Drying and Wetting Effects on Carbon Dioxide Release from Organic Horizons

,

temperature, so that variation in soil moisture potentially becomes more important at higher temperatures.Low water contents can limit the growth rate and activ-T he O horizon of many temperate and boreal forests ity of soil microorganisms as well as the diffusion of stores a large amount of soil organic matter that is nutrients and C substrates in water films, whereas high not physically protected against microbial decomposiwater contents can restrict the oxygen supply of microtion.It is thought that the turnover time of soil organic organisms (Skopp et al., 1990; Howard and Howard, matter is mainly a result of litter quality, soil fertility, 1993).In many temperate and boreal regions, O horimineralogy, soil texture, temperature, and soil moisture zons are subjected to rapid changes in water content (Davidson et al., 2000).In general, the storage of soil during the growing season.Despite the importance of organic matter increases at high water contents and the O horizon as an organic C reservoir, most drying/ decreasing temperature, resulting in thicker O horizons wetting studies have focused on mineral soils.From with increasing latitude (Trumbore, 2000).In undislaboratory studies it is well known that wetting of dry turbed, mature forests, the amount of soil organic matter mineral soils causes a pulse of CO 2 for a few days (Birch, is usually near steady state, with similar rates of decom-1958;Orchard and Cook, 1983;Kieft et al., 1987).The position and C inputs on decadal time scales.However, CO 2 pulse has been attributed to improved C availability variation in temperature and soil moisture may alter from killed microbial biomass and increased availability rates of decomposition and litter production on seasonal of soil organic C (SOC) due to increased exposure of and annual time scales.Hence, soil respiration, which organic surfaces during drying (Birch, 1958; Orchard is the largest flux of CO 2 from the terrestrial biosphere and Cook, 1983, van Gestel et al., 1991).The response to the atmosphere (Schimel, 1995), is likely to respond to wetting may vary with soil texture, quality and quantity of soil organic matter, temperature, frequency of had a strong effect on the intensity of CO 2 pulses in MATERIALS AND METHODS a temperate spruce forest.When the spruce soil was The study was conducted on three 5 by 5 m plots in a mixed gradually wetted in summer at high soil temperature, deciduous forest at the Prospect Hill tract of the Harvard soil respiration increased by 144% compared with the Forest (42Њ 32Ј N lat., 72Њ 11Ј W lat.) in Petersham (340 m control, while wetting in autumn at low soil temperature elevation), MA.The forest developed after a hurricane in 1938 and is dominated by red maple (Acer rubrum L.) and had no significant effect on soil respiration.Conversely, red oak (Quercus rubra L.).Before the hurricane, a mixed in arid climates it has been demonstrated that the temdeciduous forest had developed on pastures abandoned since perature sensitivity of soil respiration is expressed only the late 19th century (Foster, 1992).As a result of 19th century when the soil is adequately moist (Wildung et al., 1975; agricultural use, the upper mineral soil is partly disturbed as Xu and Qi, 2001).
indicated by varying depths of the A horizons from 2 cm down The response of soil respiration following wetting of to 15 cm.The thickness of the mineral soil varies from 40 to dry soils has rarely been reported from forests, although 65 cm and has rock contents up to 40% of soil volume.The this is a frequent phenomenon.Wetting events have soil, a fine sandy loam, is well aerated and has been classified as a well-drained Typic Dystrochrept.The O horizon is 3 to been attributed to increases in soil moisture of the min-8 cm thick, stores about 7 kg m Ϫ2 of organic matter, and eral soil (Borken et al., 2002)  coarse roots Ͼ2 mm in diameter were removed.The water Time domain reflectometry (TDR) and frequency docontent of the Oa horizon was not separately measured bemain reflectometry (FDR) sensors have been successcause its thickness was usually Ͻ1 cm and did not allow the fully used to record volumetric soil water content of installation of DC half-bridge sensors.Fresh samples were mineral soils (Roth et al., 1990;Veldkamp and O'Brien, weighed, and then dried at 60ЊC for 1 wk.

2000)
. However, the probes of these sensors require Relative changes in water content of the Oi and Oe/Oa complete contact with the surrounding soil matrix and horizon were continuously measured on hourly basis from June throughout October 2002 using DC half-bridge sensors therefore are problematic for O horizons.The total pore (Fig. 1).This method, described by Hanson et al. (2003), was space of the O horizon might vary with water content modified by using one 1.59 mm thick and 9 cm 2 basswood due to shrinking and swelling processes.Destructive (Tilia americana L.) piece (cut out from 1/16″ by 4″ by 24″ samples are generally taken to determine the gravimetbasswood, Midwest Products CO., Inc., Hobart, IN) per sensor ric water content of O horizons.One important disadinstead of a dead oak leaf.Two stainless steel alligator clips, vantage of destructive sampling is that short-term events connected to 15 to 25 m lengths of six conductor 22 AWG on an hourly or daily basis are usually missed.Recently, cables, were attached to each basswood piece with a spacing DC half-bridge sensors were developed for continuous of 2 cm between the paired clips.At hourly intervals a data logger (CR10X, Campbell Scientific, Logan, UT) supplied a measurement of litter water content of the Oi/Oe horiconstant 2.5 V signal across the wood and logged the voltage.zons in a temperate deciduous forest (Hanson et al., The electronic setup of DC half-bridge sensors followed the 2003).This approach is based on the electrical imped-AM416 Relay Multiplexer instructions by Campbell Scientific ance grids for recording wetness of leaves (Gillespie Inc. (1996) for half-bridge measurements.The voltage reading and Kidd, 1978) and monitors the resistance or voltage on the data logger can be described by the following equation: across an intact dead oak (Quercus prinus L.) leaf as a V 1 ϭ Vx Ϫ Vx ϫ [Rs/(Rs ϩ Rf)], function of litter water content (Hanson et al., 2003).
The objective of this study was to evaluate drying and where V 1 is the data logger voltage reading in V, Vx is 2.5 V, wetting effects of the O horizon on soil respiration of Rf is the resistance of the fixed resistor (390 k⍀), and Rs is the resistance of basswood (Fig. 1).A DC voltage of 2.5 V a temperate deciduous forest at Harvard Forest.The (Vx) is applied across the half-bridge circuit and the return approach of Hanson et al. (2003) using DC half-bridge signal (V 1 ) is related to the resistance of basswood (Rs).As sensors was modified by (i) using basswood pieces inmoisture increases in the O horizon, the basswood quickly stead of intact oak leaves, (ii) installing DC half-bridge equilibrates and its resistance decreases, thereby causing an sensors separately in the Oi and Oe/Oa horizons and (iii) calibrating the volt signals of DC half-bridge sensors using gravimetric water content from destructive samples of the Oi and Oe/Oa horizons.In addition to field measurements of soil respiration and water content, the effect of repeated drying and wetting events on hetero- perature.
increase in V 1 .Different resistors with known resistance (10 k⍀, 413 k⍀, 835 k⍀, 1M⍀) were put in place of the basswood to test the half-bridge set up.In each of the three plots, four and eight DC half-bridge sensors were installed in the Oi and Oe/Oa horizons, respectively.The installation depth in the Oe/Oa horizon was 2 to 6 cm, varying with the thickness of the O horizon.Calibration curves were obtained for DC half-bridge sensors by performing linear regressions across dates between the mean voltage signal of the 12 (Oi) or 24 (Oe/Oa) sensors and mean gravimetric soil water contents of five destructive samples of each horizon taken randomly from the study area.
In each of the three plots, four PVC collars, 10 cm tall and 25 cm in diameter, were installed approximately 4 cm into the soil for the duration of the experiment.Soil respiration was measured weekly from June to October 2001 between 0900 and 1200 h using a vented chamber with a volume of 5.2 L that was placed over a collar and connected to a portable infrared gas analyzer (IRGA, Licor 6252, LI-COR, Lincoln, NE).Air was circulated in this closed system by a pump with a constant rate of 0.5 L min Ϫ1 , and the increase in CO 2 concentration was recorded every 12 s for a 5 min period.Linear regressions were performed to calculate the CO 2 flux rates, which were corrected by air pressure and air temperature within the headspace.A certified standard of 523 L CO 2 L Ϫ1 ( 1998; 2002).Soil temperature at the 10-cm depth was manually measured next to each collar using a temperature probe.Volumetric water content of the mineral soil at the 5-cm depth each microcosm, which is 1 mm more than average maximum was determined hourly using one horizontally installed TDR water-holding capacity of the O horizon at our study site.All sensor (CS615 Water Content Reflectometer, Campbell Scien-O horizons were water saturated after 5 d, and drained water tific, Logan, UT) per plot.The TDR sensors were connected was removed from microcosms to avoid anaerobic conditions.to the data logger (CR10X, Campbell Scientific, Logan, UT).
Microcosms were continuously incubated at 17 Ϯ 1ЊC during To study the response of repeated drying and wetting on this experiment.Carbon dioxide fluxes of the dry and wetted heterotrophic respiration, three intact O horizon samples, O horizon were measured in the same way as in the forest, each of 700 cm 2 , were taken from the same mixed hardwood except that the microcosms were sealed with a lid fitted with stand in the summer of 2000.The O horizon had a thickness sampling ports similar to the chamber used for respiration of 6.3 to 6.7 cm and dry weights of 800 to 1050 g.The samples measurements in the field.were placed in plastic boxes (hereafter, "microcosm") and Soil respiration rates, soil temperatures, and DC half-bridge dried at ambient air temperature for 4 mo.In each microcosm, signals from the field study were means of three replicate three DC half-bridge sensors were permanently installed at plots.Linear and nonlinear regressions were performed (Sysdepths of 0.5, 2, and 5 cm to measure gravimetric water contat Software, 2001) (i) to evaluate the relationships of DC tent.The field calibration curves for the Oi horizon (Fig. 2a) half-bridge signals and gravimetric water contents of the Oi were used to convert voltage signals of the DC half-bridges and Oe/Oa horizon, (ii) to model the temperature dependency in the leaf litter into gravimetric water contents.Voltage sigof soil respiration using the Arrhenius function, and (iii) to nals from the 2-and 5-cm depth were converted into gravimetevaluate the relationship between residuals of the temperature ric water contents using the field calibration of the Oe/Oa model and gravimetric soil water content of the Oi and Oe/ horizon (Fig. 2b).Each microcosm had a sealable lid that Oa horizon.Multiple linear regressions were performed to could be closed for measurements of CO 2 production.
analyze the combined effects of soil temperature and soil water The O horizon was allowed to dry at room temperature content of the Oi, Oe/Oa and A horizons on soil respiration.before the wetting experiment began and between each of the water additions in the first experiment.Each of the three replicate microcosms received the same sequence of water

RESULTS
additions over a 79-d period in the following order: 0.5, 1, 2,

Field Measurements
4, 8, and 0.5 mm.At least 1 wk of drying was permitted between water additions and until the CO 2 flux decreased The DC half-bridge sensors were calibrated using below 7 mg C m Ϫ2 h Ϫ1 , when the next wetting treatment was gravimetric water contents of samples from the Oi and then applied.The water was sprayed onto the surface in Ͻ5 Oe/Oa horizons.Voltage signals of DC half-bridge senmin, except in the case of the 8.0 mm wetting treatment, which sors were linearly correlated (r 2 ϭ 0.72, P Ͻ 0.001) to required a gradual addition of water sprayed over a 1-h period.gravimetric water content of the Oi horizon (Fig. 2a).
Incubation temperature (17 Ϯ 1ЊC) was adjusted 24 h before The correlation between gravimetric water content and CO 2 measurements were performed.
the voltage signal of the Oe/Oa horizon was weaker In a second experiment, a sequential, stepwise wetting with (r 2 ϭ 0.68, P Ͻ 0.001) than for the Oi horizon (Fig. 2b).
2, 2, 4, 4, and 8 mm d Ϫ1 was performed within 5 d to measure Soil respiration varied from 65 to 266 mg C m Ϫ2 h Ϫ1 the short-term increase in CO 2 flux from dry to water saturated O horizon.In total, an amount of 20 mm water was added to during the growing season, mainly following the tempo- The large fluctuation of water content in the Oi horizon between 0.03 and 1.81 g g Ϫ1 shows this horizon is subjected to frequent and extreme drying and wetting cycles (Fig. 3b).Direct current half-bridge sensors responded within a few hours to precipitation events, depending on the previous moisture level and the amount of precipitation.In total, the Oi horizon showed 10 pronounced wetting events from Julian day 177 to 289.The water content of the Oe/Oa horizon ranged between 0.65 and 2.64 g g Ϫ1 ; however, the fluctuation in water content was less pronounced than in the Oi hori- Volumetric water content of the mineral soil as measured by TDR ranged between 0.11 and 0.30 cm 3 cm Ϫ3 ral course in soil temperature (Fig. 3a).The first maxi-(Fig.3c).In contrast to the O horizon, the mineral soil mum soil respiration rate of 265 mg C m Ϫ2 h Ϫ1 occurred did not respond to precipitation events of 4.3 mm on on Julian day 192, corresponding with high water con-Julian day 244 and 247, indicating that the mineral soil tents in the Oi and Oe/Oa horizon (Fig. 3b) and top is less subjected to wetting events.Despite the precipitamineral soil (Fig. 3c), following precipitation of 11.5 mm tion of 105 mm between Julian day 253 and 268, the on Julian day 191 and 192 (Fig. 3d).Subsequently, soil volumetric water content of the top mineral soil rerespiration dropped continuously to 143 mg C m Ϫ2 h Ϫ1 until Julian day 212 although soil temperature increased mained below 0.21 cm 3 cm Ϫ3 .by 0.8ЊC, suggesting that low soil moisture reduced soil respiration on these four sampling days.During this Relationship between Soil Respiration, 20-d period, two rainfall events of 11.7 and 16.0 mm on

Temperature, and Water Content
Julian days 198 and 207, respectively, wetted the Oi Using the Arrhenius equation, soil temperature meaand Oe/Oa horizon and mineral soil (Fig. 3b and 3c).
sured at the 10-cm depth explained 47% (P Ͻ 0.001) However, the water content of the Oi and Oe/Oa horiof the variation in soil respiration and revealed a Q 10 zon dropped again to the previous level within 4 d.The value of 2.90 for the temperature range between 10.5 second maximum soil respiration rate of 266 mg C m Ϫ2 and 20.5ЊC (Fig. 4).The residuals of the Arrhenius h Ϫ1 on Julian day 219 corresponds with the seasonal model were best explained by gravimetric water content maximum soil temperature of 20.6ЊC as well as to previous rainfall of 37.4 mm on Julian days 215 to 217.During of the Oi horizon (r 2 ϭ 0.72, P Ͻ 0.0001) obtained from water was added to the dry O horizon (Fig. 6).Peak CO 2 fluxes of similar magnitude were observed at water additions of 0.5 (32 mg C m Ϫ2 h Ϫ1 ), 1.0 (44 mg C m Ϫ2 DC half-bridge measurements (Fig. 5a).The gravimetric h Ϫ1 ), and 2.0 mm (35 mg C m Ϫ2 h Ϫ1 ).Larger peak CO 2 water content of the Oe/Oa horizon determined by DC fluxes were observed at 4 (50 mg C m Ϫ2 h Ϫ1 ) and 8 mm half-bridge sensors explained 56% (P Ͻ 0.001) of the (68 mg C m Ϫ2 h Ϫ1 ) water additions.The rate of CO 2 residuals (Fig. 5a).The linear relationship between rerelease dropped as soils dried to prewetting levels in 1 siduals and volumetric water content of the mineral soil to 10 d, with the slowest decline occurring after the at the 5-cm depth (Fig. 5b; r 2 ϭ 0.65, P Ͻ 0.001) indicates largest water additions.These pulses of 30 to 50 mg that responses of either or both the O horizon and the C m Ϫ2 h Ϫ1 following laboratory wetting are smaller than top mineral soil could contribute to variation in soil the CO 2 pulses of about 125 mg C m Ϫ2 h Ϫ1 measured respiration during drying and wetting events.
in the field (Fig. 3a), but clearly are large enough to Another approach for analyzing the combined effects contribute significantly to post-wetting pulses in situ.In of temperature and water content on soil respiration is the microcosm experiment, we repeated the 0.5-mm addition at the end of the experiment, after the O horimultiple linear regression.The temperature response zon had dried again.The first and last CO 2 pulses followis nearly fitted equally well by linear (r 2 ϭ 0.45) and ing addition of 0.5 mm water were similar, although the exponential functions (r 2 ϭ 0.47), and so a linear func-CO 2 flux of dry O horizon was higher at the beginning tion was used to simplify multiple linear regression.
of the experiment than after 77 d (Fig. 6).Temperature and water content were not correlated, so In the second microcosm experiment, the stepwise these factors are independent in the multilinear regreswetting of the O horizon to water saturation by adding sion analysis.Combining a linear temperature term with 20 mm within 5 d increased the CO 2 release to a maxia linear moisture term explained 80% (y ϭ 16.2 ϫ mum of 67 mg C m Ϫ2 h Ϫ1 (Fig. 7).Each water addition Temp ϩ 73.6 ϫ Water Ϫ 161, P Ͻ 0.0001), 72% (y ϭ caused a CO 2 pulse that was probably related to wetting 9.33 ϫ Temp ϩ 76.4 ϫ Water Ϫ 107, P Ͻ 0.0001), and of previously dried organic matter.After the last water 74% (y ϭ 15.5 ϫ Temp ϩ 572 ϫ Water Ϫ 213, P Ͻ addition, the CO 2 release and the water contents of the 0.0001), of the variation in soil respiration using the gradually drying O horizon were measured during the water contents of the Oi, Oe/Oa or A horizons, respecfollowing 20 d.During these incubations at constant temperature, the gravimetric water contents of the leaf tively.litter layer, the 2-and 5-cm depth explained 60 (P Ͻ measured water contents were 0.15 g g Ϫ1 for the Oi horizon and 0.47 g g Ϫ1 Oe/Oa horizon, the DC half-0.001),71 (P Ͻ 0.001), and 73% (P Ͻ 0.001), of the variation in CO 2 fluxes, respectively (Fig. 8a-c).The bridge calibration is less certain for lower water contents.The sensors responded immediately to drying and maximum water content decreased from 2.15 g g Ϫ1 in the leaf litter layer to 1.57 g g Ϫ1 at the 5-cm depth, wetting events, indicating rapid equilibration with the surrounding organic matter.However, our results show whereas the minimum water contents of 0.15 to 0.13 g g Ϫ1 were consistent for all three depths.Due to that due to the large heterogeneity in the thickness and composition of O horizon, many DC half-bridge sensors the limitation of DC half-bridge sensors at low water contents Ͻ0.15 g g Ϫ1 , the relationships between CO 2 are required to obtain an accurate estimate of the water content on a 25-m 2 scale.The calibration curve for the release and low water contents were weak (Fig. 8a-c).
Oe/Oa horizon (Fig. 2b) was rather weak (r 2 ϭ 0.68) although twice as many DC half-bridge sensors were

DISCUSSION
installed in this horizon compared with the Oi horizon.

Direct Current Half-Bridge Sensors
One explanation is that the spatial and vertical variation in gravimetric water content of the Oe/Oa horizon is Determination of gravimetric water content of the O higher than that of the Oi horizon, as indicated by larger horizon is often limited by destructive sampling and by error bars (Fig. 3a and 3b).In fact, the water-holding small plot areas, resulting in the lack of information for capacity of the O horizon decreases with increasing studying drying/wetting cycles in forests.Except at very depth as a result of degree of decomposition and delow water contents, DC half-bridge sensors equipped creasing organic matter content.with basswood represent an inexpensive and reasonably In contrast to our approach, Hanson et al. (2003) reliable technique to continuously measure the water installed intact leaves from Q. prinus for DC half-bridge content of the Oi and Oe/Oa horizon without disturbing sensors and obtained a linear calibration (r 2 ϭ 0.87) for the O horizon.Our field calibration showed intercepts the Oi horizon by collecting separate Q. prinus leaves of 0.01 g g Ϫ1 for the Oi horizon and Ϫ0.16 g g Ϫ1 for the Oe/Oa horizon.Because the lowest gravimetrically at varying water contents over time.It is not known 2.90 for the temperature range between 10.5 and 20.5ЊC.This is lower than the Q 10 values of 3.4 and 3.5 reported by Davidson et al. (1998) for adjacent well-drained soils at Harvard Forest calculated for a temperature range of 3 to 20ЊC in 1995 to 1996.Including soil respiration rates from additional years, Savage and Davidson (2001) explained the negative residuals of their temperature model by soil moisture of the mineral soil, indicating a linear decrease in soil respiration with decreasing mineral soil matric potential.However, positive residuals of the temperature model could not be explained by mineral soil matric potential.In European beech and conifer forests, Borken et al. (2002) found a linear trend between residuals from their temperature model and the relative change in soil matric potential.In other words, the relative change in soil matric potential from wet to dry soil or from dry to wet soil was used as an indicator of drying and wetting processes that covaried with CO 2 fluxes.This approach, however, explained only a small part of the residuals from the temperature model.By contrast, our results here reveal linear relationships between the residuals (both positive and negative residuals) from the temperature model and water content of the Oi horizon (r 2 ϭ 0.72), Oe/Oa horizon (r 2 ϭ 0.56) and top mineral soil (r 2 ϭ 0.65), indicating that soil respiration was strongly affected by drying and wetting of the O horizon and/or the top mineral soil.
Both microbial respiration and root respiration may have responded to drying and wetting cycles, and their relative contributions to soil respiration at very dry soil of the Oi horizon, pulses of CO 2 emissions from the mineral soil following wetting events may also contribhow well a calibration based on intact leaves of a single ute to variation in soil respiration.However, short species or one based on commercially available bassdrought periods as well as small amounts of rainfall are wood sticks accurately reflects the water content of O more likely to alter respiration of the O horizon, while horizon material that is comprised of litter from several respiration of the mineral soil may require larger precipspecies at varying degrees of decay.However, our caliitation events and longer drought periods to be affected.bration using basswood sticks relates the stick wetness Taking a maximum water-holding capacity of 19 L to an integrated measure of the gravimetric water conm Ϫ2 (the equivalent of 19 mm precipitation) for the O tent of each sampled horizon.This approach allows horizon into account, the water deficit of the O horizon monitoring of the water content of different layers in varied between 0 and 13 L m Ϫ2 (13 mm precipitation) the O horizon and might also be applicable in coniferous during the growing season.Because precipitation events forests.The disadvantage of using either basswood or are usually Ͻ20 mm and summertime evapotranspira-Q.prinus leaves is that both materials may change their tion of the Harvard Forest is usually on the order of 3 water-holding capacity during microbial decomposition to 4 mm d Ϫ1 (Greenland, 1996), it is not surprising that over time.Hanson et al. (2003) observed only a small the water content of the O horizon is very dynamic.shift in the calibration with decomposition of oak leaves Precipitation and DC half-bridge sensors indicated 10 throughout the summer.The basswood was visually not pronounced drying/wetting cycles in the Oi horizon and decomposed during the growing season at Harvard Forto a lesser extend in the Oe/Oa horizon.In only two of est.Nevertheless, we recommend replacement of the these 10 cases did our weekly schedule of soil respiration basswood at the beginning of each growing season or measurements happen to fall within 1 d after the wetting more often if decomposition of basswood is observed.Moreover, a specific calibration curve may be required event.Assuming that each wetting of dry O horizon for each site because the water-holding capacity of O caused a pulse in soil respiration, weekly measurements horizons may vary from forest to forest.may underestimate annual soil respiration fluxes.According to the model simulation by Hanson et al. (2003),

Effect of Drying and Wetting on Soil Respiration
inclusion of decomposition associated with the wetting and Heterotrophic Respiration dynamics of the surface litter layer increased estimated annual soil respiration by 23% in a upland oak forest Soil respiration at our study site was correlated with soil temperature (r 2 ϭ 0.47), resulting in a Q 10 value of in Tennessee.
Our laboratory study suggests that soil microorgan-ganic matter is frequently limited by low summertime water content at our study site.The CO 2 pulse following isms have the ability to respond within a few minutes to wetting.This is consistent with the observation by repeated wetting of dry O horizon can be explained by water content only, indicating that easily decomposable Orchard and Cook (1983), whereby the duration of the post-wetting CO 2 pulse may depend on many factors C is abundant.Because of the demonstrated link between O horizon water content and soil respiration, such as the duration of drought and the amount of water added.Not only did the maximum CO 2 pulse increase interannual variability in precipitation could affect interannual variation in net ecosystem exchange of C. The with the intensity of wetting, but also the period of elevated CO 2 release increased with increasing wetting O horizon is likely to act as a temporary sink for C in dry years and a temporary source of atmospheric C in intensity.Similar CO 2 pulses were measured in the second laboratory experiment, where the O horizon was wet years, which may help explain interannual variation in net ecosystem exchange of C measured by eddy covar-wetted daily to achieve saturation within 5 d.Pulses of CO 2 releases in the range of 30 to 70 mg C m Ϫ2 h Ϫ1 iance at our Harvard Forest study site (Barford et al., 2001).Long-term trends in climate related to total pre-represent 11 to 26% of the maximum in situ soil respiration rate at our study site.These CO 2 pulses observed cipitation and the seasonality of precipitation could also affect the storage of soil organic matter over the long-term.in the laboratory are in the range of most of the positive residuals from the temperature model used to analyze From our DC half-bridge measurements it may be deduced that many of the CO 2 pulses due to wetting field data (Fig. 5), indicating that the increase in soil respiration following wetting in the field could plausibly were not captured by our weekly field measurements of soil respiration, resulting in underestimation of an-have resulted from increased microbial respiration of the O horizon.nual soil respiration rates.We are currently measuring soil respiration hourly with an automated system to The CO 2 pulse following wetting of dried soil has been related to the amount of easily decomposable C quantify the possible underestimate caused by missing pulse events in a weekly manual sampling protocol (Sav-due to the death of microorganisms or exposure of easily decomposable C on drying (van Gestel et al., 1991).It age and Davidson, 2003).Direct current half-bridge sensors can be used for continuous monitoring of water is unclear if the frequency of drying/wetting cycles affects CO 2 pulses and C mineralization of O horizons in content in layers of the O horizon, allowing the modeling of short-term and long-term responses of biogeo-the long-term.Degens and Sparling (1995) demonstrated that repeated wetting-drying cycles had no effect chemical processes to changes in water content.on the C mineralization of a sandy loam soil.By contrast, Fierer and Schimel (2002)  experiments reported here demonstrated that the source Birch, H.F. 1958.The effect of soil drying on humus decomposition of available C apparently was not exhausted even after and nitrogen availability.Plant Soil 10:9-31.and, therefore, that microbial decomposition of soil or- since no reliable technique consists of Oi, Oe, and Oa horizons.An average maximum has been developed to continuously record the moisture water-holding capacity of 19 L m Ϫ2 was measured for the total conditions of O horizons.However, small rain events O horizon.Mean annual air temperature is 8.5ЊC and mean may wet up only the O horizons or parts of it while annual precipitation is 1050 mm. the covered mineral soil remains unaffected.Such small Soil moisture of the O horizon was measured directly by wetting events may last only for a few hours and are gravimetric water content and indirectly with DC half-bridge difficult to measure by weekly or biweekly measuresensors.Gravimetric water content was measured on 17 sampling days from June to October 2001 by taking five random ments.Moreover, partial wetting of the O horizon might samples of 100 cm 2 area outside the study plots.The O horizon cause only small CO 2 pulses compared with overall was separated into Oi and Oe/Oa horizons from which the soil respiration.

Fig. 1 .
Fig. 1.General description of the direct current half-bridge sensor.

Fig. 2 .
Fig. 2. Relationship between gravimetric water content (n ϭ 5) and volt signal from direct current half-bridge sensors installed in the

Fig. 6 .
Fig. 6.Carbon dioxide release of O horizon, and water contents of leaf litter layer, the 2-and 5-cm depth during drying and wetting cycles.The O horizon material was initially air-dry.Stepwise additions of 0.5, 1, 2, 4, 8, and 0.5 mm of water were applied sequentially over 79 d in laboratory microcosm experiments.Error bars represent the standard error of the mean of three replicate microcosms (n ϭ 3).Fig. 5. Residuals of the Arrhenius equation versus (a) gravimetric water contents predicted by direct current half-bridge sensors in Laboratory Incubation: Drying and Wetting

Fig. 7 .
Fig. 7. Carbon dioxide release from the O horizon and water contents in the leaf litter layer, and at the 2-and 5-cm depths in a second laboratory microcosm experiment.O horizon material was initially air-dried.Stepwise additions of 2, 2, 4, 4, and 8 mm of water were applied sequentially over 5 d.Error bars represent the standard error of the mean of three replicate microcosms (n ϭ 3).

Fig. 8 .
Fig. 8. Relationship between CO 2 release and water content of (a) reported that the total CO 2 ACKNOWLEDGMENTS loss of sieved mineral oak soil increased with the number Werner Borken acknowledges the financial support reof rewetting events during 2 mo.However, disturbance ceived by the Deutsche Forschungsgemeinschaft.The authors of soil by sieving or root cutting may increase the availthank Harvard Forest summer interns, Chris Arabia, Linda ability of the light C fraction, resulting in higher CO 2 Wan, and Rosa Navarro for their hard work in the field.We thank David Foster and the staff of the Harvard Forest for fluxes.their assistance and cooperation.This research was supported Franzluebbers et al. (2000) found a strong relationship by grants from the Harold Whitworth Pierce Charitable Trust between the CO 2 pulse within the first 24 h after wetting and from the Office of Science, Biological and Environmental of dry mineral soils and soil microbial biomass.Gener-Research Program (BER), U.S. Department of Energy, through ally, microbial biomass decreases strongly from the leaf the Northeast Regional Center of the National Institute for litter layer down the soil profile and thus, the impor-Global Environmental Change (NIGEC) under Cooperative tance of drying/wetting processes for microbial respira-Agreement No. DE-FC03-90ER61010.tion may decrease with depth.Because of the relatively strong response at only 0.5 mm water addition, which REFERENCES wetted only the surface layer of intact leaf litter, it is Barford, C.C., S.C. Wofsy, M.L. Goulden, J.W. Munger, E.H. Pyle, likely that a significant part of the easily decomposable S.P. Urbanski, L. Hutyra, S.R. Saleska, D. Fitzjarrald, and K. C that can potentially be mineralized after wetting re-Moore.2001.Factors controlling long-and short-term sequestration of atmospheric CO 2 in a mid-latitude forest.Science 294: sides in the leaf litter layer.Moreover, the laboratory 1688-1691.

5
or 6 wetting and drying events.The supply of mineral-Borken, W., Y.J. Xu, R. Brumme, and N. Lamersdorf.1999.A climate izable C to microorganisms may be less limiting in the change scenario for carbon dioxide and dissolved organic carbon fluxes from a temperate forest soil: Drought and rewetting effects.organic-rich O horizon than in mineral soils, resulting in Soil Sci.Soc.Am.J. 63:1848-1855.a larger potential for repeated pulses of CO 2 production Borken, W., Y.J. Xu, E.A. Davidson, and F. Beese.2002.Site and after numerous wetting and drying events.temporal variation of soil respiration in European beech, Norway spruce, and Scots pine forests.Global Change Biol.8:1205-1216.Bunnell, F., and D.E.N.Tait.1974.Mathematical simulation models CONCLUSIONS of decomposition processes.p. 207-225.In A.J. Holding et al. (ed.)Soil organisms and decomposition in tundra.Tundra Biome We conclude that microbial respiration of the O hori-Steering Committee, Stockholm, Sweden.zon decreases strongly with decreasing water content, Davidson, E.A., E. Belk, and R.D. Boone.1998.Soil water content and temperature as independent or confounded factors controlling