-FI_ Agricul turc Ecosystems 8% ...
-FI_
Agricul turc
Ecosystems 8%
Enwronment
Agriculture. Ecoqsterns and Environment 6.5 (1997) 9% 106
-_I_
Rehabilitation of a. semiarid ecosystem in Senegal z
1. Experiments at the hillside scale
P. Perez ‘*-, J. Albergel ‘, M. Diatta ‘, M. Grouzis ‘. M. Sene ”
-.-----~-
-
.-
~--~-.----._ . .
1 . i n t r o d u c t i o n
sible for this phenomenon (Le Borgne, 1990: Sir-
coulon, 1992; Cirouzis and Albergel, 1989: Lericol-
In Mrica, environmental degradation is now ~$1
lais, 1990).
documnted (Pieri, 1989; Richard. 1990; Poulsen
The agricultural capacity of lands in the Sudano--
and La ~‘esso~i,
1991) and charactsrized by intense
Sahelian zone is reduced by this ecosystem degrad&-
soi! eroçion. soi1 fertility Ioss and marked reductions
tion. with a consequent negative impact on i0d
in ~!:!I~I covet. N~tunl (mainly climatic) fxtcsrs arld
inhabitants. Ecosystems have to be restored in ordcr
anthrq~ic factors are considered to IX jointly respon-
to a~.ure sustainable socioeconomic activity in th::
zone. Extensive programmes have b’een set up :*:
control land degradation, but they often fail becausc
they are conducted on a strictiy technical bas&
without addressing the widespread problcm
.- -.-. --
1
-*
.1
96
P. Pere;: et a!. / Agricdture, Ecosystem ad Enc~irovwnent 65 I /Y 9 7) Y$- ](.&j
(Ro&:t& 1989; Tybirk, 1991). Moreover, there are
cleared, degrade vcry quickly afr:er cropping: The
very fiew results conceming actual effects, at the
ter-race has leachctf. disturbed ferruginous soils on
hillside scaie, of land management on croi produc-
colluvial/alluvial
deposits. This is the traditional
tion improvement or soi1 and water conserva.tion (La1
area of human occupancy. There has been a Sharp
and Stewart, 1990; Vlaar. 1992).
drop in soil fertility in recent years induced by
The present study was carried out in the ground-
continuous cropping and fertiliser .and manure short-
nut cropping basin of Senegal from 1983 ta 1993.
age (Ange, 199 1).
This region supplies almost half of the groundnut
In the lowlan&.. the soils are quite primitive
and millet produced in the country. In the light of the
hydromorphic aIlu\\ i:li deposits. co:lonized by Acacitr
diverse: activities and interactions that occur in rural
.sc~~~~~ and Mitrtrg~~~~rr iwrmis. The main factors lim-
areas, our rehabilitation operations were conducted
iting africultural de\\ clopment ure violent tloods and
on the basis of ecological features and human uses
sand buildup due to crosion on the hillslopes (Al-
(Perez and Sene, 1995). The‘ objective was to give
berge1 a n d Percz. 1’493).
evidence of the actual effects, in terms of hydrology
The population cjcnsity \\vas cstim;ited. in 1983, at
and agronomy, of rehabilitation practices used in a
6 1 inhabitants/km’: this population (mainty Wolofs)
small cultivated watershed (60 ha).
bas almost doublrd ol’cr the past 10 years. The local
We first review the features of the zone and
far-ming systcni i5 chiel’ly based o n
groundnut
techniques used for watershed management and sur-
( A r-irchis - h!~pogt~c~ 1
and m i l l e t ( Punnisetirm
vey. Next. we present results from the hydrological
/i~~l~+&.~), \\Vit11 ;I trcnd towards combining agricul-
survey.. the hedges and field monitoring. Discussion
turc and livestock ;)rot.luction. From 1970 to 1990.
ernphasizes the cost/bencfit balance of such land
cropping arcas incr-ea:,cd from 40% to 707~ of thr
management for farmers.
total surface. They arc: now characterized by continu-
ous cultivation and animal draught (Sene and Perez.
1994).
2. Material and methods
2. I. A reu description
The study area (Fig. 1) fias a Sudano-Sahelian
III 1983. t\\vo \\v:ltcrshcds werc delimited in thc‘
climate marked by a long dry season (7 to 9 months)
Rural Community of Kn>mor. Though two c;;scs
and a short summer rainy season (3 to 5 months).
absolutely alikc could net he found. we strcssed
The mean annual precipitation Ier the 1930 to 1990
zimilarity of =eornorpholocic~II and land occupancy
period was around 820 mm. and dropped to 661 mm
<haracteristics. Thc outlets v.‘cre fi1.c kilornctcrs apart.
over lhe last two decades ( 1970 to 1990). Despite
located close by v,.llagcs t o facilitate hydrological
this rainfall deficit, violent rainstorms bave ‘occurrcd
<urvcys. Thc t\\\\‘o b,isins \\\\‘c‘rc stuclied Ilncler natur;tl
regularly (Dacosta, 1992).
1:onditions over an initial pt’riod ( 1983 to 1957) :tncl
The region is characterized by a series of vast
Ihen conservation nic;IsurcS u’crtf set up on one ot
tablelands (30 to 60 m elevation), interlaced with a
them in 1988. From 1989 to 199.3, we studied the
networlc of valleys with gradua1 slopes. In thle central
hydrologicat result of these soi1 conservation prac-
plateau area, the soils are ferruginous wit’hout hard-
tices, in comparison with the traditional land use.
pan. Formerly densely wooded savannas, these lands
This paired watershed technique (B,osch and Hewlett,
have been cleared. Only talus is now covered with
1982; Fr:itsch, 1992) is based on the built-in assump-
quite poor brushwood, mainly C’omhretacea~r species
tion that other factors are nearly constant.
(Bertrand, 1972).
The Ndierguene watershed (ND), of 0.9 km’,
At the base, the hillslope includes an upstream
presents a shaltow relief (relief/Iength ratio = 5.7
slope and a vast terrace. On the slope, there are quite
rn/km). The tributaries, mainly tracks and ephemeral
primitive soils, colluvial deposits, with fine grave1
*:ullies, -feed temporarily the downstream channet
and fenric hardpan. These soils, which were recently
during the rainy season (June to October). In 1983,
:
I
P. Pue: et al. /Agriculture, Eqosytems and Akironment 65 ( 1997) 95- 106
the cn>pping area represented 42% of the basin area
‘The Keur Dianko watershed (KD), of 0.6 km’, i>.
(Fig. :!a). The outlet is equiped with a tain gauge and
situated on the same hiliside. It presents a more
a water-stage recorder located in a concrete-liried
elongated shape and a more distinct relief
ditch. From 1983 to 1985, suspended sediments were
(relief/length ratio = 12.4 m/km) than the. former
coflected with an automatic sampler. Aftetwards, due
one. In 1983, the cropping area represented 34% of
to a Neak reliability of resuhs, sediments were mrtn-
the basin area (Fig. 2h). The outlet equipmcnt ii
ually :;ampled. The ND catchment is considered as a
identical to the previous one. Unfortunately, sus-
contra1 basin during the whole period.
pended sediments were not manually sampled untii
I
L..
. .
.
c
z
i
+
.
.
f
3
Fig. 1. Location of ~the experimental site in Senegal.
. . ---_
--
.-
-
-y-
.-.._ - ._
‘WL
,
98
P. Perez et 01. /Agriculture. Ecoqstems ond Encironment 65 (IW7) 95-106
1988. Conservation nieasures started in 1988, so we
rangeland productivities (La1 and Stewart, 1990;
cari ;:onsider a fîrst period of five years (1983 to
Pierce, 1991). SO, we began, from 1988, to introduce
1987,) for ,original conditions and a second one (19891
soi1 and water conservation practices to the farmers
to 1993), for soif and water conservation.
working in the Keur Dianko watershed (Fig. 3).
First, we tried to partition hillside area in order to
2.2.2. Sd and warer conserL*ation pracfices
efficiently deal with runoff and concentmted flow
Land degradation was found in both watersheds.
and SO, to reduce erosion process. In the croppinp
Interne runoff and sheet erosion characterized up-
area 13 live-hedges, covering a total of 3000 linear
stream areas, with gully erosion and sand deposits in
meters, were set up by villagers between 1988 anal
the dtlwnstream areas. As generally accepted, we
1990. Multiple shrub species were planteSd to insure
thought that management that alleviated these con-
the best on-site selection, ac-xorcfinp to ~il cc,ndi.
straints would have an impact on trop yields 4nd
tions and farmers remarks: Acacia rdoticvr, Bmhiniu
--_
Basin boundaries
/-rJ
Rangeland cleared and croppedl between 1983 and 1990
cl
Areas already cropped
i n 19813
Remaining rangeland in 1990
l3i.i
Fig. 2. Land use evolution within the two watersheds, bctwee:n 3983 and 1990. (A> Contrai basin; (B): IIeveloped basin.
. - - * .
_ - _ - - - - - -
---_
--y---
- - -
.._”
. - /
.‘“.,~M.-”
..
P. Perez et al./Agricui’ture, Ecosystenxs and Enciranment 65 f 1997) 9S-Jo6
9 9
rufescens, Piliostigma reticulata, Ziziphus maurita-
and the absence of suitable formula, was based opon
nia, Prosopis julifora, Acacia seyal, Acacia mellif-
the field pattem and the landowners agreement.
cru, Dichrostachys glomerata (Rautureau et al.,
Planting techniques were described by Ruelle et al.
1991). The hedge spacing, due to the shallow siope
( 1990). TO decreast: nmoff energy at the base of the
cropping area iimits
-
track
- - Iive-funcc
???????
stone hund
A
micro-basin
Fig. 3. Watershed management plan of the Keur Dianko basin (0.6 km’ 1.
P. Perez et aL /Agriculture. Ecosystem ami Endronment
65 (JYY7) 95-1~
QhS, a stone line was constructed in a brushwood
Beyond the hydrological response to tic water-
zone with material available on the site.
shed management, we were also interested in ,
‘Illen, inside this network, we tried to incneas,e
long-term survey of the trop yieIds. For this purpuse.
infih-ation and water storage in the soi1 using the
an exhaustive fîeld monitoring was realijted, since
techniques adapted to traditional rangelands or trop-
1987% in the Keur Dianko watershed. Because (hi
ping mas. Hence, two exclosure plots (0.5 ha) were
time consuming this survey couid not be achicveti il%
locatc,d in the rangelands to furtber natural vegeta--
the second watershed, except for the cadastal c;ur~i:y
tjon tenovation (Diatta, 1994). On the edge of the
IIe monitoring comprised for each of the 6’7 ~?CI~;
talus, some microbasins planted to shrub species,
plots:
( Acac?a nilotica, Bauhinia rufescens, Ziziphus taarc-
* Annual cadastral survey.
ritii& permitted the revegetation of degraded atiens
* Recording the dates of farming practicc:, :i!~if ;I:
with lutcroping hardpan. I n t h e trop fields, ,the
types of tools
folloumg cropping practices were proposed to fab-
* Observations of weeding levels, trop L’III~‘I~~c‘I:~:~
ers:
and flowering.
- C3ntour cultivation.
* Measurements of yield components from ;I:F II:’
- Dry season decompacting. This technique corne-
sampling plots.
spondr to a 10 cm deep subsoiling done by animal
The abject of the hedge monitoring HX tc:, C>I~
traction with a single excavating pick. Because there
mate, according to soif and rain conditions, the avtï‘r-
is no ab,ailable time during the beginning of the raiiny
age rates of survival and grouth for each spi:cich
season (seeding operations), it is the only solution to
This was important for explaining farmer’s mari\\ .I
hclp ir filtration of water from the first, and oflen
tion and for planning shrub pruning. Twice a yc,~r
violent rainstomls (Le Thiec and Bordet, 1990; Seine
(June and October) each of the 7000 plants L~;IS-
and Pesez, 1994).
controlled and measured (Rautureau et al., 199 1 ;b.
- Shallow ridging.before emergence, for grould-
nut cro3. It is a slight modification of the traditiorlal
weetiin:; being practiced just after sëeding. SOI~~
3. Hesults
rags ari’ placed over the heads of the local hoe. Tl$s
is a clever and not expensive way to increase s&l
relief on the seeding line (5 cm high). This techniqae
cari also be combineci with a localized manure appli-
Comparison of the annual hydrological b~~lances
cation (Sene, 1995).
of the two watersheds shows similar runoff coc”ffi-
- Shiillow earthing up, for millet trop. When the
cients before applying consenation measures in Kcui
trop begins to develop suckers a small ridger is used
Dianko (Table 1). During the period from 1983 to
to crcatc: mounds on the seeding line (10 cm hi&.
1987, cumulative runoff totalizès Sa.7 mm in Keur
AS ab0i.e. it cari b e combined w i t h a localizad
Dianko and 72.6 mm in Ndierguene watershcd. Thc:
manure appliCiltiOn (Perez et al., 1996).
average runoff coefficients wert’ Close: 2. I % (KDi
At la:;t, brushwood checkdams and stone pavments
and 2.6% (ND).
were usr:d to stabilize the main gully and transform it
After 1988. the annual hydrological balance of‘ the
in a permanent waterway and, therefore, secure the
Ndierguene watershed notably evolved, as shown by
adjacent tcack. This work was done by villagers.
a marked increase in flooding; this contrasts with the
stabilizing trend noted for the Keur Dianko water-
2.2.3. Fleld and hedge moniroring
shed (Table 1). During the period from 1989 to
The ],xal impact of individual conservation mea-
1993, cumulative runoff was 87.9 mm in Keur Di-
sures (livre-hedge, stone line, cropping practices) and
anko and 277.8 mm in Ndierguene watershed. The
the mecianisms involved were not studied at th$
average runoff coefficients were distinct: 2.9%~ (KD)
hillside !Cale. They were taken into account during
and 7.0% (ND).
farm plot scale experiments, described further ii
As shown in the Fig. 4, this difference can’t be
another paper (Perez et al., 1996).
attributed to any major variation between the two
101
P. Perez et a/. /Agriculture. Ecosystems and Encironment 65 (1987) 95-lt16
Table l
CanPison of the annuaJ hy&o]ogical balances between the pilot basin, (ND; 90 ha) and tbe developed basin (KD; 60 ha)
--.
-
Penod
Year
ND basin
KD basin
-
-
-
-
Annual rain
Annual runoff
Runoff
Annual rain
Annual runoff
Runoff
(mm)
(mm)
coefficienl. (%)
(mm)
(mm)
coefficient (Sl
Before
1983
429.2
3.1
-
0.7 - -
391.4
6.8
1.7
1984
420.8
7.8
1.9
354.6
6.1
1.9
1985
493.1
14.3
2.9
589.5
18.3
3.1
1986
670.5
28.2
2.7
658.1
17.9
2.7
1987
767.8
19.2"
2.5"
663.1
5.0"
0.8" *
Total
2784.4
72.6
2.6"
2656.7
54.7
2.1b
Planning
1988
825.2
22.7
2.8
891.2
24. I
2.7
After
1989
631.9
5.7
0.9
722.0
7.4
I .O
1990
488.6
12.3
2.5
442.5
10.3
2.3
1991
455.6
22.8
5.0
498.8
14.5
2.9
1992
615.0
43.5
7.2
578.1
12.7
2.2
1993
741.9
119.9
16.2
799.0
43.0
i.-l
Total
2933.0
205.2
7.0"
3040.4
87.9
?.Y”
Watershrd management started in 1988.
“Miwng data.
“Mean value.
-----~-- ____ -.--.-.~--- _^_. --& -.-. - -.
-.-_--
..-_ y
/
-a--- KO cumulative runoff (1988 to 1993)
j
#
*--4
1
.__o- __ ND wmulative runoff (1988 to 1993) i
? ?
1
???
?
j ?? ND cumulative rainfall (1988 to 1993) 1
0
0
L
--_-
<B
-__-
-----
---_ -.--- ..- I-“-i- z-----.-----Y -
T----. -- ~-_~ ~~~~ --_
-8
500
0
d
*
8
’
*
’
b
Li_-_~_-_i---L--I<
.
,
/
i-.u- 0
0
500
1000
1500 *
2000
2500
KD cumulativt! rainfall (mm)
Fig. 4. Cumulative runoff amounts recorded from 1988 to 1993 ar, th:: cxutories of the control (ND) and developed (KD) basins. The ND
basin cumulative rainfall amounts are comparable with the KD basin.
102
P. Perez et al./Agriculrure,
&xqvstems and’ Environment 65 (1997) 95-106
rainfsill series. For this representation of the cumula-
wiith a runoff coefficient of 9% for the developed
tive runoff values, from 1988 to 1993, we extratted
basin (KD) and 11% for the control one (ND).
from the data set the events generating no flood in
Flloodings caused by the next two showers (very low
bath Twatershed, the missing or uncertain records and.
infiltration conditions) were markedly higher and
the errents for which rainfall amounts differed dore
intense in the Ndierguene watershed (KD: 10% and
t&n 40% between the two watersheds. One Can
14%; ND: 50% and 42%). The solid transports were
notice the increase, from 1988, of the variation ;be-
1.5 t/ha for the developed basin (KD) and 3.5 t/ha
tween the two basins.
,
for the contrai basin (ND).
in the years with high runoff, the conserva$on
meascres had a marked effect on the water balahce
3.2. Field nwnitoring
and the solid tmnsportsW For example, in 1991, lelels
of solid transports were the same in both basins
The results of the cadastral survey show a com-
(aboul 650 kgjha), but in * 1992, they represenked
mon trend. in both watersheds, to increase cropping
488 kj;/ha for the Keur Dianko watershed and more
areas. In 1990, cropping area represented 55% in the
than 1 t/ha for the Ndierguene watershed.
developed basin CKD) and 62% in the control one
Tht: impact of the watershed management I is
(ND) (Fig. 2). In Keur Dianko, even after 1988. frorn
clearlq demonstrated by an event that occurred ion
6 to 9 ncw farm plots were slashed from rangeland
4-5 August 1993 (Fig. 5). Three consecutive rain-
every year. During the same rime, 5 fïelds, seriously
storms totaled 179 mm in 24 heurs. The floods after
damaged by erosion, are abandonned in the control
the first shower were the same for the two basibs,
basin (ND) an4 no onz in the developed basin &CD).
0
e?I -...-. .~-.._--_-- --.-.
50
1
/
m
&J . . ..-__-_--
____ i’r-- . . -.---_---_--- -1
- - - - - _i-
100
/
?r
5
2
- -+- KD discharge
/
I
01
1
E
m
--o-- ND discharge
1
1
s
f 6r) ----
.i
’
.2
'so ?!
u
.ELm
iii
.a
- NR rainfall
,
2
---.----~-
g
4(1 T -_---- A..---~-+
-_ - -..:.Ly---- -.~------- - - - -
200
: ;
;
,
/
1 1
:
)
/
!
I
I
/
:
;
1
.Y
i
-.._
a
2tl 4’--le----- ----LA
-~
---_
250
t
j
è
I
\\
.
1 300
1932
oo:ao
04:40
09:36
14:24
19:12
0o:oo
Tipe (hh:mm)
Fig. 5. Rtinfall amounts (mm) and specific discharge (I/‘s/ha) during the 4th and 5th of August 1993. Comparison belween the control
(ND) and the developed (KD) basins.
P. Perez et ai. /Agriculture. Ecosystems and Erwironment 65 f 19971 Y5-- 106
1 on
nough some trials were condocted, eveqy year,
Table 3
with farmers of Keur Dianko, the proposed cropping
Average survival rates and average growing rates of the seedlings
-.~
pm&es; were diversly adopted. Contour cultivation
Date of Date of monitor&
increased from 20% to 80% of the flelds, between
p’lanting
Rainfall
Nb
1988
1989
1990
1991
1987 and 1993, thanks to the establishment of the
198X
894.2
1700
-
91
135
8 4
- -
]ive-hedges. But dry season decompacting, shallow
3 4 $ 1 0 79&35 137f58 145f66
ridging for groundnut trop or shallow earthing up for
1989
71!2.0
2100
-
‘74
7 0
31&11
ólt-28 63k43
millet trop increased only from 0% to 20% during
1900
442.5
4 1 0
-
-
-
3 6
the same time. Localized manure application reached
-
31*10
35*12
only 16% of the plots. TechnicaI and socio-eco-
nomic reasons of such a failure are analysed in a
Rcsults are givrn for each date of planting and each date ot
previous paper CPerez et al., 1996).
nionitoring.
liirst line: avcrsgc wrvival rate from the plantins l%); secou<:
Due to this weakness of spreading, but also to
lin<: average cumulative height from the planting (cm).
variable rainfall conditions, grain yields shoswed no
(Kamfall): annuel rainfall (mm); (Nb): numher of plants.
trend during the whole period (Table 2). Considering
Al! Oie specie< art m-r&.
annual mean yield or its coefficient of variation, it is
obvious that production remained at a very low
potential (inter-annual mean yield was 687 kg/ha
iutt:r’t: of 11w ii\\-e-hcdge. Once establishccf thc dearh
for bath crops) and that heterogeneity between the
rate of the plants was much iower. ‘Table 3 sho\\vs
fann plots tvas high. The surprising results achieved
rhat the ~ro~~~th r;tte depended aIsol on the planting
in 1988., drspite favorable rainfall amounts, were due
{conditions. Rcsults given for 1991 rnust be read with
to intense Ieaching and flowers aborting.
I::N~ because pruning activities began during tllc‘
3.3. Lir:e-lvedge rvvovvitoring
previ’ous year-.
In fact. these global results hide differences be-
The growth of the seedlings of shrub species was
tween species (Table 4). For example. Gliv-icidia
strongly linked with climatic conditions. In 1!388, the
sepiuvn gave bad results from the fîrst year (survival
exceptional precipitation permitted achieving an av-
rate: 33%) and Prosopis julifïora remained steady
erage survival rate of 9 1% (from 4700 plants:), but in
only one more year (survival rate: 43%). On the
1989 and moreover in 1990, for the last setting up of
other hand, Acacia nilotica and Bauhiniu rvq%escevvs
hedges, the survival rates dropped to 74% (from
confirmed. in local conditions, their strength and
2100 plants) and 36% (from 410 plants), respectively
their adaptability, described by ot.her authors (La1
(Table 3). It must be emphasized that neirher irriga-
nnd Steuiart. 1990: Vlaar, 1992). Farmers were ver-y
tion n(:>‘r protection were given to the Young shrubs.
sensitive to the survival rate of the plants. considered
The planting conditions were then crucial for the
3:; 3 mte of profitability of their work. But they
Table 2
Groundnut ( Aruchis h?ï>ogeo) and millet (Pennisetum fiphaïdes) annual grain yields
Year
Rainfall (mm)
Groundnut grain yield
-~
Millet grain yield
Nb
Mean (kg/ha)
c v (%)
Nb
Mean (kg/ha)
cv (%)
- - - - - -
1987
663.1
15
1069
45.3
1 0
821
43.0
1988
894.2
18
6 2 9
39.0
7
4 1 2
65.0
1989
722.0
19
9 2 9
30.6
1 1
6 8 2
4.6
1990
442.5
28
4 8 3
43.7
17
4l8
53.5
1991
498.8
26
5 6 9
43.3
2 1
5 2 7
46.2
1992
578.1
24
5 3 6
52.6
I I
7 5 7
27.1
1993
799.0
2 0
5 9 5
34.3
1 7
1163
33.6
@ainfall): annual rainfall; (Nb): number of plots; (Mean): average value; (CV): coefficient of variation.
Fiel& m’onitoring of the KD basin.
a
104
P. Ferez et al. /Agriculture. Ecosptems and Encironmenr 65 (1997,J 95- 106
Table 4
had been modified. In this case, the paired, watershec.
Avaage survival rate from che pianting (1988)
-
technique was the only experimental method tha,
Specas
Nb
Date of monitoring
could pive evidence of this result.
1989
19!xl
1s
‘1
Even in recent reviews on the subject (Amir.
1996; Unger, 19%) there are very few results of
Acacia niluticn
804 98
96
%
Bau&nia rufescens
155J 96
Y5
95
studies realised at this scale (1 km’) comparin_c
Dichwstachp glomerata
118 76
6 5
62
developed and control bnsins. At inferior scales (t
Glirkidia sepium
60 33
2 0
18
ha, 100 m2), many studies gave evidence of positive
Acucia mellifera
148
92
a 7
85
results - in terms of restoration -- of de\\~eiopment:
Parknsonia aculeata
43
9R
83
8 3
PiliO.itigma
reticulutum
641
8-l
83
IÏ
applied in comparable climatic and human condi-
Pr0stvi.r jufiflora
306
83
13
35
tions (t’laar, 1992; Roose. 199-l).
Acuciu senegal
76 87
8 3
82
The farmers took a polite interest in h>‘drological
Aaxb seFa!
177- 97
93
93
results but they were more responsive to the main
Zi:iflrus mauritar;ia
365
79
7 8
77
gully and adjacent tracks stabilisation. The relation
Total
4300
91
85
83
between rills, track and gullies became SO evident
Thesc are the main species plan& in 1988.
that every one tried to control the water\\\\.ays in bis
(Nb)r number of plants.
field with the help of stone pavments or uooci
trunks. On the other hand. despite the predictable
consequences, farmers continued to slaxh thc uppe;
noticed that it was usrless to gct two merers high
parts of the hillside. Between the land &=r~J:t,:icr~.
thonly shrubs to control runoff and erosion! FcIf this
and the social pressure on tenure, thev chose fhc
reasçn, a programme of pruning was sta.rted s ince
lesser of two eviis. This cari partly explain thc
1990.
limited effect of the conservation measures In thc
A~?other aspect of.tRe shrubs management wi ; thie
Keur Dianko watershed.
annu; replacing of the ones that died into the xist-
After eleven years of monitoring, ir s~ml; incrd-
ing hedges. From 1989 to 1991. the average r; e of
ible that we began the planning duriris thc tnc~J
substitution wus nearly 18%. For most of the S~I zies,
rainy year of the decade. By a mere chance. the
the survival rates in this case were equivale t to
major part of the plantings prolited h! i‘;.~~urabl<
those recorded in the case of a first planting For
conditions. What would have bcen thc situation. il’
exaniple, during the year 1989, -13 1 plants of A Cl rrcia
we had started two years later? The avc’r;tgt: sur\\ii.;li
dofica were newly planteci, with a survival r; It!e of
ral:e would have dropped to 40% and rhr furmt’r’!.
82Q1, while 154 plants were planted as substit:utes
enthusiasm would have disappeared. Te&>,. 1iL.e.
with 3 survival rate of S4Yc (Rautureau et al., 1 9 91).
hedges bave two main abjects for the fa:-mers: tts
control runoff (and run-on) during ttlc rainy srason
and to supply fodder during the dry st’ason. For bath
4. Dtwussion
abjects, they get into the habit of crcating swaths, :k[
the base of the hedges, with trop and weed residucs.
Since we expected hydrological effects of the
AI1 the labour time requirementfs were recorded
projec$ the fîrst results were quite disappointin$. It
(Ruelle et al., 1990). For the entire project planting
took iour, years to become obvious thàt we had to
hedges required 160 men X day, excluding the nurs-
analy:;e them in terms of conservation rather than
ery activities. Stone line and pavment building se-
restomtion. The soi1 and water conservation practjces
quired 50 men X day. This theoretically represents
only ljermitted the siowing down of some medha-
350000 FCFA (1700 FCFA/day in 198S), adding
nisms involved in the watershed degradation process.
to, approximately, 50000 FCFA for littlc equip-
For that reason, there is no rendency to decrease the
ments. The nursery yard required 10 men >: day for
global runoff ra.te of the developed basin (KD). But
preparation (fencing, digging) and 70 mett x day for
the increasing d.ifference recorded in comparison with
seeding, that is to say an amount of 14OQOO FCFA.
the control basin (ND) show-ed that the natural trend
SO, the cost of this collective work, done between
- “.
-
-
-<
-..p* - - -
-
-
-
.
P. Perez et ai. /Agriculture. Ecosystems arrd EncYronment 65 II 9971 95-106
10.5
1988 amd 1990, is 540000 FCFA. in fact, this is a
number of individuctl farmers and village communi-
theoretical amount because we didn’t pay for any
tics from the neighborhood are now asking for tech-
salary; but, on thti other hand, we didn’t incllude the
rticai assistance SO a!; to appiy conservation measures
cost of the support from the scientifïc staff.
on their land.
At the hillside scale, the collective benefit cannot
be detived from an increasing trop production, as we
References
have seen it. The positive results mentioned above
are the tracks and the fields stabilization. By compar-
Allwgcl. J.. Pcrcz. P.. 1993: Fonctionnement hydrologique et
ison wiiih the control basin (ND), we cari take as an
arnL:na(-crwnt tlcb hw-fonds des formations sédimentaires du
hypothesis that four farm plots were saved from
continental terminal: exemple du bassin arachider du Sénégal.
irrevers;ible damage. The average field area is 0.5 ha
In: Raunet. M. (Ed.). Bas-fonds et riziculture. CIRAD Publ..
hlontpellier- France. pp. I5S- 164.
(KD basin: 67 plots on 33 ha) and the average grain
Amir. J., 1996: Impact of trop rotation ànd land management on
yield is 700 kg/ha (ix pod yield is 1000 kg/ha).
wil erosion ami rchabilitation.
III: Agnssi. M. ( E d . ) , Soi1
Using the unit price of groundnut pods (1988: 90
c’rw~ion. c<>nwrvation and rehabilitation.
hlarcel Decker III~:..
FCFA/kg). WC cari estimate the benefit as 180000
Ntw \\1’orl, I.SA. pp .?75-399.
FCFA/year. Thc theoretical cost/benefit Ibalance
hecamc stcady after 3 years. These results are close
to thosc given by Vlxx ( 1992) lor Rochette (II 989).
Beyond this ovcr-simple demonstration, it is a
facr that somc of the 20 fxmers involved in the
watershed management felt somehow swindled. Even
though they agrced that there were actual affects.
BI)S~~. 1.51.. Hr‘\\vlztt. 1.1) , 1982: (t rc‘vic\\c of catchmcnt expcri-
thcy said they werc’ net totaly paid back for their
rnsnt~ to d~tcrminc th. c!tCct of \\‘egctar+on changes on water
contribution to the collective effort. This reaction is,
yirld and evapotranspiration.
J. Hfdrol., n”55. pp. 3-73.
partly, due to the relative failure of our cropping
Dacosta. H.. 1992: Economie de l’eau CI DRS sur les bassins
practices proposals. Facing the good results of the
versants de ThyssL: Kaymor. Syntht’se hydrologique 1983-
1988. ORSTOM/URZB, Dakar Scnegal. 67 p.
trials (Rrrez et al., 1996), it is clear that the major
Diatta. M.. 199-l: R&$nération des ZOIWS mlarginales et rôle de
error bas been to neglect the linkage between indi-
l’arbre dan.\\ les programmes de DRS, application aux bassins
viduai and collective actions, the fann plot amd the
versants de Thyss: Kaymor. Doctoral thesk, Un. Louis Pas-
hillside scales.
teur, Strasbourg France. 183 p.
iFritsch. J.M., 1992: Les effets du d6frichcment de la forêt atna-
zoniennc et dc la rnist: en culture >ur l‘hydrologie de petits
hakiins ver-ants.
Coll. Etudcz and Th&es. ORSTOXI E d . .
5. ConcEusion
Pari5 France. 391 p.
Growis. hl.. Alhergcl. J.. 19X9: l)u ribquc climatique à la con-
Various techniques aimed at controlling runoff
trainte écologique. Incidence dc la ~i;hercsw sur les pro-duc-
and stabilizing water erosion phenomena were ap-
tiens vé&~Ics et le mil@ PLI Burhina Faso. In: Eldin. ht..
Millcville, F. (E~S.). Le risque cn agriculture. ORSTOM Publ..
plied in a degraded part of the groundnut cropping
P&is Frwce. pp. 743-253.
basin of Senegal. The conservation measure!; were
Lai. R.. Stewart. B.A.. 1990: Agroforestry for soi1 management in
set up in different landscape units, with their ecologi-
tropics. In: Lai. R., Stewart, B.A. (Eds.), Soi1 degradation, A
cal features and human uses taken into account.
global threat. Springer-Verlag. New York USA, pp. XIIE-
Using the paired watershed technique, it has been
X V I I .
Le Borgne. J.. 1990: La dégradation actuelle du climat en Afrique.
demonstrated that this land management strategy had
entre Sahara et Equateur. In: Richard. J.F. (Ed.), La dégrada-
actual hydrological effects. After five years, the only
tion des paysages en Afrique de l‘Ouest. Ministère de la
agronomie effect was due to the land stabilization.
Coopération et du Développement. Pari5 France, pp. 17-36.
The synergy between runoff control and fertility
Lericollais, A.., 1990: La gestion du paysage? Sahélisation, surex-
restoration should have been emphasized to ensure
ploitation et délaissement des terroirs sereer au Sénégal. In:
Richard, J.F. (Ed.), La dégradation des paysages en Afrique de
farmer participation and benefïts sharing (Roose,
l’Ouest. Ministère de la Coopération et du Développement,
1994). But results are not insignificant: an increasing
Paris France. pp. 152-169.
1Oti
P. Pcrez et al. /Agriculture, Ecosystems cd Ensironmrnt 65 (1997195- 104
Le Thiec, G., Bordet, D., 1990~ Etude de deux outils de, traction
Roose. E.. 1993: Introduction à la gestion conservatoire de I’e:
animale visant à reduire les contraintes de travail du SOI en
de la biomasse et de la fertilité des sols (GCES). RuIl. PCC.~,
zones s&ches. In: Starkey. P.H., Faye, J. (I%ls.:), Animal trac-
de la FAC), n”70, Roma Italia, 520 p.
tion for agriculture development @roc. of WAATN workshop.
Ruelle, P., Sene, M.. Juncker, E.. Diatta, M., Perez. P. (Eds.
Saly. july 1988). WAATN Pubi., pp. 56-64.
1990: Défense et restauration des sols. Coll. Fiches Te<l
Perez. P., Sene, M., 199.5: Evolution des structures agmires’et
niques. vol 1 n“I. ISRA/LJNIVAL. Dakar Senegal. 64 p.
érosion dans le sud Saloum (Senegal). Reseau Erosion n”15,
Sene, IM., 1905: Influence de l’état hydrique et du comportemci
ORSTOM Montpellier France, pp. 59-68.
mécanique du sol sur l’implantation et la fructification
Perez. P., Boscher, C., Sene, M., 1996: L’amélioration dbs tech-
l’arachide. Doctoral thesis, ENSA Montpellier Frsn~e. 127
niques culturales pour une meilleure gestion de l’eau pluviale
Serre, hl.. Percz. P., 1994: Contraintes et possibililZs (Ic v:iIori’
(sud Saloum, Senegal). Agriculture and Développem$nt. n”9.
tion des ressources naturelles dans le sud du bassin ara&itli
20-79.
(Siné Saloum. Sénégal). In: Reyniers, F.N.. Netoycb. 1.. (Ed,.
Picrce. P.J.. 1991: Erosion productivity impact prediction. In: Lai.
Bilan hydrique agricole et sécheresse en Afriqtrtn trol*ical? t.
R.. Picrce. F.J. (Eds.). Soi1 management for sustainabil~ity. Soi1
John Libbey Eurotext, Paris France, pp. 217-233.
and water Conservation Society. Iowa USA. pp. 35-512.
Sircoulon. J.. 1993: Evolution des climats et des rc~\\ourc~~
Pieri. C. (Ed.). 1989: Fertilité des terres de savanes. Rilan de
eau. In: Pontie, G.. Gaud, hi. (Eds.1. L’cnvironncment i
trente ans de recherche et de développement agricole au sud
Afrique. Afrique contemporaine, no 161. La Documentation
du Sahara. CIRAD Publ., Montpellier France. 4l3 p. ~
Fran<aise. Paris France, pp. 55-76.
Poulscn, E., Lawesson, J.E. (Eds.). 1991: Dryland degradation.
Tyhirk. K., 1991: Planting trees in Sahel-doomcd to IXlurr‘. I
Causes and consequences. Danish Sahel Workshop. ~ Aarhus
Poulsen. E.. Lawesson. J.E. (E~S.). Dr+nd tlrgrad.rti~
University Prcss. Aarhus, 134 p.
Causés and consequences. Danish Sahel W’»rksh~!p. ,\\arhi
Rarrfurcau. J.. Ferez, P., Diatta, M., 1991: Implantation ct~ gestion
lrniversity Press, Aarhus. pp, 71-78.
de haie:,-vives. Rapport technique, CIRAD Puhl., hlontpellier
I -riFer. P.W., 1996: Common soi1 and water con~~n.a!iorr 1.r:.
Frmce, 52 p.
tices. In: A_oassi. hl. (Ed.), Soi1 erosion, conservation 21.
Richard, J.F. (Ed.), 1990: La dégradation des paysages en ;Afrique
rehahilitation. Marcel Decker Inc.. New York lS.4, pp. 77:
de l’Ouest. Ministère de la Coopération et du Développement,
399.
Paris France, 130 p.
Vlaar. J.C.J. (Ed.), 1992: Les techniques de conservati(ln de> Eau
Roehette, R.M. (Ed.), 1989: le Sahel en lutte contre la désertifica-
et des sols dans les pays du Sahel. CIEH/‘U,-X\\I’, L;Ai
tion. Leçons d’expérience. CILSS PAC. GTZ, 592 p. ~
W’agcningcn Nederlands, 99 p.
-
-_----
Agricul turc
Ecosystems 8%
Enwronment
Agriculture. Ecoqsterns and Environment 6.5 (1997) 9% 106
-_I_
Rehabilitation of a. semiarid ecosystem in Senegal z
1. Experiments at the hillside scale
P. Perez ‘*-, J. Albergel ‘, M. Diatta ‘, M. Grouzis ‘. M. Sene ”
-.-----~-
-
.-
~--~-.----._ . .
1 . i n t r o d u c t i o n
sible for this phenomenon (Le Borgne, 1990: Sir-
coulon, 1992; Cirouzis and Albergel, 1989: Lericol-
In Mrica, environmental degradation is now ~$1
lais, 1990).
documnted (Pieri, 1989; Richard. 1990; Poulsen
The agricultural capacity of lands in the Sudano--
and La ~‘esso~i,
1991) and charactsrized by intense
Sahelian zone is reduced by this ecosystem degrad&-
soi! eroçion. soi1 fertility Ioss and marked reductions
tion. with a consequent negative impact on i0d
in ~!:!I~I covet. N~tunl (mainly climatic) fxtcsrs arld
inhabitants. Ecosystems have to be restored in ordcr
anthrq~ic factors are considered to IX jointly respon-
to a~.ure sustainable socioeconomic activity in th::
zone. Extensive programmes have b’een set up :*:
control land degradation, but they often fail becausc
they are conducted on a strictiy technical bas&
without addressing the widespread problcm
.- -.-. --
1
-*
.1
96
P. Pere;: et a!. / Agricdture, Ecosystem ad Enc~irovwnent 65 I /Y 9 7) Y$- ](.&j
(Ro&:t& 1989; Tybirk, 1991). Moreover, there are
cleared, degrade vcry quickly afr:er cropping: The
very fiew results conceming actual effects, at the
ter-race has leachctf. disturbed ferruginous soils on
hillside scaie, of land management on croi produc-
colluvial/alluvial
deposits. This is the traditional
tion improvement or soi1 and water conserva.tion (La1
area of human occupancy. There has been a Sharp
and Stewart, 1990; Vlaar. 1992).
drop in soil fertility in recent years induced by
The present study was carried out in the ground-
continuous cropping and fertiliser .and manure short-
nut cropping basin of Senegal from 1983 ta 1993.
age (Ange, 199 1).
This region supplies almost half of the groundnut
In the lowlan&.. the soils are quite primitive
and millet produced in the country. In the light of the
hydromorphic aIlu\\ i:li deposits. co:lonized by Acacitr
diverse: activities and interactions that occur in rural
.sc~~~~~ and Mitrtrg~~~~rr iwrmis. The main factors lim-
areas, our rehabilitation operations were conducted
iting africultural de\\ clopment ure violent tloods and
on the basis of ecological features and human uses
sand buildup due to crosion on the hillslopes (Al-
(Perez and Sene, 1995). The‘ objective was to give
berge1 a n d Percz. 1’493).
evidence of the actual effects, in terms of hydrology
The population cjcnsity \\vas cstim;ited. in 1983, at
and agronomy, of rehabilitation practices used in a
6 1 inhabitants/km’: this population (mainty Wolofs)
small cultivated watershed (60 ha).
bas almost doublrd ol’cr the past 10 years. The local
We first review the features of the zone and
far-ming systcni i5 chiel’ly based o n
groundnut
techniques used for watershed management and sur-
( A r-irchis - h!~pogt~c~ 1
and m i l l e t ( Punnisetirm
vey. Next. we present results from the hydrological
/i~~l~+&.~), \\Vit11 ;I trcnd towards combining agricul-
survey.. the hedges and field monitoring. Discussion
turc and livestock ;)rot.luction. From 1970 to 1990.
ernphasizes the cost/bencfit balance of such land
cropping arcas incr-ea:,cd from 40% to 707~ of thr
management for farmers.
total surface. They arc: now characterized by continu-
ous cultivation and animal draught (Sene and Perez.
1994).
2. Material and methods
2. I. A reu description
The study area (Fig. 1) fias a Sudano-Sahelian
III 1983. t\\vo \\v:ltcrshcds werc delimited in thc‘
climate marked by a long dry season (7 to 9 months)
Rural Community of Kn>mor. Though two c;;scs
and a short summer rainy season (3 to 5 months).
absolutely alikc could net he found. we strcssed
The mean annual precipitation Ier the 1930 to 1990
zimilarity of =eornorpholocic~II and land occupancy
period was around 820 mm. and dropped to 661 mm
<haracteristics. Thc outlets v.‘cre fi1.c kilornctcrs apart.
over lhe last two decades ( 1970 to 1990). Despite
located close by v,.llagcs t o facilitate hydrological
this rainfall deficit, violent rainstorms bave ‘occurrcd
<urvcys. Thc t\\\\‘o b,isins \\\\‘c‘rc stuclied Ilncler natur;tl
regularly (Dacosta, 1992).
1:onditions over an initial pt’riod ( 1983 to 1957) :tncl
The region is characterized by a series of vast
Ihen conservation nic;IsurcS u’crtf set up on one ot
tablelands (30 to 60 m elevation), interlaced with a
them in 1988. From 1989 to 199.3, we studied the
networlc of valleys with gradua1 slopes. In thle central
hydrologicat result of these soi1 conservation prac-
plateau area, the soils are ferruginous wit’hout hard-
tices, in comparison with the traditional land use.
pan. Formerly densely wooded savannas, these lands
This paired watershed technique (B,osch and Hewlett,
have been cleared. Only talus is now covered with
1982; Fr:itsch, 1992) is based on the built-in assump-
quite poor brushwood, mainly C’omhretacea~r species
tion that other factors are nearly constant.
(Bertrand, 1972).
The Ndierguene watershed (ND), of 0.9 km’,
At the base, the hillslope includes an upstream
presents a shaltow relief (relief/Iength ratio = 5.7
slope and a vast terrace. On the slope, there are quite
rn/km). The tributaries, mainly tracks and ephemeral
primitive soils, colluvial deposits, with fine grave1
*:ullies, -feed temporarily the downstream channet
and fenric hardpan. These soils, which were recently
during the rainy season (June to October). In 1983,
:
I
P. Pue: et al. /Agriculture, Eqosytems and Akironment 65 ( 1997) 95- 106
the cn>pping area represented 42% of the basin area
‘The Keur Dianko watershed (KD), of 0.6 km’, i>.
(Fig. :!a). The outlet is equiped with a tain gauge and
situated on the same hiliside. It presents a more
a water-stage recorder located in a concrete-liried
elongated shape and a more distinct relief
ditch. From 1983 to 1985, suspended sediments were
(relief/length ratio = 12.4 m/km) than the. former
coflected with an automatic sampler. Aftetwards, due
one. In 1983, the cropping area represented 34% of
to a Neak reliability of resuhs, sediments were mrtn-
the basin area (Fig. 2h). The outlet equipmcnt ii
ually :;ampled. The ND catchment is considered as a
identical to the previous one. Unfortunately, sus-
contra1 basin during the whole period.
pended sediments were not manually sampled untii
I
L..
. .
.
c
z
i
+
.
.
f
3
Fig. 1. Location of ~the experimental site in Senegal.
. . ---_
--
.-
-
-y-
.-.._ - ._
‘WL
,
98
P. Perez et 01. /Agriculture. Ecoqstems ond Encironment 65 (IW7) 95-106
1988. Conservation nieasures started in 1988, so we
rangeland productivities (La1 and Stewart, 1990;
cari ;:onsider a fîrst period of five years (1983 to
Pierce, 1991). SO, we began, from 1988, to introduce
1987,) for ,original conditions and a second one (19891
soi1 and water conservation practices to the farmers
to 1993), for soif and water conservation.
working in the Keur Dianko watershed (Fig. 3).
First, we tried to partition hillside area in order to
2.2.2. Sd and warer conserL*ation pracfices
efficiently deal with runoff and concentmted flow
Land degradation was found in both watersheds.
and SO, to reduce erosion process. In the croppinp
Interne runoff and sheet erosion characterized up-
area 13 live-hedges, covering a total of 3000 linear
stream areas, with gully erosion and sand deposits in
meters, were set up by villagers between 1988 anal
the dtlwnstream areas. As generally accepted, we
1990. Multiple shrub species were planteSd to insure
thought that management that alleviated these con-
the best on-site selection, ac-xorcfinp to ~il cc,ndi.
straints would have an impact on trop yields 4nd
tions and farmers remarks: Acacia rdoticvr, Bmhiniu
--_
Basin boundaries
/-rJ
Rangeland cleared and croppedl between 1983 and 1990
cl
Areas already cropped
i n 19813
Remaining rangeland in 1990
l3i.i
Fig. 2. Land use evolution within the two watersheds, bctwee:n 3983 and 1990. (A> Contrai basin; (B): IIeveloped basin.
. - - * .
_ - _ - - - - - -
---_
--y---
- - -
.._”
. - /
.‘“.,~M.-”
..
P. Perez et al./Agricui’ture, Ecosystenxs and Enciranment 65 f 1997) 9S-Jo6
9 9
rufescens, Piliostigma reticulata, Ziziphus maurita-
and the absence of suitable formula, was based opon
nia, Prosopis julifora, Acacia seyal, Acacia mellif-
the field pattem and the landowners agreement.
cru, Dichrostachys glomerata (Rautureau et al.,
Planting techniques were described by Ruelle et al.
1991). The hedge spacing, due to the shallow siope
( 1990). TO decreast: nmoff energy at the base of the
cropping area iimits
-
track
- - Iive-funcc
???????
stone hund
A
micro-basin
Fig. 3. Watershed management plan of the Keur Dianko basin (0.6 km’ 1.
P. Perez et aL /Agriculture. Ecosystem ami Endronment
65 (JYY7) 95-1~
QhS, a stone line was constructed in a brushwood
Beyond the hydrological response to tic water-
zone with material available on the site.
shed management, we were also interested in ,
‘Illen, inside this network, we tried to incneas,e
long-term survey of the trop yieIds. For this purpuse.
infih-ation and water storage in the soi1 using the
an exhaustive fîeld monitoring was realijted, since
techniques adapted to traditional rangelands or trop-
1987% in the Keur Dianko watershed. Because (hi
ping mas. Hence, two exclosure plots (0.5 ha) were
time consuming this survey couid not be achicveti il%
locatc,d in the rangelands to furtber natural vegeta--
the second watershed, except for the cadastal c;ur~i:y
tjon tenovation (Diatta, 1994). On the edge of the
IIe monitoring comprised for each of the 6’7 ~?CI~;
talus, some microbasins planted to shrub species,
plots:
( Acac?a nilotica, Bauhinia rufescens, Ziziphus taarc-
* Annual cadastral survey.
ritii& permitted the revegetation of degraded atiens
* Recording the dates of farming practicc:, :i!~if ;I:
with lutcroping hardpan. I n t h e trop fields, ,the
types of tools
folloumg cropping practices were proposed to fab-
* Observations of weeding levels, trop L’III~‘I~~c‘I:~:~
ers:
and flowering.
- C3ntour cultivation.
* Measurements of yield components from ;I:F II:’
- Dry season decompacting. This technique corne-
sampling plots.
spondr to a 10 cm deep subsoiling done by animal
The abject of the hedge monitoring HX tc:, C>I~
traction with a single excavating pick. Because there
mate, according to soif and rain conditions, the avtï‘r-
is no ab,ailable time during the beginning of the raiiny
age rates of survival and grouth for each spi:cich
season (seeding operations), it is the only solution to
This was important for explaining farmer’s mari\\ .I
hclp ir filtration of water from the first, and oflen
tion and for planning shrub pruning. Twice a yc,~r
violent rainstomls (Le Thiec and Bordet, 1990; Seine
(June and October) each of the 7000 plants L~;IS-
and Pesez, 1994).
controlled and measured (Rautureau et al., 199 1 ;b.
- Shallow ridging.before emergence, for grould-
nut cro3. It is a slight modification of the traditiorlal
weetiin:; being practiced just after sëeding. SOI~~
3. Hesults
rags ari’ placed over the heads of the local hoe. Tl$s
is a clever and not expensive way to increase s&l
relief on the seeding line (5 cm high). This techniqae
cari also be combineci with a localized manure appli-
Comparison of the annual hydrological b~~lances
cation (Sene, 1995).
of the two watersheds shows similar runoff coc”ffi-
- Shiillow earthing up, for millet trop. When the
cients before applying consenation measures in Kcui
trop begins to develop suckers a small ridger is used
Dianko (Table 1). During the period from 1983 to
to crcatc: mounds on the seeding line (10 cm hi&.
1987, cumulative runoff totalizès Sa.7 mm in Keur
AS ab0i.e. it cari b e combined w i t h a localizad
Dianko and 72.6 mm in Ndierguene watershcd. Thc:
manure appliCiltiOn (Perez et al., 1996).
average runoff coefficients wert’ Close: 2. I % (KDi
At la:;t, brushwood checkdams and stone pavments
and 2.6% (ND).
were usr:d to stabilize the main gully and transform it
After 1988. the annual hydrological balance of‘ the
in a permanent waterway and, therefore, secure the
Ndierguene watershed notably evolved, as shown by
adjacent tcack. This work was done by villagers.
a marked increase in flooding; this contrasts with the
stabilizing trend noted for the Keur Dianko water-
2.2.3. Fleld and hedge moniroring
shed (Table 1). During the period from 1989 to
The ],xal impact of individual conservation mea-
1993, cumulative runoff was 87.9 mm in Keur Di-
sures (livre-hedge, stone line, cropping practices) and
anko and 277.8 mm in Ndierguene watershed. The
the mecianisms involved were not studied at th$
average runoff coefficients were distinct: 2.9%~ (KD)
hillside !Cale. They were taken into account during
and 7.0% (ND).
farm plot scale experiments, described further ii
As shown in the Fig. 4, this difference can’t be
another paper (Perez et al., 1996).
attributed to any major variation between the two
101
P. Perez et a/. /Agriculture. Ecosystems and Encironment 65 (1987) 95-lt16
Table l
CanPison of the annuaJ hy&o]ogical balances between the pilot basin, (ND; 90 ha) and tbe developed basin (KD; 60 ha)
--.
-
Penod
Year
ND basin
KD basin
-
-
-
-
Annual rain
Annual runoff
Runoff
Annual rain
Annual runoff
Runoff
(mm)
(mm)
coefficienl. (%)
(mm)
(mm)
coefficient (Sl
Before
1983
429.2
3.1
-
0.7 - -
391.4
6.8
1.7
1984
420.8
7.8
1.9
354.6
6.1
1.9
1985
493.1
14.3
2.9
589.5
18.3
3.1
1986
670.5
28.2
2.7
658.1
17.9
2.7
1987
767.8
19.2"
2.5"
663.1
5.0"
0.8" *
Total
2784.4
72.6
2.6"
2656.7
54.7
2.1b
Planning
1988
825.2
22.7
2.8
891.2
24. I
2.7
After
1989
631.9
5.7
0.9
722.0
7.4
I .O
1990
488.6
12.3
2.5
442.5
10.3
2.3
1991
455.6
22.8
5.0
498.8
14.5
2.9
1992
615.0
43.5
7.2
578.1
12.7
2.2
1993
741.9
119.9
16.2
799.0
43.0
i.-l
Total
2933.0
205.2
7.0"
3040.4
87.9
?.Y”
Watershrd management started in 1988.
“Miwng data.
“Mean value.
-----~-- ____ -.--.-.~--- _^_. --& -.-. - -.
-.-_--
..-_ y
/
-a--- KO cumulative runoff (1988 to 1993)
j
#
*--4
1
.__o- __ ND wmulative runoff (1988 to 1993) i
? ?
1
???
?
j ?? ND cumulative rainfall (1988 to 1993) 1
0
0
L
--_-
<B
-__-
-----
---_ -.--- ..- I-“-i- z-----.-----Y -
T----. -- ~-_~ ~~~~ --_
-8
500
0
d
*
8
’
*
’
b
Li_-_~_-_i---L--I<
.
,
/
i-.u- 0
0
500
1000
1500 *
2000
2500
KD cumulativt! rainfall (mm)
Fig. 4. Cumulative runoff amounts recorded from 1988 to 1993 ar, th:: cxutories of the control (ND) and developed (KD) basins. The ND
basin cumulative rainfall amounts are comparable with the KD basin.
102
P. Perez et al./Agriculrure,
&xqvstems and’ Environment 65 (1997) 95-106
rainfsill series. For this representation of the cumula-
wiith a runoff coefficient of 9% for the developed
tive runoff values, from 1988 to 1993, we extratted
basin (KD) and 11% for the control one (ND).
from the data set the events generating no flood in
Flloodings caused by the next two showers (very low
bath Twatershed, the missing or uncertain records and.
infiltration conditions) were markedly higher and
the errents for which rainfall amounts differed dore
intense in the Ndierguene watershed (KD: 10% and
t&n 40% between the two watersheds. One Can
14%; ND: 50% and 42%). The solid transports were
notice the increase, from 1988, of the variation ;be-
1.5 t/ha for the developed basin (KD) and 3.5 t/ha
tween the two basins.
,
for the contrai basin (ND).
in the years with high runoff, the conserva$on
meascres had a marked effect on the water balahce
3.2. Field nwnitoring
and the solid tmnsportsW For example, in 1991, lelels
of solid transports were the same in both basins
The results of the cadastral survey show a com-
(aboul 650 kgjha), but in * 1992, they represenked
mon trend. in both watersheds, to increase cropping
488 kj;/ha for the Keur Dianko watershed and more
areas. In 1990, cropping area represented 55% in the
than 1 t/ha for the Ndierguene watershed.
developed basin CKD) and 62% in the control one
Tht: impact of the watershed management I is
(ND) (Fig. 2). In Keur Dianko, even after 1988. frorn
clearlq demonstrated by an event that occurred ion
6 to 9 ncw farm plots were slashed from rangeland
4-5 August 1993 (Fig. 5). Three consecutive rain-
every year. During the same rime, 5 fïelds, seriously
storms totaled 179 mm in 24 heurs. The floods after
damaged by erosion, are abandonned in the control
the first shower were the same for the two basibs,
basin (ND) an4 no onz in the developed basin &CD).
0
e?I -...-. .~-.._--_-- --.-.
50
1
/
m
&J . . ..-__-_--
____ i’r-- . . -.---_---_--- -1
- - - - - _i-
100
/
?r
5
2
- -+- KD discharge
/
I
01
1
E
m
--o-- ND discharge
1
1
s
f 6r) ----
.i
’
.2
'so ?!
u
.ELm
iii
.a
- NR rainfall
,
2
---.----~-
g
4(1 T -_---- A..---~-+
-_ - -..:.Ly---- -.~------- - - - -
200
: ;
;
,
/
1 1
:
)
/
!
I
I
/
:
;
1
.Y
i
-.._
a
2tl 4’--le----- ----LA
-~
---_
250
t
j
è
I
\\
.
1 300
1932
oo:ao
04:40
09:36
14:24
19:12
0o:oo
Tipe (hh:mm)
Fig. 5. Rtinfall amounts (mm) and specific discharge (I/‘s/ha) during the 4th and 5th of August 1993. Comparison belween the control
(ND) and the developed (KD) basins.
P. Perez et ai. /Agriculture. Ecosystems and Erwironment 65 f 19971 Y5-- 106
1 on
nough some trials were condocted, eveqy year,
Table 3
with farmers of Keur Dianko, the proposed cropping
Average survival rates and average growing rates of the seedlings
-.~
pm&es; were diversly adopted. Contour cultivation
Date of Date of monitor&
increased from 20% to 80% of the flelds, between
p’lanting
Rainfall
Nb
1988
1989
1990
1991
1987 and 1993, thanks to the establishment of the
198X
894.2
1700
-
91
135
8 4
- -
]ive-hedges. But dry season decompacting, shallow
3 4 $ 1 0 79&35 137f58 145f66
ridging for groundnut trop or shallow earthing up for
1989
71!2.0
2100
-
‘74
7 0
31&11
ólt-28 63k43
millet trop increased only from 0% to 20% during
1900
442.5
4 1 0
-
-
-
3 6
the same time. Localized manure application reached
-
31*10
35*12
only 16% of the plots. TechnicaI and socio-eco-
nomic reasons of such a failure are analysed in a
Rcsults are givrn for each date of planting and each date ot
previous paper CPerez et al., 1996).
nionitoring.
liirst line: avcrsgc wrvival rate from the plantins l%); secou<:
Due to this weakness of spreading, but also to
lin<: average cumulative height from the planting (cm).
variable rainfall conditions, grain yields shoswed no
(Kamfall): annuel rainfall (mm); (Nb): numher of plants.
trend during the whole period (Table 2). Considering
Al! Oie specie< art m-r&.
annual mean yield or its coefficient of variation, it is
obvious that production remained at a very low
potential (inter-annual mean yield was 687 kg/ha
iutt:r’t: of 11w ii\\-e-hcdge. Once establishccf thc dearh
for bath crops) and that heterogeneity between the
rate of the plants was much iower. ‘Table 3 sho\\vs
fann plots tvas high. The surprising results achieved
rhat the ~ro~~~th r;tte depended aIsol on the planting
in 1988., drspite favorable rainfall amounts, were due
{conditions. Rcsults given for 1991 rnust be read with
to intense Ieaching and flowers aborting.
I::N~ because pruning activities began during tllc‘
3.3. Lir:e-lvedge rvvovvitoring
previ’ous year-.
In fact. these global results hide differences be-
The growth of the seedlings of shrub species was
tween species (Table 4). For example. Gliv-icidia
strongly linked with climatic conditions. In 1!388, the
sepiuvn gave bad results from the fîrst year (survival
exceptional precipitation permitted achieving an av-
rate: 33%) and Prosopis julifïora remained steady
erage survival rate of 9 1% (from 4700 plants:), but in
only one more year (survival rate: 43%). On the
1989 and moreover in 1990, for the last setting up of
other hand, Acacia nilotica and Bauhiniu rvq%escevvs
hedges, the survival rates dropped to 74% (from
confirmed. in local conditions, their strength and
2100 plants) and 36% (from 410 plants), respectively
their adaptability, described by ot.her authors (La1
(Table 3). It must be emphasized that neirher irriga-
nnd Steuiart. 1990: Vlaar, 1992). Farmers were ver-y
tion n(:>‘r protection were given to the Young shrubs.
sensitive to the survival rate of the plants. considered
The planting conditions were then crucial for the
3:; 3 mte of profitability of their work. But they
Table 2
Groundnut ( Aruchis h?ï>ogeo) and millet (Pennisetum fiphaïdes) annual grain yields
Year
Rainfall (mm)
Groundnut grain yield
-~
Millet grain yield
Nb
Mean (kg/ha)
c v (%)
Nb
Mean (kg/ha)
cv (%)
- - - - - -
1987
663.1
15
1069
45.3
1 0
821
43.0
1988
894.2
18
6 2 9
39.0
7
4 1 2
65.0
1989
722.0
19
9 2 9
30.6
1 1
6 8 2
4.6
1990
442.5
28
4 8 3
43.7
17
4l8
53.5
1991
498.8
26
5 6 9
43.3
2 1
5 2 7
46.2
1992
578.1
24
5 3 6
52.6
I I
7 5 7
27.1
1993
799.0
2 0
5 9 5
34.3
1 7
1163
33.6
@ainfall): annual rainfall; (Nb): number of plots; (Mean): average value; (CV): coefficient of variation.
Fiel& m’onitoring of the KD basin.
a
104
P. Ferez et al. /Agriculture. Ecosptems and Encironmenr 65 (1997,J 95- 106
Table 4
had been modified. In this case, the paired, watershec.
Avaage survival rate from che pianting (1988)
-
technique was the only experimental method tha,
Specas
Nb
Date of monitoring
could pive evidence of this result.
1989
19!xl
1s
‘1
Even in recent reviews on the subject (Amir.
1996; Unger, 19%) there are very few results of
Acacia niluticn
804 98
96
%
Bau&nia rufescens
155J 96
Y5
95
studies realised at this scale (1 km’) comparin_c
Dichwstachp glomerata
118 76
6 5
62
developed and control bnsins. At inferior scales (t
Glirkidia sepium
60 33
2 0
18
ha, 100 m2), many studies gave evidence of positive
Acucia mellifera
148
92
a 7
85
results - in terms of restoration -- of de\\~eiopment:
Parknsonia aculeata
43
9R
83
8 3
PiliO.itigma
reticulutum
641
8-l
83
IÏ
applied in comparable climatic and human condi-
Pr0stvi.r jufiflora
306
83
13
35
tions (t’laar, 1992; Roose. 199-l).
Acuciu senegal
76 87
8 3
82
The farmers took a polite interest in h>‘drological
Aaxb seFa!
177- 97
93
93
results but they were more responsive to the main
Zi:iflrus mauritar;ia
365
79
7 8
77
gully and adjacent tracks stabilisation. The relation
Total
4300
91
85
83
between rills, track and gullies became SO evident
Thesc are the main species plan& in 1988.
that every one tried to control the water\\\\.ays in bis
(Nb)r number of plants.
field with the help of stone pavments or uooci
trunks. On the other hand. despite the predictable
consequences, farmers continued to slaxh thc uppe;
noticed that it was usrless to gct two merers high
parts of the hillside. Between the land &=r~J:t,:icr~.
thonly shrubs to control runoff and erosion! FcIf this
and the social pressure on tenure, thev chose fhc
reasçn, a programme of pruning was sta.rted s ince
lesser of two eviis. This cari partly explain thc
1990.
limited effect of the conservation measures In thc
A~?other aspect of.tRe shrubs management wi ; thie
Keur Dianko watershed.
annu; replacing of the ones that died into the xist-
After eleven years of monitoring, ir s~ml; incrd-
ing hedges. From 1989 to 1991. the average r; e of
ible that we began the planning duriris thc tnc~J
substitution wus nearly 18%. For most of the S~I zies,
rainy year of the decade. By a mere chance. the
the survival rates in this case were equivale t to
major part of the plantings prolited h! i‘;.~~urabl<
those recorded in the case of a first planting For
conditions. What would have bcen thc situation. il’
exaniple, during the year 1989, -13 1 plants of A Cl rrcia
we had started two years later? The avc’r;tgt: sur\\ii.;li
dofica were newly planteci, with a survival r; It!e of
ral:e would have dropped to 40% and rhr furmt’r’!.
82Q1, while 154 plants were planted as substit:utes
enthusiasm would have disappeared. Te&>,. 1iL.e.
with 3 survival rate of S4Yc (Rautureau et al., 1 9 91).
hedges bave two main abjects for the fa:-mers: tts
control runoff (and run-on) during ttlc rainy srason
and to supply fodder during the dry st’ason. For bath
4. Dtwussion
abjects, they get into the habit of crcating swaths, :k[
the base of the hedges, with trop and weed residucs.
Since we expected hydrological effects of the
AI1 the labour time requirementfs were recorded
projec$ the fîrst results were quite disappointin$. It
(Ruelle et al., 1990). For the entire project planting
took iour, years to become obvious thàt we had to
hedges required 160 men X day, excluding the nurs-
analy:;e them in terms of conservation rather than
ery activities. Stone line and pavment building se-
restomtion. The soi1 and water conservation practjces
quired 50 men X day. This theoretically represents
only ljermitted the siowing down of some medha-
350000 FCFA (1700 FCFA/day in 198S), adding
nisms involved in the watershed degradation process.
to, approximately, 50000 FCFA for littlc equip-
For that reason, there is no rendency to decrease the
ments. The nursery yard required 10 men >: day for
global runoff ra.te of the developed basin (KD). But
preparation (fencing, digging) and 70 mett x day for
the increasing d.ifference recorded in comparison with
seeding, that is to say an amount of 14OQOO FCFA.
the control basin (ND) show-ed that the natural trend
SO, the cost of this collective work, done between
- “.
-
-
-<
-..p* - - -
-
-
-
.
P. Perez et ai. /Agriculture. Ecosystems arrd EncYronment 65 II 9971 95-106
10.5
1988 amd 1990, is 540000 FCFA. in fact, this is a
number of individuctl farmers and village communi-
theoretical amount because we didn’t pay for any
tics from the neighborhood are now asking for tech-
salary; but, on thti other hand, we didn’t incllude the
rticai assistance SO a!; to appiy conservation measures
cost of the support from the scientifïc staff.
on their land.
At the hillside scale, the collective benefit cannot
be detived from an increasing trop production, as we
References
have seen it. The positive results mentioned above
are the tracks and the fields stabilization. By compar-
Allwgcl. J.. Pcrcz. P.. 1993: Fonctionnement hydrologique et
ison wiiih the control basin (ND), we cari take as an
arnL:na(-crwnt tlcb hw-fonds des formations sédimentaires du
hypothesis that four farm plots were saved from
continental terminal: exemple du bassin arachider du Sénégal.
irrevers;ible damage. The average field area is 0.5 ha
In: Raunet. M. (Ed.). Bas-fonds et riziculture. CIRAD Publ..
hlontpellier- France. pp. I5S- 164.
(KD basin: 67 plots on 33 ha) and the average grain
Amir. J., 1996: Impact of trop rotation ànd land management on
yield is 700 kg/ha (ix pod yield is 1000 kg/ha).
wil erosion ami rchabilitation.
III: Agnssi. M. ( E d . ) , Soi1
Using the unit price of groundnut pods (1988: 90
c’rw~ion. c<>nwrvation and rehabilitation.
hlarcel Decker III~:..
FCFA/kg). WC cari estimate the benefit as 180000
Ntw \\1’orl, I.SA. pp .?75-399.
FCFA/year. Thc theoretical cost/benefit Ibalance
hecamc stcady after 3 years. These results are close
to thosc given by Vlxx ( 1992) lor Rochette (II 989).
Beyond this ovcr-simple demonstration, it is a
facr that somc of the 20 fxmers involved in the
watershed management felt somehow swindled. Even
though they agrced that there were actual affects.
BI)S~~. 1.51.. Hr‘\\vlztt. 1.1) , 1982: (t rc‘vic\\c of catchmcnt expcri-
thcy said they werc’ net totaly paid back for their
rnsnt~ to d~tcrminc th. c!tCct of \\‘egctar+on changes on water
contribution to the collective effort. This reaction is,
yirld and evapotranspiration.
J. Hfdrol., n”55. pp. 3-73.
partly, due to the relative failure of our cropping
Dacosta. H.. 1992: Economie de l’eau CI DRS sur les bassins
practices proposals. Facing the good results of the
versants de ThyssL: Kaymor. Syntht’se hydrologique 1983-
1988. ORSTOM/URZB, Dakar Scnegal. 67 p.
trials (Rrrez et al., 1996), it is clear that the major
Diatta. M.. 199-l: R&$nération des ZOIWS mlarginales et rôle de
error bas been to neglect the linkage between indi-
l’arbre dan.\\ les programmes de DRS, application aux bassins
viduai and collective actions, the fann plot amd the
versants de Thyss: Kaymor. Doctoral thesk, Un. Louis Pas-
hillside scales.
teur, Strasbourg France. 183 p.
iFritsch. J.M., 1992: Les effets du d6frichcment de la forêt atna-
zoniennc et dc la rnist: en culture >ur l‘hydrologie de petits
hakiins ver-ants.
Coll. Etudcz and Th&es. ORSTOXI E d . .
5. ConcEusion
Pari5 France. 391 p.
Growis. hl.. Alhergcl. J.. 19X9: l)u ribquc climatique à la con-
Various techniques aimed at controlling runoff
trainte écologique. Incidence dc la ~i;hercsw sur les pro-duc-
and stabilizing water erosion phenomena were ap-
tiens vé&~Ics et le mil@ PLI Burhina Faso. In: Eldin. ht..
Millcville, F. (E~S.). Le risque cn agriculture. ORSTOM Publ..
plied in a degraded part of the groundnut cropping
P&is Frwce. pp. 743-253.
basin of Senegal. The conservation measure!; were
Lai. R.. Stewart. B.A.. 1990: Agroforestry for soi1 management in
set up in different landscape units, with their ecologi-
tropics. In: Lai. R., Stewart, B.A. (Eds.), Soi1 degradation, A
cal features and human uses taken into account.
global threat. Springer-Verlag. New York USA, pp. XIIE-
Using the paired watershed technique, it has been
X V I I .
Le Borgne. J.. 1990: La dégradation actuelle du climat en Afrique.
demonstrated that this land management strategy had
entre Sahara et Equateur. In: Richard. J.F. (Ed.), La dégrada-
actual hydrological effects. After five years, the only
tion des paysages en Afrique de l‘Ouest. Ministère de la
agronomie effect was due to the land stabilization.
Coopération et du Développement. Pari5 France, pp. 17-36.
The synergy between runoff control and fertility
Lericollais, A.., 1990: La gestion du paysage? Sahélisation, surex-
restoration should have been emphasized to ensure
ploitation et délaissement des terroirs sereer au Sénégal. In:
Richard, J.F. (Ed.), La dégradation des paysages en Afrique de
farmer participation and benefïts sharing (Roose,
l’Ouest. Ministère de la Coopération et du Développement,
1994). But results are not insignificant: an increasing
Paris France. pp. 152-169.
1Oti
P. Pcrez et al. /Agriculture, Ecosystems cd Ensironmrnt 65 (1997195- 104
Le Thiec, G., Bordet, D., 1990~ Etude de deux outils de, traction
Roose. E.. 1993: Introduction à la gestion conservatoire de I’e:
animale visant à reduire les contraintes de travail du SOI en
de la biomasse et de la fertilité des sols (GCES). RuIl. PCC.~,
zones s&ches. In: Starkey. P.H., Faye, J. (I%ls.:), Animal trac-
de la FAC), n”70, Roma Italia, 520 p.
tion for agriculture development @roc. of WAATN workshop.
Ruelle, P., Sene, M.. Juncker, E.. Diatta, M., Perez. P. (Eds.
Saly. july 1988). WAATN Pubi., pp. 56-64.
1990: Défense et restauration des sols. Coll. Fiches Te<l
Perez. P., Sene, M., 199.5: Evolution des structures agmires’et
niques. vol 1 n“I. ISRA/LJNIVAL. Dakar Senegal. 64 p.
érosion dans le sud Saloum (Senegal). Reseau Erosion n”15,
Sene, IM., 1905: Influence de l’état hydrique et du comportemci
ORSTOM Montpellier France, pp. 59-68.
mécanique du sol sur l’implantation et la fructification
Perez. P., Boscher, C., Sene, M., 1996: L’amélioration dbs tech-
l’arachide. Doctoral thesis, ENSA Montpellier Frsn~e. 127
niques culturales pour une meilleure gestion de l’eau pluviale
Serre, hl.. Percz. P., 1994: Contraintes et possibililZs (Ic v:iIori’
(sud Saloum, Senegal). Agriculture and Développem$nt. n”9.
tion des ressources naturelles dans le sud du bassin ara&itli
20-79.
(Siné Saloum. Sénégal). In: Reyniers, F.N.. Netoycb. 1.. (Ed,.
Picrce. P.J.. 1991: Erosion productivity impact prediction. In: Lai.
Bilan hydrique agricole et sécheresse en Afriqtrtn trol*ical? t.
R.. Picrce. F.J. (Eds.). Soi1 management for sustainabil~ity. Soi1
John Libbey Eurotext, Paris France, pp. 217-233.
and water Conservation Society. Iowa USA. pp. 35-512.
Sircoulon. J.. 1993: Evolution des climats et des rc~\\ourc~~
Pieri. C. (Ed.). 1989: Fertilité des terres de savanes. Rilan de
eau. In: Pontie, G.. Gaud, hi. (Eds.1. L’cnvironncment i
trente ans de recherche et de développement agricole au sud
Afrique. Afrique contemporaine, no 161. La Documentation
du Sahara. CIRAD Publ., Montpellier France. 4l3 p. ~
Fran<aise. Paris France, pp. 55-76.
Poulscn, E., Lawesson, J.E. (Eds.). 1991: Dryland degradation.
Tyhirk. K., 1991: Planting trees in Sahel-doomcd to IXlurr‘. I
Causes and consequences. Danish Sahel Workshop. ~ Aarhus
Poulsen. E.. Lawesson. J.E. (E~S.). Dr+nd tlrgrad.rti~
University Prcss. Aarhus, 134 p.
Causés and consequences. Danish Sahel W’»rksh~!p. ,\\arhi
Rarrfurcau. J.. Ferez, P., Diatta, M., 1991: Implantation ct~ gestion
lrniversity Press, Aarhus. pp, 71-78.
de haie:,-vives. Rapport technique, CIRAD Puhl., hlontpellier
I -riFer. P.W., 1996: Common soi1 and water con~~n.a!iorr 1.r:.
Frmce, 52 p.
tices. In: A_oassi. hl. (Ed.), Soi1 erosion, conservation 21.
Richard, J.F. (Ed.), 1990: La dégradation des paysages en ;Afrique
rehahilitation. Marcel Decker Inc.. New York lS.4, pp. 77:
de l’Ouest. Ministère de la Coopération et du Développement,
399.
Paris France, 130 p.
Vlaar. J.C.J. (Ed.), 1992: Les techniques de conservati(ln de> Eau
Roehette, R.M. (Ed.), 1989: le Sahel en lutte contre la désertifica-
et des sols dans les pays du Sahel. CIEH/‘U,-X\\I’, L;Ai
tion. Leçons d’expérience. CILSS PAC. GTZ, 592 p. ~
W’agcningcn Nederlands, 99 p.
-
-_----