The effects of elevated atmospheric CO2 and nitrogen amendments on subsurface CO2 production and concentration dynamics in a maturing pine forest Authors: Edoardo Daly, Sari Palmroth, Paul Stoy, Mario Siqueira, A. Christopher Oishi, Jehn-Yih Juang, Ram Oren, Amilcare Porporato, and Gabriel G. Katul The final publication is available at Springer via https://doi.org/10.1007/s10533-009-9327-7. Daly, Edoardo, Sari Palmroth, Paul Stoy, Mario Siqueira, A. Christopher Oishi, Jehn-Yih Juang, Ram Oren, Amilcare Porporato, and Gabriel G. Katul. “The Effects of Elevated Atmospheric CO2 and Nitrogen Amendments on Subsurface CO2 Production and Concentration Dynamics in a Maturing Pine Forest.” Biogeochemistry 94, no. 3 (May 7, 2009): 271–287. doi: 10.1007/ s10533-009-9327-7. Made available through Montana State University’s ScholarWorks scholarworks.montana.edu The effects of elevated atmospheric CO2 and nitrogen amendments on subsurface CO2 production and concentration dynamics in a maturing pine forest Edoardo Daly Æ Sari Palmroth Æ Paul Stoy Æ Mario Siqueira Æ A. Christopher Oishi Æ Jehn-Yih Juang Æ Ram Oren Æ Amilcare Porporato Æ Gabriel G. Katul Abstract Profiles of subsurface soil CO2 concentra- tion, soil temperature, and soil moisture, and through- fall were measured continuously during the years 2005 and 2006 in 16 locations at the free air CO2 enrichment facility situated within a temperate loblolly pine (Pinus taeda L.) stand. Sampling at these locations followed a 4 by 4 replicated experimental design comprised of two atmospheric CO2 concentration levels (ambient [CO2] a, ambient ? 200 ppmv, [CO2] e) and two soil nitrogen (N) deposition levels (ambient, ambient ? fertilization at 11.2 gN m -2 year-1). The combination of these measurements permitted indirect estimation of belowground CO2 production and flux profiles in the mineral soil. Adjacent to the soil CO2 profiles, direct (chamber-based) measurements of CO2 fluxes from the soil–litter complex were simultaneously conducted using the automated carbon efflux system. Based on the measured soil CO2 profiles, neither [CO2] e nor N fertilization had a statistically significant effect on seasonal soil CO2, CO2 production, and effluxes from G. G. Katul e-mail: gaby@duke.edu P. Stoy School of GeoSciences, Department of Atmospheric and Environmental Science, University of Edinburgh, Edinburgh EH9 3JN, UK e-mail: paul.stoy@ed.ac.uk M. Siqueira Departamento de Engenharia Mecaˆnica, Universidade de Brası´lia, Brası´lia, Brazil J.-Y. Juang Department of Geography, National Taiwan University, Taipei, Taiwan e-mail: jjuang@ntu.edu.tw A. Porporato  G. G. Katul Department of Civil and Environmental Engineering, Duke University, Durham, NC, USA E. Daly (&) Department of Civil Engineering, Monash University, Building 60, Clayton Campus, Clayton, VIC 3800, Australia e-mail: edoardo.daly@eng.monash.edu.au S. Palmroth  M. Siqueira  A. C. Oishi  R. Oren  A. Porporato  G. G. Katul Nicholas School of the Environment, Duke University, Durham, NC, USA S. Palmroth availability (i.e., of photosynthate and decomposable material). The ability of decomposers to access the substrate is, in turn, controlled by additional variables including soil water content. Once produced, the diffusive transport of CO2 to the surface depends on the CO2 concentration gradient formed in the soil and the soil CO2 diffusivity, which, in turn, depends on the air-filled porosity of the soil determined by soil water content (e.g., Chen et al. 2005; Jassal et al. 2005). Thus, although not all the processes that control soil surface CO2 efflux can be readily quantified in a typical field experiments, many studies have demonstrated that much of the variation in the measured bulk soil CO2 efflux was attributable to the variation in soil temper- ature and/or moisture (Davidson et al. 1998, 2006; Palmroth et al. 2005; Gaumont-Guay et al. 2006). Atmospheric CO2 concentration and soil nutrient availability also affect carbon cycling in forest ecosystems. Based on a large number of mainly single-factor manipulation experiments (see Hyvonen et al. 2007, for review), increasing atmospheric CO2 concentration generally increases plant growth through increased canopy carbon uptake, whereas enhanced nutrient availability, especially of nitrogen (N), often causes shifts in carbon partitioning from below to aboveground. Results from forest free air CO2 enrichment (FACE) experiments suggest that elevated atmospheric CO2 concentration enhances soil CO2 efflux (Zak et al. 2000; Butnor et al. 2003; King et al. 2004; Bernhardt et al. 2006; Palmroth et al. 2006). On the other hand, fertilization has been suggested to decrease the total soil CO2 efflux by suppressing both autotrophic and heterotrophic soil respiration (Makipaa 1995; Bowden et al. 2004; Olsson et al. 2005; Palmroth et al. 2006; Phillips and Fahey 2007). At the Duke Forest FACE prototype experiment in a loblolly pine (Pinus taeda L.) plantation, a two factor experiment combined ele- vated CO2 treatment (ambient ? 200 ppmv) with a nitrogen (N) fertilization treatment (Oren et al. 2001). A shift in carbon partitioning due to fertilization produced a reduction in the total soil CO2 efflux, even under elevated atmospheric CO2 (Butnor et al. 2003; Palmroth et al. 2006). However, differences between the CO2 treatments in soil moisture due to topography and litter buildup (Scha¨fer et al. 2002) makes it difficult to infer respiratory CO2 production from efflux, and thus to assess potential interaction effects of elevated CO2 and N availability on respiration. the mineral soil over the study period. Soil moisture and temperature had different effects on CO2 concen- tration depending on the depth. Variations in CO2 were mostly explained by soil temperature at deeper soil layers, while water content was an important driver at the surface (within the first 10 cm), where CO2 pulses were induced by rainfall events. The soil effluxes were equal to the CO2 production for most of the time, suggesting that the site reached near steady-state conditions. The fluxes estimated from the CO2 profiles were highly correlated to the direct measurements when the soil was neither very dry nor very wet. This suggests that a better parameterization of the soil CO2 diffusivity is required for these soil moisture extremes. Keywords Soil CO2 dynamics  Climate change  Elevated atmospheric CO2  Nitrogen deposition  Fertilization  Loblolly pine Introduction Forests contribute to the terrestrial carbon budget by removing CO2 from the atmosphere through photo- synthesis and releasing CO2 through respiration by vegetation and decomposers in the soil (e.g., Schle- singer 1997; Brady and Weil 2002). Annual soil respiration is a major source of CO2 into the atmosphere (e.g., Raich and Schlesinger 1992) exceeding the annual anthropogenic fossil fuel emis- sions by an order of magnitude (Schimel 1995; Schlesinger 1997). Consequently, changes in these fluxes must be considered when assessing the rate at which atmospheric CO2 concentration increases. Soil CO2 efflux has a large potential for amplifying global climate change, yet its role is still debated despite a substantial amount of experimental and theoretical work on the controls of the flux performed in a wide range of ecosystems (Trumbore et al. 1996; Hungate et al. 1997; Cao and Woodward 1998; Cox et al. 2000; Giardina and Ryan 2000; Heath et al. 2005; Palmroth et al. 2005). Many interacting factors affect soil CO2 efflux by either controlling the production of CO2 (during root respiration or microbial decomposi- tion) or its transport to the surface. The CO2 production rate is a function of the amount and activity of the respiring biomass of both roots and soil heterotrophs, with the mass-specific respiration rate (i.e., the activity) depending on temperature and substrate Because of the difficulty in performing detailed measurements within the soil, most experiments mea- sure the total soil CO2 efflux with chambers placed on the litter layer. The observed flux is typically related to soil temperature and moisture measured at some arbitrary depth assumed to be representative of biolog- ical activity responsible for respiration. Consequently, the contribution to the soil CO2 efflux of sources at different soil depths, and the sensitivity of CO2 produc- tion to soil moisture and temperature conditions, which themselves vary vertically, have been rarely investi- gated and are the subject of this study. Air-phase soil CO2 has been measured manually at rather low time resolutions (e.g., weekly or monthly) in a number of ecosystems (Winston et al. 1997; Billings et al. 1998; Rustad and Fernandez 1998; Pumpanen et al. 2003; Bernhardt et al. 2006; Taneva et al. 2006; Hashimoto et al. 2007). However, recent advances in solid-state sensor technologies permit measurements at finer temporal resolution (Hirano et al. 2003; Tang et al. 2003; Jassal et al. 2004, 2005; Chen et al. 2005; Baldocchi et al. 2006). The finer time and depth resolution offered by these datasets permit resolving the interplay between transient environmental factors (e.g., pulsed rainfall) and subsurface CO2 production and fluxes (e.g., Jassal et al. 2005; Chen et al. 2005). Here, the simultaneous effects of elevated atmo- spheric CO2 and nitrogen amendments on both subsurface CO2 concentration and production dynam- ics are quantified at the Duke Forest FACE experi- ment. In each treatment combination, soil CO2 production and fluxes at different depths and efflux from the soil to the atmosphere were estimated from the measured profiles of subsurface CO2, soil mois- ture and temperature. The estimated CO2 efflux, representing the fluxes from the mineral soil, was compared at daily and longer time scales to efflux directly measured with chambers positioned on the litter surface. Methods Site description The FACE experiment is located in the Blackwood division of the Duke Forest, Orange County, North Carolina (35580N, 79080W). Site characteristics are described elsewhere (Oren et al. 1998; Andrews and Schlesinger 2001) and only a brief description is provided here. The site is located on slightly acidic soils of the Enon series, characterized by relatively low fertility and classified as clayey loam in the upper 30 cm and clay from 30 to 70 cm. Details on the properties of different soil horizons at the site are reported in Oh and Richter (2005). The experiment consists of eight plots (rings), each 30 m in diameter, established in a loblolly pine (Pinus taeda L.) plantation. Four treatment plots were fumigated with CO2 to increase atmospheric concentration by 200 ppmv above ambi- ent concentration levels, while the other four plots were kept at ambient concentration levels (Hendrey et al. 1999). Enrichment commenced in June 1994 in the prototype plot (current plot 7) and was extended to the remaining plots in August 1996. The fumiga- tion was continuous—with the exception of low temperatures and high wind speed conditions—until 2003, when it was reduced to daylight hours only. Litterfall composition and the effect of treatments on its decomposition are discussed by Finzi et al. (2001) and Finzi and Schlesinger (2002). In 1998, the prototype plot and its reference (current plot 8) were halved, and an impermeable barrier was inserted in the soil up to a depth of 70 cm (more than twice the depth of the fine roots at the site) to conduct a nitrogen by CO2 manipulation experi- ment (Oren et al. 2001). The remaining six plots were similarly partitioned in February 2005. One half of each plot (hereafter referred to as a subplot) was fertilized using ammonium-nitrate at a rate of 11.2 gN m -2 year-1, appreciably higher than the local addition of N via atmospheric deposition (*0.8 gN m -2 year-1) but typical of N fertilization treatments in loblolly pine plantations. Accordingly, in terms of belowground activities, four different conditions can be studied: ambient atmospheric [CO2] a without fertilizer addition (AU) and with nitrogen fertilizer (AF), and elevated atmospheric [CO2] e with ambient conditions (EU) and nitrogen enriched (EF). Measurements In each subplot (16 in total), CO2 concentration (C) in the air phase, soil temperature (Ts), and volumetric soil moisture (h) were continuously monitored Efflux System (ACES, U.S. Patent 6,692,970) devel- oped at the USDA Forest Service, Southern Research Station Laboratory in Research Triangle Park, NC and described in Butnor et al. (2003) and Butnor and Johnsen (2004). Data processing The concentration readings of the CO2 sensors were corrected to account for temperatures different from the reference temperature, 25C. Such a correction was computed from the empirical relationship pro- vided by the manufacturer (personal communication, Vaisala Inc.) C ¼ Cm  CT; ð1Þ where the corrected CO2 concentration, C (in ppmv), is evaluated from the measured concentration, Cm (in % CO2), by subtracting the term CT ¼ 14000ðK  K2Þ 25  Ts 25 ; ð2Þ with K ¼ A3C3m þ A2C2m þ A1Cm þ A0; A3 = 7.9 9 10-6, A2 = -10 -3, A1 = 6.7 9 10 -2, and A0 = 8.4 9 10-3. The correction term CT is of the same order of magnitude as the instrument error during most of the measurement period, becoming only relevant during winter. Hence, the correction does not significantly affect the main period of our analyses— the growing season. Statistical analyses The treatment effects on the annual soil CO2 concentration and respiration were evaluated using a split-plot ANOVA analysis (Steel et al. 1997). The atmospheric CO2 enrichment, randomly assigned to four of the plots, was considered as the main factor with N fertilization assigned at random to one half of each plot (i.e., subplot) as a nested factor. In the following analyses, missing values are replaced by the average of the other subplots subjected to the same combination of treatments. CO2 fluxes and production The measured soil CO2 concentration, temperature, and soil moisture time series can be used to compute CO2 fluxes and CO2 production profiles at each subplot beginning May 2005 at five depths within the upper *50 cm of soil, thus producing profiles in the zone in which most fine roots are located. The sensors, a total of 240 (80 for each measured variable), were inserted into horizontal holes drilled into the first 50 cm of the vertical face of a pit, which was then carefully back-filled layer-wise to minimize soil disturbance. Given the presence of large roots and rocks, the precise sensor depths and position with respect to the other sensors varied slightly from plot to plot. Soil CO2 concentration time series were measured with solid-state infrared CO2 transmitters (GMT 221 model, Vaisala, Finland; e.g., Tang et al. 2003; Jassal et al. 2005; Chen et al. 2005), with a measurement range from 0 to 50,000 ppmv (10,000 ppmv = 1% CO2 = 10 mmol mol -1) and accuracy of ±200 ppmv ? 2% of the reading. The sensors were tested with reference gases at 0 and 10,000 ppmv. A laboratory test conducted with sand showed that continual operation of the CO2 sensors increases the temperature around the head of the sensor by up to 2C, as has also been reported by Jassal et al. (2004). Although such heating has the potential to increase soil respiration rates in the immediate proximity of the probe (Baldocchi et al. 2006), it helps prevent water from condensing inside the sensor element, thus reducing the possibility for probe malfunctioning. Soil temperature profiles were constructed from thermocouple measurements (105T-L16, Campbell Scientific Inc.) vertically installed about 30–40 cm apart from the CO2 sensors at the same soil depth. Volumetric soil water content was measured with Theta Probes (model ML2x, Delta-T Devices, Cam- bridge, UK, distributed by Dynamax Inc., Houston, TX, USA). Given the high rock content and the fact that the soil is mainly composed of clay, some vertical profiles of soil water content exhibited complex vertical patterns, possibly due to localized interferences from nearby rocks, or local air pockets. Throughfall (Pt) was also measured with tipping bucket gages (TE525, Campbell Scientific Inc.) in each of the eight plots near the place where the two vertical sensor arrays were installed. All sensors were interrogated every 30 s and 30 min averages (sums for Pt) were recorded. From July 2005, forest floor CO2 efflux (Fm) time series were also measured about 30 cm far away from the vertical arrays using the Automated Carbon by using the continuity equation. Given the large CO2 concentration vertical gradients, horizontal transfer can be neglected. Accordingly, the continuity equation for CO2 concentration, both in the gas (C) and dissolved phase (Cw), can be modeled as (Simunek and Suarez 1993; Fang and Moncrieff 1999) o ot faC þ hCwð Þ ¼  ooz F þ qwCwð Þ  ECw þ S; ð3Þ where t is time, z is the vertical direction (positive upward), fa is the air-filled porosity, F describes the flux by diffusion in the gas phase, qw is the soil water flux, E is the water uptake by plant roots, and S is a source/sink term accounting for biological produc- tion (i.e., root and microbial respiration) and other possible mechanisms of CO2 movements, such as air fluxes. Flux caused by dispersion in the dissolved phase is neglected because it is 10-4 of that due to diffusion in the gas phase (Glinski and Stepniewski 1985; Simunek and Suarez 1993; Fang and Moncrieff 1999). In one-dimensional approaches, the modeling of air-displacement is difficult and arbitrary (Simunek and Suarez 1993) and thus was not included in this analysis. However, the contribution of this term is not expected to be significant except during strong rainfall events (as we show later). Dissolved CO2 can be related to the CO2 concen- tration in the gas phase by assuming that the total dissolved CO2 is present as carbonic acid, H2CO3 (e.g., Fang and Moncrieff 1999), and that the equilibrium between CO2 and H2CO3 is sufficiently fast to permit the application of Henry’s law (Glinski and Stepniewski 1985; Simunek and Suarez 1993; Flechard et al. 2007). Accordingly, the CO2 concen- tration in the liquid phase becomes Cw ¼ aHRTsC; ð4Þ where aH is Henry’s law constant, the value of which was extrapolated from the data in Glinski and Stepniewski (1985), and R is the universal gas constant (8.314 kg m2 s-2 K-1 mol-1). Given the acidity of the soil in this forest (pH % 5–6, Oh and Richter 2005), dissociated (bicarbonate and carbon- ate) forms of CO2 in water are at least one order of magnitude lower than H2CO3 (Flechard et al. 2007) and are thus neglected. The gas-phase fluxes are assumed to follow Fick’s law (e.g., Patwardhan et al. 1988; Simunek and Suarez 1993; Chen et al. 2005), given by F ¼ D h; Tsð Þ oCoz ; ð5Þ where D is the diffusivity of CO2 in the soil gas phase. The adoption of the Fick’s law appears adequate for CO2, while Stefan–Maxwell equations should be applied for a multi-component diffusion process (see Thorstenson and Pollock 1989; Fang and Moncrieff 1999; Freijer and Leffelaar (1996) for a comparison of the two formulations). The diffusivity coefficient in Eq. 5 is modeled using the classical tortuosity formulation derived by Millington (1959) and Millington and Quirk (1961) D ¼ Da Ts 293  1:75 f aa n2 ; ð6Þ with Da = 15.7 mm 2 s-1 being the molecular diffu- sion coefficient for CO2 in free air at 293 K (e.g., Campbell and Norman 1998) and a = 10/3 a semi- empirical coefficient. The form of the diffusivity in Eq. 6 with a = 10/3 was earlier employed at the same site by Suwa et al. (2004), who reported good agreement between chamber measured and modeled F near the surface. Soil porosity in Eq. 6 is evaluated as n = 1 - qb/qp, with qb and qp being the bulk density and particle density for the mineral soil, respectively. Bulk densities were estimated by fitting a second order polynomial to the values for soil of the Enon Series reported in Oh and Richter (2005), while typical qp of 2.65 g cm-3 was used (e.g., Brady and Weil 2002). The transport of dissolved CO2 due to soil water fluxes and transpiration, qwCw and ECw, was neglected because they are commonly much lower than the other terms in Eq. 3. The highest water fluxes occur during and following rainfall events. If we assume that under these conditions, the top-soil is entirely saturated, the fluxes become of the same order of magnitude as the soil hydraulic conductivity at saturation, which has been estimated to be about 0.08 m3H2O m 2 soil day 1 (Oren et al. 1998; Suwa et al. 2004). In the upper soil layers (*5–6 cm) the largest CO2 concentrations are about 10 4 lmolCO2 mol 1 air ; which correspond to about 4  105 lmolCO2 m3H2O of dissolved CO2 concentrations (see Eq. 4 where aHRTs % 1 m3air m 3 H2O ; e.g., Glinski and Stepniewski 1985). Accordingly, the product qwCw is approxi- mately 0:4 lmolCO2 m 2 soil s 1; an order of magnitude lower than F at the soil surface, Fs. Considering that the hydraulic conductivity diminishes dramatically as soil moisture decreases, the expected water fluxes are at least an order of magnitude lower than those estimated at saturation, and therefore qwCw can be usually neglected. With respect to the term ECw, transpiration rates on the order of 2  103 m3H2O m2soil s1have been reported using sapflow measurements (Oren and Pataki 2001; Phillips and Oren 2001). Assuming for simplicity that the root profile is uniformly distrib- uted over the first 40 cm of soil, E is about 6  108 m3H2O m 3 soil s 1: The product ECw is thus about 2  102 lmolCO2 m3 s1; which, as will be demon- strated later, is two orders of magnitude lower than the common values of the terms S and the divergence of the fluxes, -qF/qz. Based on the analyses above, Eq. 3 can be re- written to provide estimates of the production (S) o ot fa þ haHRTsð ÞC½   ooz D h; Tsð Þ oC oz   ¼ S; ð7Þ where the terms on the left hand side can be estimated using the time series of C, h, and Ts collected at different depths (e.g., Tang et al. 2003; Suwa et al. 2004; Chen et al. 2005; Jassal et al. 2005). The estimate of the fluxes, DoC=oz; at the surface, assumed equal to respiration from the mineral soil, Fs, is calculated by assuming constant concentration at the soil surface equal to that in atmosphere, which is either 380 or 580 ppm depending on the treatment. Because depth and time gradients are required in the calculation of S, data conditioning is necessary. In the time domain, the series were smoothed with a Savitzky–Golay filter, which preserves the height and the width of the peaks in the series (Savitzky and Golay 1964; Chen et al. 2005). Such a filter adopts a moving average procedure, according to which the smoothed values, Xk, are evaluated using Xk ¼ Pi¼þm i¼m Bixkþi   N ; ð8Þ Moreover, the vertical CO2 concentration profiles were fitted as a quadratic function of depth (e.g., Takle et al. 2004; Jassal et al. 2005). We also experimented with other interpolation schemes and found the following: (a) a linear interpolation of the data leads to unrealistic fluxes and mostly negative values of S indicating net CO2 uptake, an unlikely case in the soil pores; (b) a third order polynomial resulted in an overestimation of the fluxes and unrealistically small concentration values at the surface; and (c) other higher-order interpolation schemes, such as cubic spline, also generated source profiles with too many artificial oscillations. Results and discussion The effects of the two treatments, elevated atmo- spheric [CO2] e and N fertilization, on subsurface CO2 and production are evaluated here. The subsurface CO2 profiles during the years 2005 and 2006 are presented first, followed by estimates of CO2 pro- duction and fluxes. The estimates of production and fluxes are restricted to 2006 because, unlike the discontinuous data in 2005, there were few short gaps in the data during this year. In addition, measure- ments of the soil fluxes with the ACES chambers, used in evaluation of flux estimates, were available only since the latter part of 2005 growing season. Subsurface CO2 concentrations Effects of the treatments The mean values of C and soil temperature (Ts) over the first 30 cm and the total throughfall, Pt, during the two monitored growing seasons are reported in Table 1. The growing season is defined as the period of time during the year in which the soil temperature in the first 30 cm remains consistently higher than 15C. Thus, the length of the season varies slightly among the plots and between the 2 years. The differences in throughfall among the plots may be due to spatial variability in plant area density (Oren et al. 2006), which also causes different solar radiation interception. Despite such spatial variabil- ity, soil temperatures were similar in all plots over the entire period of observation (data not shown). where xk?i are sequential events from the original data series, Bi are convoluting integers depending on the number chosen for m (here 8) and on the degree of the polynomial adopted to interpolate the points xk?i, and N is a normalizing factor. We employed Bi and N for a cubic polynomial (e.g., Savitzky and Golay 1964; Chen et al. 2005). A split-plot ANOVA (Steel et al. 1997) showed that the growing-season averages of C were not significantly affected by the treatments or their interaction. C was *30% higher in 2006 (Fig. 1) because of the higher h. The averages were primarily controlled by local conditions, as suggested by the consistent differences among plots in both years. This result is in agreement with that obtained by Taneva et al. (2006) and Bernhardt et al. (2006), who found no statistically significant impact of enriched atmospheric CO2 on soil C sampled at the same site at depths from 0.15 up to 2 m during the years 1996– 2004. On the other hand, it contrasts the findings of Suwa et al. (2004) that subsurface CO2 concentration was often greater in elevated CO2 FACE rings from 1997 to 2000 (from 0.15 m up to 2 m, sampled monthly). However, in this regard we note that Suwa et al. (2004) found that the stimulating effect of enriched atmospheric [CO2] e in the first 30 cm decreased in time from 1997, when the experiment started, to 2004. According to this observed decreas- ing trend, the enhanced annual mean soil CO2 concentration next to the surface related to the CO2 fumigation almost disappeared in 2004. This is consistent with our observation in 2005 and 2006. C dynamics at different depths and time-scales Figure 2 shows examples of measured time series from the fertilized subplot of one of the ambient plots (plot 1, AF, see Table 1). These series are represen- tative of trends at all subplots. While C often increases with increasing depth, shallower depths can occasionally reach concentration levels higher than those at the deeper soil layers (Fig. 2a). Such conditions are generally limited to short periods after Table 1 Time averages of soil CO2 concentration (C) and soil temperature (Ts) vertically averaged over 30 cm depth (vertical averaging indicated by \[) and total throughfall (Pt) during the 2005 and 2006 growing seasons (column DOY reports the period during which soil temperatures are consistently higher than 15C) for each treatment FACE ring 2005 2006 hCi (mmol mol-1) hTsi (C) Pt (mm) DOY hCi (mmol mol-1) hTsi (C) Pt (mm) DOY AU 1 6.97 (1.85) 21.42 (3.25) 241.5 130–310 7.84 (2.28) 20.45 (3.47) 338.9 147–298 5 8.70 (3.00) 22.53 (3.61) 241.7 154–322 9.99 (2.15) 22.44 (2.86) 405.5 147–298 6 10.74 (4.22) 21.95 (3.78) 328.0 133–322 14.08 (4.50) 21.05 (3.91) 464.5 145–320 8 11.78 (5.38) 22.21 (3.63) 240.3 143–323 15.79 (6.11) 21.45 (3.70) 504.7 144–321 9.94 (3.37) 12.4 (3.31) AF 1 7.22 (2.30) 21.38 (3.44) 241.5 130–310 8.43 (3.07) 20.36 (3.69) 338.9 147–298 5 9.80 (2.65) 22.07 (3.64) 241.7 154–322 10.45 (1.70) 22.12 (2.88) 405.5 147–298 6 8.12 (2.75) 22.13 (3.47) 328.0 133–322 9.32 (2.47) 21.19 (3.44) 464.5 145–320 8 7.26 (4.12) 22.25 (3.87) 240.3 143–323 8.66 (3.21) 21.38 (4.00) 504.7 144–321 8.39 (2.75) 9.58 (2.41) EU 2 12.16 (3.50) 22.00 (3.88) 256.0 136–322 15.32 (3.65) 22.10 (2.92) 379.0 147–298 3 7.24 (2.76) 21.02 (3.47) 207.3 155–322 11.59 (4.26) 21.42 (3.74) 480.4 146–321 4 9.84 (3.77) 22.41 (2.90) 193.8 150–310 – 20.86 (3.50) 373.4 146–321 7 7.61 (3.25) 21.88 (3.51) 215.4 150–323 7.34 (2.59) 21.04 (3.61) 492.3 144–321 9.57 (3.03) 11.68 (3.42) EF 2 12.09 (3.16) 21.48 (3.43) 256.0 136–322 14.50 (3.59) 21.79 (2.58) 379.0 147–298 3 8.41 (2.53) 21.55 (3.31) 207.3 155–322 9.55 (2.49) 21.57 (3.61) 480.4 146–321 4 7.04 (2.18) 22.56 (3.10) 193.8 150–310 – 20.86 (3.72) 373.4 146–321 7 6.24 (1.92) 21.70 (3.47) 215.4 150–323 7.46 (1.99) 20.79 (3.54) 492.3 144–321 8.76 (2.18) 10.71 (2.38) Standard deviations are reported between parentheses. Treatment averages and standard deviations of C are reported in bold AU ambient unfertilized, AF ambient fertilized, EU enriched unfertilized, EF enriched fertilized rainfall events. In fact, rainfall events are commonly followed by C pulses (Xu et al. 2004; Jassal et al. 2005; Palmroth et al. 2005; Daly et al. 2008), which occur almost immediately in the shallow subsurface zone, and are often delayed by up to 1 h at deeper soil layers, likely reflecting the reduction in CO2 diffusivity with the propagation of the water infiltra- tion front (Chen et al. 2005; Jassal et al. 2005). Apart from fluctuations related to rainfall events, there is a weak seasonal oscillation in C at shallow depths (Fig. 2a), as was observed in other field studies (e.g., Hirano et al. 2003; Pumpanen et al. 2003; Jassal et al. 2005). In contrast, a seasonal cycle is evident at deeper layers (Fig. 2a) in phase with that of soil temperature (Fig. 2c). C also shows cyclic behavior at the daily scale. Figure 3 presents a comparison between daily oscil- lations of Ts and C typical of different seasons in 2005 and 2006 at different depths. The periods during which soil temperature averaged over 30 cm increases from 10 to 20C and decreases from 20 to 10C are defined as ‘spring’ and ‘fall’, respectively, while the periods when temperature is commonly higher than 20C or lower than 10C are defined as ‘summer’ and ‘winter’, respectively. The curves in Fig. 3 are obtained by (1) subtracting from each rain- free day the midnight-to-midnight daily trend and the daily average of Ts and C, respectively; (2) grouping the days into four different seasons, and then (3) constructing the ensemble averages over all the selected days. The midnight-to-midnight daily trend is assumed linear. Maxima and minima of daily averaged temperatures show an approximate 2 h delay at 25 cm compared to Ts nearest to the soil surface during the entire year (Fig. 3a). As expected, Fig. 1 hCi (growing season average of air-phase soil CO2 concentrations averaged over 30 cm of depth) during 2005 against 2006 (see Table 1). The intercept of the regression line is not significantly different from 0 (P [ 0.1). Circle and triangle refer to unfertilized and fertilized conditions, respec- tively, and open and filled symbols to ambient and elevated atmospheric [CO2], respectively. The inset shows the relation- ships between total throughfall during the growing seasons 2005 and 2006 a b c Fig. 2 Examples of time series of a measured soil CO2 concentration (C), b soil volumetric water content (h), c soil temperature (Ts), and throughfall (Pt) for the fertilized plot in plot 1 (AF1) of Table 1. One of the time series of soil moisture is not reported because of sensor malfunction the amplitude of the oscillations is more pronounced near the surface. In contrast to soil temperature, daily C oscillations vary with depth and season (Fig. 3b). A daily cycle is not evident close to the surface, where C is strongly impacted by soil water and C fluxes next to the soil surface. Moving downward within the soil column, the oscillations in C gradually shift from being in phase with temperature during most of the year to being out of phase from spring to fall. These oscillations return to be in-phase with temperature during winter, possibly because of the low microbial and root activities (see Fig. 3b). Although daily cycles of C and Ts are less related at deeper soil layers, the daily mean values of C at these depths have clear relationships with Ts. As shown in Fig. 4, variations in C are related to soil temperature changes at 32 cm. Moving upward to 9 cm depth, this relation becomes less evident and large variations in C can be seen for the same values of temperature. In contrast, daily C is less related to soil moisture at the deeper layers, where C shows large variations, despite low variability in h (Fig. 4). In summary, the relationships between C and Ts and h vary depending on the timescale and depth. Both C and Ts have daily oscillation at all depths, while h monotonically decreases in time between rain events at each depth, showing little seasonal fluctu- ations at the deeper soil layers. At shallow depths, daily average C is related to h. At deeper soil layers (below *25 cm), C and Ts show strong seasonal cycles and the variation in C is weakly related to variations in h. This might suggest that CO2 concen- tration is affected more by soil moisture near the surface, where soil water pulses following rainfall events introduce a strong perturbation, while at deeper soil layers, where soil moisture does not significantly vary, CO2 variations can be mainly related to temperature. a b Fig. 3 Ensemble-averaged daily oscillations in Ts and C for different seasons in 2005 and 2006 in plot AF1 (see Fig. 4). Atmospheric temperature (divided by 5) is also reported in a Fig. 4 Examples (plot AF1 of Table 1) of the relationships between daily soil temperature (Ts), daily soil moisture (h), and CO2 concentration (C) at three different depths: 46 cm (circle), 15 cm (?), and 9 cm (square) during the years 2005 and 2006 Rainfall-induced variation in C near the soil surface The relationships between C and h near the surface (i.e., less than 10 cm) are reflected in surface fluxes and their intermittent behavior with rainfall (e.g., Daly et al. 2008). At the daily time scale, when C and total infiltrating water are considered, consistent linear trends among different plots emerge. Because daily throughfall (Pt,d) and the conse- quent increase in soil surface h are linearly related (not shown), we analyze and discuss C directly with respect to Pt,d. We define a rainfall-induced C pulse (DC) at the daily timescale as the difference between the daily [CO2] measured the day after a rainfall event and that measured the day before the same rain event. The relationship between the magnitudes of C pulses and Pt,d tends to be quasi-linear with constant slope for the 2 year study period and in almost all the plots (Fig. 5; Table 2). The differences among the slopes of the lines in different plots are likely to be related to local soil-vegetation conditions and partly to the uncertainty in determining the exact vertical position of the sensors. There is uncertainty in determining the location of the transition zone between litter and soil, and integrating volumes of the soil moisture and C sensors themselves. The responses of ring 7 and 8 in 2005 were very different from that in 2006. These differences are probably due to similar local soil properties (the two rings are very close to each other). At the daily timescale, DC appears independent of the various values of daily h prior to rainfall events. This suggests that DC is due to reductions in gas-phase diffusivity rather than to a sudden increase in respiration following the rainfall events (e.g., Lee et al. 2004; Chen et al. 2005). The two treatments did not affect the relationships in Fig. 5, as was supported by the split-plot ANOVA analysis (lowest P = 0.05). CO2 production and fluxes Next, the subsurface CO2 concentration data along with the h, and Ts profile measurements are used to estimate subsurface CO2 production (S) and fluxes (F) during 2006. For illustration, we report in detail the analyses for two subplots of one plot (AU1 and AF1). Fig. 5 Relationships between daily throughfall (Pt,d) and CO2 concentration pulses (DC) in all the FACE rings (R1,…, R8) during 2005 (circle) and 2006 (star). The regression slopes and coefficients of determination are reported in Table 3 Effect of the treatments Table 3 reports the average amount of carbon respired per day in the plots using Eqs. 5 and 6 with a = 10/3. As with hCi, our spatial resolution could be insufficient to capture the spatial variability and, thus, the treatment effects. Such effects were captured in other studies (Butnor et al. 2003; Palmroth et al. 2006) in the case of ring 7 and 8 with a higher spatial resolution. Soil CO2 storage The estimation of the first term in Eq. 7 shows that the subsurface CO2 storage usually varies slowly, with values that are commonly largely lower than the flux divergence (i.e., the second of Eq. 7), and thus also smaller than the C production, S (Fig. 6). Changes in CO2 storage in both air and water phases are usually low during the entire year except during rainfall events, when large oscillations occur (Fig. 6). When water infiltrates the soil, air-phase (faC) and dissolved CO2 (hCw) have opposite dynamics: faC decreases because of the reduction of air-filled porosity, while the soil contains more dissolved CO2 because of higher volumetric soil moisture. The changes of fa and h during and right after rainfall events appear more important than the increase of C and the decrease of Cw in determining the variations of total CO2 storage. The variations are lower at the Table 2 Regression slopes (mmol CO2 mol -1 mm-1 H2O) and coefficients of determination, r 2, for the regression lines in Fig. 5 FACE ring AU AF FACE ring EU EF Slope r2 Ring r2 Slope r2 Slope r2 1 2005 0.15 0.86 0.22 0.66 2 2005 0.12 0.89 0.10 0.76 2006 0.09 0.76 0.15 0.63 2006 0.05 0.86 0.12 0.85 0.11 0.71 0.18 0.60 – – 0.11 0.83 5 2005 0.07 0.80 0.04 0.92 3 2005 0.19 0.90 0.09 0.85 2006 0.07 0.79 0.04 0.64 2006 0.25 0.78 – – 0.07 0.79 0.04 0.78 0.23 0.79 – – 6 2005 0.27 0.93 0.06 0.92 4 2005 0.29 0.82 0.72 0.83 2006 0.29 0.81 0.04 0.87 2006 0.20 0.59 – – 0.28 0.89 0.05 0.81 0.23 0.63 – – 8 2005 0.46 0.82 0.45 0.76 7 2005 0.27 0.91 0.09 .74 2006 0.17 0.80 0.11 0.72 2006 0.11 0.76 – – – – – – – – – – Data are not available in one plot (plot 4, fertilized) for 2006. Data from plots 7 and 8 and plot 2 unfertilized had different behavior than the other rings during the measurement period. Data from plots 7 and 3 in the fertilized treatment could not be fitted with a linear relationship in 2006 Table 3 Averaged daily carbon (gC m -2 day-1) respired in the year 2006 in the different plots as estimated from the vertical arrays FACE ring 2006 FACE ring 2006 AU AF EU EF 1 2.85 (2.25) 2.61 (1.89) 2 4.54 (4.05) – 5 – 2.74 (1.69) 3 3.55 (2.02) 4.66 (2.73) 6 3.83 (2.69) 4.34 (2.63) 4 3.42 (2.48) 4.65 (2.51) 8 5.42 (3.99) 1.64 (1.18) 7 2.1 (1.30) 3.6 (2.20) 4.04 (2.76) 2.83 (1.74) 3.41 (2.34) 4.31 (2.41) The averages are calculated over 365 days. The vertical arrays in the fertilized plot in ring 2 and in the unfertilized plot in ring 5 are not reliable because of external disturbances. The fluxes in ring 4 during 2006 have been evaluated using only the sensor near the top, since most of the other sensors did not work properly during this year. Standard deviations are reported between parentheses. Treatment averages and standard deviations are in bold top-soil layers, where the wetting and drying events occur more quickly, and reach maximum values at depths of 15–21 cm. The dynamics appear delayed when moving downward. During rainfall events, the CO2 concentration measurements may be subjected to errors, due to the amount of water that enters the macro-pores where the sensors are located. Under these circumstances, the estimates of CO2 production, calculated as the difference between storage rate and vertical flux gradients (Eq. 7), are uncertain, as suggested by the abrupt reductions of S during rainfall events (Fig. 7). Similar problems have likely been encountered in other studies. For example, Chen et al. (2005) measured large positive increments in soil CO2 concentrations during rainfall events (see Fig. 3 of Chen et al. (2005)), but only showed their estimates of soil CO2 production during a particular soil dry-down (see Fig. 5 of Chen et al. (2005)). On the other hand, Jassal et al. (2005) did not report high rainfall-induced oscillations in CO2 production prob- ably because of their better estimate of the diffusivity. Also, in that study, the amount of precipitation per rain event was generally small and the soil well drained. These possible errors notwithstanding, the computed long-term rate in storage changes in this study remains negligible when compared to the flux divergence and CO2 production. Jassal et al. (2005) reported a similar finding for a 54 years old Douglas- fir stand in Canada. Moreover, because the fluxes at depths below the root zone, zR (*40–50 cm), are very small, the effluxes from the soil to the atmosphere, Fs, are close to the total S over the root depth, as can be seen by integrating Eq. 7 over depth, i.e. 100 200 300 −1 0 1 2 3 DOY (µm o l m − 3 s− 1 ) ∂(f a C+θC w )/∂ t 31 cm21 cm 15 cm 9 cm Fig. 6 Rate of change of daily CO2 storage in both aqueous and gas phase at different soil depths in the plot AF1 during 2006 Fig. 7 Sample time series of estimated daily fluxes (top) and calculated daily production rates (bottom) at different depths in the plot AF1 (Table 1) during 2006 Z0 zR dF dz dz ¼ Fs ¼ Z0 zR Sdz: ð9Þ Therefore, the measurement of the fluxes from the soil to the atmosphere can be considered roughly as an estimate of the CO2 production within the root zone. Suwa et al. (2004) also estimated monthly CO2 production over the entire 2 m soil column and found that the forest floor efflux is nearly balanced by the CO2 production from the top 30 cm. We stress that the forest floor is in steady-state condition, and the balance between production and fluxes is commonly reached at time scales shorter than 30 min, as observed in Fig. 6. This is a major result when interpreting chamber measurements as an integrated measure of CO2 production. Vertical distribution and dynamics of CO2 fluxes and production The near-surface soil layer is the most active in terms of CO2 production and fluxes, and both F and S rapidly decrease with increasing z (Fig. 7). The decrease in production with depth is consistent with the concomitant reduction in respiring root biomass (Matamala and Schlesinger 2000), while fluxes diminish because of the reduction of both production and soil diffusivity, the latter due to the higher water content. Decreasing flux with depth is the main reason CO2 is higher in deeper layers. Moreover, both F and S fluctuate temporally at two different timescales, daily and seasonal. Both follow daily cycles that mirror those of soil temperature (not shown), and exhibit strong seasonal cycles, especially at the surface, reaching maxima during the peak of the growing season in July and August. The seasonality of the fluxes is mainly due to the larger production during the growing season and to a high soil CO2 diffusivity related to lower soil moisture levels, especially at the soil surface. The latter also explains why CO2 concentrations close to the soil surface remain fairly stable during the growing season even though the production increases (Daly et al. 2008). Figure 8 shows an example of the relationship between CO2 production and h and Ts. As expected, S decreases with depth because of the lower root density and high soil moisture. Relationships between production and both soil moisture and temperature can be identified. Higher temperatures (Ts [ 20C) lead to larger production, while low temperatures (Ts \ 12C) inhibit S. On the other hand, wetter conditions (h[ 0.26) limit the production because they may generate an oxygen-limited environment, while when the soil is relatively dry (h\ 0.13, which corresponds to relative soil moisture lower than 23%) the production is higher. Limitations to comparisons between direct measurements and indirect estimates of CO2 flux The flux at the soil surface, Fs, calculated using Eqs. 5 and 6, can be used as an estimate for soil respiration, not including respiration occurring in the litter layer. It should be emphasized that this modeled Fs may be biased towards larger values because of the use of ambient atmospheric CO2 concentration (enriched or control depending on the subplot) as a top boundary condition, when in fact, the CO2 concentration just below the litter layer should be used. A comparison between daily fluxes estimated from Eqs. 5 to 6 and those measured with ACES, Fm, is shown in Fig. 9. At intermediate soil moisture in the upper 10 cm (0.12 \ h\ 0.25), a condition dominating the study period (see Juang et al. 2007, Fig. 1 for soil moisture histogram), modeled fluxes strongly correlate with the chamber-based estimates (Fig. 9a, b). Note that Fs only describes the efflux that would occur if the mineral soil were exposed to the Fig. 8 Soil CO2 production (S) as a function of local soil moisture (h) and local soil temperature (Ts) at three different depths in the plot AF1 (Table 1) during the year 2006 atmosphere in ambient or enriched conditions. Con- versely, Fm includes litter respiration. Thus, consis- tent with our finding, Fm is expected to exceed Fs, though Fs is likely to be over-estimated by this approach. A sensitivity analysis on a (=10/3 in the formu- lation of tortuosity used in modeling diffusivity in Eq. 6) was also conducted. By reducing this param- eter to 9/3, flux estimates from Eqs. 5 to 6 were always higher than the chamber measurements, while a similar increase in a set to 11/3 = 3.67 resulted in estimated fluxes lower then the measurements by about 600–800 gC m -2 year-1. Because these differ- ences are mainly related to the litterfall respiration, they cannot be much higher than 400 gC m -2 year-1 (Lichter et al. 2005; Palmroth et al. 2006). Hence, the value of a = 10/3 appears reasonable for the site (consistent with the analysis in Suwa et al. 2004). However, Fs tends to be higher than Fm in dry conditions, becoming lower than Fm in wet condi- tions (Fig. 9). Potential reasons for this discrepancy between modeled and measured efflux are numerous, but three plausible reasons can be readily identified: (1) the vertical resolution of sampling, (2) uncertain- ties in the parameters of the soil diffusivity (espe- cially near the surface), and (3) the use of atmospheric CO2 concentration rather than the use of the CO2 concentration just beneath the litter. The selected sampling depths and the limitation of only one sensor at each depth reflect a compromise between cost of instruments, their volume averaging, and the objective to monitor the entire rooting depth. As a result, near-surface soil properties were not precisely known. These properties are probably influenced by macroporosity induced by the dense root system in the upper layers. Macroporosity affects the performance of Eqs. 5 to 6 differently for wet and dry soil moisture states. For wet soil moisture states, some of the CO2 produced locally, especially at the surface, can escape horizontally and exit the soil system in localized air flushing events in the macropores adjacent to the large roots. A one- dimensional model cannot capture such lateral trans- port. Moreover, in wet conditions, the value of the diffusivity is likely to be underestimated by Eq. 6. Hence, the CO2 efflux predicted by the model is smaller than the model prediction. In contrast, in dry conditions Fs tends to be higher than Fm. In fact, according to Eq. 6, dry conditions lead to diffusivity coefficients that solely depend on soil porosity (D * n4/3). The estimate of diffusivity thus needs accurate information about n. This parameter is difficult to obtain, especially near the soil surface where the macroporosity is high and litterfall can have an important influence on diffusivity (Maier and Kress 2000). Moreover, according to the model of D of Eq. 6, soil with similar porosities, such as, e.g., sand and clay, are associated to same values of diffusivity in spite of their different textures. This is an unlikely situation, especially in a stratified soil such as that at the site (e.g., Thorbjørn et al. 2008). Conclusions High frequency subsurface soil CO2 concentration and its relationship to soil water and temperature were analyzed in a temperate pine forest at two levels of atmospheric CO2 concentration and soil nutrient availability during 2005 and 2006. Over these first 2 years, no significant treatment effect (enriched atmospheric CO2, enriched soil N, or their interac- tion) was detected in either the values of C (averaged over the first 30 cm) or in Fs estimated from the C profiles. Vertical profiles of C were found to be strongly affected by both temperature and soil water content. Fig. 9 Comparison between daily averages of chamber measured fluxes, Fm, and fluxes estimated from the CO2 profiles, Fs, in the plots AF1 and AU1 (top). Wet (h[ 0.26, triangles) and dry (h\ 0.13, squares) conditions are shown separately (bottom panels). The intercepts of the regression lines are not significantly different from zero (P [ 0.1) At the hourly time scale, soil moisture and soil temperature interacted to determine daily C oscilla- tions. At the daily and seasonal scales soil temper- ature appeared to control C at the deeper soil layers, while soil moisture was the main forcing near the surface, where C pulses can be primarily attributed to rainfall events. A linear relationship between the magnitude of C pulses and throughfall at the daily scale was observed for all the plots. Because these pulses appear independent of antecedent moisture conditions, they cannot be attributed to enhanced microbial activity, in as much as microbial activity is expected to be dependent on re-wetting cycles. The origin of these pulses can be attributed mainly to reductions in diffusivity following rainfall events. In terms of fluxes and subsurface production, the forest was shown to be in a near steady-state condition (with very low changes in CO2 storage). Such steady-state condition can be assumed to be reached at time scales shorter than 30 min. Both production and fluxes have a strong seasonal cycle. The larger CO2 production during the growing season coupled to the higher CO2 diffusivity, due to higher temperatures and lower soil moisture, generate rather stable CO2 concentrations near the surface. There- fore, the modeling of CO2 diffusivity plays a key role when evaluating CO2 production and fluxes indi- rectly. The model of diffusivity becomes particularly important in wetter and drier than average conditions, as suggested by the comparison between direct measurements of soil efflux and the estimates obtained using the CO2 concentration profiles. How- ever, this finding needs to be further analyzed because the flux estimates from the profiles were calculated assuming values of CO2 atmospheric concentrations that might be not reflective of the actual situation. However, we should note that atmospheric concentration is at least one order of magnitude smaller than soil CO2 concentration; hence, quantities requiring the differencing between these two concentrations are always going to be dominated by the soil concentration, not the atmospheric. Acknowledgments The authors would like to thank Judd Edeburn and the Duke Forest staff, and Keith Lewin and the Brookhaven National Laboratories staff, in particular Robert Nettles, for their assistance at the Duke Forest FACE site. E. D. thanks Luca Grossini for his help with some of the analyses. The authors also thank T. Christensen and an anonymous reviewer for their suggestions. This research was supported by the Office of Science (BER), U.S. Department of Energy, Grant no. DE-FG02-95ER62083. References Andrews JA, Schlesinger WH (2001) Soil CO2 dynamics, acidification, and chemical weathering in a temperate forest with experimental CO2 enrichment. Global Biogeochem Cycles 15:149–162. doi:10.1029/2000GB00 1278 Baldocchi D, Tang J, Xu L (2006) How switches and lags in biophysical regulators affect spatial-temporal variation of soil respiration in an oak-grass savanna. J Geophys Res 111:G02008. doi:10.1029/2005JG000063 Bernhardt ES, Barber JJ, Pippen JS, Taneva L, Andrews JA, Schlesinger WH (2006) Long-term effects of free air CO2 enrichment (FACE) on soil respiration. Biogeochem 77:91–116. doi:10.1007/s10533-005-1062-0 Billings SA, Richter DD, Yarie J (1998) Soil carbon dioxide fluxes and profile concentrations in two boreal forests. Can J Res 28:1773–1783. doi:10.1139/cjfr-28-12-1773 Bowden RD, Davidson E, Savage K, Arabia C, Steudler P (2004) Chronic nitrogen additions reduce total soil res- piration and microbial respiration in temperate forest soils at the Harvard Forest. For Ecol Manage 196:43–56. doi:10.1016/j.foreco.2004.03.011 Brady NC, Weil RR (2002) The nature and properties of soils. Prentice Hall, Upper Saddle River Butnor JR, Johnsen KH (2004) Calibrating soil respiration measures with a dynamic flux apparatus using artificial soil media of varying porosity. Eur J Soil Sci 55:639–647. doi:10.1111/j.1365-2389.2004.00642.x Butnor JR, Johnsen KH, Oren R, Katul G (2003) Reduction of forest floor respiration by fertilization on both carbon dioxide-enriched and reference 17-years-old loblolly pine stands. Glob Change Biol 9:849–861. doi:10.1046/j.1365- 2486.2003.00630.x Campbell GS, Norman JM (1998) An introduction to envi- ronmental biophysics, 2nd edn. Springer, New York Cao M, Woodward FI (1998) Dynamic responses of terrestrial ecosystem carbon cycling to global climate change. Nat- ure 393:249–252. doi:10.1038/30460 Chen D, Molina AE, Clapp CE, Venterea RT, Palazzo AJ (2005) Corn root influence on automated measurement of soil carbon dioxide concentrations. Soil Sci 170:779–787. doi:10.1097/01.ss.0000190512.41298.fc Cox PM, Betts RA, Jones CD, Spall SA, Totterdel IJ (2000) Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model. Nature 408:184– 187. doi:10.1038/35041539 Daly E, Oishi AC, Porporato A, Katul GG (2008) A stochastic model for daily subsurface CO2 concentration and related soil respiration. Adv Water Resour 31:987–994. doi:10.1016/j.advwatres.2008.04.001 Davidson EA, Belk E, Boone R (1998) Soil water content and temperature as independent or confounded factors con- trolling soil respiration in a temperate mixed hardwood forest. Glob Change Biol 4:217–227. doi:10.1046/j.1365- 2486.1998.00128.x Davidson EA, Janssens IA, Luo Y (2006) On the variability of respiration in terrestrial ecosystems: moving beyond Q10. Glob Change Biol 12:154–164. doi:10.1111/j.1365-2486. 2005.01065.x Fang C, Moncrieff JB (1999) A model for soil CO2 production and transport. 1: model development. Agric For Meteorol 95:225–236. doi:10.1016/S0168-1923(99)00036-2 Finzi AC, Schlesinger WH (2002) Species control variation on litter decomposition in a pine forest exposed to elevated CO2. Glob Change Biol 8:1217–1229. doi:10.1046/j.1365- 2486.2002.00551.x Finzi AC, Allen AS, DeLucia EH, Ellsworth DS, Schlesinger WH (2001) Forest litter production, chemistry, and decomposition following two years of free air CO2 enrichment. Ecology 82:470–484 Flechard CR, Neftel A, Jocher M, Ammann C, Leifeld J, Fuhrer J (2007) Temporal changes in soil pore space CO2 concentration and storage under permanent grassland. Agric For Meteorol 142:66–84. doi:10.1016/j.agrformet. 2006.11.006 Freijer JI, Leffelaar PA (1996) Adapted Fick’s law applied to soil respiration. Water Resour Res 32:791–800. doi: 10.1029/95WR03820 Gaumont-Guay D, Black TA, Griffis TJ, Barr AG, Jassal RS, Nesic Z (2006) Interpreting the dependence of soil respira- tion on soil temperature and water content in a boreal aspen stand. Agric For Meteorol 140:220–235. doi:10.1016/ j.agrformet.2006.08.003 Giardina CP, Ryan MG (2000) Evidence that decomposition rates of organic carbon in mineral soil do not vary with temperature. Nature 404:858–861. doi:10.1038/35009076 Glinski J, Stepniewski W (1985) Soil aeration and its role for plants. CRC Press, Boca Raton Hashimoto S, Tanaka N, Kume T, Yoshifuji N, Hotta N, Ta- naka K, Suzuki M (2007) Seasonality of vertically parti- tioned soil CO2 production in temperate and tropical forest. J For Res 12:209–221. doi:10.1007/s10310-007- 0009-9 Heath J, Ayres E, Possell M, Bardgett RD, Black HIJ, Grant H, Ineson P, Kerstiens G (2005) Rising atmospheric CO2 reduces sequestration of root-derived soil carbon. Science 309:1711–1713. doi:10.1126/science.1110700 Hendrey G, Ellsworth D, Lewin K, Nagy J (1999) A free-air enrichment system for exposing tall forest vegetation to elevated atmospheric CO2. Glob Change Biol 5:293–309. doi:10.1046/j.1365-2486.1999.00228.x Hirano T, Kim H, Tanaka Y (2003) Long-term half-hourly measurement of soil CO2 concentration and soil respira- tion in a temperate deciduous forest. J Geophys Res 108(D20):4631. doi:10.1029/2003JD003766 Hungate BA, Holland EA, Jackson RB, Chapin FSIII, Moo- neyk HA, Field CB (1997) The fate of carbon in grassland under carbon dioxide enrichment. Nature 388:576–579. doi:10.1038/41550 review. New Phytol 173:463–480. doi:10.1111/j.1469- 8137.2007.01967.x Jassal RS, Black TA, Drewitt GB, Novak MD, Gaumont-Guay D, Nesic Z (2004) A model of the production and trans- port of CO2 in soil: predicting soil CO2 concentrations and CO2 efflux from a forest floor. Agric For Meteorol 124:219–236. doi:10.1016/j.agrformet.2004.01.013 Jassal R, Black A, Novack M, Morgenstern K, Nesic Z, Gaumont-Guay D (2005) Relationship between soil CO2 concentrations and forest-floor CO2 effluxes. Agric For Meteorol 130:176–192. doi:10.1016/j.agrformet.2005. 03.005 Juang J-Y, Porporato A, Stoy PC, Siqueira MBS, Oishi AC, Detto M, Kim HS, Katul GG (2007) Hydrologic and atmospheric controls on convective precipitation events in a southeastern US mosaic landscape. Water Resour Res W03421. doi:10.1029/2006WR004954 King JS, Hansonw PJ, Bernhardt E, Deangelis P, Norby RJ, Pregitzer KS (2004) A multiyear synthesis of soil respi- ration responses to elevated atmospheric CO2 from four forest FACE experiments. Glob Change Biol 10:1027– 1042. doi:10.1111/j.1529-8817.2003.00789.x Lee X, Wu H-J, Sigler J, Oishi C, Siccama T (2004) Rapid and transient response of soil respiration to rain. Glob Change Biol 10:1017–1026 Lichter J, Barron S, Finzi A, Irving K, Roberts M, Stemmler E, Schlesinger WH (2005) Soil carbon sequestration and turnover in a pine forest after six years of atmospheric CO2 enrichment. Ecology 86:1835–1847. doi:10.1890/04- 1205 Maier CA, Kress LW (2000) Soil CO2 evolution and root respiration in 11 year-old loblolly pine (Pinus taeda) plantations as affected by moisture and nutrient avail- ability. Can J For Res 30:347–359. doi:10.1139/cjfr-30-3- 347 Makipaa R (1995) Effect of nitrogen input on carbon accu- mulation of boreal forest soils and ground vegetation. For Ecol Manage 79:217–226. doi:10.1016/0378-1127(95) 03601-6 Matamala R, Schlesinger WH (2000) Effects of elevated atmospheric CO2 on fine root production and activity in an intact temperate forest ecosystem. Glob Change Biol 6:967–979. doi:10.1046/j.1365-2486.2000.00374.x Millington RJ (1959) Gas diffusion in porous media. Science 130:100–102. doi:10.1126/science.130.3367.100-a Millington RJ, Quirk JM (1961) Permeability of porous solids. Trans Faraday Soc 57:1200–1207. doi:10.1039/tf96157 01200 Oh NH, Richter DD (2005) Elemental translocation and loss from three highly weathered soil–bedrock profiles in the southeastern United States. Geoderma 126:5–25. doi:10. 1016/j.geoderma.2004.11.005 Olsson P, Linder S, Giesler R, Hogberg P (2005) Fertilization of boreal forest reduces both autotrophic and hetero- thropic soil respiration. Glob Change Biol 11:1745–1753. doi:10.1111/j.1365-2486.2005.001033.x Oren R, Pataki DE (2001) Transpiration in response to varia- tion in microclimate and soil moisture in southeastern deciduous forests. Oecologia 127:549–559. doi:10.1007/ s004420000622 Hyvonen R, Agren GI, Linder S et al (2007) The likely impact of elevated [CO2], nitrogen deposition, increased tem- perature and management on carbon sequestration in temperate and boreal forest ecosystems: a literature Oren R, Ewers BE, Todd P, Phillips N, Katul G (1998) Water balance delineates the soil layer in which moisture affects canopy conductance. Ecol Appl 8:990–1002. doi:10.1890/ 1051-0761(1998)008[0990:WBDTSL]2.0.CO;2 Oren R, Ellsworth DS, Johnsen KH, Phillips N, Ewers BE, Maier C, Schafer KVR, McCarthy H, Hendrey G, McN- ulty SG, Katul GG (2001) Soil fertility limits carbon sequestration by forest ecosystems in a CO2-enriched atmosphere. Nature 411:469–472. doi:10.1038/35078064 Oren R, Hsieh C-I, Stoy P, Albertson J, McCarthy HR, Harrell P, Katul GG (2006) Estimating the uncertainty in annual net ecosystem carbon exchange: spatial variation in tur- bulent fluxes and sampling errors in eddy-covariance measurements. Glob Change Biol 12:883–896. doi: 10.1111/j.1365-2486.2006.01131.x Palmroth S, Maier CA, McCarthy HR, Oishi AC, Kim H-S, Johnsen KH, Katul GG, Oren R (2005) Contrasting responses to drought of forest floor CO2 efflux in a Lob- lolly pine plantation and a nearby Oak-Hickory forest. Glob Change Biol 11:1–14. doi:10.1111/j.1365-2486. 2005.00915.x Palmroth S, Oren R, McCarthy HR, Johnsen KH, Finzi AC, Butnor JR, Ryan MG, Schlesinger WH (2006) Above- ground sink strength in forests controls the allocation of carbon below ground and its CO2-induced enhancement. Proc Natl Acad Sci USA 103:19362–19367. doi:10.1073/ pnas.0609492103 Patwardhan AS, Nieber JL, Moore ID (1988) Oxygen, carbon dioxide, and water transfer in soils: mechanisms and crop response. Trans Am Soc Agr Eng 31:1383–1395 Phillips RP, Fahey TJ (2007) Fertilization effects on fineroot biomass, rhizosphere microbes and respiratory fluxes in hardwood forest soils. New Phytol 176:655–664. doi:10.1111/j.1469-8137.2007.02204.x Phillips N, Oren R (2001) Intra- and inter-annual variation in transpiration of a pine forest. Ecol Appl 11:385–396. doi:10.1890/1051-0761(2001)011[0385:IAIAVI]2.0.CO;2 Pumpanen J, Ilvesniemi H, Hari P (2003) A process-based model for predicting soil carbon dioxide efflux and con- centration. Soil Sci Soc Am J 67:402–413 Raich JW, Schlesinger WH (1992) The global carbon dioxide flux in soil respiration and its relationship to vegetation and climate. Tellus 44B:81–99 Rustad LE, Fernandez IJ (1998) Experimental soil warming effects on CO2 and CH4 flux from a low elevation spruce- fir forest soil in Maine, USA. Glob Change Biol 4:597– 605. doi:10.1046/j.1365-2486.1998.00169.x Savitzky A, Golay MJE (1964) Smoothing and differentiation of data by simplified least squares procedures. Anal Chem 36:1627–1639. doi:10.1021/ac60214a047 Scha¨fer KV, Oren R, Lai CT, Katul GG (2002) Hydrologic balance in an intact temperate forest ecosystem under ambient and elevated atmospheric CO2 concentration. Glob Change Biol 8:895–911 Schimel DS (1995) Terrestrial ecosystems and the carbon cycle. Glob Change Biol 1:77–91. doi:10.1111/j.1365- 2486.1995.tb00008.x Schlesinger WH (1997) Biogeochemistry: an analysis of global change. Academy Press, San Diego Simunek J, Suarez DL (1993) Modeling of carbon dioxide transport and production in soil. I. Model development. Water Resour Res 29:487–497. doi:10.1029/92WR02225 Steel GD, Torrie JH, Dickey DA (1997) Principles and pro- cedures of statistics: a biometrical approach, 3rd edn. McGraw-Hill, New York Suwa M, Katul GG, Oren R, Andrews J, Pippen J, Mace A, Schlesinger WH (2004) Impact of elevated atmospheric CO2 on forest floor respiration in a temperate pine forest. Global Biogeochem Cycles 18:GB2013. doi:10.129/2003 GB002182 Takle ES, Massman WJ, Brandle JR et al (2004) Influence of high-frequency ambient pressure pumping on carbon dioxide efflux from soil. Agric For Meteorol 124:193– 206. doi:10.1016/j.agrformet.2004.01.014 Taneva L, Pippen JS, Schlesinger WH, Gonzalez-Meler MA (2006) The turnover of carbon pools contributing to soil CO2 and soil respiration in a temperate forest exposed to elevated CO2 concentration. Glob Change Biol 12:983– 994. doi:10.1111/j.1365-2486.2006.01147.x Tang J, Baldocchi DD, Qi Y, Xu L (2003) Assessing soil CO2 efflux using continuous measurements of CO2 profiles in soils with small solid-state sensors. Agric For Meteorol 124:193–206 Thorbjørn A, Moldrup P, Blendstrup H, Komatsu T, Rolston DE (2008) A gas diffusivity model based on air-, solid-, and water-phase resistance in variably saturated soil. Vadose Zone J 7:1276–1286. doi:10.2136/vzj2008.0023 Thorstenson DC, Pollock DW (1989) Gas transport in unsat- urated zones: multicomponent systems and the adequacy of Fick’s laws. Water Resour Res 25:477–507. doi:10. 1029/WR025i003p00477 Trumbore SE, Chadwick OA, Amundson R (1996) Rapid exchange between soil carbon and atmospheric carbon dioxide driven by temperature change. Science 272:393– 396. doi:10.1126/science.272.5260.393 Winston GC, Sundquist ET, Stephens BB, Trumbore SE (1997) Winter CO2 fluxes in a boreal forest. J Geophys Res 102(D24):28795–28804. doi:10.1029/97JD01115 Xu L, Baldocchi DD, Tang J (2004) How soil moisture, rain pulses, and growth alter the response of ecosystem res- piration to temperature. Global Biogeochem Cycles 18:GB4002. doi:10.1029/2004GB002281 Zak DR, Pregitzer KS, King JS, Holmes WE (2000) Elevated atmospheric CO2, fine roots and the response of soil microorganisms: a review and hypothesis. New Phytol 147:201–222. doi:10.1046/j.1469-8137.2000.00687.x