Contents - Previous - Next

Biomass production in dry tropical zones: How to increase water productivity


All forms of agriculture can be described as water-dependent land use, where crop growth is preconditioned by the availability of substantial volumes of root zone water for evapotranspiration. Regions where this has a particular performance-determining role are in the dry tropics (semi-arid and arid tropics), where water is often identified as the principal limiting factor in biomass production. The extremely variable and dry hydroclimate in these regions, combined with fragile, inherently low-fertile soils, implies a high degree of environmental vulnerability, seriously complicating human activities in the landscape (Falkenmark et al., 1990).

These drylands, which together with sub-humid lands constitute approximately 40 percent of the land area of the world, are inhabited by some 700 million people. Around 60 percent of these drylands are located in developing countries, lands that probably will continue to produce most of the global demand for food grains for expanding populations (FAO, 1978; Parr et al., 1990). An accelerating population growth (with an annual growth rate attaining 3 percent in many regions), reduces the available water and soil resource per caput, at the same time as demand for food, fuel, fibre, fodder and timber increases dramatically. One of the effects of this increased human pressure on land and water is a human-induced degradation of the landscape, which in turn has contributed to the tendency of a relative decline of cereal yields per hectare during recent decades, especially in Sub-Saharan Africa (Brown and Goldin, 1992; FAO, 1988a; World Bank, 1986; World Bank, 1989). This decline is located in regions that, from the starting point are characterized by extremely low yields (generally below 1 ton/ha for cereal crops in Africa).

As a result of a broad awareness of the growing difficulty of achieving livelihood security in poverty-stricken tropical countries, a large number of reports have been presented in the last 10-15 years (for example by the FAO, the World Bank, UNESCO, IIASA), pointing out the necessity of (i) introducing new technologies, mineral fertilizers and high-yielding varieties in order to "intensify" the agriculture, as well as (ii) the need for an extension of the cultivated lands (i.e. diminished fallow periods). This, in order to increase yields in pace with population growth. The unpleasant truth is that these predictions and development strategies, having an overall aim of attaining a food production growth of 3-4 percent per year during the past and the coming decades, have not been realized, with the most serious failures being experienced in Sub-Saharan Africa (World Bank, 1989).

Johan Rockström, Research Associate, Natural Resources Management Institute, Stockholm University

The hypothesis here is that one central cause behind these failures is that the proposed, essentially technology-based, development strategies have been founded on a simplified and faulty conceptual framework of the environmental preconditions defining the limits of human activity in the landscape. An important issue in this context is the need for developing a deep understanding of soil and water interactions in biomass production, in order to make possible an increase in water productivity, i.e. produce more biomass per unit of water, which probably is necessary if agricultural production is to increase in these water-scarce landscapes.

The objective of this paper is to clarify the interactions between soil, water and vegetation, preconditioning biomass production in rainfed dryland agriculture, and to present options for increasing water productivity.

The focus on rainfed agriculture is explained by the following: (i) irrigated land is still very unevenly distributed (5 percent in the Sudano-Sahelian belt in Africa compared with 50 percent in India) despite expectations of a rapid expansion; (ii) investment costs are high, and (iii) irrigation schemes have been associated with severe management problems, especially in poverty-stricken countries. This indicates that rainfed agriculture will remain the dominating source of food production during a foreseeable future (Parr and Stewart, 1990). Furthermore, rainfed agriculture consumes most of the available water resources in semi-arid and arid regions (80-90 percent), which is important as it is this same agriculture that, according to different prognoses, will have to increase its production by at least 40 percent (implying a corresponding increase in water requirement) by the year 2000 in order to feed projected populations (FAO, 1988a). Added to this is the severe degradation of crop lands, threatening the future possibilities for satisfying soil moisture and fertility levels for biomass production.


Recurrent droughts with subsequent production failures, leading to high order effects like famine/migration, are not new phenomena in the rural communities in tropical drylands. Severe social effects of the extremely variable hydroclimate probably date all the way back to the definitive establishment of rural societies and the development of a hydroclimate more or less similar to the present, around 5000 years ago (Rapp, 1988; Nicholson and Flohn, 1980; Maley, 1981). Coping with water scarcity should therefore be seen as a normal, integrated part of the social structure of rural societies in tropical drylands, rather than an intermittent boundary condition. What has changed is the gravity of the effects of these "normal" hydroclimatic fluctuations, in pace with the escalating imbalance between a relative decrease of crop yields and a rapidly increasing population pressure. The principal core reason for diminished biomass productivity is the combined effect of fertility depletion and soil dessication, i.e. the decline in plant-available soil moisture and nutrients, which eventually develops into a permanent productivity crisis in the agriculture. The result is that less food is produced per hectare in a situation of escalating demand; the landscape is not capable of sustaining its population; a rural societal crisis is inevitable. This gradually aggravated crisis, which can be defined as an accentuated disequilibrium in the complex interaction between an unavoidable population growth and a fragile landscape with finite land and water resources, is to a large extent a 20th century phenomenon.

The lack of plant-available soil water, i.e. "green" water scarcity, is due either to diminished quantity of water supplied (normally rainfall) or due to altered rainwater partitioning, i.e. diminished infiltration and water retention in the soil, or the combination of both. A high rainfall variability in time (intra-,and interannual) and space is a normal characteristic in the dry tropics, as mentioned above, even if many scientists have found evidence of a worrying tendency to a steady aridification in the Sudano-Sahelian zone in Africa since the mid-1960s, with isohyets moving 100-150 km southwards (Sivakumar, 1992; AGRHYMET, 1991).

Altered water partitioning is principally due to human-induced land degradation, mainly through diminished or abolished fallow periods, increased exportation of biomass (for fuel, fodder, construction, etc.), upstream deforestation, and cultivation of marginal lands. The result is physical and chemical soil-surface crusts, and structural soil profile destruction, which has diminished infiltration capacity and the water holding capacity (WHC). Water and wind erosion increase, as does soil evaporation, which in turn increases the mechanical destruction of soils and diminishes the available crop water. The continuous risk for crop failure is accentuated.

To understand this evolution, which can be illustrated by a "vicious spiral" of population-driven land degradation accelerated by droughts (see Figure 1), it is necessary to analyse the dynamics of agricultural systems. The biomass productivity has been, through history, (and is still to a large extent), based on the reproduction of the fertility through fallow and "slash and burn" practices. With diminished soil availability per caput, the system based on long fallow periods (>20 years) has been abandoned in favour of short fallow periods, which in turn, with increased demographic pressure, has resulted in very short (1-5 years) to completely abolished fallow.

Parallel with diminished fallows, the exportation of biomass has increased (due to cattle pressure, increased demand for fuel, construction wood, etc.). This often complete abandoning of in-situ biomass-rotation has led to a fast depletion of organic matter, depletion of macro- and micro-nutrients, and increased effects of erosion. For the semi-arid regions in West Africa the continuous cultivation has the effect of drastically reducing the content of organic matter and minerals in the soil (Bationo and Mokwunye, 1991a).

This evolution is probably the most significant structural transition of the dryland agriculture since its introduction. It is particularly serious as the previously "sustainable" system, based on fallows in general, as opposed for example to the abandoning of fallows in temperate regions, has not been replaced by an equivalent system capable of sustaining soil productivity. The agriculture turns into a purely "nutrient mining" system. Pichot et al. (1981) conclude that problems on the medium and long term evolution of tropical soils do not appear acutely until the practice of long fallows, which permits the soil to return to its equilibrium between each growing period, is abandoned.

In an evaluation of the agriculture in a Sudano-Sahelian region in Burkina Faso (annual precipitation P = 750 mm) it was estimated that pearl millet yields attained approximately 1 ton/ha until the 1960s. Fields were cultivated during 5 years, then left as fallows more than 20 years. In the end of the 1960s and during the 1970s fallow periods declined, without being compensated by fertilizers (with exceptions for rich farmers). Yields declined to 200-400 kg/ha in many zones, cultivation was extended to marginal lands, and rural migration increased. Degradation of the soils combined with the effects of the 1982-84 drought has increased food deficits.

Figure 1. The "vicious spiral" of rural dynamics in dry tropical regions

A similar pattern can be identified for an agricultural system in the Sahelian region of Niger. The traditional rainfed agriculture was forced, as late as the 1930s, to abandon the long fallow periods of 20-25 years, which were necessary to guarantee yields of 400-500 kg/ha. Progressively, in pace with diminished soil availability, fallows were shortened or even abolished. Soils were depleted and desiccated, marginal lands were put under cultivation. Present yields have fallen to around 200 kg/ha.

This evolution towards an accelerated "biomass-scarcity" in tropical drylands, resulting in chronic soil moisture and nutrient deficits (due to reduced organic and mineral recycling), has led to alarming tendencies of declining yields in many semi-arid regions: up to 50 percent lower over a 10 year period in low-input rainfed agriculture for areas in the Sahel (Bationo, 1993; pers comm.). The effects of this physical degradation can be particularly severe in semi-arid regions, due to the hydroclimate and the fragile soils. The diminishing vegetation cover, and consequently the lowered rate of organic matter, can quite easily result in a collapse of the soil structure and the formation of impermeable soil crusts (Valentin et al., 1991).


Crop water requirement varies between different hydroclimatic zones, which is explained essentially by the differences in atmospheric demand, i.e. the potential evapotranspiration (PET). Hydroclimatic conditions differ, not only between but also within zones. In the Sudano-Sahelian zone for example, with its latitudinal aridity gradient, the annual precipitation schematically varies from 200 mm in the north to 1 200 mm in the south, with PET varying respectively between 2 200 mm and 1 600 mm, largely exceeding precipitation. The evaporative demand during the crop season varies between 5 and 8 mm/day, in a zone with cropping seasons varying from 75-180 days, resulting in seasonal PET of around 600-900 mm (Dancette, 1983; FAO, 1986).

For humid natural ecosystems, the return flow of water to the atmosphere is in the order of 200 m³/ton biomass dry matter (DM), while around 1 000 m³/ton is required in dry tropical ecosystems (Falkenmark, 1986 after L'vovitch, 1979). On a global level, Le Houérou et al. (1988) conclude that the "efficiency of rains" (CEP: Coefficient d'efficacité pluviale) for arid and semi-arid regions is in the order of 4 ± 0.3 kg DM/ha/year/mm, which gives a water requirement of 2 300-2 700 m³/ton.

Crop water requirement varies during the growing season, constituting a low fraction of PET in the early vegetative stages of growth, reaching its peak during stages of flowering and grain filling (actual crop water demand normally exceeding PET), and finally dropping during the final stages of ripening and wilting (drying), before harvest (FAO, 1979; Dancette, 1983). Water requirement varies also between different crops (with major differences being correlated to the maturity time, i.e. length of growing period), as does the capacity for surviving periods of crop water deficit (often defined as "drought tolerance"). Here the discussion will focus on cereal crops, especially pearl millet (Pennisetum americanum) and Sorghum (Sorghum bicolor), which are the most important food crops in semi-arid regions.

Trials indicate that for pearl millet cultivated in semi-arid regions in West Africa the seasonal crop water requirement is around 410 mm/ha for a 90-day variety (Souna) (number of days to harvest), and around 500 mm/ha for a 120-day variety (Sanio) (Dancette, 1983). The annual precipitation in these regions fluctuated during the years of trial (1973-1977) between 374 mm and 510 mm, illustrating the extreme risk of crop failure in semi-arid regions. The experience from these trials is that fast-growing pearl millet (75-90 days) requires around 1 500 m3 of water per ton of grain produced (dry matter), and the 120-day varieties require 3 000-4 000 m3/t dry matter. However, looking at the total production of overground biomass, the water requirement is almost the same for all varieties; around 400 mm/t/ha, thanks to the large production of straw for the 120 day variety(around 10-13 t/ha compared with 5-6 t/ha for the fast-growing varieties).

For sorghum, FAO (1979) report a water requirement of 1 000 to 1 700 m3/t grain and 530-1250 m3/t grain for maize. For pearl millet, Klaij and Vachaud (1992) present results from trials in Niger where low-fertilized millet consumed around 2 000 m3/t dry matter yield and high-fertilized millet around 900 m3/t.

Le Houérou (1992) points out the determining role of rainfall variability and seasonal modality on water requirement for crop production. In North Africa with a coefficient of variation CV (the ratio between the standard deviation and the mean precipitation level) of 30-35 percent, the threshold for commercial cereal production corresponds to the 400 mm isohyet (assuring a crop yield 4 years out of 5). In arid regions of Australia, cereal production is practiced in regions receiving not more than 280 to 300 mm rainfall when the CV is in the order of 25 percent, while in regions with CV attaining 30-35 percent cereal cultivation does not descend below 400-450 mm isohyets. In the Middle East, where rainfall variation is lower than in western and northern parts of Africa, cereals are cultivated in regions receiving only 275-300 mm.

The fact that the rainy season is monomodal and concentrated during the winter in the Middle East, compared with the bimodal seasons in North Africa, also influences this difference in general water requirement. Similar differences can be noted when comparing cereal production (sorghum and millet) in West and East Africa. In the West, with a short monomodal crop season, the rainfall variability is lower than in the Eastern parts. Millet and sorghum production in East Africa requires around 650-700 mm of rain, compared with around 400-500 mm in West Africa.


With the water and soil resource being finite, the only options for increasing biomass production in rainfed dryland agriculture is to increase the water productivity, i.e. the water use efficiency (WUE), by producing more biomass per unit of water. This is clearly illustrated by using Gregory's (1989) definition of WUE (Equation 1), based on the actual rainwater supplied (yield per unit of water supplied), and which takes into account all the water flows of the hydrological cycle involved in biomass production. This expression of WUE is motivated by the large water losses in dryland agriculture due to the erratic and unpredictable rains combined with physically fragile, crust-prone soils, which result in large water "losses" in the form of surface runoff (S) and deep percolation (D).

Y = yield

S = surface runoff

T = transpiration

D = deep percolation

E = evaporation

The rain will be divided into "productive" and "unproductive" water flows depending on soil and plant conditions prevailing in the two partitioning points of the hydrological cycle: (1) rain partitioning at the soil surface determining the fraction of P divided into infiltration, surface runoff and soil evaporation (plus interception); and (2) flow partitioning in the rootzone determining the ratio between root water uptake and deep percolation (plus eventual capillary flow of water leading to evaporation) (Figure 2).

To attain an increased water productivity, the aim must be to minimize "unproductive" water losses at the two partitioning points. This means minimizing the evaporation losses in favour of productive transpiration, and minimizing "losses" in the form of runoff (surface and deep percolation). The key to success here is to a large extent to be found in the application of adequate soil and water management techniques that guarantee a maximum of infiltration and transpiration. To these are added techniques aimed at "harvesting" and reusing runoff water for protective irrigation.


Reduced crop yields are often caused by short periods of crop water deficit during critical growth stages. Severe periods of water stress can even lead to complete crop failures. The determining role on yield levels of water availability during early vegetative stages, flowering and grain filling, has been widely documented (see for example FAO, 1979; Dancette, 1983). Forest and Lidon (1984) present a strong correlation between "crop water satisfaction", defined as the ratio between actual evapotranspiration (ETR) and potential evapotranspiration (ETA/PET), and crop yield during the first 50 days of growth. An average ETA/PET ³; 0.8 is suggested as a necessary threshold to attain high sorghum yields ( > 3 t/ha), and a ratio ETA/PET ³; 0.65 to attain yields greater than 2 t/ha. Sivakumar (1992) suggests that the tendency to reduced crop yields in the Sahel zone since the mid-1960s is correlated to a significant decline in August rainfall (corresponding to the critical growth stages of flowering and grain filling), and the shortening of the growing season by 5-20 days (increased variation of the onset and the date of ending of the rains).

The strong intra-annual rainfall fluctuations have to be accepted as a normal part of the hydroclimatic reality. Under these circumstances, protective irrigation during periods of water stress, based on locally available water surplus, constitutes an important method for the droughtproofing of biomass production (Stewart, 1989). Yields have shown to increase substantially when crop water supply is stabilized by the use of combinations of small-scale supplementary irrigation schemes and runoff collection (Perrier, 1988). The crucial question here is of course to what extent there actually exists a surplus of runoff water which it is possible to collect and store. This has to be evaluated on the basis of the actual agroclimatic conditions and the hydrological distribution on the watershed (availability of "blue" and "green" water as a result of different water needs of the vegetation, animals and humans, and different amounts of water lost to the atmosphere).

FIGURE 2. Water use efficiency

Different water-use efficiency estimates as seen from an agricultural perspective. The diagram shows the two principal partitioning points in the soil profile (circles), the resulting water flows and their relation to respectively the water consumed (WUET. WUEET) and water supplied (WUEP, WUEirri, WUEtot). P = precipitation, T = transpiration, E = evaporation, Ei = evaporation via interception. S = surface runoff, D = deep percolation. Qin = river inflow from upstream watershed, Qut = river outflow. Qin - Qut = potential water availability for irrigation (Falkenmark and Rockström. 1993)

The potential of protective irrigation in dryland agriculture is presumably related to the socioeconomic and cultural environment. Being plagued by poor soils and an unreliable rainfall, the main concern of subsistence farmers in the drylands is to guarantee a stable, minimum yield all years, rather than high, but more risky (drought sensitive) and costly yields through cost-intensive mechanized and fertilizer-based cultivation of improved varieties (Brouwer et al., 1992). Even if high input rainfed agriculture is unlikely to be widespread in the semi-arid tropics in a foreseeable future, a more efficient use of existing water through protective irrigation can eventually make the introduction of yield-increasing measures more viable (use of mineral fertilizers, of crop varieties which respond better to fertilizer, etc.).


Water productivity in biomass production is closely interlinked with water availability to nutrients (especially phosphate (P) and nitrogen (N) for the African dryland agriculture). Water supply in the Sahel and similar semi-arid regions cannot be effectively managed for production without addressing soil fertility constraints (Payne et al., 1992). With adequate amounts of soil moisture, yield response to nutrients is significant, but during severe droughts mineral application can actually reduce yields (Power, 1990; Christianson and Vlek, 1991). Trials on the yield response to nitrogen for food crops in the sub-humid (maize) and semiarid regions (pearl millet, sorghum) of West Africa have resulted in strong yield increases, and therefore in increased water productivity (Christianson and Vlek, 1991).

One of the, perhaps not surprising, conclusions from these trials is that the N-rate required to reach the point of optimum economic response to fertilizer use decreases as the rainfall available to the crop is reduced. For humid regions, application of up to 120 kg N/ha gives good economical yield returns for maize, while it does not seem profitable to exceed levels of 80 kg/ha in sub-humid regions. For semi-arid regions the optimal N-level drops to 30 kg/ha for millet and 50 kg N/ha for sorghum. Even if the response to N-fertilizing is much lower in semi-arid, compared with more humid regions, the optimal N-level drops to 30 kg/ha for millet and 50 kg N/ha for sorghum. Even if the response to N-fertilizing is much lower in semi-arid compared with more humid regions, the yields increased by 75 percent when applying 30 kg of N/ha; from around 600 kg grain/ha to 1 050 kg/ha (adequate phosphate, sulphur and other nutrient levels being guaranteed, as well as good management).

The combined effect of different levels of soil moisture and plant-available minerals on pearl millet yields is clearly illustrated from trials at the ICRISAT research station in Niger (Table 1) (Christianson and Vlek, 1991). Even during a "dry" year (defined by the amount of mid-season rainfall) the yields increased by 25 percent when applying 30 kg of N/ha. When going from a "dry" to a "medium" rainfall year, yields increased by 50 percent only as an effect of increased soil moisture (zero N-input), while a 30 kg input of N increases yields by 230 percent. Under "humid" conditions 30 kg of N (an amount which under these conditions lies under the optimum level) resulted in a 320 percent yield increase. Here the "moisture effect" corresponds to a 90 percent increase of yields. This means that even though yields almost double when going from a "dry" to a "humid" year the sole fact of applying 30 kg of N leads to a more than threefold increase of yields, and 1.7 times higher yields under the same "humid" conditions (going from zero input to 30 kg input in a "humid" year).

FIGURE 3 Evolution of sorghum yield under different fertility levels

TABLE 1. Millet grain yield response to Nitrogen under different amounts of mid-season rain. Yield level for a "dry" year with no N applied (Y = 480 kg grain/ha) is set to 1 (adapted from Christianson and Vlek, 1991)

Hydroclimatic characteristic of growing season1

Yield response to N-level (kg)












1 Based on mid-season rainfall; dry = 100 mm, medium = 235 mm, and humid = 350 mm.

Table 1 data took into account only the yield response to nitrogen, all other aspects being assumed more or less optimal. With the present situation of shortened, and often abandoned, fallow periods in large parts of the semi-arid and arid tropics, it is of interest to analyse the long-term effects, on yields, of different types of management and fertility reproduction practices in continuous cereal cultivation.

In West Africa, at the IRAT (Institut de Recherche Agronomique Tropicale) research station in Burkina Faso, the effects of soil fertility and yields for a monoculture of sorghum has been studied during 18 years (1962-1981), for different levels of mineral fertilizer, mulching, and input of animal dung (Forest and Lidon, 1984; Pichot et al., 1981). The evolution of the sorghum yield is presented in Figure 3 for four treatments (out of six performed): (1) high input of mineral fertilizers combined with high input of manure (40 t/ha until 1976, once every two years thereafter) (FERT + ORG); (2) low input of mineral fertilizer combined with a lower annual input of manure (5 t/ha until 1976, once every two years thereafter) (Fert + Org); (3) low input of mineral fertilizer combined with application of crop residues every two years (Fert + Mulch); and finally (4) the Control with no inputs.

In the same figure is plotted the rate of crop water satisfaction (ratio of real evapotranspiration (ETR) to the potential evapotranspiration (PET) for each year; (i) during the first 50 days of crop growth, and (ii) from 5 September to 15 October. There is, not surprisingly, a correlation between crop water deficit and grain yield, especially during the first 50 days of crop growth. But what is also interesting to note is the large permanent difference in yield level, independent of crop water deficit, between the different soil management techniques. Monoculture of sorghum with no fallows and no external mineral inputs (Control) leads to a yield decline from around 1 500 to 200 kg/ha already after 2 years of cultivation. The yield seems to stabilize on this lower level, thereafter only slowly decreasing, and attains at the end of the series not more than around 150 kg/ha. All the other treatments present large fluctuations, due to the hydroclimate, but the yields are "stabilized" on a higher yield level, where even the absolute lowest yields are at least three times higher than the purely "nutrient mining" sorghum crop. The average crop yield for the strongly fertilized sorghum crop is 2 295 kg/ha (FERT + ORG), i.e. more than 10 times higher than the control (average yield = 208 kg/ha). The crop with low input of mineral fertilizers and low levels of manure (Fert + org) attains an average yield of 1 250 kg/ha, around 6 times higher than the control. Finally, the crop with input of crop residues (Fert + mulch) reaches a similar average yield level to the crop with low inputs; around 6 times higher than the control (Y = 1180 kg/ha).

A result worth noting is the residual effect on crop yields when combining the application of crop residues and fertilizers. In trials at ICRISAT, Niger, the yield level for pearl millet three years after application of crop residues (5 t/ha) and fertilizers (P + N), was nine times higher than the control (1 800 kg/ha compared with 200 kg/ha) (Bationo and Mokwunye, 1991b). Residual effects have also been demonstrated for P-fertilizing (Bationo et al., 1992). With zero input, yields fell to around 100 kg/ha after three years of cultivation, while 30 kg of P/ha applied in Year 1 resulted in a yield of around 450 kg/ha in Year 2 and 170 kg/ha in Year 3.

An important effect of fertilizing is that the plant density at harvest is higher than without fertilizers. This suggests an increased survival of fertilized plants. Root and shoot growth is faster, the root system more developed, which enables the plant better to withstand early drought stress. Survival of young plants is higher, which suggests that applying fertilizers constitutes a strategy for droughtproofing crops.

All the results presented above, on the integrated effect of water and nutrients on crop yields, are carried out under rainfed conditions. This illustrates the potential of producing more biomass on the same amount of water supplied. Every increase in yield is thus equivalent to an increase in water productivity, with increased WUE. The crucial question here is of course which water flows are converted to productive transpiration-flow needed to meet increased crop water requirement, in pace with increased biomass production? Is it water previously lost as evaporation or as runoff water, that is converted to productive transpiration? The former is a net water loss while the latter constitutes the source of river and groundwater recharge, which form the basis for household and other water supply. It is also worth noting that experiments indicate that even the transpirational WUE (WUEt = amount of grain produced per unit of water transpired), can be increased by applying phosphate (Payne et al., 1992). Satisfying the nutrient demand of a crop that is not water-stressed increased the WUEt by a factor of four (outdoor trials in pots), which illustrates the potential for increasing the biomass produced per unit of water consumed.

Despite the positive effects on crop yields, and on WUE, of applying phosphate and nitrogen, it is not evident that the use of fertilizers is economically viable in the drier climate zones. Especially for nitrogen, which in semi-arid regions frequently is not the limiting factor, at least not initially, it is stated that its application is economic only in areas receiving more than 900 mm of annual rainfall. Application of NPK-fertilizers and urea on millet and sorghum is risky, and only marginally profitable on average, except in the Guinean zones (World Bank, 1986).


As described above, the two principal options to increase water productivity in dryland agriculture are: (i) to maximize infiltration and water retention in the soil, i.e. converting unproductive water flows to productive transpiration and (ii) to supply water during periods of crop-water deficit through protective irrigation. This necessitates an integrated approach to soil and water management, where the choice of soil management techniques (tillage and ploughing systems, nutrient transfer systems, etc.) have to be adapted to the hydroclimatic environment (e.g. the difficulty of introducing ploughs in areas with coarse textured soils in the Sahel, which is partly explained by the increased depletion of organic matter and greater risk of soil loss due to wind or water erosion which follows from the increased turning of the soil). Also, the choice of hydrological techniques (ridging, water harvesting, drainage, terraces, etc.) has to take into account pedological aspects, not only the physical and chemical composition of the soils, but also the dynamics of these soils when submitted to different types of agricultural practices (vulnerability to crust formation, wind and water erosion, carbon depletion, soil pans, etc.).

Management of soil and water in agriculture obviously includes cropping techniques. Traditional systems of contour cultivation have the double objective of: (i) concentrating relatively more nutrient-rich surface soil around the roots of cultivated plants, and (ii) preventing surface runoff. Different cropping systems (rotations), integrating different annuals eventually with perennials, e.g. alley-cropping and agroforestry, have the double function of (i) increasing the vegetative cover both in time and space, in order to reduce runoff and evaporation losses and (ii) contributing to the fertility reproduction and structural maintenance of the soils, by combining cereals with N-fixing pulses, trees, etc.

Another cropping aspect worth mentioning is the density of the crop grown. Generally the cropping density in traditional dryland agriculture decreases with increased aridity, due essentially to the lack of soil moisture. On-farm trials in the southern parts of Niger, a region with mean annual P = 551 mm, with sandy soils (>95 percent sand) having low organic content (average 0.3 percent), suggest that increased planting density, from the original low density of around 5 000 pockets/ha, has a large effect on yields when phosphate and nitrogen are applied (30 kg of each) (Bationo et al, 1992). Each 1500 pocket/ha increase (starting from a low density of < 3 500 pockets/ha) resulted in a yield increase of approximately 200 kg/ha (up to around 10 000 pockets/ha).

Finally, to the more technical questions considered above, must be added the perhaps most important aspect; proposed soil and water management techniques have to be economically feasible and socio-culturally acceptable by the present rural communities, not only in the short but also in the long term.

The literature covering different soil and water management techniques is extensive (see for example FAO, 1988b; Jackson, 1989; Reij et al., 1988; Sivanappan, 1989; Unger et al., 1988; World Bank, 1986). Many experimental results (see for example FAO, 1986; Vanketeswarlu, 1987) indicate that there is a potential for increasing biomass production within a given water resource, by combining water conservation, soil management, protective irrigation, and the use of external inputs (fertilizers, improved crop varieties, etc.). The techniques applied are often based on an evolution of soil and water management practices of indigenous origin.


Water scarcity is an everyday reality for dryland farmers, and water can be considered as the principal limiting factor in biomass production. It is thus not surprising that attaining maximum water productivity and coping with recurrent droughts are both ancient and present endeavours of rural societies. In large parts of the semi-arid and arid tropics, more than 80 percent of the population are subsistence farmers, making their living from a rainfed agriculture based on traditional, often manual techniques. Severe droughts, even as late as the beginning of the 20th century, have been met largely through pioneer migration of a certain number of family lineages from established villages to "downstream" virgin lands, which were then put under cultivation. Famines functioned as a seed to new villages. Security storage of grains from "good" years for consumption in "bad" years, transhumance, spreading of risks for crop failure by cultivating different species in different parts of the landscape, are some examples of "drought-adapting strategies" of these rural societies.

The recent evolution seems to indicate that a rural crisis, manifested among other things by human-induced landscape degradation, famines and urban migration, has forced farmers gradually to abandon traditional security systems.

This situation is aggravated by the problems that these farmers normally have a very low purchasing power; property rights are poorly defined; and the institutional framework is weak and incapable of adequately handling price mechanisms, legal and commercial issues, credit systems, education and the diffusion of agronomic techniques.

The objective of this brief description of some central aspects of the socio-economical environment, is to point out the importance of these factors, not only as catalysts behind rural crisis, but also to identify conditions that may determine the life span of development efforts.


Water scarcity in dryland agriculture cannot be explained only by hydroclimatic fluctuations but is to a large extent caused by soil and water management failures. The major structural transition contributing to this failure is suggested to be the abandoning of an agricultural system based on fertility reproduction through long fallows, without introducing a sustainable alternative.

Diagnostics on the causes behind, and the effects of, imbalances between human carrying capacity and demographic pressure, should take into consideration the four interacting domains influencing the "health" of rural societies; the environmental preconditions (hydroclimate, land, vegetation), the agricultural techniques used, the socio-economic environment, and finally the dynamics, or driving forces, behind the evolution of agricultural systems.

In order to increase water productivity, it seems clear that both soil fertility and crop water availability during all growth stages, have to be improved. The challenge is, on the basis of an integrated approach to land and water, to develop management strategies that: (i) make the best possible use of a limited water resource, i.e. minimize unproductive water losses, especially evaporation from the soil and from interception, and (ii) that guarantee the long term productivity of the soil. One crucial point, from an agrohydrological point of view, is that nutrient and moisture issues cannot be addressed apart; the one being the limiting factor for growth when the other is abundant. Proposed management practices to tackle this precondition for successful agricultural performance can form a sustainable solution only if accepted within the socio-economic context.


AGRHYMET. 1991. Atlas agroclimatique des pays de la zone de CILSS.

Bationo, A. 1993. Personal communications. The Sahelian Centre of the International Crops Research Institute of the Semi-Arid Tropics (ICRISAT), BP 124 04, Niamey, Niger.

Bationo, A. and Mokwunye, A.U. 1991a. Alleviating soil fertility constraints to increased crop production in West Africa: The experience in the Sahel. Fertilizer Research 29: 195-215.

Bationo, A. and Mokwunye, A.U. 1991b. Role of manure and crop residue in alleviating soil fertility constraints to crop production: With special reference to the Sahelian and Soudanian zones of West Africa. Fertilizer Research 29: 217-225.

Bationo, A., Christianson, C.B., Baethgen, W.E. and Mokwunye, A.U. 1992. A farm-level evaluation of nitrogen and phosphate fertilizer use and plant density for pearl millet production in Niger. Fertilizer Research 31: 175-184.

Brouwer, J., Bouma, J. and Renard, C. 1992. Improving water and nutrient use efficiency in subsistence farming in the Sahel: Making the most of a poor and very variable environment. Progress report February 1990 - August 1992. ICRISAT, Niamey, Niger/Wageningen, Netherlands.

Brown, M. and Goldin, I. 1992. L'avenir de l'agriculture: Incidences sur les pays en développement. Etudes du centre de développement. OECD, Paris, 1992.

Christianson, C.B. and Vlek, P.L.G. 1991. Alleviating soil fertility constraints to food production in West Africa: Efficiency of nitrogen fertilizers applied to food crops. Fertilizer Research 29: 21-33.

Dancette, C. 1983. Estimation des besoins en eau des principales cultures pluviales en zone soudano-sahelienne. Agronomie Tropicale 38 (4): 281-294.

Falkenmark, M. 1986. Fresh water - Time for a modified approach. AMBIO. XV(14)

Falkenmark, M. et al. 1990. Coping with water scarcity - Implications of biomass strategy for communities and policies. Water Resources Development. 6(1). March 1990.

Falkenmark, M. and Rockstrom, J. 1993. Curbing rural exodus from tropical drylands. AMBIO. 22(7): 427-437.

FAO. 1978. Reports of the agro-ecological zones project. World Soil Resources Report 48, vol 14. FAO, Rome.

FAO. 1979. Yield response to water. Doorenbos, I., and Kassam, A.H. FAO Irrigation and Drainage Paper 33. FAO, Rome, 1981.

FAO. 1986. African agriculture: the next 25 years. FAO, Rome.

FAO. 1988a. World Agriculture: Toward 2000, An FAO study. Ed. Alexandratos N. London.

FAO. 1988b. Soil and water conservation in Semi-arid areas. FAO Soils Bulletin 57. FAO, Rome.

Forest, F. and Lidon, B. 1984. Influence de régime pluviométrique sur la fluctuation du rendement d'une culture de sorgho intensifiée. Agronomie Tropicale. 1984: 247-260.

Gregory, P.J. 1989. Water-Use-Efficiency of crops in the semi-arid tropics. In: Soil, Crop and Water Management in the Soudano-Sahelian zone. ICRISAT.

Jackson. 1989. Climate, Water and Agriculture in the Tropics. Longman, London.

Klaij, M.C. and Vachaud, G. 1992. Seasonal water balance of a sandy soil in Niger cropped with pearl millet, based on profile moisture measurements. Agric. Water Measurements 21: pp 313-330.

Le Houérou, H.N., gingham, R.L. and Skerbek, W. 1988. Relationship between the variability of primary production and the variability of annual precipitation in world arid lands. Journal of Arid environments 15: pp 1-18.

Le Houérou, H.N. 1992. Relations entre la variabilité des précipitations et celle de productions primaire de l'Est. In: L'aridité: Une contrainte au développement. Caractérisation, Réponses biologiques, Stratégies des societes. Ed Le Floc'h, E., Grouzis, A., Bille, J.C., ORSTOM, Paris.

L'vovitch, M.L. 1979. World Water Resources and their future. Translation by the American Geophysical Union, Lithocrafters Inc. Chelsea, Michigan, USA.

Maley, J. 1981. Etudes palynologiques dans le bassin du Tchad et paléoclimatologie de l'Afrique nord-tropicale de 30 000 ans à l'époque actuelle. Travaux et Documents de l'ORSTOM, Paris, 1981.

Nicholson, S.E., and Flohn, H. 1980. African environmental and climatic changes in late Pleistocene and Holocene. Climate Change. 2 (4): 313-349.

Parr, J.F. and Stewart, B.A. 1990. Improving rainfed farming for semi-arid tropics. implications for soil and water conservation. Soil Conservation Society of America, Ankeny, Iowa, USA.

Parr, J.F. et al. 1990. Improving the sustainability of dryland farming systems: A global perspective. In: Advances in Soil Science. vol 13. Dryland agriculture strategies for sustainability. Ed. Singh R.P. et al. New York.

Payne, W.A., Drew, M.C., Hossner, L.R., Lascano, R.J., Onken, A.B. and Wendt, C.W. 1992. Crop ecology, production & management. Soil phosphorous availability and pearl millet water-use efficiency. Crop Sci. 32: 1010-1015.

Perrier, E.R. 1988. Opportunities for the productive use of rainfall normally lost to cropping for temporal or spatial reasons. In: Drought Research Priorities for the Dryland Tropics. Bidinger, F.R. and Johansen, C. (eds). ICRISAT, Patancheru, India: pp 113-131.

Pichot, J., Sedogo, M.P., Poulain, J.F. and Arrivets, J. 1981. Evolution de la fértilité d'un sol ferrugineux tropical sous l'influence de fumure minérales et organiques. Agronomie Tropicale. 36(2): 122-133.

Power. 1990. Fertility management and nutrient cycling. In: Advances in Soil Science, vol 13. Dryland agriculture strategies for sustainability. pp 131-147. Ed Singh R.P. et al. New York.

Rapp, A. 1988. Mellanösterns klimathistoria under 10 000 år och klimatanpassat ökenjordbruk. SvenskGeografisk årsbok årg. Lund 64: 9-25.

Reij, C., Mulder, P. and Begemann, L. 1988. Water harvesting for plant production. World Bank Technical paper No. 91. World Bank, Washington, DC.

Sivanappan, R.K. 1989. State of the art - On soil and water conservation, water harvesting and eco-restauration in Western Ghats. SIDA, Workshop on Water Harvesting. Tamil Nadu, India.

Sivakumar, M.V.K. 1992. Agroclimatology of West Africa: Niger. Information Bulletin No. 5. ICRISAT, 2nd edition. Niamey, Niger.

Stewart, B.A. 1989. Conjunctive use of rainfall and irrigation in semi-arid regions. In: Soil, Crop and Water Management in the Sudano-Sahelian Zone. ICRISAT, Patancheru, India.

Unger, P.W., Jones, O.R. and Steiner, J.L. 1988. Principles of crop and soil management procedures for maximizing production per unit rainfall. In: Drought Research Priorities for the Dryland Tropics. Bidinger, F. R. and Johansen, C. (eds). ICRISAT, Patancheru, India.

Valentin, C., Hoogmoed, W. and Andriesse, W. 1991. The maintenance and enhancement of low-fertility soils. In: Evaluation for sustainable land management in the developing world. IBSRAM Proceedings: Technical papers. vol 2.

Venketeswarlu, J. 1987. Efficient resource management systems for drylands in India. Advances in Soil Science. 7: 209-216.

World Bank. 1986. The environment of rainfed agriculture under arid and semi-arid conditions. In: Development of Rainfed Agriculture under Arid and Semi-arid Conditions. Ed Davis, T.J., pp 82. Proceedings of the Sixth Agricultural Sector Symposium. World Bank, Washington DC.

World Bank. 1989. Sub-Saharan Africa - From crisis to sustainable growth. A long term perspective study. Washington DC.

Contents - Previous - Next