Geoderma 189–190 (2012) 81–86

Contents lists available at SciVerse ScienceDirect

Geoderma
journal homepage: www.elsevier.com/locate/geoderma

Carbon stocks and dynamics under improved tropical pasture and silvopastoral
systems in Colombian Amazonia
Octavio Mosquera a,⁎, Peter Buurman b, Bertha L. Ramirez c, Maria C. Amezquita d
a
Laboratory of Analytical Services, CIAT, Cali, Colombia
b
Earth System Science Group, Wageningen University, P.O. Box 47, 6700 AA Wageningen, The Netherlands
c
Universidad de la Amazonia, Florencia Caquetá, Colombia
d
Fundación Universitaria Católica Lumen Gentium, Cali, Colombia

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

Article history: To evaluate the effect of land use change on soil organic carbon, the carbon contents and stocks of primary forest,
Received 11 January 2011 degraded pasture, and four improved pasture systems in Colombian Amazonia were compared in a flat and a
Received in revised form 21 December 2011 sloping landscape. The improved pastures were Brachiaria humidicola, and Brachiaria decumbens, either in mono-
Accepted 29 April 2012
culture or in combination with native legumes. The age of the treatments was 30 years for degraded pasture and
Available online 5 June 2012
10 or 15 years for each of the improved pastures. Carbon fractions were Total C, Oxidizable C, and Non-Oxidizable
Keywords:
(stable) C. Stocks were compared using a fixed soil mass base. The degraded pasture in the flat landscape was
Land use change abandoned and dominated by weeds, while that in the sloping area was overgrazed. The latter had much
Soil C fractions lower C stocks than the former. B. humidicola monoculture had the highest stocks both in flat and sloping
Carbon cycling areas, while the effect of the other three treatments varied. C replacement based on δ13C indicated that after
30 years, the degraded pasture still contained more than 50% forest-derived C in its topsoil. The fraction in the
topsoil that is not replaced roughly coincides with the Stable C fraction. δ13C values suggest that the changes
in carbon stocks ascribed to differences in land use may be – at least partially – inherited from the previous
land use, thus confusing the interpretation of land use effects. Nevertheless, the introduction of improved pas-
tures on degraded grassland is a feasible alternative of land use both for carbon sequestration and as an attractive
economic alternative to farmers.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction
The present paper complements the results presented in a previous
one by the same authors dealing with carbon dynamics in soils of the
Colombian Amazonia (Mosquera et al., 2012).
The largest potential for carbon sequestration in tropical America is
in the savanna ecosystem which covers some 250 million ha, and in the
tropical forest, with some 44 million ha (Amezquita et al., 2008).
Since the studies by Lugo and Brown (1993) and Fisher et al.
(1994) it has been recognized that soils of well-managed tropical pas-
ture systems may contain amounts of soil organic carbon (SOC) equal
or even superior to those under native tropical forest.
Land use change in the Amazonia strongly influences soil organic
matter dynamics. Thus, understanding the major effects that influence
soil organic matter under cleared lands is relevant in predicting the con-
sequences of continued conversion of tropical forest to cattle pastures
⁎ Corresponding author at: Apartado Aéreo 6713, Cali, Colombia. Tel.: +57 2 4450100.
E-mail address: o.mosquera@cgiar.org (O. Mosquera).
and agriculture (Cerri et al., 2007). In addition, this is also important
to devise management technologies/strategies to enhance the sustain-
ability of these areas and thus slow down further deforestation.
An appropriate way to assess SOC changes in soils that have been
under consecutive vegetations with different photosynthetic pathway
is the relative abundance of the stable isotopes 13C and 12C, expressed
as δ13C (Balesdent and Mariotti, 1996). Plants with a C3 photosynthetic
pathway (virtually all trees, shrubs, and grasses of temperate climates)
discriminate more against 13C than those with a C4 photosynthetic
pathway (largely tropical grasses). SOC inherits the isotopic signature
of the litter, and thus reflects the standing vegetation, both above-
ground and belowground. In C3 systems there is a significant increase
of δ 13C from litter to topsoil, and therefore, calculations of replacement
by C3 humus should be based on the δ 13C value of the topsoil rather
than that of the litter (e.g., Roscoe et al., 2002).
Land use conversions in tropical savannahs or pastures are suited for
isotope studies of C dynamics (Ehleringer et al., 2000). For the Amazo-
nian systems, this is true, because 1) the forest to pasture conversion
is a C3–C4 transition; 2) the degradation of pasture causes an increase
of C3 shrubs in a C4 pasture, 3) the establishment of improved grass
monocultures is a reversion to C4, and 4) establishment of fodder
banks is a reversion to C3 vegetation.

0016-7061/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.geoderma.2012.04.022
82 O. Mosquera et al. / Geoderma 189–190 (2012) 81–86

In general, land use transition studies employing δ 13C data have
focused on δ 13Csoil in forest to pasture transitions. There have been
a few studies concerning recovery of degraded pastures, but little
work has been done with δ 13Csoil in these systems (Schedlbauer and
Kavanagh, 2008). The present work presents differences in carbon
contents and stocks related to changes in land use and tries to link
such differences to SOC dynamics.
2. Materials and methods
Areas of representative land use systems were chosen on four farms
in the Amazonian humid tropical forest, both on flat and mildly sloping
topography, at 800 m.a.s.l., 1.8° N, 75.7° W, with an annual precipitation
of 3500–4000 mm. Soils of the region are acid (pH b 5) and have low
phosphorus content and high aluminum saturation. The soils of the
flat areas are Haplic Acrisols, while those in the mild slope areas are
Haplic Ferralsols (Mannetje et al., 2008). General soil characteristics
are given in Table 1.
Blocks of different land uses (treatments) were chosen within the
farms where the history of land use had been documented. These
were “La Guajira”, a well managed commercial livestock farm
(250 ha) and “Santo Domingo” a poorly managed experimental farm
(60 ha) with degraded pastures, for the flat area, and “Los Balcanes” a
poorly managed experimental farm (60 ha) with degraded pastures
and “Pekin” a well managed commercial livestock farm (800 ha) for
the sloping landscape; the four farms were located nearby Florencia,
Caquetá, Colombia. The different land uses (treatments) were natural
forest, degraded pasture, and four improved pastures, i.e., Brachiaria
humidicola in monoculture; Brachiaria decumbens in monoculture;
B. humidicola + legume; B. decumbens + legume. The improved pastures
were all established in previously degraded pasture. The age of the var-
ious systems was different: More than 80 years for the forest, about 40
for the degraded pastures and 10–15 for the improved pastures. On
each land use system, linear transects of 1000 m were marked in the di-
rection of the predominant slope gradient. Four main sampling points
(major profile pits of 150 × 100 × 100cm) were located along the tran-
sect and eight secondary sampling points, were located on each side
of the main sampling points perpendicular to the gradient. Soil samples
were taken at four depths 0–10, 10–20, 20–40 and 40–100 cm, for a
total of 48 samples per treatment. Triplicate samples for bulk density
were taken from the same layers by the core method. Surface litter
was also sampled at each sample location.
Carbon determinations were carried out in the Analytical Services
Laboratory at CIAT (Centro Internacional de Agricultura Tropical) Cali,
Colombia. Total and Oxidizable Carbon were determined according to
Walkley–Black modified by Kurmies (Temminghoff et al., 2000). In
short, Oxidizable C was determined by oxidation with K2Cr2O7–H2SO4
at room temperature, and measurement of the resulting chromic ion
by spectrophotometer at a wavelength of 585 nm. Total C was mea-
sured using the same oxidant at a temperature of 120 °C during 2 h.
The isotope (δ13C) analyses were carried out at the University of
Davis-Isotope Facility, California. The 13C content of a sample is
expressed relative to a geological standard, the PeeDee belemnite, or
its gas equivalent, the Vienna PDB, as follows: δ 13C (per mile) =
1000 * (13Rsample − 13Rstandard) / 13Rstandard, in which 13R is the 13C/12C
ratio. As the δ13C of any soil sample is a linear mixture of the contribu-
tions of C3 and C4 vegetations, the relative contributions of each can be
calculated from δ13Csample = x * δ13CC3 + (1− x) * δ13CC4, in which x is
the fraction of C3-derived SOC, (1− x) the fraction C4-derived, and
the δ 13C values denote the typical means of the relevant C3 and C4 litter.
For a valid comparison of Carbon stocks between land uses, the
calculations were based on fixed soil mass (Amezquita et al., 2008;
Ellert et al., 2002). The results are given in t.ha − 1.meter-equivalent − 1
(m-equiv is the soil mass of the profile that had the lowest mass to
1 m depth). The formula for this calculation is as follows:

C stock ¼ 10  C0−10  BD0−10 þ C10−20  BD10−20 þ C20−40  BD20−40
þðWR −W0−40 Þ  C40−100 Þ; in which :
C stock = t.ha − 1.m-eq − 1
C =C content of specific layer (g.kg − 1),
BD =bulk density of specific layer (kg.dm − 3),
WR =weight of reference 1 m depth section (lowest of all
investigated profiles), in kg.m − 2,
W0-40 =weight of upper 40 cm of present profile in kg.m − 2.
Data were statistically analyzed by SPSS 11.5. An Anova was used
to determine statistical differences between treatments.
3. Results and discussion
3.1. C contents
Carbon contents for each of the four depth sections are presented in
Table 2. Surprisingly, in the flat area the lowest carbon contents in all
layers were found under primary forest, while in the sloping area, the
primary forest has the highest contents in all layers. It is possible that
the forest plot in the flat area is not a proper reference because it was lo-
cated far from the other plots – no forest remained close by – and
exhibited bad drainage which was not found in the other plots of the
flat area.
If degraded pasture is taken as a reference, in the flat area all im-
proved pastures but B. decumbens monoculture showed some increase
in C contents in the surface layer. The smallest effect was found under
B. decumbens + legume and the biggest under B. humidicola monocul-
ture. In the 10–20 cm layer no difference between C contents was
detected; in the 20–40 cm layer the values for B. decumbens were lower
than those for degraded pasture, while no difference was detected be-
tween B. humidicola and degraded pasture.

Table 1
Initial chemical properties of the degraded pasture soils at the two trial areas. (Means of 16 measurements).

Depth pHH2O TOCa PBrayII Exch. Al Exch. Ca Exch. Mg Exch. K B.D.b

−1 −1 −1 −1 −1 −1
(cm) g.kg mg.kg cmolc.kg cmolc.kg cmolc.kg cmolc.kg kg.dm− 3

Flat topography
0–10 4.7 25.8 3.09 3.12 0.13 0.06 0.18 1.05
10–20 4.6 17.3 0.85 n.d.c n.d. n.d. n.d. 1.22

Sloping topography
0–10 4.6 24.7 1.93 8.53 0.64 0.22 0.26 1.12
10–20 4.9 15.1 0.82 n.d.c n.d. n.d. n.d. 1.31
a
TOC = total organic carbon.
b
B.D. bulk density.
c
n.d. = not determined.
 O. Mosquera et al. / Geoderma 189–190 (2012) 81–86 83

Table 2
Total SOC contents (g.kg− 1) under different land uses on flat and sloping topography.

Depth (cm) Forest Brachiaria humidicola Brachiaria decumbens B. humidicola + legume B. decumbens + legume Degraded pasture

Flat topography
0–10 20.92 (3.55) c 30.52 (4.93) a 24.93 (3.57) b 30.36 (4.59) a 29.16 (4.61) a 24.96 (1.63) b
10–20 13.23 (1.79) c 17.87 (1.84) a 14.70 (1.46) b 17.88 (2.92) a 16.19 (1.90) b 17.30 (1.82) ab
20–40 10.91 (1.53) dc 14.69 (1.43) a 10.50 (0.95) d 13.03 (1.78) b 11.73 (1.08) c 13.87 (1.83) ab
40–100 8.13 (1.64) d 11.04 (1.15) a 9.13 (1.33) c 10.17 (1.00) b 9.08 (0.83) c 10.20 (0.93) b

Sloping topography
0–10 31.57 (12.77) a 29.92 (3.38) a 29.03 (3.40) a 29.43 (5.13) a 28.24 (2.86) a 22.64 (2.95) b
10–20 23.15 (6.59) a 17.92 (2.04) b 14.68 (1.91) c 15.21 (3.20) c 19.85 (2.04) b 13.99 (3.13) c
20–40 16.57 (2.34) a 13.06 (1.10) b 8.74 (1.36) c 8.30 (1.72) d 15.33 (1.35) a 9.92 (1.09) c
40–100 10.41(1.58) a 10.14 (1.06) a 6.43 (1.17) c 5.43 (1.30) c 11.25 (0.62) a 8.01 (1.40) b

Different letters denote statistically significant differences (p b 0.05) within one topography and within one row. (Standard deviations between brackets).

In the sloping area, all improved pastures showed a significant in- while it belongs to the lower group in the sloping area. This is mainly
crease in C, with respect to degraded pasture, in the top layer, while due to the fact that ‘degraded pasture’ is a varying concept. In the flat
both B. humidicola monoculture and B. decumbens + legume showed area, as shown by the 13C signature of the vegetation (Table 4), it de-
an increase throughout the profile. notes a pasture that is largely colonized by shrubs; while in the sloping
In both landscapes, most improved pastures showed a significant in- area it is still dominated by grasses and over-grazed.
crease of C contents in the 0–10 cm layer with respect to the degraded In the flat landscape, the primary forest had the lowest Total
grassland. Increases in C content in surface horizons after conversion to Carbon stocks, while the degraded pasture and both improvements
pasture have been often reported for the Amazon area (e.g., Neil et al., with B. humidicola had the highest. It was already mentioned that
1997; Trumbore et al., 1995). de Moraes et al. (1996) found that, in the flat land forest, because of its impeded drainage, is not a proper
the western part of Brazilian Amazonia, total soil C contents to 30 cm reference for the other flatland soils. Poorly drained soils may have
in 20-year-old pasture were 17–20% larger than in the original forest lower C contents than better drained ones close by, as also found
sites. Some studies report soil C content decline (Desjardins et al., for grassland soils invaded by shrubs in the US (Albrecht and
2004) or a conservation of the initial C content (Buschbacher et al., Kandji, 2003).
1988). Pasture management probably plays an important role in C accu- Both B. decumbens pastures had significantly lower C stocks than
mulation or loss, as reported by Trumbore et al. (1995) and Fearnside the degraded pasture. Because TC and OxC contents are highly corre-
and Barbosa (1998). lated (r 2 = 0.93–0.98), Oxidizable C showed the same pattern. The
highest stock of NOxC was found under B. humidicola monoculture.
3.2. C stocks In the sloping area, highest stocks were found under forest,
B. decumbens + legume and B. humidicola in monoculture. Stocks
Carbon stocks based on fixed soil mass are given in Table 4. Separate under both B. decumbens monoculture and B. humidicola + legume
calculations were made for Total C (TC), easily Oxidizable C (OxC) and were significantly lower but not different from that under degraded
Non-Oxidizable (stable) C (NOxC). Stocks of NOxC are obtained by sub- pasture.
tracting stocks of OxC from those of TC. NOxC stocks are between 42 and 52% of TC stocks in the flat area
In both areas there are wide differences between stocks under the and between 40 and 48% in the sloping area. The level of the NOxC
various treatments. In the flat area, stocks range from 104 to 137 and stocks increases with that of the total stocks.
in the sloping area from 94 to 153 t.ha − 1 meter-equivalent (Table 3). The various treatments do not appear to have the same effect in the
Highest stocks were found in soils where the C contents remain higher two topographies, except for the B. humidicola and B. decumbens
in the layers below 20 cm, i.e. B. humidicola, B. humidicola + legume, and monocultures, each of which had similar stocks in both topographies.
degraded pasture in the flat landscape, and forest, B. humidicola, and The forest treatment showed the lowest content in the flat area. In
B. decumbens + legume in the sloping landscape. In the flat topography, sloping areas however, forest showed the highest TC content and
the degraded pasture belongs to the group with the highest C stocks, B. humidicola + legume and B. decumbens, the lowest but similar to
that in degraded pasture. With respect to the NOxC contents, in flat
Table 3
landscape all treatments have stocks significantly higher than that in
Soil carbon stocks (t.ha− 1) of 1 m equivalent depth under various land covers. forest, with that in B. humidicola the highest. In the sloping landscape,
Recalculated from Amezquita et al. (2008). degraded pasture and B. humidicola + legume showed the lowest and
forest and B. decumbens + legume the highest.
Total C stocks Oxid C stocks Stable C stocks
Comparing the Total C stocks of the improved pastures with those
Mean SD Mean SD Mean SD
of the degraded pasture, it appears encouraging that in sloping areas
Flat forest 104.0 d 12.6 55.0 c 10.9 49.0 d 6.5 B. humidicola and B. decumbens + legume showed increases of 40 to
B. humidicola 137.2 a 10.5 73.5 a 10.4 63.6 a 5.7 50 t.ha − 1. This difference was obtained in (less than) 15 years. In
B. decumbens 114.2 c 10.0 55.1 c 9.2 59.1 b 5.3
B. hum + legume 131.1 a 12.8 76.1 a 9.8 55.0 c 5.2
both the flat and sloping area, high stocks are related to high C con-
B. dec + legume 121.6 b 8.7 64.5 b 7.9 57.0 bc 2.8 tents throughout the profile.
Deg. pasture 132.5 a 9.3 76.8 a 16.0 55.7 bc 13.3 In flat areas, improved pastures did not increase C stocks when com-
Slope forest 151.5 a 23.8 74.4 a 19.6 64.6 b 8.6 pared with degraded pasture, but this was mainly due to a vigorous
B. humidicola 141.1 a 10.9 72.0 a 9.1 58.0 c 7.7
regrowth of native shrubs in this degraded pasture.
B. decumbens 102.0 b 12.0 61.5 bc 11.4 40.4 d 5.2
B. hum + legume 93.9 b 9.5 56.1 c 5.4 37.8 d 7.9 In general, decreases of soil carbon stocks occur when forests are
B. dec + legume 153.4 a 7.5 80.2 a 9.0 73.2 a 4.9 converted to other land uses, but there are contrasting reports on
Deg. pasture 103.8 b 18.5 60.7 bc 15.2 43.1 d 15.2 the effects of forest to pasture conversion. Powers and Veldkamp
Different letters denote statistically significant differences between land uses (p b 0.05) (2005) found that mean values of soil C stocks were similar under pri-
within one topography. mary forest (80.5 Mg C ha − 1) and pasture (76.7 Mg C ha − 1) across a
84 O. Mosquera et al. / Geoderma 189–190 (2012) 81–86

Table 4
Soil δ13C values (‰) under different land use systems on flat and sloping areas.a

Flat Land use

Depth (cm) Forest Brachiaria humidicola Brachiaria decumbens B. hum + legume B. dec + legume Degraded pasture

Litter measured − 30.37 (0.48) − 13.33 (0.11) − 22.22 (2.98) − 14.37 (0.61) − 16.79 (1.34) − 29.37 (0.60)
Litter used − 30.37 − 13.33 − 13.33 − 16.80 − 16.80 − 13.96
0–10 − 28.23 (0.37) a − 18.29 (0.25) c − 22.31 (0.18) b − 17.76 (1.88) c − 18.28 (1.44) c − 21.98 (0.45) b
10–20 − 27.20 (0.58) a − 23.08 (0.68) bc − 24.21 (0.33) b − 20.10 (1.21) d − 20.64 (0.94) d − 22.79 (0.29) c
20–40 − 26.98 (0.48) a − 24.08 (0.60) c − 25.27 (0.35) b − 23.05 (0.76) d − 23.82 (0.40) cd − 23.80 (0.17) cd
40–100 − 26.78 (0.28) a − 25.07 (0.30) b − 25.50 (0.23) b − 24.20 (0.63) c − 25.05 (0.05) b − 24.91 (0.04) b

Sloping
Litter measured − 30.42 (0.27) − 12.93 (0.31) − 16.28 (0.62) − 19.23 (5.68) − 19.78 (3.61) − 13.96 (0.29)
Litter used − 30.42 − 12.93 − 13.33 − 16.80 − 16.80 − 13.96
0–10 − 27.83 (0.21) a − 17.31 (0.57) d − 20.01 (1.81) bc − 18.14 (0.87) cd − 20.93 (1.12) b − 21.43 (2.02) b
10–20 − 27.04 (0.32) a − 21.40 (0.88) bc − 21.79 (1.21) bc − 20.98 (0.61) c − 22.29 (1.65) bc − 23.17 (0.68) b
20–40 − 26.14 (0.43) a − 23.76 (0.14) c − 24.05 (0.18) bc − 23.47 (0.74) c − 24.54 (0.26) b − 24.65 (0.19) b
40–100 − 25.76 (0.11) a − 24.46 (0.07) c − 24.33 (0.40) c − 24.47 (0.42) c − 25.13 (0.20) b − 25.72 (0.13) a
a
Different letters denote statistically significant differences (p b 0.05) within one topography and within one row. (Standard deviations between brackets).

large region in Costa Rica (1400 km 2), though variation was high than C3 litter (e.g. Dijkstra et al., 2006). It cannot be ascribed to an ac-
within the region and best explained by topographic features and cumulation of relatively recalcitrant compounds, because such com-
soil mineralogy. In our case, there were differences in soil properties pounds tend to have a lower δ 13C than the less recalcitrant ones
between the flat and the sloping area (Table 1). In the sloping area, al- (Boutton and Yamasaki, 1996).
though the soils are generally more weathered than in the flat area, The litters, reflect the standing vegetation, which is mostly a mixture
both nutrient and Al contents were higher. The first is conducive to of C3 and C4 plants. The B. humidicola litter appears rather pure, as it has
biomass production, while the second helps SOC preservation in the a similar δ13C in both the flat and the sloping area. The B. decumbens lit-
soil through organo-mineral associations. Whether a soil under pas- ter shows a large variation, depending on the admixture of C3 vegeta-
ture behaves as a net C sink or will probably depend upon manage- tion. For replacement calculations, we used only the value (−13.33)
ment. Previous studies in grasslands under ‘typical’ management that appears to reflect a pure C3 vegetation. 13C signatures of the
practices indicate that pasture soils in Brazilian Amazonia behave as Brachiaria + legume mixture are necessarily quite variable. We report
a net C source, with an estimated average release of 12.0 t C.ha − 1 the measured values, but for replacement calculations we used the
(Fearnside and Barbosa, 1998). However, a study of the 0–30 cm value of −16.80, which is the mean value encountered in the B.
layer in soils in Rondônia, Brazil by Moraes (cited by Fearnside and decumbens+ legume mixture of the flat landscape. This is, of course,
Barbosa, 1998) indicated that, over the long term, there is an increase an arbitrary choice but suitable for our approach.
in the stock of soil carbon in well-managed pasture as compared to Also the variation in litter 13C of the degraded pastures was signifi-
nearby primary forest. In a soil under very well-managed pasture at cant, reflecting the difference between degraded grassy vegetation
an agricultural station near Manaus, carbon in the top 20 cm returned and regrowth of C3 shrubs. For calculations of remaining forest C in
to approximately the level of the original forest, eight years after the degraded grasslands, we used the value of −13.96 that was encoun-
clearing (90 t C/ha under forest vs. 96 t C/ha under pasture, without tered in the sloping area, because the conversion was from forest to
correction for soil compaction), with a decline of 21.4% two years grassland.
after clearing. In addition, Trumbore et al. (1995) compared C budgets The replacement calculations encompass the following comparisons:
for forest and pastures in the Eastern Amazonia, finding C stocks gains
relative to forest soil of over 20 t.ha − 1 in the first 100 cm of soil for a • Forest to degraded pasture: the amount of remaining forest-C after
rehabilitated and fertilized pasture of Brachiaria brizantha. These con- 40 years of pasture. For C4 contribution, the δ 13C value of the litter
trasting patterns are suggestive of variable soil C dynamics following of the degraded pasture of the sloping area was used.
land use change and/or the partitioning of soil C between active and • Degraded grassland to improved pasture: the influence of the new
passive pools across soil types and ecosystems. C4 contribution can only be properly measured under the monocul-
In our case, B. humidicola monoculture had the highest stocks in the tures. For contribution of the new vegetation, the values given in
flat area. C stocks for B. decumbens were significantly higher than that
in the forest; its combination with legumes had stocks as high as that
under forest in the sloping area. B. humidicola and B. decumbens+
Table 5
legume showed an increase of 37 and 50 t.ha− 1 C with respect to the de- Remaining C fraction, means and standard deviations.a
graded pasture and soil C stocks similar to that in forest.
Depth (cm) Degraded Brachiaria Brachiaria B. hum. + B. dec. +
13 pastureb humidicola decumbens legume legume
3.3. C signature and replacement
Flat landscape
0–10 0.56 (0.03) 0.58 (0.03) 1.06 (0.02) 0.19 (0.37) 0.29 (0.29)
The values of the 13C signature (Table 4) for the litters of the dif- 10–20 0.67 (0.02) 1.02 (0.07) 1.13 (0.03) 0.54 (0.20) 0.63 (0.15)
ferent land uses were typical for the kind of vegetation they represen- 20–40 0.76 (0.01) 1.05 (0.06) 1.16 (0.03) 0.92 (0.11) 1.03 (0.06)
ted: − 30.37 and −30.42 for forest in flat and sloping areas, and 40–100 0.85 (0.00) 1.00 (0.05) 1.06 (0.02) 0.92 (0.08) 1.02 (0.00)
−13.13 and − 12.93 for B. humidicola in monoculture. Values of
Sloping landscape
−30‰ and −27‰ have been found for Brazilian forest (Feigl et al.,
0–10 0.54 (0.15) 0.52 (0.07) 0.83 (0.22) 0.29 (0.19) 0.89 (0.24)
1995) and woody savannah litters (Roscoe et al., 2000). The two for- 10–20 0.70 (0.05) 0.83 (0.09) 0.86 (0.12) 0.66 (0.10) 0.86 (0.26)
est profiles behave similarly. As common in forest soils, there is a sig- 20–40 0.88 (0.12) 0.92 (0.01) 0.95 (0.02) 0.85 (0.09) 0.99 (0.03)
nificant increase in 13C from litter to topsoil, after which 13C contents 40–100 1.00 (0.01) 0.90 (0.01) 0.89 (0.03) 0.86 (0.05) 0.93 (0.02)
increase gradually with depth (Roscoe et al., 2000). This increase is a
Remaining degraded pasture fraction.
largely due to an admixture of microbial C, which has a higher δ 13C b
Remaining native forest fraction.
 O. Mosquera et al. / Geoderma 189–190 (2012) 81–86
Table 4 were used, as explained above. The results of the calcula-
tions are presented in Table 5.
After about 40 years, the degraded pasture in both topographies re-
tains about 55% of forest C in the topsoil. This is consistent with 40–50%
non oxidisable C as indicated above. The influence of the pasture de-
creases with depth, but reaches deeper in the flat than in the sloping
area. A replacement of about 50% in the topsoil in a period of
10–30 years is quite common. Similar values were found for converted
Brazilian Amazon forest (Chonè et al., 1991; Trumbore et al., 1995) and
for converted Brazilian Cerrado forest (Roscoe et al., 2000).
The replacement under the improved pastures with respect to the
degraded pasture can only be properly measured in the monocultures.
The replacement in the sloping area appears to be larger and deeper,
but this is probably due to the fact that the improved pastures in the
flat area had an age of 10 years, while in the sloping area this was
15 years. In addition, the B. decumbens pasture in the flat area had
been severely neglected. There is a significant replacement under
B. humidicola in the sloping area, but only in the topsoil in the flat area.
Similarly, the effect of B. decumbens cannot be traced in the flat area
and is rather low in the sloping area. The B. humidicola +legume pas-
tures showed significant replacement at all depths for both topographies.
In the flat area, the effect of B. decumbens+ legume is significant in the
upper two layers — and this is a minimum effect because legume-C
cannot be identified. In the sloping area, the replacement under
B. decumbens+ legume is low in the topsoil, but this may be due to a
large contribution of C4 weeds in the present vegetation (compare
Table 4).
Because C replacement in the soil under pasture is primarily linked
to deposition of root litter carbon, a relation between replacement
and changes in stock should be expected, but the results are not
unequivocal. In three cases, i.e. B. humidicola monoculture and
B. decumbens + legume in the sloping area, and B. humidicola + legume
in the flat area, considerable C replacement throughout the profile
depth coincides with high C stocks, but B. humidicola + legume in the
sloping area shows a significant replacement without the accompany-
ing high C stock, while B. humidicola in the flat area combines low re-
placement with high C stock. One complicating factor is that an
addition of new C without any replacement of old C would also result
in a calculated ‘replacement’. Such effects can only be found when C
contents are taken into account and not just the fraction of old C
remaining in the system. This, however, is not a major factor, because
higher C contents (with respect to degraded grassland) in subsoils do
not coincide with significant replacement. Comparing Tables 2, 3
and 5, we can conclude that high C stocks under B. humidicola and B.
decumbens+ legume in the sloping area, and B. humidicola in the flat
area, do not show significant replacement in the high-C subsoil layers.
We must therefore conclude that these differences were inherited
from the previous land use.
4. Conclusions
Although improved pastures tend to have carbon contents and
stocks that are superior to those in degraded pastures, the differences
are sometimes obscured by variation in the concept of degraded pas-
ture. If pastures are overgrazed, they tend to have low carbon stocks,
but if they are abandoned or undergrazed, weeds tend to establish
and carbon stocks can go up considerably.
Of the four improved systems investigated here (B. humidicola and
B. decumbens in monoculture and in combination with legumes), only
the B. humidicola monoculture increased C contents and stocks both
in the flat and the sloping landscape. The other combinations had
varying success, but the variation is partly due to different time
from establishment and different management.
Replacement calculations indicate that, after 30 years, 50% of the C
in the topsoil of the degraded pasture was still forest derived. This
85
suggests that the fraction that is here called ‘stable C’ is not affected
by this medium term replacement.
Significant increases in C stock should be reflected in a change of
the 13C signature of the Total C. Some treatments did not show such
a coincidence, and this suggests that differences in C stock ascribed
to the cultivation system may in fact reflect inherited variations in
the soil. This is a major – and so far unexposed – problem in the inter-
pretation of land use effects on C stocks. To fully understand the car-
bon dynamics in the tropical environment, not only C stocks and
replacement should be studied, but also the effect of topography, cli-
mate, soil characteristics, and differences in management, which all
influence the cycling of soil organic matter.
Acknowledgments
We are particularly grateful to the project “Research Network for the
Evaluation of Carbon Sequestration Capacity of Pasture, Agropastoral
and Silvopastoral Systems in the American Tropical Forest Ecosystem”
The Netherlands Cooperation CO-010402 which funded this research.
Furthermore special thanks to Jader Muñoz and to Hernan Giraldo for
the field work and to the staff of CIAT Analytical Services Lab for the
chemical analysis.
References
Albrecht, A., Kandji, S.T., 2003. Carbon sequestration in tropical agroforestry systems.
Agriculture, Ecosystems and Environment 99, 15–27.
Amezquita, M.C., Amezquita, E., Casasola, F., Ramirez, B.L., Giraldo, H., Gomez, M.E.,
Llanderal, T., Velásquez, J., Ibrahim, M.A., 2008. C stocks and sequestration. In:
Mannetje, L.'t, Amezquita, M.C., Buurman, P., Ibrahim, M.A. (Eds.), Carbon Seques-
tration in Tropical Grassland Ecosystems. Wageningen Academic Publishers, The
Netherlands, pp. 49–63.
Balesdent, J., Mariotti, A., 1996. Measurement of soil organic matter turnover using 13C
natural abundance. In: Boutton, T.W., Yamasaki, S. (Eds.), Mass Spectrometry of
Soils. Marcel-Dekker, New York, pp. 83–111.
Boutton, T.W., Yamasaki, S.A., 1996. Mass Spectrometry of Soils. Marcel Dekker, New
York.
Buschbacher, R., Uhl, C., Serrao, E.A.S., 1988. Abandoned pastures in eastern Amazonia.
II. Nutrients stocks in the soil and vegetation. Journal of Ecology 76, 682–699.
Cerri, C.E.P., Easter, M., Paustian, K., Killian, K., Coleman, K., Bernoux, M., Falloon, P.,
Powlson, D.S., Batjes, N.H., Milne, E., Cerri, C.C., 2007. Predicted soil organic carbon
stocks and changes in the Brazilian Amazon between 2000 and 2030. Agriculture,
Ecosystems and Environment 122, 58–72.
Chonè, T., Andreux, F., Correa, J.C., Volkoff, B., Cerri, C.C., 1991. Changes in organic matter in
an oxisol from the Central Amazonian forest during eight years as pasture, deter-
mined by 13C isotopic composition. In: Berthelin, J. (Ed.), Diversity of Environmental
Biogeochemistry. Elsevier, NY, pp. 397–405.
de Moraes, J.F.L., Volkoff, B., Cerri, C.C., Bernoux, M., 1996. Soil properties under Amazon
forests and changes due to pasture installation in Rondonia, Brazil. Geoderma 70,
63–81.
Desjardins, T., Barros, E., Sarrazin, M., Girardin, C. And, Mariotti, A., 2004. Effects of
forest conversion to pasture on soil C content and dynamics in Brazilian Amazonia.
Agriculture, Ecosystems and Environment 103, 365–373.
Dijkstra, P., Ishizu, A., Doucett, R., Hart, S.C., Schwartz, E., Menyailo, O.V., Hungate, B.A.,
2006. 13C and 15N natural abundance of the soil microbial biomass. Soil Biology and
Biochemistry 38, 3257–3266.
Ehleringer, J.R., Buchmann, N. And, Flanagan, N.B., 2000. Carbon isotopes ratios in be-
lowground carbon cycle processes. Ecological Applications 10, 412–422.
Ellert, B.H., Janzen, H.H., Entz, T., 2002. Assessment of a method to measure
temporal change in soil C storage. Soil Science Society of America Journal 66,
1787–1795.
Fearnside, P.M., Barbosa, R.I., 1998. Soil carbon changes from conversion of forest to
pasture in Brazilian Amazonia. Forest Ecology and Management 108, 147–166.
Feigl, B.J., Melillo, J., Cerri, C.C., 1995. Changes in the origin and quality of soil or-
ganic matter after pasture introduction in Rondônia (Brazil). Plant and Soil
175, 21–29.
Fisher, M.J., Rao, I.M., Ayarza, M.A., Lascano, C.E., Sanz, J.I., Thomas, R., Thomas, J., Vera, R.R.,
1994. C storage by introduced deep-rooted grasses in the South American savannas.
Nature 371, 236–238.
Lugo, A.E., Brown, S., 1993. Management of tropical soil as sinks or sources of atmo-
spheric carbon. Plant and Soil 149, 27–41.
Mannetje, L'.t, Amezquita, M.C., Buurman, P., Ibrahim, M.A. (Eds.), 2008. Carbon se-
questration in Tropical Grassland Ecosystems. Wageningen Academic Publishers,
The Netherlands.
Mosquera, O., Buurman, P., Ramierez, B.L., Amezquita, M.C., 2012. Carbon replacement
and stability changes in short-term silvo-pastoral experiments in colombian
amazonia. Geoderma 170, 56–63.
86 O. Mosquera et al. / Geoderma 189–190 (2012) 81–86
Neil, C., Melillo, J.M., Seudler, P.A., Cerri, C.C., 1997. Soil C and nitrogen stocks following
forest clearing for pasture in the southwestern Brazilian Amazon. Ecological Appli-
cations 7, 1216–1225.
Powers, J.S., Veldkamp, E., 2005. Regional variation in soil C and d13C in paired forests
and pasture of Northeastern Costa Rica. Biogeochemistry 72, 315–336.
Roscoe, R., Buurman, P., Velthorst, E.J., Pereira, J.A.A., 2000. Effects of fire on soil organic
matter in a ‘cerrado sensu-stricto’ from Southeast Brazil as revealed by changes in
δ13C. Geoderma 95, 141–160.
Roscoe, R., Buurman, P., Stein, A., Velhorst, E.J., Vasconcellos, C.A., 2002. Effects of fire
and cultivation on d13C and soil organic matter spatial variability in a cerrado
sensu stricto (Brazil). in: Roscoe, R. (Ed.) 2002. Soil organic matter dynamics in a
cerrado oxisol. PhD Thesis. Wageningen University, Wageningen, pp. 69–86.
Schedlbauer, J.L., Kavanagh, K.L., 2008. Soil carbon dynamics in a chronosequence of
secondary forests in northeastern Costa Rica. Forest Ecology and Management
255, 1326–1335.
Temminghoff, E.J.M., Houba, V.J.G., van Vark, W., Gaikhorst, G.A., 2000. Soil and Plant Anal-
ysis Part 3. Plant Analysis Procedure. Wageningen University, Wageningen. 195 pp.
Trumbore, S.E., Davidson, E.A., de Camargo, P.B., Nepstad, D.C., Martinelli, L.A., 1995.
Belowground cycling of carbon in forests and pastures of Eastern Amazonia. Global
Biogeochemical Cycles 9 (4), 515–528.