eschweizerbartxxx
Gully erosion in South Eastern Tanzania:
spatial distribution and topographic thresholds
by
Wouter M. J. Achten, Stefaan Dondeyne, Samweli Mugogo,
Elly Kafiriti, Jean Poesen, Jozef Deckers, and Bart Muys
with 4 figures and 3 tables
Summary. Though gully erosion is often mentioned as a major process of land degradation
in South Eastern Tanzania, little information is available on its distribution. The Makonde
plateau and adjacent inland plains are of particular concern as they are densely populated and
are major areas of cashew nuts production. The occurrence of gully erosion was assessed in 66
villages selected by stratified random sampling in an area of 13,000 km2. Difference in suscep-
tibility to gully erosion between landscape units was assess by determining topographic
threshold parameters of 22 gullies on the Makonde plateau and 14 in the inland plains. Over-
all, gullies are common and spread equally over the different landscape units. Their occurrence
is positively associated with terrain roughness (Cramér’s V = 0.30; P = 0.05) and negatively with
population density (V = 0.44; P
0.01). On the Makonde plateau occurrence of gully erosion
is associated with the presence of roads, while on the inland plains it is predominantly found
in fields (V = 0.37; P
0.05). This association is explained by the high susceptibility of the
Makonde plateau to gully erosion and is due to the particular nature of its deep, highly weath-
ered, sandy soils. Appreciating differences in susceptibility to gully erosion between landscape
units is most relevant for targeting soil conservation measures.
Résumé. L’érosion en ravine dans le Sud Est de la Tanzanie: répartition spatiale et seuils topo-
graphiques. – Bien que l’érosion en ravine soit souvent mentionnée comme processus impor-
tant de dégradation de la terre dans le Sud Est de la Tanzanie, peu d’information est disponi-
ble sur sa répartition. Le plateau Makonde et les plaines intérieures adjacentes sont d’intérêt
particulier étant densément peuplé et étant l’aire principale de production des noix d’anacar-
dier. L’occurrence de l’érosion en ravine a été évaluée dans 66 villages répartis sur une super-
ficie de 13.000 km2 et sélectionnés par échantillonnage aléatoire stratifié. La différence de
susceptibilité à l’érosion en ravine entre les unités paysagiques a été évaluée en déterminant
les paramètres du seuil topographiques de l’érosion de 22 ravines sur le plateau Makonde et
14 dans les plaines intérieures. En général, l’érosion en ravine est courante et est également
répartie dans les diverses unités paysagiques. Leur occurrence est positivement associée à la
rugosité du terrain (V = 0.30 de Cramér; P = 0.05) et négativement avec la densité de popula-
tion (V = 0.44; P
0.01). Sur le plateau Makonde l’occurrence de l’érosion en ravine est asso-
ciée aux routes, alors que sur les plaines intérieures on la trouve principalement dans les champs
(V = 0.37; P
0.05). Cette association s’explique par la susceptibilité élevée du plateau
Makonde à l’érosion en ravine et est principalement due à la nature particulière de ses sols pro-
Z. Geomorph. N. F. 52 2 225–235 Berlin · Stuttgart June 2008
DOI: 10.1127/0372-8854/2008/0052-0225 0372-8854/08/0225 $ 2.75
© 2008 Gebrüder Borntraeger, D-14129 Berlin · D-70176 Stuttgart

eschweizerbartxxx
fonds, fortement altérés et sableux. Apprécier ces différences de susceptibilité à l’érosion en
ravine entre les unités paysagiques est important pour pouvoir cibler des mesures de conser-
vation de sol.
1 Introduction
Gully erosion is the process whereby runoff water accumulates, and often recurs over
short periods, in narrow channels removing the soil to considerable depths. It is one
of the most important processes contributing to the sculpturing of the earth surface.
Areas with high gully density can become unsuitable for agricultural land-use. Gully
erosion can cause major damage to infrastructure and can cause problems by siltation
of rivers and reservoirs (Poesen et al. 2003).
The initiation of gullies is a threshold phenomenon occurring when, for a given
catchment area, a critical slope gradient has been exceeded; or when, for a given slope,
a critical catchment area has been exceeded. The critical values of slope and catchment
area vary according to climate, soil, terrain and land-use (Vandekerckhove et al.
2000). Patton and Schumm (1975) formulated the relationship as
S = a A–b (1)
where S is the slope of the soil surface at the gully head, A the area of the catchment,
and a and b are parameters depending on environmental characteristics. The curve
defined by this equation, corresponds with threshold values for the initiation of gul-
lies in terms of slope and catchment area.
Changes in land-use will influence the formation of gully erosion by affecting
the catchment area. For example, changes in field sizes may directly lead to an expan-
sion of the catchment area, as may alteration in water runoff due to paths or roads.
Because of differences in geomorphologic processes and soil properties, topographic
thresholds can be expected to vary across landscape units. Insights into variation in
susceptibility should be useful to target strategies for controlling or preventing gully
erosion. However, relatively few field studies on gully erosion have been conducted
at large spatial scales, as these can be difficult and time demanding (Poesen et al.
2003).
Although gully erosion is considered a major process of land degradation in
Tanzania (Majule 2004), little is known about the actual extent of the phenomenon.
This study aimed at (i) assessing the occurrence of gully erosion across different land-
scape units, (ii) investigating relationships between the occurrence of gully erosion
with geomorphologic and demographic variables, and (iii) evaluating differences
between landscape units in susceptibility to gully formation.
W. M. J. Achten et al.226

eschweizerbartxxx
2 Materials and methods
2.1 Study area
Bennett et al. (1979) mapped landscape units of South Eastern Tanzania at a recon-
naissance scale (1 : 250,000). From east to west, the landscape can be divided into a
narrow coastal plain, an area of large plateaux and an area of inland plains (fig. 1).
This study focussed on the Makonde plateau and the adjacent inland plains, as they
are the most populated, and the most important production areas for cashew nuts,
one of Tanzania’s major export commodities. The Makonde plateau is part of a chain
of similar plateaux, with the Rondo plateau to the north and the Mueda plateau in
Mozambique to the south. Soils of the Makonde plateau are deep, highly weathered,
well drained and with a sandy topsoil and sandy loam or sandy clay loam in the sub-
surface horizons. Soil structure is usually weakly developed. The dominant soils are
Xanthic, Veti-Acric Ferralsols (Dondeyne et al. 2003).
The inland plains are gently undulating with broad flat-topped interfluves and
wide shallow valleys, and are derived from Precambrian Basement rocks, mostly
gneiss. Soil changes reflect variations in lithology, drainage and erosional history. On
the interfluvial crest, least affected by erosion, deep, highly weathered, red sandy clay
loam or sandy clay soils occur (Rhodic, Veti-Acric Ferralsols). On the slopes, less
weathered, often shallow, coarse textured soils occur. These can be less than a metre
Gully erosion in South Eastern Tanzania 227
Fig. 1. Landscape units of the study area and location of the study sites (Source: adapted from
Bennett et al. 1979).

eschweizerbartxxx
deep and as varied as Chromic Luvisols, Leptic Cambisols and Petric Plinthosols
(Dondeyne et al. 2003).
The climate of South Eastern Tanzania is influenced by the south-eastern trade
wind in mid-year and the north-eastern trade wind during the turn of the year. Air
temperatures vary little: the mean is 24.3°C in July and 27.5°C in December; mean
annual air temperature is 26°C in the coastal area and 24°C in the inland areas (Ben-
nett et al. 1979). Rainfall pattern is uni-modal, but very erratic as can be seen from
table 1. The inland plains, in the rain shadow of the plateaux, have a distinct drier cli-
mate than the plateaux. Hourly rainfall records necessary to determine the rainfall
erosivity are not available. Average rainfall intensity is also higher on the Makonde
plateau (table 1), indicating that erosivity can be expected to be highest on the
plateaux.
2.2 Field survey
In a first phase, a random sample of 66 villages was selected in an area of about
13,000 km2, after stratification according to landscape units: 23 from the Makonde
dissected plateau, 22 from the Makonde high plateau, and 21 from the inland plains
(fig. 1). In each of these villages, representatives of the village authorities and farmers
were asked whether gullies occur within the village area. If so, at least one was visited
of which the geographical coordinates of the gully head was recorded using a hand-
held global positioning system (GPS). The location of the gully was checked to be in
agricultural fields, fallow fields, bush or along roads.
In a second phase, 18 of these villages were revisited to determine the slope-
catchment area relationship of the gully heads, whereby not more than five gullies
were studied per village. The villages were selected to be representative for the land-
scape units, based on the information obtained from the first phase. On the Makonde
dissected plateau 15 gullies were studied, on the Makonde high plateau 7 and in the
inland plains 14. The slope gradient (S) of the soil surface at the gully head was meas-
W. M. J. Achten et al.228
Table 1 Rainfall characteristics (mm/year) in South Eastern Tanzania. Rainfall values for a
dry and wet year have a 10 year return period.
Annual rainfall Rainfall intensity
Climatic station Dry year Average Wet year Median IQR
Makonde plateau
Mtwara airport (n = 45) 844 1137 1429 187 82
Newala (n = 11) 688 1245 1912 243 65
Inland plains
Nachingwea (n = 14) 605 823 1025 189 43
Masasi (n= 11) 538 873 1550 179 55
n = number of years of observation; Rainfall intensity = S(RM2)/RA, with RM monthly rain-
fall, RA annual rainfall; IQR, interquartile range (adapted from Dondeyne et al. 2003).

eschweizerbartxxx
ured with a clinometer over a stretch of 5 m up and 5 m down the gully head, as pro-
posed by Nyssen et al. (2002). The surface area of the catchment was determined by
walking through it and visually evaluating whether runoff water would flow to the
gully head or not, while recording boundary points with a GPS. Farmers helped in
the field assessment and proved to have a detailed knowledge on the local hydrogra-
phy. Subsequently, using a geographic information system, the boundary points were
converted to polygons for calculating the actual catchment area.
2.3 Geomorphologic characteristics
River density and terrain roughness were calculated for a circle with a radius of 5 km
around each of the study sites: for villages where more than one gully occurred, a
gully head taken randomly; for sites where no gullies occur, the village centre was
taken as a reference. The river density was calculated as the total length of rivers
occurring within this area. The data was derived from the digitised drainage network
taken from the landscape maps of Bennett et al. (1979).
The terrain roughness was calculated as the standard deviation of the topo-
graphic elevation, but corrected for the overall slope using least square linear regres-
sion. The elevation was assessed from a digital elevation model derived from contour
lines of the topographic maps (scale 1 : 50,000) with contour lines at a height interval
of 15.2 m (50 feet). ArcView’s Surface Tool extension (Jenness 2004) was used for
estimating the elevation of 49 points spaced on a 800
800 m square grid with the
same centre point as described above.
2.4 Demographic data
Data on the human population per ward1 was taken from the Tanzania National Web-
site (2003) for the 2002 census, and from the official report of the 1988 census (United
Republic of Tanzania 1991). Population density was determined by linking the 2002
population data with a digital map of the wards. Differences in the population
between 2002 and 1988 per ward were used to calculate the average annual popula-
tion growth. Subsequently, a digital map overlay allowed linking the study sites to
annual population growth rate per ward.
2.5 Data analysis
ANOVA with the Fisher’s least significant difference test (LSD) was used to check if
landscape units differ in terms of geomorphologic and demographic characteristics.
The Cramér’s V coefficient was calculated to assess the association between gully ero-
sion and the different landscape units, the geomorphologic variables (river density,
terrain roughness) and demographic variables (population density, population
growth rate). For this analysis, the continuous variables were grouped into three cat-
egories corresponding to the 1⁄3 highest, 1⁄3 medium and 1⁄3 lowest observation. The
mean topographic thresholds were determined as the orthogonal linear regression
Gully erosion in South Eastern Tanzania 229
1 A ward is the smallest administrative unit in Tanzania above village level.

eschweizerbartxxx
W. M. J. Achten et al.230
T
ab
le
2
Su
m
m
ar
y
st
at
is
ti
cs
o
f
ge
om
or
ph
ol
og
ic
a
nd
d
em
og
ra
ph
ic
c
ha
ra
ct
er
is
ti
cs
a
ro
un
d
66
o
bs
er
va
ti
on
p
oi
nt
s
on
t
he
M
ak
on
de
p
la
te
au
a
nd
ad
ja
ce
nt
in
la
nd
p
la
in
s,
S
ou
th
E
as
te
rn
T
an
za
ni
a.
L
an
ds
ca
pe
u
ni
t
G
eo
m
or
ph
ol
og
y
D
em
og
ra
ph
y
R
iv
er
d
en
si
ty
T
er
ra
in
r
ou
gh
ne
ss
P
op
ul
at
io
n
de
ns
it
y
G
ro
w
th
r
at
e
(m
/k
m
2 )
(m
)
(p
er
s/
km
2 )
(%
)
M
ak
on
de
h
ig
h
pl
at
ea
u
33
0
a*
A
26
.4
a
A
10
6
a
A
1.
1
%
a
A
(n
=
22
)
M
ak
on
de
d
is
se
ct
ed
p
la
te
au

42
3
a
A
15
.8
b
B
90
a
A
1.
9
%
ab
B
(n
=
23
)
In
la
nd
p
la
in
s
73
0
B
15
.6
B
80
A
1.
8
%
B
(n
=
21
)
L
ul
in
di
p
la
in
s
89
8
b
13
.1
b
96
a
1.
4
%
ab
(n
=
6)
N
ac
hi
ng
w
ea
-M
as
as
i p
la
in
s
73
5
b
13
.6
b
93
a
2.
2
%
ac
(n
=
8)
So
ut
he
rn
M
as
as
i p
la
in
s
58
0
ab
20
.5
ab
50
ab
1.
9
%
ab
(n
=
7)
*
di
ff
er
en
t
le
tt
er
s
in
di
ca
te
s
ig
ni
fi
ca
nt
d
if
fe
re
nc
es
b
et
w
ee
n
th
e
la
nd
sc
ap
e
un
it
s
(P


0.
05
);
t
ho
se
in
c
ap
it
al
c
on
ce
rn
s
th
e
an
al
ys
e
w
it
h
th
e
la
nd
sc
ap
e
un
it
s
of
t
he
p
la
in
s
ge
ne
ra
lis
ed
t
o
in
la
nd
p
la
in
s.

eschweizerbartxxx
lines between the logarithm of the slope at the gully head (S) and the logarithm of the
catchment area (A), as proposed by (Vandekerckhove et al. 2000).
3 Results
The landscape units mapped by Bennett et al. (1979) and used to stratify the random
sampling of the study sites clearly correspond to distinct geomorphologic character-
istics (table 2). River density on the Makonde plateau is significantly smaller than that
of the inland plains. The opposite is true for the terrain roughness, though the differ-
ence between the Makonde dissected plateau and inland plains is not significant,
while terrain roughness of the Southern Masasi plain gets to levels comparable as to
those of the Makonde high plateau.
The differences between the landscape units in terms of demographic character-
istics are less clear, which could be expected, as the study area was targeted at the more
populated parts of South Eastern Tanzania. As in terms of geomorphologic and
demographic characteristics, the differences between the Lulindi, Nachingwea-
Masasi and Southern Masasi plains are not or hardly significant (table 2), they are fur-
ther referred to as the “inland plains”.
Overall, gullies are found in 73% of the visited villages (48 out of 66 villages).
In percentage of occurrence there is only a small difference between the Makonde
high plateau 68% (15 out of 22 villages), the Makonde dissected plateau 70% (16 out
of 23 villages) and the inland plains 81% (17 out of 21 villages). Occurrence of gul-
lies is significantly associated with terrain roughness and population density, and not
with any other geomorphologic or demographic characteristics (table 3). As shown
in fig. 2, occurrence of gullies is positively associated with terrain roughness, and neg-
atively with population density. On the Makonde plateau the occurrence of gullies is
however significantly associated with the presence of roads, while on the inland
plains they rather occur in fields (fig. 3).
For the 36 gullies studied in more detail, the relationship between slope at the
gully head (S) and catchment area (A) is clearly different for the Makonde plateau and
inland plains (fig. 4). The observations of the Makonde dissected plateau and
Gully erosion in South Eastern Tanzania 231
Table 3 Strength of association between occurrence of gully erosion and geomorphologic
and demographic characteristics.
Cramér’s V df P
Landscape units* 0.24 4
0.44
Major landscape unit† 0.13 2
0.59
River density 0.17 2
0.40
Terrain roughness 0.30 2
0.05
Population density 0.44 2
0.01
Population growth rate 0.06 2
0.89
* mapping units of Bennett et al. (1979) (see also fig. 1).
† Landscape units generalised to inland plains, Makonde dissected plateau and Makonde high
plateau.

eschweizerbartxxx
Makonde high plateau, seem to represent one population as can be seen in fig. 4. The
slope of the 22 gullies on Makonde plateau ranges from 0.02 to 0.36 m/m, while on
the inland plains it ranges from 0.02 to 0.015 m/m. The larger variability of the slope
measured at the gully head on the Makonde plateau is consistent with the higher ter-
rain roughness as derived from the DEM. In contrast the range for the catchment area
is smaller on the Makonde plateau (0.08 to 7 ha) than on the inland plains (0.05 to
18 ha). Most strikingly, the slope of the regression line on the Makonde plateau is
much steeper than that for the inland plains, indicating that soils of the Makonde
plateau are more susceptible to gully erosion (fig. 4).
W. M. J. Achten et al.232
Fig. 2. Frequency of occurrence of gully erosion in relation to terrain roughness and popula-
tion density in 66 villages South Eastern Tanzania.
Fig. 3. Relationship between occurrence of gully erosion in fields or along roads per land-
scape units, South Eastern Tanzania.

eschweizerbartxxx
4 Discussion and Conclusions
Gully erosion is common and evenly spread across the different landscape units of
South Eastern Tanzania. In contrast to the perception that gully erosion would be
principally associated to roads as had been reported by Vermang (2004), this study
shows that gullies are equally common in fields. Still, a distinction can be made
between landscape units, as on the Makonde plateau gullies are more often associated
to roads than on the inland plains (fig. 3). Values from field observations reported in
literature for the slope of the regression line – the b parameter of equation 1 – typi-
cally falls within the range between 0.1 and 0.6 (Nyssen et al. 2002, Vandekerck-
hove et al. 2000). Compared to those values, one can conclude that soils of the
Makonde plateau (with b = 0.8) are very susceptible to gully erosion.
The high susceptibility for gully erosion of the Makonde plateau could partly
be attributed to the more erosive rainfall (table 1), but for the major part it seems to
be linked to geomorphologic characteristics in combination with the particular
nature of the deep, highly weathered and sandy soils. The soils of the Makonde
plateau have weakly developed structures and are generally sandy, but with distinct
higher clay content in the subsoil (Dondeyne et al. 2003) which explains their high
susceptibility to gully erosion. Terrain roughness as determined from the digital ele-
vation model is highest on the Makonde high plateau (table 2) and is positively asso-
ciated with the occurrence of gully erosion (fig. 2). This relationship can however
be seen as both cause and consequence of gully erosion: deep gullies result in higher
terrain roughness and in turn may enhance runoff processes leading to more gully
erosion.
The association between gully erosion and roads on the Makonde plateau is
understandable when taking into account the high susceptibility of this unit as indi-
cated by the topographic threshold. Any change in the size of the catchment area,
which typically happens when roads are made, will easily lead to the formation of
gullies. Reciprocally, any soil conservation measure reducing surface water runoff,
Gully erosion in South Eastern Tanzania 233
Fig. 4. Average topographical threshold of 22 gullies on the Makonde plateau and 14 gullies
on the inland plains in South Eastern Tanzania.

eschweizerbartxxx
can be expected to be more effective on the Makonde plateau than on the inland
plains. Appreciating these differences between landscape units seems most relevant
to local policy makers and rural development agencies when elaborating strategies for
soil conservation.
Somewhat counter intuitive is the negative association between the occurrence
of gully erosion and population density. At first sight, it could be explained by the
“more people, less erosion” hypothesis where population growth and agricultural
intensification results in improving rather than deteriorating the use of soil and water
resources (Boyd & Slaymaker 2000). In the case of South Eastern Tanzania, how-
ever, it has to be attributed to, fewer people living in areas of the drier inland plains
commonly affected by gully erosion and where soils are shallow and stony.
As soils of the Makonde plateau are shown to be particularly sensitive to gully
erosion, this should equally be expected to be so for the similar Rondo and Mueda
plateaux. Moreover, when taking into account that soils of the Makonde plateau have
been shown also to be sensitive to soil acidification when sulphur is used as a fungi-
cide in the cashew groves (Ngatunga et al. 2003), this area appears to be particular
vulnerable to land degradation.
Acknowledgements
This study was conducted as part of the “Soil Conservation Management project, South East-
ern Tanzania” (project No ZEIN2002 PR253) financed by the Ministry of Agriculture and
Food Security, Tanzania with support from the Flemish Inter-University Council, Belgium.
The authors like to thank Mr Laurence B. Emmanuel and Ms Elisa Mapua for their contribu-
tion in preparing the digital terrain model, Mr Musa Mapua, Mr Laurence B. Emmanuel and
Mr Abdallah Nachundu for their assistance in the fieldwork.
References
Bennett, J. G., Brown, L. C., Geddes, A. M. W., Hendy, C. R. C., Lavelle, A. M., Innes,
R. R. & Sewell, L. G. (1979): Report of the Zonal Survey team phase 2. Vol. 1: The phys-
ical environment. – Project report 67, Mtwara/Lindi Regional Integrated Development
Programme, Land Resources Development Centre, Surrey, 242pp.
Boyd, C. & Slaymaker, T. (2000): Re-examining the “more people less erosion” hypothesis:
special case or wider trend? – Natural Resource Perspectives 63: 1–6.
Dondeyne, S., Ngatunga, E. L., Cools, N., Mugogo, S. & Deckers, J. (2003): Landscapes
and soils of South Eastern Tanzania: their suitability for cashew. – In: Topper, C. P.
& Kasuga, L. J. (eds.): Knowledge transfer for sustainable tree crop development: 229–
239; BioHybrids Agrisystems, Reading.
Jenness, J. (2004): Surface Tools extension for ArcView 3.x, v. 1. 5. Jenness Enterprises. Avail-
able at http://www.jennessent.com/surface_tools.htm. (accessed 19 Aug. 2004)
Majule, A. E. (2004): Gateway to land and water information: Tanzania. FAO Rome,
http://www.fao.org/ag/agl/swlwpnr/reports/y_sf/z_tz/tz.htm (accessed 16 June 2005).
Ngatunga, E., Dondeyne, S. & Deckers, J. (2003): Is sulphur acidifying cashew soils of
South Eastern Tanzania? – Agriculture, Ecosystems and Environm. 95: 179–184.
Nyssen, J., Poesen, J., Moeyersons, J., Luyten, E., Veyret-Picot, M., Deckers, J., Haile,
M. & Govers, G. (2002): Impact of road building on gully erosion risk: a case study from
the Northern Ethiopian Highlands. – Earth Surf. Proc. and Landf. 27: 1267–1283.
W. M. J. Achten et al.234

eschweizerbartxxx
Patton, P. C. & Schumm, S. A. (1975): Gully erosion, North-western Colorado: a threshold
phenomenon. – Geology 3: 88–90.
Poesen, J., Nachtergaele, J., Verstraeten, G. & Valentin, G. (2003): Gully erosion and
environmental change: importance and research needs. – Catena 50: 91–133.
United Republic Of Tanzania (1991): Population census. Takwimu, Bureau of Statistics, Pres-
ident’s Office, Planning Commissions, Dar es Salaam (volumes per administrative
region).
Tanzania National Website (2003): 2002 Population and housing census, available at
http://www.tanzania.go.tz/census/index.html. (accessed 13 Sep. 2004)
Vandekerckhove, L., Poesen, J., Oostwijk Wijdenes, D., Nachtergaele, J., Kosmas, C.,
Roxo, M. J. & De Figueiredo, T. (2000): Thresholds for gully initiation and sedimenta-
tion in Mediterranean Europe. – Earth Surf. Proc. and Landf. 25: 1201–1220.
Vermang, J. (2004): Gully erosion on the roads of the Makonde Plateaux in Southeast Tanza-
nia. – Master dissertation, Catholic University of Louvain, Leuven, Belgium, 133pp.
Addresses of the authors: Wouter M. J. Achten, Bart Muys, Laboratory for Forest, Nature
and Landscape Research, Catholic University of Leuven, Celestijnenlaan 200, B-3001 Leuven,
Belgium. – Stefaan Dondeyne (Corresp. author), Chimanimani National Reserve, Ministry of
Tourism, C. P. 121, Chimoio, Mozambique. E-Mail: stefaan_dondeyne@yahoo.co.uk –
Samweli Mugogo, Elly Kafiriti, Naliendele Agricultural Research Institute, PO Box 509,
Mtwara, Tanzania. – Jean Poesen, Physical and Regional Geography Research Group, Catholic
University of Leuven, Celestijnenlaan 200, B-3001 Leuven, Belgium. – Jozef Deckers, Labo-
ratory for Land and Water Management, Catholic University of Leuven, Celestijnenlaan 200,
B-3001 Leuven, Belgium.
Gully erosion in South Eastern Tanzania 235