Concentration and δD of molecular hydrogen in boreal forests: Ecosystem‐scale systematics of atmospheric H2

We examined the concentration and δD of atmospheric H2 in a boreal forest in interior Alaska to investigate the systematics of high latitude soil uptake at ecosystem scale. Samples collected during nighttime inversions exhibited vigorous H2 uptake, with concentration negatively correlated with the concentration of CO2 (−0.8 to −1.2 ppb H2 per ppm CO2) and negatively correlated with δD of H2. We derived H2 deposition rates of between 2 to 12 nmol m−2 s−1. These rates are comparable to those observed in lower latitude ecosystems. We also derive an average fractionation factor, α = D:Hresidual/D:Hconsumed = 0.94 ± 0.01 and suggestive evidence that α depends on forest maturity. Our results show that high northern latitude soils are a significant sink of molecular hydrogen indicating that the record of atmospheric H2 may be sensitive to changes in climate and land use.


Introduction
[2] Molecular hydrogen (H 2 ) is the second most abundant reduced gas in the atmosphere (after methane) with a globally averaged mixing ratio of $530 ppbv [Novelli et al., 1999]. Its largest source is believed to be photochemical oxidation of methane and non-methane hydrocarbons; other recognized sources include biomass burning, fossil fuel burning, and ocean degassing. Hydrogen is unusual for a trace gas with anthropogenic sources in that the concentration in the southern hemisphere is greater (by $3%, relative) than that in the northern hemisphere; furthermore, seasonal amplitudes increase with latitude in the northern hemisphere. Combined, these observations indicate that land surface in the northern hemisphere is an important sink of atmospheric H 2 . Oxidation by OH radicals is the final recognized component of the H 2 budget.
[3] The budget and spatial and temporal distribution of atmospheric H 2 suggest that it is a unique tracer of the interplay of processes relating climate and land use change.
Previous studies have examined the uptake of H 2 in flux chamber experiments on tropical and temperate soils [Seiler, 1978;Conrad and Seiler, 1985;Yonemura et al., 1999;Yonemura et al., 2000a;Yonemura et al., 2000b], and the isotopic effects associated with that uptake [Ehhalt et al., 1989;Gerst and Quay, 2001]. However, there are no prior observations of ecosystem-wide rates and isotopic systematics of H 2 uptake by soils, and no observations of any type of soils from high northern latitudes. We report here the results of a study of the concentration and stable isotope systematics of H 2 in interior Alaskan boreal forests during July of 2001. This study is one in a series examining the atmospheric H 2 budget by way of isotopic analyses using a new method of molecular hydrogen recovery and mass spectrometry on sub-liter-sized air samples [Rahn et al., 2002].

Site Description: Delta Junction, Alaska
[4] Delta Junction is in the interior of Alaska (lat. 63°48 0 N, long. 145°06 0 W, elevation 470 m) in the Yukon-Tanana Uplands on the north eastern flank of the Alaska Range. It is a sub-arctic boreal region with both semi-dry and wet, permafrost-laden soil zones. The area experiences moderate temperatures and precipitation during the summer months (May-August) and exceedingly cold and dry conditions during the winter. Average maximum and minimum annual temperatures between 1937 -1999 are 15.6°C (July) and À20°C (January), with annual precipitation of 30 cm falling as rain and 113 cm as snow [O'Neill, 2000]. Fire and agriculture are the disturbance regimes that dominate this area, with current fire recurrence intervals averaging 100 years. Black spruce (Pinus mariana), feather moss (Hylocomium), and cranberry (Vaccinium macrocarpon) dominate ecosystems that have not been disturbed by fire within 60 years. Quaking aspen (Populous tremuloides), firemoss (Marchantia polymorpha), blueberry (Vaccinium ovaliflorum), and willow (Salix sp.) dominate ecosystems recovering from more recent burns [Viereck et al., 1983]. We chose four sites that represent a chronosequence of ecosystem age after fire: two mixed forests of ages 14 and 45 years (hereafter referred to as sites 87B and 56B, or 'burn' sites) and two mature spruce forests of 140 and 160 years age (sites 94C and 12C, or 'mature' sites). Soil moisture and soil temperature varied with age such that 87B was the driest, warmest site and 12C was the wettest, coolest site. Canopy height in both mature spruce stands was uniform at $3 to 4 m. The canopy of burn site 87B was open and uneven with maximum aspen height of $3m and scattered snags and deadfall remaining from the original fire. Burn site 56B was a more developed stand of aspen GEOPHYSICAL RESEARCH LETTERS, VOL. 29, NO. 18, 1888, doi:10.1029/2002GL015118, 2002 Copyright 2002 by the American Geophysical Union. 0094-8276/02/2002GL015118 with maximum height of $5 m and a relatively closed canopy.

Sample Collection and Analysis
[5] We collected samples over the course of one week in mid-July of 2001. Flask samples were collected for measurements of atmospheric H 2 concentration ([H 2 ]) and dD during nighttime inversions in the canopies of both mature and burn sites; concurrent sampling of CO 2 concentration ([CO 2 ]) was performed in the field with a LiCor 6200 IR analyzer. Air was pulled from heights of 0 to 2 meters in the forest canopy through double-valve glass flasks of ca. 800 ml volume, and finally into the IR cell of the Licor analyzer. A second set of twelve samples was collected in the same manner over a period of 24 hours at a tower site in a 14 year old mixed forest; in this case, samples were collected at four hour intervals at ground level (0.8 m) and above the canopy (10 m).
[6] Flask samples were analyzed for concentration and dD of H 2 within one month of return to the laboratory using the method of Rahn et al. [2002]. Briefly, we condense most components of a whole air sample in a cold trap held at 30 K, leaving only H 2 , He, Ne and a small fraction of N 2 as vapor. The non-condensed fraction is then collected on molecular sieve in a valved glass finger at liquid N 2 temperature and removed from the extraction line. It is then connected in line to the He flow at the inlet of a Finnigan Delta Plus XL mass spectrometer where it is focused on an additional molecular sieve trap before entering the mass spectrometer ion source as a single, time resolved peak. The ion beam intensities are compared to those of a reference gas to determine the D/H ratio of the sample. The uncertainty of our isotopic results is ±7% [ibid]. In addition, the integrated area of the sample peak is compared to the areas of a series of reference aliquots of known molar quantities, allowing calculation of molarity of the H 2 sample. Given that sample flask volume and pressure are known, this constrains H 2 concentration in the whole air sample. Uncertainty for analysis of H 2 concentration is $5% [ibid].

Variability of [CO 2 ], [H 2 ], and DD of H 2
[7] The diel variations of [CO 2 ], [H 2 ] and dD of H 2 within and above the forest canopy at site 87B from July 18, 1630 hr., to July 19, 1300 hr., are shown in Figure 1. Nighttime highs of [CO 2 ] reached 635 ppm at 80 cm height at 0149 hr., coincident with an [H 2 ] minimum of 135 ppb. Above the canopy, [CO 2 ] was also elevated to >400 ppm during nighttime while [H 2 ] above canopy was <400 ppb, indicating that nighttime inversion over this period was stable enough to influence even the above-canopy air mass. dD of H 2 at 80 cm was high compared to the average for the whole data base at 0149 hr., coincident with sub-ambient [H 2 ]. Variations in dD of H 2 in the above-canopy air mass is not as apparent. Samples from sites 56B, 94C and 12C were collected within the forest canopy at heights less than 2 m, typically between the hours of 2200 and 0200 and display relationships among [CO 2 ], [H 2 ] and dD of H 2 similar to those described for samples collected at night within the canopy at site 87B. Data for samples collected from all sites are combined in Figure 2, although only data for samples collected between 2000 and 0600 hr are included for 87B in order to emphasize samples collected during nighttime inversion conditions.

H 2 Uptake Rates
[8] Samples collected during nighttime inversions exhibited strong negative correlations between H 2 concentration and CO 2 concentration ( Figure 2) with slopes for the burned and mature sites averaging À1.26 and À0.84 ppb H 2 per ppm CO 2 , respectively ( Table 1). Accumulation of CO 2 in near-surface air masses during nighttime inversions in our study area reflects root and microbial respiration in soils, whereas H 2 uptake by soils under any conditions is believed to be due to abiotic soil hydrogenases [Conrad et al., 1983]. Therefore, the relationship between H 2 and CO 2 observed in our samples likely does not reflect a single chemical or biological reaction (e.g., the metabolism of a single organism). However, the consistent relationship between the two at any one site strongly suggests that they do share a common mechanism for interaction between the biological communities responsible for H 2 consumption and CO 2 production. We propose that this common mechanism is gas exchange between soil and overlying air. Based on this assumption and given known fluxes of CO 2 emission from soils at our study site and the physics of air-soil gas  exchange, we can use the correlations in Figure 2 to estimate rates of H 2 uptake at our sites.
[9] We estimate the rate of H 2 uptake through the following steps: First, we accept that CO 2 emission from soils in environments and under conditions relevant to our study sites are rate limited either by molecular diffusion (in the case that exchange is rate limited by diffusion through static soil gas) or eddy diffusion (in the case that exchange is rate limited by eddy diffusion in the boundary layer immediately above the soil-atmosphere horizon). In either case, its flux from the soil into overlying air can be described by Fick's first law: where F is the flux (in mmoles m À2 s À1 ), D is the diffusion coefficient (in m 2 s À1 ), C is the concentration (in m moles m À3 ), and z is vertical distance (in m). If we assume that the flux of H 2 into soil is also diffusion limited, and that diffusion coefficients for H 2 and CO 2 are independent of each other, the flux of H 2 can be approximated by the ordinary differential equation: Thus, we can determine F H2 if we know F CO2 , the ratios D H2 / D CO2 , and d[H 2 ]/d[CO 2 ]. F CO2 was measured by flux chamber techniques at or near the sites where the flask samples are collected; over the time period of our study, F CO2 in the recently burned, mixed forests averaged 1.4 ± 0.1 mmol m À2 s À1 and in the mature forest 3.5 ± 0.7 mmol m À2 s À1 . The relative gradients, d[H 2 ]/d[CO 2 ], we take from Figure 2 and Table 1. The ratio of diffusion coefficients for H 2 and CO 2 depends on whether the mixing is via molecular or eddy diffusion. In the case of eddy diffusion, the ratio of diffusion coefficients should be near unity and the H 2 fluxes calculated from Equation 2 would be À1.8 ± 0.5 and À2.9 ± 0.6 . nmol m À2 s À1 at the burned and mature stands respectively. If instead soil gas-atmosphere exchange is rate limited by molecular diffusion in the soil, the ratio of diffusion coefficients will be set by the respective soil gas diffusivities. Soil gas diffusivity is a function of molecular diffusivity in air as well as soil properties and water content. In the case where soils are dry, D H2 /D CO2 will reduce to the ratio of the respective diffusivities in air. At 20°C, the molecular diffusivities in air of H 2 and CO 2 are 0.627 and 0.160 cm 2 s À1 , respectively, yielding a ratio of diffusion coefficients of 3.92. As soils approach saturation, D H2 /D CO2 will approach the ratio of the respective diffusivities in H 2 O. At 20°C, the molecular diffusivities in H 2 O of H 2 and CO 2 are 4.58 Â 10 À5 and 1.67 Â 10 À5 cm 2 s À1 , respectively, yielding a ratio of diffusion coefficients of 2.74. Diffusion coefficient ratios may approach this saturated case although it should be considered an extreme end member since the absolute values of the diffusion coefficients will essentially lead to a shutdown in soil -atmosphere exchange under fully saturated conditions. H 2 fluxes calculated for the dry soil case are À6.9 ± 2.0 nmol m À2 s À1 and À12 ± 2.3 nmol m À2 s À1 for burned and mature forests, respectively and for the saturated soil case are À4.8 ± 1.3 and À7.9 ± 1.6 nmol m À2 s À1 . The results calculated for the different diffusive regimes are tabulated in Table 2. Yonemura et al. [2000b] showed, based on experiments, that variations in soil properties lead to predicted variations in H 2 uptake rates and soil gas diffusivity; given this result and the fact that both respiration of CO 2 and uptake of H 2 take place in the soil column, we conclude that the calculations assuming molecular diffusion as the rate limiting step are the most plausible results.
[10] The destruction rate of trace gases in soils depends on their concentration in overlying air; as a result, rates of gas uptake by soils are commonly reported in terms of deposition velocity (v d ). Gerst and Quay [2001] report deposition velocity for H 2 of 0.058 cm s À1 in second growth coniferous forest in Seattle, Washington and Yonemura et al. [2000a] report v d of H 2 in a mixed forested area of Japan to be 0.05 to 0.08 cm s À1 over a 12 month period. We calculate v d for the burn and mature forests at our Alaskan sites to be 0.044 ± 0.013 cm s À1 and 0.073 ± 0.015 cm s À1 , respectively, in the case of molecular diffusion control in dry soil; agreement of these values with previous estimates for similar soil types further supports our preference for the values calculated assuming molecular diffusion is the rate limiting step for soil gas-atmosphere exchange. Deposition velocities of 0.011 ± 0.003 and 0.019 ± 0.004 cm s À1 are calculated for the burn and mature sites, respectively, in the alternative case of eddy diffusion control. Previous estimates of deposition velocities for H 2 in other soil types include 0.131 cm s À1 in a South African savanna [Conrad and Seiler, 1985], 0.010 and 0.016 cm s À1 in a South African and Namibian desert soils [ibid ], and 0.00 to 0.09 cm s À1 in an arable field in Japan; in this last case, deposition velocity decreased with increasing pore water and became negligible in freshly irrigated, saturated soils [Yonemura et al., 1999].

Isotopic Fractionation During H 2 Soil Uptake
[11] We find that decreasing H 2 concentrations during nighttime inversions are correlated with increasing dD of the remaining H 2 (Figure 3). If we consider this to be a single stage, irreversible loss process, then the fractionation associated with uptake can be estimated by a Rayleigh distillation model. In simplified form, the Rayleigh model can be expressed as where dD and dD o are the measured and initial dD values of H 2 , 'f' is the fraction of initial H 2 remaining after uptake, and  e is the enrichment factor, which is related to the fractionation factor, a (defined as D/H consumed /D/H residual ), by the equation e = 1000 . ( aÀ1). Based on linear regressions of data for each of the individual sites in Figure 3, we find e = À74.5 ± 15% (R 2 = 0.79) and À78.0 ± 26% (R 2 = 0.70) for burn sites 87B and 56B and e = À27.0 ± 31% (R 2 = 0.20) to À49.7 ± 7.6% (R 2 = 0.91) for mature sites 12C and 94C. The calculated fractionation factor associated with uptake appears to be greater for burned vs. mature sites, perhaps indicating that low uptake rates involve greater isotopic fractionation than fast uptake rates. However, the standard deviations of regressions for each site in Figure 3 are large enough that this result, though suggestive, cannot be considered statistically significant. If data for all sites are regressed together, we find e = À59 ± 11% (R 2 = 0.59), comparable to the values of À57 ± 24% reported by Gerst and Quay [2001] and À60% reported by Ehhalt [1989].

Discussion
[12] We draw several general conclusions based on our observations of H 2 and dD of H 2 in the boreal forest system of interior Alaska. First, H 2 levels during nighttime inversions within boreal forest canopies fell to as low as 25% of the average tropospheric concentration indicating that these ecosystems act as net sinks of atmospheric H 2 . Furthermore, correlation of H 2 with respired CO 2 in these canopies provides us with the means to interpret the results in terms of net flux with rates that are comparable to those previously determined for soils in temperate and tropical regions, whether eddy diffusivity or molecular diffusivity is the controlling factor. Finally, the data in Table 2 suggest that there may be observable differences in the rate of uptake between young mixed forests and mature stands. Given that fire frequency may increase with increasing average temperatures and changing land-use patterns in boreal regions [Kasischke et al., 1995], our results suggest that the capacity for H 2 soil uptake may be modified in the future.
[13] The cause of the difference in rate of H 2 uptake between recently burned and mature forests could have several explanations. We note Conrad and Seiler [1985] found H 2 deposition velocities are positively correlated with soil organic carbon. Total carbon in soils from our 'mature' sites averages 217 Mg ha -1 whereas that in our 'burned' sites averages 177 Mg ha À1 [O'Neill, 2000]. Furthermore, 70% of the carbon in the mature sites near Delta Junction is in the organic horizon, whereas 92% of carbon in recently burned sites is in the mineral horizon and may have a significant contribution of inorganic carbon. Thus, recent burn sites have systematically lower soil organic carbon inventories than mature sites and would therefore be predicted to have lower deposition velocities for H 2 , all other parameters being equal. On the other hand, mature spruce stands typically retain more soil moisture which would tend to reduce the H 2 uptake at these sites. The true variability between and within these sites will require more detailed future field studies investigating additional influencing components of the ecosystems such as the potential for in situ production of H 2 via nitrogen fixation.
[14] The isotopic fractionations associated with uptake of H 2 in both 'burned' and 'mature' forests in this study are comparable to those observed in previous studies of midlatitude forests and cultivated grass [Ehhalt et al., 1989;Gerst and Quay, 2001]. This similarity across ecosystems and between flux chamber experiments and our field study suggests that the common value of e observed in these studies (ca. À60%) is a reasonable choice for modeling the role of soil uptake in the global isotopic budget of H 2 . However, we re-iterate that our results suggest variability in e that should be examined with more detailed study covering a wider range of ecosystems.