Root exclusion through trenching does not affect the isotopic composition of soil CO2 efflux

REGULAR ARTICLE

Root exclusion through trenching does not affect the isotopiccomposition of soil CO2 efflux

Nicolas Chemidlin Prévost-Bouré & Jérome Ngao &

Daniel Berveiller & Damien Bonal &Claire Damesin & Eric Dufrêne &

Jean-Christophe Lata & Valérie Le Dantec &

Bernard Longdoz & Stéphane Ponton &

Kamel Soudani & Daniel Epron

Received: 22 April 2008 /Accepted: 18 November 2008 / Published online: 10 December 2008# Springer Science + Business Media B.V. 2008

Abstract Disentangling the autotrophic and hetero-trophic components of soil CO2 efflux is critical tounderstanding the role of soil system in terrestrialcarbon (C) cycling. In this study, we combined astable C-isotope natural abundance approach with thetrenched plot method to determine if root exclusionsignificantly affected the isotopic composition (δ13C)of soil CO2 efflux (RS). This study was performed in

different forest ecosystems: a tropical rainforest andtwo temperate broadleaved forests, where trenchedplots had previously been installed. At each site, RS

and its δ13C (δ13CRs) tended to be lower in trenchedplots than in control plots. Contrary to RS, δ

13CRs

differences were not significant. This observation isconsistent with the small differences in δ13C mea-sured on organic matter from root, litter and soil. The

Plant Soil (2009) 319:1–13DOI 10.1007/s11104-008-9844-5

Responsible Editor: Per Ambus.

N. Chemidlin Prévost-Bouré (*) :D. Berveiller :C. Damesin : E. Dufrêne : J.-C. Lata :K. SoudaniUniv Paris-Sud, Laboratoire Ecologie Systématiqueet Evolution, AgroParisTech, CNRS,UMR 8079, Orsay,Paris F-75231, Francee-mail: [email protected]

J. Ngao : B. LongdozINRA, UMR INRA UHP 1137 Ecologie et EcophysiologieForestière, Centre INRA Nancy,Champenoux F-54280, France

D. Bonal : S. PontonINRA, UMR Ecofog, French Guiana,BP 709,97387 Kourou, Cedex, France

V. Le DantecUMR CESBIO, Équipe Modélisation du Fonctionnementdes Écosystèmes,BPI 2801,31401 Toulouse, France

D. EpronNancy Université, UMR INRA UHP 1137 Ecologieet Ecophysiologie Forestière, Université Henri Poincaré,Vandoeuvre-lès-Nancy, Cedex F-54506, France

Present address:J. NgaoUniv Paris-Sud, Laboratoire Ecologie Systématiqueet Evolution, AgroParisTech, CNRS,UMR 8079, Orsay,Paris F-75231, France

Present address:S. PontonINRA, UMR INRA UHP 1137 Ecologie et EcophysiologieForestière, Centre INRA Nancy,Champenoux F-54280, France

lack of an effect on δ13CRs by root exclusion could befrom the small difference in δ13C between autotrophicand heterotrophic soil respirations, but further inves-tigations are needed because of potential artefactsassociated with the root exclusion technique.

Keywords Stable carbon isotopes .

Natural abundance . Soil respiration . Trenched plot .

Rainforest . Temperate forest

Introduction

Soil plays an important role in the ecosystem carbon(C) cycle by sequestrating as much as 70% of totalforest ecosystem C as organic matter (Malhi et al.1999) and contributing 40–70% to total annual forestrespiration through soil respiration (RS) (Chambers etal. 2004; Epron et al. 2004a; Janssens et al. 2001;Bonal et al. 2008). As a consequence, RS is a majorcomponent determining the source or sink status of aforest ecosystem.

Several studies have focused on RS in variousecosystem types and all reported it was highly spatiallyand temporally variable (Buchmann 2000; Epron et al.2004a, b; Fang and Moncrieff 2001; Gaumont-Guay etal. 2006; Högberg et al. 2001; Longdoz et al. 2000;Saiz et al. 2006; Salimon et al. 2004; Yim et al. 2003).Both spatial and temporal variability of RS were relatedto variations in environmental factors and in C inputsvia roots or litter (Epron et al. 2004b; Fang et al. 1998;Högberg et al. 2001; Longdoz et al. 2000) affecting theuse of two C pools in soil (Epron et al. 2001): (1) ashort residence-time C pool (fast C pool) of photosyn-thetic assimilates and root exudates that are respired asautotrophic respiration (Ra) by roots and associatedmicroorganisms; and (2) a long residence-time C pool(slow C pool) of litter and soil organic matter (SOM)that are respired as heterotrophic respiration (Rh)mainly by microorganisms not associated with roots.Disentangling the processes using these two C pools isdecisive in understanding soil functioning and model-ing soil respiration in terrestrial ecosystems in thecontext of global environmental change (Baggs 2006).

To separate the components of soil respiration thatuses either the slow or fast C pools, different methodsare applied. Non-isotopic methods are based on cuttingoff the fast C pool inputs either by trenching (Boone etal. 1998; Buchmann 2000; Lee et al. 2003; Li et al.

2006) or tree girdling (Gottlicher et al. 2006; Högberget al. 2001). Applying these methods on differentecosystems significantly reduces RS and allows calcu-lation of the heterotrophic contribution to RS (see thereview of Subke et al. 2006). However, these methodsare still subject to uncertainties linked to decomposi-tion of severed roots or soil water content differencesamong plot types (Ngao et al. 2007).

C isotope-based methods are extensively used tostudy soil functioning. There have been many in situmeasurements of the isotopic composition of soil-respired CO2 in both C3 and C4 ecosystems (Buch-mann et al. 1997; Davidson 1995; Ekblad et al. 2005;Fessenden and Ehleringer 2003; Steinmann et al.2004). These methods allow targeting of specific soilprocesses via isotopic labeling (Andrews et al. 1999;Kuzyakov et al. 2001; Ngao et al. 2005) or picturingthe whole soil functioning at natural abundance. Thislast method successfully separated the Ra and Rh

components of RS in ecosystems characterized by C3 /C4 successions (Cheng 1996; Rochette et al. 1999).This separation was possible because the isotopiccompositions (through the δ13C notation) of the fastand slow C pools were very different (> 10‰).However, most terrestrial ecosystems do not havesuch large differences between the fast and slow Cpools, i.e. differences in δ13C < 2‰ are common(Balesdent et al. 1993; Bowling et al. 2008). In areview on C3 ecosystems, in vitro root respiration wasclearly depleted compared to in vitro soil respiration(Bowling et al. 2008). On the contrary, microbiallyrespired CO2 seems to be 13C enriched by 1–4‰compared to bulk SOM (Andrews et al. 1999;Bowling et al. 2008; Tu and Dawson 2005). Suchobservations should be confirmed in situ, but similardiscrepancies would be expected between autotrophicand heterotrophic soil respirations. According toPhillips and Gregg (2001), such discrepancies wouldlead to accurate estimates of autotrophic and hetero-trophic contributions to total RS. Therefore, it is validto determine if natural abundance of stable C-isotopescan separate Ra and Rh in ecosystems with smalldifferences of isotopic composition between the slowand fast C pools. To our knowledge, only two studiestackled this question using soil static-chamber meth-ods. In a boreal forest, Subke et al. (2004) showedthat CO2 efflux from soil without root respiration(girdled plots) was 13C-depleted compared to rootedsoil (ungirdled plots), but differences were not

2 Plant Soil (2009) 319:1–13

significant. In another study, Formanek and Ambus(2004) performed an in vitro / in situ experiment andshowed that δ13C of root respiration was in—betweenthose of soil humus and mineral horizons respiration.

The main objective of this study was to determinewhether δ13C of soil CO2 efflux measured in situ wassignificantly affected by root exclusion in differentecosystems. Keeling plots (Keeling 1958) made withclosed dynamic systems in a tropical evergreen forest(Paracou, French Guyana) and in two temperatedeciduous forests (Barbeau and Hesse, France) wereperformed to compare the isotopic composition ofCO2 evolved from control and trenched plots.

Material and methods

Experimental sites

The Paracou site (French Guiana, 35 m elevation) islocated in the Paracou Research Station (Gourlet-Fleury et al. 2004) within the “Guyaflux experimentalunit” that covers 200 ha of tropical wet forest (Bonalet al. 2008). Mean annual air temperature and rainfallover the past 10 years (1998–2008) were 25.7°C and3,041 mm, respectively. Soils are mostly acrisols(FAO-ISRIC-ISSS 1998) developed over a Precam-brian metamorphic formation called the “Bonidoroseries” and composed of schist and sandstone.

The Hesse experimental sites (Hesse 1 and Hesse 2)are in the Hesse National Forest (CARBOEUROPE site,northeastern France, 300 m elevation; Granier et al.2000). The Hesse 1 site is in the center of a 65 ha standof 35 year-old European beech (Fagus sylvatica L.).The Hesse 2 site is in a mixed 20 year-old standdominated by European beech. Mean annual airtemperature and precipitation are 9.2°C and 820 mm,respectively. The soil is a stagnic luvisol (FAO-ISRIC-ISSS 1998) of 120 cm depth covered by an oligo-mullhumus.

The Barbeau experimental sites (Barbeau 1 andBarbeau 2) are located in Barbeau National Forest(CARBOEUROPE site, southeast of Paris, France,90 m elevation). The Barbeau 1 site is located in anoak (Quercus petraea L.) high forest stand. TheBarbeau 2 site is located in an oak forest with cop-piced hornbeam stand. Mean annual air temperatureand rainfall are 10.7°C and 690 mm, respectively(1980–1996). Soil is a gleyic luvisol (FAO-ISRIC-ISSS

1998) of 80 cm depth, developed on millstone bedrockand covered by an oligo-mull humus.

All sites belong to the French network of forestecosystems (Observatoire de Recherche en Environ-nement “fonctionnement des écosystèmes forestiers”).

Experimental design

Trenched plots (TP) were installed at each experi-mental site to suppress the autotrophic component oftotal soil respiration. Trenches were dug around areaswithout trees and the delimited plots were lined usinga thick plastic film to prevent external root ingrowth.One control, untrenched plot (CP), was selected neareach TP. TP and CP characteristics for each experi-mental site are summarized in Table 1. At Paracousite, trenches were dug down to a stone linecomposed of coarse fragments of lithorelics (60 cmdepth). At Barbeau sites, they were dug down to thebedrock (80 cm depth). At Hesse sites, trenches weredug down to the clay horizon (90 cm depth). In allcases, these layers are almost impermeable to roots,preventing root ingrowth. At the time of the measure-ments, at any site, there was no evidence of rootingrowth into the TP plots. At Paracou sites, rootingrowth was observed only several months after thecompletion of this study, associated to a sharpincrease of soil respiration in the TP plots (data notshown). At Barbeau and Hesse sites, such an increaseof soil respiration was not observed (data not shown).

Measurements started at least 1 month after trench-ing (Lee et al. 2003), with details in Table 1. Themeasurements were carried out during the mainphenological phases: leafy and unleafy seasons intemperate forests (Hesse and Barbeau); or during themain climatic periods: wet and dry periods in thetropical forest (Paracou). At Hesse sites, measurementswere in March, May and September 2005. At Barbeausites, measurements were from March 2005 to May2006. At the Paracou site, measurements were at theend of the long dry season of August–November 2004,and during the short dry period of March 2005 thatinterrupts the rainy season of December–July.

Isotopic composition of soil CO2 efflux

The isotopic composition of soil CO2 efflux (δ13CRs)was determined on CP and TP using the Keeling plotmethod (Keeling 1958). Soil CO2 efflux was mea-

Plant Soil (2009) 319:1–13 3

sured by an accumulation chamber connected to aninfrared gas analyzer (EGM 1 or EGM4, PP Systems,Hitchin, UK, in Barbeau and Paracou, respectively;and LI-6200, Li-Cor Inc, Lincoln, NE, USA, inHesse) in closed circuit mode. The Li-Cor chambermodel Li-600-9 was used at Hesse, while laboratory-made chambers were used at Barbeau and Paracousites. These two chambers were constituted of aPerspex (acrylic resin) cylinder (25.4 dm3, 12 cmheight in Barbeau and 5 dm3, 52 cm height inParacou) equipped with a small fan (airflow 13.5 m3

h−1). During measurements, soil respiration chamberswere laid on permanent collars inserted 2-3 cm intosoil (one per TP and one per CP at Paracou andBarbeau; three per CP and two per TP at Hesse).Chambers were equipped to allow air sampling duringmeasurements

Air samples were collected every 50–100 ppmCO2 increase in the range 400–1,000 ppm. AtBarbeau sites, five air samples were collected duringCO2 increase using 50 cm3 valved syringes (SGE,Australia) directly connected to the chamber. At bothHesse and Paracou sites, 5-6 air samples werecollected during CO2 increase using a specificsampling device allowing air to be driven from thechamber into 10 cm3 Exetainer glass vials (Labco,High Wycombe, UK), and closing of the Exetainervial in airtight conditions. This system is described indetail by Ngao et al. (2005). At the time of syringecollection (Barbeau) or Exetainer cap closure (Hesseand Paracou), the CO2 concentration was recorded.The time lag between detection in the gas analyzerand the vial collection was < 1 s due to the flow rateof the instruments (0.3–1.5 L min−1). The expected[CO2] changes during these time lags were within therange of instrumental error (1 ppm).

The isotopic composition of sampled CO2

(δ13CCO2) was determined using isotopic ratio massspectrometers (IRMS): a VG Optima IRMS (Fison,Villeurbanne, France) connected to an elementalanalyzer (model NA-1500, Carlo Erba, Milan, Italy)for the Barbeau samples, and a Delta S IRMS (DeltaS, ThermoFinnigan, Bremen, Germany) connected toa gas purification device (Gas-Bench II, ThermoFin-nigan) for the samples from Hesse and Paracou. Themethods cited above are described in Maunoury et al.(2007) for the Barbeau samples; and in Ngao et al.(2005) for Hesse and Paracou. Analyses were per-formed within a week after each field session.T

able

1Site

characteristicsanddescriptionof

trenched

(TP)andcontrol(CP)plots.trefers

tothetim

edelaybetweentrenchingandmeasurements.nd

:no

tdeterm

ined

Site

Paracou

Barbeau

1Barbeau

2Hesse

1Hesse

2

Location

5°16'N

;52°54'

W48°29’N,02°47’E

48°29’N,02°47’E

48°40’N,7°05’E

48°40’N,7°05’E

Clim

ate

Tropical

Modifiedtemperate

maritime

Modifiedtemperate

maritime

Contin

ental

Contin

ental

Stand

type

Mixed

forest

Oak

high

forest

Oak

forestwith

coppiced

hornbeam

Beech

high

forest

Mixed

beechforest

Stand

agein

2006

(years)

nd100to

150

100to

150

3520

Groundarea

(mha

−1)

nd20.7

20.7

25.3

25.3

Max

tree

height

(m)

35–4

030

3020

20Dom

inantspecies

ndQuercus

petraeaL.

Quercus

petraeaL.

Fagus

sylvaticaL.

Fagus

sylvaticaL.

Quercus

petraeaL.

Quercus

petraeaL.

Understorey

>140sp.ha

−1

CarpinusbetulusL.

CarpinusbetulusL.

BetulapendulaL.

BetulapendulaL.

Represented

species

CarpinusbetulusL.

CarpinusbetulusL.

t(m

onth)

61

115

15Num

berof

TP/CP

4/4

1/1

1/1

1/1

1/1

Areaof

TP/CP(m

2)

0.64

/0.64

6/6

6/6

9/35

9/35

Trenching

depth(m

)0.6

0.8

0.8

0.9

0.9

Num

berof

fieldsessions

210

53

3Measurementsperfieldsession(R

S/δ1

3CRs)

4/4

10/1

10/1

2to

4/2to

42to

4/2to

4

4 Plant Soil (2009) 319:1–13

Isotopic analysis standard error was 0.2‰. Allisotopic composition values are expressed relative tothe international PDB standard.

Isotopic composition of organic matter

Soil (0–15 cm) and aboveground litter were sampledduring each field session at Paracou and Hesse. AtBarbeau, soil and aboveground litter were sampledduring three field sessions. For each sampling session,one measurement per CP plot was performed. Eachmeasurement was performed on one soil sample atParacou and on a composite sample (3 soil cores) atHesse and Barbeau. Soil was sieved to 1 mm. All fineroots (< 1 mm diameter) were separated from bulksoil after sieving. Samples were dried at 60°C andground. Ground litter, root and soil samples wereanalyzed for δ13C determination using an elementalanalyzer (model NA-1500) connected to the IRMS.

Soil respiration (RS) measurements

At Hesse and Paracou sites, RS was measured before airsample collection using the same chamber as previ-ously described for the determination of δ13CRs. At theBarbeau sites, RS was measured on 10 collars each inthe CP and TP. These collars were inserted 2 cm intothe soil in the vicinity of the Keeling plot collar.Measurements were performed with a soil respirationchamber (SRC-1, PP systems) linked to a CIRAS-1 gasanalyzer (PP systems). Wind speed, measured insidethe chamber with a thermal anemometer (Testo, ModelLenzkirch, Germany) was 0.4 m s−1 as recommendedin Le Dantec et al. (1999). In addition, no systempresented any difference in pressure between theheadchamber and the outside, arguing for no majorleak and no perturbation of the CO2 diffusion gradientdue to over/underpressure (Longdoz et al. 2000).

Data analysis

In each forest, RS recorded on TP were not correctedfor dead root decomposition, and arithmetic means ofRS were calculated for each field session: 10 collars atBarbeau, 4 at Paracou, and 2–4 at Hesse sites.

The δ13CRs was estimated from a linear regressionfitted through “transformed” CO2 concentration([CO2]) and δ13CCO2 values. We tested two differenttransformation methods. Firstly, a linear regression was

fitted through the inverse of [CO2] (independentvariable) and δ13CCO2 (dependent variable) whereδ13CRs was the intercept (Keeling 1958). Secondly, alinear regression was fitted through [CO2] (indepen-dent variable) and the product of [CO2] and δ13CCO2

(dependent variable) where δ13CRs was the slope(Miller and Tans 2003). In both cases, we used bothordinary least squares linear regression (model I) andgeometric mean linear regression (model II) (Sokal andRohlf 1995). Model I assumes that the independentvariable is measured without random error and theoptimization of the regression parameters is made tominimize errors on the dependent variable. Model II ismore robust since it assumes that both dependent andindependent variables are measured with some error.Model II optimizes the regression parameters tominimize errors on both dependent and independentvariables (Sokal and Rohlf 1995). The correlationcoefficient (r) was calculated in each case.

The standard error of δ13CRs was determined as thestandard error of either the intercept or the slope ofthe regression when Keeling or Miller-Tans methodswere used, respectively.

In Hesse and Paracou, mean δ13CRs (d13CRs)values obtained over replicate collars and the standarderror of the mean (SEðd13CRsÞ) were calculated asdescribed in Murtaugh (2007). In this method, meanδ13CRs and its standard error are corrected for theerror made on each estimation.

d13CRs ¼Xn

i¼1

wi*d13CRs;i ð1Þ

with wi ¼1

SE d13CRs;ið Þ2Pn

i¼1

1

SE d13CRs;ið Þ2

0BB@

1CCA ð2Þ

SE(δ13CRs,i) is the standard error of δ13CRs ith

estimation, and n the total number of observations(n=4 in Paracou; n=2–4 in Hesse).

SE d13CRs

� � ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1

n

� �*Xn

i¼1

wi* d13CRs;i � d13CRs

� �2� s

ð3ÞOverlapping of confidence intervals calculated

from the standard error of the mean in Hesse and

Plant Soil (2009) 319:1–13 5

Paracou, and from the standard error of the parameterin Barbeau (no replication), was used as a criterion todetermine if δ13CRs was significantly different be-tween TP and CP for each date of measurement. Thesame criterion was used to test if δ13C differedbetween root, soil and litter bulk organic matter.Treatment effect on RS was tested by one wayanalysis of variance for repeated measurements (oneway RM-ANOVA) on raw data. Significance levelwas P<0.05.

Results

Comparison of δ13CRs estimation methods

The different transformation methods and regressionmodels were compared using the complete datasetfrom the three forests. First, Miller–Tans and Keelingtransformations were compared using the same model(linear regression model I in Fig. 1a; model II in

Fig. 1b). δ13CRs estimated by Keeling method wassimilar to that estimated by Miller–Tans method eitherwhen linear regression model I or model II wereapplied (slopes=0.98 and r=0.98). In each case,slopes were not significantly different from 1.

Model I and II linear regressions were comparedfor the same variable transformation method (Keelingin Fig. 1c; Miller–Tans in Fig. 1d). In both case,estimations of δ13CRs were equivalent, whatever thelinear regression model used (Fig. 1c and d; slope=0.99, r=0.98). Thus, whatever the chosen transfor-mation method and linear regression model, therewere similar estimates of δ13CRs.

Several outliers were excluded from the 95%confidence interval of the regression line (Fig. 1)and were characterized by a standard error > 5%(corresponding approximately to a 1‰ error here) ofthe estimated δ13CRs. Afterwards, these outliers wereexcluded from the dataset. Since both transformationmethods and both regression models gave similarestimates of δ13CRs; that obtained by the Keeling

Fig. 1 Relationships between δ13CRs estimated using the twotransformation methods (Keeling or Miller–Tans) and the twolinear regression models (Model I or model II). δ13CRs,K I isδ13CRs estimated using Keeling method associated to linearregression model I. δ13CRs,K II is δ13CRs estimated usingKeeling method associated to linear regression model II.

δ13CRs,MT I is δ13CRs estimated using Miller–Tans methodassociated to linear regression model I. δ13CRs,MT II is δ13CRs

estimated using Miller–Tans method associated to linearregression model II. Open squares correspond to outliers thatwere excluded from the 95% confidence interval

6 Plant Soil (2009) 319:1–13

transformation fitted with linear regression model Iwas further used to compare δ13CRs of CP and TP.

Soil respiration

As expected, over the entire experiment period, RS

was always higher in CP than in TP at all experi-mental sites (Fig. 2). RM-ANOVA analysis on rawdata showed that over this period trenching had asignificant effect at Paracou and Barbeau (P<0.05and 0.01, respectively), but not at Hesse site (P<0.3and 0.6 at Hesse 1 and 2, respectively).

Isotopic composition of soil CO2 efflux

In Paracou, the difference between CP and TP wasonly −0.4‰ in October 2004 and +0.3‰ in March2005, δ13CRs being 1.3‰ less negative in March thanin October (Fig. 3a). In Hesse 1 and Hesse 2,differences between δ13CRs values in CP and TPranged between −1.0 and +1.9‰, depending on thesampling date (Fig. 3b and c). In Barbeau 1 andBarbeau 2, the difference in δ13CRs between CP andTP also showed a temporal variability with the δ13CRs

in TP being most often lower than CP (Fig. 3d and e).

Fig. 2 Temporal variationsof soil respiration in a Par-acou (n=4), b and c Hesse 1and Hesse 2 (n=2 to 4); dand e Barbeau 1 and Bar-beau 2 (n=8 to 10). Controlplots (CP) measurementscorrespond to the filledsymbols and trenched plots(TP) measurements corre-spond to the open symbols.Error bars represent ± 1standard error of the mean

Plant Soil (2009) 319:1–13 7

The difference between CP and TP ranged from −1.5to +4.5‰. However, at each site, these differenceswere not significant, except at the end of April 2005in Barbeau 1, and in mid-June in Barbeau 2.

δ13C of bulk organic matter

Mean δ13C of root, litter and soil bulk organic matterare reported in Table 2 (ROM, LOM and SOM,respectively). At all sites, ROM and LOM weresignificantly different from SOM (P<0.05). ROMwas significantly different from LOM at Paracou andHesse (P<0.01).

Discussion

Partitioning soil respiration into autotrophic andheterotrophic components is necessary to analyze theresponse of soil respiration to disturbances or changesin climate; however, most available methods arebased on root exclusion and have strong potentialdrawbacks (Balesdent and Mariotti 1996; Ngao et al.2007). The usefulness of stable isotope signatures forpartitioning autotrophic and heterotrophic compo-nents has been recognized (Kuzyakov 2006) but it isstill scarcely used (Rochette et al. 1999) and itsapplicability to forest ecosystems not tested. In such

Fig. 3 a Temporal varia-tions of soil respired δ13C-CO2in (a) Paracou (n=4), band c Hesse 1 and Hesse 2n=2 to 4; d and e Barbeau 1and Barbeau 2 (n=1). Con-trol plots (CP) measure-ments correspond to thefilled symbols and trenchedplots (TP) measurementscorrespond to the opensymbols. At Paracou andHesse sites, error bars rep-resent ± 1 standard error ofthe mean. At Barbeau site,error bars correspond to ± 1standard error of the esti-mated δ13C

8 Plant Soil (2009) 319:1–13

approaches, a single isotope linear-mixing modelthat is based on mass conservation equations is com-monly applied (Balesdent and Mariotti 1996) andreported here:

Ra

Rs¼ dRs � dhð Þ

da � dhð4Þ

RaRs is the contribution of the autotrophic component

of soil respiration; δa and δh are the respectiveisotopic composition of autotrophic and heterotrophicsources; and δRs is the isotopic composition of RS.

Applying this equation requires accurate estimatesof the isotopic composition of soil respiration;differences in stable C-isotopic composition betweentotal soil respiration and its heterotrophic compo-nents; and contrasted isotopic compositions of thedifferent sources. These requirements are evaluatedbelow.

Towards accurate estimates of the isotopiccomposition of soil CO2 efflux

The use of the Keeling or Miller–Tans transformationmethods and the linear regression models I or II toderive the isotopic signature of a respiratory flux arestill debated. Some authors recommend model II(Pataki et al. 2003) while others state that only modelI provides an unbiased estimate of the isotopicsignature of a respiratory flux (Zobitz et al. 2006).In this study, both transformation methods and bothregression models led to very similar estimates. Thiswas partly due to high r values and the large [CO2]ranges used to establish the relationship between[CO2] and δ13CCO2. A large [CO2] range reduces theeffect of low [CO2] values on the estimation ofregression parameters (Pataki et al. 2003; Sokal andRohlf 1995). The comparison of the different methods

also showed that they did not converge when thestandard error of δ13CRs estimate was > 5% of theestimated value. This convergence could be a criteri-on to evaluate accuracy of δ13CRs estimates.

Evaluating the difference of stable C-isotopecomposition between total soil respirationand its heterotrophic component

RS measured in CP and TP in Paracou are in the rangefound in tropical forests (Buchmann et al. 1997;Chambers et al. 2004; Epron et al. 2004b). Thosemeasured in Hesse and Barbeau are in agreement withprevious studies of temperate forests (Boone et al.1998; Buchmann 2000; Epron et al. 2001). Over theyear, RS in CP were higher than in TP at all sites, butdifferences were only significant in Paracou andBarbeau. At Hesse, the decrease in RS by trenchingwas not significant, probably because of the smallnumber of replicates; only three dates of measure-ments were considered in this study. When consider-ing the whole set of measurements made on theseplots (every 2 weeks from March 2004 to May 2005),the difference in RS between TP and CP weresignificant (Ngao et al. 2007). The differences wereconsistent with previous reports (Bowden et al. 1993;Epron et al. 2001; Lalonde and Prescott 2007; Subkeet al. 2006). This highlights that the CO2 releaseduring the decomposition of severed roots only partlycompensates for decreased respiration due to theremoval of root respiration. Therefore trenchingefficiently suppressed root respiration.

The number of replications in isotope data wasconstrained by cost and the time required for δ13CRs

measurements. Whether the studied forest ecosystemwas temperate or tropical, the δ13C of soil respiredCO2 (δ

13CRs) of CP were consistent with the literature.δ13CRs ranges from −28 to –21‰ (Bhudinperpal-Singh

Table 2 Mean (± SE) isotopic composition of root (ROM), litter (LOM) and soil (SOM) bulk organic matter, at Paracou (n=7), Hesse(n=3), and Barbeau (n=3)

Site plot Root organic matter (ROM) Litter organic matter (LOM) Soil organic matter (SOM)

Paracou Mixed forest −29.4±0.3 −30.9±0.3 −28.6±0.1Barbeau 1 Oak high forest −28.2±0.2 −28.2±0.2 −27.0±0.3Barbeau 2 Oak forest with coppiced hornbeam −28.1±0.3 −28.1±0.2 −27.0±0.3Hesse 1 Beech high forest −27.4±0.2 −29.0±0.2 −26.4±0.1Hesse 2 Mixed beech forest −28.0±0.1 −29.5±0.2 −26.9±0.2

Plant Soil (2009) 319:1–13 9

et al. 2003; Ekblad and Hogberg 2001; Fessenden andEhleringer 2003; Mortazavi et al. 2005; Ngao et al.2005; Steinmann et al. 2004) and from −28 to –26‰(Buchmann et al. 1997; Salimon et al. 2004), intemperate and tropical forests, respectively.

The comparison of δ13CRs measured in control androot exclusion (trenching) conditions has never beenreported. This difference of δ13CRs between CP andTP plots could be used to approximate the difference(δRs−δh) in Eq. 4. Here, despite the difference ofstable C-isotope composition observed in vitro be-tween root and microbial CO2 (Andrews et al. 1999;Bowling et al. 2008; Tu and Dawson 2005), therewere no pronounced differences between CP and TP,in agreement with observations of Subke et al. (2004)on girdled plots. This lack of marked difference inδ13C between TP and CP is unlikely to be due torespiration of severed roots. The delay betweentrenching and measurements was very variable amongsites and soil respiration was already reduced by 12–43% in TP without correcting for severed rootdecomposition. The difference of δ13CRs betweenCP and TP could have been significant with morereplications (n=20, standard method based on thenormal distribution of differences; Sokal and Rohlf1995). However, the difference would remain small,showing that the stable C-isotopic compositions ofCO2 from the autotrophic and heterotrophic compo-nents of soil respiration are likely to be similar.Consequently, measuring δ13CRs in— and outsidetrench plots likely provides no additional information,beyond measuring RS, to estimating the relativecontribution of Ra and Rh to RS. Nevertheless,although this observation is a worthy methodologicalpoint on the use of δ13C in root exclusion studies,further studies are needed. Indeed, the decompositionof root organic matter added by trenching maysmooth δ13CRs differences between the CP and theTP plots. On another hand, trenching is known formodifying soil moisture in the TP plot, which couldaffect organic matter decomposition. At Barbeau sites,no correlation was found between δ13CRs and soilmoisture throughout the season. Moreover, smalldifferences were also observed during girdling experi-ments where soil moisture was not affected by thetreatment (Bhudinperpal-Singh et al. 2003; Högberget al. 2001; Subke et al. 2004). Therefore, soilmoisture modification may have little influence onδ13CRs differences between TP and CP plots.

Evaluating the difference of stable C-isotopecomposition between autotrophic and heterotrophicsources

At all sites, δ13CRs was higher than δ13C for any kindof organic matter, as frequently reported (Buchmannet al. 1997; Ekblad and Hogberg 2001; Fessenden andEhleringer 2003; Mortazavi et al. 2005; Ngao et al.2005). These discrepancies between soil CO2 effluxand bulk organic matter are considered as apparentisotopic fractionation underlying several activelystudied mechanisms that are far beyond the scope ofthis study. Assuming similar fractionations for allrespiratory sources, the difference between ROM andeither SOM or LOM could be used to approximate thedifference (δa−δh) in Eq. 4. This approximation couldalso be done by the difference in stable C-isotopecomposition between root and soil respired CO2

measured in vitro (Tu and Dawson 2005). In anycase, this difference would be small.

Conclusion

In the present study, we showed that the δ13C of soilCO2 efflux was not significantly affected by rootexclusion in three C3 ecosystems. This result is inagreement with the small δ13C differences observedbetween bulk organic matter of root, litter and soil.Applying the differences (δRs−δh) and (δa−δh) esti-mated above in Eq. 4 would lead to large uncertaintiesin the partitioning of soil respiration (Phillips andGregg 2001; 2003); because differences are smalldespite their potential statistical significance. There-fore, the lack of marked difference would greatly limitthe applicability of stable C-isotopes as a tool forseparating Ra and Rh in C3 ecosystems. However,seasonal changes in δ13CRs and in the difference inδ13CRs between CP and TP probably reflects temporalchanges in the isotopic composition of availablesubstrates for respiration, especially for roots andmicrobes in the rhizosphere that rely on recentlyassimilated carbon. The analysis of this temporalsignal in soil respiration would be an opportunity toquantify the contribution of recently assimilatedcarbon to soil respiration, but will require frequentor even continuous measurements of soil respirationand its isotopic signature, certainly by means of 13Clabeling.

10 Plant Soil (2009) 319:1–13

Acknowledgements The authors wish to acknowledge the“Office National des Forêts” for facilitating experimental workin Hesse and Barbeau forests, and CIRAD in Paracou forest.They also thank Jean-Yves Goret, Nathalie Têtefort, ElliLentilus, Bernard Clerc, François Willm, Dyane Franey,Laurent Vanbostal and Jean-Yves Pontailler for their supportduring the field experiments in the different forests.

This work was supported by the French Nationalprogramme ‘ACI/FNS ECCO, PNBC’ and by the ‘Observatoirede Recherche en Environnement “fonctionnement des écosys-tèmes forestiers’ (F-ORE-T).

Authors also acknowledge the ‘Métabolisme-Métabolome’platform of the IFR87 and the ‘isotopic spectrometry’ platformof the IFR 110 for the isotopic facilities and Christan Hossan,Claude Brechet and Caroline Lelarge for isotopic analyses. Thestudy complies with current French law.

References

Andrews JA, Harrison KG, Matamala R, Schlesinger WH (1999)Separation of root respiration from total soil respirationusing 13C labeling during free-air carbon dioxide enrich-ment (FACE). Soil Sci Soc Am J 63:1429–1435

Baggs EM (2006) Partitioning the components of soil respira-tion: a research challenge. Plant Soil 284:1–5 doi:10.1007/s11104-006-0047-7

Balesdent J, Mariotti A (1996) Measurement of soil organicmatter turnover using δ13C natural abundance. In: BouttonT (ed) Mass spectrometry (pp. 83–111). New York.

Balesdent J, Girardin C, Mariotti A (1993) Site-Related δ13C oftree leaves and soil organic-matter in a temperate forest.Ecology 74:1713–1721 doi:10.2307/1939930

Bhudinperpal Singh, Nordgren A, Ottosson Lofvenius M,Högberg M, Mellander P, Högberg P (2003) Tree rootand soil heterotrophic respiration as revealed by girdlingof boreal Scots pine forest: extending observations beyondthe first year. Plant Cell Environ 26:1287–1296doi:10.1046/j.1365-3040.2003.01053.x

Bonal D, Bosc A, Goret JY, Burban B, Gross P, Bonnefond JM,Elbers J, Ponton S, EpronD, Guehl JM, Granier A (2008) Theimpact of severe dry season on net ecosystem exchange in theneotropical rainforest of FrenchGuiana. GlobChange Biol 14(8):1917–1933 doi:10.1111/j.1365-2486.2008.01610.x

Boone RD, Nadelhoffer KJ, Canary JD, Kaye JP (1998) Rootsexert a strong influence on the temperature sensitivity ofsoil respiration. Nature 396:570–572 doi:10.1038/25119

Bowden RD, Nadelhoffer KJ, Boone RD, Melillo JM, GarrisonJB (1993) Contributions of aboveground litter, below-ground litter, and root respiration to total soil respiration ina temperature mixed hardwood forest. Can J Res 23:1402–1407 doi:10.1139/x93-177

Bowling DR, Pataki DE, Randerson JT (2008) Carbon isotopesin terrestrial ecosystem pools and CO2 fluxes. New Phytol178:24–40 doi:10.1111/j.1469-8137.2007.02342.x

Buchmann N (2000) Biotic and abiotic factors controlling soilrespiration rates in Picea abies stands. Soil Biol Biochem32:1625–1635 doi:10.1016/S0038-0717(00)00077-8

Buchmann N, Guehl JM, Barigah TS, Ehleringer JR (1997)Interseasonal comparison of CO2 concentrations, isotopic

composition, and carbon dynamics in an Amazonianrainforest (French Guiana). Oecologia 110:120–131doi:10.1007/s004420050140

Chambers JQ, Tribuzy ES, Toledo LC, Crispim BF, Higuchi N,dos Santos J, Araujo AC, Kruijt B, Nobre AD, TrumboreSE (2004) Respiration from a tropical forest ecosystem:partitioning of sources and low carbon use efficiency. EcolAppl 14:S72–S88 doi:10.1890/01-6012

Cheng W (1996) Measurement of rhizosphere respiration andorganic matter decomposition using natural 13C. Plant Soil183:263–268 doi:10.1007/BF00011441

Davidson GR (1995) The stable isotopic composition andmeasurement of carbon in soil CO2. Geochim CosmochimActa 59:2485–2489 doi:10.1016/0016-7037(95)00143-3

Ekblad A, Hogberg P (2001) Natural abundance of 13C in CO2

respired from forest soils reveals speed of link betweentree photosynthesis and root respiration. Oecologia127:305–308 doi:10.1007/s004420100667

Ekblad A, Boström B, Holm A, Comstedt D (2005) Forest soilrespiration rate and δ13C is regulated by recent aboveground weather conditions. Oecologia 143:136–142doi:10.1007/s00442-004-1776-z

Epron D, Le Dantec V, Dufrene E, Granier A (2001) Seasonaldynamics of soil carbon dioxide efflux and simulatedrhizosphere respiration in a beech forest. Tree Physiol21:145–152

Epron D, Ngao J, Granier A (2004a) Interannual variation ofsoil respiration in a beech forest ecosystem over a six-yearstudy. Ann Sci 61:499–505 doi:10.1051/forest:2004044

Epron D, Nouvellon Y, Roupsard O, Mouvondy W, Mabiala A,Saint-Andre L, Joffre R, Jourdan C, Bonnefond J-M, BebigierP, Hamel O (2004b) Spatial and temporal variations ofsoil respiration in a eucalyptus plantation in Congo. ForEcol Manage 202:149–160 doi:10.1016/j.foreco.2004.07.019

Fang C, Moncrieff JB (2001) The dependence of soil CO2

efflux on temperature. Soil Biol Biochem 33:155–165doi:10.1016/S0038-0717(00)00125-5

Fang C, Moncrieff JB, Gholz HL, Clark KL (1998) Soil CO2

efflux and its spatial variation in a Florida slash pineplantation. Plant Soil 205:135–146 doi:10.1023/A:1004304309827

Fessenden JE, Ehleringer JR (2003) Temporal variation in δ13Cof ecosystem respiration in the Pacific Northwest: links tomoisture stress. Oecologia 136:129–136 doi:10.1007/s00442-003-1260-1

Formanek P, Ambus P (2004) Assessing the use of δ13C naturalabundance in separation of root and microbial respiration ina Danish beech (Fagus sylvatica L.) forest. Rapid CommunMass Spectrom 18:897–902 doi:10.1002/rcm.1424

Gaumont-Guay D, Black TA, Griffis TJ, Barr AG,Morgenstern K,Jassal RS, Nesic Z (2006) Influence of temperature anddrought on seasonal and interannual variations of soil, boleand ecosystem respiration in a boreal aspen stand. Agric ForMeteorol 140:203–219 doi:10.1016/j.agrformet.2006.08.002

Gottlicher SG, Steinmann K, Betson NR, Hogberg P (2006) Thedependence of soil microbial activity on recent photosyn-thate from trees. Plant Soil 287:85–94 doi:10.1007/s11104-006-0062-8

Gourlet-Fleury S, Laroussinie O, Guehl J (2004) Ecology andmanagement of a Neotropical rainforest. Lessons drawn

Plant Soil (2009) 319:1–13 11

from Paracou, a long- term experimental research site inFrench Guiana. Elsevier, Paris, p 350

Granier A, Ceschia E, Damesin C, Dufrene E, Epron D, Gross P,Lebaube S, Le Dantec V, Le Goff N, Lemoine D, Lucot E,Ottorini JM, Pontailler JY, Saugier B (2000) The carbonbalance of a young Beech forest. Funct Ecol 14:312–325doi:10.1046/j.1365-2435.2000.00434.x

Högberg P, Nordgren A, Buchmann N, Taylor A, Ekblad A,Högberg M, Nyberg G, Ottoson-Löfvenius M, Read D(2001) Large-scale forest gridling shows that currentphotosynthesis drives soil respiration. Nature 411:789–792 doi:10.1038/35081058

Janssens IA, Lankreijer H, Matteucci G, Kowalski AS, Buch-mann N, Epron D, Pilegaard K, Kutsch W, Longdoz B,Grunwald T, Montagnani L, Dore S, Rebmann C, MoorsEJ, Grelle A, Rannik U, Morgenstern K, Oltchev S,Clement R, Gudmundsson J, Minerbi S, Berbigier P, IbromA, Moncrieff J, Aubinet M, Bernhofer C, Jensen NO,Vesala T, Granier A, Schulze ED, Lindroth A, Dolman AJ,Jarvis PG, Ceulemans R, Valentini R (2001) Productivityovershadows temperature in determining soil and ecosystemrespiration across European forests. Glob Change Biol7:269–278 doi:10.1046/j.1365-2486.2001.00412.x

Keeling CD (1958) The concentration and isotopic abundanceof atmospheric carbon dioxide in rural areas. GeochimCosmochim Acta 13:322–334 doi:10.1016/0016-7037(58)90033-4

Kuzyakov Y (2006) Sources of CO2 efflux from soil andreview of partitioning methods. Soil Biol Biochem38:425–448 doi:10.1016/j.soilbio.2005.08.020

Kuzyakov Y, Ehrensberger H, Stahr K (2001) Carbon partition-ing and below-ground translocation by Lolium perenne.Soil Biol Biochem 33:61–74 doi:10.1016/S0038-0717(00)00115-2

Lalonde RG, Prescott CE (2007) Partitioning heterotrophic andrhizospheric soil respiration in a mature Douglas-fir(Pseudotsuga menziesii) forest. Can J Res 37:1287–1297doi:10.1139/X07-019

Le Dantec V, Epron D, Dufrêne E (1999) Soil CO2 efflux in abeech forest: comparison of two closed dynamic systems.Plant Soil 214:125–132 doi:10.1023/A:1004737909168

Lee MS, Nakane K, Nakatsubo T, Koizumi H (2003) Seasonalchanges in the contribution of root respiration to total soilrespiration in a cool-temperate deciduous forest. Plant Soil255:311–318 doi:10.1023/A:1026192607512

Li YQ, Xu M, Zou XM (2006) Heterotrophic soil respiration inrelation to environmental factors and microbial biomass intwo wet tropical forests. Plant Soil 281:193–201doi:10.1007/s11104-005-4249-1

Longdoz B, Yernaux M, Aubinet M (2000) Soil CO2 effluxmeasurements in a mixed forest: impact of chamberdisturbances, spatial variability and seasonal evolution.Glob Change Biol 6:907–917 doi:10.1046/j.1365-2486.2000.00369.x

Malhi Y, Baldocchi DD, Jarvis PG (1999) The carbon balance oftropical, temperate and boreal forests. Plant Cell Environ22:715–740 doi:10.1046/j.1365-3040.1999.00453.x

Maunoury F, Berveiller D, Lelarge C, Pontailler JY, VanbostalL, Damesin C (2007) Seasonal, daily and diurnalvariations in the stable carbon isotope composition ofcarbon dioxide respired by tree trunks in a deciduous oak

forest. Oecologia 151:268–279 doi:10.1007/s00442-006-0592-z

Miller JB, Tans PP (2003) Calculating isotopic fractionationfrom atmospheric measurements at various scales. TellusB Chem Phys Meterol 55:207–214 doi:10.1034/j.1600-0889.2003.00020.x

Mortazavi B, Chanton JP, Prater JL, Oishi AC, Oren R, Katul G(2005) Temporal variability in 13C of respired CO2 in apine and a hardwood forest subject to similar climaticconditions. Oecologia 142:57–69 doi:10.1007/s00442-004-1692-2

Murtaugh PA (2007) Simplicity and complexity in ecologicaldata analysis. Ecology 88:56–62 doi:10.1890/0012-9658(2007)88[56:SACIED]2.0.CO;2

Ngao J, Epron D, Brechet C, Granier A (2005) Estimating thecontribution of leaf litter decomposition to soil CO2 effluxin a beech forest using 13C-depleted litter. Glob ChangeBiol 11:1768–1776 doi:10.1111/j.1365-2486.2004.01014.x

Ngao J, Longdoz B, Granier A, Epron D (2007) Estimation ofautotrophic and heterotrophic components of soil respira-tion by trenching is sensitive to corrections for rootdecomposition and changes in soil water content. PlantSoil 301:99–110 doi:10.1007/s11104-007-9425-z

Pataki DE, Ehleringer JR, Flanagan LB, Yakir D, Bowling DR,Still CJ, Buchmann N, Kaplan JO, Berry JA (2003) Theapplication and interpretation of keeling plots in terrestrialcarbon cycle research. Global Biogeochem Cycles17:1022–1037 doi:10.1029/2001GB001850

Phillips DL, Gregg JW (2001) Uncertainty in source partition-ing using stable isotopes. Oecologia 127:171–179doi:10.1007/s004420000578

Phillips DL, Gregg JW (2003) Source partitioning using stableisotopes: coping with too many sources. Oecologia136:261–269 doi:10.1007/s00442-003-1218-3

Rochette P, Flanagan LB, Gregorich EG (1999) Separating soilrespiration into plant and soil components using analysesof the natural abundance of 13C. Soil Sci Soc Am J63:1207–1213

Saiz G, Green C, Butterbach-Bahl K, Kiese R, Avitabile V,Farrell EP (2006) Seasonal and spatial variability of soilrespiration in four Sitka spruce stands. Plant Soil 287:161–176 doi:10.1007/s11104-006-9052-0

Salimon CI, Davidson EA, Victoria RL, Melo AWF (2004)CO2 flux from soil in pastures and forests in southwesternAmazonia. Glob Change Biol 10:833–843 doi:10.1111/j.1529-8817.2003.00776.x

Sokal RR, Rohlf FJ (1995) Biometry: the principles andpractice of statistics in biological research. W. H. Freemanand Co., New York, p 887

Steinmann KTW, Siegwolf R, Saurer M, Korner C (2004)Carbon fluxes to the soil in a mature temperate forestassessed by 13C isotope tracing. Oecologia 141:489–501doi:10.1007/s00442-004-1674-4

Subke J-A, Hahn V, Battipaglia G, Linder S, Buchmann N,Cotrufo MF (2004) Feedback interactions between needlelitter decomposition and rhizosphere activity. Oecologia139:551–559 doi:10.1007/s00442-004-1540-4

Subke JA, Inglima I, Cotrufo MF (2006) Trends and method-ological impacts in soil CO2 efflux partitioning: ametaanalytical review. Glob Change Biol 12:921–943doi:10.1111/j.1365-2486.2006.01117.x

12 Plant Soil (2009) 319:1–13

TuK, Dawson T (2005) Partitioning Ecosystem Respiration UsingStable Isotope Analysis of CO2. In: Flanagan LB, EhleringerJR, Pataki DE, Mooney M (eds) Stable isotopes andbiosphere–atmosphere interactions: Processes and biologicalcontrols. Elsevier Academic, San Diego, pp 125–135

Yim MH, Joo SJ, Shutou K, Nakane K (2003) Spatialvariability of soil respiration in a larch plantation:

estimation of the number of sampling points required.For Ecol Manage 175:585–588 doi:10.1016/S0378-1127(02)00222-0

Zobitz JM, Keener JP, Schnyder H, BowlingDR (2006) Sensitivityanalysis and quantification of uncertainty for isotopic mixingrelationships in carbon cycle research. Agric For Meteorol136:56–75 doi:10.1016/j.agrformet.2006.01.003

Plant Soil (2009) 319:1–13 13