Agricultural Systems 79 (2004) 55–81
www.elsevier.com/locate/agsy
Spatial carbon, nitrogen and phosphorus budget
of a village in the West African savanna—I.
Element pools and structure of a
mixed-farming system
Raphaël J. Manlaya,b,*, Alexandre Ickowiczc,
Dominique Massea,1, Christian Floreta, Didier Richardd,
Christian Fellere,2
a
Institute for Research and Development (IRD, ex-ORSTOM), BP1386, Dakar, Senegal
Institute of Forestry, Agricultural and Environmental Engineering (ENGREF), BP 44494,
34033 Montpellier Cedex 5, France
c
Ecosystèmes Naturels et Pastoraux (ECONAP), CIRAD-EMVT, ISRA-LNERV, BP 2057, Dakar, Senegal
d
Projet Concerté de Recherche et Développement pour l’Elevage en Afrique de l’Ouest (CIRDES),
01 BP 454, Bobo Dioulasso, Burkina Faso
e
Centro da Energia Nuclear na Agricultura (CENA-USP), Institute for Research and Development
(IRD, ex-ORSTOM), Caixa Postal 96 13400-970 Piracicaba, SP, Brazil
b
Received 1 April 2001; received in revised form 12 February 2003; accepted 5 March 2003
Abstract
The viability of mixed farming systems in West African savannas relies largely on the
management of endogenous organic resources. Assessment of the organic balance at both plot
and village territory scales is needed as an indicator of this viability. Distribution of carbon
(C), nitrogen and phosphorus in soil and plant biomass was thus quantified for a village in
southern Senegal across the different land use systems (LUS) and farm holdings. The village
exhibited ring-like organisation, including a compound ring, a bush ring, and lowland paddy
fields, with positive gradients of intensification and food production from the savanna to the
dwellings. Marked contrasts were found between holdings, especially in livestock availability.
Clear relationships were evidenced between the spatial distribution of C and nutrients and the
* Corresponding author. On leave from ENGREF, France. Present address: IRD, MOST, BP 64501,
34394 Montpellier Cedex 5, France. Tel.: +33-467-416-267; fax: +33-467-416-294.
E-mail address: [email protected] (R.J. Manlay).
1
Present address: IRD, 01 BP182, Ouagadougou, Burkina-Faso.
2
Present address: IRD, MOST, BP 64501, 34394 Montpellier Cedex 5, France.
0308-521X/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0308-521X(03)00053-2
56
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
agricultural functions of LUS and holdings, while multi-scale diversity appeared to be the
main factor that ensured the functioning of the system. Intensification schemes at the village
level aimed at increasing organic resources and their cycling efficiency must thus take into
account their impact on this diversity.
# 2003 Elsevier Ltd. All rights reserved.
Keywords: Carbon; Fallow; Manure; Mixed-farming system; Nitrogen; Phosphorus; Plant biomass;
Savanna; Senegal; Soil
1. Introduction
Most traditional farming systems in West Africa—and especially those in the
savanna—share two main features (Ruthenberg, 1971):
1. Low requirements in exogenous inputs (fertiliser, energy, pesticide), often only
enabling extensive agriculture.
2. High social cohesion, leading to communal land tenure and management and
allowing for spatially restricted intensification.
To overcome energy, fertiliser and labour shortages, smallholders have generally
organised their village territory in a ring-like scheme. From the compounds to peripheral areas there is a decreasing gradient in agricultural intensification and land
tenure control, which usually allows the distinction of three main rings (Ruthenberg,
1971): (1) a compound ring, devoted to continuous cultivation, thanks to manuring
and spreading of household wastes, (2) a bush ring under less stable, semi-permanent cropping patterns of both cash and staple crops, (3) a forest or savanna ring,
uncropped for several decades and under common land tenure.
On the wettest fringe of the West African Savanna (WAS) belt, farmers have used
endogenous organic matter (OM) as a multi-purpose tool for decades and it can
thus be considered a resource. The organic matter pool is a continuum of carbon
forms from live biomass to humified compounds. It provides people with food, fuel
wood and construction materials. It supplies livestock with forage. In the sandy,
ferruginous soils that prevail in the WAS, soil organic matter controls many specific
chemical, physical and biological properties (Jones and Wild, 1975; de Ridder and
van Keulen, 1990; Feller and Beare, 1997).
However, such practices have only proved operational where the population density is low (van der Pol, 1992). Rapid population growth in the WAS and unsecured
national land tenure policies have resulted in an increasing need for land, wood and
forage, which calls into question organic fertility management practices such as fallowing and manuring and instead favours mining agriculture (van der Pol, 1992;
Ker, 1995).
Assessing organic resources at the field scale is the first step towards defining new
viable agricultural systems. Studies carried out in a village in the WAS belt in
southern Senegal have contributed to the understanding of the role of organic matter in sustaining the productivity of local agro-ecosystems (Manlay et al., 2002a–c).
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
57
However, more information—especially concerning the sustainability of the farming
system—can be obtained by studying carbon (C), nitrogen (N) and phosphorus (P)
dynamics at the village territory scale rather than at the plot scale (Landais and
Lhoste, 1993; Izac and Swift, 1994; Krogh, 1997). The reasons for this are:
. the functional connection between the land use systems (LUS) (bush and
compound rings), through livestock management for instance;
. communal control of fertility management practices and of land tenure, at
least at the level of the holding; and
. the need for accurate carbon and nutrient budgets representative of each LUS,
as inferred from plot budgets, and weighted by area, soil properties, and
cropping history.
Where better access to chemical inputs is impossible, quantifying carbon and nutrients
budgets at the village territory scale is the first step towards assessing the sustainability of
agricultural systems in the WAS. First, carbon and nutrients allocation among the various spatio-functional units must be analysed and this is the concern here. Second,
amounts of carbon and nutrients stored in the soil–plant system must be compared with
the use that is made of them (Manlay et al., 2003). These two studies are also relevant in
the context of global change, for they help estimate the capacity of savanna farming
systems to mitigate anthropogenic gaseous carbon emissions.
The aims of the present work were to (1) analyse the social, functional and spatial
organisation of a mixed-farming village territory in southern Senegal to determine
the influence it has on OM production and C, N and P storage, (2) assess the global
C, N and P budgets of this system related to land use and fertility management, and
(3) predict changes in the carbon balance in future years.
2. Methods
2.1. Site characteristics
The study was carried out in Senegal, in the village of Sare Yorobana, Department of Kolda, Region of High Casamance. Sare Yorobana is located at 12 490 N
and 14 530 W. In the village representatives of the common mixed-farming system
encountered on the southern fringe of the region can be found, but diversity in
population density, herd ownership and agricultural practices is the rule in High
Casamance (Fanchette, 1999).
A detailed description of the site with its climate, smooth topography, soil distribution, Sudanian-like native vegetation, and farming system can be found in
Manlay et al. (2002a–c) and Manlay (2000), and is summarised in Fig. 1. The climate
is tropical sub-humid, with a mean annual rainfall of 1120 mm (standard error: 35.3
mm) over the 1938–1997 period.
The village of Sare Yorobana belongs to the most ancient nucleus of Fulani
(Peulh) settlement in the region of High Casamance (Pélissier, 1966; Fanchette,
58
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
1999). Before the war of independence against their Mandingue masters in the midnineteenth Century, a distinction had already been made among Peulh people
between the nobility—the first, authentic Peulh migrants—and those they had
enslaved during the wars waged against other ethnic groups with the help of the
Fig. 1. Spatial organisation and land use in the village of Sare Yorobana.
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
59
Mandingues. These slaves were progressively culturally integrated into the Peulh
ethnic group, but formed a new caste, that of the captives (Pélissier, 1966). Although
slavery was abolished a long time ago, a hierarchy still remains between noble and
captive lineages and mixing seldom occurs between castes. At the present time
farmers have adopted a mixed-farming system associating diversified agriculture
(rainfed and flooded cereals, groundnut and cotton cash crops) and extensive
livestock raising. From June to November (rainy season and beginning of the
dry season), cattle are kept in the savanna adjoining the bush ring or moved to
the protected forest of Dioulacolon 20 km from the village. As crops are
harvested, cattle are left to graze freely during the day and stray onto crop residues
as well.
2.2. Characterisation of the farming system
2.2.1. Spatial organisation
Extensive cattle husbandry frequently spreads well beyond the village boundaries.
However only the territory owned by the community (in accordance with customary
law) was considered in this study. Mapping was initiated in 1996 (teodolith) and
updated in 1997 (compass, and a global positioning system for the palm grove—
Elaeis guineensis Jacq.). Fields (whether cropped or not) were identified as land
tenure units, while plots referred to land use (one field comprising at least one
plot, usually several, with in-between plot boundaries that change from one year
to another). Fallows less than 18 years old were mapped; fixing boundaries for
older stands was almost impossible, due to vegetation regrowth and unreliable
information from farmers, which evidenced weak appropriation of such plots.
Rice (Oryza sativa L.) fields were mapped as a unit due to the highly complex
land tenure.
The coordinate database was managed with the Atlas Geographic Information
System (SMI, 1993), with which topological variables were computed (surface, distance to the dwelling).
Plot history and management were characterised through on-field enquiries
among farmers. Cropping history, ownership, usership, and reasons for fallowing
were investigated, as well as the local typology of rainfed fields. Generally speaking,
this typology distinguished (1) fields under continuous cultivation of cereals (the
compound fields), crops being mainly pearl millet (Pennisetum glaucum L.), maize
(Zea mays L.) and sorghum (Sorghum bicolor L. Moench), and (2) the fields under
semi-permanent cultivation (the bush fields), groundnut (Arachis hypogea L.) and
cotton (Gossypium hirsutum L.). Although the criteria put forward to classify plots
as belonging to the compound or to the bush ring varied between farmers, enquiries
about the cropping history of each plot indicated that compound fields could be
defined as those that had been continuously cropped with cereals for the past 10
years.
These enquiries enabled computing of the cropping intensity or CI (defined as the
proportion of years during which the plot had been cropped over the last 10 years)
of each plot.
60
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
2.2.2. Social organisation
In local Casamance communities—as is mostly the case in the WAS—social
cohesion and decision making occur at the farm holding level, not at the household
level (Pélissier, 1966; Achterstraat, 1983). A holding is defined here as a community
of neighbouring people, who may or may not be related to one another, and who
share the same staple crop fields, granary, and, most often, herd. The village of Sare
Yorobana consisted of 17 holdings; one of them was not considered in the multivariate analysis since it had very small socio-economical weight and was too atypical. Multivariate analysis was carried out to establish an agro-social typology of
farm holdings in Sare Yorobana on the basis of population structure, means of
production and land use (the relative proportion of food crops ‘‘%foodS’’, cash
crops ‘‘%cashS’’, and fallow ‘‘%fallowS’’).
A census was carried out of the population of every holding. Four bulk class
variables were generated: total (‘Tot’), permanent (‘Prm’), seasonal (‘Sea’) and adult
(individuals aged over 14, denoted ‘‘Adu’’) population sizes. These bulk data were
used to generate: (1) the food needs indicator ‘‘Food’’, computed from the population size weighted as follows: 0.5 unit assigned to any inhabitant aged under 15, and
one unit to any adult, (2) the labour requirements indicator ‘‘Lab’’, derived from the
population size weighted by age and sex of individuals belonging to the holding as
follows (FIA-SLE, 1990): (a) person aged 15–59: female=0.5; male=1 (b) person
aged under 15: female=0.2; male=0.5.
Means of production investigated were: (1) livestock availability ‘‘Liv’’ expressed
in tropical livestock units (1 TLU=250 kg of live weight or LW), (2) ownership of
ploughs, hoes and sowers (pooled in one equipment variable ‘Equ’), (3) employment
of seasonal workers ‘‘Sea’’, (4) owned surface ‘‘OS’’.
Farm typology was defined using principal component analysis (PCA) with ADE
4 Software (Thioulouse et al., 1997). PCA is an unsupervised exploratory tool that
yields the simple identification of individuals sharing similarity among a large individualsvariables data set (Sokal and Rohlf, 1994). Identification is largely based on
the superimposition of the correlation circle of variables with the projection of a
compound replicate for a given principal component (PC) plane. Here the data set
consisted of 16 lines (holdings) and 10 columns (simple and mixed variables described
above, see legend of Fig. 5 for full list of variables and abbreviations). Some of the
variables were standardised with respect to population size variables Adu Lab or
Food.
2.3. Carbon, nitrogen and phosphorus storage for the village territory
2.3.1. On-field measurement of harvested plant biomass
All harvested biomass (cereal panicles, groundnut pods, cotton bolls) was weighed
on site at harvest time, plot by plot when possible. When harvested biomass could
not be measured in a plot (which happened for 9% of the village cropped area), the
weight was estimated from observation of crop residues, questioning the farmer, and
assessment of the level of fertilization. Water content was determined at the laboratory for one sample [Dry matter (DM) weight: 0.1–1.2 kg] per plot at the laboratory
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
61
(drying at 70 C to constant weight). Results for each plot were expressed as dry
matter (DM) of the harvested yield.
2.3.2. Dry matter, carbon, nitrogen and phosphorus storage in other components of
plant biomass
Regression relationships were used to estimate total plant biomass storage for
each plot. The yield of the harvested component in cropped plots and the length of
fallow in fallow fields were used as predictors for the estimation of plant biomass for
(1) stover, herbaceous advents, crop roots (down to 40 cm) in cropped plots, (2) tree
and grass layers, litter, stump, fine and coarse roots (down to 40 cm) in fallow
stands. Linear regressions were established for crops using data from Manlay et al.
(2002c) and from four added field plots, two being cropped with millet intercropped
with maize, and two being cropped with sorghum (for methods of sampling and
yield measurement see Manlay et al., 2002c). Stover and haulm yields were estimated
as linear functions (slope and intercept 6¼ 0) of panicle and pod yields (Table 1).
Mean values had to be used for nearly all other components (see the ‘‘Remark’’
column in Table 1).
Non-linear regression relations (logistic-like functions) were used in fallow fields
to predict C, N and P storage in plant biomass as a function of the length of fallow
(Manlay et al., 2002a); when regression parameters had no statistical significance,
mean values reported in both previously cited works were used.
Estimations of the yields of above- and below-ground biomass (AGB and BGB)
of bush regrowth were based on field observations, and were related to cropping
intensity CI in the following way (with CIt=0.75 being a threshold value of CI set
from qualitative observations): CI > CIt: woody AGB, coarse root and stump biomass equal to 0; CI= < CIt: AGB=0.32 t ha1, coarse root biomass=3.38 t ha1,
stump biomass: 7.48 t ha1 (derived from Manlay et al., 2002a). Since the CIt value
was derived from a partly subjective choice, a sensitivity test was applied by recomputing total amounts of plant biomass at the village scale after modifying CIt
initial value by less or more than 10%.
Carbon, nitrogen and phospherus contents from Manlay et al. (2002a,c) were then
applied to convert DM into C, N and P values.
Because of the lack of data for the palm grove and some of the sorghum and
cotton biomass components, data were also used from the literature (see Table 4,
footnote). Local palm tree density (138 trees ha1, unpublished data) was very close
to the value reported by Jaffre (1984). Some of the biomass estimates for Sare
Yorobana were consequently derived from his work in Côte d’Ivoire.
2.3.3. Carbon, nitrogen and available phosphorus storage in soils
Soil C, N and available P (Olsen method modified by Dabin, 1967, denoted POD)
amounts were computed as discrete functions of land use, land management, vicinity
of a compound, and cropping intensity (see decision rules in Table 2). The decision
rules were established from measurements from 25 plots across the village territory
(crop and fallow) from Manlay et al. (2002b,c). Figures for C, N and P pools in the
soil were estimated down to 20 cm only, since it was found that C, N and P values
62
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
Table 1
Relationsa between harvested yields of crops and other crop components
Crop
Component
Groundnut Haulm
Advent bush
Weed
Fine root
Coarse root
Millet
Stover (pure stand)
Weed (pure stand)
aa
0.967
0.366
0.00579
0.153
3.587
0.770
Stover (milletmaize) 3.310
Weed (milletmaize) 0.830
Fine root
0.389
Coarse root
0.173
ba
Mean F
(t ha1) (t ha1)
0.602
0.0183
0.490
0.425
1.07
0.353
0.485
0.569
0.231 7.53
0.7477 0.818
0.1408
3.318
0.123
0.0764
R2
n
Remarkb
40.92 *** 0.61 28
1.8
0.06 28 Mean used
0.0
0.00 28 Mean used
4.5
* 0.17 24
Set to 0
16.7
10.3
4.112
2.322
0.914
0.275
9.6
0.5
5.2
0.5
53.3
0.0
0.4
0.0
** 0.54 16
** 0.42 16 Regression used:
Weed=0.628 *
Panicle (F:14.7**)
* 0.62 8
0.08 8 Mean used
* 0.27 16
0.03 16 Mean used
Maize
Stover
Weed
Fine root
Coarse root
0.812
0.0152
0.0170
0.00041
0.877
1.245
0.397
0.0196
3.43
1.175
0.319
0.0215
*** 0.79 16
0.00 8 Mean used
0.06 8 Mean used
0.00 8 Mean used
Sorghum
Stover
Weed
3.404
0.337
1.146
2.124
9.80
1.27
6.9
1.2
* 0.54
0.17
Rice
Stover
Weed
Fine root
0.624
0.073
0.591
1.462
0.027
5.026
3.306
0.244
3.277
16.8
6.0
12.5
** 0.74
0.50
* 0.68
Coarse root
0.022
0.132
0.067
15.8
** 0.72
8
8 Mean used
8
8 Mean used
8 Valid for 1.16<
Panicle<5.38
8 Valid for 1.16<
Panicle<5.38
a
Model tested: Yieldcomponent=aYieldharvest+b. ‘‘Harvest’’ stands for: panicle for millet, sorghum
and rice, ear for maize, pod for groundnut.
b
Model adopted for calculation of plant biomass when the linear model tested was not statistically or
agronomically suitable. Fine roots: diameter ranging from 0-2 mm. Coarse roots: diameter over 2 mm
(stump not included). p{Ho: F=0}:*< 0.05; **<0.01; ***<0.001.
were only slightly influenced by land use below this depth (except for P in the
compound ring).
2.3.4. Establishment of a simplified agropastoral budget
The self-sufficiency of each of the 17 holdings with respect to forage and manure
was estimated for the 1996–1997 dry season. Self sufficiency in forage was defined
for each holding as the following ratio:
SSFH ¼
AFPH
NFI HSH
ð1Þ
63
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
Table 2
Quantities of C, N and available P (POD ) stored in soil used for the calculation of budgets at the village
scale, related to land use, land management ring, vicinity of a compound, and cropping intensity (CI)
Land use
Ring
Adjacent to
compound
CI
Fallow
Rice
Other
Compound
Bush
Yes
No
40.5
>0.5
C
(t ha1)
N
(kg ha1)
POD
(kg ha1)
15.4
41.2
1282
3796
7.2
56.3
21.3
14.5
2000
1175
77.8
17.9
12.9
11.7
989
989
6.4
6.4
Layer considered: 0–20 cm. Source: Manlay et al. (2002b,c).
where SSFH is the potential self sufficiency value in forage (unitless), AFPH is the
available forage production of the holding (defined as the sum of the biomass of
stover and weeds in cropped fields, and of herbaceous biomass in fallow plots
belonging to the holding) (in t DM year1), NFI is the normalised forage intake by
livestock (15.4103 t DM year1 per kilo of metabolic weight—noted MW—,
Ickowicz et al., 1998), and HSH is the herd size (computed in kg MW).
Self sufficiency in manure was defined for each holding as the following ratio:
SSMH ¼
NFP HSH
MMRCP ACH
ð2Þ
where SSMH is the potential self sufficiency value in manure (unitless), NFP is normalised faecal production (8.74103 t DM kg MW1 year1, Ickowicz et al., 1998;
since part of the manure produced during day straying is returned to fields not
cropped with cereals, NFP is an overestimate of actual amounts of manure returned
to cereal fields, and SSMH is an indicator for potential self sufficiency in manure),
HSH is defined as previously, MMRCC is the minimum manure requirement for a
cereal crop (2.5 t DM ha1 year1, de Ridder and van Keulen, 1990; Berger, 1996)
and ACH is the area cropped in cereal (in ha).
2.4. Prediction of changes in the carbon status of the village
To predict changes in the organic status of the village territory, a simplified
representation of its dynamics was made using a spreadsheet relating C storage to
land use, which in turn was linked to manure availability and farmers’ needs (Fig. 2).
Because of assumptions made in the model, the simulation was run over a 15 years
period only (1997–2012). Initialisation was made for year 1997 as a simplified
representation of the territory from data presented here and in Manlay et al. (2003).
64
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
Fig. 2. Flowchart diagram of the spreadsheet model for the prediction of the carbon budget of the village
of Sare Yorobana (1997–2012). AG, above-ground; BG, below-ground; C, carbon; DM, dry matter; MI,
manuring intensity; TLU, tropical livestock unit. y: measured in 1997.
Main initial simplifications consisted in pooling millet, sorghum and maize in a single
land use denoted ‘‘rainfed cereal’’ and considering groundnuts as the only cash crop.
Annual growth of the population was set to 2.5% (UNDP, 1999). The yield of
rainfed cereals was set as a linear function of manure availability (estimates for
regression parameters were computed as a function of the fact that DM grain yield
of cereal in all manured plots averaged 0.79 t ha1 in 1997 for an available amount
of manure equal to 1.99 t DM ha1, while non-manured plots yielded only 0.66 t
ha1). A threshold was set to 6 t DM of manure ha1, beyond which yields could
not increase (Gueye and Ganry, 1981; Fernandes, 1999). In the case of land shortage, priority was given to the satisfaction of the need for food, followed by the need
for cash.
The following variables were assumed to remain constant compared with values
derived from the actual situation in 1997 (this study and Manlay et al., 2003): need
for food and cash, groundnut yield, flooded rice production, ratio of the area of
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
65
fallows less than 10 years old (labelled as ‘‘Young fallows’’) to that of rainfed cropped plots, and TLU availability per capita (the control of livestock demographic
growth by forage availability, which would have required a more complex, at least
spatially explicit model, was not taken into account).
The palm grove was not considered in the model, since practically no real carbon
measurements were made in it, and also because it is unlikely to be cropped in the
coming years due to poor soil fertility (see Section 4.2.3).
Output variables were land-use distribution (among cereal fields, groundnut fields,
young and old fallows) and C storage in the plant–soil system (above-ground biomass, below-ground biomass and plant biomass+soil).
3. Results
3.1. Spatial organisation of the farming system
Sare Yorobana extends over 256 ha, of which 46% were fallow, 25% under staple
crops and 17% under cash crops. Land use and management were clearly related to
distance to the village and to geomorphology (Figs. 1 and 3).
Four LUS owned by the village people (thus not including the forest and savanna
ring) could be distinguished:
1. The bush ring, mainly located on the plateau and covering nearly 75% of the
surface owned by the village (SOV). The bush ring was the farthest ring from
the dwellings. Only slightly more than 40% of this land was cropped (60% of
which as cash crop), the rest being under short or long fallow. Various reasons
for fallowing were given during enquiries and two kinds of fallows could be
roughly distinguished. On the one hand, 60% of the area under fallow
remained uncropped due to the relocation of the dwellings from the edge of the
Fig. 3. Spatial distribution of land use as illustrated by the distance of crops to the compound owning them.
66
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
plateau to the mid glacis as a consequence of lower annual rainfall since the
early 1980s. On the other hand, other reasons put forward for keeping land
under fallow were fertility management (55%), and labour and seed shortage
(35 and 23%) (one or more reasons were cited by smallholders).
2. The compound ring was limited to the glacis and represented less than 10% of
the total surface area of the village, but a third of the whole surface of the village under cereals. The choice of the cereal crop was mainly influenced by the
distance from the plot to the dwelling (Fig. 3), the closest plots being planted
with maize either in pure stands or mixed with millet.
3. The palm grove, down-slope from the glacis, covered 12% of the SOV.
4. Paddy fields represented only 6% of the SOV, but accounted for a fourth of
the area cropped with cereals. Water management was restricted to light
embankments, since usually only women grow rice.
Fallowing (whose intensity is the inverse of cropping intensity) and manuring were
spatially dissociated among village plots (Fig. 4). Most plots under regular fallow were
not manured, while only continuously cropped plots received the highest manure inputs.
3.2. Social organisation
Adults accounted for less than half of the permanent population of Sare
Yorobana (Table 3). Additional seasonal workers contributed significantly to real
Fig. 4. Spatial complementarity between cropping intensity and manuring in the mixed-farming system
in the village of Sare Yorobana. *Rates applied during the 1996–1997 dry season (night corralling only).
yDefined as the rate at which the plot was cropped between 1987 and 1997.
67
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
Table 3
Elementary statistics for 16 holdings in the village of Sare Yorobana
Variable
Population (human beings)
incl. adults
Labour power (working unit)b
Livestock (TLU)c
Land management (%surface)
food crop
cash crop
fallow
Owned land (ha)d
Equipmente
Yield
groundnut (t pod ha1)
millet (monocropped)
(t grain ha1)
millet (mono- and intercropped)
(t grain ha1)
a
b
c
d
e
All holdings Noble holdingsa
Captive holdingsa
Mean
SE
Median Range
Mean
SE Mean
SE
15.9
2.3
7.7
1.0
7.3
1.0
26.0
8.2
13.5
8.0
7.6
14.3
3–32
19.0
3.9
2–16
7.7
1.0
1.5–15.2 8.7
1.7
0–99.6 54.8
14.6
14.1
2.8
7.7
1.5
6.5
1.3
8.6
4.1
268
129
124
415
34
3
34
3
32
5
8.8
1.5
4.3
0.9
33
38
26
7.6
3.5
16–70
37
3
9–56
40
1
0–73
22
3
1–22.6 11.6
2.8
0–14
7.2
1.6
32
5
30
5
38
8
7.2
1.5
2.6
0.5
30
32
38
144
76
1.28
0.13
0.91
0.13
1.30
0.83
0.4–2.3
0.3–2.1
1.31
0.11
1.17
0.26
1.26
0.21
0.75
0.12
1.03
0.12
0.91
0.4–2.0
1.26
0.2
0.96
0.25
Village
See Section 2.1.
See Section 2.2.2 for definition of working units.
Tropical Livestock Unit; 1 TLU=250 kg of live weight.
Per holding (does not include oldest fallows owned communally by the village).
Defined as the sum of the number of ploughs, hoes and sowers held by the holding.
manpower (15%, data not shown). Vast areas of land on the plateau that was
exploited by, but was not belonging to the village, had to be taken into account to
compute meaningful human and livestock densities. For instance, the total surface
area explored by cattle during the dry season to meet their forage needs was 812 ha
(Manlay et al., 2003). As a result, realistic densities averaged 33 inhab. km2 and 51
TLU km2.
A wide range of values was found among holdings for various variables (Table 3)
and indicators (see Section 2.2.2). In the PCA, land (‘‘OS’’, ‘‘OS:Adu’’) and livestock (‘‘Liv’’, ‘‘Liv:Adu’’) availability, and the balance between labour availability
and the food needs of the permanent population (‘‘PrmLab:PrmFood’’) (PC1, Fig. 5
a) provided the highest loadings of PC 1. They could thus be interpreted as the
major sources of variability among holdings. The second PC was a gradient of
cropping intensity (highest loadings by the share of land under fallow or cash crops).
Comparing analysis of Fig. 5a and b (as well as statistics in Table 3) indicates that
holdings belonging to noble lineages were generally characterised by higher land and
livestock availability, use of seasonal workers and share of cropped area, and a
lower ‘‘labour availability: food need’’ ratio of the permanent population. A high
level of variability was found among holdings belonging to captive lineages.
68
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
Fig. 5. Principal component (PC) analysis of the structure of 16 out of the 17 holdings in Sare Yorobana:
(a) correlation circle of the variables, and (b) projection of the compound replicates on plane PC12.
‘‘%cashS’’: proportion of surface area devoted to cash crops (in% of OS). ‘‘%fallowS’’: proportion of
surface area devoted to fallow (in% of OS). ‘‘%foodS’’: proportion of surface area devoted to food crops
(in% of OS). ‘‘Capt’’: ‘‘captive’’ holding. ‘‘Equ:PrmLab’’: equipment per permanent labour unit. ‘‘Liv’’:
livestock (in tropical livestock units or TLU). ‘‘Liv:adu’’: livestock (in TLU) per adult capita. ‘‘Noble’’:
‘‘noble’’ holding. ‘‘OS’’: owned surface area. ‘‘OS:adu’’: owned surface area per adult capita.
‘‘PrmLab:PrmFood’’: permanent labour unit per permanent feed unit. ‘‘SeaLab:TotLab’’: ratio of seasonal labour unit to total labour unit. See Section 2.2.2 for the calculation of ‘‘Equ’’, ‘‘PrmFood’’,
‘‘PrmLab’’, ‘‘SeaLab’’ and ‘‘TotLab’’ variables.
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
69
3.3. Dry matter, carbon, nitrogen and phosphorus storage at the village scale
3.3.1. Manageable amounts of carbon, nitrogen, and phosphorus at the village
scale
Carbon stored in plant biomass and in the 0–20 cm soil layer averaged 29.7 t ha1
(Table 4). Sixty per cent of plant C was stored above ground. Vertical allocation was
as follows (in%): soil: 52, woody AGB: 22, stump: 8, coarse root: 8, the rest consisting of fine roots, herbaceous AGB (crop and weeds) and litter. Above-ground
biomass of cereal crops (not including pure maize and rice) reached 1.89 t C ha1,
with a low harvest index (millet: 0.13). The uncropped area of the bush ring was the
main carbon pool (60% of total C stored in the village) (Table 4). The highest carbon densities were found in rice fields (44 t ha1, 94% in the soil) and in fallow
stands on the plateau (39 t ha1, 40% in the soil).
The mean amount of nitrogen was 1.52 t ha1, mainly stored in the soil (87%)
(Table 4). The second contributor was tree AGB, which represented only 6% of the
total stock. Fallows in the bush ring accounted for 46% of nitrogen stored in the
village, the other main spatial pool being cropped fields in the bush ring. Highest N
spatial densities were found in rice fields (3.8 t ha1) and in fallow fields in both bush
and compound rings (range: 1.5–1.6 t ha1).
Phosphorus (Ptotal in plant biomass plus POD in soil; palm grove not included)
amounts showed somewhat different vertical and horizontal distribution patterns
(Table 4). Mean stocks amounted to 28.6 kg ha1 in the whole village territory.
Main reservoirs were the soil (44%) and tree AGB (26%). Other components each
accounted for less than 10% of total amounts. Though low (4%), the contribution
of harvested biomass (panicle, ear, and pod) was four times higher when expressed
in P than in C or N units. More than half of P was stored in fallows in the bush ring.
Highest P spatial densities were found in rice (63.2 kg P ha1) and compound (38.5
kg P ha1) fields, and in fallows on the plateau (32.5 kg P ha1).
Modifying the cropping intensity criteria CIt used to assess tree biomass in cropped fields by 10% did not much affect the C budget of the cropped ring (data not
shown), except for the BGB component (+15%). When non-cropped hectarage was
included, variations did not exceed 2.5%.
3.3.2. A simplified agro-pastoral budget
Self-sufficiency in manure and forage in the 17 holdings varied considerably and
was related to herd size (Fig. 6). Eight of the holdings, each owning at least 30 TLU,
were able to meet their need for manure in 1997, but the biggest livestock owners
were not self-sufficient in forage. And only four holdings produced both enough
manure and forage.
3.4. Outlook on future carbon stocks
In the simulation old fallows would disappear by the end of the period 1997–2012
(Fig. 7). Enough groundnut and cereal fields would remain to satisfy food and cash
needs up to then.
70
Table 4
C, N and P amounts in some agro-ecosystems in Sare Yorobana with respect to ring management and land use
Component
Bush ring
fallow
Surface (ha)
117
10.6
1.3
–
–
0.7
4.3
4.1
0.8
15.4
37.2
4337
Nitrogen (kg ha1)
Tree
Grass/weed
Harvest
Stover
Litter
Stump
Coarse root
Fine root
Soil
Total
Total (t)
118
23
–
–
11
46
43
20
1282
1543
180.1
groundnut cotton
35
sorghum millet
7
4
0.1a
0.0
0.2
0.4
1.5
0.5
0.3
0.2
11.7
14.9
336
0.0
0.2
0.6
0.6
0.6
0.7i
0.1b
0.4
0.2
1.3
1.0
0.4
0.2
12.1
15.1
528
1.9
0.9
0.4j
11.9
16.6
113
1.4
0.7
0.3k
11.8
16.2
60
5
5
37
31
10a
21
26i
9
4
10
989
1090
38.1
18
8
10j
989
1081
7.4
7b
16
8
9
12
8
10
989
1061
3.9
23
2
9
13
11
4
5
6
989
1039
23.5
ring
187
6.6
0.9
0.2
0.4
0.4
3.1
2.7
0.6
14.1
29.0
5405
76
17
10
8
7
32
29
16
1173
1367
255.0
millet
11
Palm grove Rice field Village
Milletmaize maize
6
2
0.2
0.4
1.5
0.8
0.4
1.3
0.3
0.4
0.6
0.1
0.2
15.2
17.7
187
0.1
0.2
16.8
19.5
108
0.0
0.1
18.1
19.6
35
9
14
11
30
12
9
18
14
15
3
6
1263
1305
13.8
3
5
1451
1510
8.4
0
4
1611
1663
3.0
ring
21
0.0
0.4
0.4
1.3
0.0
0.0
0.1
0.2
15.9
18.2
390
0
15
14
12
0
0
2
6
1341
1391
29.8
32
16
13.8c,d,e
1.1g,h
–
–
0.7h
2.4c,e
6.6c,e
9.9l
34.5
1101
249
20h
–
–
11h
42c
118c
860l
1300
41.5
0.1
0.4
0.7
0.0
1.3
41.2
43.7
710
2
8
9
1
32
3796
3849
62.4
256
6.5
0.8
0.2
0.4
0.4
2.2
2.3
1.4
15.4
29.7
7607
86
16
9
8
6
23
27
29
1314
1517
388.8
(continued on next page)
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
Carbon (t ha1)
Tree
Grass/weedf
Harvest
Stover
Litter
Stump
Coarse root
Fine root
Soil
Total
Total (t)
Compound ring
Table 4 (continued)
Component
Bush ring
fallow
Compound ring
groundnut cotton
sorghum millet
ring
millet
Palm grove Rice field Village
Milletmaize maize
ring
1
13.8
2.1
–
–
0.0
3.2
3.5
1.1
7.2
30.9
3.60
0.5
0.3
2.2
1.5
0.0
0.2
0.4
6.4
11.6
0.41
1.0m
3.9
6.9
0.7b
2.3
1.6
2.3
0.2
1.2
2.6
2.8
0.0
0.5
4.4
6.4
22.9
0.16
0.0
0.5
0.7
6.4
14.6
0.05
0.0
0.3
0.4
6.4
13.9
0.31
8.8
1.6
0.9
1.0
0.0
2.0
2.3
1.0
6.9
24.5
4.57
1.3
2.6
2.8
4.2
2.4
2.3
4.5
2.4
3.6
0.2
0.4
24.3
31.6
0.33
0.2
0.4
37.9
47.4
0.26
0.0
0.2
49.5
60.3
0.11
0.0
2.3
2.5
2.9
0.0
0.0
0.2
0.7
30.0
38.5
0.83
nd
nd
–
–
nd
nd
nd
nd
nd
nd
nd
0.5
1.8
2.2
0.1
2.3
56.3
63.2
1.02
7.3n
1.6
1.1
1.2
0.0
1.7
1.9
1.0
12.7
28.6
7.34
Figures for plant below-ground and soil components down to 40-cm and 20-cm depths, respectively. Fine roots: diameter ranging from 0 to 2 mm. Coarse
roots: diameter over 2 mm (stump not included).
a
For all plant components, C and N contents taken from Jacquemard (1995).
b
Element contents of plant components set to those of millet plots.
c
DM biomass or N content from Dufrène et al. (1990).
d
DM biomass from Jaffre (1984).
e
DM biomass or C content from Jacquemard (1995).
f
Carbon content set to that of grass biomass of fallows aged 1 to 9 years from Manlay et al. (2002a).
g
DM biomass from Blanfort (1991).
h
DM biomass or C and N contents derived from Manlay et al. (2002a).
i
DM biomass from Déat et al.(1976).
j
DM biomass from Poulain (1977).
k
DM biomass from Chopart (1980).
l
Baldensperger et al. (1967) and Manlay (2000).
m
All total P contents from Déat et al. (1976).
n
For P, not including the palm grove.
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
Phosphorus (kg ha )
Tree
Grass/weed
Harvest
Stover
Litter
Stump
Coarse root
Fine root
Soil (avail. P)
Total
Total (t)
71
72
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
Carbon stocks in plant AGB and BGB, and in the whole plant-soil system, would
decline sharply over the next 15 years (Fig. 8). Amounts of carbon in AGB will have
decreased by 50% in 15 years’ time, resulting in a loss of 0.26 t C ha1 year1.
However, this loss would only affect the woody component since C amounts in the
herbaceous layer would increase from 0.98 to 1.19 t C ha1 during the simulation
period. The loss would reach 0.20 t C ha1 year1 for BGB. Carbon loss for the
whole system (including soil) would amount to 24%, but soil contribution to C
depletion of the village system would be only 0.02 t C ha1 year1.
4. Discussion
4.1. Social organisation and dynamics of the farming system
History and inherited know-how account for the concentration of livestock ownership among holdings belonging to noble lineages in Sare Yorobana (Pélissier,
1966; Fanchette, 1999). Two significant social features of noble holdings—resulting
from a significantly higher standard of education than among captive ones—influence their farming systems: higher cash needs, and frequent migration of young
people to town. Satisfying their high food and cash needs with the limited labour
available is a major challenge. Such constraints are usually overcome with the help
of livestock ownership, which enables: (1) high manuring potential allowing good
cereal yields at a low cost of labour, (2) a good supply of animal labour, which is
crucial at ploughing, sowing and weeding bottlenecks, (3) consistent working capital
Fig. 6. Self-sufficiency in manure and forage availability in holdings of Sare Yorobana as derived from
simplified agro-pastoral budgets. *Tropical livestock unit=250 kg of live weight.
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
73
Fig. 7. Predicted changes in land use in the village of Sare Yorobana for the 1997–2012 period. See
description of model in text. Young fallows: aged 0–9 years. Old fallows: aged 10 years or more.
Fig. 8. Predicted changes in carbon storage in the plant–soil system (only soil layer considered: 0–20 cm)
in the village territory of Sare Yorobana for the 1997–2012 period. See description of model in text.
74
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
enabling better equipment and the employment of a seasonal work force. Holdings
with small herds (mostly captive holdings) on the other hand, rely on permanent
high labour availability and fallowing (spatially complementary to manuring).
Agro-pastoral budgets testify to complementary relationships between holdings
owning small or large numbers of cattle (Fig. 6). Livestock-mediated organic matter
exchanges generated by browsing of crop residues and manure deposit during the
day are subject to strict common rules, such as the ban on removing crop residue
from a neighbour’s plot in order to feed one’s own herd in the farmyard. Nevertheless, organic transfers lead to misbalanced nutrient exchanges, due to the absence
of night corralling agreements between holdings, de facto leading to nutrient concentration in compound fields of big cattle owners (Dugué, 1985); the latter thus
maintain their animal wealth—i.e. capital and means of production—at low cost,
with the help of a resource created by the work of others.
The balance of power between the holdings that own the forage and those that
own the animals may be reversed in the future, since the privatisation of land and
organic resources that is usually observed in the case of increasing population densities invariably calls into question the practice of straying and can lead to reduced
cattle ownership (Lericollais and Milleville, 1993; Dugué, 1998).
4.2. Land use systems: agro-ecological functions and carbon, nitrogen and
phosphorus status
4.2.1. The bush ring
The bush ring is the most distant ring from the dwellings and has poorly secured
land tenure; it therefore receives little attention with respect to the management of
its fertility, as testified by scant manuring (Manlay et al., 2003). Highest C, N and P
densities (per ha) recorded in the plant biomass illustrate the three main agro-ecological functions of the bush ring. First, it supplies livestock with herbaceous forage—from crop and fallow stands—and tree forage (Richard et al., 1991), and
farmers with wood and some of the crop harvest. Second, it replenishes its own fertility despite nearly no exogenous mineral input, thanks to fallowing. Plot studies
suggest that this aptitude relies on the proper maintenance of perennial rooting
systems (Manlay et al., 2002a,b). Fallowing may also be the result of lack of labour
or of available seed, or even strategies for securing land tenure (Fanchette, 1999).
Third, the bush ring acts as a source of nutrients for the compound fields thanks to
organic transfers mediated by livestock and crop and wood harvest (Manlay et al.,
2003).
4.2.2. The compound ring
Positive gradients of investment in fertility and labour—and in land tenure security—from the bush to the compound ring have been widely reported throughout
West Africa (Kowal and Kassam, 1978; Prudencio, 1993; Peters and Schulte, 1994).
The compound ring ensures a secure supply of food to smallholders. Intensified
fertility management enables steady production of food in the form of cereals,
mainly through organic practices (manuring during night tethering and day straying,
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
75
spreading of household waste, recycling of cooking stove ashes). As a result,
amounts of available soil P are much bigger in the compound ring than in any other
LUS, except rice fields (Table 4). A careful choice is made at sowing between three
cereals (maize, millet and sorghum) to overcome the tide-over period at the end of
the dry season. Maize for instance has the shortest growing cycle, and highest
nutrient requirements (Kowal and Kassam, 1978). It is thus sown closest to the
dwelling, where soil chemical properties are at their best, and care of plants is easiest
(Fig. 3).
4.2.3. The palm grove
Though spatially restricted, the palm grove represents an important source of
construction wood and forage. But the feed value of the herbaceous biomass is poor
and is not available during the cropping season. Palm trees are grown to provide oil,
and their use to meet wood needs is thus restricted. The palm grove is unlikely to be
converted to cropping at a future date, due to high soil constraints (chemical fertility
and the risk of erosion), but here and there orchards are gradually replacing ageing
palm groves.
4.2.4. Rice fields
Rice fields represent a source of steady food production with only little investment
for labour and fertility management, thanks to high inherent soil fertility and a free
supply of water and nutrients during seasonal flooding. However, paddy production
for home-consumption is available in December only because of the long growing
cycle of local rice varieties. Rice stover is of low foraging quality, but the limited
grass re-growth attracts livestock during the dry season.
4.3. The limited current carbon storage capacity of the village farming system
Few carbon budgets have been assessed at the micro-regional scale for dry Africa,
and sampling schemes often differ from one to another, thus limiting possible comparisons of soil data. Furthermore, the present study was solely concerned with soil
layers whose organic status would only be affected by land management; consequently assessment of C stored in the whole soil component was not undertaken.
The mean amount of C in plant biomass in Sare Yorobana (14 t C ha1) was
similar to that found for farming systems in the East African highlands (12.6 t C
ha1, Woomer et al., 1998). Figures for uncropped areas of the village territory
compared with those of dry woodlands and savannas in South Africa (22.4–27 t C
ha1, Woomer, 1993). Yet they were much lower than those reported for dry
savannas and forests (50 t C ha1, Tiessen et al., 1998), reflecting both local soil
restrictions (Manlay et al., 2002b) and the intensity of disturbances caused by
human activities such as biomass clearing or burning.
The fast carbon depletion of the village system predicted by the model (Fig. 8)
results from the conversion of old fallow stands to young ones, and from young
fallows to cropping and the subsequent drop in C stored in woody and in fine root
components (Fig. 7). At the same time, soil may contribute little to the variation,
76
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
and may only occur in young fallow stands replaced by cash crops. Later, the rate of
carbon depletion in the village territory is expected to decrease, as the quantity of
biomass to be cleared from remaining ecosystems is much lower than that of mature
fallows, and as soil organic carbon (SOC) amounts in plots cropped with cereals are
slightly higher than those under cash crops as a result of manuring. However, the model
does not take into account the impact of the simulated increase in cropping intensity on
carbon amounts in plant and soil components for a given length of fallow or cropping
as compared with the amounts measured in 1997. Higher cropping intensity will certainly reduce C pools in perennial plant biomass; its impact on SOC pool is also likely
to be negative as a result of reduced C inputs to root biomass and SOC mineralization
induced by tillage and nutrient depletion (Diels et al., 2002), while reduction of faunal
and microbiological activity may reversely reduce SOC mineralisation.
From the perspective of global change, modifications in land use in Sare
Yorobana—like in the rest of the WAS—should not be responsible for a high proportion of global anthropogenic atmospheric C release (Houghton, 1995). Early in
the simulation period, values for net C release (0.48 t C ha1 year1) for the village
during the next 15 years are certainly higher than those estimated by Woomer (1993)
for a range of similar cropping systems in dry tropical southern Africa (0.07–0.17 t C
ha1 year1). Yet they remain less than a twentieth of those estimated for wet
tropical forests in southern Cameroon (Kotto-Same et al., 1997), the discrepancy
originating from different initial C amounts in plant biomass.
In southern Senegal, communal social organisation of labour and land use, national
tenure policies and the need for land might impede the efficiency of management programmes aimed at replenishing the carbon resource, as compared to what could be
expected in East Africa (Woomer et al., 1998). A better securisation of land title should
help replenish plant biomass and thus carbon and nutrient stocks of the farming systems of High Casamance by (1) dissuading people from clearing fallows and extending
crop production merely for the purpose of extending their right of use over more land,
(2) stimulating perennial investments favouring carbon sequestration (Izac, 1997).
4.4. Methodological considerations on the assessment of carbon and nutrient budgets
at the village scale
One of the main factors accounting for the scarcity of carbon and nutrient budgets
for African villages is methodological. Soil heterogeneity (inherent and biologically
induced), vegetation physiognomy and crop yields require careful and intensive
sampling schemes, and the use of regression relationships linking variables that are
easily measurable in the field to the amount of each element in the different components of the ecosystem. Establishing regression relationships is labour-intensive
and expensive, which is the reason for the following discussion on the relevance of
applying them to other subregions of sub-Saharan Africa.
4.4.1. Assessing carbon, nitrogen and phosphorus storage in cropped fields
Several factors influence crop and weed yields and may restrict the application of
regression relationships established here for crops to other campaigns or to other
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
77
places (Powell, 1984). Climatic conditions and plant health in sampled plots suggest
that the value of regression parameters between crop plant components was not
significantly affected by any special climatic or pest hazard, thus validating their use
for fields cropped with similar cultivars on the widespread Lixisols of the Sudanian
ecozone. The discrepancy found for sorghum and groundnut regression parameters
between the present study and Powell’s (1984) suggests that intra-specific variability
may more seriously limit the possibility of extrapolation.
Quantifying bush re-growth dynamics under cropping remains a difficult task
since it relies on crop history, clearing methods and cultivation practices (Bohringer
et al., 1996). Furthermore, woody BGB is poorly correlated to above-ground components due to repeated clearing of vegetation during cropping. The cropping
intensity criterion adopted here is intuitive and controversial, and still needs to be
validated with methods other than sensitivity tests.
4.4.2. Assessing carbon, nitrogen and phosphorus storage in fallow plots
Applying time functions defined for secondary successions in Sare Yorobana
(Manlay et al., 2002a) as tools to estimate element pools in secondary successions
from more crowded areas of the subregion should be done with care and validated
by field measurements. Combretaceae formations on Lixisols are certainly widespread in West Africa, but the dynamics of woody biomass and even of soil organic
status after crop abandonment depend to a large extent on plot history and cropping
intensity.
4.4.3. Spatial boundaries for the assessment of carbon, nitrogen and phosphorus
stocks of the village
Sampling depths for soil and roots in the present study were chosen taking into
account the literature and results of local plot studies. These limits may well apply to
Lixisols in general, but should be adapted to each subregion. Setting horizontal
limits for the correct assessment of the carbon and nutrient budget of a village territory is even more delicate, since the area exploited by the community does not
necessarily match the area it owns.
5. Conclusion
From a local agro-ecological point of view, diversity is one of the main determining factors in the functioning of the village farming system of Sare Yorobana at all
spatial and functional scales. Previous plot studies stressed the role of tree- and
fauna-induced heterogeneity in the ability of savanna-derived natural fallows to
replenish soil fertility. The present work shows that diversity is preserved at the
farm-holding level through the maintenance of various land-use systems that play
different roles related to specific carbon and nutrient distribution in plant and soil
components.
The village community is itself a mosaic of contrasted types of holdings, especially
with regard to livestock availability. Past experience suggests that sustainable
78
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
patterns of intensification at the plot scale are those that enhance diversity or
heterogeneity. This should be applied at the village and farm-holding levels by
promoting corralling agreements to avoid unbalanced manuring rates. However,
any development programme aimed at intensifying animal husbandry—stalling for
instance—should take into account their potential ability to modify the traditional
rules set for the communal management of organic matter, and thus the balance of
power between holdings.
Analysis of spatial and functional diversity provides some understanding of the
structure of the system. Diversity initiates matter and energy flows—including
anthropogenic uptakes—whose study is indispensable in assessing the sustainability
of the system.
From the perspective of global change, the most easily manageable carbon stocks
still remaining in savannas are the main reservoirs of easily available energy, nutrients, construction material and other economic goods. Since the need for cropped
land leaves little hope for their conservation, global environmental considerations in
intensification schemes should aim at settling people in savannas through the more
efficient use of these reservoirs to prevent migration towards more fragile carbon
deposits such as wet forests (Brown and Lugo, 1990).
A significant reduction in C depletion could result from intensification of production schemes, albeit within realistic limits. For instance, mineral fertilisers stabilise soil organic matter and slow down deforestation. However, such improvements
rely heavily on the subsidisation of fertiliser prices and on rural investment policies,
and these decisions are taken at levels over which peasant farmers have but little
control.
Acknowledgements
The authors acknowledge critical and helpful comments by two anonymous
reviewers, J. Aronson, A.-M. Izac, C. Millier and J.-C. Remy. This work received
financial support from the EEC Project ‘‘Reduction of the Fallow Length, Biodiversity and Sustainable Development in Central and West Africa’’ (TS3-CT93-0220,
DG12 HSMU) (Floret, 1998) and from the following French institutions: Institute
for Research and Development (IRD, ex-ORSTOM), International Centre for
Agricultural Research in Hot Regions (CIRAD), and Institute of Forestry, Agricultural and Environmental Engineering (ENGREF).
References
Achterstraat, A., 1983. Agriculture d’autosubsistance ou agriculture de rente? La prise de décision des
paysans entre différentes cultures, et l’impact de l’intervention de la SODEFITEX et de la culture
cotonnière sur les sociétés Peul en Haute-Casamance. L’exemple de Lenguewal. Inst. of Cult. Anthr.,
Free University, Amsterdam, Netherlands.
Baldensperger, J., Staimesse, J.P., Tobias, C., 1967. Notice Explicative de la Carte Pédologique du
Sénégal au 1/200000 - Moyenne Casamance. ORSTOM, Dakar.
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
79
Berger, M., 1996. Fumure organique: des techniques améliorées pour une agriculture durable. Agriculture
et Développement 10, 37–46.
Blanfort, V., 1991. Phytosociologie et production de la végétation des parcours d’une petite région agropastorale de Moyenne-Casamance (Sénégal). Master’s Thesis, INA-PG—Univ. Paris XI et Paris VI,
Paris.
Bohringer, A., Leihner, D.E., Bocker, R., 1996. Shoot biomass of natural stump regrowth in cropping
systems in the subhumid forest savanna mosaic zone of West Africa. Tropenlandwirt 97 (2), 225–239.
Brown, S., Lugo, A.E., 1990. Tropical secondary forests. Journal of Tropical Ecology 6 (1), 1–32.
Chopart, J.L., 1980. Etude au champ des systèmes racinaires des principales cultures pluviales au Sénégal
(arachide, mil, sorgho, riz pluvial). PhD Thesis, Institut National Polytechnique, Toulouse.
Dabin, B., 1967. Application des dosages automatiques à l’analyse des sols. 3ème partie. Cahiers ORSTOM, Série Pédologie 5 (3), 257–286.
de Ridder, N., van Keulen, H., 1990. Some aspects of the role of organic matter in sustainable intensified
arable farming systems in the West-African semi-arid-tropics (SAT). Fertilizer Research 26, 299–310.
Déat, M., Joly, A., Dubernard, J., Sément, G., 1976. Exportations minérales du cotonnier et de quelques
cultures tropicales en zone de savane africaine. Coton et Fibres Tropicales 31 (4), 409–418.
Diels, J., Aihou, K., Iwuafor, E.N.O., Merckx, R., Lyasse, O., Sanginga, N., Vanlauwe, B., Deckers, J.,
2002. Options for soil organic carbon maintenance under intensive cropping in the West African
Savanna. In: Vanlauwe, B., Diels, J., Sanginga, N., Merckx, R. (Eds.), Integrated Plant Nutrient
Management in sub-Saharan Africa: From Concept to Practice. CAB International, Wallingford,
Oxon, UK, pp. 299–312.
Dufrène, E., Ochs, R., Saugier, B., 1990. Oil palm photosynthesis and productivity linked to climatic
factors. Oléagineux 45 (8–9), 345–355.
Dugué, P., 1985. Utilisation des résidus de récolte dans un système agro-pastoral sahélo-soudanien au
Yatenga (Burkina-Faso). In: Relations Agriculture-Elevage. Actes du Deuxième Séminaire du DSA/
CIRAD, Montpellier, France, 10-13/09/1985. CIRAD-DSA, pp. 76–85.
Dugué, P., 1998. Les transferts de fertilité dus à l’élevage en zone de savane. Agriculture et Développement 18, 99–107.
Fanchette, S., 1999. Densité de population et intensification agro-pastorale en Haute-Casamance.
Espaces, Populations, Sociétés 1, 67–81.
Feller, C., Beare, M.H., 1997. Physical control of soil organic matter dynamics in the tropics. Geoderma
79 (1–4), 69–116.
Fernandes, P. 1999. Effet des modes de gestion des terres en zone sahélienne sur les termes et l’évolution
de leur bilan organique et biologique: conséquences pour une production céréalière soutenue—cas du
Sénégal. PhD Thesis, Inst. Nat. Polytech. de Lorraine, Nancy.
FIA-SLE, 1990. L’Analyse Socio-Economique des Systèmes d’Exploitation Agricole et de la Gestion de
Terroir dans le Bas-Saloum, Sénégal. Schriftenreihe des Fachbereichs, Vol. 132. Verlag-Technische
Universität Berlin, Berlin.
Floret, C. (Ed.), 1998. Raccourcissement du temps de jachère, biodiversité et développement durable en
Afrique Centrale (Cameroun) et en Afrique de l’Ouest (Mali, Sénégal): Final Report. Comm. des
Communautés Européennes. Contrat TS3-CT93-0220 (DG 12 HSMU) and IRD Paris.
Gueye, F., Ganry, F., 1981. Recyclage des résidus de récolte par la voie fermentative. Exemple de deux
pailles: mil et maı̈s. International Foundation for Science—Institut Sénégalais de Recherches Agricoles/
CNRA, Stockholm—Bambey.
Houghton, R.A., 1995. Land-use change and the carbon cycle. Global Change Biology 1, 275–287.
Ickowicz, A., Usengumuremyi, J., Badiane, A., Richard, D., Colleye, F., Dupressoir, D., 1998. Interactions entre jachère et systèmes d’alimentation des bovins en zone soudanienne du Sénégal: choix
techniques et dynamique de développement (zone soudanienne, Sénégal). In: Floret, C., Pontanier, R.
(Eds.), Jachère et Systèmes Agraires, Niamey, Niger, 30 septembre–2 octobre. ORSTOM, Dakar, pp.
123–138.
Izac, A.-M.N., Swift, M.J., 1994. On agricultural sustainability and its measurement in small-scale
farming in sub-Saharan Africa. Ecological Economics 11, 105–125.
80
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
Izac, A.-M.N., 1997. Developing policies for soil carbon management in tropical regions. Geoderma 79
(1–4), 261–276.
Jacquemard, J.C., 1995. Le Palmier à Huile. Le Technicien d’Agriculture Tropicale. Maisonneuve et
Larose, Paris.
Jaffre, T., 1984. Evolution de la biomasse épigée et du stock de carbone d’une culture pérenne: le palmier à
huile (Elaeis guineensis Jacq.). ORSTOM, Abidjan.
Jones, M.J., Wild, A., 1975. Soils of the West African Savanna. Technical Communications, Vol. 55.
Commonwealth Agricultural Bureaux, Farnham Royal.
Ker, A., 1995. Farming Systems in the African Savanna. A Continent in Crisis. International Development Research Centre (IDRC), Ottawa.
Kotto-Same, J., Woomer, P.L., Appolinaire, M., Louis, Z., 1997. Carbon dynamics in slash-and-burn
agriculture and land use alternatives of the humid forest zone in Cameroon. Agriculture Ecosystems &
Environment 65 (3), 245–256.
Kowal, J.M., Kassam, A.H., 1978. Agricultural Ecology of Savanna: a Study of West Africa. Clarendon
Press, Oxford, UK.
Krogh, L., 1997. Field and village nutrient balances in millet cultivation in northern Burkina Faso: a
village case study. Journal of Arid Environments 35 (1), 147–159.
Landais, E., Lhoste, P., 1993. Systèmes d’élevage et transferts de fertilité dans la zone des savanes
africaines. II. Les systèmes de gestion de la fumure animale et leur insertion dans les relations entre
l’élevage et l’agriculture. Cahiers Agricultures 2, 9–25.
Lericollais, A., Milleville, P., 1993. La jachère dans les systèmes agropastoraux Sereer au Sénégal. In:
Floret, C., Serpantié, G. (Eds.), La Jachère en Afrique de l’Ouest. Atelier International, Montpellier,
France, 2–5 decembre 1991. ORSTOM, Paris, pp. 133–145.
Manlay, R.J., 2000. Organic Matter Dynamics in Mixed-farming Systems of the West African Savanna: A
Village Case Study from South Senegal. PhD Thesis, ENGREF, Paris. Available from http://www.
engref.fr/thesemanlay.htm.
Manlay, R.J., Kaı̈ré, M., Masse, D., Chotte, J.-L., Ciornei, G., Floret, C., 2002a. Carbon, nitrogen and
phosphorus allocation in agro-ecosystems of a West African savanna I. The plant component under
semi-permanent cultivation. Agriculture Ecosystems and Environmen 88 (3), 215–232.
Manlay, R.J., Masse, D., Chotte, J.-L., Feller, C., Kaı̈ré, M., Fardoux, J., Pontanier, R., 2002b. Carbon,
nitrogen and phosphorus allocation in agro-ecosystems of a West African savanna II. The soil component under semi-permanent cultivation. Agriculture Ecosystems and Environment 88 (3), 233–248.
Manlay, R.J., Chotte, J.-L., Masse, D., Laurent, J.-Y., Feller, C., 2002c. Carbon, nitrogen and phosphorus allocation in agro-ecosystems of a West African savanna III. Plant and soil components under
continuous cultivation. Agriculture Ecosystems and Environment 88 (3), 249–269.
Manlay, R.J., Ickowicz, A., Masse, D., Feller, C., Richard, D., 2003. Spatial carbon, nitrogen and phosphorus budget of a village of the West African savanna—II. Element flows and functioning of a mixedfarming system. Agricultural Systems. doi:10.1016/S0308-521X(03)00054-4.
Pélissier, P., 1996. Les Paysans du Sénégal. Les Civilisations Agraires du Cayor à la Casamance. Ministère
de l’Education Nationale—Centre National de la Recherche Scientifique, St Yrieix.
Peters, J.B., Schulte, E.E., 1994. Soil test survey of the Gambia—an overview. Communications in Soil
Science and Plant Analysis 25 (9–10), 1713–1733.
Poulain, J.F., 1977. Crop residues in traditional cropping systems of West africa. Effects on the mineral
balance and level of organic matter in soils. Proposals for their better management. In: Organic
Recycling in Africa. FAO/SIDA Workshop on the Use of Organic Materials as Fertilizers in Africa,
Buéa, Cameroon, 5–14 mai. FAO, pp. 38–71.
Powell, J.M., 1984. Assessment of dry matter yield from grain yield in the West African Savannah Zone.
Journal of Agricultural Science 103, 695–698.
Prudencio, C.Y., 1993. Ring management of soils and crops in the West African semi-arid tropics—the
case of the Mossi farming system in Burkina Faso. Agriculture Ecosystems and Environment 47, 237–
264.
Richard, D., Ahopke, B., Blanfort, V., Pouye, B., 1991. Utilisation des zones agricoles et pastorales par
les ruminants en zone soudanienne (Moyenne Casamance, Sénégal). In: Gaston, A., Kernick, M., Le
R.J. Manlay et al. / Agricultural Systems 79 (2004) 55–81
81
Houérou, H.N.(Eds.) (Eds.), Actes du Quatrième Congrès International des Terres de Parcours,
Montpellier, France, 22–26 avril. CIRAD, Montpellier, France, pp. 754–756.
Ruthenberg, H., 1971. Farming Systems in the Tropics. Clarendon Press, Oxford.
SMI, 1993. ATLAS GIS—3.1 Manuel de référence. Le système d’information géographique pour
Windows. Strategic Mapping Inc, Santa Clara.
Sokal, R.R., Rohlf, F.J., 1994. Biometry: The Principles and Practice of Statistics in Biological Research.
W.H. Freeman & Co.
Thioulouse, J., Chessel, D., Dolédec, S., Olivier, J.M., 1997. ADE-4: a multivariate analysis and graphical
display software. Statistics and Computing 7 (1), 75–83.
Tiessen, H., Feller, C., Sampaio, E.V.S.B., Garin, P., 1998. Carbon sequestration and turnover in semiarid savannas and dry forest. Climatic Change 40 (1), 105–117.
UNDP, 1999. Human Development Report. United Nations Development Programme, New York.
van der Pol, F., 1992. Soil Mining. An Unseen Contributor to Farm Income in Southern Mali. Bulletin of
the Royal Tropical Institute, Vol. 325. KIT Press.
Woomer, P.L., 1993. The impact of cultivation on carbon fluxes in woody savannas of southern Africa.
Water Air and Soil Pollution 70, 403–412.
Woomer, P.L., Palm, C.A., Qureshi, J.N., Kotto-Same, J., 1998. Carbon sequestration and organic
resource management in African smallholder agriculture. In: Lal, R., Kimble, J.M., Follett, R.F.,
Stewart, B.A. (Eds.), Management of Carbon Sequestration in Soil. CRC Press Inc, Boca Raton, pp.
153–173.