Effects of Atmospheric Co 2 Enrichment on Soil Co 2 Efflux in a Young Longleaf Pine System

The southeastern landscape is composed of agricultural and forest systems that can store carbon (C) in standing biomass and soil. Research is needed to quantify the effects of elevated atmospheric carbon dioxide (CO 2) on terrestrial C dynamics including CO 2 release back to the atmosphere and soil sequestration. Longleaf pine savannahs are an ecologically and economically important, yet understudied, component of the southeastern landscape. We investigated the effects of ambient and elevated CO 2 on soil CO 2 efflux in a young longleaf pine system using a continuous monitoring system. A significant increase (26.5%) in soil CO 2 efflux across 90 days was observed under elevated CO 2 ; this occurred for all weekly and daily averages except for two days when soil temperature was the lowest. Soil CO 2 efflux was positively correlated with soil temperature with a trend towards increased efflux response to temperature under elevated CO 2. Efflux was negatively correlated with soil moisture and was best represented using a quadratic relationship. Soil CO 2 efflux was not correlated with root biomass. Our data indicate that, while elevated CO 2 will increase feedback of CO 2 to the atmosphere via soil efflux, terrestrial ecosystems will remain potential sinks for atmospheric CO 2 due to greater biomass production and increased soil C sequestration.


Introduction
The rural southeastern landscape is dominated by three vegetation types (crops, forests, and pastures), all of which have the ability to store atmospheric carbon (C) as standing biomass (including plant roots) or in soil.One particularly important ecosystem is longleaf pine savannahs.Prior to European settlement, the coastal plains of the southeastern United States were dominated by nearly pure stands of longleaf pine (Pinus palustris Mill.) with a diverse understory plant community; some longleaf ecosystems have the highest reported values for species richness, including many threatened and endangered species, in the temperate Western Hemisphere [1].This system now occupies only 2% of its former range [2], a loss comparable to or exceeding that of most endangered communities throughout the world including the North American tallgrass prairie, the moist tropical coastal forest of Brazil, and the dry forests along the Pacific coast of Central America [3].Longleaf pine forests in the southeast currently occupy sites at the more xeric end of the moisture continuum and are often found on soils with low N availability.In fact, it is not unusual to find disjunct longleaf pine communities in the rural farm landscape.Thus, landowner interest in this species has increased dramatically over the last decade, not only due to its ecological significance but also because of superior lumber quality, fire tolerance, and resistance to some of the more devastating southern forest insects (e.g., bark beetles) and diseases (e.g., fusiform rust).Given that longleaf pine systems may become a more important component of the rural farm landscape, it is important to determine how the rising level of carbon dioxide (CO 2 ) in the atmosphere [4] will impact these systems.
The rising level of atmospheric CO 2 has prompted speculation on the ability of terrestrial ecosystems to sequester C as a means of mitigating this rise and its potential impacts on climate.However, as the ability of terrestrial ecosystems to store C (in biomass and/or in soil) is not based solely on net primary productivity [12], elevated atmospheric CO 2 may also impact terrestrial ecosystem C storage through alterations in plant tissue quality, which will impact soil microbes, decomposition processes, and subsequent soil C storage.Plant tissue produced under high CO 2 often has higher C : N ratios [13,14] and may be structurally different, with alterations in leaf anatomy [15] and epicuticular waxes [16,17].Plants grown under elevated CO 2 may also exhibit altered tissue chemistry, including lower N concentrations [18,19], higher concentrations of carbohydrates [19,20], and increased levels of defense compounds such as phenolics [21,22].
The fate of C within plant systems is affected by a chain of biological events starting with transfer of C from air to leaf, transformation within the plant, translocation within the plant/soil system, return of plant residue to the soil, and decomposition and is impacted by the effects of other environmental factors (e.g., temperature, nutrients, and water) on these processes.Therefore, the ability of terrestrial ecosystems to sequester C will depend on C cycling among the various biomass and soil pools and on the residence time of the C within these pools [23].
At many stages in the cycling of C within terrestrial ecosystems, CO 2 is transferred back to the atmosphere by both autotrophic and heterotrophic respiration.Soil respiration is a significant source of CO 2 flux from terrestrial ecosystems to the atmosphere [24], with global estimates ranging from 68 to 100 Pg C yr −1 [25,26].Therefore, even small shifts in soil CO 2 efflux could have serious implications for increasing or decreasing atmospheric CO 2 concentration and the resulting impacts on climate change [27].Through its impact on the quantity and quality of C within the plant/soil system, elevated CO 2 can affect this feedback of C to the atmosphere.For example, increased root growth under elevated CO 2 could increase root respiration [28], while changes in root exudation and/or quality might enhance [23,29] or suppress [30] microbial respiration.The combined effects on total soil CO 2 respired back to the atmosphere, and the potential for C sequestration, are difficult to predict.
One review of soil and microbial respiration demonstrates that elevated atmospheric CO 2 generally increases belowground respiration, with overall estimates ranging from 40 to 50% for soil respiration and from 20 to 35% for microbial respiration [31]; these estimates agree with another review that reported an overall increase of 37% for forest species [32].Other elevated CO 2 studies report stimulation of root or total soil respiration in the range of 15-50% [33][34][35], with even greater stimulation reported in some cases [36,37].Enhanced root or soil respiration under high CO 2 is often related to increased root biomass, that is, autotrophic respiration [31,34,36,37] and/or increases in the size or activity of the microbial community, that is, heterotrophic respiration [31,34,37,38].However, some cases [30,33] showed elevated CO 2 to suppress soil respiration or to have no effect [39,40].Soil CO 2 efflux can be highly variable on temporal and spatial scales within a single field experiment [34,41] and among experiments; therefore, even relatively large increases in soil efflux under elevated CO 2 may not be statistically significant [31].Some of the variation among individual studies may be due to differences in plant species, experimental conditions, or methods used for determination of CO 2 efflux.
A major drawback of most methods for determining soil CO 2 efflux concerns the timescale of measurements (i.e., cumulative totals across hours to days with NaOH traps or discreet points in time with soil collars and gas exchange devices); efflux between measurement periods is then generally assumed to be linearly integrative across the intervening time periods [34,37].Given the varying responses of soil CO 2 efflux to elevated atmospheric CO 2 and the limitations of current measurement technology, more research is needed before we can confidently predict the impacts of elevated atmospheric CO 2 on the ability of terrestrial ecosystems to sequester C. The objective of this experiment was to assess the response of soil CO 2 efflux (root plus microbial respiration) to three years of atmospheric CO 2 enrichment in a model regenerating longleaf pine community using a novel, continuous CO 2 efflux monitoring system; correlations of efflux with changes in root biomass, populations of microbes and micro-and mesofauna, and soil C were also investigated.

Study Site.
A model regenerating longleaf pine-wiregrass ecosystem was constructed in Spring 1998 at the National Soil Dynamics Laboratory in Auburn, AL; descriptions of the study site and model ecosystem have been previously reported [42].Briefly, an assemblage of five early successional forest species representing major functional guilds within a typical longleaf pine-wiregrass community was chosen for study: longleaf pine (Pinus palustris, a C 3 evergreen conifer), wiregrass (Aristida stricta, a C 4 bunch grass), sand post oak (Quercus margaretta, a C 3 broadleaf tree), rattlebox (Crotalaria rotundifolia, a C 3 perennial herbaceous legume), and butterfly weed (Asclepias tuberose, a C 3 , nonleguminous, herbaceous perennial).These species are common associates throughout the southeastern USA.The model forest community was assembled in April 1998 on an outdoor soil bin (2 m deep, 6 m wide, and 76 m long) containing a Blanton loamy sand (loamy, siliceous, and thermic Grossarenic Paleudults).The planting regime used [42] reflected densities found in naturally regenerating longleaf pine-wiregrass ecosystems [43,44].
Open top chambers [45], encompassing 7.3 m 2 of ground surface area, were used to deliver target CO 2 concentrations of 365 µmol mol −1 (ambient) or 720 µmol mol −1 (elevated) beginning June 1998 using a delivery system described by Mitchell et al. [46].The study area was divided into six blocks, and each CO 2 treatment was randomly assigned to one open top chamber within each block; therefore, the experimental design was a randomized complete block design, with blocks occurring along the length of the soil bin.

Soil Respiration
Measurements.Soil CO 2 efflux was measured using the Automated Carbon Efflux System (ACES) (US Patent 6,692,970), developed at USDA Forest Service, Southern Research Station Laboratory in Research Triangle Park, NC; a description of the ACES has been previously reported [47].Briefly, ACES is a chamber-based, multiport respiration measurement system, which uses open system, dynamic soil respiration chambers measuring 25 cm diameter (491 cm 2 ) equipped with air and soil thermocouples (soil thermocouples were inserted to depth of 5 cm).The soil chambers are designed with pressure equilibration ports to ensure that differences in chamber pressure do not compromise the quality of the respiration measurement [48].Each ACES has 15 sample chambers and one null calibration chamber, which are measured sequentially for 10 minutes each, allowing a complete run every 2 hours and 40 minutes or nine complete runs per day.When not being actively sampled, all chambers are refreshed with reference air to prevent buildup of CO 2 .The ACES units constructed for our study were modified to allow use of reference air from two sources, owing to the differential atmospheric CO 2 concentrations employed; soil chambers in ambient CO 2 open top chambers were refreshed with ambient CO 2 air, while those in elevated open top chambers were refreshed with elevated CO 2 air.Ambient CO 2 reference air was obtained by placing an air compressor in an additional, empty, ambient open top chamber located on an adjacent soil bin and using the same CO 2 delivery system as the main study; elevated CO 2 reference air was similarly obtained by placing a second air compressor in an additional, empty, elevated open top chamber.The air compressors replace the ballast tanks commonly used with the ACES, which provide reference air for the ACES that is buffered against fluctuations in atmospheric CO 2 concentration [47].
Constraints on distance between soil respiration chambers and the main ACES unit (housing the infrared gas analyzer and datalogger) necessitated use of two ACES units in this study; one was used for blocks 1-3 and a second for blocks 4-6.Two soil chambers were placed into each of the 12 open top chambers; the three additional soil chambers for each system were placed outside of open top chambers.Calibration chambers were placed into the ambient open top chamber nearest to each main ACES unit.A soil moisture probe was placed adjacent to each calibration chamber and inserted to a depth of 20 cm.
To minimize the effect of precipitation exclusion on the soil substrate within the soil chambers, soil chambers were moved every 3-4 days between two sample points (A and B) within each open top chamber.Litter on the soil surface was not removed from each sample point, but all points were kept free of live vegetation.The ACES units were installed on March 6, 2001, at which time the study had been continuously exposed to CO 2 treatments for 33 months.The ACES units were run continuously until June 4, 2001 (day of year (DOY) 65 through 155), with the exception of brief periods for maintenance or due to system/power failures; at this time they were removed to allow for a complete destructive harvest of the study.Details of the harvest, along with associated biomass and plant and soil C data, have been previously reported [49].

Soil Biology
Assessments.Root-zone soil, from the 0-15 cm depth increment, was collected using large soil cores (24.5 cm diameter × 60 cm deep) and an extraction method of our own design [50].The soil was then passed through a 2 mm mesh stainless steel sieve until 10-20 g of sieved soil was collected.Dehydrogenase activity, a reliable index of microbial activity in soil [51], was determined from modified procedures described by Tabatabai [52].Sieved soil (≈1 g) for triplicate subsamples from each plot was placed in test tubes (15 × 100 mm), covered with 1 mL of 3% aqueous (w/v) 2,3,5-triphenyltetrazolium chloride and stirred with a glass rod.After 96 hr incubation (27 • C), 10 mL of methanol was added to each test tube, and the suspension was vortexed for 30 sec.Tubes were then incubated for 1 hr to allow suspended soil to settle.The resulting supernatant (≈5 mL) was carefully transferred to clean test tubes using Pasteur pipets.Absorbance was read spectrophotometrically at 485 nm, and formazan concentration was calculated using a standard curve produced from known concentrations of triphenyl formazan.One subsample of sieved soil (≈1 g) from each soil sample was used for determination of soil moisture so that formazan concentrations could be expressed per gram soil dry weight.
Soil taken from the previously described cores was extracted for relative populations of Collembola and Acari by a modified version of the Tullgren system as described by Wiggins et al. [53].Soil samples in large funnels, with stems positioned over water in a collecting tube, were arranged in series under 40 W light bulbs.The animals, migrating in advance of the slowly drying soil (5-7 days), were collected live.Populations, counted under a dissecting microscope, were expressed as numbers per kg of air-dried soil.A subsample of the soil collected for soil animals was sent to the Department of Entomology and Plant Pathology, Auburn University for assessment of nematode populations.

Data Analysis.
All soil respiration data were analyzed for system and power failures; obvious "systematic" errors were parsed from the data set.A total of 18,813 soil CO 2 efflux observations were taken over the 90-day measurement period; of these, 94.4% were deemed acceptable for analysis.Data analysis was conducted using the mixed model procedures (Proc Mixed) of the Statistical Analysis System [54].Data were initially analyzed to determine if differences existed between the two ACES units employed in the study or between soil chamber positions (A versus B); as no significant unit or positional effects were noted, data were not segregated prior to analysis.Effects of CO 2 concentration were determined using the analysis of variance statistics derived from the mixed procedure of SAS; effects were determined by day, by week, and across the entire measurement period.All data from each ACES chamber (total = 24) were then averaged for 1.0 • C intervals of soil temperature measured at a depth of 5 cm at each ACES chamber, regardless of DOY; averaging served to reduce the influence of outliers on the response of soil CO 2 efflux to temperature throughout the experiment.Linear regression [55] was then used on the averaged data to determine the relationship between soil CO 2 efflux and soil temperature; a similar procedure was used to investigate the relationship between soil CO 2 efflux and soil moisture.The relationship of soil CO 2 efflux with components of root biomass and with soil C and N (previously reported by Runion et al. [49]) was also investigated using linear regression.Specific respiration rates were calculated by dividing cumulative soil CO 2 efflux (g C m −2 ) over the entire measurement period by total root biomass, total root C and N, and total soil C and N (g m −2 ) to give g C respired per g root dry weight, root C or N, or per g soil C or N.

Results
Soil CO 2 efflux, averaged across the entire 90 day measurement period, from ambient plots was 2.54 (±0.008; n = 8884) µmol CO 2 m −2 s −1 , while elevated plots averaged 3.22 (±0.011; n = 8878) µmol CO 2 m −2 s −1 ; this represented a significant increase (P < 0.0001) of 26.8% or a total increase of ≈60 g C m −2 .When averaged on a weekly basis (Figure 1), elevated CO 2 plots consistently had significantly higher (P < 0.0001) soil respiration rates than did ambient plots, with the increase ranging from 15 to 33%.Further, when analyzed on a daily average basis (data not shown), elevated CO 2 significantly increased (P < 0.05) soil respiration on all but two days (DOY 79, P = 0.08, and DOY 80, P = 0.99); these two days had the lowest daily average soil temperatures recorded during the entire duration of the study (7.3 and 8.0, resp.).Regression of averaged soil CO 2 efflux on soil temperature (Figure 2) showed strong positive linear relationships for both ambient and elevated CO 2 plots (r 2 = 0.90 and 0.86 for ambient and elevated, resp.).Assessment of nonlinear models did not improve the fit of these data over the linear models.The slope of the line for elevated CO 2 plots was significantly steeper (P < 0.01) than for ambient plots; Y-intercepts for these two regression lines did not differ (P = 0.15).Using these regressions, we calculated the change in soil CO 2 efflux for a 10 • C change in soil temperature for each set of plots; these values were 0.77 and 1.18 µg m −2 s −1 for ambient and elevated CO 2 plots, respectively, indicating a 53% increase in the response of efflux to increasing temperature under elevated CO 2 .Soil moisture, collected only in two ambient CO 2 plots (Figure 3), also showed a strong linear correlation with soil CO 2 efflux, albeit a negative relationship (r 2 = 0.76).However, these data showed a better fit when a quadratic function was employed (r 2 = 0.96).
Soil CO 2 efflux was not correlated with fine, coarse, or total root biomass (r 2 range = 0.01 to 0.35), whether analyzed for each plant species or for total across all species.Similar trends were observed when correlating soil CO 2 efflux with either root N or C (data not shown).
No effects of CO 2 treatment on dehydrogenase were observed (P = 0.40).Soil CO 2 efflux was not correlated with dehydrogenase (r 2 = 0.40 and 0.05 for ambient and elevated CO 2 treatments, resp.).While numbers of both nematodes and soil animals were higher under elevated CO 2 , these effects were not significant (P = 0.40 and 0.15 for nematodes and soil animals, resp.).Soil CO 2 efflux was, again, not correlated with numbers of nematodes or soil animals (r 2 range = 0.01 to 0.35).Soil C and N content did not differ among plots at initiation of the study (1852 and 197 g m −2 for soil C and N, resp.).These variables also varied little at study termination (3042 and 2975 g C m −2 and 165 and 163 g N m −2 for ambient and elevated plots, resp.).Soil CO 2 efflux was positively correlated with both soil C and N content measured at the end of the study; the significance of these correlations varied by soil profile depth (Table 1).Correlations of soil CO 2 efflux with soil C tended to be stronger than correlations with soil N. Specific soil CO 2 efflux rates per g root dry weight or per g root C were significantly lower in elevated than in ambient CO 2 plots; specific respiration rate per g root N was not different between CO 2 treatments (Table 2).However, specific respiration rates per g soil N and C were significantly higher in elevated than in ambient CO 2 plots.

Discussion
The observed increase in soil CO 2 efflux under elevated CO 2 in this study is consistent with other reports in the literature (e.g., [31]).The ≈60 g m −2 increase observed across the 90-day measurement period is also comparable with the ≈178 g C m −2 increase reported by Butnor et al. [47] who used ACES over a 220-day period in a 17-year old loblolly pine (P.taeda) stand.The consistency of the increase we observed, on a weekly or daily basis, further demonstrates that growth under high CO 2 had a sustained impact on soil CO 2 efflux in the model longleaf pine community following 33-36 months of constant exposure to a twice ambient concentration of atmospheric CO 2 .
Temperature is known to strongly influence soil respiration [56,57], with efflux increasing as temperature increases, as observed in this study.Soil CO 2 efflux has generally been shown to increase as an exponential function of temperature [57]; however, in the present study this relationship was more than adequately described using a linear function for both ambient and elevated CO 2 treatments.Under elevated CO 2 , the increased responsiveness (i.e., steeper slope) of soil CO 2 efflux to temperature might suggest increased feedback of C to the atmosphere under global warming.However, when we attempted to fit quadratic relationships of soil respiration to soil temperature (data not shown), we observed differences in the inflection points of the curves (30 and 24 • C for ambient and elevated plots, resp.), as well as a slight increase in fit statistics (r 2 = 0.92 and 0.94 for ambient and elevated plots, resp.).This analysis suggested that, at soil temperatures above 24 • C, the increase in soil CO 2 efflux under elevated CO 2 was reduced; extrapolation of these curves indicated that efflux for both CO 2 treatments would be nearly equal at ≈33 • C. Additional research is needed to verify these extrapolations.
Soil moisture is known to affect soil CO 2 efflux through both physical (displacing soil gases) and biological (impacts on root and microbial activity) means [56].Although the relationship is generally positive, negative relationships (as observed in the present study) have been reported [56,58].The improved fit of our data to a quadratic relationship suggests the existence of a soil moisture content at which soil CO 2 efflux is maximized; this would, obviously, be dependent on soil type.Therefore, as most prior soil CO 2 efflux data have been collected using a series of spot measurements or measurements integrated across relatively short time scales, it is possible that the varying responses (i.e., positive versus negative) of soil CO 2 efflux to soil moisture in previous studies [56,58] might be explained by knowing where data fell on the quadratic response curve (Figure 3).
Increased rates of soil CO 2 efflux under high CO 2 have often been shown to be related to increases in root biomass [31,34,36,37].Course, fine, and total root biomass were all increased by elevated CO 2 in this experiment [49].Therefore, the lack of strong correlations between soil CO 2 efflux and root biomass, root C, or root N was unexpected.Most likely, this lack of correlation was due to high variability within the data, particularly variability among species [49].The lower specific respiration rates for root dry weight and root C under elevated CO 2 are primarily due to the fact that root biomass increased more than soil CO 2 efflux.In contrast, the higher specific respiration rates per g soil C or N are primarily due to the fact that elevated CO 2 increased soil CO 2 efflux to a greater degree than soil C or N.
Assessments of soil microbial activity and populations of soil micro-and mesofauna at study termination showed no differences between CO 2 exposure treatments, again suggesting that root respiration was primarily responsible for the observed differences in soil CO 2 efflux.However, since microbial parameters are highly variable even on short temporal scales, it is likely that assessment of these parameters solely at the end of the study does not accurately reflect their overall contribution to soil CO 2 efflux across the 90-day measurement period.Further, the strong correlations of CO 2 efflux with soil N and, especially, C might also indicate a greater contribution of heterotrophic respiration to total soil efflux than the microbial assessments suggested.Separation of heterotrophic and autotrophic respiration would have aided explanation of these trends.Previous research with container-grown longleaf pine seedling showed that N was the controlling factor; under low N conditions, longleaf growth response to elevated CO 2 was negligible [14].In the current study, Torbert et al. [59] found increased soil N mineralization under elevated CO 2 , indicating that N resources should not be limiting for either microbial decomposition of residues or plant growth in future regenerating longleaf pine systems.Therefore, despite this study receiving no N additions throughout the three years, a positive growth response to elevated CO 2 was observed for longleaf pine [49].
We assessed the overall impact of CO 2 enrichment on this model longleaf pine community through its first three years of growth by extrapolating biomass (above-and belowground, as well as litter), soil C and N [49], and soil respiration data from this study.Elevated CO 2 resulted in a significant increase of 4.07 Mg C ha −1 yr −1 sequestered in standing biomass (ambient CO 2 = 6.29 Mg C ha −1 yr −1 ; elevated CO 2 = 10.36Mg C ha −1 yr −1 ) with an additional significant increase of 0.54 Mg C ha −1 yr −1 in litter (ambient CO 2 = 0.72 Mg C ha −1 yr −1 ; elevated CO 2 = 1.26 Mg C ha −1 yr −1 ).The change in soil C was not significantly different between CO 2 treatments at termination of the study (3.97 and 3.74 Mg C ha −1 yr −1 for ambient and elevated CO 2 , resp.).Therefore, the entire system showed a gain of 4.38 Mg C ha −1 yr −1 due to exposure to elevated atmospheric CO 2 (ambient CO 2 = 10.98 Mg C ha −1 yr −1 ; elevated CO 2 = 15.36Mg C ha −1 yr −1 ).It should be noted that soil respiration rates from the final three months of exposure were used to estimate soil CO 2 efflux over the threeyear study period.Also, we did not assess nighttime plant respiration; it is unlikely this would have significantly impacted the analysis since plant respiration has been shown to be relatively unresponsive to elevated CO 2 [60].Despite increased soil respiration of 2.54 Mg CO 2 -C ha −1 yr −1 under elevated CO 2 , our estimates suggest a net increased storage of 1.84 Mg C ha −1 yr −1 with the majority of the added C residing in plant biomass.Torbert et al. [59] found decreased soil C turnover under elevated CO 2 , suggesting that increased C sequestration in soil is possible in these longleaf systems.
In general, elevated CO 2 increases soil CO 2 efflux due to increases in autotrophic respiration from increased root growth and/or increased heterotrophic respiration associated with microbial use of increased C inputs [31,34].However, despite increased soil CO 2 efflux under elevated atmospheric CO 2 , terrestrial ecosystems can still be potential sinks for atmospheric CO 2 due to greater biomass production and increased soil C sequestration.This may be particularly true for forest systems.For example, our research indicates that regenerating longleaf pine systems have the potential to be sinks for atmospheric CO 2 in a future elevated CO 2 environment.These findings are especially important given that longleaf pines currently occupy less productive, low N sites and given the increasing landowner interest in this species due to its superior economic and ecological attributes.

Table 1 :
Regression parameters and statistics for relationships of total soil CO 2 efflux (g C m −2 ) over the 90-day measurement period with soil N and C content.

Table 2 :
Specific respiration rates (g C respired per g) for root biomass, root C and N and soil C and N.