![Fig. 2. Relationship between enrichment ratio of organic C, EROC, and unit sediment discharge, qs, for the data of Polyakov and Lal (2004b) compared with Eq. [1].](/figures/fig-2-relationship-between-enrichment-ratio-of-organic-c-2b7x9jv1.webp)
50 SSSAJ: Volume 72: Number 1 • January–February 2008
Enrichment of Organic Carbon in Sediment
Transport by Interrill and Rill Erosion Processes
Soil Sci. Soc. Am. J. 72:50-55
doi:10.2136/sssaj2007.0201
Received 4 June 2007.
*Corresponding authors (donald.gabriels@UGent.be).
© Soil Science Society of America
677 S. Segoe Rd. Madison WI 53711 USA
All rights reserved. No part of this periodical may be reproduced or
transmitted in any form or by any means, electronic or mechanical,
including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher.
Permission for printing and for reprinting the material contained
herein has been obtained by the publisher.
S
oil losses by overland fl ow can remove a signifi cant por-
tion of biologically active organic matter, because a large
part of the organic matter is located on or near the soil surface.
Consequently, soil erosion decreases the chemical and physi-
cal soil fertility, affecting soil productivity (Williams, 1981).
Although the erosional loss of nutrients can be replaced by
fertilizers, the loss of organic C (OC) is not easily restored.
Furthermore, OC is more important in terms of soil physi-
cal characteristics like porosity and aggregate stability (Le
Bissonnais and Arrouays, 1997; Rhoton et al., 2002). Fullen
(1991) found that a decline in soil organic matter to less
than about 2% makes soils especially vulnerable to erosion.
Furthermore, the OC loss by erosion affects nutrient avail-
ability both directly by loss of associated nutrients and indi-
rectly by loss of cation exchange capacity. Erosion may result in
decreased biomass production, which in turn adversely affects
C storage in the soil because of the reduced quantity of OC
returned to the soil as plant residues (Gregorich et al., 1998).
Total soil loss by erosion is the result of different processes
such as deposition, splash, and interrill, rill, and gully ero-
sion. These processes can cause a preferential transport of OC.
Ghadiri and Rose (1991a) found that the outer layers of soil
aggregates contained more N, organic matter, and pesticides
than the inner core. Therefore, “stripping” of the aggregates
by raindrop impact caused an enrichment of these soil com-
ponents in the eroded material. Most research has shown that
the enrichment of OC in suspended sediment is related to a
selective transport of clay particles, indicating that the enrich-
ment mechanism is mainly controlled by the transport capacity
of overland fl ow (e.g., Sharpley, 1985; Weigand et al., 1998).
Enrichment of OC can also be caused by deposition: large
particles and aggregates settle out more quickly, resulting in a
physical enrichment of smaller particles and less dense aggregates
in runoff (Davis et al., 1983; Lu et al., 1988; Beuselinck, 1999).
The selectivity of the erosion processes is often expressed
as an enrichment ratio, ER (Massey and Jackson, 1952), cal-
culated as the ratio of the concentration of a soil constituent
in the eroded material to its concentration in the original soil.
The enrichment ratio of OC, ER
OC
, has been related to OC
content in the topsoil (Young and Onstad, 1976; Avnimelech
and McHenry, 1984) and to soil loss (Sharpley, 1985; Ghadiri
and Rose, 1991b). In general, ER
OC
values in suspended sedi-
ment have been found to be >1. For a sandy loam soil, Cogle
et al. (2002) reported ER
OC
values between 1.3 and 4.1, with
larger variations in ER
OC
at low sediment yields. Jacinthe et
al. (2002) conducted laboratory rainfall simulations on small
soil pans with undisturbed samples from fi elds under different
tillage techniques. For no-till, moldboard plow, and chisel till,
they reported ER
OC
values of 1.4, 1.7, and 2.3 at sediment
delivery rates of 17.4, 34.1, and 48.9 g m
−2
h
−1
, respectively.
This indicates an increase in the ER
OC
with increasing soil loss
rate, which is rather unusual and may be a result of the differ-
ence in tillage technique; however, the results of Cogle et al.
(2002) indicated that the kind of tillage technique affected the
sediment yields but did not directly infl uence the ER
OC
. This
was also observed by Basic et al. (2002) and Kisic et al. (2002)
for enrichment of clay and OC, respectively. On the other
hand, Fullen (1991) and Fullen et al. (1996) found lower OC,
clay, and silt concentrations in eroded material compared with
the loamy sand topsoil and attributed it to eluviation of clay
and organic matter into the topsoil during runoff events. This
in-washing process resulted in a “fi ltration pavement” immedi-
ately beneath the surface crust.
W. Schiettecatte
D. Gabriels*
W. M. Cornelis
G. Hofman
Department of Soil Management and Soil Care
Ghent Univ.
Coupure Links 653
B-9000 Ghent
Belgium
Erosion and loss of organic carbon (OC) result in degradation of the soil surface. Rill and
interrill erosion processes on a silt loam soil were examined in laboratory rainfall and fl ume
experiments. These experiments showed that rill and interrill erosion processes have contrast-
ing impacts on enrichment of OC in transported sediment. Rill erosion was found to be
nonselective, while for interrill erosion the enrichment ratio of OC, ER
OC
, varied between
0.9 and 2.6 and was inversely related to the unit sediment discharge. At unit sediment dis-
charge values >0.0017 kg s
−1
m
−1
, the ER
OC
remained equal to 1. The enrichment process
was not infl uenced by raindrop impact. Enrichment of OC by “aggregate stripping” was
found to be unimportant in our study. This was attributed to the low aggregate stability of
the soil and the equal distribution of OC within the different soil aggregate classes.
Abbreviations: Ek, kinetic energy of the rain on the soil surface; ER, enrichment ratio; OC, organic
carbon; qs, unit sediment discharge.
SOIL PHYSICS
SSSAJ: Volume 72: Number 1 • January –February 2008 51
The enrichment in OC is also affected by the storm
intensity. The results of Jacinthe et al. (2004) suggest a more
selective detachment and sorting of labile OC fractions during
low-intensity storms than high-intensity storms. They stated
that the high soil losses associated with intense rainstorms can
determine the overall impact of erosional events on losses of
labile OC. On the other hand, fi eld plot experiments con-
ducted by Quinton et al. (2001) showed that the enrichment
of sediment during small events should not be neglected in
estimating the loss of soil constituents by erosion.
The cited research results indicate that, in general, sus-
pended sediment is enriched in OC. To what extent differ-
ent erosion processes account for this enrichment mechanism,
however, is still unclear. This knowledge is required for model-
ing OC losses. Polyakov and Lal (2004a) concluded in their
review that only a few soil organic matter models incorporate
an erosion component in their routine. These erosion com-
ponents are based on empirical models like the Universal Soil
Loss Equation, and are mostly unable to represent erosion-
induced soil organic matter loss and deposition as spatially and
temporally distributed phenomena. Erosion and deposition
processes are dependent on spatially variable factors. Therefore,
assessment of total soil loss in a lumped way does not take into
account the spatial variation in enrichment processes. Moreover,
many models use constant values for ER or (usually logarithmic)
equations that relate ER to average soil loss during a rainfall event
or a whole year. In this way, the effect of temporal variations in
erosion intensity on the ER is not taken into account.
The purpose of this study was to examine the enrichment
of OC caused by rill and interrill erosion processes simulated
in the laboratory. Flume experiments and rainfall simula-
tions were conducted on small soil pans to fi nd a relationship
between OC losses and erosion. Based on these experiments,
also the infl uence of rainfall impact and interrill and rill ero-
sion on ER
OC
was investigated. In this study, we only consid-
ered suspended material in runoff. Therefore, the term sedi-
ment refers to suspended sediment and not deposited material.
MATERIALS AND METHODS
General Setup
For all the laboratory experiments, topsoil samples (0- to 0.05-m
depth) were taken from fi eld plots located on a silt loam soil at
Maarkedal, Belgium (50°46′29″ N, 3°35′11″ E). The fi eld was mold-
board plowed to the depth of 0.25 m, followed by harrowing before
planting. The particle size distribution of the soil obtained using full-
dispersion and minimal-dispersion techniques is given in Table 1. The
soil has a pH (KCl) (1:2.5 w/w) of 5.4 and does not contain CaCO
3
.
The OC content is 11 g kg
−1
.
Interrill and rill erosion have different driving forces: in interrill
erosion, the interaction of rainfall impact and overland fl ow is impor-
tant, while the main agent in rill erosion is overland fl ow. Therefore,
a different setup was used for the two experiments: interrill erosion
was examined by rainfall simulations, while rill erosion was mainly
investigated by overland fl ow experiments.
Interrill Erosion Experiments
The soil was air dried and ground (<8 mm) before use in the
experiments. The sieved soil was spread out equally in soil pans (0.55
m long, 0.20 m wide, and 0.035 m deep) and slightly compacted with
a wooden plate to obtain a bulk density of 1.33 Mg m
−3
, equaling
the conditions of the fi eld plots at Maarkedal. Each soil pan had a
perforated bottom, allowing free drainage of percolating water. The
soil pans were set at slopes of 10 and 30%.
To examine the infl uence of rainfall kinetic energy on the ER,
a plastic cover was applied: small plastic sheets (20 by 20 mm) were
placed on wooden toothpicks, which were put into the soil according
to an alternating grid based pattern, leaving 10-mm space between
the plastic cover and the soil surface. The use of plastic sheets with
known dimensions allowed an accurate measurement of the covered
area and better resembled the natural cover of germinating plants
compared to mesh screens. The following surface covers were applied:
0, 0.25, 0.40, 0.50, 0.75, and 0.9 m
2
m
−2
. The kinetic energy of the
rain on the soil surface, Ek (J m
−2
s
−1
), was calculated by multiplying
the kinetic energy of the rain (J m
−2
s
−1
) by the fraction of uncovered
soil surface (m
2
m
−2
).
The experiments were done using two different kinds of rainfall
simulators to have a wider range of rainfall intensities and rainfall
kinetic energies. In the fi rst series of experiments, a rainfall simula-
tor with capillary tubes was used. The water level above the capillary
tubes determined the rainfall intensity. Two different water levels were
used, resulting in intensities of 25 ± 5 and 40 ± 5 mm h
−1
, respec-
tively. The diameter of the raindrops was 4.92 ± 0.12 mm and their
fall height was 2.75 m. The nomograph constructed by Laws (1941)
was used to estimate the terminal velocity of the drops. The calcula-
tion of the rainfall kinetic energy was based on the normal velocity of
raindrop impact. The rainfall intensities of 25 and 40 mm h
−1
pro-
vided rainfall kinetic energies of 0.15 and 0.23 J m
−2
s
−1
, respectively.
More details about the rainfall simulator were given by Gabriels et
al. (1973). Rainfall intensities were kept constant during the experi-
ments, which lasted for 100 min. The fi rst runoff sample was taken
after 20 min; afterward, sampling was done every 10 min. Runoff
volume was measured and analyzed for sediment content. During the
second half of the experiment, a steady runoff and sediment discharge
was obtained. Rainfall simulations at 40 mm h
−1
indicated that the
ER of the suspended sediment reached a steady state when sediment
discharge became constant (Schiettecatte, 2006). Therefore, only the
runoff sampled at the end of the experiment was used to analyze the
OC content in the sediment.
To provide higher rainfall intensities (and consequently higher
runoff discharges), the rainfall simulator in the ICE wind tun-
nel (International Centre for Eremology, Ghent University, Ghent,
Belgium) was used. A detailed description of the wind tunnel was
given by Gabriels et al. (1997) and Cornelis et al. (2004a). Instead of
using capillary tubes, the rainfall was simulated by pressurized nozzles
(type TeeJet TG SS 14w), fi xed at a height of 1.8 m. A nozzle pres-
sure of 1.5 × 10
5
Pa was applied. The kinetic energy was determined
Table 1. Fully dispersed and minimally dispersed particle size
distribution of the silt loam soil used in the experiments.
Fraction Fully dispersed Minimally dispersed
µm ————————— g kg
−1
—————————
0–2 170 74
2–10 48 108
10–20 112 111
20–50 380 437
50–100 135 110
100–200 128 128
200–500 20 24
500–2000 7 8
52 SSSAJ: Volume 72: Number 1 • January –February 2008
indirectly using the splash-cup technique (Ellison, 1947; Salles and
Poesen, 2000; Salles et al., 2000; Cornelis et al., 2004b,c). Because
the median diameter of the raindrops produced by the nozzles was
only 1 mm, the kinetic energy of the raindrops was much lower than
the simulator with the capillary tubes. The rainfall simulation experi-
ments lasted for 30 min, with a rainfall intensity of 70 mm h
−1
and a
rainfall kinetic energy of 0.035 J m
−2
s
−1
for conditions in which one
nozzle was used. When two nozzles were used (at 1-m distance from
each other), an intensity of 120 mm h
−1
with a rainfall kinetic energy
of 0.079 J m
−2
s
−1
was applied for 20 min. Every 2 min, the runoff
volume was measured and analyzed for sediment content. The runoff
sampled at the end of the experiment was used to determine the OC
content in the sediment.
The combination of two slopes, six different covers, and four
different intensities resulted in a total of 48 experiments.
Rill Erosion Experiments
Rill Experiment 1: Addition of Water to a Preformed
Erodible Gully
A soil pan (0.96 m long, 0.56 m wide, and 0.20 m deep) was
fi lled with air-dried soil aggregates (<8 mm) to obtain a bulk density
of 1.33 Mg m
−3
. An artifi cial V-shaped rill was made with slopes of
30% at both sides. The soil pan was placed at slopes of 10, 20, and
30%. Sediment-free tap water was added at a constant discharge of
0.01 and 0.017 L s
−1
at the upper edge of the rill. Runoff volume
was measured and analyzed for sediment content every 60 s for the
low discharge and every 30 s for the high discharge. At the end of the
experiment, one more runoff sample was taken to determine OC con-
tent. Every experiment lasted for 5 or 10 min depending on the high
or low runoff discharge. At the end of the experiments, the rill width
was measured for calculation of the sediment discharge per unit width.
Rill Experiment 2: Addition of Water and Sediment to
a Preformed Erodible Gully
A similar setup (same slopes and soil pan confi guration) was
also used with dry soil aggregates (<2 mm) added to the water infl ow,
resulting in an average input sediment concentration of 163 g L
−1
.
Sampling and measurements were done following the same procedure
as described above.
Rill Experiment 3: Rainfall Simulation on a Preformed
Erodible Gully
An analogous experiment was done using the same soil pan
confi guration, but neither water nor sediment was added to the rill.
Instead, rainfall simulations were done in the ICE wind tunnel, using
intensities of 100 mm h
−1
. Kinetic energy of the rainfall was mea-
sured using the splash-cup technique. The soil pan was set at slopes
of 10, 20, and 30%. The rainfall simulations lasted for 30 min and
runoff was sampled every 30 s. At the end of the experiment, one
more runoff sample was taken to determine OC content. At the end
of the experiments, the rill width was measured for calculation of sediment
discharge per unit width.
Analysis of Samples and Data
The sediment concentration of the runoff samples was deter-
mined by drying (at 105°C), using tared recipients. The OC con-
tent was determined using the method of Walkley and Black (1934).
Because this method yields, on average, 75% of the total OC, the
results of the analysis were divided by 0.75 to obtain the total OC
content (De Leenheer and Van Hove, 1958). The fully dispersed par-
ticle-size analysis was obtained using the sieve pipette method (De
Leenheer, 1966). The minimally dispersed particle-size distribution was
obtained in a similar way without chemically dispersing the samples.
Aggregate stability was determined by wet sieving according to
the procedure of De Leenheer and De Boodt (1959), using a fi xed
aggregate-size distribution with a mean weight diameter (MWD) of
4.45 mm. Briefl y, the soil aggregates were prewetted by raindrops and
incubated for 24 h. After wet sieving for 5 min, the MWD of the
resulting aggregate size distribution was determined. The stability
index was calculated as the reciprocal value of the difference between
the original MWD (4.45 mm) and the MWD of the wet-sieved soil
sample. After wet sieving and drying (at 105°C), the OC content
of the different aggregate size fractions (0.3–0.5, 0.5–1.0, 1.0–2.0,
2.0–2.8, 2.8–4.8, and 4.8–8.0 mm) was determined.
Unless stated otherwise, the ER of a soil constituent was calcu-
lated by dividing the content of the constituent in the suspended sedi-
ment by its content in the original soil material. Analysis of variance
and regression analysis were done using the statistical software SPSS
Version 11.0.1 (SPSS, 2001).
RESULTS AND DISCUSSION
Interrill Erosion
In Fig. 1, the ER
OC
is shown as a
function of the unit sediment discharge,
qs, measured during the last sampling
interval of both the interrill and rill ero-
sion experiments. For the interrill erosion
experiments, the ER
OC
values varied
between 0.93 and 2.61, with an average
value of 1.53, while qs ranged between
6.8 and 790 mg s
−1
m
−1
, with an average
of 140 mg s
−1
m
−1
. It can be observed
that the ER
OC
decreased with an increase
in qs. Higher sediment transport rates
will decrease the selectivity of the erosion
process, as larger aggregates and particles
are also transported. This is supported by
fi ndings of Olarieta et al. (1999), who
observed that on steep slopes the eroded
Fig. 1. The enrichment ratio of organic C, ER
OC
, as a function of unit sediment discharge,
qs, for different rainfall kinetic energies, Ek (J m
−2
s
−1
), compared with Eq. [3], the
relationship given by Sharpley (1985).
SSSAJ: Volume 72: Number 1 • January –February 2008 53
material was not signifi cantly enriched in organic matter or
nutrients due to the high erosive power of the runoff.
When correlating ER
OC
to qs, we found a Pearson r value
of −0.425, which was signifi cant at the 0.01 level. Apart from
qs, raindrop impact is expected to be the most important vari-
able that might control the enrichment mechanism. For ER
OC
vs. Ek, however, a Pearson r of only −0.028 was found, which
was not signifi cant at the 0.01 level. This was further con-
fi rmed by a stepwise linear regression on log-transformed data
showing that only qs had a signifi cant infl uence on the ER
OC
.
The hence obtained regression equation, which was signifi cant
at the 0.01 level despite the rather low r value (r = 0.574), was
ER
OC
= 0.455 qs
-0.123
[1]
where ER
OC
is the enrichment ratio of OC and qs is unit sedi-
ment discharge (kg s
−1
m
−1
).
In conducting laboratory rainfall simulations on small soil
pans (0.3 by 0.45 m) but with a rainfall intensity of 30 mm h
−1
,
Jacinthe et al. (2002) found, for undisturbed samples from
moldboard-plowed plots, ER
OC
values close to 1.7 for qs rates
of 3.7 and 4.9 mg s
−1
m
−1
. Although these qs values are outside
the range of our observations, the ER
OC
values measured by
Jacinthe et al. (2002) seem to be slightly lower than expected
based on our observations.
Several (usually logarithmic) relationships have been given
to predict ER values from soil loss data (e.g., Menzel, 1980).
These are diffi cult to compare with Eq. [1] and the results in
Fig. 1, however, as soil loss is usually expressed in kilograms
per hectare. Rescaling soil losses from experimental plot areas
to hectares may produce different results because of differences
in plot width. Moreover, soil loss expressed in kilograms per
hectare does not indicate the intensity of the erosion process
because no time component is included. Low-intensity rainfall
for a long period may cause as much erosion as a short, intense
rainfall event, but enrichment processes will be different in the
two cases. Weigand et al. (1998) found that for the same soil
loss of 60 kg ha
−1
, the ER values of
137
Cs varied between 0.51
and 2.46. Therefore, the applicability of an ER in models that
assess soil loss on a long-term (annual) basis can be
questioned, especially if the runoff events for a longer
period show large variations in erosion intensities.
The reason for using unit sediment discharge,
qs (kg s
−1
m
−1
), which is the sediment transport
per unit width and unit time, rather than soil loss
(kg ha
−1
), in Eq. [1] is that it is a better expression
for “erosion intensity” and can also be measured eas-
ily during experiments. Furthermore, qs is commonly
used in sediment transport equations, relating runoff
to erosion (e.g., Nearing et al., 1997). Equation [1]
enables linking nutrient losses to runoff and sediment
transport, which is of particular interest in the case of
event-based modeling.
Despite the diffi culties in comparing data due to
the different expressions of soil loss, we verifi ed our
results with a relationship given by Sharpley (1985),
who related the ER
OC
to soil loss (kg ha
−1
) for different
soil types. For a Durant soil, which has similar textural
properties as the soil in our study, he found
ER
OC
= 8.943 (soil loss)
-0.317
[2]
To verify our results with Eq. [2], the soil loss value in Eq. [2]
was converted into unit sediment discharge, qs (kg s
−1
m
−1
),
taking into account the experimental setup used by Sharpley
(1985), i.e., 60-min rainfall simulations on soil pans of 1-m
length and 0.3-m width. Equation [2] can hence be rewritten as
ER
OC
= 0.036 qs
-0.317
[3]
As illustrated in Fig. 1, Eq. [3] predicts lower ER
OC
values
for our measurement range than Eq. [1]. Sharpley (1985)
worked with pans with a lower slope, on slowly prewetted
samples, which may partly explain the difference in results.
Because slaking does not occur in slowly prewetted samples,
the soil used by Sharpley (1985) was less susceptible to com-
plete aggregate breakdown than dry samples directly exposed
to rain as reported in this study. This is also refl ected in the
differences in sediment discharge between Sharpley’s study and
ours. Sharpley (1985) observed soil losses ranging between 10
and 800 kg ha
−1
, or 0.278 and 24 mg m
−1
s
−1
, respectively,
when converted to unit sediment discharge. In our experi-
ments, the unit sediment discharge values ranged from 6.8
to 790 mg m
−1
s
−1
. These fi ndings indicate that relationships
between ER and soil loss may vary not only for various soil
types but also across different soil loss ranges.
To further validate our fi ndings, we used data published
by Polyakov and Lal (2004b). They conducted rainfall simula-
tions on small soil pans (0.3 m wide and 1 m long) fi lled with
sieved (<8-mm) silt loam soil. Four soil pans were positioned
in a cascade system in which runoff from the upper soil pan
fl owed into the lower pan. A subsample of the runoff from
each soil pan was taken at regular time intervals and analyzed
for sediment and OC content. The subsequent pans were put
under different slopes (8, 7, 1, and 0.5%), resembling a con-
cave hillslope. This setup resulted in erosion on the 8 and 7%
slopes, of which the sediment was partly deposited on the 1%
slope. Because deposition is a different enrichment process
Fig. 2. Relationship between enrichment ratio of organic C, ER
OC
, and unit
sediment discharge, qs, for the data of Polyakov and Lal (2004b) com-
pared with Eq. [1].
54 SSSAJ: Volume 72: Number 1 • January –February 2008
than interrill erosion, we only used the data from the 8 and 7%
slopes to validate our fi ndings. As shown in Fig. 2, the obser-
vations of Polyakov and Lal (2004b) agree well with ours: the
best-fi t power function for the data in Fig. 2 closely resembles
Eq. [1]. Polyakov and Lal (2004b) further found that the ER
OC
decreased with rainfall duration. They attributed the higher
ER
OC
values during the initial stage of the rainfall experiment
to a fl ushing of loose organic particles and fi ne soil fractions.
Based on Fig. 2, we may conclude that the higher ER
OC
values
are a result of the lower qs values. Due to the lower transport
capacity of the runoff, a selective transport of small particles
rich in OC occurs.
Not only in laboratory experiments, but also at the fi eld
scale, several researchers have revealed a relationship between
erosion rate and ER
OC
. Using fi eld rainfall simulations on silt
loam and loamy soils, Young et al. (1986) observed an increase
in ER
OC
with decreasing soil loss: ER
OC
values varied between
1.41 and 2.21 for a soil loss range of 7.32 to 20.58 Mg ha
−1
.
Lowrance and Williams (1988) also conducted fi eld rainfall
simulations and found that the suspended sediment under
four-row peanut (Arachis hypogaea L.) cover was more enriched
in OC than the sediment from a continuous fallow plot. This
may be attributed to the lower soil losses from the covered
plots, being less than half of the soil losses from the bare plots.
Andreu et al. (1994) measured soil and nutrient losses on fi eld
plots under natural rainfall and observed that, with respect to
the texture of the suspended sediment, all plots showed enrich-
ment in clay. This phenomenon was more obvious for the plot
without vegetation, which would indicate that cover may infl u-
ence the ER. On the other hand, the enrichment in OC was,
on average, lower on the bare plot than the vegetated plots,
which was attributed to the accumulation of plant biomass on
the vegetated plots. Furthermore, the ER may be infl uenced by
the erosion intensity, because soil losses on the vegetated plots
were always lower than on the bare plot. Consequently, the
estimation of ER values from large fi eld plots and watersheds
is diffi cult due to the interaction of different enrichment pro-
cesses and the larger within-fi eld variability of OC.
Ghadiri and Rose (1991a) found that “stripping” of soil
aggregates was the main mechanism in nutrient enrichment
of eroded particles in the case of a well-aggregated soil. This
was also confi rmed by Wan and El-Swaify (1998) in the case
of a well-aggregated Oxisol. The soil used in our experiments
has a stability index of 0.36, indicating a low aggregate sta-
bility. This is also confi rmed by the particle size data (Table
1) showing a similar distribution for minimally dispersed and
fully dispersed samples, except for the clay fraction, which
was mainly aggregated to 2- to 10-µm sizes in the case of the
samples that were minimally dispersed. The use of air-dried
soil in the experiments may have enhanced the rapid break-
down of the aggregates, as dry aggregates are more susceptible
to slaking and breakdown by rainfall impact than prewetted
aggregates. Therefore “stripping” is considered to be an unim-
portant enrichment mechanism for this soil. If “stripping”
would have been important, the ER
OC
should decrease with
increasing Ek because of a more complete breakdown of the
aggregates. This trend could not be observed in our results (Fig.
1), however, and Ek was found to be an insignifi cant variable
in the stepwise regression analysis. Analysis of the OC content
in the different aggregate fractions obtained after wet sieving
did not show a signifi cant increase in OC (at the 0.05 level)
with decreasing aggregate size (Table 2). This indicates a rather
homogeneous distribution of OC within the aggregates. This
was also observed by Schiettecatte (2006) for the OC content
and textural composition of different aggregate classes, which
had been separated by dry sieving the same soil used in our
study. Therefore, we can conclude that for the soil under study,
detachment by raindrop impact is an unselective process. This
agrees with the results of Gabriels and Moldenhauer (1978),
who found that for different soil types (fi ne sand, silt loam,
silty clay loam, and silty clay), the particle size distribution of
detached material by splash erosion was not signifi cantly dif-
ferent from that of the original soil.
Rill Erosion
The ER
OC
as a function of qs for the rill erosion experi-
ments is given in Fig. 1 as well. The ER
OC
values ranged
between 0.80 and 1.07 with an average of 0.95, while qs
ranged between 0.002 and 0.224 kg s
−1
m
−1
, with an average
of 0.07 kg s
−1
m
−1
. Linear regression on log-transformed val-
ues of ER
OC
vs. qs for rill erosion showed that both the slope
and intercept were not signifi cantly different from 0 (at the
0.01 level), which indicates that the ER
OC
equals 1 regard-
less of the sediment discharge. This agrees with the general
statement that interrill erosion causes selective transport of
fi ner sediment particles and nutrients, while no selectivity is
observed in sediment transport by rill erosion (Meyer et al.,
1975; Alberts et al., 1980; Proffi tt and Rose, 1991). When
extrapolating Eq. [1] beyond the measured qs range of the
interrill erosion experiments, ER
OC
becomes equal to 1 at a qs
value of 1700 mg s
−1
m
−1
. This value corresponds to the lowest
qs values observed during the rill erosion experiments. For qs >
1700 mg s
−1
m
−1
, Eq. [1] does not hold and ER
OC
= 1.
The results given in Fig. 1 further indicate that detach-
ment by overland fl ow in rills is not infl uenced by raindrop
impact. This is in agreement with the results of the interrill
erosion experiments.
CONCLUSIONS
Laboratory experiments showed that enrichment of OC
in suspended sediment is dependent on the erosion process.
Rill erosion was found to be unselective, while ER
OC
values
for interrill erosion decreased from 2.6 toward 1 as the ero-
sion rate increased. At qs values >1700 mg s
−1
m
−1
, the ER
OC
remained equal to 1. The enrichment process was not infl u-
enced by raindrop impact or “aggregate stripping.” The trend
in ER
OC
values indicated that the intensity of the erosion pro-
Table 2. Concentration of organic C in different aggregate
size classes obtained after wet sieving (n = 5; values in
parentheses indicate standard deviations).
Aggregate class Organic C
mm g kg
−1
0.3–0.5 12 (1)
0.5–1.0 11 (1)
1.0–2.0 10 (1)
2.0–2.8 11 (1)
2.8–4.8 10 (1)
4.8–8.0 10 (1)