4. MINIMIZING WATER STRESS AND IMPROVING WATER RESOURCES

Improving restricted rainfall infiltration

Infiltration depends on there being sufficient porosity in the surface soil for rainfall to infiltrate, and in the subsoil and parent material (if shallow) for rainwater to percolate. When the porosity of the surface soil is too low to accept rainfall, or subsoil porosity is too low to allow rainwater percolation (i.e. permeability is too slow), then infiltration will be restricted and rainwater will be lost as runoff.

The porosity of surface soil may have been reduced by clogging of pores with particles detached from soil aggregates under the impact of raindrops, or by the deposition of detached particles on the soil surface as impermeable crusts or seals. The porosity of subsurface soil may be naturally low, or may have been reduced by compaction and tillage practices that have disrupted or destroyed pore spaces causing a zone of low permeability at the base of the tilled layer. The degree to which soil porosity is reduced by tillage is frequently sufficient to limit root penetration, but is less often so severe that permeability to water is significantly diminished.

The overriding approach should be to instil in society, and in farmers, extensionists and researchers in particular, the will to create and sustain soil conditions that encourage the infiltration of rainfall where it falls, and to counteract the causes of runoff (Jonsson et al., 1999). This implies that the porosity of the soil must be at least maintained, or increased.

The approaches for overcoming restricted infiltration may be categorized according to the cause of the problem as shown in Table 13 at the end of this chapter. Where soil has already been damaged, a combination of two or more of these approaches may be necessary to initiate soil improvement at and beneath the surface.

Improving the infiltration capacity of the soil surface

Porosity of the soil surface is best maintained by first protecting it from the disruptive action of raindrops through a protective cover, usually of residues from the previous crop, a cover crop or mulch, and by ensuring the soil is not disturbed by tillage. This is best accomplished through what is called Conservation Agriculture, which is described in Chapter 5. The effects of conservation agriculture on higher infiltration and reduced runoff and flooding have been well documented in Brazil in particular (FAO, 2000e).

If the whole concept cannot be applied immediately, improvements in soil moisture status of the soil can still be achieved, though probably not to the same extent, by other measures aimed at prolonging the useful life of rainwater. These include the use of surface residue covers alone, fallow periods under cover crops or natural vegetation, protection or temporary closure of grazing lands and forests from overgrazing, and operations on the contour, complemented by physical measures to detain rainwater.

The regular use of shallow tillage with disc or tined implements to break-up surface crusts to increase surface porosity and enhance rainfall infiltration is not recommended. The increase in surface porosity is only temporary and on crusting-susceptible soils tillage will need to be repeated after every rainstorm. Tillage leads to the disruption of pore spaces in the soil, and the use of discs, in particular, often causes compaction, which may impede root growth and rainwater percolation. Tillage also accelerates the loss of soil organic matter leading to a progressive deterioration of soil architecture and a reduction in the number and stability of pores that allow growth of roots and movement of rainwater.

Regular tillage therefore is not recommended as a solution to restricted infiltration caused by low porosity of the soil surface.

Using surface residue covers to increase infiltration and reduce runoff

A residue cover absorbs most of the energy of the raindrops that fall on it and by the time this rainwater reaches the soil below, its ability to disintegrate soil aggregates and detach fine particles is greatly reduced. Consequently, there is little or no clogging of surface soil pores by detached particles, and little deposition of soil particles that would form a crust on the surface.

The benefits of a residue cover are most apparent on soils initially in reasonable physical condition, but even under these conditions runoff can sometimes occur despite a good soil cover. For example, runoff will occur when rainfall intensity is greater than the soil's infiltration rate, or when the soil's pore spaces are already filled with water because the soil is shallow, its water holding capacity is low, or its subsoil is only slowly permeable.

When a residue cover is applied to a soil with a very degraded surface of low porosity, the beneficial effect of the cover on infiltration may be initially limited. In such situations, it is advisable to accelerate the recuperation of surface porosity before applying residue covers by tilling the soil once to break-up the crust and any subsurface pans, followed by a fallow period under a cover crop to enhance the formation and stabilization of soil porosity. Annex 9 provides a list of publications about cover crops.

The choice of a cover material depends on what is locally available. Residue covers may consist of:

• Crop residues left in the field after harvesting the previous crop.

• Cover crops sown the previous season and left on the soil surface after slashing or applying a herbicide.

• Leaves and branches lopped from trees growing within the cropping area.

• Mulches of grasses, shrubs, weeds, litter, husks and other organic waste materials.

The last option (mulches) requires residues to be collected from elsewhere, transported to the cropping area and then applied in the field, whereas in the other options, the residues are produced within the cropping area.

Examples of materials that may be used as mulches are grasses and sedges, banana leaves and pseudostems (Plate 38), shrubs such as Lantana and wild sunflower (Tithonia), forest litter and tree loppings (Plate 39). Other materials occasionally used are weeds, rotten thatch and coffee husks. Where soils have a cover of stones, these may be left on the surface as a protective cover provided they do not interfere with planting or weeding operations. Mulching is most commonly practised on horticultural crops that produce negligible residues (foliage), or are completely harvested for their foliage, or are completely harvested (e.g. tuber + foliage).

PLATE 38. Mulching of bananas with their own leaves and pseudostems and with grasses in western Uganda

[R.G. Barber]

PLATE 39. Example of tree loppings used as a mulch in the Quesungual system (Honduras) to reduce the loss of rainwater through runoff and evaporation

[R.G. Barber]

In the steeply sloping Guaymango area of El Salvador, efforts were made in the 1960s and 1970s to improve crop production by encouraging small farmers to adopt a combination of hybrid seeds, nitrogenous and phosphatic fertilizers, increased plant densities and application of herbicides and insecticides to maize, sorghum, sesame, rice, beans. These recommendations were not particularly successful and in 1973 recommendations for soil conservation were added. These included no burning of crop residues; uniform distribution of residues across the field; use of living or dead barriers; and sowing on the contour in a zero tillage system.

Improvements in crop yield and quality of soil occurred and a high proportion of farmers adopted these measures. Although erosion control was cited as the farmers' main reason for not burning crop residues any more, an important pointer to the benefits due to improved soil moisture conditions was evident in 1997. In that year there was a serious drought during the rainy season associated with the El Niño weather phenomenon. But according to the farmers, they were able to harvest almost as much maize as in a more normal year because of conservation of moisture in the soil as a consequence of the better soil status, while neighbours who had not adopted the system lost their crops to drought. Nor did they lose their crops the following year during hurricane Mitch and the associated torrential rainfall, which caused disastrous flooding. The farmers noticed that the same mulch prevented the seeds from being washed away by rainstorms and facilitated rainwater infiltration so that they did not have problems of decaying plants during the heavy rains (FAO, 2000c). A cross-check on this beneficial effect under the same extreme weather conditions comes from Honduras, where hurricane Mitch caused much erosive devastation on many hillsides, but less on those hillsides where soils were well protected by crop residues (Hellin et al., 1999).

In a limited area of western Honduras the Quesungual traditional agroforestry system has been used by small farmers to produce maize, sorghum and beans. As rising population pressure makes the traditional slash-and-burn system increasingly unsustainable, there is increasing interest among farmers in the Quesungual system. It combines pruning of naturally regenerating indigenous trees and shrubs with normal agroforestry methods for growing high-value timber and fruit-trees. Before sowing, vegetation is cut down by hand without burning and is spread across the field together with the branches and leaves from pollarding. Crop seeds are then scattered or jab-planted with a stick through the mulch layer. Weeds are cleared infrequently, by hand or using a herbicide.

Mechanisms by which surface residue covers enhance rainwater infiltration

The physical contacts between a residue cover and the soil surface obstruct the movement of the runoff, slowing it down, giving more time for infiltration and so reducing the volume of runoff. Thus two aspects of surface cover can be distinguished:

• all surface cover absorbs the energy of raindrops and so prevents the loss of pore spaces into which rainwater can infiltrate;

• contact cover slows down any runoff, giving more time for infiltration.

The degree of contact cover is important especially on steep slopes, on soils with naturally low infiltration rates, and on degraded soils with surface crusts or seals of low porosity. Furthermore, it is the contact cover that is immediately accessible to soil macro-organisms and can stimulate their activity. Thus greater numbers of biopores are likely to be formed, leading to more rapid infiltration and percolation. This is why major disturbances such as tillage or incorporation of residues, mulches or other organic matter drastically reduces these positive effects.

Pliable materials of short length, such as leaf or grass mulch, which can be easily flattened by raindrops, will develop a high degree of contact cover and will substantially slow down the speed of runoff flow, generally resulting in reduced volumes of runoff. In contrast, inflexible long materials, such as woody branches of tall bushes that are not easily flattened by raindrops, will develop a low contact cover and so have less influence on the speed of runoff flow (Plate 40).

PLATE 40. Maize on a steep slope with a degraded soil surface covered by stiff long-strawed stems of a bush. Despite a 90 percent aerial cover there was high runoff because of the restricted contact between the vegetation and the soil surface, and the low surface porosity of the soil, Morazan, El Salvador

[R.G. Barber]

Advantages of surface residue covers

The advantages of mulches are the same as for crop residue, i.e. increased infiltration, decreased runoff (Lal, 1976), and greater soil water availability. They both provide additional benefits, notably less soil water losses by evaporation, less weed incidence and water losses by transpiration, softer and more workable soils, increased earthworm activity (Lal et al., 1980), the incorporation of additional nutrients (FAO, 1999b) and frequently increased yields.

In western Kenya, mulching with Tithonia has given substantial yield increases of maize, kale, tomatoes and French beans. Net profits from mulching kale ranged from US$91 to US$1 665 per ha (ICRAF, 1997). In the semiarid zone west and north-west of Mount Kenya, maize yields increased by a factor of 4.4 when 3 t/ha of mulch were applied (Liniger, 1990). Termites were not a problem in this area, probably because of the cool climate.

Constraints to using surface residue covers

The main disadvantage of applying mulch is the cost or labour of collecting, transporting and applying the mulch. This is not the case with crop residues, which are produced on-site. Often, there will be no suitable mulching materials in the vicinity of the farm, or there is insufficient labour available. Transporting large quantities of mulch for large-scale cropping is seldom economic and mulches cannot be applied after emergence to closely spaced crops.

When a cover crop is used as mulch, there is the cost of slashing the cover crop or applying a herbicide. Similarly, lopping trees and distributing the branches and leaves over the cropping area requires considerable labour. On steep slopes, the application of residue covers is not easy and requires much labour as well. Moreover, these materials are easily washed downhill on steep slopes.

Mulching materials and crop residues are often grazed by cattle belonging to the farmer, the community or the landowner (in the case of tenant farmers), fed to livestock, or sold as fodder. Sometimes these materials are in demand for thatching or fuel; in many semiarid areas they are rapidly consumed by termites, and in hot humid climates, they decompose rapidly. Another disadvantage of mulches is a progressive decrease in soil fertility where the mulching materials are produced, unless manures or fertilizers are applied. In parts of Uganda, the residues of cereals grown on hillsides are used to mulch bananas on the lower slopes or valley bottoms, which become enriched in nutrients at the expense of the cereal areas. Soil erosion may also degrade the source areas when the cover provided by the vegetation is removed for use as mulch.

The amount of residues needed

In relation to increasing infiltration, studies over two seasons in Nigeria on slopes of 1 to 15 percent have shown that 4 t/ha of rice straw mulch, equivalent to about 80 percent cover, will reduce runoff to 5 percent of the seasonal rainfall (Lal, 1976). A similar result has been found on a 12 percent slope with a well-structured freshly cultivated soil in Kenya, where 4 t/ha of grass mulch equivalent to 79 percent cover, reduced the runoff from simulated rainfall to 5 percent. On the basis of these data an 80 percent cover, equivalent to about 4 t/ha maize straw, would appear to be appropriate for increasing rainwater infiltration.

Conditions favouring the adoption of surface residue covers

The use of soil covers is more common in subhumid and humid zones because of the greater availability of vegetative materials. Nevertheless, they are particularly suited to semiarid areas when materials are available and in the absence of severe termite problems. Mulches are often applied to limited areas of high-value horticultural crops and home gardens in easily accessible fields with gentle slopes.

Fallowing under cover crops or natural vegetation

When soils are so badly degraded that they must be taken out of production, soil porosity can be restored through the action of biological processes. This can be achieved by fallowing for 1 or several years under natural vegetation, natural vegetation enriched with fast-growing leguminous trees, or planted fallows. The accumulation of large amounts of biomass on the soil surface from the fallow vegetation associated with high biological activity and strongly developed root systems promote the biological recuperation of soil porosity. Biological incorporation of residues into the surface soil results in higher soil organic matter in the upper few millimetres, which progressively extends into deeper layers overtime. The permanent cover of surface residues encourages soil faunal activity, which combined with higher soil organic matter contents leads to improved soil porosity (FAO, 1995c).

A well-adapted, deep-rooting leguminous cover crop often speeds up the recuperation of soil porosity compared with a natural vegetation fallow because larger amounts of biomass are rapidly produced by the cover crop. Whereas a natural vegetation fallow may require 3-5 years, a cover crop may recuperate soil porosity in 1 year. When degraded soils are severely compacted, deep tillage with a subsoiler immediately prior to sowing the cover crop encourages establishment and development of the cover crop. If the degraded soil is severely deficient in phosphorus the application of P fertilizer will be necessary to encourage the establishment of the cover crop.

A constraint of soil recuperation by natural vegetation fallows in mechanized production systems is the problem of eliminating trees and excavating roots before returning to cropping. If a manual system is to be adopted, the problem is less serious. Herbaceous and shrubby cover crops can be eliminated much more easily by slashing, mowing or application of a systemic herbicide, and the subsequent crop may be sown directly into the residues of the cover crop.

Temporary closure of grazing lands and subsequent protection

Low infiltration and high runoff can occur on grazing lands even on slopes less than 2 percent, as for example at Sebele, in Botswana. In this area, vegetation cover was considered to be the most important factor controlling infiltration and runoff, and catchments with a cover in excess of 70 percent generally had lower runoff compared with those with less than 70 percent cover (LWMP, 1992).

Although the percentage of grass cover in grazing lands has an important influence on rainfall infiltration, soil surface porosity can be more important, especially when overgrazing has degraded the soil, resulting in surface compaction and very low porosity (Plate 41). On degraded grazing lands at Iiuni, Kenya, for example, even with 57 percent vegetative cover the runoff was in excess of 60 percent (Moore et al., 1979). The presence of algae growths on bare surfaces that were resistant to wetting encouraged runoff, whereas stone covers reduced runoff due to the creation of water-storage areas between the stones where the rainwater is detained, allowing more time for infiltration (Barber and Thomas, 1981).

PLATE 41. The land in the foreground is a clay soil with an unstable surface and an argillic horizon; it was previously arable land, but after developing a plough pan at 12-15 cm depth, it was abandoned to grazing. A surface crust of very low porosity has developed which encourages runoff, strongly reducing the amount of moisture available for use by grasses. This combined with heavy grazing has resulted in a denuded land surface - Machakos, Kenya

[R.G. Barber]

Importance of forest protection for water infiltration

Forest provides an excellent protective cover made up of the canopy, low-storey bushes, herbs and surface litter, which combine to protect the soil surface from loss of porosity by direct impact of raindrops. The litter also serves as a food and energy source for soil organisms, which encourages the formation of soil organic matter and faunal passages leading to high infiltration rates.

Where forests are not protected from grazing and litter is consumed by livestock, removed for use as mulch as in parts of Nepal, or is lost in fires, the surface cover may be diminished to such an extent that the soil becomes bare. This is likely to be more serious under trees that discourage the growth of understorey herbs and shrubs, such as teak (Tectona grandis) and some species of Eucalyptus due to shade, high water use - especially by Eucalyptus - and to a lesser extent because of the acid nature of the litter. If the tree canopy is high, accumulated rainwater drops that fall off the leaves may be larger than normal raindrops and can fall with sufficient velocity to cause more damage to the soil than if there were no tree cover. This can lead to a pronounced loss of soil porosity, as can trampling by livestock, resulting in restricted infiltration and high runoff despite the high canopy cover (Plate 42).

PLATE 42. Example of compacted soil beneath a Tectona (teak) plantation, which resulted in high runoff and erosion - Jocoro, El Salvador

[R.G. Barber]

The protection of forests from overgrazing is an important management issue in overcoming restricted infiltration, and the establishment of forest user groups is often a crucial step in effectively controlling overgrazing and the loss of surface soil porosity. Forest user groups are most likely to be successful where indigenous forest management systems have existed (Kandel and Wagley, 1999).

Increasing the period for infiltration by detaining runoff with physical structures

Alternative, but less favourable solutions to restricted infiltration are the use of physical structures, which may be necessary under certain situations:

• When it is not immediately feasible to implement conservation agriculture or simple soil cover because, for example, crop residues are used as fodder.

• As backup measures to support conservation agriculture where the problem of restricted infiltration is due to rainfall intensities that are higher than soil infiltration rates even in the presence of a residue cover.

In these situations, the volume of water soaking into the soil may be increased by giving more time for infiltration by slowing down runoff, by means of physical or vegetative structures constructed across the slope and parallel to the contour.

Closely spaced structures on the contour (e.g. ridge and furrow series of planting lines and irregularities formed by contour tillage and crop management operations) may be formed over the whole field so that rainfall is detained where it falls. Widely spaced structures at intervals down the slope (e.g. fanya juu terraces, stone walls, earth bunds, live barriers and trash lines) used on their own without contour field operations between them will result in rainwater running downslope until it is detained or slowed down at the next barrier.

Details of the layout, design, construction and maintenance of these structures appear in many Soil and Water Conservation (SWC) handbooks, such as Soil conservation (Hudson, 1995), Soil and water conservation manual for Kenya (Thomas, 1997), FAO Soil Bulletin 70 (FAO, 1996a), A land husbandry manual (Shaxson et al., 1977) and other documents produced by governmental and other agencies for specific countries or particular environmental conditions.

Contour field operations

On sloping land all field operations such as tillage, planting, weed control, spraying and harvesting should be carried out along the contour. Ridges and mini-depressions along the contour create small storage volumes where rainwater can accumulate, allowing more time for infiltration (Plate 43). Field operations conducted in a downslope direction can cause a devastating impact resulting in high runoff losses and soil erosion (Plate 44).

PLATE 43. Contour cultivation creating small ridges and depressions parallel with the low marker-ridge at top right - Umuarama, Brazil

[T.F. Shaxson]

PLATE 44. Despite earth bunds constructed approximately parallel to the contour, planting tobacco in a downslope direction has led to serious gully formation from right to left, resulting in breakage of the bunds - Kasungu, Malawi

[T.F. Shaxson]

Narrowly spaced contour planting ridges with and without cross ties have the advantage of detaining rainwater where it falls so that there is more time for soak-in, and can be an effective means of encouraging infiltration and preventing runoff in semiarid and the drier subhumid areas. An additional advantage is that working along the contour makes operations such as harvesting easier and quicker.

Constraints of surface irregularities formed by contour field operations

The surface depressions have limited capacity to retain water and on sloping land the effective storage volume rapidly diminishes as slope increases. On slopes greater than 5 percent the effective storage volumes are considerably reduced. Reductions in storage volume will also occur on soils with a low structural stability, as the small ridges slump into depressions on becoming wet. Substantial runoff can occur even on land of 1-2 percent slope when the soils are of low stability and susceptible to crusting. Even on structurally stable soils, depressions may be quickly overtopped by the accumulation of rainwater from all but the lightest of rainstorms.

Conditions favouring the adoption of contour field operations

The only exceptions to contour cultivation may be in high-rainfall areas where the soils have high infiltration rates and high susceptibilities to mass movements, e.g. landslides and mudflows. In these situations high soil water content increases the risk of mass movements, and so it may be better to encourage controlled runoff of some of the rainfall. Since the effectiveness of contour field operations in reducing runoff is limited on all but the gentlest slopes, it should be considered as just one of the practices necessary to increase water availability.

Narrowly spaced contour planting ridges and tied ridges

In tied ridges the ties are constructed at intervals across the furrows formed by the contour ridges (Plate 45). These structures are usually constructed with animal traction or tractor power and may be formed annually or can be semi-permanent (Plate 46). They may also be made by hand but labour demands are high. The precise form and management of contour ridges and tied ridges vary considerably, with the optimum design and management being dependent on the crop, rainfall and soil type.

Contour ridges run the risk of being overtopped if too much rainwater accumulates within the furrows. They also may be breached or collapse at low points where large volumes of runoff accumulate from along the furrows (Plate 47). If large volumes of water frequently accumulate a subsurface pan or horizon of restricted permeability may be present beneath the furrows.

PLATE 45. Example of graded contour ridges with cross ties lower than the main ridges to retain water between the cross ties, but allow excess rainwater to flow between the ridges rather than spill over or break the main ridges

[T.F. Shaxson]

PLATE 46. Making cross ties - Makoka, Malawi.

[T.F. Shaxson]

PLATE 47. Example of the effects of excessive rainwater breaching contour ridges at low points resulting in loss of rainwater by runoff and severe soil erosion - Mua, Malawi

[T.F. Shaxson]

These risks can be reduced by carefully laying out and maintaining the ridges and furrows to ensure there are no low points and by constructing tied ridges to prevent lateral movement of water along the furrows towards any low points that may exist. The ties should be spaced at 1 to 3 metre intervals along the furrows and no more than half to two-thirds the height of the ridges. Although tied ridges require additional work, they provide good insurance against the collapse of ridges at low points during heavy rains and the loss of rainwater by discharge from the ends of the furrows if a slight gradient exists.

The furrows of contour ridges are normally aligned parallel to the contour. However, if very large volumes of runoff are periodically expected, tied ridges should be installed and the furrows constructed on a slight gradient (never steeper than 2 percent in the direction of a natural watercourse) so that excess rainwater is discharged along the furrows to prevent overtopping of the ridges. In these circumstances well-designed discharge points will be necessary at the furrow outlets. The size and spacing of the ridges should coincide with the crop's recommended spacing, furrow width and depth.

Ridges and tied ridges may be constructed prior to, or after, planting. Maize is often planted on the flat, and the ridges constructed at the time of the first weeding about 30 days after planting, which saves labour. Clearly, the earlier ridges are constructed the more rainwater they will be able to detain, and the greater the probability of a good yield. The time when ridges are constructed is also a convenient time to simultaneously incorporate manures.

Advantages of narrowly spaced contour ridges and tied ridges

The main advantage of contour ridges and tied ridges is the greater accumulation of rainwater within the furrows due to the retention of potential runoff (Njihia, 1975) (Figure 15). The concentration of water in the furrows encourages deeper percolation, but for this to be useful to the crop, the soil's AWC must be sufficiently high to retain the accumulated water within reach of the crop roots. Sandy soils with a low AWC may permit a large proportion of the rainwater to drain beyond the zone penetrated by the roots.

FIGURE 15. Fate of rainwater for three soil management practices (Morse, 1996, adapted from Moyo and Hagmann, 1994)

Constraints of narrowly spaced contour ridges and tied ridges

The continual formation of ridges each year by hand or by mechanization, combined with trampling along the furrows, may result in the formation of a compacted horizon at the base of the ridges, which can prevent roots from penetrating into deeper layers. This will counter the advantages provided by the ridges of increasing the supply of available water. The exposure of the soil surface leads to an accelerated loss of soil organic matter and surface crusting due to the effects of tillage, raindrop action and direct exposure to the sun, and very little macrofaunal activity. Consequently, the soils rapidly become degraded.

Another constraint is the time required to construct contour ridges, with even more time needed for tied ridges. The manual construction of contour ridges needs about 100 hours per hectare (Morse, 1996), and heavy textured soils will be even more demanding. To form ridges by hand or by animal traction in hardsetting soils will generally only be possible once the first rains have moistened the soil. The process of manually constructing contour ridges on sloping land, where the farmer faces uphill and pulls the soil into ridges with a hoe, causes soil to move downhill, so encouraging soil erosion (Plate 48).

Conditions favouring the adoption of narrowly spaced contour ridges and tied ridges

Contour ridges and tied ridges are most suited to areas suffering from water deficits where it is not feasible to provide a soil cover to enhance infiltration and reduce runoff through the use of crop residues, mulching materials or cover crops (Plate 49).

The manual implementation of these structures will only be possible where sufficient labour is available and where farmers consider that the high labour requirement is justified by the value of the crop. These structures are particularly suitable for the production of tuber crops.

Impermeable and permeable contour barriers at discrete intervals downslope

These structures include stone lines, walls, earth banks, fanya juu terraces, trash lines, live barriers and similar constructions. They have usually been installed to prevent small rills developing into gullies by limiting the area over which runoff collects, with or without sideways diversion into prepared waterways for safe disposal downslope. The barriers, which they provide may, if well maintained, accumulate soil which has been eroded from upslope.

PLATE 48. Farmer constructing ridges for potato cultivation in southwest Uganda. The action of pulling the soil downhill to form ridges is contributing to soil movement and erosion

[R.G. Barber]

PLATE 49. Mulch, cross tied planting ridges and an earth bund (as backup) provide multiple means of catching rainwater for the benefit of young tea. Weeping lovegrass has been planted along the bund to provide future mulch. Mulanje, Malawi

[T.F. Shaxson]

PLATE 50. Earth bund stabilized with Phalaris sp. and combined with zero tillage - Chapecó, Brazil

[R.G. Barber]

PLATE 51. A broad-based earth bund set on the true contour can cause local waterlogging if the collected runoff cannot soak in or flow away. Kasungu, Malawi

[T.R.Jackson]

In many situations, the chief benefit of laying out structures along the contour, at discrete intervals downslope, is their use as guidelines for the alignment of contour field operations in the cropping areas between them (Plate 50). The capture and soak-in of runoff along the upper sides of these structures may be considered as an added, rather than a primary, benefit.

The more closely spaced the banks, the more frequently runoff will be intercepted, but the more of the farmer's land will be taken out of production, unless some useful crop is planted along the earth bank. In semiarid areas structures can be designed for the purpose of water harvesting, which provides the extra water needed for adequate yields, if only from a relatively narrow strip.

If the strip of land immediately upslope of the barrier has been made impermeable by passage of machinery or of feet, or the soil itself is relatively impermeable, temporary or more long-lasting local waterlogging can be induced (Plate 51).

If more runoff accumulates than the structure can hold back on its upslope side, it will overtop and may break, with the resulting concentrated flow of accumulated water often causing more damage downslope than if the structure had not been there at all.

Conditions favourable for adoption of impermeable cross-slope barriers for water conservation

In areas of moderate to high rainfall, such barriers may be appropriate where they complement water-absorptive conditions, good surface cover and/or ridge and furrows, with or without tied ridges. If they are laid out on the level contour they may have some small additional effect on increasing water in the soil (Figure 16).

FIGURE 16. Fanya Juu terrace at construction and after several years (Thomas, 1997)

Permeable cross-slope barriers

Permeable barriers, which may be accumulations of stalks, branches, crop residues, leaves (trash lines) without or with a line of one or more crops, forage grasses, shrubs or trees (live barriers) may impede but not stop runoff. The lower speed as runoff passes tortuously through the material provides an opportunity for infiltration. The live barrier may benefit from the additional soil moisture, but the additional transpiration through deep-rooted plants may minimize the volume, which could flow beyond the roots to groundwater. If the farmer receives no benefit from the live barriers, then the competition effects for light and moisture would be a disincentive (Plate 52).

Bench-type terraces

These structures are a total modification of natural land slopes into a series of platforms which are almost level or slope at shallow gradient across or along the terrace. Controlling the gradient in this way allows management of water movement on what were formerly steep slopes. The cultivation platform may be continuous along the slope (bench terrace, Plate 53) or discontinuous (orchard or platform terrace). The surface of the intervening uncontrolled slopes is preferably covered with a close-growing grass or legume as a soil-protecting cover.

PLATE 52. Example of a live grass barrier of a hybrid between Pennisetum sp. and Phalaris sp. associated with an earth bund. Chapecó, Brazil

[R.G. Barber]

PLATE 53. Bench terracing for horticultural corps - Costa Rica

[R.G. Barber]

Bench-type terraces are arduous and expensive to construct, requiring up to 700 person-days per hectare. Their capacity to receive and store rainwater depends on the depth, condition and quality of the soil into which they have been constructed. In semiarid areas they may be able to catch and detain all the rain that falls. In places with greater volume and frequency of rainfall, provision may have to be made for disposal of excess water down very steep waterways, and there is also an added danger of landslips if the benches become saturated.

Deep tillage to increase subsoil porosity and permeability

Rainwater infiltration may be restricted in soils where the pore spaces rapidly become saturated with water because of the presence of dense subsoil horizons of low permeability. In these situations an initial deep tillage of the whole field with a tined implement, subsoiler or paraplow to break-up the dense horizon may improve subsoil permeability and so allow more rainwater to infiltrate. By improving subsoil permeability the rate of oxygen supply to the crop roots will also improve. However, the beneficial effects of deep tillage may only last 2 to 3 years.

Reducing water losses from evaporation and excessive transpiration

The most effective solution to high evaporation losses of soil water is a cover of plant residues on the soil surface. Agronomic practices that increase shading of the soil surface, and physical structures that concentrate rainwater, encouraging percolation to deeper layers, also reduce evaporation losses. Wasteful transpiration losses may be the result of weeds or excessive crop transpiration in hot windy conditions, and can be reduced by appropriate weed control practices and windbreaks, respectively.

Minimizing evaporation from the soil surface

Surface residues reduce soil water losses through evaporation by acting as an insulating layer. This diminishes the temperature of the surface soil and eliminates the effect of wind. Heat from the sun is only slowly transmitted from the surface of the residues through the air trapped within the layer of residues to the soil surface. Consequently the soil surface remains cooler and the rate of evaporation of soil water is slowed down. The thicker the layer of trapped air, the greater will be the insulating effect, and the quantity of residues required to reduce evaporation losses is considerably greater than the quantity needed to ensure that most rainfall infiltrates where it falls.

For example, in Uganda, farmers traditionally apply between 8 and 40 t/ha of mulch to bananas (Briggs et al., 1998), whereas 4-5 t/ha are probably sufficient to minimize runoff and allow most of the rainfall to infiltrate. Banana yields respond very favourably to mulching, and applying 5-10 cm depth of maize stover and Paspalum sp. to bananas at Sendusu, Uganda, increased yields from 4.3 to 10.8 t/ha (Speijer et al., 1998). Experiments in Uganda have shown that yields approximately doubled when 30-40 t/ha of mulch were applied compared with applications of 10-20 t/ha (Briggs et al., 1998). This yield increase was mainly attributed to lowered evaporation losses. Protecting the soil surface from wind also slows down evaporation by reducing the rate at which water vapour is removed from the soil surface.

The use of residue covers for conserving soil moisture in the topsoil and increasing yields is particularly important in regions with limited rainfall and high evaporation rates. It is also important for shallow-rooted crops, e.g. bananas, tea, coffee, pineapple, and vegetables such as onion, lettuce, cabbage and carrots. Residue cover can also be very beneficial in reducing water losses by evaporation from soils with a shallow water table (less than 1 to 2 metres), from which there may be capillary rise of the subsurface water. Such soils are often used in horticultural production.

However, the main disadvantage of using residue covers for reducing direct evaporation is the large quantities of residues required to significantly reduce evaporation. Often, the regions with high evaporation losses also suffer from a shortage of rainfall, which restricts the production of vegetative matter. Frequently there are also other demands on residues, which take priority such as fodder, thatching and construction.

Reducing excessive transpiration

In hot windy weather, the rate of loss of water through plants by transpiration can be very high and can result in early depletion of limited soil moisture reserves. This in turn can lead to serious water stresses developing in plants - both crops and weeds - before their cycle of growth to maturity has been completed.

Weed control

Loss of soil water through weed transpiration can seriously reduce the amount of water available to crops. Consequently, timely and effective weed control practices are essential. The presence of a thick layer of residues on the surface is a very effective way of controlling weeds. Where weed control measures are needed, the use of herbicides or appropriate crop rotations is often preferable from a conservationist perspective to mechanical weed control, unless it is practised with no soil disturbance. Post-emergence herbicides leave weed residues on the soil surface as a protective cover whereas cultivation leaves soil exposed to the impact of raindrops and sun, accelerates drying of the surface soil and tends to disrupt and destroy soil porosity through smearing and compaction.

Windbreaks

When crops are exposed to strong winds in a dry environment the water that has been transpired by the crop is rapidly removed from the leaf surfaces into the atmosphere. This encourages a more rapid movement of water up through the crop and much greater absorption of water from the soil. Strong winds can therefore cause excessive crop transpiration rates and an unnecessary loss of soil water.

Windbreaks will significantly reduce wind speed and so reduce crop transpiration rates and the unnecessary loss of soil water. Windbreaks are usually established by planting single, double or triple rows of trees, but sugar cane or tall grass species may also be used. In areas where forests are being cleared for agricultural development, strips of the original forest may be left as natural windbreaks.

Important considerations in the design of planted windbreaks are their composition, orientation, height, porosity and spacing (McCall et al., 1977; Barber and Johnson 1993). Windbreaks should be oriented at right angles to the direction of the prevailing winds during the growing season. As a general rule, they should occupy no more than 5 percent of the cropped area. For small production units a single row of trees is usually most appropriate. Paths and roads should not cross windbreaks to avoid channelling of the wind through the openings at high velocities. The tree species selected should be adapted to the climate and soils of the area (Shigeura and McCall 1979; Johnson and Tarima 1995). The foliage should not be so dense that most of the wind is forced to pass over the top of the windbreak, as this will cause severe turbulence on the downwind side of the windbreak, which can seriously damage the crop. The porosity of the windbreak vegetation should ideally be 40 percent so that part of the wind passes through the windbreak. This will give a 50 percent reduction in the velocity of the wind within a distance of ten times the height of the trees (Skidmore and Hagen, 1977). When there is sparse protection in the lower part of the windbreak, as shown in Plate 54, it is advisable to allow regeneration of shrubs within the windbreak or plant tall grasses (e.g. Pennisetum purpureum) or sugar cane to ensure a more uniform protection from top to bottom. Maintenance of the windbreaks is important to ensure that no holes appear, to regulate the porosity of the vegetation to wind and to avoid excessive shading and weed infestation of adjacent crops.

Plate 54. A single row windbreak of Leucaena leucocephala with little foliage at 0-2 m height, providing inadequate protection to the crop - Santa Cruz, Bolivia

[R.G. Barber]

Natural windbreaks are strips of forest left after deforestation. Since a much drier and windier microclimate develops in these strips of forest compared with that in the undisturbed forest, many trees in natural windbreaks often die, sometimes leaving holes through which the wind passes at increased velocity. The important guideline for natural windbreaks, as for planted windbreaks, is that the porosity of the vegetation should be about 40 percent. In open forests in particular, natural windbreaks may need to be substantially wider than planted windbreaks to allow for the death of some trees. Alternatively, planting individual trees to fill gaps, or enriching the natural windbreak with one or two rows of additional trees may be necessary to produce a protective cover of 40 percent porosity.

Well-designed windbreaks will significantly reduce evapotranspiration rates of crops in windy conditions resulting in the conservation of soil water and less subsequent moisture stress when water is limiting. A 50 percent reduction in wind velocity (from 32 to 16 km/h) will reduce evapotranspiration rates by 33 percent (McCall and Gitlin, 1973). Windbreaks may provide additional benefits to crops by reducing mechanical damage and the loss of flowers, and by creating better conditions for insect pollination. They are also beneficial in reducing wind erosion, especially in fine-sandy and silty soils, and in diminishing air pollution problems. Depending on the tree species selected, windbreaks may also provide fruit, nuts, fodder and timber, but the harvesting of these products must not result in pronounced gaps being formed within the windbreak.

The main disadvantage for farmers with small plots is the loss of cropping area due to the windbreak and the risks of competition between the windbreak and the crop for water, nutrients and light leading to lower crop yields. This zone of competition may extend over a distance equal to 1.5 times the height of the windbreak.

In areas where there are severe shortages of fodder, fuelwood and timber, windbreaks may need to be fenced to prevent indiscriminate grazing and harvesting. To ensure that wind cannot pass around the ends of individual windbreaks, the establishment of windbreaks should be planned on a community basis.

Conditions favouring the adoption of windbreaks

Windbreaks will be favoured in areas subject to strong dry winds during the growing season, and where windbreaks cause a net gain in soil water (i.e. where the gain in soil water due to reduced crop transpiration exceeds the loss of water due to windbreak transpiration). Windbreaks are also likely to be favoured where they consist of species that provide additional benefits, such as fodder, fruit, nuts, fuelwood and timber that can be harvested without damaging the windbreak.

Shade

Shade can be provided by all manner of materials, whether artificial such as nets, cloths, plastic sheets and others, or plant-derived, such as cut branches, cut grass supported on nets, or living trees which provide high-level and wide-spreading shade. Shade is necessary in plant nurseries in hot regions to protect seedlings and other plants with shallow roots from rapid desiccation. While shade may ameliorate the severity of hot dry conditions and limit undesirable losses of soil moisture, it can also be so dense as to limit solar energy reaching leaf surfaces and limit photosynthesis and growth rates.

Where shade may be desirable, its density should be adjusted to provide an appropriate balance between losing water too fast, limiting sunlight intensity and avoiding scorching of leaves due to temporary dehydration and cell-damaging high temperatures. Using living shrubs and trees to provide long-term shade for tea and coffee can cause difficulties in maintaining the desired degree of shade above the crop over the long term.

Reducing rainwater drainage beyond the rooting zone

Soils without restricted rooting

In regions where much of the rainfall occurs as light showers, the concentration of rainwater as near as possible to the crop will cause more of the rainwater to infiltrate deeply, where it is less susceptible to evaporation. In order not to lose this water by drainage beyond the crop's rooting zone and where there is no rooting restriction some solutions can be adapted, such as increasing the capacity of soils to retain water within the rooting zone, early planting to accelerate root development or changing to deeper-rooting crops.

Increasing available water capacity of soil

The addition of large quantities of organic manure will increase the available water capacity (AWC) of soils and in theory this is a useful practice for reducing deep drainage losses. However, even in temperate climates the quantities of organic materials required to markedly increase AWC are very high, applications must be continued over many years and usually affect only the plough-layer depth (Russell, 1988). In tropical zones, where organic matter decomposition rates are much higher, the influence of organic manures on AWC is likely to be even less. Nevertheless, this practice may be feasible for small-scale farmers growing high-value crops where large quantities of organic manures and labour are readily available.

Dry planting

In low rainfall areas, it is frequently difficult to know when the rains have truly started, as initial rains are often followed by a dry period. Many farmers wait until the topsoil has been moistened to a depth of about 15-20 cm before planting, so that even if there is a subsequent short dry period there is sufficient water within the soil. However, this results in a delay in planting and for every day's delay yields will decrease (by about 5-6 percent for maize in eastern Kenya, Dowker, 1964), largely due to the loss of rainwater by drainage and evaporation, together with the loss of some released nutrients.

To overcome this problem and to allow crops to develop deeper rooting systems earlier on so that more of the rainfall can be utilized during the initial stages of the season, some farmers "dry plant" when soils are dry prior to the onset of the rains. To avoid premature germination before sufficient rain has fallen, the seeds are usually placed deeper than normal. Dry planting also has the advantage of spreading labour over a longer period. Crops may also benefit from this practice by being able to utilize the nitrogen released at the start of the rains from the decomposition of soil organic matter, which reduces leaching and pollution of groundwater. However, there are a number of problems associated with dry planting, notably that some soils, and in particular hardsetting soils, are difficult if not impossible to till when dry. If seeds are not planted sufficiently deeply, they may germinate at the first rains and then die during a subsequent dry period.

Improving plant nutrition for early root development

Applying fertilizer to speed up crop canopy development and increase the shading of the soil surface will decrease the soil water lost by evaporation so that more is available to the crop. Planting crops equidistantly (i.e. with between-row spacing similar to within-row spacing) so that the soil surface becomes shaded more quickly would also be expected to reduce the proportion of soil water lost by evaporation. However, the effects of these agronomic practices on reducing evaporation losses will be much less than applying surface residues.

Introducing deep-rooting crops

On permeable sandy soils that retain small quantities of available water for crop use, it is preferable to introduce deep-rooting crops that can utilize soil water at depth that would not be available to shallow-rooting crops. Examples of deep-rooting crops are almond, barley, cassava, citrus, cotton, grape, groundnut, olive, pearl millet, pigeon pea, safflower, sisal, sorghum, sunflower, sweet potato and wheat.

Improving soils with restricted rooting

The type of solution to be applied will depend on the cause of root restriction. The most frequent cause is physical root restriction due to a lack of pores that are large enough to be readily penetrated by roots or which can be sufficiently widened by the growing roots. This condition occurs in dense layers, such as plough pans formed by tillage, but also in naturally occurring dense layers as found in hardsetting soils. Root restriction may be overcome, at least temporarily, by biological or mechanical means. In addition to eradicating the causes of root restriction it is also important to take steps to avoid future recurrence of the problem by, for example, introducing conservation agriculture where dense layers have been formed by tillage.

Less common causes of restricted rooting are chemical restrictions due to the presence of toxic concentrations of aluminium or manganese, high salinity or severe nutrient deficiencies, especially of phosphorus. A lack of oxygen due to a fluctuating water table may also restrict root development. While the water table is high, root development for most crops will be restricted to the soil immediately above the upper level of the water table but the crop will not suffer from a lack of moisture. If the water table then falls relatively quickly to a substantially lower level, for example at flowering, when the crop has still to reach physiological maturity but the roots have ceased growing, the roots may be left stranded in the dry soil without access to the moisture in deeper layers.

The causes of restricted rooting given above can, where appropriate, be overcome by the application of lime, or lime and the more mobile gypsum, to eradicate aluminium and manganese toxicities; leaching to reduce salinity hazards; fertilizers to rectify nutrient deficiencies; or drainage to remedy the lack of oxygen from a fluctuating water table.

The principal biological method of restoring the porosity of root-restricting layers is to place the land in fallow and utilize the roots of natural vegetation or planted cover crops to act as biological subsoilers penetrating the dense root-restricting horizons (Elkins, 1985). The stability of root channels created by plant roots will be greater than that of channels formed by mechanical methods because of the release of organic substances from the roots that stabilize the channel surfaces. Once the roots have died and shrunk, these pores will be sufficiently large and stable to enable the roots of subsequent crops to penetrate.

Land may be left in fallow for 2-3 years for natural bush or forest vegetation to regenerate. Alternatively, planting selected species that are effective in regenerating soil structure can enrich the natural fallow. A cover crop may be sown to serve as a planted fallow. Promising cover crop species that have been shown to have potential as biological subsoilers are the grasses Bahia grass (Paspalum notatum), Festuca elatior (Elkins et al., 1977), Guinea grass (Panicum maximum) (Lugo-Lopez, 1960), and alfalfa (Medicago sativa) (Meek et al., 1992), pigeon pea (Cajanus cajan) and cowpea (Vigna unguiculata) (Maurya and Lal, 1979). Radish (Raphanus sativus)^[2] and the nitrogen-fixing shrubs Tephrosia vogelii (Plate 55), Sesbania sesban and Gliricidia sepium have also been identified as potentially useful (Baxter, 1995; Douglas et al., 1999). Some weeds with pronounced tap-roots, such as Amaranthus sp., may possibly also have potential to act as biological subsoilers, as Mennonite farmers in eastern Bolivia have observed much higher crop yields on compacted soils after high infestations with Amaranthus.

PLATE 55. Tephrosia vogelii for regenerating soil fertility through its effect as a biological subsoiler in breaking up the hardpan, and of producing high biomass and fixing nitrogen to increase soil organic matter and nitrogen contents - Zomba, Malawi

[T.F. Shaxson]

Biological methods are generally much cheaper to implement and their benefits are longer-lasting than mechanical methods. An important advantage of vegetative fallows is that they greatly improve the physical, chemical and biological fertility of the soil due to the large quantities of organic matter produced and added to the soil. Tree fallows can be beneficial in supplying fuelwood, construction materials and other products, provided the harvesting of these materials does not reduce the beneficial effects of the fallow on soil chemical fertility.

The main disadvantage is the 2 to 3 years required for natural fallows when the land is taken out of production while the recuperation takes place. A disadvantage of tree fallows is the difficulty of returning to annual cropping after the fallow period because of the problem of extracting the tree roots and the longer the fallow period the more difficult the problem. However, the extraction of the roots of Sesbania after 2 years of fallow has not been a problem in Zambia. It is also necessary to protect the vegetation from grazing, burning and harvesting during the 2-3 year fallow period, which may involve additional costs for fencing.

Planted fallows of cover crops with tap-roots may be difficult because of the lack of available seeds and their cost, since a high plant population is necessary to ensure an adequate density of tap-root penetration of the root-restricting layer. For very dense root-restricting layers, even Cajanus cajan may have only a limited effect^[3].

Conditions favouring the adoption of biological methods

The use of natural biological methods will be favoured by farmers who have sufficient land. They can take some of it out of production and place it into fallow while the slow process of natural regeneration of soil porosity takes place. The use of cover crop fallows is often a rapid process which enables land to be more quickly returned to production. Natural fallows in which there is a regeneration of tree vegetation are more likely to be adopted by farmers who wish to change the land use of the recuperated area to forest or perennial tree crops.

Mechanical solutions to physical root restriction

The aim of mechanical methods is to break-up the compacted or naturally dense root-restricting layer in order to create larger pores through which crop roots can penetrate. This is accomplished by the implement slightly lifting and breaking the compacted or dense layer. The operation may be carried out over the whole of the field, or merely along the rows where the crop is to be planted. The latter, known as in-row subsoiling, is much quicker and requires less draught power, but the crop must be sown with precision directly over the loosened rows. The most appropriate method will depend on the depth to the root-restricting layer, its thickness and hardness, and the source of power available.

Mechanical disruption of shallow root-restricting layers

Shallow root-restricting layers such as hoe pans are typically produced at 5 to 8 cm depth, and the easiest means of breaking them up are with ox-drawn rippers or tractor-mounted chisel ploughs. Most farmers relying on manual tillage will probably have to use hand tools to break the hoe pans by methods such as double digging, which are very labour-intensive (Box 5). To break-up compacted layers in the dry season when the soil is very hard may require robust tools different from those the farmer normally uses for tillage, such as pickaxe, mattock, three-tined hoe (jembe) or a long crowbar.

In central and western Kenya, small resource-poor farmers intensified their production, both in yield and diversity, by using double-dug (to 50 cm depth) composted beds on small areas, generally near to their houses. Positive results were achieved from the concentration of organic materials onto the beds, which received focused attention, plus improved rainwater capture (Plate 56). Improved conditions in the root zone, including excellent moisture-holding capacity, enabled a range of vegetables (and field crops) to be grown well into the dry season, and these were less affected by drought than those grown in unimproved plots

PLATE 56. Double-dug composted beds with a crop of Capsicum - Kerugoya, Kenya

[Association of Better Land Husbandry]

While the total area of land managed in this way is often only a small proportion of the total cropped land, overall output from the beds rose sharply due to higher yields and diversification of crops. The system provides many benefits which were recorded during a survey of farm families' comments (Box 6).

Mechanical disruption of moderately deep root-restricting layers

Deeper root-restricting layers such as plough pans are formed at the lower limit to which the soil is tilled, and usually occur within the upper 10 to 25 cm of the soil profile. Plough pans formed by ox-drawn implements can usually be broken up using two passes of an ox-drawn ripper, whereas those formed by tractor-drawn or -mounted implements usually require a tractor-mounted subsoiler or paraplow (Plates 57 and 58).

PLATE 57. Subsoiler used to break-up naturally occurring dense horizons or compacted layers caused by tillage

[R.G. Barber]

PLATE 58. The use of tractors with steel-rimmed wheels and metal fins to subsoil dense root-restricting layers is likely to be counterproductive because of the compacting effect of the metal wheels and fins - Santa Cruz, Bolivia

[R.G. Barber]

Paraplows are similar to subsoilers except that the shanks are slanted sideways to the direction of travel, which enables soil to flow over the shanks (Figure 17). They are preferable to subsoilers as they bring fewer subsoil clods to the surface, require less draught power and cause less incorporation of surface residues that should ideally be left on the surface. Disc ploughs are less suitable because they invert the soil, incorporate most of the crop and weed residues and bring subsoil clods to the surface, resulting in the need for additional tillage.

FIGURE 17. Example of a paraplow (R.G. Barber)

If the root-restricting layer is to be disintegrated over the whole field, then as a rule of thumb the subsoiler or paraplow should penetrate to 1.5 times the depth to the lower limit of the root-restricting layer, and the spacing of the shanks should not be greater than this value. For example, if the root-restricting layer occurs at 10-24 cm depth, the shanks of the subsoiler or paraplow should penetrate to 36 cm and the spacing between the shanks should be no more than 36 cm. If the shanks are more widely spaced, there is a likelihood that the root-restricting layer will not be fully disrupted in the region midway between where the shanks passed. To avoid compaction from the wheels of the tractor, shanks should be positioned immediately behind the tractor's wheels. For in-row subsoiling, the shanks need only penetrate to the lower limit of the root-restricting layer, and the shank spacing should coincide with the planned row spacing of the crop (Figure 18).

FIGURE 18. a) Shows the depth of shank penetration for in-row subsoiling in relation to the root-restricting horizon, b) shows a cross-sectional view of the effect on crop root development (FAO, 2000d)

Subsoiling should be carried out perpendicular to the normal direction of tillage and the soil should be dry to the depth of subsoiling to obtain good shattering. If the soil is moist or wet, there will be no shattering, merely the formation of channels gouged out where the subsoiler's points have passed. Further information on subsoiling procedures is given in FAO Land and Water Bulletin No 8 (FAO, 2000d).

Mechanical disruption of very deep root-restricting layers in the subsoil

Subsoil compaction at 40 cm depth and greater is caused by the passage of very heavy equipment with high axle loads of at least 6 tonnes, such as combine harvesters and lorries laden with grain. At this depth the use of conventional subsoilers to loosen deep compacted layers is difficult and expensive because of the very high traction power needed. Vibratory and rocking subsoilers, in which the subsoiler points vibrate or rock using the tractor's power takeoff can work to 80 cm depth, but require 75-100 HP. New implements have been developed employing elliptically moving blades or rotary hoes, which utilize a break-off-loosening mechanism to disintegrate compacted layers. They can be used to depths of 60 to 120 cm and at higher soil moisture contents than conventional subsoilers, but are very expensive and require high traction power^[4].

The shattering and lifting of root-restricting layers by mechanical means creates larger pore spaces through which roots can penetrate, enabling them to reach and take advantage of soil moisture and nutrients stored in deeper layers. Consequently, crops are able to make more efficient use of the rainfall. The main effect of subsoiling is usually that of promoting deeper root growth, but if the root-restricting layers are so dense that rainwater movement is also limited, subsoiling may also facilitate the percolation of rainwater into deeper layers.

The development of improved rooting frequently increases crop and pasture yields (Plate 59). In Babati District, Tanzania, breaking up hardpans by subsoiling has almost tripled maize yields and quadrupled maize dry matter production (Jonsson et al., 1999). Increased yields from subsoiling are most likely in areas where yields are limited by rainfall, and the drier the season the greater the probable response to subsoiling (Box 7).

PLATE 59. Contrasting performance of Brachiaria brizantha in a compacted soil - Las Brechas, Santa Cruz, Bolivia

[R.G. Barber]

PLATE 59. Contrasting performance of Brachiaria brizantha after subsoiling - Las Brechas, Santa Cruz, Bolivia

[R.G. Barber]

In-row subsoiling, especially when it is combined with planting in a single operation, is particularly beneficial for hardsetting soils that rapidly form root-restricting layers on drying after being saturated with rain. This technique is most likely to be successful when associated with precision planting and controlled traffic, in which the passage of all machinery wheels is restricted to permanent tracks. The benefits of subsoiling are likely to be greatest when immediately followed by the establishment of a dense cover crop with a strong rooting system that helps stabilize the new pore spaces created. The cover crop should then be followed by a system of conservation agriculture in which the absence of tillage reduces the recurrence of further compaction.

The principal disadvantage of mechanically breaking up root-restricting soil layers is the high power requirement, whether it is manual, animal or mechanical. Since most farmers do not have access to more than that which they use for land preparation, the process is inevitably slow.

Some soils become so extremely hard during the dry season, that the farmer's normal draught power is incapable of penetrating the soil in order to break-up the root-restricting layer. It is then necessary to wait for the beginning of the rains to moisten and soften the soil before it becomes possible to break-up the compacted layer, but this may coincide with the critical time of land preparation and planting. This problem can apply equally to farmers using animal traction or tractors and to those using hand tools as their source of power. Subsoiling operations are ineffective when the dense or compacted layers are wet or very moist as no shattering effect takes place.

Farmers often lack the necessary implements, whether they are pickaxes for farmers relying on manual power, rippers for animal traction farmers, or subsoilers or paraplows for mechanized farmers. The use of normal land preparation implements will generally be less satisfactory. For example, disc ploughs can be used to break-up plough pans, but they invert the soil bringing large clods of subsoil to the surface and form an uneven surface that needs additional tillage to create a seed bed. Disc ploughs also incorporate the residues of crops and weeds, when ideally they should be left as a protective layer on the soil surface. Repeated use of disc equipment, especially heavy-duty disc harrows, can produce an almost impermeable compacted pan in only a few seasons. These pans have been the cause of increasingly severe runoff and erosion from millions of hectares in Brazil, before the use of disc equipment was abandoned in favour of minimum tillage with tines, and subsequently by no-till systems. When bulky crop residues are left on the surface, especially the stiff residues of maize, sorghum and cotton, the performance of subsoilers and paraplows is considerably impaired unless they are fitted with front cutting discs.

If subsoiling is followed by conventional tillage, the beneficial effects are only likely to persist for 2 or possibly 3 years and so the subsoiling has to be regularly repeated. The speed with which the root-restricting layer reform will depend on the number of tillage and other field operations, the moisture content of the soil at the time of these operations and the susceptibility of the soil to compaction. Fine-sandy and silty soils and those with impeded drainage are most susceptible to compaction.

To improve the physical conditions of hardsetting soils requires the incorporation of large quantities of organic material into the dense layers and the regeneration process is likely to be slow. For hardsetting soils, in-row subsoiling may be necessary each year. These disadvantages can be overcome by adopting reduced tillage, or preferably zero tillage as in conservation agriculture, or by controlled traffic in which all machinery follows the same tracks year after year, leaving the cropped strips untouched. Thorough loosening of soils by subsoiling may render them more susceptible to compaction if they are subsequently subjected to high pressures, as from excessive tillage or the passage of very heavy machinery. The recompaction may be worse than the original state of compaction.

Subsoiling heavy textured soils, such as vertisols, can greatly increase the quantity of rainwater that reaches the subsoil, resulting in a marked reduction in the soil's bearing strength, i.e. its capacity to support heavy machinery. It should be noted that subsoiling to any given depth produces a high proportion of very large soil pores and fissures, a situation favouring better penetration of roots and of rainwater. It will not however produce any significant increase in the range of smaller soil pores, which make up the water-retention capacity of the soil.

Conditions favouring the adoption of mechanical methods

The adoption of mechanical methods to overcome physical root restriction will be favoured where yields are frequently limited by low rainfall. Under such conditions it becomes important that as much of the rainfall as possible is stored within the soil profile, and that the crop's roots have access to all of the stored soil moisture. Mechanical methods will be favoured where farmers have access to tractors and subsoilers or paraplows, and where land cannot be taken out of production and put down to fallow for 2 to 3 years.

Chemical solutions to restricted root growth

Root development is sometimes restricted by unfavourable soil chemical conditions, such as severe nutrient deficiencies, aluminium or manganese toxicity and salinity. The nutrient which most commonly restricts root development is phosphorus and the application of P fertilizers to phosphorus-deficient soils frequently encourages deeper rooting, enabling the crop to access more soil moisture and so increase productivity. The incorporation of lime without or with gypsum will reduce toxic concentrations of aluminium and/or manganese to non-toxic levels and so encourage deeper rooting. The greater solubility of gypsum compared with lime makes the former more suited to soils with aluminium or manganese toxicity problems in the subsoil, whereas the slowly soluble lime is most effective in topsoils. When high salt concentrations inhibit root development in irrigated soils, excess quantities of water should be applied sufficient to leach the salts out of the crop's rooting zone.

Maximizing usefulness of low and erratic rainfall

Several approaches may be used to diminish the impact of low and erratic rainfall, viz. match land use to soil characteristics; use drought-resisting and drought-escaping crops; increase the efficiency with which crops utilize rainwater; concentrate rainfall by water harvesting; divert river water; intercept floodwater; and apply supplementary irrigation.

Match land use to soil characteristics

Matching land use to the most suitable soil types within a farm may increase the efficiency with which the available soil water in the different soil types is utilized for crop production. Crop water requirements vary, as do the capacities of soils to retain and supply water to crops. Moreover, the variations in available water capacities (AWC) of soils often occur over short distances. Soils with high AWC will be expected to suffer less water loss from deep drainage and possibly from runoff. Consequently, greater quantities of rainwater will remain in the soil and so the potential crop-growing season will be longer assuming an adequate amount, distribution and infiltration of rainfall (Table 9).

TABLE 9
Length of growing period for different soil available water capacities in bimodal rainfall areas of semiarid India (Virmani, 1980)

The longer the expected duration of dry periods and the more sensitive the crop to drought, the more important it will be to use soils of high AWC. For soils to be considered suitable for maize in semiarid areas of Arusha, Tanzania, they must be of sufficient depth and AWC for the maize to be able to tolerate dry periods of up to four weeks (Jonsson et al., 1999). Farmers can take advantage of variations in AWC by locating moisture-sensitive crops and crops with longer growing periods on soils of high AWC and crops tolerant to drought and early-maturing crops on soils of low AWC. This approach is applicable at farm level (Plate 60) and also at field level, especially for farmers with very smallholdings where differences in soil AWC between small areas within a field can still permit diversification. Some localized areas may occur within a field where runoff accumulates and provided the soil's AWC is adequate to retain the moisture, the soil will be suitable for more water-demanding crops.

PLATE 60. Example of the matching of land use to land suitability based on differences in soil available water capacity and other land characteristics. From foreground to background: citrus, terraced vegetables, natural forest, grain crops, citrus and Eucalyptus woodlot. Chapecó, Brazil

[R.G. Barber]

Seasonally waterlogged low-lying, grassy areas, known as dambos, are commonly found at the head of watercourses in southern and central Africa. Their high soil water content makes them highly suitable for crop production, even in semiarid areas, because they are relatively unaffected by mid-season droughts. Even in dry years, yields up to 2.5 t/ha of maize can be obtained (Morse, 1996). Traditionally, dambos were used for rice, maize and vegetable production, dry season grazing and sources of domestic water. In Zimbabwe the cultivation of dambos was banned because of concern about environmental degradation, but recent research has shown that with environmental safeguards, present levels of yield could be increased threefold (Bell et al., 1987).

Matching crops with weak root systems, such as beans, to soils lacking root-impeding layers, would be expected to increase crop water use efficiency. Beans are more suited to freshly tilled soils, or to mature no-tilled soils where large numbers of channels suitable for root penetration have been created through the decomposition of old roots and soil faunal activities (FAO, 2000e).

Allocating land use to suitable soil types may enable production to be intensified, leading to benefits in addition to that of higher water use efficiency. Thus, intensifying subsistence food crop production may liberate land for producing cash crops. Alternatively, it may allow land previously used for inappropriate, extensive and degrading forms of land use, to revert to natural vegetation, thereby reducing land degradation.

Use of drought-resistant and drought-escaping crops and varieties

Some crops can tolerate drought because they are able to resist a shortage of water, i.e. they are said to be drought-resistant. This is because either:

• they can stop growing when water is unavailable by becoming dormant. When rain occurs they resume growing and developing as though nothing had happened; or

• they have deep rooting systems, such as pigeon pea, and can absorb water from deep within the soil. This is important where occasional rainstorms wet the soil to great depth followed by long dry periods.

Pineapples and sisal resist the effects of drought due to their thick leaves that slow down water loss by transpiration. These crops, as well as sorghum, pearl millet, pigeon pea, cassava, groundnut and cowpea, are drought-resistant and suited to climates with a defined mid-season drought.

Drought-escaping crops are those that can tolerate droughts because they have short growing periods and mature quickly before all the soil water has been used up. Early-maturing cultivars have been successfully bred in Kenya for dry areas, such as Katumani Composite maize and "Mwezi moja" beans. Cowpeas mature early and are both drought-escaping and drought-resistant (Squire, 1990).

A drawback of drought-escaping crops is that their short growing season restricts yields compared with long-season cultivars, although under dry conditions they will outyield the long-season cultivars. For example, improved pearl millet varieties in Tanzania, which mature two weeks earlier than farmers' local varieties, have yielded 43 percent more (2.31 t/ha) than local varieties (1.62 t/ha) (Letayo et al., 1996). Applying fertilizers to counteract nutrient deficiencies can speed up crop maturity and so enable them to escape droughts more easily.

Drought-escaping crops are more suited to short rainy seasons, or to soils which can only store a limited quantity of water. It is therefore important to select drought-escaping crops and varieties whose maturation period matches the expected length of growing season. If possible very determinate varieties should be avoided so that the risks of the whole crop being adversely affected by dry periods is reduced^[5]. Unfortunately, farmers often do not have access to varieties that match the expected length of growing season, and the length of growing season may vary widely from year to year.

It must be borne in mind that the choice of crops and varieties depends not only on their ability to resist or escape droughts, but also on their susceptibility to pests and diseases, labour requirements, availability of seed, ease of grain processing (threshing, dehulling and grinding), fuel requirements for cooking, and palatability.

Increase crop water use efficiency

Crop water use efficiency refers to the amount of dry matter produced for each millimetre of water that is transpired by the crop or evaporated by the soil, i.e. for each millimetre of evapotranspiration. Clearly, in dry areas the more efficient use the crop can make of the rainfall that infiltrates (referred to as the effective rainfall), the higher will be the yield. The following management practices influence crop water use efficiency:

Selecting water-efficient crops

A group of crops referred to as C₄ crops, which include maize, sugar cane, sorghum and pearl millet, are physiologically much more efficient at producing dry matter for each millimetre of transpired water than other crops, referred to as C₃ crops. But this distinction is most important in situations where rainfall is adequate. For areas where water deficits are common, the use of drought-resistant and drought-escaping crops is much more important.

Adjusting plant population to expected rainfall

A high plant population will use large amounts of water for transpiration during early growth provided sufficient water is available in the soil. Because of rapid shading of the soil by the crop foliage, less water will be lost by direct evaporation, ensuring a higher water use efficiency compared with low plant populations. High plant populations, and especially those with a more square planting arrangement, also increase water use efficiency through the quicker development of cover and therefore less weed growth.

Although evaporation losses are greater for low plant populations, soil texture and the frequency of rainfall events also influence the amount of water lost. Sandy soils in areas where rainfall occurs in few heavy storms will suffer less evaporation than medium or fine textured soils in areas with frequent rainfall events.

Where rainfall is erratic the situation is complicated, and becomes more than just a matter of water use efficiency. Farmers then face the dilemma of whether to sow at a low density to ensure some yield in bad years but underperforming in good years, or to use a high population to maximize yields in good rainfall years but to harvest very little, if anything, in bad years (Morse, 1996). If farmers have sufficient land they can opt for both options, i.e. an area with low population and another area with high population, but many small-scale farmers possess insufficient land for this to be feasible.

Response farming is an approach for matching crop management to estimated seasonal rainfall in variable rainfall zones (Stewart, 1988). Plant populations and N fertilizer applications are adjusted after the crop has been established on the basis of information about the expected rainfall. Initially, the crop is sown at a high population assuming a good rainfall season, and with a low application of N fertilizer. The expected potential of the season (good, fair, poor) is determined on the basis of the anticipated amount of rainfall during the first 30-50 days, derived from as many years' records as are available. Decisions are then made according to the amount of rainfall early in the season on whether or not to thin or to apply additional N fertilizer. So far, this practice has not been adopted by farmers because of the great variations in seasonal rainfall over short distances, because farmers usually intercrop, and because of the initial wastage of water that occurs if crops are thinned after 30 or more days (Morse, 1996).

Applying fertilizers

Applying modest amounts of N and P fertilizers to soils lacking these nutrients is a very effective way of increasing the efficiency of crop water use in semiarid areas, so that more dry matter and grain are obtained from the same amount of rainfall (Gregory et al., 1997). Phosphorus particularly helps in dry conditions by increasing root development and so enabling greater water uptake, whereas nitrogen tends to increase foliage production and hence transpiration in the presence of adequate water.

The effect of modest P fertilizer applications on sorghum yields and water use efficiency on P-deficient soils in Botswana is illustrated in Table 10. The efficiency of rainwater use and grain yield were greatly increased by P fertilization on deep soils, but not on shallow soils. This was presumably due to the low available water capacity of shallow soils, with greater rainwater losses by deep drainage.

TABLE 10
Effect of P fertilizer and soil depth on rainwater use efficiency and sorghum grain yield in Botswana (Adapted from Morse, 1996)

Values are the means of six tillage-planting treatments.

Rotations with legumes can have a similar effect to the application of N fertilizers. The higher water use efficiencies of fertilized crops are largely due to increased growth and transpiration, causing greater shading of the soil surface and less water loss by evaporation (Squire, 1990).

Increasing soil fertility through fertilizer applications may also increase the speed of crop development so that crops mature earlier, and so become more drought-escaping. As an example, the addition of P fertilizer to very P-deficient soils in northern Syria accelerated the maturity of sorghum by two weeks, enabling the crop to mature while water was still available in the soil (Shepherd et al., 1987). However, speeding up crop maturity can sometimes expose crops to water stress later on at a more critical growth stage (Morse, 1996).

Weed control

Another important management practice for increasing crop water use efficiency and yields in areas with water deficits is weed control. Competition from weeds in pearl millet reduced yields by 25-50 percent in northern Namibia (Spencer and Sivakumar, 1986), and complete weed control in the USA increased the water use efficiency of sorghum by 10 kg/ha/mm (Clegg, 1996). Good weed control during the first 30 days is an essential practice if water use efficiency is to be maximized.

Seed priming

Seed priming refers to soaking seeds in water before sowing to hasten germination and emergence, which leads to greater crop water use efficiency and higher yields. Soaking the seed for as little as 5-10 hours can reduce the time to emergence by 10 hours (LWMP, 1992), which may be crucial in enabling seedling roots to grow down to below a rapidly drying or crusting soil surface. For most crops soaking the seed for 12 hours is usually sufficient, but up to 24 hours are needed for rice and maize. Seed priming apparently does not work for finger millet (Village notes, 2000).

Early planting

Early planting at the beginning of the rains has several advantages. It increases the chances of a crop reaching maturity before the rains end, and as a result of early growth shading the soil surface, evaporation is reduced enabling more water to become available for transpiration. This increases the efficiency of water use by the crop and so increases yields. These effects are also favoured by the flush of inorganic nitrogen and other nutrients liberated at the beginning of the rains from the decomposition of dead soil micro-organisms. Interaction between the additional nutrients and soil water enhances crop growth and yield. Crops planted early usually also benefit from less pest problems.

Farmers who rely on hand weeding prefer to allow the weeds to germinate at the first rains and only when the weeds have been controlled will they sow the crop. Some farmers favour staggered planting, as the different stages of growth of the crop will spread the risk of the crop suffering from drought at a critical stage of growth.

Accumulate moisture from one season to the next

The amount of water in a soil available to crops can be increased in bimodal rainfall areas by keeping the soil in a clean-weeded fallow condition during the first season, in order to store rainwater for the next season. In this way the crop benefits from rainfall from two seasons, provided the water losses during the fallow period from weed transpiration, evaporation and runoff are negligible. There will be some inevitable losses by evaporation and probably by deep drainage. In some situations fallowing may be a feasible practice as it ensures a yield, but considerable labour is required to maintain the fallow free of weeds and to prevent serious weed problems the following season. Research from Zimbabwe has shown much higher yields after fallow than after the same or another crop, although the total yield as measured over a run of years was not significantly greater (Nyamudeza and Maringa, 1993).

Clean-weeded fallows are most feasible where extensive land areas are available and where weeds can be controlled mechanically. For these reasons it is regularly practised in large-scale highly mechanized systems in Australia and South Africa (Morse, 1996). However, the exposure of bare soils during the fallow period is not consistent with the principles of conservation agriculture and so is not a very desirable system. It will increase the loss of soil structure and organic matter and may give rise to serious erosion problems.

Water harvesting

Water harvesting encompasses many different practices based on the utilization of runoff from uncropped areas to supplement the rain falling on cropping areas, or to store water for irrigation, or domestic or livestock use. Emphasis is placed on the use of runoff for crop production. Water harvesting practices are appropriate in semiarid and arid areas where droughts are common and irrigation is not feasible. If doubts exist about whether or not the seasonal rainfall is adequate for cropping, efforts should first be made to minimize rainfall losses from low infiltration and evaporation.

In situations where water harvesting practices are appropriate and practised, runoff is considered as a valuable resource. This is in marked contrast to the other water management systems considered in this Bulletin, for which the approach is to avoid runoff by maximizing infiltration and to encourage farmers to develop an aversion to runoff. Water harvesting methods may be separated into:

• Runoff harvesting which refers to the harvesting of runoff from bare or sparsely vegetated areas and its collection for use in cropped areas. These may be as small as single planting positions, as in the case of "zaï" pits (see below). Two forms may be identified:

  a) Sheet-flow runoff harvesting where runoff occurring as sheet-flow is collected from gently sloping land surfaces.

  b) Concentrated runoff harvesting where runoff is collected from narrow channels such as footpaths, cattle tracks or transient streams in which runoff has been concentrated (Figure 19).

• Floodwater harvesting is the diversion of floodwater from watercourses for storage in farm ponds or microreservoirs.

• Water spreading which refers to the diversion of floodwater from watercourses for spreading over land that is to be cultivated (Figure 20).

• Rooftop harvesting which is the direct harvesting of rainfall from roofs, generally for domestic or livestock use (not considered further).

FIGURE 19. Example of concentrated runoff harvesting by diverting ephemeral flows into retention ditches or basins (Thomas, 1997)

FIGURE 20. Permeable rock dams with contour stone bunds for floodwater harvesting and water spreading (Thomas, 1997)

The capacity of the soil in the cropping or receiving area to retain runoff and rainfall is of crucial importance for crop production by water harvesting. Consequently deep soils and loamy textures, with high available water capacity, should be selected rather than shallow, sandy or very stony soils. There is little point in harvesting runoff for crop production if no attention is paid to the other aspects (chemical, biological and physical) of soil fertility. Substantial yield increases can only be obtained if nutrients are not limiting, so the addition of organic materials, manures, or fertilizers will often be essential. Good agronomic practices to control weeds, pests and diseases are also important.

Drought-resistant cereals, such as sorghum and millet, should be sown. Sorghum is particularly suited to water harvesting because it also tolerates temporary waterlogging. Legumes are much more susceptible to waterlogging, but should be encouraged when possible because of their ability to fix nitrogen. Suitable legumes in northern Kenya are cowpeas, green grams, black grams and pigeon peas. Chickpeas do well on black cotton soils (vertisols) (Thomas, 1997).

Social and land tenure factors are frequently very important in determining the degree of adoption of water harvesting practices. The labour required to construct the collecting areas and maintain bare runoff areas, the amount of land needed, the rights of individuals to the land and the feasibility of restricting grazing to avoid damaging the collection structures will often preclude their implementation. Successful implementation of water harvesting schemes is most often achieved when based on traditional water harvesting practices and when the whole community participates. More detailed information on the selection, implementation and management of water harvesting systems is given in manuals by Thomas (1997), FAO, (1991), Pacey and Cullis (1986), TAJAS (1999).

In sheet-flow runoff harvesting systems sheet-flow runoff is collected from a larger catchment (collection) area and is concentrated into a smaller cropping area. The lower the rainfall and the more water needed by the crop, the greater should be the catchment area compared with the cropping area. For catchment areas less than 10 metres, the ratios of catchment to cropping area generally vary from 1:1 to 3:1.

It is recommended that the slope of the catchment area does not exceed 5 percent in sheet-flow runoff harvesting. Bare catchment areas yield most runoff, but work is needed to maintain the land in this condition. In many situations catchment areas are left under natural vegetation and may sometimes be sown to short-season crops, but the efficiency of runoff generation will be considerably less. Diversion ditches may be necessary upslope of the area used for runoff harvesting to prevent excessive runoff damaging the water harvesting structures. There are many variations in the form and design of water catchment structures, but essentially they are pits, ditches or basins, or formed by earth or stone barriers. Concentration of runoff into smaller areas encourages deeper percolation of rainwater into the soil from where it is less susceptible to loss by evaporation. This increases the efficiency of crop water use and raises productivity.

The results of water harvesting compared with traditional cropping systems are very variable. In dry seasons yields can increase by as much as 300 percent compared with yields without runoff harvesting, but in wet seasons yields are likely to be reduced because only part of the land is cropped, or because waterlogging in the cropping areas has reduced yields (SUA, 1993). The adoption rates of runoff harvesting are often low because of the following factors (Morse, 1996):

• farmers' reluctance to maintain clean weeded runoff areas;

• high costs or labour requirements for constructing and maintaining pits, ditches and barriers;

• farmers' reluctance to crop only a fraction of the field as the productivity gain may not compensate for the higher yields obtained from the whole field in good rainfall seasons;

• farmers have limited available land;

• land is used for communal grazing which can damage the water retention structures.

Other adverse features of runoff harvesting are:

• risks of crops suffering from waterlogging in the cropped areas from excess runoff;

• high risks of erosion and other forms of soil degradation in the runoff catchment area;

• the greater the quantities of runoff harvested, the higher the risk of serious erosion problems;

• risks of collapse of barriers and infilling and overflowing of pits and ditches from heavy rainstorms, and of breaching of earth barriers by rodents or by the formation of cracks during clay shrinkage.

The following examples of water harvesting are mainly from the Soil and water conservation manual for Kenya (Thomas, 1997).

Zaï pits or Tassa

This is an example of one of the many traditional forms of planting pits practised in the arid and semiarid zones of the Sahel (FAO, 1996a). Zaï pits, about 15 cm deep, 40 cm in diameter and spaced every 80 cm, are constructed during the dry season by digging out the soil and placing it on the downslope side (Plate 61). Stones may be placed on the upslope side of the earth around the pits to help control runoff. Termites quickly attack organic residues that are blown into the pits and the formation of termite galleries from the surface of the pit deep into the subsoil encourages rainfall infiltration. Two weeks before the rains, one or two handfuls of dry dung (1-2.5 t/ha) are applied to the bottom of the pits and covered with earth. Millet is sown in the pits when the rains begin, and some runoff from the crusted soil surface upslope of the pits runs into the pits. The millet sends roots deep into the bottom of the pits where they find stores of water and nutrients recycled by the termites.

PLATE 61. Zaï pits or Tassa, for water harvesting - Illela, Niger

[C.P.Reij]

Zaï pits enable farmers to use small quantities of rainwater, manure and compost very efficiently and rapidly restore the productivity of degraded lands (Hassane et al., 2000). They are recognized as the most cost- and time-efficient technique for rehabilitating very degraded lands in the Sahel, and are an excellent means of establishing tree seedlings so that agroforestry practices can be introduced (Ouedraogo and Sawadogo, 2000). In Tigray province, Ethiopia, infiltration pits have tripled crop yields (Abay et al., 1998). The main constraints to Zaï pits are the labour needed to construct the pits in the dry season and the scarcity of manure. In Mali, yields of sorghum on test plots treated with improved zaï were far higher than on control plots with the conventional flat-planting method (Table 11).

TABLE 11
Effects of improved zaï on sorghum yields over 2 years (Wedum et al., 1996)

*= Optimum sowing date

Similar effects of improved zaï have been noted by farmers in Burkina Faso (Ouedraogo and Kaboré, 1996):

• by concentrating rainfall and runoff, crops are less susceptible to dry periods within the rainy season;

• economizes on scarce manure by concentrating its use at planting positions;

• encourages reintroduction of soil fauna (termites, etc.), which improves soil structure;

• because land can be prepared well in advance, planting can take place on time;

• enables rehabilitation of badly degraded land (important where there is large population pressure on land);

• possible to get a yield even in the first year and generally higher than yields obtained from fields already under cultivation;

• contribute locally to replenishing the groundwater table.

A study in Niger on yields and farmers' returns to labour in a "wet" year (1994-613 mm) and a "dry" year (1996-439 mm), compared yields of millet from the traditional planting procedure without planting pits (T0) with the use of tassa/zaï or demi-lunes (larger, half-moon-shaped), each with manure alone (T1), or with manure and fertilizer (T2) (Table 12).

TABLE 12
Yields, net value of production and returns to labour from existing tassa/(zaï) and demi-lunes, Niger (after Hassane et al., 2000)

The yields achieved by early adopters on only 4 ha in 1989 encouraged others to try, and the method spread rapidly to about 3 800 ha by 1995 and has continued to increase since.

Half moons (demi-lunes)

Sheet-flow runoff is collected from catchment areas of 10 to 20 m² areas by banks of earth constructed in the form of half moons 2 to 6 metres wide, which are constructed along contour lines in an offset arrangement (Plate 62) The spacing between contour lines will depend on the required ratio of catchment to cropping area. In Ouramiza in Niger, 20 cm deep half moons are 2 m wide and set at 4 m intervals along the contour, with a 4 m spacing between contours (FAO, 1996a).

PLATE 62. Examples of half moons for water harvesting - Illela, Niger

[C.P. Reij]

The half moon bunds guide runoff into their centre where it accumulates in pits, and excess runoff can escape around the ends of the half moons. For tree establishment the pits may be 60 cm deep and 60 cm square. Half moons may be planted to grain crops, forage grasses or trees, with the tree seedlings planted just above the pit or just below the bund to avoid waterlogging. Half moons are usually made by hand. Consequently their construction requires considerable amounts of labour. A further disadvantage for millet and some trees is that the large amounts of sediment deposited within the half moons form fairly impermeable crusts, which can impede emergence.

Contour stone lines

Contour stone lines refer to a single line of stones placed along the contour, whereas stone bunds are built up of stones to a height of 25 cm and about 35-40 cm wide. The base may be set in a shallow trench 5-10 cm deep to prevent the stones being swept downhill by the runoff. Bunds are permeable but slow down runoff, and by positioning smaller stones on the upslope side and larger stones on the downslope side, some sediment is filtered out and deposited behind the bunds. With time there can be a slow development of terraces. Spacing of the lines and bunds is generally 15-30 m.

Stone bunds have been very effective in Burkina Faso and Ethiopia for crops and rangeland rehabilitation. On slopes of 1-3 percent stone lines at 25 m spacing have doubled sorghum yields and reduced runoff by 23 percent (Zougmore et al., 2000). In some parts of Burkina Faso the stone bunds are constructed so as to be continuous with permeable rock dams created across gullies, which divert water from the gully and spread it over the land.

Contour earth ridges and bunds

Contour earth ridges are generally 15-20 cm high, constructed parallel to the contour and spaced 1.5 to 3 m apart, and have been found to be technically successful for producing crops and trees. They are constructed by digging a furrow along the contour and throwing the soil on the downslope side to form ridges. Prior cultivation of the land beneath the ridges promotes the binding of the ridge to the soil below. Cross ties are constructed in the furrow every 4-5 m to prevent runoff from accumulating at the lowest point and overtopping or breaking through the ridge. Sorghum or bulrush millet is often planted on both sides of the furrow, with the land between the ridges being left bare to encourage runoff generation.

Contour earth bunds are large ridges, at least 20-40 cm high, constructed with a road grader or tractor and plough. The bunds are spaced every 5 to 10 m and cross ties are constructed at 10 m intervals. The ridges should be rebuilt every season, and can be periodically moved downslope for a short distance to ensure a fresh supply of nutrients. Earth bunds may be faced with stones positioned on the upslope side. Earth ridges and bunds only work well when the soil is reasonably permeable so that infiltration can occur. If the soil is compacted or naturally impermeable, the buildup of water behind the ridge or bund can cause collapse or overtopping, resulting in the loss of water and soil erosion.

Another essential requirement is that the ridges or bunds do not form cracks and are sufficiently stable that they do not collapse when wetted by the runoff. Earth bunds have been successfully used for establishing trees at about two metre intervals together with grasses in denuded lands. The grasses assist in stabilizing the bunds. Although they have been technically successful, adoption by farmers without assistance in northern Kenya has been limited (Thomas, 1997).

Retention ditches

The most common and successful concentrated runoff harvesting practice in Kenya is the harvesting of road runoff in retention ditches. These are usually about 50 cm deep, 50 cm wide, and constructed along the contour. The excavated soil is either thrown uphill to form an enlarged fanya juu terrace, or downhill as in a cutoff drain. The base of the ditch is usually level, but may be graded to allow water to flow from one end to the other. Retention ditches are often used for bananas. Since bananas need large amounts of water and can tolerate temporary waterlogging, it is only necessary for the ditches to be large enough to retain the expected runoff. Alternatively, a spillway should be constructed so that excess water can escape without causing damage.

Retention pits

Small retention pits (or microcatchments) of 0.5 to 2 m³ capacity and lined with concrete are being investigated by farmers and researchers in Honduras for harvesting runoff from patios, footpaths and natural temporary waterways (Lopez and Bunch, 2000). The aim is to use harvested water for supplementary irrigation or for extending the cropping season.

Retention basins

Retention basins collect the runoff from roads, footpaths or transient streams. They may be rectangular or square, surrounded by small earth bunds and located adjacent to individual bananas or trees. Small basins may be used for individual trees or range reseeding, and larger basins for annual crops or small woodlots.

The Majiluba system is an example of traditional retention basins, which are extensively and successfully used by farmers in the semiarid lowlands of the lake zone of Tanzania for paddy rice cultivation (Gowring, pers. comm. 2000, and Morse, 1996). The main sources of runoff are ephemeral streams, paths and residential areas and the runoff is diverted into paddy-fields with earth bunds in the bottoms of the valleys. The grass Cynodon dactylon protects the bunds of the retention basins. This system requires collective organization by the community.

Farm ponds

Harvesting runoff from concentrated flows and storing it in farm ponds of 150 to 300 m³ capacities is being investigated in Burkina Faso and Kenya (Rockstrom, 1999). The aim is to use the harvested water for the supplementary irrigation of staple grain crops.

Floodwater harvesting and water spreading

Floodwater harvesting and water spreading refer to the utilization of water from watercourses (Thomas, 1997). There are two approaches: interception of floodwater behind large bunds with stone spillways in the floor of a flat valley so that the water is retained and spread laterally, and diversion of spate flow from an ephemeral watercourse, over adjacent land.

Temporary structures such as bunds are used to divert water from a watercourse and guide it over the land to be cultivated. Alternatively, the water is diverted into a series of basins, the water passing from basin to basin through spillways. The main problems are the unpredictability of floods, the dangers of structures being washed away, and the uneven depth of the spread water.

Collaborative stakeholder participation

A participatory approach should be adopted to ensure that the real causes of the problems are identified and that possible solutions are appropriate, feasible and acceptable to all concerned. Participants should include representatives of the whole community that is affected by the problem, i.e. men and women, young and old, rich and poor. Representatives of government and private organizations who can contribute to the solution of the problems should also participate. These may include government and NGO extensionists, commercial agricultural suppliers, credit and marketing organizations and technical specialists in soil and water management, agronomy, irrigation and groundwater hydrology as appropriate. Annexes 1-6 provide information and suggestions on collaborative activities and the participatory approach.

Need for a facilitator

An extensionist with whom the community is familiar should act as facilitator. It is the facilitator's responsibility to ensure that all participants have the opportunity to express their views and that no undue emphasis is given to the community's more prosperous and influential members. Further details of facilitators' roles are given in "Guidelines and reference materials on integrated soil and nutrient management and conservation for farmers field schools" (FAO, 2000a).

Need to tackle root causes

To successfully resolve problems, the underlying root causes of the problem need to be identified and addressed. Failure to tackle the root cause would result in the symptoms of the problem being treated rather than the underlying cause, which would greatly diminish the chances of successfully resolving the problem.

Participatory identification and prioritization of soil water problems

Evidence of crop water stress problems and their causes should be obtained from the field transects and soil pit examinations, using the indicators described and prioritized using a ranking method.

Participatory identification of the root causes

The root causes of the crop water stress problems are discussed and identified in a participatory manner by developing a problem-cause tree, such as that shown in Figure 21. Problem-cause trees illustrate the relationships between problems, causes, and the causes of the "causes" in a logical hierarchical arrangement, with the observable problem at the top and the ultimate root causes of the problem at the bottom. The causes of the observable problem are usually themselves problems for which causes can be identified, and so the process continues until the root cause(s) is (are) identified.

To produce a problem-cause tree, participants are asked to write the immediate causes of the observable problem on cards. Those cards corresponding to the main and most immediate causes of the problem are arranged in a line immediately beneath the observable problem. These causes then in turn may be considered as problems. The process is repeated to obtain the most substantial and immediate causes of these problems, which are arranged in a line below the problem to which they correspond. After each line has been established the tree is discussed and any necessary modifications are made. The process continues until the root causes have been established and agreed. This procedure is useful for all types of problems. For example, in Figure 21, the root cause of runoff is identified as the lack of fodder for livestock in the dry season.

FIGURE 21. Example of a problem-cause tree for high runoff

Note: Arrows point to the causes of each problem.

Participatory identification of possible solutions for testing

Possible solutions are emphasized at this stage, as farmers will normally need to assess the suitability and appropriateness of solutions by carrying out simple trials to evaluate or validate them. Farmers will frequently need to adapt these possible solutions to their own particular farming, social, economic and environmental conditions.

Possible solutions are identified through participatory discussions that draw upon the experiences and suggestions of all participants. The problem-cause tree diagram is a useful framework for focusing thoughts and discussion on possible solutions to each of the causes or problems identified, starting with the root cause(s), and working up the tree. An example is shown in Figure 22. Technical specialists and the facilitator may also need to propose solutions to the problems (Table 13), but whenever possible emphasis should be placed on modifications of farmers' existing technologies. Visits to innovative farmers who have successfully adopted or adapted possible solutions are highly desirable, as this enables farmers to freely discuss their advantages and disadvantages.

FIGURE 22. Example of possible solutions to the problem of high runoff (FAO Soils Bulletin No. 75)

Note: Causes are indicated by the direction of the arrows, and possible solutions are given in parentheses for each cause

TABLE 13
Checklist of possible solutions to soil water problems that will need validating and adapting with farmers

The fact that each individual type of action may also have more than one effect is illustrated by a visual approach to matching possible solutions to soil water problems as given in FAO Soils Bulletin No.75, pp. 56-57 (FAO, 1999a).

Participatory selection of possible solutions for testing

The possible solutions are discussed according to their suitability to the farming system and farmers' circumstances on the basis of the resources needed (labour, land, cash, on-farm materials and external inputs), their availability within the household or community, and other practical limitations. Some possible solutions will require changes to the farming system and household activities. For example, the introduction of silage (as a possible solution in Figure 22) may require silage crops to be sown on land that was previously used for food crops, and allocation of labour for collecting, making and distributing the silage. In this way the most promising possible solutions suitable for testing can be selected, and any changes required to the farming system or household activities can be identified.

Participatory testing and evaluation of possible solutions

The final step is for farmers to test and evaluate the possible solutions that have been selected to assess whether they are technically, socially, economically and environmentally acceptable to the farmer, his or her family and the community. Because of the highly variable nature of soils, even within a limited area, it is important that several farmers from different parts of the community carry out the same test on their farms. In this way it is possible to avoid atypical or strange results being obtained from one or two locations where the soil type or management was exceptionally good or bad. Farmers should carry out the initial tests on a small area only.

The testing of possible solutions by farmers on their own farms, perhaps following in-field demonstrations of the validity of the most likely ones, under guidance by field staff in conjunction with researchers, also encourages farmers to become more innovative, which is considered to be the key to sustaining agricultural development, especially in areas with inadequate advisory services (Bunch, 1995).