Upload
clement-peltre
View
216
Download
1
Embed Size (px)
Citation preview
at SciVerse ScienceDirect
Soil Biology & Biochemistry 52 (2012) 49e60
Contents lists available
Soil Biology & Biochemistry
journal homepage: www.elsevier .com/locate/soi lb io
RothC simulation of carbon accumulation in soil after repeated applicationof widely different organic amendments
Clément Peltre a, Bent T. Christensen b, Sophie Dragon c, Christian Icard c, Thomas Kätterer d,Sabine Houot a,*a INRA, UMR 1091 Environment and Arable Crops, INRA-AgroParisTech, F-78850 Thiverval-Grignon, FrancebDepartment of Agroecology, Aarhus University, AU Foulum, P.O.Box 50, DK-8830 Tjele, DenmarkcCtifl/SERAIL Experimental Station, 123 chemin du Finday, F-69126 Brindas, FrancedDepartment of Soil and Environment, Swedish University of Agricultural Sciences, P.O. Box 7014, 750 07 Uppsala, Sweden
a r t i c l e i n f o
Article history:Received 1 February 2012Received in revised form14 March 2012Accepted 19 March 2012Available online 21 April 2012
Keywords:Animal manureOrganic amendmentsCompostsC storageAnalytical characterizationRothC modelLong-term experiment
* Corresponding author. Tel.: þ33 1 30 81 54 01; faE-mail addresses: [email protected] (C.
agrsci.dk (B.T. Christensen), [email protected] (slu.se (T. Kätterer), [email protected], hou
0038-0717/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.soilbio.2012.03.023
a b s t r a c t
Multi-compartment soil carbon (C) simulation models such as RothC are widely used for predictingchanges in C stocks of arable soils. However, rigorous routines for establishing entry pools that accountfor the diversity of exogenous organic matter (EOM) applied to croplands are still lacking. We obtaineddata on changes in soil C stocks after repeated applications of EOM from four long-term experiments(LTEs): Askov K2 (Denmark, 31 yrs), Qualiagro (France, 11 yrs), SERAIL (France, 14 yrs) and Ultuna(Sweden, 52 yrs). The adjustment of the partition coefficients of total organic C in EOM (EOM-TOC) intothe labile, resistant and humified entry pools of RothC (fDPM, fRPM, fHUM, respectively) provideda successful fit to the accumulation of EOM-derived C in the LTE soils. Equations estimating the EOMpartition coefficients in the RothC model were based on an indicator (IROC) of the EOM-TOC potentiallyretained in soil. IROC was derived from the C found in the soluble, lignin þ cutin-like and cellulose-likeVan Soest fractions and the proportion of EOM-TOC mineralized during 3 days of incubation. Usingthe EOM partition coefficients derived from these laboratory analyses resulted in RothC simulations withonly slightly larger errors than simulations based on partition coefficients fitted from LTE soil data, exceptfor EOMs that caused very large accumulations of C in soil (e.g. peat) possibly due to factors notaccounted for in the RothC model, such as change in soil pH. The proposed partitioning of EOM-TOCallows the potential soil C storage after EOM applications to be predicted regardless of field locationand specific composition of EOMs.
� 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Loss of soil organic C (SOC) may lead to reduced soil fertility andincreased soil erosion (Matson et al., 1997; Ciais et al., 2010). Theloss of SOC has been identified as a major threat towards the soilresource (European Commission, 2006), and small but consistentincreases in SOC could mitigate climate change effects by storingatmospheric CO2eC (Lal et al., 2007).
Annual application of exogenous organic matter (EOM) to culti-vated landmay lead to long continued accumulation of SOC (Marmoet al., 2004). We define EOM as crop residues, animal manures, andorganic wastes from urban areas, forestry and industry as these
x: þ33 1 30 81 53 96.Peltre), Bent.T.Christensen@C. Icard), Thomas.Katterer@[email protected] (S. Houot).
All rights reserved.
materials are subject to similar transformations in soil and to similarmanagement methods (Marmo et al., 2004). In Europe, recycling ofbiodegradable wastes is expected to increase in the future(European Commission, 2010). Different EOMs differ in theirpotential contribution to SOC, depending on their origin and degreeof transformation before being added to soil (Christensen andJohnston, 1997; Gerzabek et al., 1997; Bipfubusa et al., 2008). SinceC stocks change slowly, long-term field experiments are needed toevaluate the effects of repeated applications of EOM (IPCC, 1997).
Multi-compartment models of C turnover in soil (Jenkinson andRayner, 1977; Parton et al., 1987; Andren and Kätterer, 1997; Bruunet al., 2003) accurately simulate SOC dynamics in long-term fieldexperiments under different climatic conditions and soil types(Smith et al., 1997b). RothC is one of the most widely used modelthat simulates SOC dynamic based on relatively few parametersand input data. In RothC, total organic C in EOM (EOM-TOC) isdistributed into pools of labile (DPM), resistant (RPM) and humified
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e6050
(HUM) organic matter by partition coefficients (fDPM, fRPM and fHUM,respectively). The RothC model was initially calibrated with resultsfrom several Rothamsted long-term experiments leading to parti-tion coefficients for crop residues (fDPM ¼ 59% and fRPM ¼ 41% ofEOM-TOC) and farmyard manure (FYM; fDPM ¼ 49%, fRPM ¼ 49% andfHUM ¼ 2% of EOM-TOC) (Coleman and Jenkinson, 1999). Such par-titioning has frequently been used to simulate the effect of agri-cultural management (e.g. manure application) on changes in SOCstocks (Coleman et al., 1997; Yokozawa et al., 2010). In some studies,partition coefficients for manure are not specified (Falloon andSmith, 2002; Yang et al., 2003; Guo et al., 2007) or are modifiedwithout proper documentation (Kamoni et al., 2007). A range ofEOMs with contrasting chemical composition and decomposabilityhave been applied to soil (Thuriès et al., 2002; Lashermes et al.,2009) and specific partition coefficients have been used to simu-late changes in SOC under different farming systems (Leifeld et al.,2009). Establishing RothC partition coefficients based directly onanalytical characteristics of the applied EOMs would facilitate thesimulation of long-term changes in SOC after successive applica-tions of EOM.
In the CENTURY model, EOM-TOC is split into metabolic andstructural pools based on the lignin:N ratio (Parton et al., 1987),while in the STICS model, the behaviour of plant residues is basedon their C:N ratio (Nicolardot et al., 2001). However, theseapproaches have been shown to be inadequate to simulate changesin SOC in long-term experiments with manure application (Kamoniet al., 2007) and have not been tested for EOMs such as compostsand sludge. The stabilization of N in composts changes the bio-logical meaning of the C:N ratio (Paré et al., 1998). Moreover, thelignin content may not represent the recalcitrant fraction of com-posted EOM because lignin is partly degraded during compostingwhile other recalcitrant compounds are formed and recovered insoluble forms (Morvan and Nicolardot, 2009; Peltre et al., 2010).Thus, Hyvönen et al. (1996) argued that the lignin in EOM was nota valid indicator of EOM quality in a model of continuous qualitydistribution of soil organic matter.
Lashermes et al. (2009) developed an indicator of the fraction ofEOM-TOC that potentially is retained in soil in the long-term (IROC).The IROC indicator is calculated from analytical characteristics of theEOM and has been calibrated against long-continued C minerali-zation kinetics in laboratory incubations for a wide range of EOMsincluding plant materials, animal manures and various types offresh or composted wastes. We surmise that this indicator can beused to estimate the partition of EOM-TOC in the RothC model.
Our objective was to devise a method to establish the partitioncoefficients of EOMs in the RothC model based on analytical char-acteristics of EOMs. We first ran the model to reproduce theaccumulation of EOM-derived C in soil by simulating the additionalC in amended plots compared to un-amended reference plots infour long-term field experiments. We adjusted the partition coef-ficients of the RothC entry pools for contrasting EOMs applicationsto fit the changes in EOM-derived soil C. Then we developedequations to derive the partition coefficients of EOMs in RothCentry pools from their chemical composition. Finally, we tested thefeasibility of simulating the accumulation of EOM-derived C in soilwhen using the RothC partition coefficients derived from labora-tory analyses.
2. Materials and methods
2.1. Field experiments
Data was assembled from four differently sited long-term fieldexperiments (LTEs) where various types of EOMs had been appliedregularly. The LTEs differed in their soil types, climatic conditions
and management. The LTEs were the Askov K2 (Christensen andJohnston, 1997), Qualiagro (Houot et al., 2002), SERAIL (Berryet al., 2008) and Ultuna (Gerzabek et al., 1997). Table 1 summa-rizes selected characteristics of the LTEs, including average climaticdata, initial soil characteristics, crop rotation and EOM applied.
2.1.1. The Askov K2 experimentThe Askov K2 experiment was situated at Askov Experimental
Station, Denmark. The soil was retrieved from the 50e100 cm soildepth and adopted as topsoil in large concrete cylinders (diam.0.986 m; area 0.76 m2; depth 0.50 m). The depth of the soil layeramended with EOM (0e25 cm) was delineated by a coarse-meshednet to ensure a constant amount of soil into which the EOM wasincorporated and from which soil samples were taken. The exper-iment grew a four-course crop rotation that received annualapplications of 4 different EOMs (Table 1): physiologically maturecereal straw (STR-Ask), sawdust (SAW-Ask), dry FYM (FYM-Ask)and white sphagnum peat (PEA-Ask). Each treatment was in tworeplicates and reference plots without EOM amendment wereincluded. All plots received additional mineral N fertilization toensure similar crop production in all treatments. At harvest, allabove-ground plant parts were removed, leaving 4 cm of stubbles.Soils were sampled from 0 to 25 cm soil depth every four years andanalysed for C content. A soil bulk density of 1.5 Mg m�3 wasadopted (Bruun et al., 2003). Further details of the experiment aregiven by Christensen and Johnston (1997).
2.1.2. The Qualiagro experimentThe ongoing Qualiagro experiment, located in Feucherolles, near
Paris, France, was initiated in 1998. It is cropped with a maize-wheat rotation and includes 4 EOM treatments and a referencewith no EOM application. EOMs are applied every second year onwheat stubbles using 10 � 45 m2 plots separated by 6 m widestrips. The EOMs include a municipal solid waste compost obtainedby composting solid municipal wastes after removal of its contentof dry and clean packaging (MSW-Qua), an FYM (FYM-Qua),a compost derived from co-composting green wastes with sewagesludge (GWS-Qua), and a biowaste compost produced by co-composting green wastes with a source-separated organic frac-tion of municipal solid wastes (BIOW-Qua). Two mineral N fertil-ization regimes (optimum and minimum mineral N fertilization,referred as þN and eN, respectively) are applied on all organictreatments in two separate sections of the field experiment. Withineach N fertilization regime, the EOM treatments occur ina randomized block design with four blocks placed 25 m apart toprevent cross-contamination during EOM applications. Annual drymatter yields of grain and plant residues are determined manuallybefore harvest. In 2007, barley was grown due to prognoses ofa regional attack of Diabrotica virgifera to maize, causing EOMs to beapplied in two consecutive years (2006 and 2007). Soil is sampledlate August from the topsoil (0e29 cm) prior to EOM applicationand analysed for C content. Topsoil bulk density was measuredplot-wise in 1998 (start of the experiment), 2004 and 2009. Addi-tionally, bulk density and C content of the subsoil (29e35 cm) wasmeasured in September 2004.
2.1.3. The SERAIL experimentThe ongoing SERAIL experiment, located in Brindas, near Lyon,
France, was started in 1995. The soil is under a rotation of vegeta-bles (lettuce, turnip, spinach, leek, carrot, cabbage, Swiss chard,celery). Five different EOMs are applied annually: dried FYM (FMT-Ser), fresh FYM (FYM-Ser), enriched bark compost produced by co-composting bark with poultry manure, liquid manure and algae(ALG-Ser), compost produced by co-composting coffee cake (90%),sheep manure and wool waste (VGH-Ser), and a green waste
Table
1Principal
characteristicsof
thefieldex
perim
ents.
Location
Duration
(yea
rs)
Soil
type(FAO)Clay
content
(%)
Silt
content
(%)
InitialC
content
(%DM)
Mea
nan
nual
temperature
(�C)
Plou
ghed
laye
rdep
th(cm)
Mea
nan
nual
rainfall(m
m)Mea
nan
nual
poten
tial
evap
o-tran
spiration
(PET
)(m
m)
Crop
succession
EOMs
applie
dRateof
EOM
applic
ation
Freq
uen
cy/
periodof
applic
ation
Askov
K2
Askov
,Den
mark
(55�28
0 N,9
� 550
E)30
yrs
(195
6e19
86)Dystric
Arenosol
2.5
2.4
0.3
7.4�
0.7
0e25
972�
160
501�
56Sp
ringba
rley
,fibe
rflax
,wintercereals,
silage
maize
STR-A
sk6.5Mgha�
1of
dry
matter
Everyye
arin
Octob
erSA
W-A
skFY
M-A
skPE
A-A
sk
Qualiagro
Feuch
erolles,Fran
ce(48�52
0 N,1
� 570
E)11
yrs
(199
8e20
09)Lu
visol
15.2
78.7
1.0
11.0
�0.5
0e29
643�
128
761�
63W
hea
t,maize
MSW
-Qua
4MgCha�
1Ev
ery2ye
ars
inSe
ptembe
rGW
S-Qua
BIO
W-Q
uaFY
M-Q
ua
SERAIL
Brindas,F
rance
(45�43
0 N,4
� 420
E)14
yrs
(199
5e20
09)Lu
visol
16.8
17.2
1.2
12.9
�0.5
0e30
822�
118
970�
52Veg
etab
les
rotation
FMT-Se
rCin
30Mgha�
1
offreshFY
MStab
leCin
30Mgha�
1
offreshFY
M*
Everyye
arin
Aprilto
July
FYM-Ser
VGH-Ser
ALG
-Ser
GW
C-Ser
Ultuna
Nea
rUppsala,
Swed
en(59�
490
N,1
7�38
0 E)
52yrs
(195
6e20
07)Eu
tric
Cam
bisol
37.0
41.0
1.5
5.7�
1.0
0e20
541�
9358
5�
7062
%sp
ringcereals,
15%silage
maize
,33
%rapean
dother
crop
s
SAW
-Ult
4MgCha�
1
inOctob
erIn
1956
,196
0,19
63an
dev
ery
seco
ndye
artherea
fter
FYM-U
ltSL
U-U
ltPE
A-U
lt
STR:straw
,SAW
:saw
dust,FYM:farmya
rdman
ure,PEA
:pea
t,MSW
:municipal
solid
waste
compost,GW
S:gree
nwaste
andsludge
compost,BIO
W:b
iowaste
compost,FM
T:deh
ydratedFY
M,V
GH:e
nrich
edco
ffee
cake
compost,
ALG
:enrich
edba
rkco
mpost,GW
C:g
reen
waste
compost,GM:g
reen
man
ure,SLU
:sew
agesludge
;presentedfortheAskov
K2(A
sk),Qualiagro(Q
ua),SER
AIL
(Ser)a
ndUltun
a(U
lt)e
xperim
ents.*Amou
nto
fstableCin
freshFY
Mob
tained
usingthebiolog
ical
stab
ility
index
(BSI)of
FYM
(Linères
andDjako
vitch,1
993).
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e60 51
compost (GWC-Ser). The experiment has a randomized blockdesign with 3 blocks. Each block includes all organic treatments: 5EOMs � 2 application rates (equivalent to total organic C content inFYM, EqC treatments or to humified organic C in FYM, EqH treat-ments) and a reference plot without EOM amendment, using 11 �1.4 m2 plots. Soil is sampled each year from the topsoil (0e30 cm)before EOM application and analysed for C content. A bulk densityof 1.40 Mg m�3 has been used throughout the experiment, corre-sponding to the average bulk density measured in 2009 in thetopsoil prior to EOM application. The density was assumed to haveremained constant since no differences in bulk density have beenfound between treatments.
2.1.4. The Ultuna experimentThe ongoing Ultuna experiment, located near Uppsala, Sweden,
was started in 1956. Six different EOMs are applied: straw (STR-Ult), green manure (GM-Ult), sawdust (SAW-Ult), FYM (FYM-Ult),anaerobically digested sewage sludge (SLU-Ult) and whitesphagnum peat (PEA-Ult). The different treatments are laid out ina randomized block design in 2�2 m2 plots (four replicates) sepa-rated by wooden frames. For the straw, sawdust and peat treat-ments, plots without and with mineral N fertilization (referred aseN and þN, respectively) were included. Plots receiving greenmanure, FYM and sewage sludge received no mineral N fertilizer.An additional treatment was FYM enriched in P (referred asFYM þ P eN). The topsoil was sampled for analysis of the C contentbefore the EOM applications in 1956, 1967, 1974, 1975, 1977, 1979,1983, every second year from 1985 to 2001, and in 2005 and 2007.The bulk density of the topsoil was measured in 1956, 1975, 1991and 1997. The C content of the subsoil (20e25 cm) was measuredunder each treatment in 1991. A detailed description of theexperiment is given by Kätterer et al. (2011).
2.2. C stocks accumulated in soil following EOM application
Soil C concentrations were converted into C stocks (inMg C ha�1) in the topsoil using soil bulk densities. The C stockswere based on equivalent soil masses in all treatments andthroughout the experiments (Ellert and Bettany, 1995; Lee et al.,2009). Detailed calculations are given in Annex No. 1 of supple-mentary material. The depth of the considered soil layer to reachequivalent soil mass was adjusted to account for changes in bulkdensity in the Qualiagro and Ultuna experiment, whereas constanttopsoil layer depth was considered for the Askov K2 and SERAILexperiments as the bulk density remained constant throughoutthese experiments. The amount of soil C derived from EOM (DC)was calculated as the difference between the C stock in amendedand un-amended reference plots.
2.3. The RothC model
The RothC 26.3 model simulates soil C dynamics by consideringfive organic C pools: a labile pool (DPM: “decomposable plantmaterial”; mean residence time of 1.2 months), a resistant pool(RPM: “resistant plant material”; mean residence time of 3.3 yrs),a humified pool (HUM; mean residence time of 50 yrs), a microbialbiomass pool (BIO; mean residence time 1.5 yrs) and a pool of inertorganic matter that is not degraded (IOM) (Coleman and Jenkinson,1999). EOM-TOC added to the soil is split into the DPM, RPM andHUM pools according to the partition coefficients fDPM, fRPM andfHUM¼ 1� fDPM� fRPM. Using amonthly time step, a fraction of eachpool is turned over according to specific decay rate constants and iseithermineralized into CO2, transferred into the humified (HUM) orascribed to microbial biomass pools (BIO). The proportion that isconverted into CO2 and BIO þ HUM depends on the clay content of
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e6052
soil. Rate modifying factors adjust the decay rate constants toaccount for soil humidity (topsoil moisture deficit), air temperatureand soil cover (covered or not covered).
The model was run on DC (the difference in C stock betweenamended and reference treatments) meaning that only the Caccumulation due to EOM application was simulated. Conse-quently, the initial size of the BIO, HUM and IOM pools were set tozero. This is made possible by the linearity of the model inwhich allpools decompose according to first order kinetics: dCi/dt ¼ �kiCi,where Ci is the amount of C in a given pool at time t and ki is thecorresponding decay rate constant which is independent of EOMquality or pool size. Consequently the kinetic of C accumulation insoil due to EOM corresponds to the difference in C kinetic betweensoils with and without EOM applications. Thus any possiblepriming of EOM applications on native SOC turnover is excluded.
The entry data for RothC included the clay content of soils, soilcover (covered or not covered), C inputs from the EOM, and addi-tional C inputs from crop residues (stem bases, roots and rootsexudates) in amended plotswhen larger crop yields weremeasuredin amended than in reference plots. All additional C inputs fromcrop residues were assumed to enter the soil at ploughing. Monthlyclimatic data were air temperatures, cumulated precipitation andopen-pan evaporation. Potential evapo-transpiration was calcu-lated from daily climatic data recorded at meteorological stationsnear the experimental sites (solar radiation, air temperature, windvelocity and relative humidity) according to the Penman equation(Penman, 1948) and converted into open-pan evaporation afterdivision by 0.75 (Coleman and Jenkinson, 1999).
The EOM partition coefficients for the entry pools DPM, RPMand HUM (fDPM, fRPM, fHUM ¼ 100% � fDPM � fRPM, respectively,expressed as % of EOM-TOC) were adjusted to fit the kinetic ofmeasured DC accumulation in the four LTEs. The fDPM and fRPMfractions were confined between 0 and 100% of EOM-TOC, and fHUMwas set �20% of EOM-TOC. The constraint on fHUM eliminatedunrealistic estimates of the coefficients, e.g. EOM being allocatedhigh fDPM and fHUM coefficients and a fRPM coefficient equal to zero.Such a distribution of EOM-TOC would not be biologically mean-ingful. The remaining model parameters were kept as in the orig-inal publication (Coleman and Jenkinson, 1999). For each EOMtreatment, the model fitting was performed simultaneously onplots with, without or with minimum N fertilization (Ultuna andQualiagro experiments) and on plots with EqC and EqH rates (theSERAIL experiment) by minimizing the residual sum of squaresbetween measured and simulated values (using Excel solver withthe Newton method). A number of statistical indicators were usedto evaluate the goodness of fit (Smith et al., 1997b). The root meansquare error (RMSE, in Mg C ha�1) was used to evaluate thedifference between measured and predicted values. The coefficientof variation of the RMSE (CV(RMSE), %) was calculated as RMSE/y*100 (y is the mean of measured values) in order to evaluate therelative error between measured and predicted values. TheSpearman coefficient of correlation (r) was used to determinewhether the simulated kinetics displayed the same pattern as themeasured kinetics. The efficiency of the model was calculatedusing:
EF ¼ 1�Pn
i¼1ðyi � yiÞ2Pn
i¼1ðyi � yÞ2
where yi and yi are the measured and predicted values for thesample i, y is the mean of measured values and n is the number ofsamples. Model efficiency expresses the part of the variance inmeasured data that can be explained by the model. A positive EFvalue indicates that the simulated values describe the trend in themeasured data better than the mean of measured values.
2.4. Additional C inputs from crop residues
We considered all additional crop C input to the EOM plots(compared to reference plots) as input in the RothC model. Wetherefore accounted for any difference between above andbelow ground crop C returned in amended and reference plots(DCres).
In the Qualiagro, Ultuna and SERAIL experiments, crop yieldsdiffered between amended and reference plots. These differenceswere greater in plots with minimum or no additional N fertilization(Ultuna and Qualiagro). In the the Askov K2 experiment, nodifferences in C inputs from crop residues were expected becauseall plots received adequate mineral fertilizers in order to achievethe similar crop production in all treatments.
In the Qualiagro experiment, all wheat crop residues wereremoved except for 17 cm of stubbles, which accounted for 30% ofthe measured dry matter yield of above-ground residues (YS, inMg ha�1). Maize residues were all returned to the soil. In the Ultunaexperiment, all plant parts were removed except for stem bases.The C amounts derived from roots, root exudates and stems weredetermined based on the study of Kätterer et al. (1993), Bolinderet al. (2007), Jackson et al. (1996) and Machet et al. (2007) andare detailed in Annex 2 of supplementary materials.
2.5. C accumulation in soil under standard pedoclimatic conditions
The rate of C accumulation in soil tends to decrease withincreasing number of applications and duration of experiment (Sixet al., 2002; Chung et al., 2008; Thomsen and Christensen, 2010;Powlson et al., 2012). To compare the potential C accumulation insoil from the various EOMs, RothC simulations were run understandardized pedoclimatic conditions over a period of 20 years assuggested by the IPCC guidelines for greenhouse gas inventories(1997). In all runs, 2 Mg C ha�1 of EOMwas applied every two yearsin September, reaching a total of 20 Mg C ha�1 applied over 20years. The soil was plant-covered throughout the year except inSeptember. No additional C input from crop residues (compared toreference plots) was assumed, that is similar crop yield in alltreatments was considered. The soil was with 15% clay, and thepreviously fitted EOM partition coefficients for the RothC entrypools (fDPM, fRPM, fHUM) were applied.
The increase in soil C (as % of C inputs) was calculated using thedifference between C stocks in amended and reference plots(DC20y) simulated using RothC according to:
Y20y ¼ DC20y=CEOM�20y*100 (1)
where CEOM-20y are the cumulated C inputs from EOMs after 20years of application.
Three contrasting climate scenarios were chosen for the simu-lation: a Mediterranean climate (San-Giuliano, Corsica, France;average annual temperature: 16.0 �C, mean annual precipitation:822 mm), a medium temperate climate (Feucherolles, Île-de-France, France; average annual temperature: 11.0 �C, mean annualprecipitation: 643 mm), and a Nordic climate (Ultuna, Sweden;average annual temperature: 5.7 �C, mean annual precipitation:541 mm).
2.6. Laboratory characterization of EOMs
EOMs were fractionated into soluble, hemicellulose-like, cellu-lose-like and lignin þ cutin-like fractions (SOL, HEM, CEL and LIC,respectively) using the Van Soest method (Van Soest, 1963; VanSoest and Wine, 1967) as modified in the French XP U 44-162standard (AFNOR, 2009a). The C mineralization of EOMs was
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e60 53
measured after 3 days of incubation of soil-EOMmixtures in sealedjars at 28 �C (C3d, expressed as a % of TOC) and 75% of the soil waterholding capacity. The CeCO2 was trapped in 10 mL of 0.5 M NaOHand determined by colorimetry (AFNOR, 2009b).
Total organic C (TOC) was determined by dry combustion usingan elemental analyser after removal of carbonates with hydro-chloric acid (AFNOR, 1995). The IROC indicator reflects the propor-tion of EOM that may be retained in the soil for decades (Lashermeset al., 2009) and is based the SOL, CEL and LIC fractions of Van Soestfractionation and the proportion of EOM-TOC mineralized after 3days of laboratory incubation (C3d):
IROC ¼ 44:5þ 0:5 SOL � 0:2 CEL þ 0:7 LIC� 2:3 C3d (2)
The SOL, CEL and LIC and C3d are expressed as % of TOM and TOC,respectively.
For the Qualiagro experiment, all EOMs applied between 1998and 2007 (6 applications) were sampled and characterized, and thevalues presented are averages of six samples. For the SERAILexperiment, we used average values for EOMs applied in 2005 and2009 for which values for short-term C mineralization were avail-able. For the fresh and dehydrated FYM treatments (FYM-Ser andFMT-Ser), only 2005 data were used because EOMs applied to FYM-Ser and FMT-Ser in 2009 were not representative for the FYMapplied throughout the experiment. For the Ultuna experiment,three pooled samples representing EOMs applied in 1975 þ 1979,1991 þ1993 þ 1995 and 2005 þ 2007 were used in order to obtainsufficient material for laboratory analyses, assuming sufficienthomogeneity among EOMs applied over short-term periods. Thedata for the Ultuna experiment were thus average values of threesamples. The EOMs applied in the Askov K2 experiment have onlybeen analyzed for their total-C and total-N contents (Christensenand Johnston, 1997). For sawdust and peat, we assigned valuesmeasured for sawdust and peat in the Ultuna experiment as similartypes of materials were applied in both experiments. For FYM, weused an average value of all the FYMs applied in the Qualiagro,SERAIL and Ultuna experiments, while for straw, we used theresults from Djakovitch (1988) because the straw applied in AskovK2 was physiologically mature and probably less degradable thanthe straw applied in the Ultuna experiment.
Table 2Analytical characteristics of the EOMs applied during the Askov K2 (Ask), Qualiagro (Qua)dry matter at 105 �C (DM), soluble, hemicellulose-, cellulose-, and ligninþcutin- like fracexpressed as % of total organic matter (% TOM), proportions of EOM-TOC mineralized afteorganic C in soil (IROC) determined using Equation (2) (see part 2.6). Mean values � stan
TOC% DM
SOL% TOM
HEM% TOM
STR-Ask 44.6 14.1 31.2SAW-Ask 49.9 � 0.1 8.2 � 0.3 11.7 � 2.3FYM-Ask 33.8 � 5.9 34.3 � 8.3 16.8 � 7.2PEA-Ask 47.7 � 1.5 28.3 � 4.4 11.2 � 4.1MSW-Qua 31.9 � 4.7 42.5 � 8.7 6.4 � 2.4FYM-Qua 32.1 � 5.3 39.1 � 5.9 12.0 � 2.9GWS-Qua 27.1 � 6.2 45.7 � 9.4 5.1 � 2.1BIOW-Qua 18.6 � 2.4 44.4 � 2.7 4.2 � 2.8FMT-Ser 38.5 27.6 27.4FYM-Ser 37.2 32.0 27.2VGH-Ser 34.3 32.6 � 7.8 3.4 � 7.1ALG-Ser 15.7 45.0 � 6.7 5.5 � 2.2GWC-Ser 22.2 48.0 � 4.4 3.5 � 1.7STR-Ult 45.6 � 0.0 11.8 � 2.1 31.2 � 7.1GM-Ult 45.6 � 1.4 26.7 � 6.4 30.9 � 4.2SAW-Ult 49.9 � 0.1 8.2 � 0.3 11.7 � 2.3FYM-Ult 35.9 � 8.2 25.6 � 6.0 23.1 � 5.6SLU-Ult 26.2 � 6.9 46.6 � 11.2 17.3 � 6.7PEA-Ult 47.7 � 1.5 28.3 � 4.4 11.2 � 4.1
STR: straw, SAW: sawdust, FYM: farmyard manure, PEA: peat, MSW: municipal solid wFMT: dehydrated FYM, VGH: enriched coffee cake compost, ALG: enriched bark compos
2.7. Independent prediction of the partition coefficients of EOM-TOCin RothC
Linear regressions to predict the partition coefficients previ-ously fitted with field data were calculated using the IROC indicatoras an independent variable. Leave-one-out cross-validations werecalculated by predicting each sample successively with theregression model calibrated with the rest of the samples. Regres-sions and cross validations were calculated using R (R DevelopmentCore Team 2010). The statistics used to evaluate the quality ofregression models included the root mean square error (RMSE)expressed in % of EOM-TOC, the coefficient of variation of the RMSE,CV(RMSE) in % and the coefficient of determination R2.
3. Results and discussion
3.1. Characteristics of the EOMs applied
Table 2 shows the characteristics of all EOMs applied during thelong-term field experiments. The EOMs covered a wide range ofbiochemical compositions. Non-processed EOMs (i.e. plant mate-rials such as straw, green manure and sawdust) were generallyhigher in TOC (>40% of DM) and cellulose (>35% of EOM-TOM) andhad a smaller soluble fraction (<30% of EOM-TOM) compared tocomposted EOMs as previously observed (Lashermes et al., 2009).FYMs showed intermediate values reflecting their content of bothbeddingmaterials and themodification of organic matter occurringduring animal digestion and manure storage. Peat had a biochem-ical composition and TOC content close to that of plant materialsbut was decomposed more slowly during incubation (C3d ¼ 0.7% ofTOC). High acidity in the peat material is one possible explanationfor these observed differences in short-term decomposition (Leifeldet al., 2008). Although not measured in our experiment, pH isusually low in untreated Sphagnum peat. The relatively low pHrecorded in the Peat-treatment in the Ultuna experiment is con-firming the acidifying properties of the peat material (Kättereret al., 2011). Non-processed plant materials displayed contrastingC3d values, ranging from low value for sawdust (2.6%) to high forgreen manure (18.8%). Composts had low C3d values, except for the
, SERAIL (Ser) and Ultuna (Ult) experiments: total organic C (TOC), expressed as a % oftions (SOL, HEM, CEL, LIC) as determined using the Van Soest fractionation methodr 3 days of laboratory incubation in soil at 28 �C (C3d) and the indicator of remainingdard deviation.
CEL% TOM
LIC% TOM
C3d
% TOCIROC% TOC
46.9 8.0 5.5 35.256.1 � 3.0 24.1 � 3.6 2.6 � 2.3 48.3 � 3.728.5 � 7.5 20.4 � 5.2 3.9 � 1.4 61.4 � 11.238.1 � 6.2 22.5 � 6.3 0.7 � 0.2 65.2 � 2.935.8 � 10.3 15.2 � 2.4 10.6 � 2.9 45.0 � 12.825.5 � 7.7 23.4 � 2.5 2.8 � 0.3 68.9 � 6.020.8 � 10.1 28.5 � 11.9 1.8 � 0.9 78.9 � 10.520.3 � 2.3 31.1 � 8.9 2.7 � 1.3 78.3 � 5.829.0 16.0 6.3 49.329.2 11.5 5.9 49.322.9 � 4.5 41.1 � 5.4 6.8 � 2.4 69.4 � 4.518.5 � 10.9 31.0 � 2.0 1.0 � 0.3 82.7 � 6.310.1 � 0.5 38.3 � 5.6 0.5 � 0.4 92.2 � 2.549.6 � 6.7 7.6 � 2.5 8.0 � 1.1 27.4 � 2.036.8 � 2.8 5.6 � 0.6 18.8 � 0.7 11.0 � 4.856.1 � 3.0 24.1 � 3.6 2.6 � 2.3 48.3 � 3.734.1 � 5.6 17.2 � 5.2 5.3 � 0.4 50.3 � 7.216.6 � 4.0 19.5 � 1.7 7.4 � 4.5 61.2 � 15.638.1 � 6.2 22.5 � 6.3 0.7 � 0.2 65.2 � 2.9
aste compost, GWS: green waste and sludge compost, BIOW: biowaste compost,t, GWC: green waste compost, GM: green manure, SLU: sewage sludge.
Table 3Fitting of EOM partition coefficients in the RothC labile, resistant and humified entrypools (fDPM, fRPM, fHUM, expressed as a % of EOM total organic C: TOC) andmeasurements of best fit between simulated andmeasured C stock dynamics. RMSE:root mean square error between measured and simulated accumulated C values,CV(RMSE): coefficient of variation of the RMSE, r: coefficient of correlation, EF modelefficiency. Results for the Askov K2 (Ask), Qualiagro (Qua), SERAIL (Ser) and Ultuna(Ult) experiments.
Treatment fDPM% TOC
fRPM% TOC
fHUM% TOC
RMSEMg C ha�1
CV(RMSE)%
r EF
STR-Ask 58.5 38.4 3.1 1.6 13.4 0.94 0.89SAW-Ask 56.9 31.5 11.6 2.8 17.5 0.89 0.77FYM-Ask 37.7 42.3 20.0 1.3 8.6 0.98 0.96PEA-Ask 0.0 80.0 20.0 7.4 24.0 0.95 0.76MSW-Qua 62.3 37.7 0.0 0.7 13.9 0.97 0.92FYM-Qua 30.1 49.9 20.0 1.2 13.3 0.98 0.93GWS-Qua 15.3 64.7 20.0 1.3 13.3 0.98 0.95BIOW-Qua 0.0 80.0 20.0 1.6 17.0 0.97 0.92FMT-Ser 71.2 28.8 0.0 2.0 36.2 0.74 0.55FYM-Ser 70.1 29.9 0.0 3.4 65.8 0.62 0.38VGH-Ser 39.2 60.8 0.0 3.9 61.5 0.33 �0.16ALG-Ser 8.9 82.4 8.7 3.8 41.2 0.65 0.39GWC-Ser 0.0 80.0 20.0 5.14 43.6 0.81 0.51STR-Ult 88.2 11.8 0.0 2.3 24.2 0.81 0.14GM-Ult 90.3 9.7 0.0 2.5 22.8 0.84 0.07SAW-Ult 53.2 44.5 2.3 1.7 12.3 0.95 0.84FYM-Ult 49.2 40.6 10.2 3.3 17.2 0.91 0.59SLU-Ult 0.0 85.3 14.7 6.7 22.4 0.82 0.11PEA-Ult 0.0 80.0 20.0 9.2 26.4 0.92 0.31
STR: straw, SAW: sawdust, FYM: farmyard manure, PEA: peat, MSW: municipalsolid waste compost, GWS: green waste and sludge compost, BIOW: biowastecompost, FMT: dehydrated FYM, VGH: enriched coffee cake compost, ALG: enrichedbark compost, GWC: green waste compost, GM: green manure, SLU: sewage sludge.
Time (years)
Accu
mul
ated
C (M
g C
ha-
1 )
1960 1970 19800
10
20
30
40
50
600
10
20
30
40
50
60
1960 1970 1980 1990
STR-Ask SAW-Ask
CV(RMSE) = 25.4%
Δ CV(RMSE) = 12.0%
EF = 0.61
FYM-Ask
CV(RMSE) = 21.0%
Δ CV(RMSE) = 3.5%
EF = 0.68
CV(RMSE) = 22.2%
Δ CV(RMSE) = 13.6%
EF = 0.71
PEA-Ask
Fig. 1. Accumulation of soil C compared to reference plots after EOM applicationsduring the Askov K2 experiment. Measured values (-, errors bars ¼ standard devia-tion) and values simulated with RothC using partition coefficients fitted with field data(black lines) or predicted using the IROC indicator (grey dashed lines). STR: straw, SAW:sawdust, FYM: farmyard manure, PEA: peat. Coefficient of variation of the RMSE(CV(RMSE)) and model efficiency (EF) of the simulations using the predicted partitioncoefficients, D CV(RMSE): difference between the CV(RMSE) values obtained using thepredicted and fitted partition coefficients.
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e6054
ALG andMSW composts. The IROC indicator calculated from the SOL,CEL, LIC and C3d values using Equation (2) provides an estimate ofthe proportion of EOM-TOC that is potentially retained in the soilfor decades after application. IROC values ranged from 11.0% forgreen manure to 92.2% for the GWC compost. The highest IROCvalues were found for composts, except for the MSW compostwhich had a medium IROC value (45.0%). The FYMs had variable IROCvalues, ranging from 49.3% for the FYM applied in the SERAILexperiment to 68.9% for the FYM applied in the Qualiagro experi-ment. This is ascribed to the FYM being matured during storagebefore application. The IROC value of the sewage sludge applied inthe Ultuna experiment fell within the range of values reported byLashermes et al. (2009) for sewage sludge. The large standarddeviation of IROC (61.2� 15.6%) indicated a marked variability in thecomposition of the sludge applied in the Ultuna experiment, causedmainly by substantial differences in degradability during short-term incubations (C3d ¼ 7.4 � 4.5%).
3.2. Parameterization of the partition coefficients of EOMs in RothCthrough fitting of DC accumulation in soil from field experimentdata
The C stocks in the LTEs were calculated for equivalent soil massand the differences between C stocks in amended and referenceplots were calculated (DC). The additional C inputs were calculatedas the difference between C inputs in amended and reference plotsand included C inputs from repeated EOM application and addi-tional C inputs from crop residues, roots and root exudates inamended plots. Detailed results of soil C stocks and C inputs to soilare presented in Table S1 of Annex 3 in supplementary material.The LTEs had widely different initial C stocks ranging from11.4Mg ha�1 in the Askov experiment to 43.2Mg ha�1 in the Ultunaexperiment. The C accumulated following EOM application alsovaried greatly between the different amendments and LTEs. The Cinputs calculated for crop residues, roots and roots exudates for theUltuna experiment were similar to those reported by Kätterer et al.(2011).
The partition of EOM-TOC into the DPM, RPM and HUM entrypools of RothC (fDPM, fRPM, fHUM ¼ 1� fDPM � fRPM) were fitted to themeasured DC accumulation in the four LTEs, the other parametervalues being kept as in the original RothC model. Generally, thefitting of the partition coefficients for each type of EOM reproducedthe DC accumulation in the LTEs satisfactorily (Table 3, Fig. 1 toFig. 4). Peat from the Askov and Ultuna experiments and greenwaste compost from the SERAIL experiment led to a large DCaccumulation that could not be accurately reproduced by themodel(Table 3, Figs. 1, 3 and 4). For peat application this may be ascribedto a decrease in soil pH, as previously discussed (see 3.1), leading togreater soil C accumulation that could not be accounted for by themodel. For all other treatments in the Ultuna and Askov experi-ments and for all treatments in the Qualiagro experiment, thesimulated values remained within, or very close to, the standarddeviation of the measured values (Figs. 1, 2 and 4). The same sets offitted coefficients enabled a satisfactory simulation of C accumu-lations in plots with both minimum and optimum N fertilization inthe Qualiagro and Ultuna experiments. The quality of the simula-tions was poorer for the SERAIL experiment when compared to theother LTEs (CV(RMSE) of 35.2e65.8%). This was ascribed to thegreater variability in measured C stocks during this experiment(Fig. 3). However, the simulations reproduced C accumulationtrends fairly accurately, as shown by the acceptable correlationcoefficients obtained, except for plots amended with VGH. Thistreatment displayed highly variable C stocks in the EqC plotswhereas the C accumulation in the EqH plots was quite wellreproduced. The coefficients fitted in the SERAIL experiment were
thus considered as a valuable reflection of the quality of the EOMsapplied during this experiment.
The DC accumulation in soil was well reproduced for the Ultunaexperiment. However, measured C stocks in the Ultuna experimentseemed to reach steady state conditions between 1995 and 2000, at
0
5
10
15
20
25
0
5
10
15
20
0
5
10
15
20
2000 2004 20080
5
10
15
20
2000 2004 2008
CV(RMSE) = 17.7%
Δ CV(RMSE) = 3.8%
EF = 0.87
MSW-Qua -N MSW-Qua +N
FYM-Qua -N FYM-Qua +N
GWS-Qua -N GWS-Qua +N
BIOW-Qua -N BIOW-Qua +N
CV(RMSE) = 15.5%
Δ CV(RMSE) = 2.1%
EF = 0.91
CV(RMSE) = 15.3%
Δ CV(RMSE) = 2.0%
EF = 0.93
CV(RMSE) = 28.6%
Δ CV(RMSE) = 11.6%
EF = 0.76
Time (years)
Accu
mul
ated
C (M
g C
ha-
1 )
Fig. 2. Accumulation of soil C compared to reference plots after EOM applicationsduring the Qualiagro experiment. Measured values (-, errors bars ¼ standard devi-ation) and values simulated with RothC using partition coefficients fitted with fielddata (black lines) or predicted using the IROC indicator (grey dashed lines). MSW:municipal solid waste compost, FYM: farmyard manure, GWS: green waste and sludgecompost, BIOW: biowaste compost. eN: plots with minimum mineral Nfertilization, þN: plots with optimum mineral N fertilization. Coefficient of variation ofthe RMSE (CV(RMSE)) and model efficiency (EF) of the simulations using the predictedpartition coefficients calculated for the þN and eN parts of the experiment takentogether, D CV(RMSE): difference between the CV(RMSE) values obtained using thepredicted and fitted partition coefficients.
Accu
mul
ated
C (M
g C
ha-
1 )
Time (years)
0
5
10
15
20
25
30
0
5
10
15
20
25
0
5
10
15
20
25
0
5
10
15
20
25
1996 2000 2004 20080
5
10
15
20
25
1996 2000 2004 2008
CV(RMSE) = 55.3%
Δ CV(RMSE) = 19.1%
EF = -0.06
FMT-Ser EqC FMT-Ser EqH
FYM-Ser EqC FYM-Ser EqH
VGH-Ser EqC VGH-Ser EqH
ALG-Ser EqC ALG-Ser EqH
CV(RMSE) = 75.1%
Δ CV(RMSE) = 9.4%
EF = 0.19
CV(RMSE) = 73.3%
Δ CV(RMSE) = 11.8%
EF = -0.64
CV(RMSE) = 42.1%
Δ CV(RMSE) = 0.9%
EF = 0.37
GWC-Ser EqC GWC-Ser EqH
Fig. 3. Accumulation of soil C compared to reference plots after EOM applicationsduring the Serail experiment. Measured values (-) and values simulated with RothCusing partition coefficients fitted with field data (black lines) or predicted using theIROC indicator (grey dashed lines). FMT: dehydrated farmyard manure, FYM: farmyardmanure, VGH: enriched coffee cake compost, ALG: enriched bark compost, GWC: greenwaste compost. EqC: “C equivalent” dose, EqH: “humus equivalent” dose (see 2.1.3).Coefficient of variation of the RMSE (CV(RMSE)) and model efficiency (EF) of simula-tions using the predicted partition coefficients calculated for the EqC and EqH parts ofthe experiment taken together, D CV(RMSE): difference between the CV(RMSE) valuesobtained using the predicted and fitted partition coefficients.
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e60 55
different level depending on the EOM applied. The model predictedthat steady state would be reached much later, except for thesawdust treatment. The differences between measured and simu-lated values at the end of the simulated period in the Ultunaexperiment gave low model efficiency values (EF, Table 3) whereasmeasured values were accurately simulated in the period prior topossible steady state (Fig. 4).
A broad range of partition coefficients was found for thedifferent EOMs. We investigated the correlations between theanalytical characteristics of EOMs and the fitted partition coeffi-cients of RothC (Table 4). The peats and GWC-Ser were discardedfrom the correlation study since DC accumulation were poorlyreproduced by the model after fitting the partition coefficients
(Table 3, Figs. 1, 3 and 4) .The sewage sludge in the Ultuna experi-ment was also discarded from the correlation because it led to anunexpectedly large C accumulation in soil that may be ascribed toan inhibition of soil microbial activity, possibly due to increased soil
Accu
mul
ated
C (M
g C
ha-
1 )
Time (years)
0
10
20
30
40
50
0
10
20
30
40
50
60
0
10
20
30
40
50
0
10
20
30
40
50
1960 1980 20000
10
20
30
40
50
1960 1980 2000
STR-Ult -N STR-Ult +N
GM-Ult -N
SAW-Ult -N SAW-Ult +N
FYM-Ult -N FYM-Ult +P -N
CV(RMSE) = 28.1%
Δ CV(RMSE) = 3.9%
EF = -0.15
CV(RMSE) = 25.7%
Δ CV(RMSE) = 2.9%
EF = -0.18
CV(RMSE) = 14.4%
Δ CV(RMSE) = 2.1%
EF = 0.79
CV(RMSE) = 19.2%
Δ CV(RMSE) = 2.0%
EF = 0.48
SLU-Ult -N
PEA-Ult -N PEA-Ult +N
Fig. 4. Accumulation of soil C compared to reference plots after EOM applicationsduring the Ultuna experiment. Measured values (-) and values simulated with RothCusing partition coefficients fitted with field data (black lines) or predicted using theIROC indicator (grey dashed lines). GM: green manure, SLU: sewage sludge, STR: straw,SAW: sawdust, FYM: farmyard manure, PEA: peat. eN: plots without mineral Nfertilization, þN: plots with mineral N fertilization. Coefficient of variation of the RMSE(CV(RMSE)) and model efficiency (EF) of the simulations using the predicted partitioncoefficients calculated for the þN and eN parts of the experiment taken together, DCV(RMSE): difference between the CV(RMSE) values obtained using the predicted andfitted partition coefficients.
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e6056
heavy metals content or low soil pH (Witter et al., 1993; Kättereret al., 2011).
Positive significant correlations (p < 0.01) were found betweenthe labile pool fraction (fDPM) and EOM characteristics related tohigh potential biodegradability, hemicellulose-like fraction (HEM)
and proportion of EOM-TOC mineralized after 3 days of incubation(C3d) (Table 4). On the other hand, negative significant correlationswere found with more recalcitrant fraction, the lignin þ cutin-likefraction (LIC) and with the indicator of remaining C in soil (IROC). Areverse trend was found for the resistant fraction (fRPM) that waspositively correlated with IROC and LIC, and negatively correlatedwith HEM and C3d. The humified fraction (fHUM) was positivelycorrelated with IROC (p < 0.01) and negatively with C3d (p< 0.05). Ahigh EOM-TOC revealed less stabilized EOM as shown by thepositive correlationwith the labile fraction (fDPM) and negativewiththe resistant fraction (fRPM).
The EOMs with the highest fDPM coefficients were greenmanureand straw applied in the Ultuna experiment, which led to low DCaccumulation in soil. This can be ascribed to their low proportion oflignin and cutin-like Van Soest fraction and the high proportion of Cmineralized after 3 days of incubation (Table 2). Composted EOMsshowed the highest fRPM coefficients (ALG-Ser, BIOW-Qua, GWS-Qua) and contained a large soluble (SOL) fraction, as often seen formature composts (Annabi et al., 2007; Doublet et al., 2010; Peltreet al., 2010). This reflects the abundance of stabilized compoundsin the composts. The composts also showed a low proportion of Cmineralized after 3 days of incubation and a high IROC value, indi-cating a high level of stability against biodegradation.
Very variable partition coefficients were found for EOMs of thesame type. For example, the FYMs applied in the four LTEs haddifferent partition coefficients, reflecting their differentbiochemical composition and C3d values. In the SERAIL experi-ment, 70% of the C applied in FYM was allocated to the labile pool(fDPM) in accordance with its high C3d value, small proportion ofthe ligninþcutin- like fraction (LIC) and low IROC value (Table 2).In the Qualiagro experiment, the FYM had most C in the resistantand humified pools (50 and 20% of EOM-TOC, respectively) inaccordance with its lower C3d, higher LIC and higher IROC valuescompared with FYM-Ser. The coefficients obtained for FYMs inthe present LTEs differed from coefficients derived from theRothamsted experiment (fDPM ¼ 49%, fRPM ¼ 49%, fHUM ¼ 2%,Coleman and Jenkinson, 1999). This confirmed that FYM is anEOM with very variable composition and decomposability,depending on its origin, proportion and type of bedding materialsand storage conditions (Morvan et al., 2006). Such variability ofthe analytical results suggests that taking account of thecomposition of a given EOM would enable a better simulation oftheir contribution to soil C.
3.3. Yield of additional C inputs remaining in soil
The differences in climatic conditions, soil types and inputs ofEOM-C among the four LTEs prevented a between-site comparisonof DC accumulation in soil. Therefore we calculated the DC accu-mulation for a specific soil after 20 years of EOM applications usingthe partition coefficients fitted previously and for three differentclimatic regimes (see Section 2.5). The climate regime influencedthe proportion of EOM-TOC accumulating in soil substantially. Theaverage DC accumulations in soil were 1.6 times higher underNordic than under Mediterranean climate, the DC accumulationbeing intermediate under the temperate climate (Table 5). TheEOMs differed in their ability to increase soil C stock, which wasrelated to a higher C input in the DPM pool for the less stable EOMand a higher C input in the RPM and HUM pools for the most stableEOMs. With an EOM input of 2 Mg C ha�1 every second year, theannual increase in soil C ranged 0.13e0.42 Mg C ha�1 yr�1 underMediterranean climate, 0.15e0.50 Mg C ha�1 yr�1 underTemperate climate, and 0.19e0.68 Mg C ha�1 yr�1 under Nordicclimate. The composts accumulated much more C in soil than non-composted materials. For example, soil amended with the
Table 4Pearson correlation coefficients between EOM characteristics and RothC partition coefficients and yields of C accumulation under different climatic conditions. Total organic Ccontent (TOC), soluble, hemicellulose-, cellulose- and lignin þ cutin-like fractions (SOL, HEM, CEL and LIC, respectively), proportion of EOM-TOC mineralized after 3 days oflaboratory incubation at 28 �C (C3d), indicator of remaining organic C in soil (IROC), partition coefficients of EOM-TOC into the labile, resistant and humified pools of RothC (fDPM,fRPM, fHUM, respectively).
TOC SOL HEM CEL LIC C3d IROC fDPM fRPM
TOC 1SOL �0.90*** 1HEM 0.59* �0.54 1CEL 0.88*** �0.92*** 0.41 1LIC �0.54* 0.41 �0.88*** �0.48 1C3d 0.37 �0.11 0.49 0.16 �0.59 1IROC �0.79*** 0.65** �0.79*** �0.68** 0.85*** �0.78*** 1fDPM 0.80*** �0.63* 0.78*** 0.62* �0.79*** 0.71** �0.93*** 1fRPM �0.82*** 0.62* �0.80*** �0.63* 0.83*** �0.67** 0.92*** �0.96*** 1fHUM �0.49 0.43 �0.47 �0.39 0.43 �0.58* 0.64** �0.75** 0.55*
*, ** and *** indicate significance at p < 0.05, 0.01 and 0.001, respectively.
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e60 57
biowaste compost (BIOW-Qua) accumulated approximately twotimes more C than soil amended with straw (STR-Ask), the rates ofC accumulation under temperate climate being 0.50 and 0.26 MgC ha�1 yr�1, respectively.
The rates of C accumulation following FYM applications werefairly high compared to previous results of Smith et al. (1997a).They estimated accumulations of 0.16 Mg C ha�1 yr�1 for FYMapplied at the rate of 10 Mg ha�1 yr�1 during 100 years of repeatedapplication (approximately 1.75 Mg C ha�1 yr�1 with a TOC contentof 35% and a drymatter content of 50%). If our FYMswere applied ata similar rate of 1.75 Mg C ha�1 yr�1 under temperate climate, 0.19,0.31, 0.41 and 0.42Mg C ha�1 yr�1 would accumulate during similarrepeated application of the FYM applied in the Serail, Ultuna, Askovand Qualiagro experiments, respectively. The C accumulationdiffers widely among the FYMs due to the differences in theirbiochemical composition (Table 2). It confirms the importance ofaccounting for the specific composition of each EOM by performinglaboratory measurement of EOM composition.
3.4. Partition coefficients from laboratory characterizations ofEOMs
The previously fitted partition coefficients were estimated fromlaboratory characterizations of EOMs, again discarding the PEA-
Table 5Yields of C accumulation in soil and annual rates of C accumulation after 20 years of cconditions and different climates for all EOMs, with EOMs applied at a dose of 2 Mg C ha�1
(Qua), SERAIL (Ser) and Ultuna (Ult) experiments.
Y20y (% of the total C inputs)
Treatment Mediterranean climate Temperate climate Nordic clim
STR-Ask 20.9 25.5 35.1SAW-Ask 26.2 30.3 38.9FYM-Ask 34.6 40.0 50.7MSW-Qua 18.4 22.8 32.3FYM-Qua 36.0 42.1 54.2GWS-Qua 38.7 46.1 61.1BIOW-Qua 41.6 50.3 68.2FMT-Ser 16.8 20.4 28.2FYM-Ser 17.0 20.7 28.7VGH-Ser 22.7 29.1 43.0ALG-Ser 33.3 41.9 60.1STR-Ult 13.7 15.7 20.3GM-Ult 13.3 15.2 19.3SAW-Ult 21.5 26.5 37.3FYM-Ult 26.8 31.7 41.9SLU-Ult 38.5 47.5 66.3
STR: straw, SAW: sawdust, FYM: farmyardmanure, MSW:municipal solid waste compostFYM, VGH: enriched coffee cake compost, ALG: enriched bark compost, GM: green manu
Ask, PEA-Ult GWC-Ser and SLU-Ult treatments due to the poorquality of the model fitting, probably related to changes in soilfactors not accounted for in RothC such as low soil pH or high heavymetal concentrations indicating the need of EOM- or site-specificmodel re-parameterization in such specific cases as reported else-where (Todorovic et al., 2010).
IROC was the variable best correlated with RothC partitioncoefficients (Table 4) andwas therefore used to predict the fDPM andfRPM coefficients (expressed as a % of EOM-TOC). Thus fHUM wascalculated as fHUM ¼ 100% � fDPM � fRPM. The regressions (Table 6and Fig. 5) showed a satisfactory accuracy for the prediction offDPM and fRPM, the R2CV based on leave-one-out cross-validationbeing 0.84 and 0.80 for fDPM and fRPM, respectively. The slope of thefDPM and fRPM regression models (a coefficients, Table 6) differedsignificantly from 0 (p < 0.001) and were close to 1 for the fRPMregression model. The small differences between the calibrationand cross-validation results indicated a fairly good robustness ofthe regression models. Other EOM analytical characteristics orcombinations thereof were inferior in predicting the partitioncoefficients and the simulated evolution of DC accumulation in soil(results not shown).
The predicted EOM partition coefficients were used in RothC tosimulate the DC accumulation in the LTEs. Table 7 presents thepredicted partition coefficients along with statistics to assess the
ultivation (Y20y, see part 2.5 for calculation) calculated with RothC using standardevery two years (20 Mg C ha�1 in 20 years). Results for the Askov K2 (Ask), Qualiagro
Rate of C accumulation (Mg C ha�1 yr�1)
ate Mediterranean climate Temperate climate Nordic climate
0.21 0.26 0.350.26 0.30 0.390.35 0.40 0.510.18 0.23 0.320.36 0.42 0.540.39 0.46 0.610.42 0.50 0.680.17 0.20 0.280.17 0.21 0.290.23 0.29 0.430.33 0.42 0.600.14 0.16 0.200.13 0.15 0.190.21 0.26 0.370.27 0.32 0.420.38 0.48 0.66
, GWS: green waste and sludge compost, BIOW: biowaste compost, FMT: dehydratedre, SLU: sewage sludge.
Table 6Linear regression between the RothC partition coefficients fDPM and fRPM as dependent variables and the indicator of remaining organic C in soil (IROC) as the independentvariable (predictor). Slope (a) and intercept (b) of the regression line presented� standard deviation. Rootmean square error, coefficient of variation of the RMSE, coefficient ofdetermination given for the calibration results (RMSE, CV(RMSE), R2) and for the cross-validation results (RMSECV, CV(RMSECV), R2CV).
Dependent variable Predictor Regression coefficients RMSE%TOC
CV(RMSE)%
R2 RMSECV%TOC
CV(RMSECV)%
R2CV
a b
fDPM IROC �1.254 � 0.14 115.922 � 7.78 9.40 19.29 0.87 10.58 21.70 0.84fRPM IROC 0.979 � 0.12 �8.928 � 6.59 8.08 18.57 0.85 9.34 21.45 0.80
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e6058
goodness of fit of the resulting simulations of DC accumulation.Figs. 1e4 shows the measured and simulated kinetics of DC accu-mulation in soil using either the fitted or predicted partition coef-ficients. The differences between the simulations based on fittedand on predicted coefficients were assessed by D CV(RMSE),calculated as the difference between the CV(RMSE) values forpredicted and fitted coefficients.
IROC (% TOC)0 20 40 60 80 100
f RPM
(% T
OC
)
0
20
40
60
80
100
IROC (% TOC)0 20 40 60 80 100
f DPM
(% T
OC
)
0
20
40
60
80
100fDPM = -1.254 * IROC + 115.922
R² = 0.87
RMSE = 9.4
RMSECV = 10.6
fRPM = 0.979 * IROC - 8.928
R² = 0.85
RMSE = 8.1
RMSECV = 9.3
Fig. 5. Relationship between the indicator of remaining organic C (IROC, errorbars ¼ standard deviations) and partition coefficients into the labile and resistant poolsof RothC (fDPM and fRPM). Lines are regression lines, determination coefficient (R2), rootmean square error calculated using the entire sample set and using cross-validationpredictions (RMSE and RMSECV, respectively).
For the Askov K2 experiment (Fig. 1) the simulation usingpredicted coefficients remained satisfactory for the straw amen-ded plot, despite a relatively high D CV(RMSE) value, and thesimulated values remained close to the standard deviation ofmeasured values. Small differences were found for the sawdustamended plots using fitted and predicted coefficients, whereasfor FYM plots the use of predicted coefficients eventually led toan underestimation of C accumulation. This could be due tothe use of average IROC values for the FYM applied in theother experiments, the compositions of which varied markedly(Table 2).
Successful simulations were achieved for the Qualiagro experi-ment using the predicted partition coefficients for plots takingMSW compost, FYM and GWS compost, with small D CV(RMSE)values between the simulations using fitted and predicted coeffi-cients and good model efficiency values (0.87e0.93, Table 7, Fig. 2).Simulations with the predicted coefficients slightly underestimatedthe measured values for plots given BIOW compost.
In the SERAIL experiment (Fig. 3), small differences were foundbetween the predicted and fitted partition coefficients for plotstreated with FYM, VGH compost and ALG compost (D CV(RMSE) of9.4%, 11.8% and 0.9%, respectively). A higher D CV(RMSE) value wasfound for plots amended with dehydrated FYM. The CV(RMSE) andthe correlation coefficients remained better for the FMT treatmentthan for the other treatments because simulations using fittedcoefficients were superior for this treatment.
The results for the Ultuna experiment (Fig. 4) indicated smalldifferences between the simulations with fitted and predictedcoefficients (D CV(RMSE) ranging from 2.0% to 3.9%). The correla-tion coefficients indicated relatively good reproducibility of thetrend of DC accumulation in soil. However, low model efficiencieswere obtained for plots amended with straw and green manurebecause the model failed to simulate the soil C steady stateobserved at the end of the experimental period, as previouslydiscussed.
The parameterization used a classical least squares optimiza-tion method as in most studies dealing with calibrations of SOCmodels (Andren and Kätterer, 1997; Gabrielle et al., 2004; Guoet al., 2007; Ludwig et al., 2011). This method does not allow theuncertainty of model parameters or model predictions to beassessed as in Bayesian and Monte Carlo methods (Barré et al.,2010; Juston et al., 2010). However we consider the uncertaintyto be limited since the model was run on differences in soil Cbetween EOM amended and un-amended reference plots. Asa result it was not necessary to estimate the initial size of themodel pools, which is a considerable source of uncertainty formodel predictions (Christensen, 1996; Zimmermann et al., 2007),the size estimation of the inert pool being particularly sensitive(Falloon et al., 2000). Finally, the uncertainty of model results canbe seen as fairly satisfactory compared with experimental uncer-tainty, as the predictions obtained from cross-validation usingpartition coefficients predicted from analytical characterizationswere generally within the standard deviation of measured C stocks(Figs. 1 and 2).
Table 7Predictions of EOM partition coefficients in the RothC labile, resistant and humified entry pools (fDPM, fRPM, fHUM) with the regression models presented in Table 7 and thequality of simulations of C accumulation dynamics with RothC using the predicted partition coefficients. RMSE: root mean square error between measured and simulatedaccumulated C values, CV(RMSE): coefficient of variation of the RMSE, D CV(RMSE): difference of the CV(RMSE) values using the predicted and fitted partition coefficients, r:coefficient of correlation, EF model efficiency. Results for the Askov K2 (Ask), Qualiagro (Qua), SERAIL (Ser) and Ultuna (Ult) experiments.
Treatment fDPM% TOC
fRPM% TOC
fHUM% TOC
RMSEMg C ha�1
CV(RMSE)%
D CV(RMSE)%
r EF
STR-Ask 73.7 23.7 2.6 3.1 25.4 12.0 0.95 0.61SAW-Ask 55.3 38.9 5.9 3.3 21.0 3.5 0.87 0.68FYM-Ask 39.1 51.9 9.0 3.4 22.2 13.6 0.97 0.71MSW-Qua 59.3 34.9 5.8 0.9 17.7 3.8 0.97 0.87FYM-Qua 29.5 59.6 11.0 1.4 15.5 2.1 0.97 0.91GWS-Qua 17.4 69.1 13.5 1.5 15.3 2.0 0.98 0.93BIOW-Qua 21.5 65.1 13.4 2.7 28.6 11.6 0.97 0.76FMT-Ser 52.8 40.1 7.1 3.0 55.3 19.1 0.74 �0.06FYM-Ser 52.9 40.0 7.1 3.9 75.1 9.4 0.61 0.19VGH-Ser 27.6 58.8 13.6 4.6 73.3 11.8 0.33 �0.64ALG-Ser 13.2 69.2 17.6 3.9 42.1 0.9 0.65 0.37STR-Ult 80.0 19.3 0.6 2.7 28.1 3.9 0.82 �0.15GM-Ult 100.0 0.0 0.0 2.8 25.7 2.9 0.82 �0.18SAW-Ult 55.5 37.9 6.6 2.0 14.4 2.1 0.95 0.79FYM-Ult 53.1 40.3 6.6 3.6 19.2 2.0 0.91 0.48
STR: straw, SAW: sawdust, FYM: farmyardmanure, MSW:municipal solid waste compost, GWS: green waste and sludge compost, BIOW: biowaste compost, FMT: dehydratedFYM, VGH: enriched coffee cake compost, ALG: enriched bark compost, GM: green manure.
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e60 59
4. Conclusions
Fitting of the partition coefficients of exogenous organic matters(EOMs) in the entry pools of the RothC model (labile, resistant andhumified pools) generally satisfactorily reproduced the kinetics of Caccumulation in soil after repeated EOM applications compared toreference plots without EOM application in four long-term fieldexperiments, without modifying the other parameters of themodel. The EOM derived increase in SOC stocks ranged from 0.15 to0.50 Mg C ha�1 yr�1 when considering a period of 20 years,2 Mg C ha�1 of EOMs applied every two years, and a temperateclimate. The EOMs covered a broad range of materials providingcontrasting values for fitted partition coefficients that were relatedto their chemical composition. The partition coefficients could bedetermined using the indicator of potentially remaining organic Cin soil (IROC) calculated from the biochemical fractions of Van Soestfractionation and the proportions of C mineralized after three daysof laboratory incubation. The use in RothC of the partition coeffi-cients predicted from laboratory characterizations enabled a simu-lation of SOC accumulation after EOM applications with relativelysmall differences (an average increase of 6.7% in the coefficient ofvariation of the RMSE) compared to the use of fitted partitioncoefficients. The proposed partitioning of EOM-TOC makes itpossible to predict the potential C storage after EOM applicationsunder contrasted climatic conditions and allows accounting for thespecific composition of most types of EOMs.
Acknowledgements
The authors would like to thank Veolia Environnement,Research and Innovation for its technical and financial support forthe Qualiagro experiment, Guillaume Bodineau, Muriel Jolly, Vin-cent Mercier, Aurelia Michaud and Jean-Noel Rampon for their helpwith soil sampling, soil and EOMs analyses and data managementfor the Qualiagro experiment. The assistance provided by Elly M.Hansen in organizing and processing of the data from the Askov K2experiment is gratefully acknowledged. Two anonymous reviewersare acknowledged for their comments on the manuscript.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.soilbio.2012.03.023
References
AFNOR, 1995. Norme NF ISO 10694. Qualité du sol - Dosage du carbone organique etdu carbone total après combustion sèche (analyse élémentaire). In: Qualité dessols 1. AFNOR, Paris, pp. 361e371.
AFNOR, 2009a. Norme XP U 44e162. Amendements organiques et supports deculture e Caractérisation de la matière organique par fractionnement bio-chimique et estimation de sa stabilité biologique. AFNOR, Paris.
AFNOR, 2009b. Norme XP U 44-163. Amendements organiques et supports deculture e Caractérisation de la matière organique par la minéralisationpotentielle du carbone et de l’azote. AFNOR, Paris.
Andren, O., Kätterer, T., 1997. ICBM: the introductory carbon balance model forexploration of soil carbon balances. Ecological Applications 7, 1226e1236.
Annabi, M., Houot, S., Francou, F., Poitrenaud, M., Le Bissonnais, Y., 2007. Soilaggregate stability improvement with urban composts of different maturities.Soil Science Society of America Journal 71, 413e423.
Barré, P., Eglin, T., Christensen, B.T., Ciais, P., Houot, S., Kätterer, T., van Oort, F.,Peylin, P., Poulton, P.R., Romanenkov, V., Chenu, C., 2010. Quantifying andisolating stable soil organic carbon using long-term bare fallow experiments.Biogeosciences 7, 3839e3850.
Berry, D., Demarle, O., Vian, J.F., Gautronneau, Y., Estrade, J.R., 2008. Organicamendments: a report from 10 years of use in the culture of legumes. PHMRevue Horticole, 39e44.
Bipfubusa, M., Angers, D.A., N’Dayegamiye, A., Antoun, H., 2008. Soil aggregationand biochemical properties following the application of fresh and compostedorganic amendments. Soil Science Society of America Journal 72, 160e166.
Bolinder, M.A., Janzen, H.H., Gregorich, E.G., Angers, D.A., VandenBygaart, A.J., 2007.An approach for estimating net primary productivity and annual carbon inputsto soil for common agricultural crops in Canada. Agriculture, Ecosystems &Environment 118, 29e42.
Bruun, S., Christensen, B.T., Hansen, E.M., Magid, J., Jensen, L.S., 2003. Calibrationand validation of the soil organic matter dynamics of the Daisy model with datafrom the Askov long-term experiments. Soil Biology & Biochemistry 35, 67e76.
Christensen, B.T., 1996. Matching measurable soil organic matter fractions withconceptual pools in simulation models of carbon turnover: revision of modelstructure. In: al, D.P.e. (Ed.), Evaluation of Soil Organic Models Using Long-termDatasets NATO ASI Series I: Global Environmental Change. Springer-Verlag,Berlin Heidelberg, Germany, pp. 143e160.
Christensen, B.T., Johnston, A.E., 1997. Soil organic matter and soil quality e lessonslearned from long-term experiments at Askov and Rothamsted. Developmentsin Soil Science 25, 399e430.
Chung, H.G., Grove, J.H., Six, J., 2008. Indications for soil carbon saturation ina temperate agroecosystem. Soil Science SocietyofAmerica Journal 72,1132e1139.
Ciais, P., Wattenbach, M., Vuichard, N., Smith, P., Piao, S.L., Don, A., Luyssaert, S.,Janssens, I.A.,Bondeau,A., Dechow,R., Leip, A., Smith, P.C.,Beer, C., vanderWerf,G.R.,Gervois, S., VanOost,K., Tomelleri, E., Freibauer, A., Schulze, E.D., 2010. TheEuropeancarbon balance. Part 2: croplands. Global Change Biology 16, 1409e1428.
Coleman, K., Jenkinson, D.S., 1999. A Model for the Turnover of Carbon in Soil,Model Description and Windows Users Guide. IACR, Rothamsted HarpendenHerts AL5 2JQ, p. 47.
Coleman, K., Jenkinson, D.S., Crocker, G.J., Grace, P.R., Klir, J., Korschens, M.,Poulton, P.R., Richter, D.D., 1997. Simulating trends in soil organic carbon inlong-term experiments using RothC-26.3. Geoderma 81, 29e44.
Djakovitch, J.L., 1988. Mise au point d’une méthode de détermination rapide ducoefficient isohumique de matériaux organiques utilisables pour l’amendementdes sols. Thèse CNAM, Bordeaux, p. 208.
C. Peltre et al. / Soil Biology & Biochemistry 52 (2012) 49e6060
Doublet, J., Francou, C., Poitrenaud, M., Houot, S., 2010. Sewage sludge composting:influence of initial mixtures on organic matter evolution and N availability inthe final composts. Waste Management 30, 1922e1930.
Ellert, B.H., Bettany, J.R., 1995. Calculation of organic matter and nutrients stored insoils under contrasting management regimes. Canadian Journal of Soil Science75, 529e538.
European Commission, 2006. Proposal for a Directive of the European Parliamentand of the Council Establishing a Framework for the Protection of Soil. http://ec.europa.eu/environment/soil/pdf/com_2006_0232_en.pdf.
European Commission, 2010. Communication from the Commission to the Counciland the European Parliement on Future Steps in Bio-waste Management in theEuropean Union. http://ec.europa.eu/environment/waste/compost/pdf/com_biowaste.pdf.
Falloon, P., Smith, P., 2002. Simulating SOC changes in long-term experiments withRothC and CENTURY: model evaluation for a regional scale application. Soil Useand Management 18, 101e111.
Falloon, P., Smith, P., Coleman, K., Marshall, S., 2000. How important is inert organicmatter for predictive soil carbon modelling using the Rothamsted carbonmodel? Soil Biology & Biochemistry 32, 433e436.
Gabrielle, B., Silveira, J.d., Houot, S., Francou, C., 2004. Simulating urban wastecompost effects on carbon and nitrogen dynamics using a biochemical index.Journal of Environmental Quality 33, 2333e2342.
Gerzabek, M.H., Pichlmayer, F., Kirchmann, H., Haberhauer, G., 1997. The response ofsoil organic matter to manure amendments in a long-term experiment atUltuna, Sweden. European Journal of Soil Science 48, 273e282.
Guo, L., Falloon, P., Coleman, K., Zhou, B., Li, Y., Lin, E., Zhang, F., 2007. Application ofthe RothC model to the results of long-term experiments on typical upland soilsin northern China. Soil Use and Management 23, 63e70.
Houot, S., Clergeot, D., Michelin, J., Francou, C., Bourgeois, S., Caria, G., Ciesielslki, H.,2002. Agronomic value and environmental impacts of urban composts used inagriculture. In: Insam, H., Riddech, N., Klammer, S. (Eds.), Microbiology ofComposting. Springer-Verlag, Berlin Heidelberg, pp. 457e472.
Hyvönen, R., Agren, G.I., Andrén, O., 1996. Modelling long-term carbon and nitrogendynamics in an arable soil receiving organic matter. Ecological Applications 6,1345e1354.
IPCC, 1997. Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories.In: Reference Manual. Intergovernmental Panel on Climate Change, vol. 3.http://www.ipcc-nggip.iges.or.jp/public/gl/invs6.html.
Jackson, R.B., Canadell, J., Ehleringer, J.R., Mooney, H.A., Sala, O.E., Schulze, E.D., 1996.A global analysis of root distributions for terrestrial biomes. Oecologia 108,389e411.
Jenkinson, D.S., Rayner, J.H., 1977. The turnover of soil organic matter in some of theRothamsted classical experiments. Soil Science 123, 298e305.
Juston, J., Andren, O., Katterer, T., Jansson, P.E., 2010. Uncertainty analyses for cali-brating a soil carbon balance model to agricultural field trial data in Swedenand Kenya. Ecological Modelling 221, 1880e1888.
Kamoni, P.T., Gicheru, P.T., Wokabi, S.M., Easter, M., Milne, E., Coleman, K., Falloon, P.,Paustian, K., Killian, K., Kihanda, F.M., 2007. Evaluation of two soil carbonmodels using two Kenyan long term experimental datasets. Agriculture,Ecosystems & Environment 122, 95e104.
Kätterer, T., Hansson, A.C., Andren, O., 1993. Wheat root biomass and nitrogendynamics e effects of daily irrigation and fertilization. Plant and Soil 151,21e30.
Kätterer, T., Bolinder, M.A., Andrén, O., Kirchmann, H., Menichetti, L., 2011. Rootscontribute more to refractory soil organic matter than above-ground cropresidues, as revealed by a long-term field experiment. Agriculture, Ecosystems& Environment 141, 184e192.
Lal, R., Follett, R.F., Stewart, B.A., Kimble, J.M., 2007. Soil carbon sequestration tomitigate climate change and advance food security. Soil Science 172, 943e956.
Lashermes, G., Nicolardot, B., Parnaudeau, V., Thuriès, L., Chaussod, R.,Guillotin, M.L., Linères, M., Mary, B., Metzger, L., Morvan, T., Tricaud, A.,Villette, C., Houot, S., 2009. Indicator of potential residual carbon in soils afterexogenous organic matter application. European Journal of Soil Science 60,297e310.
Lee, J.H., Hopmans, J.W., Rolston, D.E., Baer, S.G., Six, J., 2009. Determining soilcarbon stock changes: simple bulk density corrections fail. Agriculture,Ecosystems & Environment 134, 251e256.
Leifeld, J., Zimmermann, M., Fuhrer, J., 2008. Simulating decomposition of labile soilorganic carbon: effects of pH. Soil Biology & Biochemistry 40, 2948e2951.
Leifeld, J., Reiser, R., Oberholzer, H.R., 2009. Consequences of conventional versusorganic farming on soil carbon: results from a 27-year field experiment.Agronomy Journal 101, 1204e1218.
Linères, M., Djakovitch, J.L., 1993. Caractérisation de la stabilité biologique desapports organiques par l’analyse biochimique. In: Decroux, J., Ignazi, J.C. (Eds.),Quatrième journées de l’analyse de terre et cinquième forum de la fertilisationraisonnée. GEMAS-COMIFER, Blois, pp. 159e168.
Ludwig, B., Bergstermann, A., Priesack, E., Flessa, H., 2011. Modelling of crop yieldsand N(2)O emissions from silty arable soils with differing tillage in two long-term experiments. Soil & Tillage Research 112, 114e121.
Machet, J.M., Dubrulle, P., Damay, N., Duval, R., Recous, S., Mary, B., 2007. Azofert:a new decision support tool for fertiliser N advice based on a dynamic versionof the predictive balance sheet method. In: 16th International Symposium ofthe International Scientific Centre of Fertilizers, Gand, p. 6.
Marmo, L., Feix, I., Bourmeau, E., Amlinger, F., Bannick, C.G., De Neve, S., Favoino, E.,Gendebien, A., Gibert, J., Givelet, M., Leifert, I., Morris, R., Rodriguez Cruz, A.,Ruck, F., Siebert, S., Tittarelli, F., 2004. Reports of the Technical Working GroupsEstablished Under the Thematic Strategy for Soil Protection. In: Organic Matterand Biodiversity. Taskgroup 4 on Exogenous Organic Matter, vol. III. Office forOfficial Publications of the European Communities, Luxembourg. http://ec.europa.eu/environment/soil/pdf/vol3.pdf.
Matson, P.A., Parton, W.J., Power, A.G., Swift, M.J., 1997. Agricultural intensificationand ecosystem properties. Science 277, 504e509.
Morvan, T., Nicolardot, B., 2009. Role of organic fractions on C decomposition and Nmineralization of animal wastes in soil. Biology and Fertility of Soils 45, 477e486.
Morvan, T., Nicolardot, B., Péan, L., 2006. Biochemical composition and kinetics of Cand N mineralization of animal wastes: a typological approach. Biology andFertility of Soils 42, 513e522.
Nicolardot, B., Recous, S., Mary, B., 2001. Simulation of C and N mineralisationduring crop residue decomposition: a simple dynamic model based on the C:Nratio of the residues. Plant and Soil 228, 83e103.
Paré, T., Dinel, H., Schnitzer, M., Dumontet, S., 1998. Transformations of carbon andnitrogen during composting of animal manure and shredded paper. Biology andFertility of Soils 26, 173e178.
Parton, W.J., Schimel, D.S., Cole, C.V., Ojima, D.S., 1987. Analysis of factors controllingsoil organic matter levels in Great Plains grasslands. Soil Science Society ofAmerica Journal 51, 1173e1179.
Peltre, C., Dignac, M.F., Derenne, S., Houot, S., 2010. Change of the chemicalcomposition and biodegradability of the Van Soest soluble fraction duringcomposting: a study using a novel extraction method. Waste Management 30,2448e2460.
Penman, H.L., 1948. Natural evaporation from open water, bare soil and grass. In:Proceedings of the Royal Society of London. Series A, Mathematical and PhysicalSciences 193, pp. 120e145.
Powlson, D.S., Bhogal, A., Chambers, B.J., Coleman, K., Macdonald, A.J.,Goulding, K.W.T., Whitmore, A.P., 2012. The potential to increase soil carbonstocks through reduced tillage or organic material additions in England andWales: a case study. Agriculture, Ecosystems & Environment 146, 23e33.
Six, J., Conant, R.T., Paul, E.A., Paustian, K., 2002. Stabilization mechanisms of soilorganic matter: implications for C-saturation of soils. Plant and Soil 241,155e176.
Smith, P., Powlson, D.S., Glendining, M.J., Smith, J.O.U., 1997a. Potential for carbonsequestration in European soils: preliminary estimates for five scenarios usingresults from long-term experiments. Global Change Biology 3, 67e79.
Smith, P., Smith, J.U., Powlson, D.S., McGill, W.B., Arah, J.R.M., Chertov, O.G.,Coleman, K., Franko, U., Frolking, S., Jenkinson, D.S., Jensen, L.S., Kelly, R.H.,Klein-Gunnewiek, H., Komarov, A.S., Li, C., Molina, J.A.E., Mueller, T., Parton, W.J.,Thornley, J.H.M., Whitmore, A.P., 1997b. A comparison of the performance ofnine soil organic matter models using datasets from seven long-term experi-ments. Geoderma 81, 153e225.
Van Soest, P.J., 1963. Use of detergents in the analysis of fibrous feeds: II. A rapidmethod for the determination of fibre and lignin. Journal Association of OfficialAnalytical Chemists 46, 829e835.
Thomsen, I.K., Christensen, B.T., 2010. Carbon sequestration in soils with annualinputs of maize biomass and maize-derived animal manure: evidence from 13Cabundance. Soil Biology and Biochemistry 42, 1643e1646.
Thuriès, L., Pansu, M., Larré-Larrouy, M.C., Feller, C., 2002. Biochemical compositionand mineralization kinetics of organic inputs in a sandy soil. Soil Biology &Biochemistry 34, 239e250.
Todorovic, G.R., Stemmer, M., Tatzber, M., Katzlberger, C., Spiegel, H., Zehetner, F.,Gerzabek, M.H., 2010. Soil-carbon turnover under different crop management:evaluation of RothC-model predictions under Pannonian climate conditions.Journal of Plant Nutrition and Soil Science 173, 662e670.
Van Soest, P.J., Wine, R.H., 1967. Use of detergents in the analysis of fibrous feeds. IV.Determination of plant cell-wall constituents. Journal Association of OfficialAnalytical Chemists 58, 50e55.
Witter, E., Martensson, A.M., Garcia, F.V., 1993. Size of the soil microbial biomass ina long term experiment as affected by different N-fertilizers and organicmanures. Soil Biology & Biochemistry 25, 659e669.
Yang, X.M., Zhang, X.P., Fang, H.J., Zhu, P., Ren, J., Wang, L.C., 2003. Long-term effectsof fertilization on soil organic carbon changes in continuous corn of NortheastChina: RothC model simulations. Environmental Management 32, 459e465.
Yokozawa, M., Shirato, Y., Sakamoto, T., Yonemura, S., Nakai, M., Ohkura, T., 2010.Use of the RothC model to estimate the carbon sequestration potential oforganic matter application in Japanese arable soils. Soil Science and PlantNutrition 56, 168e176.
Zimmermann, M., Leifeld, J., Schmidt, M.W.I., Smith, P., Fuhrer, J., 2007. Measuredsoil organic matter fractions can be related to pools in the RothC model.European Journal of Soil Science 58, 658e667.