Agricultural Systems 112 (2012) 48–54

Contents lists available at SciVerse ScienceDirect

Agricultural Systems
journal homepage: www.elsevier.com/locate/agsy

Changes in soil organic carbon and climate change – Application of the RothC
model in agro-silvo-pastoral Mediterranean systems
Rosa Francaviglia a,⇑, Kevin Coleman b, Andrew P. Whitmore b, Luca Doro c, Giulia Urracci c,
Mariateresa Rubino a, Luigi Ledda c
a
CRA, Centro di Ricerca per lo Studio delle Relazioni tra Pianta e Suolo, Via della Navicella 2-4, 00184 Rome, Italy
b
Sustainable Soils and Grassland Systems, Rothamsted Research, Harpenden, Herts AL5 2JQ, United Kingdom
c
Dipartimento di Agraria, Università di Sassari, Viale Italia 39, 07100 Sassari, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Due to the complexity of soil organic carbon (SOC), models have proved very useful in helping to improve
Received 13 February 2012 our understanding of the soil–plant–atmosphere system. In this study we used the Rothamsted carbon
Received in revised form 30 May 2012 model (RothC), which considers the effect of climate (temperature, rainfall and evaporation), soil texture
Accepted 1 July 2012
(clay content) and crop management on the decomposition processes, to predict changes in SOC under
Available online 1 August 2012
different land uses and the different climate that may occur in the future.
SOC turnover was evaluated in land uses under different levels of agricultural intensification. The agri-
Keywords:
culture of the study-area is mainly extensive and markedly agro-silvo-pastoral, and is typical of north-
C sequestration
CO2 emissions
eastern Sardinia (Italy) and similar areas of the Mediterranean basin. The following land uses were con-
Climate change sidered: vineyards, hay crops, pastures, semi-natural systems (formerly vineyards) and cork oak forests.
Emission scenarios Since the study-area is characterized by the same soil type (Haplic Endoleptic Cambisols, Dystric) derived
Land use from granitic rocks and potential native vegetation (Quercus suber L.), this was considered as the model
RothC baseline to study soil organic carbon dynamics. Climate change scenarios for 90 yr were generated from
the baseline climate with two Global Climate Models: GISS (Goddard Institute of Space Studies, USA), and
HadCM3 (Met Office, Hadley Centre, UK), for two of the Intergovernmental Panel on Climate Change
(IPCC) emission scenarios (SRES A2 and B2).
Land uses at low cropping intensity were more efficient than intensive agricultural systems in terms of
SOC storage in the order: hay crop > pasture > cork oak forest > semi-natural systems. Land use change to
vineyards decreased the carbon balance significantly between 0.09 and 0.15 t C ha1 yr1. However,
former vineyards have almost recovered from soil disturbance following the establishment of the
vineyards.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction Different management systems can alter the stocks of SOC, mit-
igating or worsening climate change through C storage or C emis-
Carbon sequestration via agricultural soils has the potential to sions. Emissions of C from the soil are mainly due to the
contribute significantly to climate change mitigation, provided that decomposition of plant residues and humus oxidation (mineraliza-
specific measures are implemented. Sound cropland management tion). On the other hand agriculture can contribute to climate
can play a positive role in reducing GHGs emissions, and in partic- change mitigation by increasing the C sequestration in soils by
ular carbon dioxide, either by decreasing soil organic carbon losses, the adoption of practices such as diversifying cropping systems,
or by increasing the organic matter input, or a combination of both. reduction of erosion, application of inorganic fertilizer, tree plant-
Globally, soil organic carbon (SOC) has a stock of C of 1550 Pg to ing, animal manure application, green-manure cropping systems,
a depth of 1 m, compared to 760 Pg C in the atmosphere and improved grassland management, cropland-grazing and rotation,
560 Pg C in the above ground vegetation (Lal, 2008), so changes optimal fertilization (Franzluebbers and Doraiswamy, 2007) and
in the SOC stock have profound implications for the mitigation or overall conservation-tillage.
exacerbation of climate change (Smith et al., 2008). Due to the complexity of the soil–plant–atmosphere system, it
is helpful to study SOC dynamics through the use of simulation
models combined with local measured data. In order to overcome
⇑ Corresponding author. Tel.: +39 06 7005299; fax: +39 06 7005711. the problems associated with extensive and costly experiments,
E-mail address: [email protected] (R. Francaviglia). most of the known factors acting on SOC dynamics have been

0308-521X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.agsy.2012.07.001
 R. Francaviglia et al. / Agricultural Systems 112 (2012) 48–54 49

included in models through mathematical algorithms that take ac- and humified organic matter (HUM). The model allows two types
count of the various factors involved (climate, pedology, cropping of simulations: ‘‘direct’’ that uses the known input of organic car-
system, soil and crop management) and their complex interactions. bon to the soil to calculate the SOC, and ‘‘inverse’’ that evaluates
If well-validated, they can be used to predict changes in SOC under the input of organic carbon required to maintain the stock of SOC.
the different management and climatic conditions that may occur
in the future (Jones and Donnelly, 2004). There is a substantial glo- 2.2. The study area
bal effort in the area of SOC modeling involving a number of differ-
ent models (Smith et al., 1997; Kirschbaum et al., 2001), including To run RothC and evaluate the effects of climate change on soil
Century (Parton et al., 1994) and RothC (Jenkinson et al., 1992). organic carbon dynamics, we used the Berchidda–Monti data set
Both of these models have a similar structure, containing pools (Olbia Tempio, Sardinia, Fig. 1), based on surveys begun in 2007
with a rapid turnover (month–year), moderate turnover (decadal), within a national research project. Six different land uses, with dif-
and slow turnover (millennial or inert). ferent levels of cropping intensification were compared in the
SOC modeling under climate change has been previously stud- study area (Lagomarsino et al., 2011), characterized by the same
ied in the Mediterranean area, but available literature mostly deals type of soil and cork oak forest (Quercus suber L.) as potential native
with intensive agricultural systems and soil tillage practices. Luga- vegetation. These were:
to and Berti (2008), using the Century model evaluated the effects
of farm manure application, reduced tillage and grassland conver-  Tilled vineyards (area 0.98 ha) established in 1993 (TV).
sion in comparison with a business as usual scenario from 2008 to  No-tilled grassed vineyards (area 3.4 ha) established in 1990
2080, in a 4-yr rotation including maize, wheat and soybean in (GV).
north-east Italy.Àlvaro-Fuentes and Paustian (2011) used Century  Hay crop (oats, Italian ryegrass and annual clovers or vetch)
to model the SOC variations of a barley-fallow rotation and contin- with sparse cork oaks (area 3.5 ha) (HC).
uous barley system with conventional tillage and no-tillage from  Pasture, covered with spontaneous herbaceous vegetation with
2010 to 2100 in Zaragoza province (Spain). Farina et al. (2011) used sparse cork oaks (area 11 ha) (PA).
WinEPIC (Gerik et al., 2006) in a hilly area of central Italy, to sim-  Semi-natural systems (scrublands, Mediterranean maquis and
ulate the effect of climate change and soil tillage (conventional vs. Helichrysum meadows), former vineyards set-aside about
no-tillage) on two rainfed crop rotations (durum wheat–sunflower 30 yr ago (SN).
and durum wheat–maize), evaluating crops yields and SOC from  Cork oak forest (Q. suber L.) established in the past century,
2010 to 2100. where cork harvest occurs every 11 yr (area 15 ha) (CO).
The aim of this study was to evaluate the effects of climate
change on SOC dynamics using RothC in different land uses at dif- The tilled vineyards (TVs) and no-tilled grassed vineyards (GVs)
ferent levels of crop intensification (vineyards, hay crops, pastures, are agricultural higher intensive land uses, whereas hay crop (HC),
semi-natural systems and cork oak forests), in an extensive agro- pasture (PA), semi-natural (SN) and cork oak forest (CO) are agro-
silvo-pastoral area typical of north-eastern Sardinia and other silvo-pastoral lower intensive land uses.
areas of the Mediterranean basin (e.g. southern and central Spain). The tilled vineyard (TV) was established in 1993, and was
ploughed to 40 cm and harrowed in March or April and occasion-
ally in July. Organic fertiliser was added to provide about 50 kg
2. Materials and methods each of N and K2O ha1 and 25 kg P2O5 ha1 and pruning residues
were removed from the field. The mean grapes yield was between
2.1. The model 7 and 8 Mg ha1 fresh weight.
The no-tilled grassed vineyard (GV) was established in 1990.
RothC (Coleman and Jenkinson, 1996, 1999) used in this study, Mineral fertilizer up to 40 kg N ha1, 50 kg P2O5 ha1 and
is a model for the turnover of organic carbon in non-waterlogged 50 kg K2O ha1 was applied in March. Pruning was carried out in
topsoils that allows for the effects of climate (temperature, rainfall January and June with the pruning residues being left on the soil.
and evaporation), soil texture (clay content) and crop management Drip irrigation (up to 100 mm) was provided between June and
on the turnover process. It is one of the most widely used models July to partially restore crop evapotranspiration. The grape yield
for the study of SOC turnover, currently there are over 1300 regis- was between 7 and 8 Mg ha1 fresh weight.
tered users from over 80 countries. The model was originally
developed using data from the long-term field experiments, 14C lit-
ter decomposition experiments and radiocarbon dating of soils
from Rothamsted and Woburn, UK and Ibadan, Nigeria (Jenkinson
et al., 1987) and later for other areas including Australia, the USA,
Costa Rica, Austria and Nigeria (Jenkinson, 1990), Australia (Skj-
emstad et al., 2004), Brazil (Cerri et al., 2003), Europe (Smith
et al., 1997; Falloon et al., 2002), Japan (Yokozawa et al., 2010),
Kenya (Kamoni et al., 2007), Zambia (Kaonga and Coleman,
2008), Spain (Nieto et al., 2010). It uses a monthly time step to cal-
culate total organic carbon (t ha1), microbial biomass carbon
(t ha1) and D14C (from which the equivalent radiocarbon age of
the soil can be calculated) on a years to centuries timescale (Jen-
kinson et al., 1987, 1992; Jenkinson, 1990; Jenkinson and Coleman,
1994). The model requires little data to run and those needed are
easier available.
In RothC, SOC is divided into four active fractions and a small
amount of inert organic matter (IOM) that is resistant to decay.
The four active fractions are: decomposable plant material
(DPM), resistant plant material (RPM), microbial biomass (BIO) Fig. 1. Experimental site location in north-eastern Sardinia (Italy).
50 R. Francaviglia et al. / Agricultural Systems 112 (2012) 48–54

The hay crop (HC) and pasture (PA) land uses have a comple-
mentary 6-yr rotation. The pasture (PA) land use was 5 yr of spon-
taneous herbaceous vegetation, and 1 yr of intercropping with
oats, Italian ryegrass and annual clovers or vetch cultivated as an
hay crop. It is not tilled and is grazed from December until June
with 6 sheep ha1. The hay crop (HC) land use was oats, Italian rye-
grass and annual clovers or vetch for 5 yr and intercropped by
spontaneous herbaceous vegetation in the sixth year. It is managed
by ploughing to 40 cm and harrowed before seeding. About 50 kg N
ha1 and 90 kg P2O5 ha1 are applied before seeding and grazing is
allowed with 3–4 sheep ha1 from January until February or
March, before being cut in May. Both pasture and hay crop in-
cluded spotted cork-oak trees, which are key components of the
‘‘Dehesa’’-type landscape typical of this area of Sardinia. The
semi-natural systems (SN) arise from the natural re-vegetation of
former vineyards which have been abandoned, probably due to
the low grape yields and the high cost of modern tillage equip-
ment. The cork oak forest (CO) is used for cork production and cat- Fig. 2. Aerial photo of the site Berchidda–Monti (1977).
tle grazing and the understory is covered by mixed herbaceous
vegetation.
The local climate is typically Pluvi-seasonal oceanic low meso- (11.4%), a carbon input to the soil of 3.74 t C ha1 was necessary to
Mediterranean low sub-humid (Rivas-Martinez and Rivas-Saenz, achieve the SOC content of 50.5 t C ha1 to a depth of 20 cm mea-
2009), with a mean annual rainfall of 630 mm (70% in October– sured in 2007 (Table 1).
March) and mean annual temperature of 14.2 °C. Having set equilibrium conditions in this way for the native
The site is within an area of about 1450 ha, characterized by vegetation (cork oak forest), specific management files were cre-
extensive agro-silvo-pastoral systems, typical of north-eastern ated to simulate the effects of different land use changes starting
Sardinia (Italy) and similar areas of the Mediterranean basin (e.g. from 1900, with the aim to match the SOC contents measured in
southern and central Spain). Based on surveys began in 2007, the 2007 and thereafter to model the effects of future climate change.
soils are mainly Haplic Endoleptic Cambisols, Dystric (WRB, For each land use the following was assumed:
2006), derived from granitic rocks, are sub-acid to acid (pH 5.0–
6.4), sandy (sand 730–830 g kg1), with sandy-loam and loamy- (a) tilled vineyard (TV): the model was kept in equilibrium con-
sand texture, 0.93–1.07 g total N kg1 in the two vineyards and ditions under the cork oak forest from 1900 to 1992, then
semi-natural systems, 1.69–2.08 g total N kg1 in cork oak, hay 2 yr with zero plant inputs (1993–1994) were simulated to
crop and pasture, 12 mg available P2O5 kg1 in cork oak forest take into account soil levelling and planting, then a transi-
and semi-natural systems, 61–81 mg kg1 P2O5 in all the other tion phase (1995–1996) characterized by reduced plant
land uses. Organic carbon content in g kg1 was 14.2 in TV, 11.0 inputs to simulate the development phase of the vineyard
in GV, 24.3 in HC, 18.7 in PA, 17.0 in SN and 21.8 in CO. The soil (0.50 t C ha1), and finally the simulation of the fully devel-
reaction (pH) was determined in 1:2.5 soil:water suspension by oped vineyard from 1997, with a carbon input of
potentiometric method using a pH meter; particle-size analysis 1.00 t C ha1;
and soil texture with the wet sieving and sedimentation procedure (b) no-tilled and grassed vineyard (GV): the model was kept in
and the USDA classification respectively; total N with the Kjeldahl equilibrium conditions under the cork oak forest from
method; available P2O5 with the Olsen method; organic carbon 1900 to 1989, then 2 yr with zero plant inputs (1990–
with the Walkley–Black method. 1991) were simulated to take into account soil levelling
Soils samples where collected along the different soil horizons and planting, then a transition phase (1992–1995) charac-
during the pedologic survey, where pits were digged with a mini terized by reduced plant inputs to simulate the development
excavator. Three replicates were collected in the vineyards and phase of the vineyard (0.50–1.00 t C ha1), and finally the
the cork oak forest, four replicates in the hay crop and the pasture, simulation of the fully developed vineyard from 1996, with
six replicates in the semi-natural systems. a carbon input of 1.75 t C ha1;
To apply RothC and allow a comparison among the different (c) hay crop (HC): the model was kept in equilibrium conditions
land uses, SOC contents were normalized at a fixed 20 cm depth under the cork oak forest from 1900 to 1933, and the land
considering the A and Ap soil horizons of each replicate. Soil bulk use change was simulated from 1934 with a carbon input
density, required to convert SOC to t ha1, was calculated after Sax- of 5.99 t C ha1;
ton and Rawls (2006) from soil texture (sand and clay) and SOC (d) pasture (PA): the model was kept in equilibrium conditions
content. under the cork oak forest from 1900 to 1933, and the land
use change was simulated from 1934 with a carbon input
2.3. Model parameterisation of 3.68 t C ha1;
(e) semi-natural systems (SN) represented by former vineyards:
The model baseline was established with cork oak forest (CO) as the model was kept in equilibrium conditions under the cork
the potential native vegetation, based on aerial photographs taken oak forest from 1900 to 1938, then the vineyard establish-
in 1977 (Fig. 2) provided by the Central Institute for Cataloguing ment (2 yr with zero plant inputs from 1939 to 1940) and
and Documentation of the Ministry of Cultural Heritage and the gradual development were first simulated with a transi-
Activities. tion phase (1941–1942) characterized by reduced plant
RothC was run to equilibrium in inverse mode (i.e. the carbon inputs (1.00 t C ha1) and the fully developed vineyard from
input was adjusted, so that the modeled SOC matched the initial 1943 to 1965 (1.50 t C ha1); and finally its subsequent
measured SOC) on the CO land use. With an IOM content of abandonment from 1966 with the gradual restoration of nat-
4.3 t C ha1 calculated from SOC after Falloon et al. (1998) and clay ural vegetation (4.20 t C ha1).
 R. Francaviglia et al. / Agricultural Systems 112 (2012) 48–54 51

Table 1
Input data and short-term simulation results. Data refers to year 2007.

Land use Clay (%) Soil C inputs SOC measured SOC modeled by Deviationa (%)
(t C ha1) (t C ha1) RothC (t C ha1)
Tilled vineyard, TV 11.7 1.00b 36.4 36.3 0.3
No tilled grassed vineyard, GV 11.6 1.75c 37.4 37.5 0.3
Hay crop, HC 12.8 5.99 57.7 54.6 5.4
Pasture, PA 13.6 3.68 53.7 52.5 2.2
Semi-natural systems, SN 9.2 4.20d 47.0 44.5 5.3
Cork oak forest, CO 11.4 3.74e 50.5 50.5 0.0
a
Deviation calculated as [100 * (modeled–measured)/measured].
b
TV inputs: 0 (1993–1994); 0.50 (1995–1996); 1.00 (1997 onwards).
c
GV inputs: 0 (1990–1991); 0.5 (1992–1993); 1.0 (1994–1995); 1.75 (1996 onwards).
d
SN inputs: 0 (1939–1940); 1.0 (1941–1942); 1.50(1943–1965); 4.20 (1966 onwards).
e
Model run to the equilibrium in ‘‘inverse mode’’.

To take into account sheep grazing on the pasture land (6 sheep  90 yr: 90 yr after baseline, mean climate change for the period
per hectare from December to June) and on the hay crop (3–4 2068–2097, with 710 (A2) and 555 (B2) CO2 concentrations
sheep per hectare from January until February or March) a monthly (ppm).
input of 0.1 and 0.025 t C ha1 as FYM during the grazing months
was assumed respectively. A FYM input of 0.5 t C ha1 was as-
sumed under the cork oak forest, which is normally grazed for 3. Results
1 month in January by 0.5 cattle heads per hectare; we assumed
sheep manure was identical to FYM. Inputs of carbon from sheep 3.1. Climate change and emission scenarios
and cattle manure were estimated from expert judgment in rela-
tion to the animal load, the length of the grazing period and the For the emission scenarios A2, there was a temperature increase
quality of pastured plant materials. of 0.5, 1.4 and 3.1 °C (GISS) and 0.7, 1.9 and 3.5 °C (HadCM3) in
Only the hay crop was assumed to have a fallow period (from ‘‘30–60–90 yr’’, with a steady reduction in rainfall in GISS (9.6%,
July to October), the other land uses were covered by vegetation 16.4% and 21.6%) but a considerable reduction in rainfall of more
for the whole year. The input data and the results of model simu- than 25% in ‘‘90 yr’’ for HadCM3 in comparison with ‘‘30–60 yr’’
lations for different land uses are shown in Table 1. (Table 2). For the emission scenario B2, the temperature increase
Simulations until 2007 were run with the long-term observed was 0.6, 1.0 and 2.0 °C (GISS) and 0.7, 1.4 and 2.3 °C in ‘‘30–60–
monthly data collected from the local climatic station. Following 90 yr’’; the reduction in rainfall was 8.6%, 13.7% and 18.4% in GISS,
this, the model was run for three periods of 30 yr under the differ- and 2.8%, 8.8% and 10.1% in HadCM3 (Table 2).
ent climate change conditions described below.

3.2. Baseline carbon stocks and model results

2.4. Climate change scenarios
The highest C stocks measured in 2007 (Table 1) following the
The climate change scenarios for the experimental site, derived land-use change from the native vegetation of cork oak
from the long-term data series of Monti station (40°480 N, 9°190 E, (50.5 t ha1) were observed on the hay crop HC (57.7 t ha1), the
296 m a.s.l.), were generated from the baseline climate using two pasture PA (53.7 t ha1) and the semi-natural systems SN
Global Climate Models (GCMs), the first GISS (3°  4°) of the God- (47.0 t ha1), the lowest in two vineyards (36.4 and 37.4 t ha1 in
dard Institute of Space Studies, USA (Hansen et al., 2002), and the TV and GV respectively). In comparison with the cork oak forest
second HadCM3 (2.5°  3.75°) of the Met Office Hadley Centre (UK) differences were +7.2 t C ha1 for the hay crop (HC), + 3.2 t C ha1
(Pope et al., 2000); the two Barker et al. (2007) climate scenarios for the pasture (PA), 3.5 t C ha1 for the semi-natural systems
(based on the emission scenarios) A2 (‘provincial enterprise’) and SN, 13.1 t C ha1 for the no-tilled and grassed vineyard (GV),
B2 (‘local stewardship’) were chosen. The scenario A2 – provincial and 14.1 t C ha1 for the tilled vineyard (TV).
enterprise – is characterized by strengthening regional cultural The simulations by RothC for the period 1900–2007 (Table 1)
identities, emphasis on family values and local traditions, high were in good agreement with the measured data since the percent
population growth and less concern for rapid economic develop- deviation between simulated and measured data is low, and
ment; B2 – local stewardship – is characterized by emphasis on lo- showed that the extensive agro-pastoral systems represented by
cal solutions, less rapid and more diverse technological change, the pasture and hay crop land uses maintain the carbon stocks
strong emphasis on community initiative and local, rather than when compared with the cork oak forest, with slight increases of
global solutions. 2.0 t C ha1 for the pasture (PA) and 4.1 t C ha1 for the hay crop
The simulations were projected on three time periods to simu- (HC) compared to the cork oak forest (CO). At the same time, a
late a very close, intermediate and fully realized climate change marked decrease of SOC in the two vineyards (14.2 and
scenario: 13.0 t C ha1 in TV and GV respectively) was evident, and a lower
SOC decrease (6.0 t C ha1) was observed on semi-natural sys-
 30 yr: 30 yr after baseline, mean climate change for the period tems consisting of former set-aside vineyards.
2008–2037, with 425 (A2) and 410 (B2) CO2 concentrations The highest modeled C stocks in 2007 were observed on the hay
(ppm); crop (54.6 t ha1) and the pasture (52.5 t ha1), the lowest in the
 60 yr: 60 yr after baseline, mean climate change for the period two vineyards (36.3 and 37.5 t ha1). The semi-natural systems
2038–2067, with 540 (A2) and 480 (B2) CO2 concentrations were in an intermediate position (44.5 t ha1), approaching the or-
(ppm); ganic C content of the cork oak forest (50.5 t ha1).
52 R. Francaviglia et al. / Agricultural Systems 112 (2012) 48–54

Table 2
Changes in the mean annual temperature and rainfall compared with the baseline climate (1985–2006).

Emission Years after CO2 level Temperature (+°C) Rainfall (%)

scenario baseline
GISS HadCM3 GISS HadCM3
A2 30 yr 425 0.52 0.73 9.6 2.3
60 yr 540 1.43 1.90 16.4 7.4
90 yr 710 3.06 3.54 21.6 25.5
B2 30 yr 410 0.56 0.74 8.6 2.8
60 yr 480 0.98 1.40 13.7 8.8
90 yr 555 1.98 2.25 18.4 10.1

Table 3
Simulated soil organic carbon in t C ha1 and changes in t C ha1 yr1 (in italics) in the A2 scenario and the three time periods (30 yr, 60 yr and 90 yr). TV tilled vineyard, GV no-
tilled grassed vineyard, HC hay crop, PA pasture, SN semi-natural systems (former vineyards), CO cork oak forest.

Land use 2007 GISS Land use HadCM3

30 yr 60 yr 90 yr D–A 30 yr 60 yr 90 yr G–A
A B C D E F G
TV 36.3 32.5 26.7 22.7 13.6 TV 32.3 26.2 22.7 13.6
0.13 0.16 0.15 0.13 0.17 0.15
GV 37.5 35.0 31.0 29.0 8.5 GV 34.7 30.4 29.2 8.3
0.08 0.11 0.09 0.09 0.12 0.09
HC 54.6 56.1 58.3 59.3 4.7 HC 55.2 55.4 57.9 3.3
0.05 0.06 0.05 0.02 0.01 0.04
PA 52.5 53.6 54.8 54.8 2.3 PA 53.1 53.3 54.9 2.4
0.04 0.04 0.03 0.02 0.01 0.03
SN 44.5 45.9 47.6 48.4 3.9 SN 45.5 46.8 48.7 4.2
0.05 0.05 0.04 0.03 0.04 0.05
CO 50.5 51.4 52.1 52.1 1.6 CO 50.8 50.9 52.1 1.6
0.03 0.03 0.02 0.01 0.01 0.02

Table 4
Simulated soil organic carbon in t C ha1 and changes in t C ha1 yr1 (in italics) in the B2 scenario and the three time periods (30 yr, 60 yr and 90 yr). TV tilled vineyard, GV no-
tilled grassed vineyard, HC hay crop, PA pasture, SN semi-natural systems (former vineyards), CO cork oak forest.

Land use 2007 GISS Land use HadCM3

30 yr 60 yr 90 yr D–A 30 yr 60 yr 90 yr G–A
A B C D E F G
TV 36.3 32.7 26.7 22.7 13.6 TV 32.4 26.5 23.0 13.3
0.12 0.16 0.15 0.13 0.16 0.15
GV 37.5 35.0 30.7 29.0 8.5 GV 34.3 29.7 28.0 9.5
0.08 0.11 0.09 0.11 0.13 0.10
HC 54.6 57.0 56.0 57.4 2.8 HC 55.2 56.0 58.3 3.7
0.08 0.02 0.03 0.02 0.02 0.04
PA 52.5 54.0 54.6 55.1 2.6 PA 53.2 54.0 55.4 2.9
0.05 0.04 0.03 0.02 0.02 0.03
SN 44.5 46.5 47.5 48.3 3.8 SN 45.8 47.2 48.8 4.3
0.07 0.05 0.04 0.04 0.04 0.05
CO 50.5 51.8 51.9 52.2 1.7 CO 51.1 51.5 52.6 2.1
0.04 0.02 0.02 0.02 0.02 0.02

The results of simulations under the emission scenarios A2 and 4. Discussion

B2 with the two GCMs in terms of SOC (t C ha1) and changes of
SOC (t C ha1 yr1) are shown in Tables 3 and 4. A graphical exam- The positive effect of the predicted mitigation scenario B2 com-
ple is given in Fig. 3 for the GISS A2 scenario. pared with A2 (Table 2) is particularly evident in ‘‘60–90 yr’’, since
In the more intensive land uses, i.e. tilled vineyard (TV) and no temperature changes are about one third lower in both periods and
tilled grassed vineyard (GV), a sharp SOC decrease between 8.3 and both GCMs. The corresponding decreases in rainfall are very simi-
13.6 t C ha1 was observed after 90 yr compared with the baseline lar in GISS ‘‘30–60–90 yr’’ and HadCM3 ‘‘30–60 yr’’, but more than
values in 2007, even with no changes in land management, soil halved in HadCM3 ‘‘90 yr’’. Moreover, HadCM3 provides more ex-
carbon inputs or crop yields. treme climate change projections for temperature than the GISS
In the other land uses there was always an increase of SOC com- model in both emission scenarios, and a lower decrease for rainfall
pared with the 2007 baseline. In the hay crop (HC) SOC increased in most cases.
by 2.8–4.7 t C ha1; in the pasture (PA) by 2.3–2.9 t C ha1; in the Under climate change (Tables 3 and 4), the lower decrease of
semi-natural systems (SN) by 3.8–4.2 t C ha1; and in the cork SOC in GV (8.3–9.5 t C ha1) compared to TV (13.3–13.6 t C ha1)
oak forest (CO) by 1.6–2.1 t C ha1. can be attributed to land cover, i.e. the grassed surface of the vine-
 R. Francaviglia et al. / Agricultural Systems 112 (2012) 48–54 53

ent climate change scenarios. According to Àlvaro-Fuentes and

Paustian (2011) differences in SOC sequestration were greater
among management systems than among climate change scenar-
ios. Farina et al. (2011) concluded that SOC varied more between
the crop rotations and the soil tillage practices than among the dif-
ferent climate change scenarios.
As already mentioned we assumed no technological improve-
ments for adaptation to expected climate change, or crop yield and
agricultural management changes during the simulation. Another
source of uncertainty to be considered is that CO2 concentration is
expected to increase with time, while fixed concentrations were
used for each time period in the simulations under climate change.

5. Conclusions

The RothC model is suitable for regional simulations because it

Fig. 3. Simulation of soil organic carbon dynamics under GISS A2 scenario. HC hay requires a limited number of input data represented by the follow-
crop, PA pasture, CO cork oak forest, SN semi-natural systems (former vineyards),
ing main factors that influence the soil organic carbon turnover:
TV tilled vineyard, GV no-tilled grassed vineyard.
climate (temperature, rainfall and evaporation), soil texture (clay
content) and crop management. RothC is therefore a suitable tool
yard, the return of pruning residues to the soil and no soil distur- for estimating long-term effects of land use and agricultural man-
bance from tillage. At the end of the simulation period (‘‘90 yr’’) agement, for assessing the contribution of agriculture to the stor-
TV and GV carbon stocks were 22.7–23.0 and 28.0–29.2 t C ha1 age of soil organic carbon, and also for comparing different
respectively, compared to 52.1–52.6 t C ha1 in the cork oak forest. options to preserve the organic carbon levels (e.g., stubble and crop
If compared to CO, TV and GV average soil carbon stocks were residues management, crop rotations, exogenous organic matter
about 44% and 55% respectively. inputs, etc.). Mondini et al. (2012) have demonstrated how regio-
The SOC increases in the other land uses can be attributed to var- nal simulations can be valuable for helping to plan deployment
ious causes. The SN land use, after the soil disturbance following the of scarce organic matter resources such as compost.
establishment of the vineyards, partially recovered the SOC content In these agro-silvo-pastoral systems, typical of north-eastern
of the native vegetation (CO) due to the natural restoration of Sardinia (Italy), changes in land use with respect to the potential
scrubs, meadows and other plants. After ‘‘90 yr’’ carbon stocks were native vegetation of cork oak forest affected the carbon balance
in the range 48.3–48.8 t C ha1, i.e. about 93% in comparison with significantly with a sharp decrease in the vineyards, with a lower
CO. SN land use was sensitive to climate change (+3.8–4.3 t C ha1) effect to be ascribed to climate change scenarios. This should be ta-
compared to the baseline value of 2007. The HC land use was more ken into account when planning measures for mitigation and adap-
sensitive to climate change due to the high C inputs from crop res- tation to climate change by adopting cropping systems at low
idues and FYM (+2.8–4.7 t C ha1) compared to the baseline value intensification which have proved to be more effective in main-
of 2007. After ‘‘90 yr’’ carbon stocks were in the range 57.4– taining the carbon stock. Moreover, in the study area the land de-
59.3 t C ha1, about 11% higher in comparison with CO. The PA land voted to vineyards is constantly increasing since this farming
use, receiving lower C inputs from the soil and the crop, showed a system is more profitable, and is a key economic driver for removal
smaller increase with climate change (+2.3–2.9 t C ha1). At the of forested and pasture land. Nonetheless, when vineyards are
end of the simulation period (‘‘90 yr’’) carbon stocks were in the introduced they should be managed with less disturbance to the
range 54.8–55.4 t C ha1, about 5% higher in comparison with CO. soil (i.e. no-tilled and grassed) in order to limit soil degradation
Lastly, the CO land use had a slow increase compared to the 2007 in a long term perspective.
baseline during the simulation period (+1.6–2.1 t C ha1), almost
reaching a plateau, which suggests it is reaching a new equilibrium
Acknowledgments
of SOC; alternatively an apparent increase with climate change
might be hiding a natural SOC decrease. After ‘‘90 yr’’ carbon stocks
The research is part of the Italian research project ‘‘SOILSINK’’,
in CO were in the range 52.1–52.6 t C ha1.
Climate change and agro-forestry systems: impacts on soil carbon
The land uses characterized by a low level of intensification such
sink and microbial diversity, funded by the Integrated Special Fund
as the silvo-pastoral systems, namely SN, PA, HC and CO, stored car-
for Research (FISR) of the Italian Ministry of University and Re-
bon in t C ha1 at the end of the 90 yr of simulation in the following
search. It used facilities funded by the UK Biotechnology and Biolog-
order: hay crop 58.4 > pasture 55.1 > cork oak forest 52.4 > semi-
ical Sciences Research Council (BBSRC). We greatly acknowledge
natural systems 48.6 (former vineyards). Regardless of the GCM
the soil profile description of Prof. Salvatore Madrau with the ana-
used, since the climate change effect was lower than land use, they
lytical support of Mario Antonello Deroma (Dipartimento di Agrar-
were able to maintain and slightly increase SOC stocks by 0.02–
ia, Università di Sassari). Rosa Francaviglia acknowledges the
0.05 t C ha1 yr1 from the initial content. The vineyards, in con-
financial contribution of CRA (Consiglio per la Ricerca e la Speri-
trast, had high losses of SOC, from 0.09 to 0.15 t C ha1 yr1. Our
mentazione in Agricoltura) with a research grant for her stay at
findings are in agreement with results from different simulation
Rothamsted Research. Kevin Coleman and Andrew P. Whitmore
models (e.g. Century and EPIC) clearly indicating that management
acknowledge the support of the BBSRC institute strategic pro-
choices linked to the level of crop intensification and/or conserva-
gramme grant Delivering Sustainable Systems.
tion tillage could affect the soil organic carbon balance more
strongly than climate change (Lugato and Berti, 2008; Álvaro-Fuen-
tes and Paustian, 2011; Farina et al., 2011), suggesting the adoption References
of improved rotations, grassland conversion and no-tillage.
Álvaro-Fuentes, J., Paustian, K., 2011. Potential soil carbon sequestration in a
Lugato and Berti (2008) concluded that SOC change was more semiarid Mediterranean agroecosystem under climate change: quantifying
pronounced among the management practices than among differ- management and climate effects. Plant Soil 338, 261–272.
54 R. Francaviglia et al. / Agricultural Systems 112 (2012) 48–54

Barker, T., Bashmakov, I., Bernstein, L., Bogner, J.E., Bosch, P., Dave, R., Davidson, O., Kamoni, P.T., Gicheru, P.T., Wokabi, S.M., Easter, M., Milne, E., Coleman, K., Falloon,
Fisher, B.S., Gupta, S., Halsnæs, K., Heij, B., Kahn Ribeiro, S., Kobayashi, S., Levine, P., Paustian, K., Killian, K., Kihanda, F.M., 2007. Evaluation of two soil carbon
M.D., Martino, D.L., Masera, O., Metz, B., Meyer, L., Nabuurs, G.-J., Najam, A., models using two Kenyan long term experimental datasets. Agr. Ecosyst.
Nakicenovic, N., Rogner, H.-H., Roy, J., Sathaye, J., Schock, R., Shukla, P., Sims, Environ. 122, 95–104.
R.E.H., Smith, P., Tirpak, D.A., Urge-Vorsatz, D., Zhou, D., 2007. Technical Kaonga, M.L., Coleman, K., 2008. Modelling soil organic carbon turnover in
summary. In: Metz, B., Davidson, O.R., Bosch, P.R., Dave, R., Meyer, L.A. (Eds.), improved fallows in eastern Zambia using the RothC-26.3 model. For. Ecol.
Climate Change 2007: Mitigation. Contribution of Working Group III to the Manage. 256, 1160–1166.
Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Kirschbaum, M.U.F., Carter, J.O., Grace, P.R., 2001. Brief description of several
Cambridge, United Kingdom and New York, NY, USA. ecosystem models for simulating net ecosystem exchange in Australia. In:
Cerri, C.E.P., Coleman, K., Jenkinson, D.S., Bernoux, M., Victoria, R., Cerri, C.C., 2003. Kirschbaum, M.U.F., Mueller, R. (Eds.), Net Ecosystem Exchange. Workshop
Modeling soil carbon from forest and pasture ecosystems of Amazon. Brazil. Soil Proceedings. CRC for Greenhouse Accounting, pp. 8–29.
Sci. Soc. Am. J. 67, 1879–1887. Lagomarsino, A., Benedetti, A., Marinari, S., Pompili, L., Moscatelli, M.C., Roggero,
Coleman, K., Jenkinson, D.S., 1996. RothC-26.3 – a model for the turnover of carbon P.P., Lai, R., Ledda, L., Grego, S., 2011. Soil organic C variability and microbial
in soil. In: Powlson, D.S., Smith, P., Smith, J.U. (Eds.), Evaluation of Soil Organic functions in a Mediterranean agro-forest ecosystem. Biol. Fertil. Soils 47, 283–
Matter Models Using Existing Long-Term Datasets. Springer-Verlag, Heidelberg, 291.
pp. 237–246. Lal, R., 2008. Carbon sequestration. Philos. Trans. Roy. Soc. B 363, 815–830.
Coleman, K., Jenkinson, D.S., 1999. RothC-26.3, A Model for the Turnover of Carbon Lugato, E., Berti, A., 2008. Potential carbon sequestration in a cultivated soil under
in Soil: Model Description and User’s Guide. Lawes Agricultural Trust, different climate change scenarios: a modelling approach for evaluating
Harpenden, UK. <http://www.rothamsted.ac.uk/aen/carbon/rothc.htm> promising management practices in north-east Italy. Agr. Ecosyst. Environ.
(accessed December 2011). 128, 97–103.
Falloon, P., Smith, P., Coleman, K., Marshall, S., 1998. Estimating the size of the inert Mondini, C., Coleman, K., Whitmore, A.P., 2012. Spatially explicit modelling of
organic matter pool from total soil organic carbon content for use in the changes in soil organic C in agricultural soils in Italy, 2001–2100: potential for
Rothamsted carbon model. Soil Biol. Biochem. 30, 1207–1211. compost amendment. Agr. Ecosyst. Environ. 153, 24–32.
Falloon, P., Smith, P., Szabo, J., Pasztor, L., 2002. Comparison of approaches for Nieto, O.M., Castro, J., Fernández, E., Smith, P., 2010. Simulation of soil organic
estimating carbon sequestration at the regional scale. Soil Use Manage. 18, 164– carbon stocks in a Mediterranean olive grove under different soil-management
174. systems using the RothC model. Soil Use Manage. 26, 118–125.
Farina, R., Seddaiu, G., Orsini, R., Steglich, E., Roggero, P.P., Francaviglia, R., 2011. Soil Parton, W.J., Schime, D.S., Ojima, D.S., Cole, C.V., 1994. A general model for soil
carbon dynamics and crop productivity as influenced by climate change in a organic matter dynamics: sensitivity to litter chemistry, texture and
rainfed cereal system under contrasting tillage using EPIC. Soil Till. Res. 112, management. In: Bryant, R.B., Arnold, R.W. (Eds.), Quantitative Modeling of
36–46. Soil Forming Processes. SSSA Special Publication 39, Madison, WI, pp. 147–167.
Franzluebbers, A.J., Doraiswamy, P.C., 2007. Carbon sequestration and land Pope, V.D., Gallani, M.L., Rowntree, P.R., Stratton, R.A., 2000. The impact of new
degradation. In: Sivakumar, M.V.K., Ndiangu’ui, N. (Eds.), Climate and Land physical parametrizations in the Hadley Centre climate model: HadAM3. Clim.
Degradation. Springer Verlag, New York, pp. 339–354. Dynam. 16, 123–146.
Gerik, T.J., Harman, W.L., Williams, J.R., Francis, L., Greiner, J., Magre, M., Meinardus, Rivas-Martinez, S., Rivas-Saenz, S., 2009. Worldwide Bioclimatic Classification
A., Steglich, E., 2006. Researcher’s Guide for WinEPIC, Version 3.0. Blackland System, 1996–2009. Phytosociological Research Center, Spain. <http://
Research and Extension Center, Temple, TX, 113 pp. www.globalbioclimatics.org> (accessed December 2011).
Hansen, J., Sato, M., Nazarenko, L., Ruedy, R., Lacis, A., Koch, D., Tegen, I., Hall, T., Saxton, K.E., Rawls, W.J., 2006. Soil water characteristic estimates by texture and
Shindell, D., Santer, B., Stone, P., Novakov, T., Thomason, L., Wang, R., Wang, Y., organic matter for hydrologic solutions. Soil Sci. Soc. Am. J. 70 (5), 1569–1578.
Jacob, D., Hollandsworth, S., Bishop, L., Logan, J., Thompson, A., Stolarski, R., Skjemstad, J.O., Spouncer, L.R., Cowie, B., Swift, R.S., 2004. Calibration of the
Lean, J., Willson, R., Levitus, S., Antonov, J., Rayner, N., Parker, D., Christy, J., Rothamsted organic carbon turnover model (RothC ver. 26.3), using measurable
2002. Climate forcings in Goddard Institute for Space Studies SI2000 soil organic carbon pools. Aust. J. Soil Res. 42, 79–88.
simulations. J. Geophys. Res.-Atmos. 107. http://dx.doi.org/10.1029/ Smith, P., Martino, D., Cai, Z., Gwary, D., Janzen, H., Kumar, P., McCarl, B., Ogle, S.,
2001jd001143. O’Mara, F., Rice, C., Scholes, B., Sirotenko, O., Howden, M., McAllister, T., Pan, G.,
Jenkinson, D.S., 1990. The turnover of organic-carbon and nitrogen in soil. Philos. Romanenkov, V., Schneider, U., Towprayoon, S., Wattenbach, M., Smith, J., 2008.
Trans. Roy. Soc. B 329, 361–368. Greenhouse gas mitigation in agriculture. Philos. Trans. Roy. Soc. B 363, 789–
Jenkinson, D.S., Coleman, K., 1994. Calculating the annual input of organic-matter to 813.
soil from measurements of total organic-carbon and radiocarbon. Eur. J. Soil Sci. Smith, P., Smith, J.U., Powlson, D.S., McGill, W.B., Arah, J.R.M., Chertov, O.G.,
45, 167–174. Coleman, K., Franko, U., Frolking, S., Jenkinson, D.S., Jensen, L.S., Kelly, R.H.,
Jenkinson, D.S., Harkness, D.D., Vance, E.D., Adams, D.E., Harrison, A.F., 1992. Klein-Gunnewiek, H., Komarov, A.S., Li, C., Molina, J.A.E., Mueller, T., Parton, W.J.,
Calculating net primary production and annual input of organic-matter to soil Thornley, J.H.M., Whitmore, A.P., 1997. A comparison of the performance of nine
from the amount and radiocarbon content of soil organic-matter. Soil Biol. soil organic matter models using datasets from seven long-term experiments.
Biochem. 24, 295–308. Geoderma 81, 153–225.
Jenkinson, D.S., Hart, P.B.S., Rayner, J.H., Parry, L.C., 1987. Modelling the turnover of WRB, 2006. World Reference Base for Soil ResourcesWorld Soil Resources Report
organic matter in long-term experiments at Rothamsted. INTERCOL Bull. 15, 1– 103. IUSS, ISRIC, FAO, Rome, 145 pp.
8. Yokozawa, M., Shirato, Y., Sakamoto, T., Yonemura, S., Nakai, M., Ohkura, T., 2010.
Jones, M.B., Donnelly, A., 2004. Carbon sequestration in temperate grassland Use of the RothC model to estimate the carbon sequestration potential of
ecosystems and the influence of management, climate and elevated CO2. New organic matter application in Japanese arable soils. Soil Sci. Plant Nutr. 56, 168–
Phytol. 164, 423–439. 176.