Internet & Technology


of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  RAPID COMMUNICATION Elevated atmospheric carbon dioxide increases soilcarbon  JULIE D. JASTROW * , R. MICHAEL MILLER * , ROSER MATAMALA * , RICHARD J. NORBY w ,THOMAS W. BOUTTON z , CHARLES W. RICE§ and CLENTON E. OWENSBY§ * Environmental Research Division, Argonne National Laboratory, Argonne, IL 60439, USA,  w Environmental Sciences Division,Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA,  z Department of Rangeland Ecology and Management, Texas A&MUniversity, College Station, TX 77843, USA,  § Department of Agronomy, Kansas State University, Manhattan, KS 66506, USA Abstract The general lack of significant changes in mineral soil C stocks during CO 2 -enrichmentexperiments has cast doubt on predictions that increased soil C can partially offset risingatmospheric CO 2  concentrations. Here, we show, through meta-analysis techniques, thatthese experiments collectively exhibited a 5.6% increase in soil C over 2–9 years, at amedian rate of 19gCm  2 yr   1 . We also measured C accrual in deciduous forest andgrassland soils, at rates exceeding 40gCm  2 yr   1 for 5–8years, because both systemsresponded to CO 2  enrichment with large increases in root production. Even thoughnative C stocks were relatively large, over half of the accrued C at both sites wasincorporated into microaggregates, which protect C and increase its longevity. Our data,in combination with the meta-analysis, demonstrate the potential for mineral soils indiverse temperate ecosystems to store additional C in response to CO 2  enrichment. Keywords:  carbon sequestration,  13 C stable isotope, FACE experiment, meta-analysis, microaggregates,open-top chamber, roots, soil organic matter, sweetgum forest, tallgrass prairie grassland Received August 3, 2005; revised version received and accepted August 30, 2005 Introduction Most field-scale CO 2 -enrichment studies have failed todetect significant changes in soil C against the relativelylarge, spatially heterogeneous pool of existing soil or-ganic matter (SOM), leading to the general conclusionthat the potentialforincreasedsoilCislimited(Hungate et al ., 1997; Gill  et al ., 2002; Hagedorn  et al ., 2003; Lichter et al ., 2005). Yet, it is currently unclear whether the lackof change in these studies is a general response or afunction of (1) the low statistical power of most experi-ments and/or (2) the magnitude of CO 2 -stimulatedC inputs relative to the duration of the experiments(Hungate  et al ., 1996; Filion  et al ., 2000; Smith, 2004). A furtherconcern is that if CO 2 -stimulated increases in soilorganic C do occur, they will be allocated to rapidlycycling, labile pools with little, if any, long-term stabili-zation (Hungate  et al ., 1997; Lichter  et al ., 2005).One of the few elevated-CO 2  experiments to report asignificant increase in soil C is our study of nativegrassland in Kansas ( Jastrow  et al ., 2000; Williams et al ., 2000, 2004). Here, we present corroborating resultsfrom a study in temperate deciduous forest, along withadditional data from the grassland experiment, to showthat similar mechanisms are responsible for the increaseand affect the potential longevity of accrued C at bothsites, even though they differ in climate, vegetation, andsoil properties. We also discuss factors that enableddetection of C accrual despite high background C con-centrations and low treatment replication. Lastly, wepresent a meta-analysis of published data from a widerange of ecosystems to support the generality of ourexperimental results. Materials and methods Site description and sampling The deciduous forest free-air CO 2  enrichment (FACE)experiment was constructed in a 10-year-old sweetgum( Liquidambar styraciflua  L.) plantation in Oak Ridge, TN(Norby  et al ., 2001). When the experiment began in Correspondence: Julie D. Jastrow, tel.  1 1 630 252 3226,fax  1 1 630 252 8895, e-mail: Global Change Biology (2005)  11,  2057–2064, doi: 10.1111/j.1365-2486.2005.01077.x r 2005 Blackwell Publishing Ltd  2057  April 1998, the average tree height was about 12m. Thesite is located on Wolftever silty clay loam (fine, mixed,semiactive, thermic Aquic Hapludult). The ongoingexperiment consists of two 25m diameter plots exposedto elevated CO 2  at 542 m molmol  1 (5-year daytimeaverage) during the growing season (April–November) and three ambient-CO 2  plots (CO 2  concen-trations averaged 384–398 m molmol  1 ).The 8-yeargrasslandexperiment usedopen-topcham- bers placed on native prairie in Manhattan, KS (Owens- by  et al ., 1993). The site, which is representative of mixedC4/C3 grasslands in the eastern Great Plains and CornBelt regions, was located on a 5% slope and was domi-nated by warm-season C4 grasses, with C3 speciescontributing only about 15% to aboveground biomass(Owensby  et al ., 1999). Although bare patches occurred between plant stems and crowns, virtually all surfacesoil was heavily occupied by roots and rhizomes. Thesoil series was Tully silty clay loam (fine, mixed, super-active, mesic, Pachic Argiustoll). Experimental treat-ments were (1) ambient CO 2 , no chamber; (2) ambientCO 2 , with chamber, and (3) elevated CO 2 , with chamber.Treatments, which began in May 1989, were replicatedthree times on 4.5m diameter plots. Elevated-CO 2  con-centrations were maintained at twice ambient on the basis of real-time measurements. Measured CO 2  concen-trations in the elevated treatment averaged 709 m molmol  1 during the day and 811 m molmol  1 at night.At both sites, soil was sampled with a 4.8cm diametercorer after removal of surface litter. Cores were dividedinto depth increments of 0–5, 5–15 and 15–30cm andpooled within plots. The forest site was sampled ran-domly in October 1997 (pretreatment; 6 cores per plot to30cm), November 2000 (12 cores per plot to 15cm) andOctober 2002 (18 cores per plot to 30cm). The prairieexperiment was sampled at its conclusion in earlyNovember 1996 (4 cores to 30cm and 2 additional coresto 5cm per plot) by using the stratified random designdescribed in Jastrow  et al . (2000). In this design, thecores were randomly placed on or near the crowns of the dominant C4 species, but roots of C3 species wereintermixed with C4 roots and were not excluded.  Analytical procedures After removal of roots and rhizomes, soil was passedthrough a 4mm sieve anddried at 65–70 1 C. Subsampleswere gently crushed to pass through a 2mm sieve, andremaining litter and root pieces longer than 3–4mmwere removed. Another subsample of 4mm sieved soilwas separated into stable microaggregates (53–250 m min diameter) and nonmicroaggregated soil by using amicroaggregate isolator (Six  et al ., 2000). Briefly, sampleswere immersed in deionized water over a 250 m m sieveandshaken with 50 glassbeadsat 180strokesmin  1 on areciprocating shaker. Continuous water flow flushedmicroaggegates and other soil components  o 250 m mthrough the device and onto a 53 m m sieve. Materialretained on the 53 m m sieve was wet-sieved (50 up-down strokes in 2min). Nonaggregated fine (53–250 m m) particulate organic matter (POM) was sepa-rated from microaggregates by flotation in sodiumpolytungstate solution (2.0gcm  3 ). Both the nonaggre-gated POM and stable microaggregates were washedwith deionized water, and the POM was combined withthe other nonmicroaggregated soil. Whole soil andfractions were dried at 65 1 C, ground, and analyzedfor C and N by dry combustion with a Carlo ErbaNC2500 elemental analyzer (Thermo Electron, Waltham,MA, USA). Soil at both sites was free of carbonates.Stable C isotope ratios in soil from the forest site weredetermined by using a Carlo Erba EA1108 elementalanalyzer interfaced with a isotope ratio mass spectro-meter (Thermo Electron, Waltham, MA, USA) operatingin continuous-flow mode. Results were expressed instandard  d 13 C notation and reported relative to theinternational Vienna Pee Dee Belemnite standard. Over-all precision (machine plus sample preparation error)was 0.1 % . The fraction (  f  ) of soil C derived from  13 C-depleted CO 2  fixed by plants in elevated-CO 2  plots wascalculated as  f  5 ( d t  d i )/( d o  d i ) where  d t  is the  d 13 C of soil after 3 years (2000) or 5 years (2002) of treatment;  d i is the  d 13 C of inputs produced under elevated CO 2 ; and d o  is the  d 13 C of pretreatment (1997) soil. The value of   d i (  37.5 % ) was estimated from roots collected from in-growth cores (Matamala  et al ., 2003). Data analyses and meta-analysis For the forest study, data were analyzed by repeatedmeasures analysis of variance ( ANOVA) . Calculation of C accrual attributed to CO 2  enrichment was adjustedfor initial conditions (accrual 5 [2002 elevated  1997elevated]  [2002 ambient  1997 ambient]). For the prair-ie study, data were analyzed by  ANOVA  with orthogonaltreatment contrasts in a randomized complete-blockdesign. Experimental blocks were positioned at differ-ent elevations along the slope, and orthogonal contrastswere used to test two preplanned, independent nullhypotheses about the CO 2  treatments: (1) no difference between ambient controls (ambient, with chamber 5 ambient, no chamber) and (2) no difference betweenelevated-CO 2  treatment and the ambient controls. Gi-ven constraints on replication and the consequent lowstatistical power of the experiments (Hungate  et al .,1996; Filion  et al ., 2000), a probability level of 0.10 waschosen as the criterion for statistical significance. Allreported errors are standard errors. 2058  J. D. JASTROW  et al . r  2005 Blackwell Publishing Ltd,  Global Change Biology ,  11,  2057–2064  The metadata (see supplementary Table S1) wereanalyzed by using MetaWin 2.0 software (Rosenberg et al ., 2000), with the effect size for each observationcalculated as the natural log of the response ratio. Themean effect size is the weighted mean of individualeffect sizes, with the reciprocal of the standard devia-tion as the weighting factor. Data were obtained frompublished sources identified by searching the Web of Science s ( data were presented graphically, values wereobtained directly from authors or estimated by digitiz-ing an image of the figure. Data reported as concentra-tions were converted to stocks by using published bulkdensities for the site; in a few cases bulk densities wereunavailable and values were estimated.Our primary criteria for inclusion in the meta-analy-sis were (1) study duration was at least two growingseasons and (2) CO 2  enrichment used FACE (or similar)technology, open-top chambers, or other outdoor fumi-gation apparatus. We excluded data from (1) naturalCO 2  springs, because of their inherent variability andlack of control; (2) studies conducted in greenhouses orcontrolled environment chambers; and (3) studieswhere plants were grown in containers with surfaceareas  o 1m 2 or depths  o 1m. Most studies were con-ducted on existing soils,  in situ , but constructed soilswere included if the studies met our large containercriteria. However, constructed soils were excluded if (1)the soil was diluted with sand or similar materials or (2)the soil was imported from outside the region (usuallyto provide SOM with a contrasting C isotope signature).We also excluded one study with treatment differencescorresponding to accrual 4 800gCm  2 yr  1 .We required that plots were the unit of replication(not samples within a plot) and that treatments werereplicated at least twice, eliminating studies that aggre-gated data over a range of CO 2  concentrations. Theanalysis was also limited to studies that reportedmeans, standard deviations or standard errors, andnumber of replicates (or for which these parameterscould be obtained from the investigators). This criterioneliminated two otherwise eligible studies that wouldhave contributed six data points to the analysis.When results from the same experiment were re-ported over time, we used data from only the latestsampling date because meta-analysis assumes that stu-dies are independent (Gurevitch & Hedges, 1999).However, we adjusted the data for reported variationsin initial C stocks by normalizing to the overall averageof pretreatment C stocks; when pretreatment differ-ences were not reported, they were assumed to benegligible. When different vegetation, soil types orelevated-CO 2  concentrations were investigated at thesame facility, these studies were considered indepen-dent and were included in the analysis as separateobservations. Similarly, when studies incorporated ad-ditional manipulations, the results were included asseparate observations if independent ambient-CO 2  con-trols were available with the same additional manipula-tions as elevated-CO 2  plots. Results and discussion Organic C in the surface 5cm of the forest soil increasedlinearly during 5 years of exposure to elevated CO 2 ,    S  o   i   l   C   (  g  m    −    2    ) 2,5002,7002,9003,1003,300 (b)    S  o   i   l   C   (  g  m    −    2    ) 1,0001,1001,2001,3001,400 (a) Year0 1 2 3 4 5    S  o   i   l   N   (  g  m    −    2    ) 80859095100105110 (c) Fig. 1  Mean forest soil C and N stocks (   SE) in elevated (solidsymbols)andambient(opensymbols)CO 2 treatmentsovertime.(a)Carbon at 0–5cm; treatment  time interaction in repeated-mea-sures analysis of variance indicated differing effects of CO 2  treat-mentsovertime( P 5 0.032).(b)Carbonat0–15cm(treatment  time P 5 0.48). (c) Nitrogen at 0–5cm (treatment  time  P 5 0.081). ELEVATED-CO 2  INCREASES SOIL CARBON  2059 r  2005 Blackwell Publishing Ltd,  Global Change Biology ,  11,  2057–2064  while C in the ambient plots remained relatively con-stant (Fig. 1a). No significant changes in soil C werefound at deeper depths for either elevated or ambientCO 2 . Consistent with vegetative effects on soil forma-tion ( Jenny, 1941), increases in soil C storage, particu-larly in forests, are more likely to occur near the surface,where inputs from roots and aboveground litter aregreatest. If we had sampled to 15cm in one increment,the C accrued in the surface 5cm would have beendiluted with the unchanged C pool at 5–15cm, and wewould not have detected a significant change (Fig. 1b).This finding raises the possibility that actual changes insoil C have been masked in some CO 2  enrichmentstudies because the surface 10–20cm was sampled inone increment.Because the CO 2  source used at the forest site wasdepleted in  13 C ( d 13 C 5  55 % ), the  d 13 C of CO 2  in theelevated treatment was altered, which decreased the  13 Csignature of the vegetation (Matamala  et al ., 2003). Weused the subsequent decrease in the  d 13 C of SOM (Fig.2a) to determine that 441  5gm  2 of FACE-derived Cwas present in the surface 5cm after 5 years. Thisamount was sufficient to account for both the observed5-year accrual of 220gCm  2 and replacement of 221gm  2 of pretreatment C (Fig. 2b). With no changesin C stocks below 5cm, decreases in  d 13 C (Fig. 2a)accounted for the replacement of pretreatment C with196  49 and 99  30gm  2 of FACE-derived C at 5–15and 15–30cm, respectively. These amounts are consis-tent with a profile of declining C inputs with increasingsoil depth.In contrast to the forest, soil C at the prairie siteincreased significantly throughout the surface 30cmwith CO 2  enrichment (Table 1), reflecting the typicallyhigher root densities and greater C inputs at depth inprairie. Because the prairie site was on a slope, experi-mental blocks were positioned at different elevations.This design was essential for distinguishing the effect of CO 2  treatment on soil C from the influence of topogra-phy, another well-known soil forming factor (see sup-plementary Table S2).The detection of soil C accrual depends not only onsampling issues and variability, but also on the rate of inputs to SOM. Smith (2004) demonstrated that the timerequired for a change in soil C to become measurabledeclines as a function of the percentage stimulation in Cinputs. In both the forest and prairie experiments, the Year0 1 2 3 4 5    S  o   i   l   C   (  g  m    −    2    ) 6008001,0001,2001,400              1   3    C − 30 − 28 − 26 − 24 − 22 (a)(b) Fig. 2  Incorporation of free-air CO 2  enrichment (FACE)-de-rived C into forest soil over time. (a) Changes in the stableisotopic composition of whole soil C (means   SE) from theelevated-CO 2  treatment at 0–5cm ( } ), 5–15cm (  ), and 15–30cm( . ). (b) Loss of pretreatment C ( . ; shaded area) and accrual of FACE-derived C (  ; hatched area) at 0–5cm underelevated CO 2 (means   SE) calculated from the mass balance of stable Cisotopes. Table 1  Mean (  SE) whole soil C and N stocks and accrual by depth in prairie soil after 8 years of CO 2  treatmentDepth andtreatmentOrganic C(gm  2 )Total N(gm  2 ) 0–5cm Elevated CO 2  2494  132 202  9Ambient CO 2  2368  61 197  4Accrual(Probability 4  F )126  32(0.0162)5  3(0.1180) 0–15cm Elevated CO 2  6153  268 505  19Ambient CO 2  5884  164 494  11Accrual(Probability 4  F )269  77(0.0249)11  5(0.1145) 0–30cm Elevated CO 2  10370  436 865  34Ambient CO 2  9900  282 837  21Accrual(Probability 4  F )470  150(0.0352)27  10(0.0580)Values for chambered and unchambered ambient controls didnot differ ( P 4 0.14) and are combined ( n 5 6); for elevated CO 2 , n 5 3. Standard errors and probability 4 F for accrual valuesdetermined after adjusting for topographic effects on soil Cand N (see supplementary Table S2). Below 30cm, no signifi-cant response to CO 2  treatment was found in an independentset of random samples taken to 90cm (data not shown). 2060  J. D. JASTROW  et al . r  2005 Blackwell Publishing Ltd,  Global Change Biology ,  11,  2057–2064  vegetation responded to CO 2  enrichment by substan-tially increasing belowground C allocation. ElevatedCO 2  stimulated net primary production in the forest by an average of 24%, but 76% of this response wasallocated to fine roots after the first 2 years. From thethird year on, the peak standing crops of fine roots inthe surface 15cm increased by an average of 140%under elevated CO 2 , while leaf litter increased by only8% (Norby  et al ., 2004). Because of the rapid turnover of sweetgum fine roots (Matamala  et al ., 2003; Norby  et al .,2004), CO 2  stimulation of root production quickly de-livered a large increase of inputs to SOM, particularlywithin the surface 5cm (where root density was aboutthree times that at 5–15cm for both elevated andambient CO 2 ; data not shown). Below 5cm, the CO 2 response was apparently insufficient to discernibly in-crease soil C within 5 years. In the prairie, CO 2  enrich-ment stimulated root production by an average of 41%,compared with a 17% increase in aboveground produc-tion (Owensby  et al ., 1999; Jastrow et al ., 2000). Although belowground standing crops were two to four timesgreater in the prairie than in the forest, sweetgum rootsturn over three to four times faster than prairie grassroots (Hayes & Seastedt, 1987; Matamala  et al ., 2003).Hence, CO 2  enhancement of belowground inputs wasof similar magnitude in both forest and prairie surfacesoils and greater than in many other experiments withlower belowground productivity and/or smaller rela-tive responses  ( e.g. Hungate  et al ., 1997; Leavitt  et al .,2001; Gill  et al ., 2002; Pendall  et al ., 2004; Lichter  et al .,2005).The annual rate of soil C accrual was similar for bothexperiments, even though differences in ecosystemrooting strategies caused C to be accumulated at differ-ent depths. Atmospheric CO 2  enrichment increased Cstocks in the forest soil at an average rate of 44  9gCm  2 yr  1 . In the prairie, the incremental in-crease in C stocks corresponded to an average accrualrate of 59  19gCm  2 yr  1 . In an independent sam-pling of the prairie experiment, Williams  et al . (2004)found a similar rate of accrual within the surface 15cm.Soil C accrual at both sites was comparable with the52gCm  2 yr  1 accrued in the litter layer of a pine forestFACE experiment, where CO 2  enrichment stimulatedfoliage production more than root production (Lichter et al ., 2005). However, unlike C accumulated in surfacelitter, which is stabilized only by its biochemical resis-tance to decomposition, C accrued in mineral soil canalso be stabilized by physical protection and chemicalassociations with soil minerals, potentially increasingthe residence times of more labile constituents (Chris-tensen, 1996).In both experiments, a portion of the accumulated Cwas associated with soil minerals in stable aggregates.In the forest, the proportion of whole-soil C found inmicroaggregated soil averaged 58% in both elevated-CO 2  and ambient plots and was unchanged over time(Fig. 3a). This suggests that additional inputs derivedfrom CO 2  enrichment were processed and protected inmuch the same manner as in ambient soil, with littleapparent saturation of this protection mechanism, evenafter 5 years. The formation of microaggregates is a keyfactor in physically protecting particulate SOM fromrapid decomposition and helps to create conditionswherein microbial residues and breakdown productscan be stabilized in organomineral complexes (Golchin et al ., 1994; Christensen, 1996; Six  et al ., 2002).Even though the prairie soil had larger C stocks, 55%of the accrued C was also incorporated into microag- Depth (cm)0-5 5-15 15-30    E   l  e  v  a   t  e   d    −    a  m   b   i  e  n   t   i  n  c  r  e  m  e  n   t   (  g   f  r  a  c   t   i  o  n   C  m    −    2    ) − 1000100200300 (b) Year543210    F  r  a  c   t   i  o  n   C   (  g  m    −    2    ) 400500600700800Nonmicroaggregated CMicroaggregated C (a) *** Fig. 3  DistributionofCstocksbetweenmicroaggregatedandnon-microaggregated soil. (a) Changes in fraction C (means   SE)over time for forest soil in elevated (solid symbols) and ambient(open symbols) CO 2  treatments at 0–5cm; treatment  timeinteraction in repeated-measures analysis of variance indicateddiffering effects of elevated- and ambient-CO 2  treatments overtime (microaggregated  P 5 0.038; nonmicroaggregated  P 5 0.078). (b) Allocation of C stocks accrued in prairie soil underelevated CO 2  between microaggregated (solid bar) and nonmi-croaggregated (hatched bar) fractions by depth ( *  P    0.10; **  P o 0.05). ELEVATED-CO 2  INCREASES SOIL CARBON  2061 r  2005 Blackwell Publishing Ltd,  Global Change Biology ,  11,  2057–2064
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks