. Chanton, 1999.Controls on CH4 emissions from a northern peatland. GlobalBiogeochem. Cycles 13: 81–91.
Blodau, C., 2001. Carbon biogeochemistry in northern peatlands:Regulation by environmental and biogeochemical factors.Ph.D. thesis, 189 p., McGill University, Montreal.
Blodau, C., 2002. Carbon cycling in peatlands: A review ofprocesses and controls. Environmental Reviews 10: 111–134.
Blodau, C. and T. R. Moore, 2002a. Macroporosity affects watermovement and pore water sampling in peat soils. Soil Sci. 167:98–109.
Blodau, C. and T. R. Moore, 2002b. Micro-scale CO2 and CH4 dy-namics in a peat soil during a water fluctuation and sulfatepulse. Soil Biol. Biochem., in press.
Blunier, T., J. Chappelaz, J. Schwander, B. Stauffer and D. Raynaud,1995. Variations in atmospheric methane concentrations duringthe Holocene epoch. Nature 374: 46–49.
Bottner, P., 1985. Response of microbial biomass to alternate moist and dry conditions in a soil incubated with 14C- and 15N-labelled plant material. Soil Biol. Biochem. 17: 329–337.
Bubier, J. L., 1995. The relationship of vegetation to methane emis-sion and hydrochemical gradients in northern peatlands. J.Ecology 83: 403–420.
Bubier, J. L., T. R. Moore and N. T. Roulet, 1993. Methane emis-sions from wetlands in the midboreal region of northern On-tario, Canada. Ecology 74: 2240–2254.
Bubier, J., T. R. Moore and S. Juggins, 1995. Predicting methaneemission from bryophyte distribution in northern peatlands.Ecology 76: 677–693.
Clein, J. S. and J. P. Schimel, 1994. Reduction in microbial activityin birch litter due to drying and rewetting events. Soil Biol.Biochem. 26: 403–406.
Cornel, P. K., R. S. Summers and P. V. Roberts, 1985. Diffusion ofhumic acid in dilute aqueous solution. J. Colloid Interface Sci.110: 149–164.
Dise, N. B. and E. S. Verry, 2001. Suppression of peatland methaneemissions by cumulative sulfate deposition in simulated acidrain. Biogeochemistry 53: 143–160.
Edwards, C., B. A. Hales, G. H. Hall, I. R. McDonald, J. C. Murrell,R. Pickup, D. A. Ritchie, J. R. Saunders, B. M. Simon and M.Upton, 1998. Microbiological processes in the terrestrial car-bon cycle: Methane cycling in peat. Atmos. Environ. 32:3427–3255.
Fechner-Levy, E. J. and H. F. Hemond, 1996. Trapped methane vol-ume and potential effects on methane ebullition in a northernpeatland. Limnol. Oceanogr. 41: 1375.
Fetzer, S., F. Bak and R. Conrad, 1993. Sensitivity of methanogenicbacteria from paddy soil to oxygen and desiccation. FEMS Mi-crobial Ecol. 12: 107–115.
Freeman, C., M. A. Lock and B. Reynolds, 1993. Fluxes of CO2,CH4 and N2O from a Welsh peatland following simulation ofwater table draw-down: potential feedback to climate change.Biogeochemistry 19: 51–60.
Fung, I., J. John, J. Lerner, M. Mathews, M. Prather, L. Steele andP. Frazer, 1991. Global budgets of atmospheric methane: resultsfrom a three-dimensional global model synthesis. J. Geophys.Res. 6: 13033–13065.
Aquat. Sci. Vol. 65, 2003 Research Article 61
Funk, D. W., E. R. Pullman, K. M. Peterson, P. M. Crill and W. D.Billings, 1994. Influence of water table on carbon dioxide, car-bon monoxide and methane fluxes from taiga bog microcosms.Global Biogeochem. Cycles 8: 271–278.
Gorham, E., 1991. Northern peatlands: Role in the carbon cycle andprobable responses to climatic warming. Ecological Applica-tions 1: 182–195.
Hall, G. H., B. M. Simon and R. W. Pickup, 1996. CH4 productionin blanket bog peat: A procedure for sampling, sectioning, andincubating samples whilst maintaining anaerobic conditions.Soil Biol. Biochem. 28: 9–15.
Hogg, H. E., 1993. Decay potential of hummock and hollow Sphag-num peats at different depths in a Swedish raised mire. Oikos66: 269–278.
Jin, Y. and W. A. Jury, 1996. Characterizing the dependence of gasdiffusion coefficient on soil properties. Soil Sci. Soc. Am. J. 60:66–71.
Kalbitz, K., S. Solinger, J.-H. Park, B. Michalzik and M. Matzner,2000. Controls on the dynamics of dissolved organic matter insoils: A review. Soil Sci. 165: 277–304.
Kelly, C. A. and D. P. Chynoweth, 1980. Comparison of in situ andin vitro rates of methane release in freshwater sediments. Appl.Environ. Microbiol. 40: 287–293.
Kettunen, A., V. Kaitala, A. Lehtinen, A. Lohila, J. Alm, J. Silvolaand P. J. Martikainen, 1999. Methane production and oxidationpotentials in relation to water table fluctuations in two borealmires. Soil Biol. Biochem. 31: 1741–1749.
Kettunen, A., V. Kaitala, J. Alm, J. Silvola, H. Nykänen and P. J.Martikainen, 1996. Cross-correlation analysis of the dynamicsof methane emissions from a boreal peatland. Global Bio-geochem. Cycles 10: 457–471.
Kieft, T. L., E. Soroker and M. K. Firestone, 1987. Microbial bio-mass response to a rapid increase in water potential when drysoil is wetted. Soil Biol. Biochem. 19: 119–126.
Lerman, A., 1979. Geochemical processes, John Wiley & Sons Inc.,New York.
Liblik, L. K., T. R. Moore, J. L. Bubier and S. D. Robinson, 1997.Methane emissions from wetlands in the zone of discontinuouspermafrost: Fort Simpson, North West Territories, Canada.Global Biogeochem. Cycles 11: 485–494.
Mayer, H. P. and R. Conrad, 1990. Factors influencing the popula-tion of methanogenic bacteria and the initiation of methane pro-duction upon flooding of paddy soil. FEMS Microbial Ecol. 73:103–112.
Millington, R. J. and J. P. Quirk 1961. Permeability of porous solids.Trans. Faraday Soc. 57: 1200–1207.
Moore T. R., N. T. Roulet and R. Knowles, 1990. Spatial and tem-poral variations of methane flux from subarctic/northern borealfens. Global Biogeochem. Cycles 4: 29–46.
Moore, T. R. and M. Dalva, 1993. The influence of temperature and water table on carbon dioxide and methane emissions from
laboratory columns of peatland soils. J. Soil Sci. 44: 651–664.
Moore, T. R. and R. Knowles, 1989. The influence of water tablelevels on methane and carbon dioxide emissions from peatlandsoils. Can. J. Soil. Sci. 69: 33–38.
Moore, T. R., A. Heyes and N. T. Roulet, 1994. Methane emissionsfrom wetlands, southern Hudson Bay Lowland. J. Geophys.Res. 99: 1455–67.
Moore, T. R., N. T. Roulet and J. M. Waddington, 1998. Uncertaintyin predicting the effect of climatic change on the carbon cycleof Canadian Peatlands. Climatic Change 40: 229–245.
Nilsson, M. and E. Bohlin, 1993. Methane and carbon dioxide con-centrations in bogs and fens- with special reference to the ef-fects of the botanical composition of the peat. J. Ecol. 81:615–625.
Nykaenen H., J. Alm, J. Silvola and K. Tolonen, 1998. Methanefluxes on boreal peatlands of different fertility and the effect oflong-term experimental lowering of the water table on fluxrates. Global Biogeochem. Cycles 12: 53–69.
Öquist, M. and I. Sundh, 1998. Effects of a transient oxic period onmineralization of organic matter to CH4 and CO2 in anoxic peatincubations. Geomicrobiology 15: 325–333.
Schimel, J. P., 1995. Plant transport and methane production as con-trols on methane flux from arctic wet meadow tundra. Biogeo-chemistry 28: 183–200.
Shannon, R. D. and J. R. White, 1994. A three year study of controlson methane emissions from two Michigan peatlands. Biogeo-chemistry 27: 35–60.
Updegraff, K., J. Pastor, S. D. Bridgham and C. A. Johnston, 1996.Environmental and substrate controls over carbon and nitro-gen mineralization in northern wetlands. Ecol. Appl. 5:151–163.
Valentine, D. W., E. A. Holland and D. S. Schimel, 1994. Ecosystemand physiological controls over methane production in northernwetlands. J. Geophys. Res. 99: 1565–1571.
Van den Pol-van Dasselaar, A. and O. Oenema, 1999. Methane pro-duction and carbon mineralisation of size and density fractionsof peat soils. Soil Biol. Biochem. 31: 877–86.
Walter, B. and M. Heimann, 2000. A process based climate sensi-tive model to derive methane emissions from natural wetlands:Application to five wetland sites, sensitivity to model param-eters, and climate. Global Biogeochem. Cycles 14: 745–765.
Whiting, G. J. and J. P. Chanton, 1993. Primary production controlof methane emissions from wetlands. Nature 364: 794–795.
Yavitt, J. B., D. M. Downey, E. Lancaster and G. E. Lang, 1990.Methane consumption in decomposing Spaghnum derived peat,Soil Biol. Biochem. 22: 441–447.
Yavitt, J. B., C. J. Williams and R. K. Wieder, 1997. Production ofmethane and carbon dioxide in peatland ecosystems acrossNorth America: Effects of temperature, aeration, and organicchemistry of the peat. Geomicrobiol. J. 14: 299–316.
62 Christian Blodau and Tim R. Moore Response of C dynamics to a water table fluctuation
