14 Implications of population growth and declining access to transhumant grazing areas for the sustainability of agropastoral systems in the semi-arid areas of Niger

Bruno Barbier and Peter Hazell

The study community

The modeling method

Model specification

The utility function

The constraints

Recursive solutions

Model simulations

The baseline scenario

The effect of droughts

Alternative scenarios for transhumance and purchased feed

Conclusions

Bibliography

Sahelian pastoralism is encountering difficult problems as population growth leads to the expansion of cropland at the expense of traditional pastures. Many analysts believe that pastoralism is bound to disappear and is likely to be replaced by mixed crop–livestock farming systems, where livestock stay near the farms and provide milk, draft power, and manure for soil fertility (Boserup 1965; Ruthenberg 1980; Pingali, Bigot, and Binswanger 1987; Beets 1990; Landais, Lhoste, and Guerin 1990;). However, other studies suggest that mixed crop–livestock farming systems are less suited than pastoralism to the low and erratic rainfall patterns of the Sahel. Sedentary farming systems have limited means for coping with drought, while traditional pastoral systems, which rely on walking animals to other regions that have received better rainfall, are much more able to cope (Scoones 1995; Toulmin 1995).

The shift from pastoralism to more settled farming is driven in part by increasing population pressure and greater commercialization of agriculture. These forces create new opportunities as well as new needs for intensifying the farming system within rural communities. At the same time, these forces might also lead to greater enclosure and settlement of traditional grazing areas, leading to diminished access rights to these areas. Thus, the decline in traditional pastoralism can be seen as a cause as well as a result of diminishing access to transhumant grazing areas, and to a shift to increased cropping.

To better understand the economic forces driving these transformations, and to assess their implications for future livelihood standards and environmental sustainability, we have developed a bioeconomic model of a typical community in the semi-arid areas of Niger.¹ The model is used to simulate the longer-term consequences of changes in population growth and reduced access rights to transhumant grazing areas. Particular attention is given to the role of drought risk in conditioning the model’s results, and how improved methods of managing drought risk affect the development pathway that the community follows.

1 We used the solver CONOPT from the GAMS Software (Brooke, Hendrick, and Meeraus 1988).

The study community

The village of Banizoumbou is a typical community of the Sahel, with low rainfall (450 millimeters) and sandy soils. Most farmers are barely self-sufficient in raising millet. Sale of animals and temporary migration are the main sources of cash income. Banizoumbou has good access to a paved road and to several active markets.

The village has a population of around 1,300 people and a total land area of 6,200 hectares. This gives a population density of 22 persons per square kilometer, which is about average for this part of the Sahel. There are 860 Zarmas and 475 Fulanis in the village territory. The Zarmas live in hamlets, while the Fulanis live in more isolated nuclear families that are scattered across the village territory. Zarmas are mainly settled farmers, while Fulanis are agropastoralists who cultivate some land but also are transhumant for large parts of the year.

The population is growing rapidly in Banizoumbou, probably at close to the national rural average of 3 percent per year. Permanent migration is restricted because Nigeria, the main place of destination, is in a deep economic crisis.

The main crop produced is millet, which is grown during the single rainy season that extends from June to September. Yields are low and vary between 200 and 400 kilograms per hectare. One worker can cultivate between 2 to 3 hectares. Availability of phosphorus and nitrogen in the soil are important limiting factors on yields, and the prime source of loss of these nutrients is through removal of the harvest. Where fertilizers are applied on millet, yields increase but rarely to more than 600 or 700 kilograms per hectare. The sandy soils are very poor but easy to work, which explains the extensive agropastoral system that has traditionally prevailed in the area. Agropastoralists usually have much smaller plots than farmers, but their yields are twice as large because of intensive manuring (Beauvillain 1977). Agropastoralists are usually less self-sufficient in grain than farmers and compensate for this potential nutrient and economic shortfall by buying millet and by drinking milk from their livestock (Collin de Verdière 1995).

The prevailing land-tenure system in Banizoumbou provides farmers and agropastoralists with relatively free access to land. They have to ask the traditional chiefs for permission to cultivate a plot, but they can still obtain this permission readily.

Agropastoralists from the area go north during the rainy season, where there are more pastures and the grass is of better quality. Herds that do not migrate seasonally stay in the village territory and have lower productivity. The areas that they can graze in the village during the rainy season are restricted because the crops are growing. Agropastoralists move back into the area after the rainy season to let their herds graze crop residues. If not enough residues and grass are available around the village during the postharvest season, some agropastoralists migrate to grazing areas around neighboring communities. The general strategy is to gain the maximum livestock weight during the rainy season and to lose as little weight as possible during the dry season.

Some authors (Scoones 1995; Toulmin 1995) suggest that, in the Sahel, overgrazing has no major long-term effect on forage production. It seems to be true for the sandy soils that characterize Banizoumbou, where grazing does not compact soils as it does in areas with soils that are richer in silt and clay. However, intensive grazing during the rainy season can lead to a change in the species composition of the pastures—toward shorter-cycle grasses and less palatable species. However, if grazing is less intense the following season, the new species composition rapidly becomes more palatable, confirming that the effect of overgrazing on subsequent pasture production is mild.

Livestock production in the area is cyclical between droughts. The two exceptional droughts of 1973 and 1984 drastically reduced the national livestock herd in Niger, perhaps by as much as 50 percent (FAO 1997). This led many farmers and agropastoralists to increase their holdings of sheep and goats, which are more drought resistant than cattle. The recovery of herd size after a drought is conditioned by the number of breeding cows that survive, and agropastoralists usually sell their bulls but keep as many fertile cows as possible. This policy is also consistent with herders’ preferences for building up larger herds as a major source of wealth and for protecting their food security in drought years. Selling a cow can mean a significant loss of future income. Also, an old cow that has survived the harsh conditions of the Sahel may well have a more drought-resistant genetic makeup than cows that do not survive. Another reason not to sell cows is that in the current context of open access to grazing areas, a larger herd is more productive than a smaller herd for an individual herder.

Labor is generally not a limiting factor in the agropastoralist community. Many agropastoralists lost their herds during the severe droughts of the 1970s and 1980s, and fewer heads of livestock are in Niger today than before those droughts. Yet the agropastoralist population has almost doubled, hence there are far fewer heads of livestock per herder today.

Until recently, crop- and livestock-production systems were considered to be symbiotic in the region because agropastoralists used to exchange animal manure for grazing rights, and milk for grain. Farmers even used to contract to have their livestock tended by agropastoralists. Now, farmers have started to manage their livestock themselves, and they increasingly restrict agropastoralists’ access to pasture, crop residues, and water points. Agropastoralists are also intensifying their own crop production because pastoralism is becoming more risky and because they have the manure to intensify crop production.

Analysts have conflicting views over which system of property rights should be implemented to help improve productivity and natural-resource management. Many analysts consider that the current open-access system for pastures leads to the overuse of these resources, while a system of individual property rights would lead to more responsible resource management and higher levels of productivity. Under individual property rights, pastoralism is more likely to be replaced by sedentary production systems where livestock are fed with locally produced or purchased forage. In contrast, other analysts argue that more settled crop–livestock systems are not environmentally sustainable in such low-rainfall and drought-prone areas as Banizoumbou, and that continued pastoralism is required.

However, it is not clear that the return to the traditional patterns of resource management is a viable alternative today given the livelihood needs of a larger and rapidly growing rural population (see Chapter 11). Resolution of these conflicting views requires a serious quantitative analysis of the options open to Sahelian communities such as Banizoumbou, including an assessment of the longer-term consequences for sustainable resource-management. The next section describes a bioeconomic model constructed specifically to analyze these issues.

The modeling method

The literature on suitable models for simulating integrated crop–livestock systems is growing (for recent reviews, see Breman 1993; Oriade and Dillon 1997). The most recent models include biophysical components for simulating the productivity of pasture areas, and the status and yield consequences of soil-nutrient balances and the amount of organic matter in the soil. Several models have also been applied to Sahelian situations. Until recently, most models were designed at the farm level. However, given the prevalence of common and open-access land, new village- and community-level models have been developed to explicitly include these lands (Kebe 1992; Deybe and Butcher 1996; Barbier and Benoît-Cattin 1997). Given also the importance of climate and price variability in the Sahel, several models have incorporated risk (production variance) and risk-averse behavior using such methods as MOTAD (Hazell 1971); Target MOTAD (Tauer 1983); Focus-Loss constrained programming (Boussard and Petit 1967); and discrete stochastic programming (Cocks 1968; Rae 1970, 1971).

In this study we use a dynamic and discrete stochastic programming model to conduct long-term simulations of alternative development pathways available to the village of Banizoumbou. The model describes the crop and livestock-production systems and their interactions at the village level. The scale of the model is the full village territory plus the open-access pastures where the pastoralists from the village migrate and that are not included in the village territory. The data for the model are taken mainly from International Livestock Research Institute (Hiernaux et al. 1998), although some technical coefficients come from the technical literature about Sahelian situations (French Cooperation Ministry 1991; Breman et al. 1986; Milleville and Serpantiér 1994; Collin de Verdiere 1995).

A key characteristic of the model is the way risk is specified. Farmers and pastoralists are assumed to be risk averse, and to conform to a decision framework with a mean standard-deviation in making their decisions at the beginning of each rainy season. However, not all decisions have to be made at the beginning of the rainy season, and many can be delayed until later in the season when more information about the season’s rainfall outcome is available. Optimal adjustments (or recourse decisions) to the emerging rainfall pattern is an extremely important part of risk management in the Sahel. For example, while many decisions about planting crops (such as type, area, seed rate, and manuring) have to be made early in the year before the rains have arrived, and hence have to be based on expectations about the forthcoming rains, other decisions (such as feeding livestock, buying and selling animals, transhumance, and storing food) do not have to be made until later in the year and can be adjusted according to the emerging rainfall pattern and the known availability of foods and feed. To model this type of sequential decision problem, we use the discrete stochastic programming with recourse (DSPR) approach developed by Cocks (1968) and Rae (1970, 1971).

DSPR models have a decision-tree structure where the nodes of the tree are the decision points and the branches correspond to different states of nature (or rainfall outcomes in our application). As such, these models can quickly become very large, and to avoid this we made the simplifying assumptions that only two decision stages occur during the year (planting and postharvest), and only two rainfall outcomes of interest (drought and normal). With a four-year planning horizon (see later), this results in a 2⁴ = 16 sequences of states of nature.

The drought event is taken to be the level of rainfall that has a 10 percent chance of occurring (a catastrophic event), whereas a normal year has a 90 percent chance of occurring. Rainfall outcomes are assumed to be independent over time.

The two decision stages in the model are the planting period and the postharvest period. All decisions made in the planting period have to be taken before any season-specific knowledge about the rainfall is available. These ex ante decisions can only be informed by prior knowledge of the probability distribution of rainfall. All decisions in the postharvest period are assumed to be taken once the actual rainfall outcome (drought or normal) is known. These ex post decisions take the form of optimal adjustments to the available crop production and fodder and grazing resources. The planting-period decisions include the amount of area to plant for each crop as well as the quantity of manure and inorganic fertilizers applied. The postharvest decisions include choices about storing, selling, buying, and consuming the harvest that are based on actual yield outcomes and market prices, and most of the livestock-management decisions.

Livestock production has much more recourse than cropping. In fact, no significant livestock decisions have to be made during the planting season in the model. In the model, animals can be bought or sold. In the postharvest period, the model will adjust the planned duration of transhumance depending on the rainfall outcome and the availability of feeding resources and market prices (especially for animals). Temporary human migration is also a recourse decision in the model. Migration of males to Côte d’Ivoire and Ghana to work there during the dry season generates valuable earnings on average.

Model specification

We adhere to the following notation in this analysis:

• Endogenous variables are capitalized.
• Coefficients are denoted by small letters.
• Indexes are subscripts.

Index t is time in years, p denotes periods within years, r the discount rate, n the two states of nature (drought and normal), and m denotes a sequence of states of nature over four years.

All variables and coefficients are listed and defined in Table 14.1. The model has three seasonal periods: the rainy season, from June to September; the harvest season, from October to January; and the hot and dry season, from February to May. Decisions made during the rainy season are made on the basis of prior expectations about rainfall, whereas decisions made during the other two seasons are based on the actual rainfall in the previous rainy season.

The utility function

The model maximizes the aggregate welfare of the community, measured as the discounted value of future income adjusted for risk, EXPUTILITY. Income is defined in the Becker sense to include the opportunity cost of leisure, while risk aversion is specified in mean-standard deviation form (Markowitz 1959; Hazell and Norton 1986; McKarl and Spreen 1997). We assume that the length of the planning horizon is four years.

The objective function is

where EXPINC is the expected value of discounted income over the four-year planning horizon, VARINC is the associated variance of discounted income, and 1.65 is an assumed risk-aversion coefficient.

To calculate these variables, we begin with the definition of the income outcome in year t, INC_t, defined as follows:

.

Table 14.1 Model notation and definitions

^a In this chapter, “ton” means metric ton.

^b Corralling is a method for producing manure from letting cattle rest and produce manure in some fields, which benefits the next crop.

^c This is actually a mixture of dung and crop residues produced by keeping cattle in stables.

Income is the sum of crop production, milk sales, wages from seasonal migrants, and livestock sales, adjusted for changes in livestock inventories and the opportunity cost of leisure, less the costs of cash expenses for farm production, transhumance, and grain purchased for the family.

The expected value of discounted income is then

In addition, the associated variance of discounted income is

.

The constraints

MILLET PRODUCTION. Total millet production is a function of yields, planted area, and fertilizer. Yields depend on the amount of organic and inorganic fertilizers applied. Organic fertilizer includes stabling manure, corralling manure, and compost, as described in Table 14.1. Currently farmers use only corralling manure. Producing manure from stabling and producing compost require much more labor than corralling. It is also assumed that when the phosphate content reaches a threshold level, yields begin to decrease:

                       

                        .

The quantity of stabling manure and compost available for millet production during year t is a function of the crop residues stored during year t – 1:

FERTMAN_t + FERTCOM_t < dmres * MILPROD_n,t–1.

Millet may be stored, consumed or sold:

MILSTORE_n,t + MILSEL_n,t + MILCONS_n,t = MILPROD_n,t + MILSTORE_t–1.

FOOD CONSUMPTION. The population (minus temporary migrants) is assumed to consume a fixed amount of millet throughout the year. Millet may be produced in the village or bought:

MILCONS_n,t + MILBUY_n,t > milcons * (POPF_n,t + POPA_n,t)

– milconsd * (POPTEMPF_n,t + POPTEMPA_n,t).

LAND USE. The village territory is either cultivated for millet or left in fallow (BUSH), which can be grazed:

MIL_t + BUSH_t = area.

The initial millet area can be increased or decreased:

MIL_t–1 + MILNEW_t – MILCUT_t = MIL_t.

The millet area can be increased by converting bush:

MILNEW_t = BUSHCUT_t

Similarly, the bush area can be increased or reduced:

.

POPULATION AND LABOUR. The local farming population is assumed to increase in accordance with the United Nations Environment Programme’s projection for Burkina Faso (Stephen et al. 1991). This implies a progressive decrease in population growth until the middle of the next century, when the population size will stabilize. However, in the model, emigration options (POPMIGF and POPMIGA) permit the size of the population to fall if this is more profitable for the village. This will happen whenever the population size reaches the point where another person consumes more than he or she produces:

Popg_t * POPF_t-1 – POPMIGF_n,t = POPF_n,t.

A similar equation is also assumed to apply for agropastoralists:

Popg_t * POPA_t-1 – POPMIGA_n,t = POPA_n,t.

The sum of labor requirements, temporary migration, and leisure (POPLEIS) has to equal the total days of labor available from the population during each seasonal time period. Labor is required during the peak time of planting and establishing millet, bush clearing, manuring, applying inorganic fertilizer, and transhumance to the northern pastures:

.

Transhumance can only be performed by agropastoralists who do not migrate:

actrans_p * LIVTRANS_n,p,t < actp_p * POPA_n,t – POPTEMPA_n,t.

PHOSPHORUS BALANCE. Phosphorus is said to be the most limiting factor for millet growth in the Sahel (Breman and de Witt 1983; Bationo and Mokwunye 1991). The soil has considerable phosphorus, but its assimilable fraction is insufficient for millet growth. Phosphorus becomes assimilable through complex processes that depend upon an equilibrium among different nutrients and organic matter. Application of organic and inorganic fertilizers increases the assimilable phosphorus. However, removal of crops depletes the amount of available phosphorus:

PHOS_t–1 * MIL_t–1 + PHOSB_t–1 / BUSH_t–1 * BUSHNEW_t

                                – PHOS_t–1 / MIL_t * MILCUT_t – phoex * MILPROD_n,t

                                + phosman * FERTMAN_t+ phoscom * FERTCOM_t

                                + phosnpk * FERTNPK_t = PHOS_n,t * MIL_t.

We assume that the same equation applies for phosphorus on land under bush, but in this case phosphorus loss from grazing does not occur, because phosphorus is mostly restored through manure.

Exports of assimilable phosphorus by crops is limited to a fraction of the available phosphorus, plus the phosphorus coming from fertilizers (which is assumed to be assimilable):

phosex * MILPROD_n,t < phosass * PHOS_t-1 + phosman

* FERTMAN_t + phoscomp * FERTCOM_t.

Below a certain level of phosphorus in the soil (PHOS), a deficit occurs (PHOSDEF) which negatively affects the millet production function:

.

LIVESTOCK AND LIVESTOCK MANURE. The amount of stabling and corralling manure available for millet production at the beginning of the year is limited by the number of livestock from the previous year that did not migrate:

.

LIVESTOCK BALANCE. Livestock production is managed by agropastoralists. Their income is the aggregation of milk sales, animal sales, and the value of herd growth (the increase in stock value). Livestock activity in the model is measured in standard tropical livestock units, where one unit is equivalent to an adult tropical cow. Changes in livestock activity are determined by selling animals and by herd growth. The latter has two components: weight growth over time and weight losses if forage deficits exist; the latter can offset the former. We assume that farmers do not consume meat themselves, but they can sell animals for meat.

Livestock carry over from one period to another in the model. Transfers occur between seasons within a year,

LIVP_n,p,t = livpot_p * LIVP_n,p–1,t – LIVSELP_n,p,t – ufstres * UFDEF_n,p,t,;

and between years,

LIVP_n,p,t = livpot_p * LIVP_n,p–1,t – LIVSELP_n,p,t – ufstres * UFDEF_n,p,t.

At a certain point of energy deficit, some animals have to be sold:

.

The number of livestock (units) that can be transhumant is calculated by dividing the forage consumed outside the village by the requirement for dry matter per animal per period:

.

LIVESTOCK AND FORAGE. Livestock per capita consumption of dry matter is fixed:

dmcons_p * LIV_n,t = DMCONS_n,p,t.

Total forage dry matter is produced by the village pastures, village crop residues, or the transhumance pastures, or comes from purchased feed:

DMTOT_n,p,t = DMBUSH_n,p,t + DMRES_n,p,t + DMFEED_n,p,t

+ DMTRANS_n,p,t + DMTRANS_n,p,t.

The amount of energy consumed depends on the amount of dry matter ingested and its energy content. If basic energy needs are not satisfied, an energy deficit (UFDEF) will occur:

ufneed_p * LIV_n,t – UFDEF_n,p,t < DMCONT_n,p,t * DMCONS_n,p,t.

Energy content is the ratio of total energy over total dry matter:

.

The total forage energy consumed is produced by the village pastures, by the crop residues, by purchased feed, and by the transhumance area:

DMUF_n,p,t.< ufbu_n,p * DMBUSH_n,p,t + ufres_n,p * DMRES_n,p,t

+ uffeed * DMFEED_n,p,t + uftrans_p * DMTRANS_n,p,t.

The next equation defines the available, edible dry matter, including transfers from one period to the next of any fraction of the grass that was not previously grazed:

DMBUSH_n,p,t + DMSURPB_n,p,t = dmbu_n,p * BUSH_n,t

+ dmlosbu_p * DMSURPB_n,p–1,t.

A similar equation applies for crop residues:

DMRES_n,p,t + DMSUPR_n,p,t = DMRESFOR_n,p,t

+ dmlosre_p * DMSUPR_n,p–1,t.

A similar equation applies for the transhumance areas. In addition, the productivity of the transhumance areas is assumed to decline over time (dmlostt_p) because of continuing population growth in the region and an associated loss of grazing areas to crop cultivation:

DMTRANS_n,p,t + DMSUTR_n,p,t = dmtran_n,p * areatrans_t

+ dmlostt_p * DMSUTR_n,p–1,t.

In the baseline scenario, forage from transhumance areas decreases at 3 percent per year.

Residues derived from millet production can be used for forage or for manure production:

dmre_p * MILPROD_n,t–1 = DMRESFOR_n,p,t + DMMANURE_n,p,t.

Recursive solutions

Although the model is solved as a dynamic four-year optimization program, it is also solved recursively each year to provide a series of moving four-year plans. This approach enables the model to be used to track much longer time periods than four years. It also provides a realistic way of simulating farmers’ ability to adjust their plans each year on the basis of outcomes of the previous year. In the recursive framework, the results of the first year of the planning horizon—in terms of livestock, millet stock, and soil phosphorus—become the initial resources of the revised model that is solved for the following year. We ran the model 100 times, representing 100 future years, for each of the simulations undertaken.

The recursive framework also enables adjustments to be made between expected and actual outcomes each year, given that production of millet and forage are affected by stochastic rainfall events. We use one of the two states of nature “drought” to introduce “climatic shocks” between some years in the various scenarios. The model adjusts total production and recalibrates the closing stocks of livestock and grain that enter the constraint set for the multiperiod model in year t+1.

Model simulations

We ran the simulations over 100 years because difficulties with soil fertility (phosphorus deficit) and with livestock only become critical in the long term. In all the simulations, we shocked the model to simulate droughts every 20 years (from 1997). Droughts are simulated by exogenously reducing millet and forage yields, forage quality, and livestock prices, and by increasing the price of millet. These shocks are based on historical data. In the baseline scenario, transhumance is allowed, but purchasing supplementary feeds is not an option. Three alternative scenarios were also simulated to help identify the effects of changing access rights to transhumance areas and of the possibility of purchasing supplementary feeds. In all scenarios, population is assumed to grow exogenously according to the United Nations Environment Programme projections, stabilizing around year 2030 (Figure 14.1).

The baseline scenario

In the baseline scenario, the millet area first expands because of population growth but later decreases again because an increasing phosphorus deficit in the soil requires a shift back to longer fallows (Figure 14.2). Millet yields (Figure 14.3) are affected from the beginning by the phosphorus deficit. The model tries to reduce the phosphorus deficit by rotating the millet area with manured pastures, but after a while the pasture area becomes too small to fulfill this role as a phosphorus provider. Similarly, the manuring technique selected by the model (corralling) cannot adequately compensate for the removal of phosphorus through crop yields. To maintain millet production, inorganic fertilizer is finally adopted by about the 40^th year (2030). Use of inorganic fertilizer allows for a regular increase in millet yield (Figure 14.3) and enables total millet production to grow in step with the population’s consumption needs (Figure 14.4). This yield increase compensates for the decrease in the millet area.

Figure 14.5 tracks the baseline evolution of livestock numbers. Livestock numbers initially trend upwards because livestock production is competitive and because some forage is still left in the community and in the transhumant areas, at least during normal years. However, as the availability of forage in the transhumance areas declines (at an assumed rate of 3 percent per year), livestock numbers eventually trend downwards.

Figure 14.6 shows forage consumption by the livestock from the community. Transhumance is always an important source of forage during the rainy season (season 1), but not during the hot and dry season (season 3), except in drought years.

Figure 14.7 shows the evolution of total income and its composition in the village. Crop income, which accounts for about a quarter of total income, initially increases but then begins a long-term decline after about the 35^th year, as production costs increase with the adoption of inorganic fertilizer. Livestock income accounts for only a small part of total income and shows little change over time. The men from the villages have to resort to greater seasonal migration to survive, although migration from the village is not permanent because temporal migration is more profitable. However, the increase in migration income is insufficient to maintain total income, hence total income begins to fall after about the 35^th year (Figure 14.7). This means that per capita income also declines (Figure 14.8 baseline case with transhumance but no feed). Even the adoption of inorganic fertilizers is insufficient to reverse this trend; apparently the system is too constrained to intensify in accordance with Boserup’s induced innovation model (Boserup 1965).

Millet yields are strongly affected by droughts (Figure 14.4), because no recourse decisions are available for reducing losses. The model chooses to buy millet during droughts instead of carrying stocks from the previous harvest (Figure 14.3). This result conforms to reality. With low productivity, a need for cash and the possibility of migration, farmers are reducing the size of the grain stocks they carry.

Millet yields do not recover immediately after droughts (Figure 14.4) because the loss of part of the livestock herd reduces the quantity of manure that is available. The model compensates for these lower yields in the immediate post-drought period by increasing the cropped area. This is possible because the lower yields obtained during the drought means that less phosphorus is removed from the soil, hence need for fallow after the drought decreases. As a result of the compensation of area for yield, crop income (which includes millet consumption) is much less affected by droughts (Figure 14.7).

The effect of droughts

The droughts significantly shock livestock numbers and production (Figure 14.5). The droughts worsen the existing energy deficit. By allowing transhumance, the model allows the livestock to obtain sufficient dry matter, but the ingested energy intake is low and productivity declines.

Figure 14.5 Evolution of livestock numbers

Figure 14.6 Forage consumption from different sources, by period

Figure 14.7 Evolution of village income, baseline scenario

  Millions of CFAF (Community of Francophone Africa, Francs

Figure 14.8 Evolution of income per capita under four scenarios

     Thousands of CFAF

The baseline simulation challenges the idea that rainfall variability affects crops more than livestock production in drought years. After a severe drought, recovery of livestock numbers and production takes several years. The problem is less severe for farmers because the effect of a severe drought can be mitigated by migration. In this sense, millet production is perhaps better adapted to rainfall variability than livestock production, and thus farmers can recuperate faster than agropastoralists. Agropastoralists usually revert to millet production in the years immediately following a drought, suggesting that livestock production is not that well adapted to droughts.

Income from seasonal migration does not increase in drought years. This is because the adult males in the village already migrate as much as they can during normal years.

Alternative scenarios for transhumance and purchased feed

As we have seen, transhumance plays an important part in the feeding and drought-management strategies of farmers and pastoralists in the Sahel. Continued access to these grazing areas is increasingly threatened by expansion of the cropped area throughout the region, and by greater privatization of land by communities and individuals (Ngaido [Chapter 11]). A key question is how villages like Banizoumbou will cope as their access to grazing areas diminishes. We considered two components to the adjustment strategy. The first is the use of supplementary feeds purchased from outside the village, particularly in drought years. We assumed that this feed would be provided at market cost. The second component is the exclusion of outsiders from using the village’s own grazing resources, that is, reciprocal privatization of the village’s own land. Excluding outside livestock would increase the availability of fodder and grazing resources available for use by the villagers’ own livestock, and it might also be expected to lead to greater intensification of the farming system within the village through increased investments in inorganic fertilizers and manure.

To examine these options, we conducted three additional model simulations. While the baseline scenario allows transhumance but not the purchase of feeds, one new scenario allows both transhumance and the purchase of feed. Two other scenarios then remove the transhumance option (and also exclude transhumant livestock from entering the village), and one of these scenarios has an option of purchasing feed while the other does not.

The bottom pair of graphs in Figure 14.9 show that a ban on transhumance in the absence of purchased feeds leads to a significant reduction in the size of the livestock herd after about 10 years. This is a clear demonstration of the value of transhumance practices for maintaining herd sizes under existing feeding practices. The graphs also show that transhumance does not smooth out the size of the downside shocks to herd size in drought years, but this is because more animals are carried into the droughts, when transhumance is allowed. Access to transhumance areas is particularly important in the harvest, and dry and hot, seasons during drought years for protecting the herd size.

Figure 14.9 also shows that access to purchased feed has a much more beneficial effect on livestock growth than transhumance and leads to very significant gains in livestock numbers in the longer term (compare the top and bottom pairs of graphs). It also leads to more stable herd sizes with greatly reduced losses in drought years. However, although a feed distribution program would have a beneficial effect on livestock numbers, its impact on per capita incomes is quite modest (Figure 14.9), and almost nonexistent in the short and medium term. This is because livestock income continues to account for but a small share of total village income.

The effect of the different scenarios on land use and yields is small because livestock numbers are too low in this village to have a significant impact on soil fertility at the village level. Even when the herd size expands sharply given a purchased feed option, the effect of manuring is low.

Conclusions

Our modeling results show that transhumance contributes importantly to maintaining the size of the livestock herd in the village, and it is particularly important in drought years for reducing herd losses. If the village were to lose all its traditional access-rights to grazing areas, the impact on livestock production would be severe. However, transhumance does not have a big impact on per capita incomes. This is partly because livestock income is only a small part of total income. (Most income comes from seasonal migration for nonfarm employment during the dry season.) However, this is also true because the village would in turn exclude others from using its own grazing resources, and this would increase the availability of local pastures and crop byproducts for the villager’s own animals.

If the villagers were to start purchasing supplementary feeds for their livestock, this could lead to a dramatic increase in the herd size. It would be a very effective way of reducing the loss and sale of animals in drought years. Use of purchased feeds would significantly reduce the need for transhumance. However, again, the impact on per capita incomes would be modest because livestock income is only a small part of total income, and little justification may exist for subsidizing the feeding program.

These results confirm that transhumance is an important risk-management strategy for villages such as Banizoumbou, but that the reciprocal cost of allowing outsiders to bring their animals into the village is also high. Given an alternative drought-management strategy, such as the use of feed supplements, the village would likely soon abandon transhumance arrangements and exclude outsiders from using its own grazing resources.

Bibliography

Barbier, B. 1999. Induced innovation and land degradation: Results from a bioeconomic model of a village in West Africa. Agricultural Economics 19 (1–2): 15–26.

Barbier, B., and M. Benoît-Cattin. 1997. Viabilité à moyen et long-terme d’un système agraire villageois d’Afrique Sudano-Sahélienne: Le cas de Bala au Burkina Faso. Economie Rurale 239 (May): 35–51.

Bationo, A., and A. U. Mokwunye, 1991. Role of manures and crop residues in alleviating soil fertility constraints to crop production with special references to the Sahelian and Sudanian zones of West Africa. Fertilizer Research . 29: 117–125.

Beauvillain, A. 1977. Les Peuls du Dallol Bosso. In Centre National de Recherche Scientifiques et Humaines. Edition Études Nigériennes, No. 42. Niamey: Institut de Recherche pour le Développement.

Beets, W. C. 1990. Raising and sustaining productivity of small holders in the tropics. Alkmaar, The Netherlands: AgBé.

Berger, M. 1990. La gestion des résidus à la ferme. Proceedings of the international conference Fertilité des Terres de Savane. December 10–14, Montpellier, France. Montpellier: Centre International de Recherche Agronomique pour le Développement.

Boserup, E. 1965. The conditions of agricultural growth: The economics of agrarian change under population pressure. New York: Allen and Uwin.

Boussard, J. M., and M. Petit. 1967. Representation of farmers’ behaviour under uncertainty with a focus loss constraint. Journal of Farm Economics 49 (4): 869–880.

Breman, H. 1993. Modélisation et simulation dans l’élaboration de systèmes de production animale durables. In Livestock and sustainable nutrient cycling in mixed farming systems of sub-Saharan Africa, Vol. 2, eds. J. M. Powell, S. Fernandez-Rivera, T. O. Williams, and C. Renard. Addis Ababa: International Livestock Centre for Africa.

Breman,, H., and C. T. de Witt. 1983. Rangeland productivity and exploitation in the Sahel. Science 221: 1341–1347.

Breman, H., N. Traoré, A. Bonfiglioli, J. H. Eriksen, J. H. M. Ketelaars, and K. Sawadogo. 1986. Analyse des conditions de l’élévage et propositions de politiques et de programmes: Burkina Faso. Report 12. Paris: Club du Sahel.

Brooke, A., D. Kendrick, and A. Meeraus. 1988. GAMS: A user’s guide. Redwood City, Calif., U.S.A.: Scientific Press.

Cocks, K. D. 1968. Discrete stochastic Programming. Management Science 15: 72–79.

Collin de Verdière, P. 1995. Etude comparée de trois systèmes agropastoraux dans la région de Filingue. Niger. Les conséquences de la sédenterisation de l’élevage pastoral au Sahel. Ph.D. diss., Institut National Agronomique Paris—Grignon, France.

Deybe, D., and, W. R., Butcher. 1996. Analysis of sustainability and credit policy using a model for a small village in the west of Burkina Faso. Paper presented at the 35^th annual meeting of the African Studies Association.

Dugué, P. 1985. Soil preparation in the Soudano-Sahelian Zone: Prospects and constraints. In Appropriate technologies for farmers, eds. H. W. Ohm and J. G. Nagy. West Lafayette, Ind., U.S.A.: Purdue University.

FAO (Food and Agriculture Organization of the United Nations). 1997. FAOSTAT database. <http://apps.fao.org/default.htm>. Accessed July 1997.

Fernandez-Rivera, S., T. O. Williams, P. Hiernaux, and J. M. Powell. 1995. Fecal excretion by ruminants and manure availability for crop production in semi-arid West Africa. In Livestock and sustainable nutrient cycling in mixed farming systems of sub-Saharan Africa, Vol. 2, eds. J. M. Powell, S. Fernandez-Rivera, T. O. Williams and C. Renard. Addis Ababa: International Livestock Centre for Africa.

French Cooperation Ministry. 1991. Mémento de l’agronome, 4^th ed. Paris.

Gavian, S., and M. Fafchamps. 1994. Land tenure and soil management in Niger. Paper presented to the annual meeting of the American Agricultural Economic Association, August 7–10, San Diego, California, U.S.A.

Hazell, P. 1971. A linear alternative to quadratic and semi variance programming for farm planning under uncertainty. American Journal of Agricultural Economics 53 (1): 53–62.

Hazell, P., and R. Norton. 1986. Mathematical programming for economic analysis in agriculture. New York: MacMillan.

Hiernaux, P., S. Fernandez-Rivera, M. D. Schlecht, M. D. Turner, and T. O. Williams. 1998. Livestock-mediated nutrient transfers in Sahelian agro-systems. In Soil fertility management in West African land use systems, ed. V. F. Magraf. Weikersheim, Germany: Verlag Josef Magraf.

Kebe, D. 1992. Modélisation de l’impact technico-économique de l’introduction de stylosanthes hamata dans les systèmes agraires villageois au Mali-Sud. Paper presented at the workshop Stylosanthes as Feed and Fallow, October, Kaduna, Nigeria.

Kruseman, G., R. Ruben, H. Hengsdijk, and M. K. van Ittersum. 1995. Farm household modeling for estimating the effectiveness of price instruments in land use policy. Netherlands Journal of Agricultural Science 43: 111–123.

Landais, E., P. Lhoste, and H. Guerin. 1990. Systèmes d’élevage et transferts de fertilité. Proceedings of the international conference Fertilité des Terres de Savane, December 10–14, Montpellier, France: Centre International de Recherche Agronomique pour le Développement.

Markowitz, H. M. 1970. Portfolio selection: Efficient diversification of investments. New Haven, Conn., U.S.A.: Yale University Press.

Matlon, P. J., and M. Fafchamps. 1988. Crop budgets for three agroclimatic zones of the West African semi-arid tropics. Report 85. Niamey: International Crop Research Institute for Semi-Arid Tropics.

McKarl, B. A., and T. H. Spreen. 1998. Applied mathematical programming using algebraic systems. Texas A&M University, U.S.A. Mimeo.

Milleville, P., and G. Serpantié. 1994. Intensification et durabilité des systèmes agricoles en Afrique Soudano Sahélienne. In Promotion des systèmes agricoles durables dans les pays d’Afrique Soudano-Sahéliennes, ed. M. Benoît-Cattin and J. C. de Grandi. Montpellier, France: Centre International de Recherche Agronomique pour le Développement.

Oriade, C. A., and C. A. Dillon. 1997. Developments in biophysical and bioeconomic simulation of agricultural systems: A review. Agricultural Economics 17 (October): 45–58.

Pieri, C. 1989. Fertilité des terres de savanes: Bilan de trente ans de recherche et de développement agricoles au Sud du Sahara. Montpellier, France: Centre International de Recherche Agronomique pour le Développement.

Pingali, P., Y. Bigot, and H. P. Binswanger. 1987. Agricultural mechanization and the evolution of farming systems in Sub-Saharan Africa. Baltimore, Md., U.S.A.: Johns Hopkins University Press.

Place, F., and P. Hazell. 1993. Productivity effects of indigenous land tenure systems in Sub-Saharan Africa. American Journal of Agricultural Economics 75 (1): 10–19.

Rae, A. N. 1970. Capital budgeting, intertemporal programming models with particular reference to agriculture. Australian Journal of Agricultural Economics 14: 39–52.

_____. 1971. Stochastic programming, utility and sequential decision problems in farm management. American Journal of Agricultural Economics 14: 39–52.

Ruthenberg, H. 1980. Farming systems in the tropics. Oxford: Clarendon Press.

Scoones, I., ed. 1995. Living with uncertainty: New directions in pastoral development in Africa. London: Intermediate Technology Publications.

Stephen, P. W., E. Bos, M. T. Vu, and R. Bulatao. 1991. Africa region population projections. PRE Working Paper WPS 598. Washington, D.C.: World Bank.

Tauer, L. W. 1983. Target Motad. American Journal of Agricultural Economics 65 (August): 606–610.

Taylor, E., and I. Adelman. 1997. Village economies. New York: Cambridge University Press.

Toulmin, C. 1995. Tracking through drought: Options for destocking and restocking. In Living with uncertainty: New directions in pastoral development in Africa., ed. I. Scoones. London: Intermediate Technology Publications.

van Keulen, H., and H. H. van Laar. 1986. Crop yields and nutrient requirements. In Modeling of agricultural production: Weather, soils, and crops, eds. H. van Keulen and J. Wolf. Simulation Monographs. Wageningen, The Netherlands: Wageningen University.

Williams, J. R., C. A. Jones, and P. T. Dyke. 1987. EPIC, the Erosion Productivity Impact Calculator. Temple, Tx., U.S.A.: U.S. Department of Agriculture—Agricultural Research Service, Economics Research Service, and Soil Conservation Service.

World Bank. 1997. World development indicators. Washington, D.C. CD-ROM.