1. INTRODUCTION

2. CONCEPTS

Water Stress and Crop Yield
Effectiveness of Rainfall
Figure 1 Possible Destinations of Rainwater
Loss of Soil Voids; Surface Sealing
Organic Matter and Recuperation Processes
The Importance of Crop Cover
Runoff Control
The Policy Environment
Soil Capacities and Soil Factors
Capacities
Factors
Introducing Change
Box 1: Relations Between Soil Capacities and Governing Factors
Box 2: Some Shifts in Emphasis Regarding Water and Soil Conservation

3. CHARACTERISING UNCERTAIN RAINFALL CONDITIONS

Rainfall Flags
Figures 2-5: Example of Rainfall "flags"
Characterizing Rainfall In Ten-day Units (dekads)

4. PROJECT DESIGN

Annex 1: Relevant Further Reading

Table 1: Field Sheet for Diagnosing a Water Conservation Problem or Characterising a Water Conserving Intervention

The success of dryland agriculture depends on there being an adequate supply of water in the rooting zone of the soil, freely available to crop plants. The designers of investment projects to support dryland agriculture often analyse rainfall statistics in some detail. However they have tended to give less attention in project designs to measures which will promote the more efficient capture of precipitation in situ, improve the availability of that moisture in the rooting zone, and optimise its subsequent use by crops. At worst, this omission has led to unnecessary and costly proposals to provide supplementary water through irrigation. Yet a couple of examples indicate the scale of possible benefits simply from making more efficient use of rainfall. It has been estimated that adoption of improved water conservation technology in the Central Great Plains of the USA made the largest single contribution (45%) to the increase in average wheat yields from 750 kg/ha to 1800 kg/ha between 1936 and 1977, ahead of improved wheat varieties (30%), equipment (20%) and fertilizer practice (5%) (Meyer, quoted in FAO 1991c). Day and Aillery (1989) suggest that disposable incomes of farmers in Mali could be more than doubled if rainfall capture could increase from 40% to 60%, and could be even further increased if 80% of rainfall were captured.

This does not imply that investment projects should be designed solely to conserve soil moisture, in the manner that projects (usually unsuccessful) to conserve the soil itself were designed in the past to address problems of erosion. Measures to reduce the frequency of sub-optimal soil moisture conditions should, however, be an underlying concern in the technical strategies of any land and water management project, and in any other rural development programmes and projects whose benefits derive from rainfed plant production. Indeed in many such projects, inclusion in the design of measures to improve soil moisture conditions will be doubly important. Responses of rainfed crops, fodders or woody species to other inputs needed to raise output may not be realisable without sufficient moisture first being available. In an upward spiral, better response from nutrient supply and improved plant material will in turn allow more efficient exploitation by the plant of available soil moisture. At the same time more water infiltration also reduces the risks of erosion by surface runoff, thus combining moisture conservation with soil conservation.

This document aims to review available technologies to improve in situ capture of rainfall and its more efficient use by crop plants. It concerns areas with restricted, irregular or markedly seasonal rainfall, mostly in places where the annual average is from 400 to 1,000 mm but also in other areas where seasonal shortages of rainfall can limit crop productivity. It is directed in particular at staff of the FAO Investment Centre and their counterparts in government teams responsible for project design. It assumes that readers have some working knowledge of agronomy, soils and plant/soil/water relationships. It should be used alongside broader Investment Centre guidelines on how technical, socio-economic and other considerations should be woven into the design of agricultural investment projects (FAO 1992, FAO 1993). By combining the technical approach advocated here with that broader guidance, it is hoped that users will be better able to decide how to promote more efficient capture and use of rainfall in specific dryland settings. It assumes that, depending on physical conditions and the socio-economic circumstances of farmers, readers will identify either technologies which can already be recommended with confidence, or ideas for testing in pilot locations or communities, or technologies for farmers to experiment with individually and perhaps modify themselves, or simply a need for more conventional research. Often the conclusion will be that a given project needs to support more than one of these lines of action simultaneously.

The intention here is to provide a framework in which to make the technical decisions involved in the above process. It is assumed that an understanding of the socio-economic factors will also be obtained, using parallel rural appraisals of the sort increasingly use by the Investment Centre in project design. Technical and socio-economic findings then need to be integrated in deciding on the specific measures to support more efficient use of soil moisture.

The document aims first to improve the user's understanding of the desirable technical capacities which the soil should have for water use to be efficient, and of the individual technical factors which affect those capacities. By this means ranges of potential interventions can be identified, allowing for more precise matching of possible solutions to particular problems. It is assumed that potential interventions will then be screened taking account of farmers' socio-economic constraints and preferences, to choose the measures to be supported with external investment.

Recurrent water stress causes a progressive reduction in the potential yield of a crop, in comparison to the yield which could have been achieved without water stress. The soil-moisture deficit (moisture tension) at which potential yield starts to be depressed varies between plant species, and the stage of growth. The difference between the first onset of a soil moisture deficit and the point at which lack of water starts to cause yield loss, determines the degree to which crop productivity is buffered against the effects of periods of insufficient rainfall. Where the water in the soil cannot be regularly replenished at will, minimising transpiration demand and maximising the retention of water freely available in the soil will delay the onset of crop stress and hence reduce drought risks and damage. Another tactic is to change the crop species or variety to one which is more physiologically tolerant of drought, or to one which passes through its more sensitive stages of growth when dry spells are least probable. It should not be forgotten, however, that the removal of water stress will not restore yields if the crop is subject to another, overriding constraint such as disease or pest attack.

On arriving at the soil surface, rainwater is destined to travel by one or more of six routes across and through the soil:plant system, as shown in Figure 1.

1. Surface runoff
2. Direct evaporation from the soil surface
3. Transpiration through non-useful plants (weeds)
4. Transpiration through useful plants (crops)
5. Storage in the profile within and below the root-range of current crops
6. Deep through-flow to groundwater or drainage.

In many areas, poor yields of crops are related less to an absolute insufficiency of rainfall, and more to an insufficiency of moisture in the rooting zone because crop and land management do not optimise water flow along those routes which favour crop growth. Rainfall is ineffective if it is lost as surface runoff, evaporates from the surface or through weeds, or is not held within the root zone (routes 1, 2, 3, 5, 6 in Figure 1). Changes in management technology should seek to minimise water losses via these five routes and strategies should seek to optimise the proportion channelled into route 4 - transpiration through useful plants.

In some circles in the past, declines in soil productivity were mainly equated with erosional loss of soil particles, and plant nutrients with them. Undue focus on falling nutrient status tends to obscure the important contribution to declining productivity also arising from the collapse or destruction of soil-structural units, and the consequent loss of voids within and between structural aggregates and soil particles. In the top few millimetres the impact of high-energy raindrops on exposed soil may compact and seal the uppermost pores, reducing the infiltration rate. Undue tillage and trampling of soils, causing pulverization and compaction, can reduce the size of voids at lower levels in the soil profile. Destruction of these voids in the soil slows the movement of water into and within the soil and reduces storage capacity. It may even be severe enough to cause waterlogging and limit aeration within the root zone. The first symptoms of land degradation may result from plants being deprived of rainwater due to loss of internal voids or surface sealing, rather than being starved of nutrients.

Cover of the soil, whether with live or dead material, to provide mechanical protection against the impact of high-energy raindrops is important to prevent breakdown of soil micro-aggregates, dislodging of soil particles and sealing of the surface layers. For soils with a fragile internal structure, organic matter, and in some cases lime, are also important for the building of stronger structural units at superficial or deeper levels in the profile.

Once a soil structure has been lost, it may take several years to restore. Organic materials and processes are of particular importance in the maintenance or restoration of a soil structure to a condition allowing easy entry of water and its storage in plant-available form. Plant residues, whether from crops or green manures, in addition to providing direct protection against surface sealing by raindrops, provide organic substrates for microbial action, leading to release of plant nutrients and formation of structure-enhancing humic gums. They also promote burrowing by meso-organisms and root penetration. Thus, acting in various ways, organic materials generally give far more lasting improvements in soil structure than mechanical treatments alone, which shatter rather than build up the soil's structural units.

On the other hand, since their degradation is dependent on temperature, organic materials are more difficult to build up and retain in tropical soils than under temperate conditions. A further problem facing the project designer is that farmers may have other priorities for the use of their organic materials such as crop residues - eg. to feed their livestock or as a source of cash income. Furthermore they may not initially have the spare time or cash to incorporate techniques like green manuring into their production systems. Inorganic fertilizer materials often may need to be added to the system (a) to replace plant nutrients exported from the farm in harvested materials, (b) to improve yield potential and root-zone quality in combination with soil organic matter, and (c) to produce sufficient plant materials both for harvesting and for return of organic plant residues onto and into the soil. Some form of outside support may therefore be needed before the "benign spiral" of rising biomass production and income which is initiated by organic matter accumulation can begin to turn.

Since most tillage and grazing systems involve some degree of damage to soil structure, they should generally include regular periods of soil recuperation for adequate structural conditions to be re-established. In sustainable production systems, recuperation is normally via bush fallows or managed grass breaks. Both bring about their benefits to productivity mainly because they replenish soil organic matter and restore soil structure. Where population is dense and cropping is continuous, more intensive measures such as agroforestry, mulching or application of organic manures and composts may become necessary. Where soil degradation is a hazard there are four main means of increasing the effectiveness of recuperative periods:

• increasing their duration;
• managing the crop or fallow to favour moist, adequately-aerated conditions which maximise the biological transformation of soil organic matter into humic gums;
• reducing the length of the cropping or grazing period during which damage occurs;
• reducing the severity of damage (e.g. by tillage or trampling) in this antecedent period.

In practice, for many small farmers the use of fallows or rotations may be their only means to maintain a soil structure which optimises water retention and crop production. On the other hand - as for agroforestry, mulching with crop residues or organic manuring - land pressure and other socio-economic constraints may impede their initial adoption by farmers.

An evenly-distributed crop stand optimises the protection given by the leaf canopy against surface sealing of the soil under the impact of raindrops. Simultaneously, it ensures that the crop makes maximum use of the available space for growth and yield accumulation. Subsequent crop residues when spread over the surface, combined with organic matter left in the soil, can further enhance rainfall infiltration in situ and water retention. Attention to crop cover can therefore improve the moisture status of the soil while at the same time initiating a benign spiral of rising productivity. Compared with a frontal attack on runoff and soil loss based on mechanical barriers, biomass-mediated "conservation by stealth" of this sort usually costs less and generates earlier and larger benefits. It is therefore much more likely to be adopted by land users.

Methods which rely on an even crop cover and increased biomass to improve the soil moisture regimes should therefore be favoured, except in areas so dry that the only way to grow a crop at all is to concentrate runoff through water harvesting structures.

Even in semi-arid areas, however, rainfall intensity will at times be greater than the maximum possible rate of infiltration of water into the soil, so that some water remains on the surface. This may cause temporary flooding if it does not move away, or may become potentially damaging runoff if it moves rapidly downslope. In semi-arid to subhumid situations, contoured cultivation - especially where tined implements are used or where contour bed-and-furrow systems are constructed - may almost completely overcome this problem in most years. Each furrow or bed becomes a "mini absorption bank" which can detain, and favour the infiltration of, most runoff. However, in all settings severe storms will occur at infrequent and irregular intervals, and additional protection of the land will need to be provided by permeable barriers of stones or vegetation for slowing and spreading concentrated runoff, even though all feasible measures to promote on-the-spot infiltration have been taken. In higher-rainfall areas it may be desirable rather to divert such runoff by impermeable earth banks across the slope to well-stabilised watercourses, although situations for achieving and adequately maintaining such a system may be hard to find.

A technical approach to the more efficient capture of rainfall which is based on an appropriate mix of the above concepts will only be adopted by farmers if it provides them with net benefits, whether in monetary or other terms. This implies a conducive policy environment, especially with regard to input and output prices, and adequate support services. Incentives and support services must furthermore also be stable and sustained if farmers are to maintain specific improved practices. Arbitrary alterations by governments of policy or the support framework may inadvertently provoke altered farm-level decisions, which in turn can encourage soil-degrading practices that are antagonistic to optimal capture and use of rainfall.

As the previous paragraphs imply, for optimum capture, percolation, storage and use of soil moisture, three physical capacities of the soil are important:

• the capacity to allow water to enter;
• the capacity to allow water to move readily through the profile;
• the capacity to store the acquired moisture in the crop rooting zone, and release it to plant roots in response to their evaporative demand.

If the roots of crop plants are to penetrate the storage layer so as to exploit the available moisture fully, three more capacities of the soil come into play:

The above capacities are governed by soil factors, many related to or capable of being modified by the way the soil is managed. They include cover over the soil, its structure and organic composition, and physical or chemical limitations within the profile. In many cases, though not necessarily all, factors which favour water penetration and storage also favour root growth. Hence if a soil's water-holding capacity declines over time as a result of poor management, yield losses will tend to be greater than those attributable merely to a physical reduction in water storage, because the volume of soil which is explored by roots may simultaneously be reduced.

Capacities may be influenced by more than one factor and a single factor may affect more than one capacity, giving a complex series of possible interrelations. These can conveniently be summarised in the form of a matrix, as indicated in Box 1 and the notes which follow it. The open circles within the matrix indicate the factors (listed down the side) which have the possible and most direct influence on the different capacities listed across the top. They also show, for a specific capacity, which of the factors in the right-hand list may have the most direct effect. The matrix can be used in two ways in devising a project technical strategy. If field diagnosis has identified deficiencies in one or more capacities as the major constraints to more efficient use of incident rainfall, the matrix can lead the user to those factors which, if amenable to modification, could ease these constraints. Alternatively, if changes in land use or management are being contemplated, their likely influence on the soil's capacity to capture and store rainfall efficiently can be anticipated.

Thus it can be seen from the matrix that several factors affect the soil's capacity, for example, for retaining plant-available water: they include particle-size distribution, the swell/shrink capacity of the clay, size of structural aggregates and their stability, pore-size distribution, amount and composition of organic materials, effectiveness of micro-organic transformations, the presence of gravel and stones in the profile and of dense horizons. Conversely one factor may affect several capacities. For example, the matrix shows that the factor "root channels" will affect the soil's capacities for allowing root extension and for aeration; such channels may traverse dense horizons and hence also increase the effective depth of the profile; and they will furthermore increase rates of percolation by providing low-tension passage for incoming water. The factors most readily accessible to outside modification relate to tillage and chemical status, and above all to organic matter content and processes - emphasising again the importance of the latter as already flagged.

Individual factors, particularly those related to organic matter and processes, may have both direct and indirect effects. For instance, the factor "organic materials on the surface" will directly protect the soil surface; but surface litter will also indirectly provide a substrate for the micro-organisms whose by-products will modify soil structural conditions; modified structure in turn will affect aeration and water movement in the soil.

In promoting measures to improve efficiency in the use of rainfall, as in introducing any other technical changes to farmers, the credibility of advisers is critical. Such credibility has to be earned. There must be a two-way flow of information between advisers and farmers for the purposes of mutual learning and understanding. Favourable factors when it comes to introducing changes to improve efficiency in rainfall use include:

• participating with farmers in discussing and deciding what changes will best fit the local context;
• respecting local priorities and decisions;

• building as far as possible on current perceptions and practices related to moisture conservation (which may be highly developed and relevant to the context) as a starting point for incremental improvements;

and perhaps most important of all,

• giving priority to those changes which show useful results quickly, are of little or no cost, simple to implement and maintain, and which can if necessary still function acceptably even under sub-optimal conditions.

Fortunately for the adviser, any change which can increase the soil moisture available to a crop in a season of drought stress is likely to give a very visible and immediate benefit.

As part of project design, Investment Centre staff must judge the ease with which a credible core of such advisers can in fact be built up, capable of propagating strategies for soil moisture conservation and management based on the largely biomass-mediated and participatory concepts just outlined. Prospects are likely to depend on the extent to which technicians and their managers have absorbed recent changes in the focus of thinking about water and soil conservation. Box 2 contrasts traditional with more recent emphasis, against which local attitudes can be compared. For further discussion see Shaxson et al, 1989.

Seen from the above viewpoint, some of the most appropriate financial investments to improve soil moisture conditions would be (a) in developing the technical capacities of advisory staff and of farmers themselves to characterise the problems and decide appropriate solutions, and (b) in assuring the availability and use of sufficient quantities of essential plant nutrients for the production of adequate biomass for harvests, for soil protection and for increasing water-holding capacity. This is likely to be more appropriate than providing grants or subsidies for construction of point-specific water-conservation structures with their associated problems of management and maintenance.

The design of measures to promote improved capture and efficient use of rainfall must start from the best possible understanding of the rainfall regime under which any investment project would be implemented. This chapter refers to two complementary approaches to improving understanding of the frequency, timing and reliability of rainfall: rainfall "flags" and rainfall analysis over successive ten-day periods ("dekads").

The mean annual rainfall figure at a specified dryland location gives a very poor indication of whether, and to what extent, crops are likely to suffer from moisture stress in a given year. A better indication of prospects - as many dryland farmers know -may be given by the date on which rains begin. The later they start, as a general rule, the greater the probability of a less-than-average total for that season.

Stewart (1988), using records from a succession of seasons, has shown that this phenomenon can be useful for decision-making by dryland farmers. If the rains break early, indicating good prospects for an average or above-average rainfall total, farmers are justified in committing themselves to such moisture-sensitive inputs as high-yielding varieties, close crop spacings and inorganic fertilizers; if the rains are late, the seasonal cropping plan should focus on minimising potential damage from drought by use of moisture-conserving tillage, wider crop spacings, early attention to weeding, and use of drought-tolerant or very short-term varieties. Stewart has shown that by plotting past annual or seasonal rainfall totals against the dates of onset of the rains, a "rainfall flag" can be drawn, characteristic of the location, which can provide the basis for such decisions on seasonal strategies for soil moisture management - an approach known as "response farming".

Flags for four locations are shown in Figures 2, 3, 4, and 5, taken from Stewart (1988), in which the method of constructing them is fully explained.

The first example plots rainfall data for 100 years versus the water requirements for growing a successful wheat crop at Davis, California. In this example, the distribution of dots in each vertical strip of the flag indicates the relative probability of crop success, partial success or failure, depending on whether winter rainfall begins in the earliest third of the 100-year set of onset dates, the middle third or the final third. The second example shows for Settat, Morocco, how such data can be used to judge the likelihood of future success or failure of wheat, for commerce and for subsistence, once the rainy season is considered to have started. The third example shows how the correlation between rainfall onset and amount is much less pronounced for the short rains at Kajiado, Kenya, than it is for the long rains at the same place. The fourth example indicates how plotting flags for different groups of years can highlight any shifts in rainfall patterns that have occurred, thus implying a need for shifts in dryland farming strategy. It is noteworthy that since flags for some locations show a much greater dispersion of annual rainfall figures than flags for others, the practical value of rainfall flags for prediction will vary from place to place.

If it is essential to make even more precise estimates of rainfall probability for given periods, other methods can be used, as described in e.g. Doorenbos and Pruitt, (1984 pp. 72-74), and Critchley and Siegert (1991 pp. 24-26).

The process of identifying measures which could improve the local capture and increase the efficient use of rainfall, and then deciding how best they might be promoted in the context of an investment project, will tend to mirror the overall process of project design (see FAO Investment Centre Guidelines for the Design of Agricultural Investment Projects, Part I, Chapter 3). Thus it is necessary to establish:

• that sub-optimum exploitation of rainfall is a genuine problem;
• the factors which contribute to the problem;
• the technical options for improvement - i.e. what might be done about the problem;
• what measures can in reality be taken in the context of a project, taking account, also, of non-technical constraints - for instance whether the time is yet ripe for widespread dissemination of new technologies or only for pilot tests; whether there are as yet any technologies ready to suggest to farmers;
• the appropriate role and actions which government institutions should take in promoting or facilitating execution of the chosen measures;
• any other residual externalities or issues which could influence the outcome.

Problem identification and thinking about possible approaches to improving the capture and efficient use of rainfall need to start early in the design process. Field work will be an essential ingredient. Whenever possible, diagnosis should be based on field-level consultation with land users and field staff of government support services, such as takes place during Investment Centre socio-economic and production systems surveys. If improved soil moisture management is thought likely to be a central project objective, then dialogue with farmers on their present problems and the probing of local views on possible solutions should be specifically included in such diagnostic surveys. The deeper knowledge of farming systems and the rural economy which a diagnostic survey brings will also be a principal means for deciding, at a later stage, what candidate technologies for improved soil moisture management might be ready for pilot testing or wider recommendation to farmers, or priorities for future research and on-farm adaptation. In general, changes which involve adjustments to existing practices or complement them should be favoured. "Add-on" technologies bearing no relation to what farmers do at present should be treated with caution.

Direct evidence that soil moisture is likely to be insufficient may come from rainfall records and their analysis. Potential problems in different parts of the country can be deduced from:

• published or unpublished records of the local meteorological service, research stations, universities, airports etc.;
• maps of annual rainfall and potential evapotranspiration;
• rainfall flags or dekad analysis (see earlier);

Another major source of direct evidence will be the knowledge of farmers, extensionists, researchers or geographers of local rainfall patterns, geographic variation, trends or local anomalies.

Soil degradation, increasingly violent flood regimes or high levels of sedimentation may indirectly indicate decreasing capacity of the land to retain incoming rainfall. Thematic maps - which would include information on soil types, topography, erosion risk and severity and other related soil constraints - may be available from agricultural research stations, geography departments of universities or hydropower corporations. Quantitative data can be checked against the recollections of older farmers or experienced extensionists of changes in these phenomena over the years. Results from research trials to measure rainfall runoff or soil loss should be probed to see to what extent they represent real-life farming situations on comparable slopes or soils. Such data may help pinpoint the soil types, farming systems or combinations of the two on which problems are most severe, or risks are greatest. It may also be possible to make some indirect estimates of the proportion of incoming rainfall which might be retained in future under improved management practices, to the benefit of rainfed crop production.

More direct estimates of possible benefits from improved moisture management may come from research (if any has been done) on improved management technologies, local experience of water harvesting (if practised in the potential project area) or from comparing rainfed yields and their variation with those under different types of irrigation.

Also important at this stage is to gain a general understanding of agro-ecological conditions in the project area, production potentials and the cropping systems most appropriate to exploit them. The design team should study the results of any land evaluation exercises (see FAO Soils Bulletins 52, 53 and 55) or comparable surveys giving data on soils, climate, hydrology and vegetation. Agricultural research organizations should be asked for available data on crop production at different levels of inputs and soil moisture regimes, both present and improved. The possible improvements in farmers' current cropping practices (particularly with regard to choice of planting dates and input levels) which might result if they had better short-term forecasts of rainfall, should be assessed.

Where important information is missing, such as rainfall probability analyses, soil surveys, crop production functions or key socio-economic details, arrangements need to be made for it to be supplied. In some cases this may involve writing-up and mapping work that has been undertaken but not yet analysed or presented. In other cases new investigations may be necessary before project appraisal. If they require major effort or will be very time-consuming, it may be necessary to carry them over into project implementation as a component to be financed.

Having ascertained that sub-optimal use of local rainfall is a problem, the next step is to reach a clear understanding of the cause or causes. Some indications may already be to hand, from diagnostic surveys or research results. However deeper field work, focused more exclusively on soil moisture problems, may also be required. In addition to seeking, in particular, as much information as possible on farmers' perceptions of the problem and their suggestions for solutions, the design team may need to:

• take a closer look at landscapes to see what proportion of moisture loss is attributable to geological conditions or other natural causes, as opposed to resulting from inappropriate land management or use;
• dig profile pits to investigate soil structure and condition, profile depth, rooting depth, topsoil loss etc., and correlate these findings with laboratory analyses;
• discuss in more detail with land users why they have not adopted any official recommendations which may have been made for soil and moisture conservation.

Useful initial work may be done on these topics at the same time as or immediately after a diagnostic survey and may not necessarily require a separate mission nor extensive extra manpower. In other cases deeper studies, and hence additional resources, may be called for. However, at a quite early stage it may be possible to identify some relatively homogenous areas where similar broad approaches to improving soil moisture management could be envisaged - some potential "recommendation domains" for technological change.

In general, it will be possible to categorise the immediate causes of the soil moisture problems which are found into those related to water entry and retention, and those related to its subsequent use by plants. Causes of soil related problems are listed in Box 3. Inadequacy of rainfall has already been discussed. Infiltration problems may be due to inherent soil characteristics (high sodium content, cracking clays etc.), to the way the soil has been managed previously, (formation of plough pans, compaction by trampling) to the direct or indirect consequences of a decline in organic matter content, (structural collapse or reduced activity of the soil fauna or flora).

• the use of plant species or varieties which are ill-adapted to the expected rainfall total or pattern of distribution (i.e. those with the wrong overall length of growth cycle or with an inappropriate pattern of evaporative demand through their growth cycle), or which are unduly susceptible to physiological stress;
• management practices, such as untimely exposure of moist soil by tillage, unchecked weed growth or inappropriate planting dates, which reduce the share of available moisture which can be transpired by the crop;
• the presence of physical or chemical hindrances to full exploration of the storage profile by crop roots.
  

The overall aims of any technical changes intended to optimise the in situ capture and use by crop plants of incident rainfall must be to minimise the soil- and plant-related problems summarised above.

To minimise the soil-related problems it is necessary to:

• detain water on the surface long enough for infiltration;
• prevent break-up or sealing of the surface under the impact of raindrops, so that runoff, splash erosion, rilling or gullying are avoided;
• minimise direct evaporative losses from the surface;
• increase capacities for surface infiltration and downward percolation of detained water;
• maximise internal storage.

To avoid plant-related problems it is necessary to:

• remove any physical or chemical impediments to root expansion through the storage profile;
• ensure adequate soil aeration throughout the soil profile in which water is stored, to allow its complete exploration by crop roots;
• minimise evapotranspiration by the crop where feasible, and eliminate losses via weeds;
• ensure optimal matching of plants to the site conditions, which could involve:

  • changing the variety (or even the species) of crop;
  • optimising planting dates.
    

By reference to Box 1, it can be seen that most of these actions amount, in their effects, to improving the capacities shown across the head of the matrix table which is given there. A checklist of possible interventions which could contribute to each of the necessities listed above is given in Box 4 for soil-related problems and in Box 5 for plant-related problems. Once the nature and causes of problems have been identified, Boxes 4 or 5 can therefore be used for a simple preliminary screening of possible approaches to solving the soil- or plant-related problems which have been identified.

In reality, however, the situation is more complex than is represented in Boxes 4 and 5. Many of the interventions listed will have more than just the one effect attributed to them there. For instance subsoiling will not only increase water infiltration and deeper percolation: if it breaks a localised plough pan it may also increase the effective depth of the soil profile for moisture storage, improve soil aeration for root growth and open new channels for root penetration. Likewise, the improved soil surface protection from raindrops which results from a denser crop cover will also contribute to detaining water longer on the surface and facilitating infiltration, and the residues of a denser crop may also supply more organic matter which, via improvements in soil structure, may eventually allow more water to be retained per unit of soil depth.

The multiple effects of some of the interventions on capacities are shown in matrix form in Box 6. Attached notes give some added explanations of the intervention and/or its effects. It will be noted that while the capacities at the top of the matrix are the same as those listed in Box 1, the factors of Box 1 have been replaced by interventions.

Referring back to Box 1, it can be seen that the interventions (left hand side of Box 6) which can be used to overcome soil-related or plant-related problems have their practical effects by influencing the factors shown down the right hand side of the Box 1 matrix. There is thus a three-way interaction between interventions and factors, and then between factors and capacities. Building on this, a more sophisticated handling of the diagnostic and screening process is possible, if refinement of the simple checklist approach is thought necessary.

Under this more sophisticated approach, a matrix table such as that in Box 1 can first be used to define or diagnose the target factors requiring intervention for the efficiency of rainfall capture and use to be improved. Box 7 gives an example of such a diagnostic matrix. It refers to badly-degraded land under low rainfall conditions in the Illela district of Niger. Even though the average rainfall is low, there is much runoff from the soil surface which has become crusted and impermeable from long-continued exposure to raindrop impact and to surface compaction associated with overgrazing in the past. The crop yields were poor even in years of above-average rainfall. When the rains were below average the crops often failed to mature and no yield was harvested. The traditional method of growing crops in such land was to break the crust by digging small pits (tassa), into which seeds of millet or sorghum were planted. These pits - at the spacing of the crop to be planted - caught and accumulated some of the runoff, but did not occupy the entire field surface. The excavated soil was scattered around in no particular pattern. The "Field Notes, Comments" down the left of the matrix in Box 7 link what is observed in the field to the specific factors (down the opposite margin) which are diagnosed as contributing to the present problems of the farmers in the area. By referring to the top of the matrix it is then possible to pinpoint the capacities which are deficient in the system, which are therefore the necessary targets for interventions to improve moisture conservation and use.

Thus, in the example, the soil surface suffers from high impact effects of rainfall energy, which reinforce the crusting by militating against the buildup of any structure at the soil surface. The soil's capacity to accept rainwater is jeopardised, as is evident from the vertical columns in the matrix: rain water is not detained on the surface to allow it to penetrate optimally (except where caught in the small planting pits), surface resistance to rain impact is weak, and rates of infiltration through the surface are low. In the example, the circles where "Impact/erosive energy" and "Intensity/rate x duration" intersect with these capacities for rainfall acceptance are therefore blacked in, identifying mitigation or modification of these capacities as necessary targets for intervention.

Further factors identified from field study of these degraded fields as contributing to the problem of excessive runoff in this particular example are the virtual absence of structural aggregates in the upper layers of soil due to compaction and pulverization through rainfall and trampling, and a very restricted range of pore sizes in the soil through which air and water might penetrate the soil profile. These poor structural conditions affect adversely not only the "Acceptance of rainwater" as already noted, but also the capacities "Storage of plant-available water" ("effective depth of soil profile" and "retention per unit depth") and "Adequacy for rooting" (especially "aeration" as it affects root functioning). There is an additional potential "Physical limitation" to root extension if the small planting pits are not sufficiently deep to traverse this compacted layer. Since under traditional management practices no cover of plant debris is left on the soil surface, and no manure is added to the planting pits, there is little or no build-up of organic matter in the tillage layer which might otherwise stabilise aggregates and hence reduce the tendency towards compaction by raindrops. Soil testing has shown lack of phosphate in these soils to be a factor limiting the growth of both roots and shoots, an effect which is increased when soil moisture is limited; this is noted under "Adequacy for rooting" as affecting both plant nutrition and root extension into deeper layers.

Although in the example the factor/capacity intersection points have simply been blacked in, it is equally possible to use symbols to show, for instance, relative importance or priorities for intervention.

Having formalised the definition of the soil moisture problems of a given setting using a diagnostic matrix of the type shown in Box 7, the next step is to black in a comparable matrix to characterise the interventions which the screening process has suggested may have some potential for overcoming these problems. Matrix sheets can also be completed to show the relevance of the farmers' own techniques for soil moisture management to the problems which have been diagnosed. Clearly, prominence should be given in the design of any investment project to developing those interventions which most influence the capacities diagnosed in the matrix as the origins of the problems in the setting concerned. In other words, the more closely the pattern of blacked-in (or otherwise marked) circles for a candidate intervention matches the pattern for the problem as diagnosed in Box 7, the more useful that intervention is likely to be in improving the capture and use of rainfall.

Boxes 8 and 9 show blacked in matrices for two candidate interventions intended to overcome the problems of the farmers of Illela district as characterised in the matrix of Box 7. Under the prevailing conditions of weather and their farming system, it is not practicable for the farmers to loosen the entire soil surface so as to favour rainwater infiltration in every part, nor to provide it with a complete cover of mulch, whether of crop residues or of manure. However, a first alternative as shown in Box 8 - a combination of enlarging the planting pits and adding enriched manure to each pit - can give significant positive results in terms of higher crop yields and better use of rainwater. Increasing the capacity of the planting pits, and placing the excavated soil in small half-moons around the downhill edge of each pit, increases the effectiveness of the system in catching and detaining rainfall and runoff water. Adding manure to each pit, rather than spreading it all over the soil surface, has a number of simultaneous effects: (1) the soil surface around the plants within each pit is protected against rainfall impact, minimising crusting and favouring infiltration; (2) the aggregation of the soil within the area of each pit is favoured by the activity of associated micro-organisms, in terms of both the size and stability of structural units; the more varied pore-size distribution which results leads to improved permeation by water and air; (3) the nutrient content of the soil is raised, and its cation-exchange capacity is increased, through the addition of the organic materials; (4) termites (as meso-organisms) are found to tunnel upwards to collect and take down organic matter to lower levels of the profile, thereby creating large continuous channels for passage of air, water and roots, as well as distributing organic matter more deeply through the profile; this also contributes to deeper growth of roots through the compacted upper soil layers. Incorporating phosphate with the manure results in less immobilization of the phosphate ion than might otherwise occur, the formation of organically-bound compounds with slow-release properties, and thus alleviation of this constraint to root growth.

Superimposing the marked matrix of Box 8 (intervention) on the marked matrix of Box 7 (problem) shows a close fit with the requirements for improved rainfall capture and conservation in this situation. Clearly this intervention is a good candidate for inclusion in a project technical strategy; indeed, it has already been enthusiastically adopted by the farmers in the area because of the impressive results in terms of higher and more reliable crop yields.

Box 9 characterises an intervention often advocated by technicians in the past for soil and water conservation - the construction of impermeable earthen banks. Superimposing Box 9 over Box 7 shows clearly the poor fit between the factors identified in the diagnosis in Box 7 and the characteristics of this intervention, indicating that it would do much less than the previous intervention to improve the situation. Water would only be detained close along the uphill side of each bund, with little if any effect on the planted areas between the bunds; if erosion of topsoil by runoff were severe, then the banks would limit removal of topsoil by runoff from inter-bund areas. This intervention would do little to benefit the crops directly by improving soil moisture conditions.

The type of matrix analysis described above is best carried out as part of mission fieldwork. A blank matrix sheet is given at the end of this guideline which can be photocopied for field use. The fit between diagnoses and candidate interventions can be readily seen by holding the superimposed, completed matrix sheets - one for the diagnosis of the problem and one for each of the possible interventions - against a bright light. In practice, it is uncommon for a single intervention to provide such a comprehensive solution as is the case in the worked example described above. Several simultaneous changes may be needed to cover an acceptable proportion of the marked circles on the diagnostic sheet.

Once appropriate techniques for getting water into the soil have been identified, either by simple use of checklists or the more comprehensive matrix approach, a second series of possibilities, in this case based on manipulation of agronomic factors, needs to be screened to identify the best means to ensure that the moisture which is captured is used by crop plants in the most efficient way. A number of practices are relevant in most or all climatic situations of dryland farming, while others will be suitable for particular conditions. A checklist is given in Box 10.

In reality, interventions to mitigate or modify the factors which influence the soil's capacity to accept and store rainfall (the examples in Boxes 8 and 9) will very frequently need to be combined with changed agronomic practices such as those in Box 10, if their full benefits are to be obtained. Often it will be an iterative process to identify the combinations most likely to be relevant. It should be remembered that the effects of one type of intervention may take longer to be felt than that of another. For example, tillage can have an immediate effect of the rate of infiltration, but changing the amount of organic matter in the tillage layer may take considerable time. It may also be necessary therefore to define the sequence in which a set of favourable changes should best be introduced. At this point socio-economic data from diagnostic surveys or other sources need to be integrated into the process. Potential costs and returns, the resources or inputs normally available to farmers, their attitudes to change and their willingness to accept risks will all have to be taken into account. In finalising the list of measures to be supported by investment, project designers also need to take into account the degree of variation in socio-economic status which exists within the farming population. Most investment projects seek to bring benefits to the deprived majority of rural people; hence a design which promotes changes that can only be adopted by the resource-rich segment of the farming population will tend to be of limited relevance.

For these reasons, in cases where screening identifies technologies which can already be recommended to farmers, or which are ready for pilot testing in farming communities, project design can only be considered as complete when the farm-level costs and benefits of the identified changes have been sufficiently well estimated to allow farm modelling and economic and sensitivity analysis during later stages of project preparation. Their compatibility with the socio-economic circumstances of the target group of intended adopters must also have been adequately demonstrated.

Whether or not farmers adopt a given practice, or package of practices, to improve soil moisture conservation will be influenced by many factors. They may be prevented from doing so by lack of implements or draught power, or not have physical access to the purchased inputs or planting material required. Alternatively, they may lack the cash or credit needed to finance changes, or have more pressing immediate uses (as livestock feed or fuel) for the organic residues which are likely to be crucial to improving the receptivity of the soil to incoming rainfall. The less familiar they are with the potential changes which have been identified, or the more these differ from existing practices, the harder it may be for farmers to accept what they may see as the risks involved. The measures which the designers propose for the improvement of soil moisture conservation under the project should reflect such considerations.

In assessing adoption prospects for interventions which have been identified as relevant to the project area, it is important for the design team to search out successful examples of the uptake or longer-term use of moisture conserving technology by farmers at nearby locations with similar physical conditions. If possible these areas should be visited. Every opportunity should be taken on such visits to identify the particular attitudes, resources or local circumstances which have had a determining influence on choices of technology by farmers and its scale of adoption, and from this to assess the prospects for replication in the project area. Questions to farmers and field-level staff of support services, if possible included in rapid rural appraisals of socio-economic conditions and farming systems of the intended project area, can also help illuminate local attitudes to improved moisture conservation.

If, despite such efforts, adoption prospects remain highly uncertain it will not be possible to predicate the project design on the assumption of routine recommendation and widespread uptake of novel conservation technology; nor, perhaps, may it be safe to assume even pilot tests of recommendations by farming communities. Instead, provision may have to be made, during project implementation, for groups of farmers from the project area to visit places where technology for improved moisture conservation is already being used, and to discuss it with the current users. Arrangements can then be included for a preliminary phase of on-farm testing of possibly relevant technologies by project farmers themselves. A group approach to dialogue with other communities, local extensionists and researchers is likely to provide more stimulus than individual contacts by farmers. On the basis of such contacts groups can be encouraged to propose variants or novel combinations of their own for on-farm testing, so that indigenous as well as outside knowledge will be drawn upon. The project design should then provide the necessary support for a programme of farmer-participatory research and development.

If, on the other hand, the design team can find neither readily-adoptable recommendations nor ideas for on-farm testing, it may be possible to take only a first step towards improved soil moisture conservation, by including a conventional research component in the project.

In deciding on the extent of initial change in moisture conservation practices likely to be accepted, it will often be useful to rank the potential acceptability of options. Interventions which farmers will find easiest to take up are likely to:

• show significant benefits quickly, such as a contoured ridge and furrow system or a new variety which combines the desirable characteristics of the old one with extra drought resistance;
• improve efficiency or stability of production, such as a mulch which improves infiltration and organic matter accumulation as well as keeping pumpkins dirt-free for market; or shorter season varieties which allow the introduction of a second crop based on residual moisture;
• have multiple effects, such as mulching with retained crop residues (see above), or multiple intercropping which combines enhanced soil protection with more root exploration of the soil, more residues to contribute organic matter, and greater output per cultivated hectare;
• be capable of disaggregation by farmers, so that they can assess them and incur extra costs and risks in steps, rather than having to confront them all at once (as in a combined package requiring simultaneous adoption of ridging, mulching, inorganic fertilizer and a new variety).

A prior phase of testing or research, or at least more conservative assumptions on the phasing of adoption, will be needed if uptake is expected to depend heavily on:

• progressive accumulation of sufficient capital or acquisition of costly equipment before farmers can embark on proposed changes;
• the recruitment and training of a new corps of extensionists to promote the new technology;
• a preceding phase of visits to learn from the example of other areas or communities.

To introduce some of the crucial changes having less immediately obvious effects - for instance mulching to initiate progressive improvements in soil structure - it may be worthwhile to link them to another, more visible, initial intervention such as water harvesting. In this way farmers will see that something is happening and gain confidence in their ability to improve their land by their own efforts.

Having decided on the mix of project measures to support the more efficient capture and use of rainfall which is most appropriate to the physical and socio-economic conditions in the project area and to the state of existing knowledge, the next step is to decide what the government could or should do to facilitate the implementation of the chosen approach. Together, the decisions taken at this stage will specify the scope, scale, costs and organization of the project component to support soil moisture conservation.

Much of what needs to be done is common ground with the design of other types of project component. It is covered comprehensively in the FAO Investment Centre publication Guidelines for the Design of Agricultural Investment Projects (Technical Paper Series No. 7). There are, however, a number of questions particularly relevant to the introduction of new techniques for soil and moisture conservation. Some are listed below:

• If it is important for farmers to have better weather forecasts, in what ways should the meteorological services be strengthened? What are the implications?
• If a particular cropping strategy is recommended - e.g mixed cropping in place of sole cropping - are the present varieties the most appropriate, and if not, are sufficient supplies of more-suitable varieties likely to be available? If new varieties are needed, are the particular characteristics required being searched for in existing plant breeding programmes?
• If shelter-belts of trees are a preferred intervention, who would be responsible for their planting and management, and who would have rights to their use for forage, firewood, poles etc.? What species are preferred by the local farming community? Would adequate supplies of planting material be available? Who would produce them? How would planting material of other woody species grown under individual control - fodder, browse or fruit trees - be produced?
• If new types of equipment or materials are needed for specific intervention - e.g ox-drawn ridgers or lime - where would they come from, how, and at what cost? Do special services - eg. for land shaping - need to be created? Who should operate them? If private sector entrepreneurs, would there be adequate incentives to go into business?
• Might farmers require special one-off incentives or subsidies to allow them to initiate new practices for soil moisture management? How would these be financed? How would continued dependence on subsidies be prevented? If subsidies for chemical fertilizers were removed, by what means would crops' ability to make use of better moisture conditions be maintained?
• If runoff-control works are needed, who would be responsible for design, construction, payment, recurrent maintenance? What should be government's role in this? Should government share initial or subsequent costs? If so, how could this be supported by the government budget?
• What would be the optimum means of providing land husbandry advice to farmers? Is an existing soil conservation agency the most, or least, appropriate group for the purpose?
• What means exist or should be created to promote group or participatory action? How should these be strengthened?
• For promising techniques which have not yet been adequately tested, what arrangements are needed for on-farm technology tests by farmers, or for conventional research and development? Who would be responsible?
• In the interests of extensionists developing credibility in the eyes of a particular farming community, are there any higher-ranking priorities which (in the view of the community) would need to be dealt with first (e.g roads, schools, drinking water etc.), before community attention is likely to be given to advice on soil moisture conservation? If so, what are the implications for other aspects of the project such as cost, phasing and overall design, and collaboration with other government agencies not so far involved?
• Where are the most appropriate places for in-service training and re-orientation of farm-level advisers and higher-level specialists towards the approaches here discussed, by means of study-tours or short courses? Would courses be better developed and run within rather than outside the country, so as to take full advantage of familiar conditions as examples; or would seeing the same problems in unfamiliar contexts be better for stimulating thought and discussion?
• What should be the role of outside technical assistance in creating local awareness of opportunities for improved soil moisture management, and promoting its uptake?
  

Finally, in designing a component for the more efficient capture and use of rainfall, it is necessary to review the main issues, external factors beyond project control or risks which are likely to influence successful implementation. Again, much of what needs to be covered is common ground with the design of other sorts of technical components: for instance questions of overall commitment by government decision-makers, by managers and staff of support services, or by farmers themselves, to the technical improvements being proposed; or the prospects for the creation of a credible corps of farm-level advisers; or for farmers to form voluntary groups to test or apply new technology. More specific points may arise, nonetheless, concerning policy, legislation and the overall attitudes of research and extension services.

National policy can have important positive or negative effects on land and resource use in general. The project design team may conclude, for instance from financial models designed to evaluate the costs and benefits of preferred interventions, that some form of incentive may be needed to secure initial adoption by farmers of the technology which is proposed. This may be justified on the grounds that it would only be a temporary measure. Furthermore it may be economically rational, in that while the incentive may raise financial (private) returns to the land user, they may still not exceed what is economically justified for the public good - i.e. external benefits. The need for the government to agree to such an innovation needs to be made clear. The design team should also note whether the introduction of such a change, however well justified in the project context, conflicts with macro-economic priorities such as an overall reduction of the government budget or raises questions as to where the fiscal resources would come from in practice. Risks of abuse of special incentives must also be realistically assessed. Commodity price effects may furthermore need to be flagged; for instance unduly high prices (whether government-controlled or in free markets) may stimulate a rush to increase production which may lead to the spread of poorly-managed cultivation or grazing onto land of inadequate quality, thus worsening runoff, erosion and loss of productivity. Contradicting policies or programmes of different government departments may have similar negative effects. In all such cases the design team should indicate precisely the adjustments or changes which would be needed for the efficiency of soil moisture management to be improved under the technical and social strategy which is proposed.

Legislation may raise a further set of issues. Punitive or coercive legislation which requires farmers to develop or manage their land in ways which give no immediate benefit may create negative attitudes to any form of conservation. This will be compounded if legislation casts the so-called advisory services in a policing role. New laws or regulations may be needed to permit communities, through local by-laws, to take the sort of joint responsibility for local resources on which a comprehensive soil moisture management strategy may depend. The same may be needed if government services are to adopt a more facilitating role or if new institutions - for instance a land husbandry service formed from elements of formerly separate extension and soil conservation divisions - are to be created.

Attitudes and priorities may also need to change in agricultural research. New means may have to be found to reflect the importance of land management for soil moisture conservation in what might currently be a research programme dominated by yield or productivity maximisation for separate crops. Individual researchers may have to be encouraged to think in terms of optimising whole-farm returns from limited reserves of cash or farmer labour, or stabilising rather than maximising output. The researcher's contribution to raising farmers' incomes may need more emphasis among promotion criteria than his or her contribution to scientific journals.

Analogous changes in attitude may be needed among extensionists. A combination of technologies aimed to optimize the capture and use of rainfall will be very different in concept and content from the sort of individual technology packages for pure-stand crops which field staff may currently be used to promoting. Moisture conservation achieved indirectly through biomass - i.e. via increased crop cover and measures to build up soil organic matter - will also be radically different from what may have been a past emphasis on physical structures. Finally, the facilitating, participative emphasis of extension suggested here to promote soil moisture conservation may clash with top-down habits of thought. Training and technical assistance may be insufficient guarantee of change without, again, some restructuring of institutions and redefinition of their terms-of-reference.

• CRITCHLEY W, SIEGERT K. 1991. Water harvesting : A manual for the design and construction of water harvesting schemes for plant production. Paper AGL/MISC/17/91. Rome: FAO.
• DAY J C, AILLERY P. 1989. Economic effects of soil- and water-management technologies: preliminary results from a case-study analysis in Mali. Proc. Internatl. Workshop: Soil, Crop and Water Management systems for Rainfed Agriculture. ICRISAT Sahelian Center, Niamey, Jan 1987. Patancheru, India: ICRISAT. pp.363-371.
• DOORENBOS J, KASSAM A H. 1979. Yield response to water. Irrigation and Drainage Paper 33. Rome: FAO. ISBN 92-5-100744-6.
• DOORENBOS J, PRUITT W O. 1984. Crop water requirements. Irrigation and Drainage Paper 24 (first pubd. 1977). Rome: FAO.ISBN 92-5-100279-7.
• FAO, 1978a. Report on the Agro-Ecological Zones Project: vol. 1: Methodology and results for Africa. World Soil Resources Report 48. Rome: FAO. ISBN 92-5-101179-9.
• FAO, 1978b. Organic materials and soil productivity. Soils Bull. 35. Rome: FAO.
• FAO, 1981. Arid zone hydrology. Irrigation and Drainage Paper 37. (quoted in Hudson 1987, q.v.) Rome: FAO.
• FAO, 1984. Improved production systems as an alternative to shifting cultivation. Soils Bull.53. Rome: FAO.
• FAO, 1985a. Guidelines: land evaluation for irrigated agriculture. Soils Bull. 55. Rome: FAO. ISBN 92-5-102243-7.
• FAO, 1985b Guidelines: land evaluation for rainfed agriculture. Soils Bull. 52 Rome: FAO.
• FAO, 1986. Early agrometeorological crop yield assessment. Plant Production and Protection Paper 73. Rome: FAO. ISBN 92-5-102419-7.
• FAO, 1991a. Manual and guidelines for CROPWAT. Rome: FAO/AGL (unpubd.).
• FAO, 1991b. World Soil Resources: an explanatory note on the FAO World Soil Resources Map at 1:25,000,000 scale. World Soil Resources Report 66. Rome: FAO.
• FAO, 1991c. Soil Conservation Notes No.25. Nov. 1991. Rome: FAO/AGLS (unpub'd).
• FAO, 1992. Guidelines on sociological analysis in agricultural investment project design. FAO Investment Centre Tech. Paper 9. Rome: FAO Investment Centre.
• FAO, 1993. Guidelines for the design of agricultural investment projects. FAO Investment Centre Tech. Paper 7. Rome: FAO Investment Centre.
• FRERE M, POPOV G F. 1979. Agrometeorological crop monitoring and forecasting. Plant Production and Protection Paper 17.
• HUDSON N W, 1987. Soil and water conservation in semi-arid areas. Soils Bull.57. Rome: FAO.
• HUDSON N W, 1992. Land Husbandry. London: Batsford. ISBN 0 7134 5976 X.
• REIJ C. 1990. Indigenous soil and water conservation in Africa: an assessment of current knowledge. Proc. Workshop 'Conservation in Africa: Indigenous Knowledge and Conservation Strategies; Harare, Zimbabwe, Dec. 1990. Amsterdam: Free University/Centre for Development Cooperation Services.
• REIJ C, MULDER P, BEGEMANN L. 1988. Water harvesting for plant production. Technical Paper 91. Washington: World Bank.
• ROCHETTE R M, (ed.) Le Sahel en lutte contre la desertification. Weikersheim, Germany: Verlag Josef Margraf/GTZ.
• SHAXSON T F, HUDSON N W, SANDERS D W, ROOSE E, MOLDENHAUER W 1989. Land husbandry : a framework for soil and water conservation. Ankeny, USA: Soil and Water Cons. Soc. ISBN 0-935734-20-01.
• STEWART J I, 1988. Response farming in rainfed agriculture. Davis, USA: WHARF Found. Press. ISBN 0-9620274-0-5.

TABLE 1