To date, very little information has been available on the build-up and release of biogenic gas bubbles in poorly-decomposed bog peats near the peatland surface (upper 1 m). We investigated the importance of ebullition of biogenic gas bubbles as a mechanism for the transport of CH4 to the atmosphere in eight cores (24 cm diameter, 22 cm depth) of poorly-decomposed, near-surface bog peat. Ebullition was recorded in all but one sample but varied greatly between samples. Maximum rates of CH4 efflux via ebullition were also highly variable, ranging from 2.2 to 83.0 mg CH4 m−2 day−1. These rates are similar to rates of diffusive CH4 efflux. Our results also show that wetland methane models are likely to need revision because they assume that unrealistically high CH4 pore-water concentrations are required before bubbles can be produced and because, in part, they do not account for gas bubble build-up prior to ebullition.
 It has been shown that ebullition of biogenic gas bubbles is an important mechanism of CH4 transfer between bogs and the atmosphere [e.g., Rosenberry et al., 2003; Glaser et al., 2004]. However, most studies that have been done on ebullition from bogs have looked at relatively deep peat in bogs in Minnesota where the bubbles are held below confining layers of woody peat of low hydraulic conductivity and apparently high resistance to bubble transfer. Under these conditions, large volumes of gas accumulate and are released in very large events where >40 g CH4 m−2 may vent to the atmosphere in a matter of minutes and hours. It is not clear how common or widespread such behaviour is or whether it is confined to peatlands such as those found in Minnesota. Equally, little is known about what happens in shallower bog peats (c. upper 1 m) where confining layers of woody peat may be absent. Therefore, the first aim of our study was to quantify ebullition in near-surface bog peats.
 Some physically-based wetland methane models, such as those of Walter et al. , Walter and Heimann , and Granberg et al.  allow for CH4 ebullition. In these models, bubbles do not form until pore-water CH4 concentrations exceed c. 440–500 μM (i.e., 7.1–8.0 mg CH4 L−1). This threshold concentration appears to be derived from classical analysis of bubble formation in lakes, where bubbles are assumed to be capable of forming when the quantity of gas in solution exceeds the equilibrium concentration (the solubility) [e.g., Hutchinson, 1957] with the added assumption of a CH4 mixing ratio in bubbles of 25 percent.
 We studied ebullition dynamics in eight c. 10-L samples of poorly-decomposed Sphagnum peat from lawns and hollows in two raised bogs: Longbridgemuir in SW Scotland (3°29′W 55°00′N), and Cors Fochno in W Wales (4°1′W 52°30′N). These sites were also used by Baird and Waldron , but new samples were collected for the study reported here. The samples from Longbridgemuir contained Sphagnum papillosum (two samples: LBMv2 and LBMv3) and S. magellanicum (two samples: LBMv1 and LBMv4) peat. The S. papillosum samples had von Post humification (H) scores ranging from 2 and 3 (top) to 4 (bottom), while the S. magellanicum samples ranged between 3 and 5 in each case. The four samples from Cors Fochno contained S. pulchrum (CFv1 and CFv3: H scores between 2 and 4) and S. magellanicum remains (CFv2 and CFv4: H scores between 2 and 4–5).
 The samples were cut from each site from the top 30 cm of peat and placed in metal cylinders 24 cm in diameter and 24 cm in length. In the laboratory, they were transferred to PVC cylinders of the same size and the growing-Sphagnum surface (2 cm) removed. The PVC cylinders were fitted with end plates which contained ports to which were connected transparent silicone rubber pipes. The samples were then slowly wetted from the base using the procedure described by Baird and Waldron . Following rewetting, the lower silicone pipe was clamped and water ponded above the peat surface in the upper pipe to simulate conditions below the water table. The upper pipe was coiled to form a gas trap, with the coil being initially water-filled. Ebullition was estimated typically every 2–3 days by measuring the volume of gas trapped in the coiled pipe. Our gas trap arrangement is the only simple way to measure ebullition. It is possible that the narrow opening of the port fitted into the upper end plate may have restricted CH4 and other gas export from the samples via diffusion. If this were to happen, one might expect higher dissolved CH4 concentrations than if no plate were fixed. However, our pore-water CH4 concentrations were within the range reported for field peats below the water table in NW Europe/Scandinavia and North America (see below).
 The samples were maintained at 12°C, a temperature found in the shallow peat in late spring, and late summer through to mid autumn. Measurements of volumetric water content (VMC) in each sample were taken using two time domain reflectometry (TDR) probes, one inserted into the upper half, and one in the lower half.
 VMC is the volume of water contained in a unit volume of peat. The TDR was calibrated for VMC for each peat type by bringing separate calibration samples of peat as close to saturation as possible by flushing them with de-aired water, after which they were weighed and a dielectric constant (ka) measurement taken. The samples were allowed to drain, and further sample weights and ka readings taken. Each sample was then oven dried and weighed which enabled porosity, VMC, and the VMC – ka relationship to be determined (r2 > 0.98 in each case). It was assumed that the porosity of each main peat sample was the same as the corresponding calibration sample. This assumption should be borne in mind when interpreting the results.
 The volumetric gas content (VGC), defined as the volume of gas per unit volume of peat, was obtained for each main peat sample by subtracting the VMC from the porosity. The main peat samples were also fitted with two gas permeation samplers from which gas samples were taken for analysis of CH4 content using a Poropak Q column on a Perkin Elmer 8700 gas chromatograph. The gas samplers consisted of silicone tubing (1.3 cm outside diameter × 9 cm length) filled with washed quartz sand inserted into the peat via a port in the PVC cylinder which was fitted with a rubber septum through which gas samples could be obtained using a gas-tight syringe. Finally, the incubations lasted for up to 125 days.
 In the results discussed below, pore-water CH4 concentrations were obtained by multiplying the solubility of CH4 (as given by Fogg and Gerrard ) by the proportion of CH4 in the gas permeation samplers. When calculating CH4 efflux from the samples it was assumed that the proportion of CH4 in the bubbles was the same as the proportion in the gas permeation samplers.
 A summary of the findings is given in Tables 1 and 2, with detailed information for two samples (LBMv3 and CFv4) in Figures 1, 2, and 3.
Table 1. Summary of the Volumetric Gas Content (VGC) and Ebullition Data Collected From Both Sites and Pore-Water CH4 Concentrationsa
Length of Experiment (days)
VGC Start Upper
VGC Start Lower
VGC Finish Upper
VGC Finish Lower
Ebullition Start Day
Ebullition Total (cm3)
CH4 (mg L−1) Bubble Build-Up Initiation
CH4 (mg L−1) Ebullition Initiation
CH4 (mg L−1) Final
NR denotes no reading. The data from this TDR probe exceeded the range of the calibration for this peat type. NC denotes not calculated (because VGC did not increase around both TDR probes – see text).
Table 2. Estimated CH4 Efflux (mg CH4 m−2 d−1) Via Ebullition for Each Sample
Denotes an estimate of rates at the beginning of the main period of ebullition.
Denotes an estimate of rates at the close of the experiment.
3.1. Biogenic Gas Bubble Storage and Ebullition
 Like Baird and Waldron  we found that each peat sample was undersaturated with water at the beginning of the experiment, despite our method of wetting being more likely to give higher levels of saturation than wetting via rainfall [cf. Baird and Waldron, 2003]. Undersaturation occurred in all samples, with initial VGCs ranging between 0.01 and 0.13 (Table 1 and Figure 1). In some samples, VGC decreased over the first few days of observation in response to dissolution of air trapped during the wetting of the sample (e.g., see upper TDR in Figures 1 and 2). Then, with the exception of CFv2, VGC increased at both the lower and upper TDR locations in response to biogenic gas production, indicated by increases in pore-water CH4 concentrations (see also section 3.2) as shown in Table 1 and in detail for LBMv3 in Figure 3. In all samples, again excepting CFv2, ebullition was recorded, with total ebullition amounts varying between 415 cm3 over 90 days (LBMv3) and 46 cm3 over 117 days (CFv4) (Table 1). After the inception of ebullition, rates of ebullition were not constant as shown by the departure from a straight line of the cumulative ebullition data in Figures 1 and 2.
3.2. Pore-Water CH4 Concentrations
 It was noted in section 1 that gas bubble formation is assumed to occur when pore-water CH4 concentrations exceed 7.1–8.0 mg CH4 L−1. To investigate whether this threshold applied to our samples, we examined pore-water CH4 concentrations at the beginning of biogenic gas bubble production as shown by the TDR readings. We calculated the timing of this as the time at which VGC in both TDR probes started to increase after any initial decrease. We also examined CH4 concentrations at the start of ebullition and the end of the experiment. To ensure our estimates were conservative, we used the higher reading from the two gas-permeation samplers. The results show that biogenic gas bubble build-up occurred at pore-water CH4 concentrations substantially below (5–10+ times) the assumed threshold (Table 1). Moreover, pore-water CH4 concentrations were below this threshold in six samples when ebullition started, and in three samples at the end of the experiment.
3.3. CH4 Fluxes Via Ebullition
 Estimates of CH4 efflux via ebullition at the start of ebullition and at the end of each experiment (when CH4 concentrations were higher) are shown in Table 2. We used the mean of the two gas permeation samplers in these estimates. Bubbles are more likely to be generated and released from parts of the peat where gas concentrations are highest. Thus, our estimates are again conservative. The values were very variable, ranging from 2.2 to 83.0 mg CH4 m−2 day−1 (Table 2).
 The CH4 losses via ebullition in this study are well within the range reported for field studies in northern peatlands measuring CH4 diffusion using the static chamber approach [e.g., Roulet et al., 1994; Waddington and Roulet, 1996]. Noting that our estimates are conservative (see also below), the losses of CH4 via ebullition would also likely be much higher if one were to consider a 1-m column of poorly-decomposed peat, because one would expect CH4–containing bubbles to be produced and released from the entire 1-m depth. Our results support both an episodic and regular pattern of ebullition. For example, once ebullition was initiated, the LBMv3 sample (Figure 1) demonstrated regular rates of ebullition, while the CFv4 sample (Figure 2) showed periods of higher rates of ebullition followed by several days of little or no gas release. In addition to the effect of confining layers [Rosenberry et al., 2003; Glaser et al., 2004], the episodic nature of ebullition has been explained by changes in pressure and temperature [Fechner-Levy and Hemond, 1996]. For example, falling atmospheric pressure or a decrease in pore-water pressures due to a falling water table may lead to gas coming out of solution and possible ebullition. Similarly, as peat temperatures increase, gas solubility decreases. However, in our study, pore-water pressure and temperature were the same for all samples, and each sample experienced ambient atmospheric pressure, which suggests that differences in peat properties are the reason for the variability in ebullition between samples. Differences in peat quality will result in differences in the rate of gas production and consumption, while differences in the pore-size distribution of the peat sample will lead to differences in the ability of the peat to trap and subsequently release bubbles. In our study the time to the initiation of ebullition ranged from about three weeks to almost 10, and there was also very large variation in ebullition volume (Table 1) and peak CH4 efflux (Table 2). However, one remarkable feature of our work was that the main period of ebullition did not commence in any sample until the VGC around at least one TDR probe was between 0.10 and 0.14, suggesting a common storage threshold before ebullition could occur.
 The pore-water concentrations observed in this study were within the range observed in field investigations [e.g., Waddington and Roulet, 1996]. The results suggest that biogenic gas bubbles form readily in peat at pore-water CH4 concentrations substantially below those previously assumed necessary. This requires some explanation. One possibility is that CH4 concentrations are highly variable spatially within the peat and at scales smaller than the permeation samplers so that concentrations are high enough for bubble formation in the immediate vicinity of a bubble but much lower elsewhere. Thus the permeation sampler gives an ‘average’ CH4 concentration which may not be representative of the situation close to bubbles. Of course, all sampling methods are prone to these sampling/scale problems. However, such problems do not detract from the fact that CH4 models do not account for heterogeneous ‘hotspot’ production: they assume a high CH4 concentration everywhere at a given depth before bubbles can form. Finally, we also propose that pockets of soil air trapped during water-table rise act as foci for bubble growth. Our results suggest that many trapped bubbles are initially maintained and then grow via inward diffusion of biogenic gas (see section 3.1). Thus, bubbles are likely always present in shallow peat after water-table rise, and simple thresholds of bubble formation will not be applicable in such circumstances.
 We have shown that ebullition can be an important mechanism of CH4 efflux from near-surface peat under realistic pore-water CH4 concentrations, and that previous assumptions about gas bubble production and ebullition in peats will need to be revised in future versions of physically-based wetland methane models. Further work is, however, needed on how changes in, for example, atmospheric pressure affect ebullition and whether preferential pathways are formed by gas bubbles moving upwards through the peat column and remain as conduits for further, later gas release. More also needs to be known about the variability of CH4 production at scales of a few cm and how individual bubbles grow and ‘scavenge’ CH4 from surrounding pore water.
 The project was funded by the UK Natural Environment Research Council (award #: GR3/12451).