Oxidation of methane in boreal forest soils: a comparison of seven measures

. Methane oxidation rates were measured in boreal forest soils using seven techniques that provide a range of information on soil CH, oxidation. These include: (a) short-term static chamber experiments with a free-air (1.7 ppm CH,) headspace, @) estimating CH, oxidation rates from soil CH, distributions and (c) 222Rn-calibrated flux measurements, (d) day-long static chamber experiments with free-air and amended (+20 to 2000 ppm CH,) headspaces, (e) jar experiments on soil core sections using free-air and (f) amended (+SO0 ppm CH,) headspaces, and (g) jar experiments on core sections involving tracer additions of I4CH,. Short-term unamended chamber measurements, "2Fh-calibrated flux measurements, and soil CH, distributions show independently that the soils are capable of oxidizing atmospheric CH, at rates ranging to < 2 mg m-2 d-'. Jar experiments with free-air headspaces and soil CH, profiles show that CH, oxidation occurs to a soil depth of 60 cm and is maximum in the 10 to 20 cm zone. Jar experiments and chamber measurements with free-air headspaces show that CH, oxidation occurs at low (<0.9 ppm) thresholds. The I1CH4-amended jar experiments show the distribution of end products of CH, oxidation; 60% is transformed to CO, and the remainder is incorporated in biomass. Chamber and jar experiments under amended atmospheres show that these soils have a high capacity for CH, oxidation and indicate potential CH, oxidation rates as high as 867 mg m-2 d-'. Methane oxidation in moist soils modulates CH, emission and can serve as a negative feedback on atmospheric CH, increases.


Introduction
Many regional estimates of emission of CH, to the atmosphere derive from "scaling up" results of enclosure experiments (Harriss 1989), where flux is estimated by changes in headspace CH, concentration over time in open-bottom chambers placed over the soil.Static chamber (no forced air circulation) flux determinations are inexpensive and require minimal equipment, but they have been criticized for disturbing conditions (advection, temperature, concentration gradients) at the soil surface (Schutz & Seiler 1989;Mosier 1990).Accurate regional estimates of soil CH, fluxes therefore depend on the validity of the commonly used static chamber method.Validation is lacking or indirect, usually involving comparison with rnicrometeorological flux estimates (e.g., Bartlett et al. 1991).However, similarity between gas flux estimates made by chamber and indirect techniques may be coincidental (Mosier 1989).
A recently introduced technique (Dorr & Miinnich 1987) employs the noble gas 222Rn as a conservative internal standard to provide an independent, point source flux for comparison with static chambers.Soil 222Rn is produced by radioactive decay of 226Ra and is removed only by diffusion to the atmosphere and radioactive decay.Soil-atmosphere exchange of any gas can be calculated from the rate of 222Rn accumulation in a chambzr and the ratios of molecular diffusivities and soil concentration gradients for the two gases.Disturbance of the soil 222Rn concentration gradient during flux measurement is negligible in soils where 222Rn has a comparatively large diffusional path length before decay."Rn-calibrated" CH, fluxes could be more accurate than direct, static chamber CH, flux determinations for these soils (Dorr & Miinnich 1990).
Static chamber experiments estimate net atmospheric CH, flux and give no information concerning methanogenesis and CH, oxidation, the soil microbial processes that influence flux.Microbial CH, oxidation is an important modulator of CH, flux, with global CH, consumption slightly in excess of emission (Reeburgh et al., submitted).Moist or dry soils from the tropics to arctic tundra show net CH, consumption (Harriss et al.  1982; Seiler et al. 1984; Steudler et al. 1989; Born et al. 1990; Keller et  al. 1990; Whalen & Reeburgh 1990b; Yavitt et al. 1990a; Whalen et al.  1991), which is indicated by a decrease in chamber headspace CH, concentration.Inundated tundra soils that experience a reduced summer water table may become a sink for atmospheric CH,, due partly to increased CH, oxidation (Whalen et al. 1991).Loci, controls and in situ and potential rates of CH, oxidation are well-characterized for aquatic environments (reviewed by Rudd & Taylor 1980; Kiene 1991), but similar, detailed terrestrial data are available only for a cultivated humisol (Megraw & Knowles 1987), moist tundra (Whalen & Reeburgh 1990b) and landfill soils (Whalen et al. 1990).The fate of oxidized CH, is important for models of climate change, as CH, is 3.7-(Lashoff & Ahuja 1990) to 30-fold (Blake & Rowland 1988) more effective as a greenhouse gas than CO,.Clearly, an improved understanding of the controls on soil CH, oxidation is essential to predict the impact of climate change on the atmospheric CH, budget.Experiments must extend beyond chamber determinations of net flux and should involve discrete soil samples.
We have measured net CH, flux at permanent sampling stations in moist taiga (boreal forest) soils for two years using a static chamber technique (Whalen et al. 1991).These soils consistently consume atmospheric CH, and show no evidence of methanogenesis throughout the thaw season, so that chamber determinations of net CH, flux are equivalent to depth-integrated rates of CH, oxidation.The aim of the present two-day field and one-week laboratory study was to confirm our chamber flux determinations and to explore the controls on the capacity of CH, oxidation in taiga soils.
We addressed these objectives by using seven techniques to directly or indirectly estimate rates of CH, oxidation in representative moist taiga soils.These methods included: (a) experiments monitoring the change in headspace CH, concentration over 0.75 h with static chambers initially equilibrated with a free-air atmosphere (1.7 ppm CH,); (b)  (f) jar experiments involving soil core sections under amended atmospheres (+500 ppm CH,); and (g) jar experiments exposing soil core sections to "CH,.We expected these experiments to give area-based estimates of in situ and potential rates of CH, oxidation, thresholds and capacities for CH, oxidation, soil distributions of CH, and CH,-oxidizing activity and the distribution of end products of CH, oxidation.This study is not a comparison of measurements of in situ CH, oxidation by seven methods.Rather, we use these seven measures because of the unique insight each provides concerning rates and controls on CH, oxidation in taiga soils.

Field sites
This study was conducted in the Bonanza Creek Experimental Forest (64"45'N, 148"18'W, a 5045-ha research area located 20 km west of Fairbanks, Alaska.The mean annual and July air temperatures in the Fairbanks area are -3 "C and 17 "C, respectively, and the average frostfree period is about 100 d.Precipitation averages 285 mm annually; 6.5% is rain.Topography is gently rolling and upland soils are well-drained.Permafrost is discontinuous and usually confined to north-facing slopes with a black spruce (Picea manana) overstory.Upland soils are stone-free and have slight morphological development.The parent material is a micaceous loess deposited during the most recent Pleistocene glaciation.The physiography of interior Alaskan taiga, including this study area, is given in Van Cleve & Dyrness (1983), Viereck et al. (1983) and Van Cleve et al. (1 99 1).
Four upland sites were selected for study.Intermediate successional stages were represented by south-facing aspen (Populus tremuloides; site AS2) and north-facing birch (Betula papyrifera; site NB2) communities, whereas advanced successional stages were represented by north-facing black spruce (BS2) and south facing white spruce (Picea glauca; site UP3A) stands.The deciduous sites have a heavy ground cover of leaf litter and an insignificant understory of shrubs and herbs.The coniferous sites show a continuous ground cover of feather mosses (predominately Pleurozium sp. and Hylocomium sp.).These are invaded by lowbush cranberry ( Vaccinium vitis-idaea) and lichens at BS2.

Field sampling
Except where noted, all measurements at sites AS2 and NB2 were taken on 9 October 1990 and measurements at sites BS2 and UP3A were taken on 11 October 1990.
Static chamber determinations of 222Rn and CH, flux were made at stations that had been sampled regularly during the thaw season.Each chamber consisted of a skirted aluminum base permanently seated in the soil and lucite vertical sections and lids that utilize a water-filled channel for a seal (Whalen & Reeburgh 1988).Samples for CH, analysis were collected over a 0.75 h period from one chamber whose lid was equipped with a septum for syringe sampling of headspace gas.A second, similar chamber located within a few meters of the first chamber was used for 222Rn flux determinations.Evacuated counting cells (Lucas 1957) were filled directly from the chamber lid with a quick-connect fitting.
Soil gas samples were obtained by inserting a perforated stainless steel tube to known depths and using a battery-powered diaphragm pump to fill 0.5 L Tedlar bags (Born et al. 1990).The bags were sampled by syringe and evacuated counting cell for CH, and 222Rn analyses, respectively.Depth distributions for soil temperature were determined at each site with a portable thermistor probe.
Duplicate 30 cm soil cores were collected from each site with a 15 cm diameter stainless steel coring apparatus.Cores were then cut horizontally into three 10 cm sections.The centers of the soil cores were sub-cored using a 6.7 cm ID X 10 cm long plastic tube.Additional 10 cm long core sections were obtained from 30 to 60 cm below the soil surface by inserting a 6.7 cm ID plastic tube into the hole created by removal of the initial 30 cm core.Bedrock limited core collection to the upper 40 cm of soil at BS2.All core sections were extruded into 1 liter Mason jars fitted with a septum for syringe sampling of headspace gas.It was impossible to maintain the integrity of the soil matrix during extrusion.Soils were returned to the laboratory and stored at 5 "C.
Methane consumption thresholds and capacities were studied using three static chambers reserved for these experiments at each site.Disappearance of CH, in each chamber was monitored following equilibration of the headspace with a free-air atmospheres ( -1.7 ppm CH,).The same chambers were used several days later in amended atmosphere experiments where headspaces were initially adjusted to about 20, 200 or 2000 ppm CH, (one chamber, each concentration).These additions increased the initial CH, concentration 10 to 1000-fold above ambient.Headspace CH, was sampled 7 or 8 times over a 24 h period in all experiments.The first sample (to) in amended atmosphere experiments was taken 0.5 h after CH, addition to allow equilibration of headspace gas.
Permanent sampling stations for all chamber flux studies described above were clustered within a -25-m2 area at each site to minimize natural variability; soil cores, gas samples and temperature measurements were also taken within this plot.

Laboratory studies
Laboratory studies of microbial CH, oxidation in soil samples were conducted in Mason jars (unless otherwise noted) at 10 "C and were completed within one week of sample collection.Two time-course experiments with periodic sampling for CH, consumption were conducted on all 10-cm core sections that were extruded into Mason jars from each 40 or 60 cm core.The first experiment lasted 24 h during which the decrease in CH, concentration was monitored after initial equilibration with the atmosphere.The second experiment was 12 h in length and initial atmospheres were adjusted to -500 ppm CH,.We also studied 14CH4 oxidation using 25 to 100 g aliquots of homogenized soil from each 10-cm section of a single 40 or 60 cm core.Soil samples in 0.25-liter jars were equilibrated with a free-air atmosphere and amended with microliter quantities of rnicrobially-produced 14CH, (Daniels & Zeikus 1983) tracer (3.5 kBq; specific activity 2005 MBq rnrnol-I).Tracer addition increased the headspace CH, concentration by about 8%.Methane oxidation was terminated after 12 h by adding 0.2 cm3 C211, (BCdard & Knowles 1989) and the jars plus contents were frozen until tracer recovery.Samples were thawed and jar headspaces were flushed with He into a stripping/oxidation line (Whalen et al. 1990) where I4CO2 was trapped directly and 14CH4 was trapped as 14C02 after combustion.Soils were freeze-dried and assayed by dry combustion for 14C incorporated in microbial biomass (Whalen et al. 1990) Physical and chemical properties of the soil cores were determined at the end of these studies.Soil pH was measured potentiometrically according to McLean (1982).Soil moisture was determined gravimetrically and is expressed as percent of oven-dried (105 "C) mass.Organic content was determined by loss on ignition (550 "C) of oven-dried samples, and particle density was measured pycnometrically.Soil bulk density was computed as the quotient, over-dried mass divided by the field volume.Air-filled porosity was computed as the field volume minus the liquid and solid volumes.Determination of soil physical properties and calculations follow Klute (1986).
Methane determinations were made by flame ionization detection gas chromatography with a precision of < 1% (Whalen & Reeburgh 1988); calibration gases are relatable to standards from the National Institute for Technology and Standards.Sample analysis for CH, was completed within a few hours of collection for field samples and immediately upon sampling in laboratory experiments.Radon-222 activity was determined by scintillation counting of gas samples contained in Lucas cells.Counting was done in either a dual channel alpha scintillation counter (Applied Techniques) or a portable radon monitor (Pylon Model AB-5), both of which accomodated our counting cells.Counting cells were constructed of pyrex or quartz bulbs ( -100 cc volume) equipped with a Swagelok quick-connect fitting to permit evacuation and introduction of soil gas samples.The interiors of the cells were coated with silver-activated ZnS (Lucas 1957), which was supported by either a thin coating of stopcock grease or a layer of clear Krylon paint.Cell backgrounds and efficiencies were monitored continuously.Backgrounds averaged 0.39 cpm.Counting cell efficiencies (75%) were determined by counting air from a sealed glass tube containing a 226Ra standard supported on moist Mn-coated acrylic fibers (Butts et al. 1988).The counting cells were equilibrated for at least 3.5 h so that measured counts are the sum of alpha decay from 222Rn and its short-lived daughters, 218Po and 2 1 4 P ~.All 222Rn data were decay-corrected to the time of sampling.The precision of a 222Rn standard (228 dpm) determination is 3%.

Soil properties
A litter layer was characteristic of all sites, but the overlying moss carpet at BS2 and UP3A resulted in lower mean soil temperatures to 60 cm at BS2 (1.5 "C) and UP3A (2.7 "C) than at AS2 (4.5 "C) and NB2 (4.1 "C).Soil temperatures showed little variation with depth at any station, ranging about 2 "C from the surface to 60 cm (Fig 1).The dark surface, organic zone graded into a tan mineral horizon below about 7 and 15 cm at UP3A and BS2, respectively, while the transition began at about 4 cm at the deciduous sites.Bulk density (,q,) was lower and soil organic content, moisture content and gas-filled porosity ($$ were higher in the 0 to 10 cm and 10 to 20 cm zones at the spruce sites (particularly BS2) than at the hardwood sites (Table 1).Soils were almost entirely mineral below 30 cm at BS2 and below 20 cm at all other sites.Soil physical properties varied little below these depths.Soils were neutral or acidic with pH values varying from 4.9 to 7.4 (Table 1).Lowest soil pH was observed at coniferous sites, and pH increased with depth at all sites.Data given here for differences in pH, moisture, temperature and organic content of deciduous and coniferous taiga soils are consistent with similar reports for soils in this biome (Flanagan & Van Cleve 1983;Viereck et al. 1983;Bonan & Shugart 1989;Van Cleve et al. 1991).

Soil C'H, distributions
Soil CH, concentrations decreased with depth at all four study sites (Fig. 2).Surface soil CH, concentrations were about 1.75 ppm, reflecting the ambient atmospheric concentration.Soil CH, concentrations at BS2 decreased rapidly with increasing depth to 0.14 ppm at 60 cm (Fig. 2a).Soil CH, concentrations at the remaining sites also showed a sharp decrease with increasing depth; however, CH, concentrations reached a minimum of about 0.10 ppm at a depth of 30 or 40 cm and remained constant at that concentration to 60 cm (Fig. 2b-d).
Soil CH, distributions in Fig. 2 suggest an extensive zone of CH, oxidation, no zone of CH, production and an atmospheric source of CH, for soil methanotrophs.In agreement with our data, Central Panamanian forest and agricultural soils had CH, concentrations at 20 and 40 cm that were less than half the atmospheric value (Keller et al. 1990), while monthly averaged soil CH, profiles in a mixed hardwood forest showed a continuous decrease from near-atmospheric values at 2 cm to < 0.25 ppm at 15 cm (Crill 1991).German forest and Canary Island volcanic soils had   CH, profiles to 70 cm (Born et al. 1990) that were remarkably similar to CH, distributions in Fig. 2. In contrast, soil CH, profiles to 20 or 50 cm in mixed mesophytic and spruce forests showed zones where CH, was depleted and enhanced relative to atmospheric values, indicating both production and consumption of CH, (Yavitt et al. 1990a).Soil CH, profiles and the following modification of Fick's first law (Campbell 1985) were used to estimate area-based rates of CH, oxidation: where JcH, is the CH, flux (g m-' s-'), DcH is the binary diffusion coefficient of CH, in air (0.194 cm2 s-I at 5 OCI Lerman 1979), AC,,, is the CH, concentration gradient at the air-soil interface determined by linear regression of the decrease in soil CH, with depth to 20 cm (g CH, cm-" cm-I), $g is fractional air-filled porosity (0 to 10 cm) and the constants 0.9 and 2.3 are recommended average values to account for tortuosity.Calculated rates of CH, oxidation at the four sites varied from 0.77 to 1.78 mg m-2 d-I (Table 2).Yavitt et al. (1990a) used soil CH, distributions to estimate a CH, oxidation rate of 2 mg m-' d-I in a mixed mesophytic forest soil.

Clzamber CH, time-course experiments
Headspace CH, decreased continuously from an initial concentration of about 1.8 ppm to a final concentration of 1.20 to 1.57 ppm in all 0.75 h static chamber time-course experiments (Fig. 3).Methane oxidation rates were calculated from chamber geometry and regression analysis of headspace CH, concentration versus time (Whalen & Reeburgh 1988).Most studies using static chamber experiments to assess rates of CH, oxidation in soils have fit a linear function to the concentration versus time data (e.g.Steudler et al. 1989), although an exponential model has also been used (Keller et al. 1990;Mosier et al. 1991).We employed a linear function because the goodness of fit (r2 = -0.96 to -0.98) was not significantly improved by use of an exponential model.
Methane oxidation rates in static chamber experiments varied from 0.22 to 0.62 mg mP2 d-l (Table 2), in close agreement with the May through September 1990 median CH, oxidation rates of 0.26 to 0.56 mg mP2 d-' reported for these same sites (Whalen et al. 1991).These CH, oxidation rates fall toward the low end of net CH, oxidation rates reported from other studies using static chamber techniques in non-waterlogged soils.Temperate evergreen and deciduous forests showed mean CH, oxidation rates of 0.4 to 4.15 mg m-2 d-' (Keller et al. 1983;Steudler et al. 1989;Crill 1991), while tropical forest soils had average CH, oxidation rates of 0.14 to 0.8 mg m-2 d-I (Keller et al. 1986; Goreau & de Mello  1988; Keller et al. 1990).Net CH, oxidation rates ranged to 2.7 mg rn-2 d-' in moist tundra (King et al. 1989; Whalen & Reeburgh 1990a, b;  Barlett et al. 1992), varied from 0.14 to 1.46 mg rnp2 d-' i n semiarid grasslands (Mosier et al. 1991) and averaged 1.25 mg rn-2 d-' in broadleafed savannah (Seiler et al. 1984).
Results of the 24 h time-course experiments in a free-air atmosphere indicated that soil microbes at all sites could consume CH, to threshold concentrations ranging from < 0.10 to 0.87 ppm (Table 3).A low threshold for CH, oxidation is not clearly correlated with a high in situ rate of CH, oxidation.Station NB2 had the lowest rate of CH, oxidation in static chamber experiments (Table I), but three additional stations at NB2 showed the lowest thresholds for CH, oxidation (Table 3).Stations at UP3A demonstrated high thresholds for CH, oxidation (Table 3), but weekly May through September sampling of four other permanent stations within each site type (Whalen et al. 1991) showed a significantly higher median rate of CH, oxidation at UP3A (0.40 mg mV2 d-') than at NB2 (0.26 mg mV2 d-I).The lack of a negative relationship between CH, oxidation thresholds and rates may result from a small sample size, local spatial variability in soil physical and biological characteristics and the small range in low rates of CH, oxidation.Overall, these data are consistent with results of similar chamber and jar experiments in moist tundra and landfill soils, which showed median thresholds for CH, oxidation ranging from 0.23 to 0.37 ppm (Whalen et al. 1992).Results of amended atmosphere experiments suggest that these soils have a high CH, oxidation capacity (Table 3).Methane concentrations in chamber headspaces were sub-atmospheric within 24 h following CH, addition in all but three experiments, despite adjustment of initial concentrations to levels -lo3-fold above ambient in some cases.Previous chamber and jar experiments on tundra and landfill cover soils (Whalen &  Reeburgh, 1990b; Whalen et al. 1990) show an immediate response to CH, amendments, and undetectable changes in response for periods up to a week.These observations suggests that CH,-oxidizing activity is present continously and that induction of activity is not important.The final headspace CH, concentration was higher than expected at NB2-CT3 and NB2-CT4, based on the results of experiments involving free-air atmospheres.Soil moisture content (0 to 10 cm) and temperature (mean to 13 cm) were similar at about 60% (w/w) and 13 "C in nearby sites on 27 July and 10 August, suggesting that differences in microbial population size or structure between dates rather than physical variables account for the observed patterns of CH, utilization.
Headspace CH, concentrations at t, and t, (between 1 and 2 h after CH, addition) in amended atmosphere experiments (+ -2000 ppm) were used to estimate potential CH, oxidation rates varying from 638 to 867 mg mP2 d-' for these sites (Table 2).

Laboratoly time-courses under a free-air atmosphere
Headspace CH, concentrations decreased in most laboratory experiments involving 10-cm core sections initially equilibrated with a free-air atmosphere.First order rate constants, k(d-I), were determined from a least squares fit of ln[CH,] versus time.Slopes were highly significant for 41 of the 44 core sections (p < 0.05; n = 5 or 6), indicating that most soil samples were capable of CH, oxidation to 60 cm.Methane oxidation rate constants were generally around -0.3 to -1.5 d-', although values as high as -25 d-' were found for some core sections at BS2.The highest (most negative) rate constants were generally observed for 10 to 20 cm or 20 to 30 cm core sections at all sites; no other pattern was noted.
Only a few soil CH, oxidation rate profiles have been reported.Methane consumption increased with increasing depth to 30 cm in generally waterlogged, moss-derived peats (Yavitt et al. 1990b) but showed no depth-dependence in peat pore water to 30 cm (Yavitt et al. 1988).
Methane oxidation was observed in core sections from the < 10 cm organic horizon in well-drained tundra soil (Whalen & Reeburgh 1990b), to the water table (12 cm) in arctic hummocks (Whalen et al. 1992) and to 8 cm in the inorganic cover soil of a retired landfill (Whalen et al. 1990).
The mass of CH, in each 10 cm soil zone was calculated from #c (Table I), the soil CH, profiles (Fig. 2) and the core volume.The in siti rate of CH, oxidation was calculated as the product of the first-order rate constant and the mass of soil CH,.Rates are normalized to soil mass and show a maximum in the 10 to 20 cm mineral zone at all sites except NB2 (Fig. 4).When data for each core segment are expressed on a volume basis, this subsurface maximum in the CH, oxidation rate is even more pronounced at BS2, AS2 and UP3A, and is evident at NB2 as well.This results from the relatively low surface p, (Table 1).Overall, the results indicate that CH, oxidation occurs to about 30 (Fig. 4a-c) or 50 cm (Fig. 4d) in these soils, and below that depth the process is limited by substrate availability (Fig. 2).
Methane oxidation rates for each 10 cm core section were summed to give area-based rates for each site.Methane oxidation rates varied from 0.04 to 1.5 1 mg m-' d-' (Table 2), with highest area-based rates occurring at BS2 due to the exceptional rate of CH, oxidation in the 10 to 20 cm soil zone (Fig. 4a).

Laboratory time-courses under an amended atmosphere
Most laboratory experiments involving 10 cm core sections amended with -500 ppm CH, showed a decrease in headspace CH, concentration over a 12 h period.A linear decrease in headspace CH, concentration indicated zero order consumption, that is, oxidation rates were maximum and independent of CH, concentration.Regression analysis gave significant slopes for 43 of 44 core sections (p < 0.05; n = 4).High slopes (k values) in free-air experiments were not always associated with high slopes in amended atmosphere experiments; coefficients of correlation (Kendall's t) were significant (n = 8 or 12; p < 0.05) only for BS2 and NB2 when data within each site type were compared.The lack of correlation between these variables at the other two sites likely reflects the small sample size coupled with the small range in data.However, we cannot discount the possible adverse effect of repeated sample handling or the possibility that genuine depth-dependent differences exist in the microbial consortium and its response to in situ and elevated substrate levels.
Maximum or potential rates of CH, oxidation for each core section were calculated from the soil mass, jar headspace volume, and slope of the linear regression of CH, concentration versus time.Potential CH, oxidation rates normalized to soil mass show a maximum in the 10 to 20 cm zone at BS2 only (Fig. 5), in contrast to calculated in situ CH, oxidation rates normalized to soil mass for these same core sections (Fig. 4).Volume-based potential rates of CH, oxidation were maximum in the 10 to 20 cm soil zone for all cores, in agreement with data for volume-based in situ rates of CH, oxidation for these cores.The data indicate that all soil intervals except the 0 to 10 and 30 to 40 cm sections at BS2 (Fig. 5a) and the 50 to 60 cm section at NB2 (Fig. 5c) are capable of rapid CH, oxidation if substrate is available.
Potential rates of CH, oxidation varied from 0 to 1350 ng (g soil)-' d-' (Fig. 5) and depended on both the size of the CH,-oxidizing microbial population and soil physical characteristics.The average oxidation rate of about 140 ng CH, (g soil)-' d-' is roughly lo3-fold lower than the rate of 130 to 280 p g CH, (g soil)-' d-' we calculate from data given by Megraw & Knowles (1987) for humisol soil not previously exposed to exogenous CH, and is lo2-fold lower than the V, (maximum oxidation rate) of 60 p g CH, (g soil)-' d-' determined by kinetic analysis of CH, utilization by composites of a landfill cover soil (Whalen et al. 1990).Data from all 10 cm depth intervals were summed to estimate area-based potential CH, oxidation rates at each site.Rates varied from 59 to 116 mg CH, m-2 d-' (Table 2).
The subsurface maximum in volume-based rates of in situ and potential CH, oxidation is surprising.It may result from the large surface area in the less porous mineral soil (Table 1) or soil moisture.We found little or no CH, oxidation in the 0 to 2 cm horizon in well-drained tundra (Whalen & Reeburgh 1990b) and landfill cover soils (Whalen et al. 1990), regardless of organic content.A reduction in soil moisture content from the ambient level of 11% to 6% severely reduced CH, oxidation in a landfill cover soil (Whalen et al. 1990), suggesting a high sensitivity to drying.Soils adjacent to the atmosphere experience wide fluctuations in moisture, which may prevent establishment of a vigorous community of methanotrophs.Atmospheric CH, (the source of energy, reducing equivalents, and cell carbon for soil methanotrophs) is available below the surface organic soil horizon.This zone may function as a highly permeable buffer to variations in subsurface soil moisture.

Laboratory experiments on 14CH4 oxidation
All 10 cm core segments exposed to 14CH4 except the two deepest (40 to 60 cm) at NB2 showed CH, oxidation, as evidenced by the appearance of 14C02 and 14C-labeled biomass (Fig. 6).These data confirm that decreases in headspace CH, resulted from microbial CH, oxidation in free-air (Fig. 4) and amended atmosphere (Fig. 5) time-course experiments involving core sections to 60 cm.An average (f 1 S.D.) of 39 + 30% of the added label was used and 101 f 7% was recovered in these experiments.Data from each site were summed to give area-based CH, oxidation rates that varied from 0.87 to 2.32 mg mV2 d-I (Table 2).
The fraction of oxidized CH, that was incorporated into microbial biomass (m2 basis) was similar among sites, ranging from 34 to 43%, with the balance respired as I4CO,.There were no depth-dependent differences in the partitioning of label between 14C-biomass and I4CO, (Fig. 6).No pattern is evident for other studies examining the distribution of end products of 14CH4 oxidation in soils, and our data are well within the range of results given in earlier investigations.For example, the fraction of oxidized 14CH4 that was converted into biomass was 15 to 22% and 50 to 68% for peat (Yavitt et al. 1988;1990b) and forest soils (Yavitt et al. 1990a) in Appalachia.In addition, 54%, 69% and > 70% of the 14CH4 oxidized in tundra (Whalen & Reeburgh 1990b), landfill (Whalen et al. 1990) and cultivated humisol (Megraw & Knowles 1987) soils, respectively, was recovered as 14C-biomass.
The CH, relaxation depth, 6, is the order of magnitude of the projected path length in the soil for a CH, atom before oxidation (Sogaard-Hansen & Darnkjaer 1987).Methane relaxation depths at each site were calculated from soil CH, distributions (Fig. 2) and the relationship C, = C, exp(-z/t) given by Born et al. (1990), where C, and C, are CH, concentrations at depth z and in the atmosphere.Values of [ ranged from 23 to 39 cm, so we expected little or no CH, oxidation below this depth.However, amended atmosphere and radiocarbon jar experiment indicated that many soils were capable of significant CH, oxidation to 60 cm (Fig. 5b-d and Fig. 6b, d).Addition of substrate to resting microbial cells can stimulate metabolic activity (Roszak & Colwell 1987), which may account for the CH, oxidation observed in core segments taken below <.
Another possible explanation for the high CH, oxidizing potential at depth is fortuitous metabolism by nitrifymg bacteria.Nitrifiers exist in soils and marine nitrifiers have been demonstrated to oxidize CH, (Ward 1987(Ward , 1990)).Diffusion and percolation may supply sufficient NH,, O,, and CO, to support chemoautotrophic NH, oxidation deep within the soil.Addition of CH, to soil cores in jar experiments may provide a competing substrate for ammonia monooxygenase, which has a wide specificity (Bedard & Knowles 1989).

222Rn-calibrated CH, Jluxes
Static chamber experiments assessing soil 222Rn flux showed a continuous increase in headspace 222Rn over the 0.5 h time-course (Fig. 7), in contrast to similar experiments measuring CH, flux (Fig. 2).Rn-222 fluxes varied from 0.14 to 0.45 atom cm-' s-' and averaged (f 1 S.D.) 0.35 f 0.14 atom cm-' s-'.The mean 222Rn emission rate agrees with the average flux of 0.35 atom cm-2 s-' reported for tropical forest soils (Trumbore et al. 1990) and lies between the mean fluxes of 0.22 and 1.07 atom cm-2 s-' found in sandy and clayey temperate forest soils (Dorr & Munnich 1990).These studies also used static chambers to determine "'Rn fluxes.
Soil 222Rn activity increased with depth (Fig. 8), consistent with the observed flux (emission to the atmosphere).The positive slopes associated with time-courses for soil "' Rn fluxes (Fig. 7) and depth distributions of soil 2"Rn (Fig. 8; z positive downward from soil surface) as well as the negative slopes observed for soil CH, profiles (Fig. 2) were also reported for German soils showing net CH, consumption (Born et al. 1990;Dorr & Munnich 1990).However, our 222Rn profiles, derived from limited data, deviate from a theoretical steady state profile that decreases exponentially to the soil surface from a constant value at depth (Dorr &  Miinnich 1987).Rn-222 profiles similar to those in Fig. 8 have also been reported for stratified tropical forest soils (Trumbore et al. 1990).
Soil gas (222Rn and CH,) profiles and 222Rn chamber experiments were used to calculate 222Rn transport-corrected rates of area-based CH, oxidation according to Dorr & Miinnich (1 987):  1990) calculated AC, from a third order polynomial fit to 222Rn distributions in stratified soils similar to those in Fig. 8, but noted that the derivative was approximately equal to the slope of a linear fit to data in the 0 to 20 cm soil zone.Methane oxidation rate estimates from the 222Rn method varied from 0.1 9 to 1.60 mg m-2 d-' (Table 2).

Comparison of information from each measure
The 222Rn technique provides an independent, indirect check on chamber CH, flux determinations.Both methods are nondestructive, so permanent sampling stations can be visited repeatedly.The 222Rn method gave CH, oxidation rate estimates that were 65 to 220% of the static chamberderived estimates (Table 2).The methods appear to compare poorly when data are cast as percent difference, but we feel that agreement is satisfactory for two reasons.First, both methods clearly indicate that CH, fluxes are low, but both methods also have the potential for large errors.An error in determining a small concentration change with time at sub-atmospheric CH, concentrations in static chamber experiments results in a relatively large error in the flux estimate.A 0.1 ppm error in CH, concentration determination at t,, or t, in a static chamber flux experiment will bias the slope of the regression equation used to calculate flux, and will reduce the CH, oxidation rate estimate, for example, for AS2.The rate would drop from 0.55 (Table 2) to 0.45 mg m-2 d-I, or 20%.The largest source of error in our 222Rn technique probably results from determining AC,, in Equation 2 using two points (0 and 20 cm) in a heterogeneous soil zone (Table 1).A 20% error here is not unreasonable and could increase the CH, flux estimate at AS2 from 0.35 (Table 2) to 0.44 mg m-2 d-'.Steps were taken in the 1991 field season to reduce both sources of error.The number of gas samples collected in the 0 to 20 cm soil zone was increased to better estimate AC,, and the sampling interval was decreased in 0.75-h static chamber CH, flux determinations to add observations and reduce the influence of outliers.
The second indication that the two techniques did not yield appreciably different CH, oxidation rate estimates relates to spatial heterogeneity in microbial activity.Experiments at each site were conducted simultaneously, but at different stations.Hence, differences in rate estimates may simply reflect local variability.The 222Rn flux estimates in Table 2 lie well within the range of estimates determined by weekly 0.75-h time-courses for chamber CH, fluxes at these same stations in 1990 and 1991 (Whalen et al. 1991 and unpublished).A better experimental design involves use of a single station to determine CH, and 222Rn fluxes simultaneously with a static chamber.Dorr and Miinnich (1990) found that the annual mean 222Rn-calibrated CH, flux exceeded the direct static chamber flux by a factor of 4 with this approach.The difference was attributed to a reduction in the concentration gradient for soil CH, during static chamber experiments.
Rn-222 flux and profile measurements also give information on the effective soil diffusivity of the gas under investigation.The effective diffusivity for CH, (PC,) is calculated (Born et al. 1990) as: Values of PCH4 were 0.012, 0.013, 0.053 and 0.078 cm2 s-' for NB2, AS2, UP3A and BS2.This rank order is consistent with the expected order based on 4g and p, (Table 1).Effective diffusivities in these soils were lower than DCH4 (0.194 cm2 S-l) by a factor of 2 to 16, in reasonable agreement with the expected reduction (factor of 2 to 10) of free-air diffusivities of gases in unconsolidated porous media (Lerman 1979).
Values of PCH4 for deciduous sites are within the range reported by Born et al. (1990) for various European soils (-0.001 to 0.03 cm2 s-I), whereas values for coniferous sites are higher.Effective diffusivities derived from 222Rn studies (Equation 3) for all sites under a range of moisture and temperature regimes can be used to estimate CH, oxidation rates from Fick's first law (JCH4 = PCH4 ACCHI) and soil CH, profiles.Calculation of PcHq for a wide range of soils provides a useful index of the ability of moist soils to consume atmospheric CH,; PcH4 limits CH, oxidation in nonwaterlogged soils where the atmosphere is the sole CH, source for soil methanotrophs.Methane oxidation rates calculated from soil CH, profiles and Equation 1 were consistently 1.5 to 3.5-fold higher than estimates from static chambers (Table 2).This is a nondestructive (non-jar) technique that gives a rough estimate of in situ CH, oxidation rates.However, this estimate requires no information beyond that needed for other CH, oxidation rate estimates, as $g and ACcH4 are necessary components of CH, oxidation rate estimates from jar and 222Rn experiments, respectively.Sources of error in Equation 1 may stem from the determination of $g in the stratified 0 to 10 cm soil zone or the application of "average" coefficients to correct DCH4 for tortuosity.Values of Dc,4(0.9)$,2.3 in Equation 1 vary from 0.041 to 0.117 cm2 s-l and overestimate PCH4 if $g is accurately determined.We feel that the error lies in the coefficients used to adjust DCH4.These depend on both soil structure and moisture; Currie (1984) demonstrated that for a single soil type of varying moisture content no single relationship was satisfactory.In contrast, we are confident in our estimates for ACcH, in Equation 1 because soil CH, distributions similar to those in Fig. 2 were observed throughout the 1 990 and 199 1 thaw seasons.
Rates of in situ CH, oxidation estimated from jar experiments were lower than chamber-derived estimates at all sites except BS2 (Table 2), in contrast to the consistently high CH, oxidation rates estimated from soil CH, profiles.Decreased CH, oxidation rates in jar experiments relative to chamber-derived estimates are probably a consequence of disturbance of the soil during sample collection.The effects of sample collection, transport, and confinement on rates of microbial activity are well-documented, but totally unpredictable (Karl 1986).This method does not allow repeated sampling at a fixed station, unlike chamber methods.However, this method identifies soil zones of microbial activity, important information for modeling the response of soil methanotrophs to climate change.
The elevated rate of in situ CH, oxidation in the jar vs. chamber experiment at BS2 (Table 2) is consistent with our suggestion that sample handling reduced microbial activity.The high CH, oxidation rate estimate in the jar experiment clearly results from a substance "hotspot" (Fig. 5a).The soil CH, profile taken in the same area provides further evidence that this hotspot is a localized phenomenon; a more rapid decline in soil CH, concentration is not apparent in the 10 to 20 cm zone (Fig. 2a).Localized zones of enhanced microbial activity have frequently been found for soil denitrification (e.g.Parkin 1987), but we are unaware of similar reports for soil CH, oxidation.
Chamber experiments involving amended atmospheres are a nondestructure method of estimating potential CH, oxidation rates.Although results of these experiments suggest a high capacity for CH, oxidation in soils at all sites (Table 3), we cannot dismiss these potential rates (Table 2) as overestimates for two reasons.First, equilibration as well as microbial oxidation decreases the headspace CH, concentration immediately after chamber amendment.Added CH, can be expected to equilibrate with gases in the chamber headspace and air-filled pore space in surface, organic soil within a few minutes (Whalen & Reeburgh 1990b), but we have no information concerning rates of equilibration with the more firmly packed subsurface mineral horizon.Therefore, we conservatively used time points > 1 h after CH, addition to the chamber when calculating potential CH, oxidation rates.Second, lateral diffusion beneath soil collars may have allowed added CH, to escape the chambers.Relaxation depths ranged from 23 to 39 cm.Chamber collars extended to a soil depth of 15 or 20 cm, so loss of chamber CH, by diffusion was possible.We feel these losses were minimal.Chamber bases were firmly placed well into mineral soils of low 4g (Table 1) and final headspace CH, concentrations (Q in all free-air and most amended atmosphere chamber experiments were well below atmospheric concentrations (Table 3).
Amended atmosphere jar experiments give an additional, conservative estimate of potential rates of CH, oxidation.The technique is destructive and sample handling may reduce rates of microbial activity, but confinement insures retention of added headspace CH,.Jar experiments gave potential CH, oxidation rates that were only 7 to 18% of rates estimated from chamber experiments (Table 2).Nonetheless, these CH, oxidation rates of 59 to 116 mg m-* d-' were at least lo2-fold higher than chamber estimates under a free-air atmosphere, providing convincing evidence for a high potential for CH, oxidation at all sites.We consider potential rates to be an intensive property of the system, while capacity is an extensive property.
The I4CH, experiment is also destructive, but it is the only technique capable of assessing the end products of CH, oxidation.Moreover, it unequivocally demonstrates that microbial CH, oxidation is responsible for loss of headspace CH, in other experiments.Sample handling may reduce the rate of CH, oxidation, and increasing the CH, pool by as much as 1O3-fold by tracer addition and equilibration with the atmosphere could enhance CH, oxidation.Clearly, I4CH4-derived rates are not directly comparable to in situ rates measured using chamber techniques (Table 2).

Summary
These measures provide a range of information about CH, oxidation in boreal forest soils.They are not directly comparable in most cases, but give unique insights into the controls on microbial CH, oxidation.The short-term unamended chamber measurements, the 222Rn-calibrated flux measurements, and the soil CH, distributions show independently that the soils are capable of oxidizing atmospheric CH,.The jar experiments with free-air headspaces and soil CH, profiles give the depth distribution of CH, oxidation rates.The jar experiments and chamber measurements with free-air headspaces show that CH, oxidation occurs at low thresholds (G0.9 ppm).The 14CH4-amended jar experiments show that the process is microbially-mediated and give the distribution of end products of CH, oxidation.The chamber and jar experiments with amended atmospheres give information on potential CH, oxidation rates and CH,-oxidizing capacity.Methane oxidation in these moist soils consumes atmospheric CH, and may serve as a negative feedback on atmospheric CH, increases.
soil gradient measurements.where CH, fluxes were estimated from soil CH, distributions; (c) static chamber experiments estimating CH, flux from soil distributions of 222Rn and CH, and the change in headspace 222Rn activity over time; (d) experiments assessing the change in headspace CH, concentration over 24 h in static chambers initially equilibrated with amended ( f 2 0 to 2000 ppm CH,) and free-air atmospheres; (e) jar experiments exposing soil core sections to atmospheric CH, concentrations (1.7 ppm);

Fig. 2 .
Fig. 2. Soil CH, distributions at experimental sites.Sites and sampling dates as in Fig. 1

Fig. 3 .
Fig. 3. Static chamber CH, time-course measurements at experimental sites.Station 1 (CTI) was sampled at each site.Sites and sampling dates as in Fig. 1.

Fig. 4 .
Fig. 4. Methane oxidation rates derived from jar experiments on 10 cm core segments from experimental sites.Each segment was exposed to a free-air atmosphere.Rates were corrected to in situ CH, concentrations and are plotted at the midpoint of each depth interval.Sites and sampling dates as in Fig. 1.

Fig. 5 .
Fig. 5. Potential methane oxidation rates derived from jar experiments on 10 cm core segments from experimental sites.Each segment was exposed to an amended (+500 ppm CH,) atmosphere.Rates are plotted at the midpoint of each depth interval.Sites and sampling dates as in Fig. 1.

Fig. 6 .
Fig. 6. Results from 14CH, addition experiments on core sections from experimental sites.Solid line is '4C02 production rate; Dashed line is 14C-biomass production rate.Sites and sampling dates as in Fig. 1.

Table I .
Soil properties at four sites in the Bonanza Creek Experimental Forest.

Table 3 .
Methane oxidation threshold and capacity experiments.below detection limit (0.10 ppm).Initial (t,,) and final (t,) headspace CH, concentrations in static chamber experiments.Chambers were initially equilibrated with a free-air atmosphere in CH, oxidation threshold experiments.Chamber headspace CH, concentrations were adjusted to -20, 200 and 2000 ppm in oxidation potential experiments.Data fort, were taken after 24 h. =