Methane fluxes from three peatlands in the La Grande Rivière watershed, James Bay lowland, Canada

4.1. Vegetated Surface Measurements

[7] At the end of May 2003, 20 collars (diameter 25 cm) were installed in the LG1 and LG2 peatlands and 19 collars in the LG3 peatland, with 2 to 4 collars covering each biotype (Table 1). Planks were installed on the peat surface next to each group of collars to minimize disturbance during flux measurements. PVC tubes were inserted in the peat to measure WTD next to each biotype where gas flux measurements were made. A meteorological tower was installed in each of the three peatlands in order to continuously measure water table depth (WTD), air temperature, peat temperature at 5, 10, 20 and 40 cm depth and photosynthetic photon flux density (PPFD). The tower in LG1 was located on biotype SeV while the towers in LG2 and LG3 were located on biotype LHuS (Table 1). Daily precipitation at La Grande Rivière airport and daily water table depth measured at the meteorological tower in the three peatlands are presented in Figure 3.

[8] Methane flux measurements on vegetated surfaces were made between June and August 2003 by sampling the headspace in an 18-L chamber placed on the collar. Chambers were covered with tinfoil to prevent heating inside and there was a water seal between chamber and collar. Air in the chamber was mixed prior to sampling, using a 60-mL syringe, and 10-mL samples were collected in syringes every 5 min for a 20-min period.

[9] Methane fluxes were measured at the LG2 site during one week in November 2003 and one week in March 2004. Samples were taken using the same collars as during the 2003 growing season, after removing the snow pack covering them. The snow was removed to allow access to the collar rim. There was inevitable disturbance to the snow within the collar. However, no measurements were made on the WS and WD biotypes during November 2003 and March 2004, and on the SHo and the SeV biotypes during March 2004. These collars were covered by a thick layer of ice, making it impossible to reach the sampling collar. Gas samples were collected using the same equipment used for CH₄ sampling during the 2003 growing season. However, 10-mL glass vials with a rubber stopper and a metal crimp were used instead of 10-mL syringes. Air contained in the sealed glass vial was removed prior to sampling, using a 60-mL syringe with a 25-gauge needle. Samples were taken every 15 min for a 60-min period. Air in the chamber was mixed prior to sampling, using a 60-mL syringe. The longer sampling period is justified by the fact that the flux rates are smaller during winter than summer [Dise, 1992; Alm et al., 1999; Panikov and Dedysh, 2000].

[10] Methane concentrations were determined within 48 hours of collection on a Shimadzu Mini-2 gas chromatograph using a 5-mL hand-injected sample, a 1-mL injection loop and a 6′ Poropak-Q column (50/80 mesh) at 45°C. Detector temperature was 100°C. N₂ was used as the carrier gas at a flow rate of 30 mL min⁻¹. CH₄ standards of 2.73 and 200 parts per million by volume (ppmv) were used before each analysis run. CH₄ fluxes were calculated by linear regression using the concentration change with time in the five samples, rejecting fluxes with coefficients of determination (r²) of <0.85.

4.2. Pool Measurements

[11] During the 2004 growing season, chamber and ebullition CH₄ flux measurements were made at 15 sites within 5 to 8 pools, representing the different types of pools in each of the three peatlands. Five sites in each peatland were chosen for ebullition flux measurements, where the water column was >30 cm deep. Wood stakes, inserted in the vegetated surface peat at both ends of each transect, were used to attach the inverted funnels and/or the floating chambers to prevent them from moving during the sampling period. Chamber CH₄ flux measurements were made approximately every 10 days from the last week of June to the end of August 2004 while ebullition CO₂ and CH₄ flux measurements were made from the last week of July to the end of August. The late start was caused by frozen sediment impeding equipment insertion and leakage from the inverted funnels.

[12] Chamber CH₄ fluxes from pools were measured with floating collars, 2.5 cm above the base of the chamber, using the same sampling method and analysis used during growing season. Methane ebullition emissions in 2004 were measured with 30-cm-diameter floating inverted funnels fitted with a 100-mL graduated cylinder and rubber septum. The gas bubbles released by the sediments enter the water-filled inverted funnel and are trapped in the graduated cylinder and withdrawn with a 10-mL syringe. Ebullition fluxes were calculated by measuring the volume of gas/bubbles accumulated in the cylinder and the CH₄ mixing ratio of gas accumulated over time between the samplings. Gas samples were analyzed as described above.

5.1. Vegetated Surface Fluxes

[13] A total of 470 flux measurements were made during summer 2003, of which 44 were rejected (see 4). The remaining 426 measurements ranged from −2.9 to 1844 mg CH₄ m⁻² d⁻¹, with a mean and median of 53 and 25 mg CH₄ m⁻² d⁻¹, respectively. LG1 peatland individual CH₄ fluxes had a mean and a median of 72 and 34 mg CH₄ m⁻² d⁻¹, respectively. Average CH₄ fluxes for each of the LG1 biotypes ranged from 5.7 mg CH₄ m⁻² d⁻¹ on the TiSpHuS biotype to 197 mg CH₄ m⁻² d⁻¹ in the SpHo biotype (Table 3). The highest biotype values of 165 and 197 mg CH₄ m⁻² d⁻¹ from the P and SpHo biotypes at LG1 are probably the result of ebullition in collars 10 and 15 during the last day of sampling (Figure 4). If these extreme fluxes are removed, the summer averages were 125 and 71 mg CH₄ m⁻² d⁻¹ for the P and SpHo, respectively. Therefore the maximum average biotype value at LG1 would be 125 mg CH₄ m⁻² d⁻¹. At the LG2 peatland, the individual fluxes had a mean of 39 and a median of 25 mg CH₄ m⁻² d⁻¹. Biotype summer average fluxes ranged from 3.5 mg CH₄ m⁻² d⁻¹ on the LHu to 70 mg CH₄ m⁻² d⁻¹ in the WS biotype (Table 3). The LG3 peatland fluxes had a mean and median of 60 and 26 mg CH₄ m⁻² d⁻¹, respectively. Biotype averages ranged from 6.4 mg CH₄ m⁻² d⁻¹ on the SpHu biotype to 124 mg CH₄ m⁻² d⁻¹ in a SpHo biotype (Table 3).

[14] CH₄ fluxes varied during the summer among the different peatlands. In all three sites, lowest emission rates were observed early in the season with averages showing similar patterns during the first half of the growing season, with emission rates starting between 10 and 19 mg CH₄ m⁻² d⁻¹ early in the growing season and increasing from day of year (DoY) 160 to 200. At the LG2 site, average emission peaked at 68 mg CH₄ m⁻² d⁻¹ around DoY 200 and emission patterns in all three peatlands differed after this date. Average emission from the LG2 site decreased until the end of the summer, following the peak around DoY 200. At the LG3 peatland, CH₄ flux peaked around DoY 220 at approximately 100 mg CH₄ m⁻² d⁻¹ and reached a maximum in LG1 of 217 mg CH₄ m⁻² d⁻¹ around DoY 230. Peak values are respectively 15, 5, and 8 times greater than the early season emissions measured respectively in the LG1, LG2 and LG3 peatlands. Individual flux ranges, within single days, increased through the summer at the LG1 and LG3 peatlands as shown by the increasing range of percentiles. At the LG2 peatland, the largest difference between fluxes was observed around DoY 200. In general, ranges between the 25th and 75th quartiles are greater in LG1 and LG3 than LG2 throughout the summer.

[15] Methane flux variations during the summer for each biotype show that, in general, average flux and average flux variation of the hollows were greater than for the hummocks biotype in all three peatlands. LG1 and LG3 hollows showed increasing average fluxes as the summer advanced. All three peatlands hummocks average fluxes ranged between 0 and 45 mg m⁻² d⁻¹ except for the SpHu biotype at LG2 and the LHuS biotype at LG3, which had greater fluxes for the last sampling of the growing season.

5.2. LG2 Peatland Winter Fluxes

[16] CH₄ fluxes in November 2003 ranged from −2.9 to 32.8 mg CH₄ m⁻² d⁻¹ with a mean and median of 2.0 and 0.7 mg CH₄ m⁻² d⁻¹, respectively (Table 4). Biotypes mean CH₄ fluxes ranged from −0.4 mg CH₄ m⁻² d⁻¹ on the LH biotype to 4.7 mg CH₄ m⁻² d⁻¹ on both SpHo and SeV biotypes. SeV biotype mean CH₄ flux is statistically greater than all other biotype, except the SpHo (p < 0.05). Only the collars located on hummocks were sampled in March 2004, with fluxes ranging from 0.3 to 6.4 mg CH₄ m⁻² d⁻¹, with a mean and median of 2.1 and 2.1 mg CH₄ m⁻² d⁻¹, respectively. LH biotype mean flux was 0.6 mg CH₄ m⁻² d⁻¹, while the LHuS biotype mean flux was 1.1 mg CH₄ m⁻² d⁻¹. SpHu and SpHuS mean CH₄ fluxes were 3.8 and 3.3 mg CH₄ m⁻² d⁻¹ (Table 3). Overall, the March 2004 mean values from the hummock biotypes are statistically greater than the November 2003 values (p < 0.001, Kruskal-Wallis).

5.3. Pool CH₄ Fluxes

[17] A total of 270 individual chamber fluxes were made, of which 17 were rejected. Flux value of the remaining 253 measurements ranged from −32 to 8192 mg CH₄ m⁻² d⁻¹ with a mean and median of 136 mg and 25 mg CH₄ m⁻² d⁻¹ (Figure 5 and Table 5). The largest fluxes were measured on the LG3 peatland pools with an individual summer mean flux of 329 mg CH₄ m⁻² d⁻¹, compared to 54 and 34 mg CH₄ m⁻² d⁻¹ in the LG1 and LG2 pools, respectively. Summer average pool fluxes from LG3 are statistically greater than LG2 (p < 0.05) but not greater than LG1 (p = 0.09, Kruskal-Wallis).

[18] Within the LG1 peatland, the summer average fluxes from the structured fen section pools were not statistically different (p > 0.05) from the average flux from the larger pools in the higher portion of the LG1 peatland. In the LG2 peatland, average CH₄ fluxes from the sites 13 to 15 at LG2 were statistically greater than average fluxes measured from the larger and deeper pools (site 1–12) in the same peatland (p < 0.05, Kruskal-Wallis).

[19] Individual ebullition CH₄ fluxes from the last week of July to the end of August 2004 range from 0.002 to 117 mg CH₄ m⁻² d⁻¹ for the three peatlands. As for chamber fluxes, the largest emissions were from the LG3 peatland, with a single measurements average of 21 mg CH₄ m⁻² d⁻¹, statistically greater than the 1.61 and 0.81 mg CH₄ m⁻² d⁻¹ mean fluxes from the LG1 and LG2 peatlands, respectively (p < 0.05). Mean ebullition fluxes from the 5 sites within the LG3 peatland ranged from 2.9 to 67 mg CH₄ m⁻² d⁻¹ and the largest fluxes were measured at site 1.

5.4. Controls on CH₄ Fluxes

[20] As expected, average seasonal flux of CH₄ from the vegetated surface increased with a rise in the average water table: There was a correlation between log₁₀CH₄ flux and water table in each peatland (r² = 0.78 to 0.93, p < 0.02, Figure 6a). There were no significant differences in regression slopes and intercepts for the three peatlands and combination of all data resulted in an r² of 0.78 (p = 0.001). Seasonal average CH₄ fluxes from the pool sites in 2004 were not as strongly related (p = 0.012 to 0.079) to average pool depth with negative slopes in LG1 and LG2 and a positive slope for LG3 (Figure 6). In LG1 and LG2, there was an increase CH₄ flux as the average water table rose from 35 cm below to 5 cm above the peat surface, followed by a decrease in fluxes with increasing pool depth. In LG3, however, there was an increase in flux from sites with a water table 35 cm below peat surface to pool depth of 80 cm. When data are combined for vegetated surface and pools at each site, and overall, there is a significant convex-up relationship (Figure 6b).

[21] Within the season, CH₄ fluxes, expressed as the average of the 20 collars in each peatland on each sampling day, increased with increasing peat temperature measured at the meteorological towers (Table 6). Relationship between the CH₄ fluxes and peat temperature at 20 cm was significant (p < 0.06) in each peatland and explained between 63 and 83% of the seasonal variation in CH₄ flux within each peatland. The relationship was also significant for peat temperature at 40 cm at LG1 and LG3 where it explained 94% and 85% of the seasonal variation in CH₄ flux, respectively.

[22] Vegetation also played a role in controlling CH₄ emissions. Above-ground biomass of individual species within each collar were combined into groups (e.g., trees, shrubs, sedges, herbs, lichens and mosses) and entered into a step-wise regression against log₁₀CH₄ flux. The relationship between the four major plant groups (sedge, hummock, Cuspidata Sphagnum and shrub) and CH₄ flux represent the “best” relationship between CH₄ flux and the biomass of plant groups with an r² of 0.59, with sedge and cuspidata Sphagnum (those species in biotypes SpHo and WS in Table 1) showing a strong positive influence on CH₄ flux (Table 7). Although the sedge biomass was correlated with CH₄ flux when all collars are included (r² = 0.30, p < 0.001), restriction to the 10 SeV collars in LG1 and LG2 improved the relationship (r² = 0.55, p = 0.014, Figure 7). Combination of water table position and above-ground, green sedge biomass explained 70% of the variance in CH₄ fluxes (Table 7).

5.5. Seasonal and Annual Peatland CH₄ Flux Estimate

[23] To spatially weight CH₄ fluxes for the three peatlands, coverage of each biotype was estimated using a multiscale approach. Biotypes present in the peatlands were identified from vegetation relevés made on the three peatlands. Surface patterns were identified and delimited and their surface area was numerized using georeferenced aerial photographs of the three peatlands in ARCMAP. For each surface pattern, biotype and pool coverage were estimated (Table 2).

[24] The growing season for the CH₄ budget in the three peatlands was arbitrarily defined as 15 May to 31 August (109 days). For the LG2 annual budget, growing season period was extended to 20 October, for a total of 159 days. Cold season corresponds to the period between 21 October and 14 May (206 days). The 15 May date is based on field observation, as no snow was left on the ground and the air temperature was warm enough to allow photosynthesis. The 15 May date is also consistent with peat temperature >0°C, 5 cm below the surface measured in 2004. The 20 October end of growing season date corresponds to peat temperature <0°C, 5 cm below the surface.

[25] The average daily CH₄ fluxes (Table 3) for each biotype were used to estimate the growing season CH₄ budget. Contributions to the CH₄ budget from the shallow pools at LG2 (2003) and the larger pools in the three peatlands (2004) were estimated using summer average daily CH₄. In order to get a more representative estimation of gas release from the pools, average values are used for the different pool types within peatlands. For example, LG1 pool daily average CH₄ values are separated into “upper pools” and “structured fen pools” (Table 4). LG3 CH₄ pool fluxes from site 1 were not included in the average daily CH₄ flux from LG3 peatlands, because of evidence of ebullition through disturbance, so that value was reduced from a high value of 345 mg CH₄ m⁻² d⁻¹ (± 215) to a more rational 134 mg CH₄ m⁻² d⁻¹ (± 43). LG3 site 1 fluxes are high and probably result from ebullition induced by accidental sediment disturbance during sampling. In the LG2 annual budget, average values calculated for each biotype from the November 2003 and March 2004 CH₄ measurements are used in the CH₄ winter budget calculation.

[26] Spatially weighted growing season CH₄ flux was significantly larger in the LG1 and LG3 peatlands (44 ± 21 and 52 ± 17 mg CH₄ m⁻² d⁻¹) than in the LG2 peatland (21 ± 9.4 mg CH₄ m⁻² d⁻¹) (Table 8). The standard error was derived from individual biotype standard error extrapolated spatially on the basis of coverage of each biotype (Table 1) and does not include errors linked to biotype coverage extrapolation. Growing season emission of CH₄ was 4.8 ± 2.3, 2.3 ± 1.0 and 5.7 ± 1.9 g CH₄ m⁻² at LG1, LG2 and LG3, respectively. In the LG2 peatland, the extended season to 20 October resulted in a flux of 3.3 ± 1.5 g CH₄ m⁻² and combination with the estimate of cold season flux resulted in an estimated annual flux of 3.8 g m⁻², with the winter contributing to 13% of the overall emission. The estimate does not include possible episodic emissions of CH₄ during spring thaw.

6.1. Methane Fluxes

[27] The CH₄ fluxes measured in the La Grande peatlands are similar to those reported in other northern peatlands. The hummock average fluxes ranged between 3.4 and 21 mg CH₄ m⁻² d⁻¹, similar to those measured on hummocks in other Canadian peatlands of −1.3 to 23 mg CH₄ m⁻² d⁻¹ [Bubier, 1995; Bubier et al., 1995; Liblik et al., 1997]. From biotypes with the water table closer to the surface, not including the shallow pools (2003), fluxes ranged from 48 to 197 mg CH₄ m⁻² d⁻¹, which is similar to values of 47 to 221 mg CH₄ m⁻² d⁻¹ presented by Liblik et al. [1997], for two peatlands in Fort Simpson, Mackenzie Valley, Canada. Although daily winter fluxes from the LG2 peatland are approximately 20 times smaller than those in the summer, the winter contribution to the annual CH₄ release is significant. The LG2 winter CH₄ fluxes averaged 2.0 and 2.1 mg m⁻² d⁻¹, smaller than the range of 5 to 23 mg m⁻² d⁻¹ presented in the literature [Dise, 1992; Alm et al., 1999; Panikov and Dedysh, 2000].

[28] The summer average fluxes of 54 and 34 mg CH₄ m⁻² d⁻¹ from the LG1 and LG2 pools are similar to 17 mg CH₄ m⁻² d⁻¹ measured by Kelly et al. [1997] while the LG3 mean flux of 329 mg CH₄ m⁻² d⁻¹ is in the same order of magnitude as fluxes measured by Hamilton et al. [1994] with 110 to 180 mg CH₄ m⁻² d⁻¹. The large variation in fluxes from the pools at LG3 may be a function of ebullition because of the floating peat debris and collapsing ridges that were observed in pools 1 and 2. The ebullition flux measurements showed only small releases of CH₄ through bubbling at LG1 and LG2 peatlands. Hamilton et al. [1994] reported no ebullition from their ponds on the Hudson's Bay lowlands. On the other hand, the LG3 peatland CH₄ ebullition fluxes are smaller than those reported by Dove et al. [1999] of 31 and 160 mg CH₄ m⁻² d⁻¹ for a beaver pond.

6.2. Controls on CH₄ Flux

[29] The strong relationships between average summer CH₄ flux and average water table depth (Figure 6) are in accord with other studies [Roulet et al., 1992; Moore and Roulet, 1993; Moore et al., 1994; Bubier, 1995; Bubier et al., 1995; Liblik et al., 1997; Nykänen et al., 1998; Huttunen et al., 2003]. The slopes and intercepts of the regressions of log₁₀ CH₄ flux against water table (0.038 to 0.044, and 1.83 to 2.20, respectively) are within the range reported by these other studies. In 2004, increasing pool depth resulted in decreased CH₄ flux in the LG1 and LG2 pools, with slopes significantly different from zero. This pattern may be explained by slower CH₄ production in the sediment of the deeper pools, which may contain well-decomposed organic matter. In a Wisconsin lake, Barber and Ensign [1979] found that methanogenesis was more rapid in shallow than deep-water sediments (1 and 3 m). Colder sediment temperature in deep portions of the pools could also reduce CH₄ production. In the LG3 peatland, CH₄ fluxes increased with pool depth, possibly caused by greater ebullition than in the other pools.

[30] Variations of CH₄ flux within the season were correlated with peat temperature and other studies have found that peat temperature at the water table position was the best predictor of CH₄ fluxes [Bubier et al., 1995; Nykänen et al., 1998; Bellisario et al., 1999]. In this study, the CH₄–peat temperature relationships only show the general summer trend effect of peat temperature on fluxes because peat temperature was not measured at each individual collars at time of sampling. The range of Q₁₀ values (0.8–5.1) is in the lower range of values reported by Segers [1998] of 1.5 to 28.

[31] The positive relationship between summer average CH₄ flux and sedge biomass for the collars within the SeV biotypes in the LG1 and LG2 peatlands suggests that the sedges act as a conduit to the atmosphere for CH₄, bypassing the oxidation layer and may stimulate methanogenesis [Whiting and Chanton, 1992; Waddington et al., 1996; Bellisario et al., 1999]. The CH₄ flux: sedge biomass relationship found in the La Grande Rivière peatlands is not significantly different from found in northern Manitoba by Bellisario et al. [1999] and suggests that end-of-season biomass may be a useful predictor, more so than for other vegetation characteristics.

[32] Peatland-average, summer estimated CH₄ fluxes of 2.3 to 5.7 g CH₄ m⁻² are similar to the estimates of Bubier et al. [1993] of 3.4 g CH₄ m⁻² for Clay Belt wetlands located in northeastern Ontario, Canada and of 2.5 to 8.2 g CH₄ m⁻² reported for hummock, transitional fens and low sedge fens by Nilsson et al. [2001] in boreal Sweden. Alm et al. [1999] found the winter contribution to represent 8 to 17% of the annual CH₄ flux and the winter contribution to CH₄ flux annual budget in LG2 represents 13%.

[33] CH₄ fluxes from these northern boreal peatlands can be significant, from the perspective of both the carbon budget of the peatland and the emission of greenhouse gas. While there are orders of magnitude variations in CH₄ flux across the peatland surface, variations on the terrestrial portions of the peatland can be related to water table position, peat temperature and vegetation cover, particularly sedges. The quantitative similarity of flux:environment relationships that we found to those from other northern peatlands suggests that overall relationships can be applied and scaled up in these landscapes, using simple remote sensing tools. CH₄ emissions from pools are not as strongly related to simple environmental properties, such as pool depth or size and differences in the evolution of the pools through degradation may lead to substantial differences in CH₄ emission. Episodic CH₄ ebullition is difficult to measure and may play a dominant role in flooded sections of the peatland. Combination of spatial estimates of CH₄ emission from peatlands such as these can then be compared with emissions from the water surface after flooding.