Reassessment of tillage erosion rates by manual tillage
on steep slopes in northern Thailand$
F. Turkelbooma,b, J. Poesenc,*, I. Ohlera,b, S. Ongprasertd
a
Soil Fertility Conservation Project (SF0),1 Mae Jo University (MJU), Chiang Mai, Thailand
Laboratory for Soil Fertility and Soil Biology, Catholic University Leuven, K.U. Leuven, Belgium
c
Fund for Scienti®c Research Flanders, Laboratory for Experimental Geomorphology, K.U. Leuven, Belgium
d
Department of Soils and Fertilizers, Mae Jo University, Chiang Mai, Thailand
b
Abstract
Changing land-use practices in northern Thailand have increased tillage intensity. This study re-assesses the rate of tillage
erosion by manual hoeing on steep slopes (17±82%) in northern Thailand. Previously collected soil translocation data during
an on-farm tillage erosion experiment and additionally collected data during an on-farm tillage erosion survey have been
analysed whereby a new calculation method (i.e. trapezoid tillage step) has been used. A comparison with previously collected
data indicates that the trapezoid tillage step method and the tracer method are the most reliable methods to assess downslope
translocation by manual tillage. Based on newly acquired understanding of the processes involved, soil ¯uxes by tillage
erosion are quanti®ed by linear functions for different slope gradient classes rather than one single diffusion-type equation for
the whole slope range. For slope gradients smaller than 3%, soil ¯uxes are close to zero as farmers do not have a preferred
tillage direction. For slope gradients between 3% and 70%, soil is tilled only in the downslope direction and soil ¯uxes range
between 16 and 67 kg mÿ1 tillage passÿ1. On slopes with gradients in excess of 70%, the angle of repose for soil clods is often
exceeded resulting in a sliding down of the complete tilled top layer. These data are used to assess the soil ¯ux for complete
cropping cycles for the most dominant cropping systems in the highlands of northern Thailand: i.e. upland rice, maize, (soy)
beans, cabbage and ginger. The on-site effects of tillage erosion will be very pronounced if parcels are short with respect to
their slope length, cultivated for upland rice or cabbage, or when weed pressure is high. Tillage erosion results in a tillage step
with low soil fertility and low in®ltration capacity. Solutions to reduce tillage erosion intensity depend on the degree that
tillage intensity can be reduced. This might happen by an improved weed management or by changing landuse to perrenial
cropping. Other strategies are concentrating nutrients on the truncated hillslope sections and retaining soil on the ®eld by
vegetative buffers. # 1999 Elsevier Science B.V. All rights reserved.
Keywords: Manual tillage; Tillage erosion; Soil ¯ux; Soil-landscape variability; Steep slopes; Northern Thailand; Angle of repose; Soil
conservation
$
Paper presented at International Symposium on Tillage Translocation and Tillage Erosion held in conjunction with the 52nd Annual
Conference of the Soil and Water Conservation Society, Toronto, Canada, 24±25 July 1997.
*
Corresponding author. Tel.: +32-1632-6425; fax: +32-16-326400
E-mail address: [email protected] (J. Poesen)
1
A co-operative project between Department of Soils and Fertilizers.
0167-1987/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 7 - 1 9 8 7 ( 9 9 ) 0 0 0 4 1 - 0
1. Introduction
Traditionally, hilltribe farmers in northern Thailand
did not till their ®elds, because long fallow periods and
subsequent buming guaranteed weed-free soils with a
good structure (Hinton, 1970; Grandstaff, 1980; Collins et al., 1991). Nowadays, many highland farmers
started to till their land in response to increasing weed
pressure (Grandstaff, 1980). In our case study village,
Pakha (Chiang Rai province, northern Thailand), tillage became widespread since 1989 (Fig. 1). Deep
tillage (or rough tillage; mean clod size equals 10±
20 cm), during the dry season is an effective way to
control perennial grasses, to retard the emergence of
annual broadleaved weeds (Van Keer et al., 1995), to
break a sealed top layer and to enhance soil fertility.
Farmers traditionally start to till from the bottom of
the ®eld and gradually moving up the slope. The tilled
soil itself is always pulled downslope. In upland rice
and cabbage ®elds the large clods are broken and the
surface is levelled (®ne tillage, mean clod size equals
2±5 cm) in order to prepare a ®ne seedbed. To kill reemerging weeds, upland rice ®elds are cleaned with a
hoe once or twice more before planting (stale seedbed
method).
Soil tillage is considered as a factor affecting the
soil's sensitivity to erosion by water: tillage has several short- and long-term effects, which may to a
certain extent counteract each other. It breaks up
the (crusted) soil surface and dramatically increases
macroporosity and soil surface roughness, thereby
increasing the surface depression storage as well as
the in®ltration capacity of the soil. However, tillage
also decreases the soil's resistance to detachment by
raindrop impact or ¯owing water (Govers et al., 1994)
and when rainfall exceeds the surface depression
storage, tillage can lead to concentrated ¯ow erosion.
An often overlooked process contributing to soil
loss is the direct effect of tillage on soil movement.
This process recently became better documented, but
is still studied on a small scale. The limited attention
probably originates from the slow and inconspicuous
nature of tillage erosion, compared to the more spectacular water erosion. Tillage erosion (also referred to
as `arable erosion' by Zachar (1982), or `mechanical
erosion' by Revel et al. (1989)) is the process of soil
movement caused by the force applied by agricultural
tools and by gravity (Roose, 1994). Redistribution of
soil by tillage has been described by a diffusion
equation (Eq. (1); Govers et al., 1994):
Qt ˆ kS;
(1)
ÿ1
where Qt is the soil flux caused by tillage (kg tillage
passÿ1); k the diffusion coefficient (kg mÿ1 tillage
passÿ1); and S the slope gradient (%) of the soil
surface prior to tillage.
Fig. 1. Manual tillage on steep slopes leads to a systematic displacement of the topsoil of the arable land.
Because the soil ¯ux is directly proportional to the
slope gradient, erosion and deposition rates will be
proportional to the slope gradient change: erosion will
occur at the convexities, while deposition will take
place as colluvium at the concavities (Govers et al.,
1994). Also zones with zero-in¯ux (such as hill tops,
spurs and upper ®eld boundaries) will face accelerated
truncation of the upper soil horizons, while soil banks
can be formed at lower ®eld boundaries (Revel et al.,
1989; Revel and Guiresse, 1995; Govers et al., 1994;
Quine et al., 1994, Lobb et al., 1995; Guiresse and
Revel, 1995), or on moderate to steep slopes (Poesen
et al., 1997).
Most studies to date have concentrated on tractorplough tillage on gentle slopes (Mech and Free, 1942;
Bollinne, 1971; Papendick and Miller, 1977; Lindstrom et al., 1992; Revel et al., 1989, 1993; Revel and
Guiresse, 1995; Govers et al., 1994, 1996; Quine et al.,
1994; Guiresse and Revel, 1995) or on moderate to
steep slopes (Poesen et al., 1997). Only few publications underline the importance of soil losses caused by
manual tillage (Wassmer, 1981; Roose, 1994; Turkelboom et al., 1997), which is presumably the most
widespread method of land preparation in hilly tropical areas.
In a previous study (Turkelboom et al., 1997), the
rate of soil movement by manual tillage under conditions prevailing in northern Thailand has been
assessed by different methods. Soil translocation
due to manual tillage on ®ve slopes (32±82%) were
measured by monitoring tracers, by measuring triangle tillage step pro®les and by collecting displaced soil
material in a trench. The easiest method to assess
tillage erosion consists of measuring triangle-step
pro®les. The tracer method is a rather accurate, but
time consuming method. The trench method was
rejected due to signi®cant border effects, which led
to underestimation of the soil translocation. Soil
translocation resulting from one manual tillage pass
ranged between 39 and 87 kg mÿ1 on the tested slopes.
On slopes up to 60%, there was modest increase in soil
translocation. However, on slopes steeper than 70%,
soil translocation increased signi®cantly because the
angle of repose for soil clods was exceeded. Based on
these results, soil ¯ux ( the net downslope translocation) was de®ned as an exponential function of slope
gradient. The soil ¯uxes were used to construct, a
nomograph for estimating soil loss rates resulting
from manual tillage erosion as a function of slope
and downslope parcel length. Rates on a typical
upland ®eld (slope 30±50%, slope length 30±50 m)
range from 8 to 18 t haÿ1 tillage passÿ1, but on short
®elds soil loss can reach up to 170 t haÿ1 tillage
passÿ1. This means that, tillage erosion dominates
on short ®elds and ®elds with buffer-strips, whereas
water erosion is the more important form of soil loss
on middle size and long ®elds.
The objective of this paper is to reassess the rate of
tillage erosion by manual hoeing on steep slopes. For
this purpose, additional data have been collected by an
on-farm tillage survey and a new calculation method
for the tillage step pro®le has been applied. These new
data will be compared with the translocation data
obtained via the tracer method in a previous study
(Turkelboom et al., 1997) and the relationship
between hillslope gradient and soil ¯ux will be reexamined. This relationship will be used to assess the
soil ¯ux for complete cropping cycles for the most
dominant cropping systems in the highlands of northern Thailand. Finally, the effect of tillage erosion on
the spatial variability of soil fertility will be described,
and potential solutions will be discussed.
2. Methodology
2.1. Experimental sites
As reported by Turkelboom et al. (1997), soil
movement by manual tillage for different slope gradients (32%, 41%, 51%, 60% and 82%) was measured
by means of an on-farm experiment during December
1994. After removing the weeds, the plots were tilled
by local farmers using normal practices. In order to
measure the soil translocation, three complementary
methods were used: the tracer method, the trianglestep pro®le method and the trench method. The latter
method will not be presented here, due to its poor
results (see Turkelboom et al., 1997).
An on-farm tillage survey was conducted while
farmers were tilling their land during February
1996. Informal discussions with farmers and ®eld
observations helped to understand the nature and
sequence of tillage operations and cultivation practices for various crops. On six ®elds, vertical tillage
depth, step dimensions and hoe size were measured.
The angle of repose of soil clods in ®eld conditions
was assessed at 15 sites by measuring the slope
gradient of the soil surface where clods had started
to roll down as a result of tillage and walking.
All observations were done at Pakha, which is an
Akha tribe village located in Chiang Rai Province,
northern Thailand (208N and 1008E). The elevation
ranges between 800 and 1100 m a.s.l., and the climate
is characterized by a six-month monsoon season and a
six-month dry season. The soil pro®les are classi®ed
as Cambisol (FAO-classi®cation) and the parent material is predominantly phyllite, which explains the
relatively high clay content (30±40% by mass).
2.2. Methods to assess tillage induced soil
translocation
Tracer method: The tracer method was only used in
the on-farm tillage experiment. For this purpose,
painted rock fragments of various sizes were used
as tracers. Tracers were placed randomly in a narrow
furrow below a reference line (marked by a string) at
the top of the plot, and the furrow was re®lled with soil
(Fig. 2). The farmers then tilled the plot moving up the
slope until they passed the furrow. Size of rock fragments was not correlated with the displacement distance (r < 0.20) suggesting that the rock fragments
were suitable tracers. After tillage, the original soil
surface above the narrow trench was marked with a
string, to measure the average tillage depth and the
displacement distance of the tracers that were moved.
Hence, tracers that were installed in the furrow below
the plough depth were not taken into account. The
tillage depth was measured vertically in the ®eld, and
transformed to tillage depth perpendicular to the soil
surface by multiplying with the cosine of the local
slope angle.
The mass of soil that passes a unit contour length for
one tillage pass is de®ned as `soil translocation' (T).
This translocation can be calculated using Eq. (2)
(which is similar to the formula given by Yatsukhno
(1976), quoted by Zachar (1982)):
T ˆ ddDBd;
where T is the soil translocation caused by tillage
(kg mÿ1 tillage passÿ1); dd the mean downslope displacement distance of the tracers (m), parallel to the
soil surface; D the mean tillage depth (m), perpendicular to the soil surface; and Bd the dry bulk density of
the soil ( = 1100 kg mÿ3).
Step pro®le method: A small step at the top of the
parcel is the most obvious indicator of soil movement
by manual tillage (Fig. 3). On this spot soil is moved
downwards but not replaced by any soil from further
upslope. Consequently, a compacted sub-soil is surfacing. Tillage steps degrade easily and need to be
measured soon after tillage operation. The step pro®le
method was used both during the on-farm tillage
experiment and the on-farm tillage survey.
Turkelboom et al. (1997) suggested a triangular
shape of the tillage step pro®le (Fig. 4a) to calculate
soil translocation Eq. (3):
T ˆ Bd
sin …ÿ†sinA2
;
2sin
(3)
where a is the slope angle of original soil surface (8); the slope angle of step (8); and A the length of the step
slope (m), measured down the angle .
After re-investigating tillage steps during the onfarm tillage survey, a trapezoid pro®le shape was
found to be more accurate in characterising the morphology of a tillage step pro®le than the triangle shape.
The trapezoid dimensions of the step pro®le were
measured six times at each site during the on-farm
tillage survey (Fig. 4b). The adapted equation to
calculate soil translocation T is
T ˆ Bd:D‰B ‡ 0:5X ‡ 0:4YŠ;
Fig. 2. Design of tracer experiment for the quantification of soil
fluxes via manual tillage.
(2)
(4)
where D is the tillage depth (m), measured perpendicular to the soil surface (n = 30); B the bottom length
of the tillage trapezoid (m); and E the length of tillage
trapezoid at the soil surface (m).
Fig. 3. The tillage step at the top of a recently tilled parcel is a
clear indication of tillage erosion.
The last term (Y) of Eq. (4). is derived from other
parameters (B, E, D, and see Fig. 4b), and is
therefore less accurate than the other terms. Therefore,
the following assumptions were made: (1) in order to
take into account the slightly convex shape of the tilled
soil surface at the downslope side of the tillage step, a
coef®cient of 0.4 instead of 0.5 is used; and (2) when
this term produced negative values, its value is set
equal to zero.
3. Results and discussion
3.1. Measurement of soil translocation by manual
tillage
Tillage depth depends on a large number of factors.
The length of the hoe blade and the slope gradient of
Fig. 4. (a) Cross-section through a tillage step, and the parameters
needed to calculate soil translocation by the triangle-step profile
method. A, and are measured in the field. The shaded area represents soil lost by manual tillage and equals ˆ C…Asin †=2 with
C ˆ …Asin†=…sin† and ˆ ÿ(b): Cross-section through a
tillage step, and the parameters needed to calculate soil translocation by the trapezoid-step method. D, B, E, and are measured in
the field. The (shaded) area represents soil lost by manual tillage.
X equals (D / tan(ÿ)), and Y equals (E-B-X).
the ®eld were expected to be important factors controlling tillage depth, but ®eld measurements showed
that there are many other factors that in¯uence tillage
depth, such as local weed pressure, rock fragment
content, soil texture, bulk density, presence of cracks
and moisture content (Turkelboom et al., 1997).
Therefore, an average depth perpendicular to the soil
surface of 8.2 cm (standard deviation = 1.9 cm,
n = 21) was calculated.
The mean displacement distance measured by the
tracer method ranged from 0.50 to 1.10 m, depending
on the slope gradient (Table 1), resulting in soil
translocation rates between 45 and 100 kg mÿ1 tillage
Table 1
Mean displacement distance of soil and soil translocation rates due to one manual tillage pass assessed by tracer method (on-farm tillage
experiment; after Turkelboom et al., 1997)
Slope
gradient (%)
Mean displacement
distance (m)
Standard
deviation (m)
Observations
(n)
Soil translocationa
(kg mÿ1 tillage passÿ1)
32
41
51
60
82
0.50
0.48
0.54
0.58
1.10
0.19
0.26
0.19
0.28
0.43
170
169
174
171
178
45.1
43.2
48.2
52.4
98.7
a
Calculated with Eq. (2) and assuming a mean tillage depth of 8.2 cm.
passÿ1. The step dimensions measured by the trianglestep and trapezoid-step method are shown in Tables 2
and 3, respectively. Despite the limitations of the
trapezoid-step method (see Section 2.2), its results
seem superior to the triangle-step method, because (1)
the translocation data obtained via the trapezoid-step
method corresponds better with those obtained with
the tracer method (which is considered the most
realistic estimate) (Fig. 5); and (2) ®eld observations
indicate that a trapezoid shape is more accurate in
describing the tillage step pro®le. Based on the data
given in Tables 1±3 it could be concluded that the
coef®cient of determination for the relation between
slope gradient and soil translocation was highest for
the triangle-step pro®le method (r2 = 0.94, p < 0.01)
followed by the trapezoid-step pro®le method
(r2 = 0.90, p < 0.01) and the tracer method
(r2 = 0.81, p < 0.05).
Table 2
Tillage step profile dimensions and soil translocation rates due to one manual tillage pass assessed by the triangle-step method (on-farm tillage
experiment; after Turkelboom et al. (1997))
Slope
Step dimensions
Gradient (S) (%)
Angle (8)
(8)
A (cm)
32
41
51
60
82
18
22
27
31
39
58
64
69
71
72
17.7
22.0
26.3
28.3
39.0
Soil translocation
(kg mÿ1 tillage passÿ1)
30.5
42.2
52.0
52.7
66.9
Table 3
Tillage step dimensions and soil translocation rates due to one manual tillage pass assessed by the trapezoid-step method (on-farm tillage
erosion survey)
Slope
Step dimensions
Gradient (S) (%)
Angle (8)
(8)
C (cm)
B (cm)
D (cm)
17
28
38
39
54
71
10
16
21
21
28
35
41
58
50
57
53
64
31.6
45.7
49.0
59.3
85.3
72.6
17.3
32.7
35.8
37.0
68.4
60.0
8.1
9.7
8.7
8.2
5.8
10.2
Soil translocation
(kg mÿ1 tillage passÿ1)
21.7
41.8
41.9
43.6
48.8
77.9
Fig. 5. Soil translocation by manual tillage assessed by different methods for different slope gradients.
3.2. Calculation of soil fluxes caused by tillage
erosion
Soil ¯ux is de®ned as the net downslope translocation (Lobb, personal communication). The soil ¯ux
equation is the relationship between slope gradient
and soil ¯ux. A simple diffusion type equation (Eq.
(1), Govers et al., 1994; Poesen et al., 1997) or an
exponential function (Turkelboom et al., 1997), is not
the most appropriate approach for this type of tillage
operation, as different relationships can be identi®ed
for different slope gradient ranges:
1. For the slope gradient range between 20% and
70%, reliable translocation values were obtained
by the tracer method (n=5) and the trapezoid-step
method (n=6). These data were combined to
develop a linear regression equation between
slope gradient and soil ¯ux Eq. (5). The ¯ux for
the 82% slope gradient was excluded from the
regression analysis, because a second process was
involved. Although no measurements were conducted at very gentle slopes, it seems reasonable
to extrapolate the linear relationship to slope
gradients down to 3% (Fig. 6):
Qt ˆ 77:0S …if 3%<S<70%†;
(5)
2. Below a slope gradient of 3%, random tillage
directions dominate; above a slope of 70% topsoil
sliding can take place. For very gentle slope
gradients (i.e. S < 3%), farmers do not have a
preferred tillage direction. Hence, over the years
the soil ¯ux will greatly balance out and approach
0 kg mÿ1. Although no measurements were conducted in this slope range, it can be expected that
there is a net movement downslope as a tillage
operation downslope still moves slightly more soil
than an upslope tillage operation. The soil ¯ux
equation at this slope gradient range Eq. (6) is
assumed to be similar as Eq. (5) minus the
intercept which cancels out in the calculation of
net downslope translocation:
Qt ˆ 77:0:S …if S<3%†:
(6)
The threshold slope gradient for random tillage direction is estimated to be around 3%. Above this thresh-
Fig. 6. The physical relationship between slope gradient and soil flux by tillage. Linear equation from this study, exponential equation from
Turkelboom et al. (1997).
old, farmers always start to till from the bottom of the
field and gradually progress in an upslope direction.
3. Above the angle of repose, clods start to slide and
roll downslope where they accumulate at the foot
of the slope as colluvium (Fig. 7). The angle of
repose for rough tilled ®elds was identi®ed to be
around 70% (Turkelboom et al., 1997). Under
these conditions, soil ¯ux does not depend on the
slope gradient anymore, but on the downslope
length of the parcel area which has a slope
gradient steeper than the angle of repose. Soil
¯uxes for these conditions can be described by
Qt ˆ L>70% DBD …if S > 70%†;
(7)
where L>70% = downslope length of the parcel having
a slope gradient > 70%.
The absolute ¯ux values predicted by the linear Eq.
(5) and the exponential equation reported in Turkelboom et al., (1997) do not differ much for slope
gradients in the range between 20% and 70% (Fig.
6). Nevertheless, the complex relationship presented
in this paper is considered an improvement as it
approaches the physical reality in the ®eld more
closely and as it clearly indicates two thresholds where
soil ¯ux signi®cantly changes with slope gradient, i.e.
3% and 70%.
3.3. Assessment of tillage erosion rates
For a given parcel, mean soil losses per hectare for
one tillage pass can be derived from the soil ¯ux Eq.
(5) and the downslope parcel length. This results in
TER ˆ
Qt10; 000m2
…if 3%<S<70%†;
Lha
(8)
where TER is the tillage erosion rate (kg haÿ1 tillage
passÿ1), Qt the soil flux caused by tillage
(=77.0S+13.3, Eq. (5) (kg mÿ1 tillage passÿ1); and
L = downslope parcel length (m).
Based on Eq. (8) a nomograph can be constructed
(e.g. Fig. 9 in Turkelboom et al., 1997). Such a
nomograph shows that the average soil losses over
the entire parcel due to tillage erosion can vary
dramatically, and that this variation is mainly dependent on downslope parcel length, and to a lesser extent
on slope gradient. For a typical highland parcel with a
can even be much higher as the complete tilled topsoil
becomes prone to sliding. In that case up to 900 t haÿ1
(which equals the soil mass of the complete tilled top
layer) can be translocated. These ®gures indicate that
tillage erosion can contribute signi®cantly to soil
degradation, gradual terracing, colluviation and hillslope evolution via soil pro®le truncation.
3.4. Soil flux caused by different cropping systems
Fig. 7. Deep tillage on very steep slopes with gradients exceeding
the angle of repose (Photo: J. Pelletier).
downslope length of 30±50 m and a slope gradient of
30±60%, 7±20 t haÿ1 would be moved on average to
the bottom of the plot by one deep tillage operation.
Soil movement can reach spectacular rates (70±
130 t haÿ1) on very short parcels ( < 5 m), such as
home gardens or parcels with buffer-strips. These soil
losses represent average ®gures for the entire parcel.
Local soil losses by tillage erosion are much more
important given that within a parcel with a uniform
slope gradient, all soil loss occurs at the upper part of
the parcel (whereas an equivalent soil mass will
accumulate at the lowermost portions of the parcel).
In the conditions where downslope translocation is not
offset by upslope translocation, there is an area of
severe soil degradation that begins at the upper most
section of the parcel and expands progessively downslope. Soil loss by tillage from parcels with a slope
gradient exceeding the angle of repose of soil clods
Tillage erosion rates will differ for each crop, as
each crop requires a speci®c soil management.
Although no measurements of tillage erosion other
than the deep tillage were made, rates can be estimated
based on the experience gained during the conducted
experiments, farmer interviews and ®eld observations.
The basic principle for assessing soil ¯uxes for each
cultivation practice is that all displacement distances
were assumed similar as the one measured during the
tillage experiment, but that tillage depth is adapted
according the type of operation. Table 4 shows the soil
¯uxes caused by the current cultivation practices for
®ve dominant ®eld crops in northern Thailand. A slope
gradient of 40% was taken for this example. The
reported values are based on average cropping practices conducted at Pakha. Large variations can occur
depending on labour availability, previous crop and
actual weed pressure.
The estimated soil ¯uxes for each of the cultivation
practices shown in Table 4 are explained below: The
soil ¯ux rate caused by deep tillage is based on Eq. (5).
All ®elds which are prepared for upland rice, cabbage
and ginger will undergo a deep tillage operation
(¯ux = 44 kg mÿ1, mean tillage depth = 8.2 cm).
Fields for maize are tilled in about 70% of the cases.
A maize ®eld will not be tilled (¯ux = 0 kg mÿ1 in
30% of the cases) when there was a good bum or when
there are few weeds (®rst part of rainy season), or
when the necessary time or labour is lacking and
herbicides are sprayed (second part of the rainy season). Deep tillage for beans is practised in 50% of the
cases. When beans are grown as a ®rst crop and weed
pressure is small, tillage is often not practised. Fields
to be planted with beans as a second crop (after maize
or fallow) need a weed control by deep tillage or by
herbicide spraying.
In upland rice and cabbage ®elds, the big clods are
smoothened to prepare a ®ne seedbed. For ginger, ®ne
Table 4
Assessment of tillage erosion risk on a 40% slope for the five most important annual cropping systems in Pakha, northern Thailand
Upland rice
Maize
(Soy) Beans
Cabbage
Ginger
Cultivation period
Soil flux (kg mÿ1 passÿ1): land preparation
Deep tillage (hoe)a
Smoothening / fine tillage (hoe)b
Cleaning (hoe or rake)d
Stale seedbed weeding (hoe)a
Construction of beds (hoe)a
Soil flux (kg mÿ1 passÿ1): Plantingd
Soil flux (kg mÿ1 passÿ1): Weeding
1st weedingb,c
6 months
3 months
3 months
3 months
5 or 10 months
44 (100%)e
11 (100%)
4 (100%)
11 (100%)
±g
Bamboo stickh
0 or 44 (30±70%)f
±g
±g
±g
±g
1 hoe
0 or 44 (50±50%)
±g
±g
±g
±g
Bamboo stick/little hoeh
44 (100%)
11 (100%)
4 (100%)
±g
9 (100%)
Little hoeh
44 (100%)
0 or 12 (50±50%)
±g
±g
±g
1 hoe
4 (100%) little hoe
Hand + knifeh
(Hand + knife)h
74
0 or 4 (50±50%)
(little) hoe
±g
±g
0!48
4 (100%)
little hoe
±g
±g
72
4 (100%) little hoe
2nd weedingb,c
3rd weedingd
Sum soil flux (kg mÿ1 cropping cycleÿ1)
0 or 8 (50±50%)
hoe or knife
0 or 8 (50±50%) hoe
a
9!61
Based on tillage experiment.
b
Based on estimated tillage depth.
c
Based on estimated % of the soil surface affected by tillage.
d
Estimate
e
Indicates the frequency of cultivation practice causing the soil flux indicated above.
f
0 or 44 (30±70%): Soil flux equals 0 kg m tillageÿ1 in 30% of the cases, whereas soil flux equals 44 kg mÿ1 tillage passÿ1 in 70% of all cases.
g
Cultivation practice does not apply to this crop.
h
Soil flux is negligible.
(Hand + knife)
±g
49!61
Fig. 8. A network of furrows and beds are constructed for cabbage cultivation.
tillage is performed only when the soil is hard. It is
estimated that the displacement distance is the same as
for deep tillage, but the average tillage depth is about
2 cm.
Organic plant residue is cleared with a rake (cleaning). The amount of soil is quite small, but the displacement distances can be important. The soil ¯ux for
cleaning is estimated to be 4 kg mÿ1.
The stale seedbed weeding is only performed in
upland rice ®elds. The purpose of this method is to kill
fast-emerging weeds before planting of the crop and to
dilute the seed source of weeds. This tillage operation
is similar to the ®ne tillage. Super®cial tillage represents about 20% of the soil ¯ux caused by a deep
tillage (Van Keer, personal communication).
In cabbage fields a network of beds and furrows are
constructed by moving the soil from the furrows to
the beds (Fig. 8). This action is similar to a deep
tillage operation, but it affects only 20% of the soil
surface.
Planting causes little soil displacement. Where a
big hoe is used, the soil flux is estimated at
1 kg mÿ1.
The first weeding is done most thoroughly, as
competition risks are most severe during the seedling stage of the crops. During weeding the topsoil
is also moved because the roots of the weeds have
to be destroyed. When a small hoe is used, 50% of
the top 2 cm is assumed to move over a distance of
40 cm. Healthy beans do not always receive a
weeding as they can quickly cover the soil. In
maize fields a big hoe is used, which supposedly
moves 50% of the top 4 cm over a length of 40 cm.
When the farmer has little time for the first weeding
in his maize fields, weeds are slashed with a knife
only. In case of slashing or when a second crop is
planted after maize, a second weeding by hoe is
essential.
Herbicides are becoming a more popular alternative for weeding and tillage, especially when there
is a labour shortage. However, local farmers consider deep tillage still a more efficient weed control
measure than herbicides.
Despite the uncertainty of the soil ¯ux caused by
some operations, this exercise allows one to make a
relative comparison between cropping systems and
cultivation operations.The deep tillage operation is
clearly the most important cause of soil movement.
If deep tillage is conducted, it contributes between
60% and 90% of the total ¯ux. If we sum up the soil
¯uxes for every cropping cycle, crops can be classi®ed
according to their susceptibility to tillage erosion.
Upland rice and cabbage are the most sensitive crops,
ginger and maize cause a moderate soil transport,
while the cultivation of beans results in the smallest
soil ¯ux.
An increasing land pressure and an improving
market access in the highlands of northern Thailand,
will probably cause tillage erosion to become a more
important process of land degradation. This is not only
due to increasing tillage frequency and weed pressure,
but also by the change in the farmers' crop choice to
more `tillage-demanding' crops such as cabbage,
potatoes and ginger. On the other hand, a decline in
tillage-demanding subsistence crops (such as upland
rice) and the increasing use of herbicides will partly
offset this trend.
3.5. Spatial variability of soil properties
Cultivation of the same parcel for several years
causes a spatial variability of soil properties, especially when the fertility is concentrated in the top layer
(Fig. 9). Soil fertility near the tillage step is lower as
the subsoil is exposed, while fertility at the bottom of
the parcel is increasing (on the condition that accumulation of soil takes place inside the parcel). This
was clearly observed at two fertilised runoff plots at
Doi Tung experimental station (conducted by the
Department of Land Development of Thailand and
IBSRAM). After 5 years of cultivation, the tillage
steps clearly show a redder colour and a more compacted topsoil than at the surrounding area. The soil
organic matter content at the step was 33±48% lower
than in the rest of the plot, while the available phosphorus content declined from 2.5 mg kgÿ1 to nondetectable levels. This caused the yield of upland rice
to decline by 55% compared to the rest of the run-off
plot. During an erosion survey at Pakha, it was
observed that due to this fertility drop, the upper most
3±5 m of some parcels were abandoned half way
through the cropping season and quickly colonised
by weeds. On parcels with a short downslope length
(e.g. ®elds with hedgerows) this process progesses
much faster. Tillage erosion leads to the gradual
formation of terraces with a steep fertility gradient,
which can undermine the long-term productivity of
these systems (Turkelboom et al., 1996, 1997).
On steep parcels, where tillage steps can become
several meters long, it was often observed that rills
were initiated at the compacted sub-surface soil. This
downward head development of rills is the opposite of
what is normally taking place. Moeyersons (1991)
observed on steep slopes in Rwanda a similar process
caused by rural paths, and called it `forward erosion'.
The impact of these rills is mostly limited as they
come to a dead end in the tilled and loose topsoil
where water transmission losses occur. In some steep
and ®ne tilled parcels, however, the rills can continue
Fig. 9. Spatial variability of soil properties as caused by tillage erosion.
their downslope development and might contribute
signi®cantly to the generation of sediment further
downslope.
3.6. Potential solutions
As the primary purpose of tillage is to control
weeds, the pressure and type of weed will in¯uence
the tillage frequency and, hence, tillage erosion rates.
Any measure that reduces the weed intensity, will
automatically lead to a reduced soil loss by tillage
erosion. Environmentally friendly methods to control
weeds are related to management of the fallow land,
the timing of slashing, buming, weeding and tilling
and the ef®ciency of the crops to cover the soil, which
depends on crop choice, crop density, mulching and
fertility management (Van Keer et al., 1998). A switch
to minimum or no-tillage cultivation may be another
alternative. However, under the current conditions in
northern Thailand, this is only possible by the massive
use of herbicides, or by switching to perennial crops
which do not need to be tilled regularly.
The net loss of soil can be reduced if the soil can be
kept at the bottom of the ®eld and protected against
water erosion. Buffer strips with weeds or crops can
act as a barrier and ®lter, if they are not becoming
buried under soil clods.
Cultivation of annual crops on ®elds steeper than
the angle of repose is probably the most dramatic form
of soil degradation as the complete topsoil is at risk.
The only solution is to avoid ®elds with slopes steeper
than 65% for annual cropping, and use them for
perennial crops or leave them for forest regeneration.
Short ®elds and ®elds with hedgerows are highly
susceptible to a fertility decline at the tillage step zone.
This could be compensated by concentrating plant
nutrients in this zone, for example, by pruning of
hedgerows or by scraping topsoil from land above
the tillage step. An alternative is to annually move
down the alleys, so that the formation of fertility
gradients can be avoided, or by growing less demanding crops in the less fertile top zone (Turkelboom et
al., 1996).
The opposite strategy can also be observed in the
®eld: tillage and water erosion are sometimes actively
stimulated in order to concentrate the fertility in a
limited zone where a high yield can be obtained
(sediment or nutrient harvesting, or controlled collu-
viation, De Ploey and Yair, 1985). This principle was
applied by a farmer at Pakha who wanted to construct
paddy ®elds (at the expense of his upper neighbours),
or in areas where the available soil fertility is too poor
to produce a reasonable yield (e.g. the highlands of
Ethiopia, Kruger, personal communication).
4. Conclusions
The switch from shifting cultivation to more permanent highland cropping systems in northern Thailand led to an increase in soil tillage intensity. Soil
translocation by manual tillage on steep slopes was
reassessed by applying a new calculation method
(trapezoid-step) on tillage steps which were measured
during an on-farm tillage survey. Comparison with
previously collected data (Turkelboom et al., 1997)
indicated that the tracer method and the trapezoid-step
method seem to be the most reliable ones. The ®rst
method is probably the most accurate, the second one
is very time ef®cient.
Based on newly acquired understanding of the
processes involved, soil ¯ux by tillage erosion is
quanti®ed by linear functions for different slope gradient classes, instead of a single diffusion-type equation (often proposed for tractor derived tillage erosion)
or an exponential function (proposed by Turkelboom
et al. (1997)). For the slope gradient class between 3%
and 70%, where soil is tilled only in the downslope
direction, soil ¯uxes range between 16 and 67 kg mÿ1
tillage passÿ1. For very gentle slope gradients (i.e.
< 3%), soil ¯uxes are approaching to zero as farmers
do not have a preferred tillage direction. On slope
gradients steeper than 70%, the angle of repose for soil
clods is often exceeded which can result in a complete
sliding down of the tilled top layer.
Soil loss rates resulting from manual tillage can be
calculated if the soil ¯ux and the upslope parcel length
is known. Erosion rates on a typical upland parcel
(slope gradient 30±60%, downslope parcel length 30±
50 m) range from 7 to 20 t haÿ1 tillage passÿ1. The onsite effects of tillage erosion will be more pronounced,
if the ®elds are short, cultivated for upland rice or
cabbage, or when weed pressure is high. Tillage
erosion results in a tillage step with low soil fertility
and low in®ltration capacity. Solutions for this form
of land degradation depend on the degree to which
tillage intensity can be reduced. This might happen
through an improved weed management or by
changing land-use to perennial cropping. Other strategies are concentrating nutrients on the truncated
zones and retaining soil on the ®eld by vegetative
buffers. In conclusion, although not at all apparent,
tillage erosion is an important process of land degradation in northern Thailand and is likely to increase
with increasing land pressure and improved market
access.
5. Notation
g
A
B
Bd
C
D
dd
E
K
L
L>70%
Qt
S
T
TER
slope angle (8) of original soil surface
slope angle (8) of step
±
length of the step slope (m), measured
down the angle b
bottom length of the tillage trapezoid (m)
dry bulk density of the soil
…Asin†=…sin†
tillage depth (m), measured perpendicular
to the soil surface
mean downslope displacement distance of
the tracers (m), parallel to the soil surface
length of tillage trapezoid at the soil surface
(m)
diffusion coefficient (kg m ÿ1 tillage
passÿ1)
downslope parcel length (m)
downslope length of the parcel having a
slope gradient > 70%
soil flux caused by tillage (kg mÿ1 tillage
passÿ1)
slope gradient (%) of the soil surface prior
to tillage
soil translocation caused by tillage (kg mÿ1
tillage passÿ1)
tillage erosion rate (kg haÿ1 tillage passÿ1)
Acknowledgements
We sincerely thank the farmers of Pakha, our
translator Kasea and the Hill Area Development
Foundation (HADF) for their kind co-operation in
the ®eld. Prof. K. Vlassak, K. Van Keer and all our
colleagues of the Soil Fertility Conservation project
are thanked for their support. The SFC research was
sponsored and assisted by the Belgian Agency for
Development Co-operation (BADC), the Flemish
Inter-University Council (VLIR), the International
Rice Research Institute (IRRI), the Flemish Organisation for Development Co-operation and Technical
Assistance (VVOB), Mae Jo University (MJU), the
Catholic University of Leuven (K.U. Leuven) and the
National Research Council of Thailand (NRCT). This
study is a contribution to the Soil Erosion Network of
the Global Change and Terrestrial Ecosystems Core
Research Programme which is part of the International
Geosphere±Biosphere Programme. Finally, we want
to thank Dr. David Lobb and two anonymous
reviewers for their profound effort to improve the
content of this paper.
References
Bollinne, A., 1971. Les rideaux en hesbaye gembloutoise. Bulletin
de la SocieÂte GeÂographique de LieÁge 7, 61±67.
Collins, M., Sayer, J., Whitinore, T., 1991. The Conservation Atlas
of Tropical Forests ± Asia and the Pacific. International Union
for Conservation of Nature and Natural Resources (IUCN),
Gland, Switzerland, 253pp.
De Ploey, J., Yair, A., 1985. Promoted erosion and controlled
colluviation: a proposal concerning land management and
landscape evolution. Catena 12, 105±110.
Govers, G., Vandaele, K., Desmet, P., Poesen, J., Bunte, K., 1994.
The role of tillage in soil redistribution on hillslopes. Eur. J.
Soil Sci. 45, 469±478.
Govers, G., Quine, T., Desmet, P., Walling, D., 1996. The relative
contribution of soil tillage and overland flow erosion to soil
redistribution on agricultural land. Earth Surface Process.
Landforms 21, 929±946.
Grandstaff, T., 1980. Shifting cultivation in northern Thailand
Possibilities for development. Resource systems theory and
methodology series, No.3, The United Nations University,
Tokyo, pp. 44.
Guiresse, M., Revel, J.C., 1995. Erosion due to cultivation of
calcareous clay soils on the hillsides of south west France: II.
Effect of ploughing down the steepest slope. Soil Till. Res. 35,
157±166.
Hinton, P., 1970. Swidden cultivation among the Pwo Karen of
northern Thailand: Present practices and future prospects. In:
Intemational seminar on shifting cultivation and economic
development in northern Thailand. Land Development Department, Bangkok.
Lindstrom, M., Nelson, W., Schumachter, T., 1992. Quantifying
tillage erosion rates due to mouldboard plowing. Soil Till. Res.
24, 243±255.
Lobb, D.A., Kachanoski, R.G., Miller, M.H., 1995. Tillage
translocation and tillage erosion rates on shoulder slope
landscape positions measured using 137Cs as a tracer. Can. J.
Soil Sci. 75, 211±218.
Mech, J., Free, G., 1942. Movement of soil during tillage operation.
Agr. Eng. 23, 379±382.
Moeyersons, J., 1991. Ravine formation on steep slopes: Forward
versus regressive erosion Ð Some case studies from Rwanda.
Catena 18, 309±324.
Papendick, R., Miller, D., 1977. Conservation tillage in the Pacific
Northwest. J. Soil Water Cons. 32, 49±56.
Poesen, J., Van Wesemael, B., Govers, G., Martinez-Feinandez, J.,
Desmet, P., Vandaele, K., Quine, T., Degraer, G., 1997. Patterns
of rock fragment cover generated by tillage erosion. Geomorphology 18, 183±197.
Quine, T.A., Desmet, P.J., Govers, G., Vandaele, K., Walling, D.E.,
1994. A comparison of the roles of tillage and water erosion in
landform development and sediment export on agricultural land
near Leuven, Belgium. In: Olive, L.J., Loughran, R.J., Kesby,
J.A. (Eds.), Variability in Stream Erosion and Sediment
Transport. IAHS Publication No. 224, 77±86.
Revel, J.C., Coste, N., Cavalie, J., Costes, J.L., 1989. Premiers
reÂsultats expeÂrimentaux sur l`entraiÏneinent meÂcanique des
terres par le travail du sol dans le Terrefort toulousain (France).
Cah. ORSTOM, SeÂr. PeÂdol. 25(1±2), 111±118.
Revel, J., Guiresse, M., Coste, N., Cavalie, J., Costes, J., 1993.
Erosion hydrique et entraiÏneinent meÂchanique des terres par
les outils dans les coÏteaux du sud-ouest de Ia France. In:
S. Wicherek (Ed.), Farm Land Erosion in Temperate Plains
Environment and Hills. Elsevier, Amsterdam, pp. 551±562.
Revel, J.C., Guiresse, M., 1995. Erosion due to cultivation of
calcareous clay soils on the hillsides of south west France: I.
Effect of former farming practices. Soil Tillage Res. 35, 147±
155.
Roose, E., 1994. L'eÂrosion meÂchanique seÁche. In: Introduction aÁ la
gestion conservatoire de l'eau, de Ia biomasse et de Ia fertiliteÂ
des sols. FAO Ð Soil Bull. (FAO, Rome), 70, 85±87.
Turkelboom F., Ongprasert, S., Taejajai, U., 1996. Soil fertility
dynamics in steep land alley farming. In: Sajjapongse A. (Ed.),
The 7th Annual Meeting on the Management of Sloping Lands
for Sustainable Agriculture in Asia Network, Organized by
DLD and IBSRAM, Chiang Mai, Thailand, 16±20 October
1995.
Turkelboom, F., Poesen, J., Ohler, I., Van Keer, K., Ongprasert, S.,
Vlassak, K., 1997. Assessment of tillage erosion rates on steep
slopes in northern Thailand. Catena 29, 29±44.
Van Keer, K., Vejpas, C., TreÂbuil, G., 1995. Effects of farmers'
practices on weed infestation in an upland rice based swidden
system in northern Thailand. In: Proceedings of the 2nd Weed
Science Congress, Peradeniya, Sri Lanka, 25±27 May, 1995,
15pp.
Van Keer, K., Cointois, J.D., Turkelboom, F., Ongprasert, S., 1998.
Options for soil and farmer friendly agriculture in the highlands
of northern Thailand. Tropical Ecology Support Program
(TOB) and Soil Fertility Conservation Project (Mae Jo
University, Thailand Ð K.U.Leuven, Belgium). Deutsche
Gesellschaft Wr Technische Zusaminenarbeit (GTZ), Eschborn, Germany, 167pp.
Wassmer, P., 1981. Etude de l'eÂrosion des sols et de ses
conseÂquences dans Ia PreÂfecture de Kibuye. These de Doctorat
de 3eÁme cycle, GeÂographie, Universite Louis Pasteur, Strasbourg, pp. 65±71.
Yatsukhno, V.M., 1976. Study of the erosion by mechanical soil
cultivation in rolling and hommocky sinking relief in
Belorussian SSR. In: Regulations Governing the Occurrence
of Erosive and River-bed Processes in Different Natural
Conditions. lzd. Moskovskogo Universiteta, Moscow.
Zachar, D., 1982. Soil erosion, Developments in Soil Science, No.
10 Elsevier, 547pp.