Contents - Previous - Next

Effect of buffer strips and contour cropping in humid and dry tropical zones (trials on erosion plots) (cf. Roose and Bertrand 1971)

Buffer strips


0 m

2 m

4 m


Adiopodoumé (1965)

Rainfall: 2300 mm

KR %






E t/ha/yr





Slope 7%

Bouaké (1965-66)

Rainfall 1180 mm

KR %






E t/ha/yr





Slope 4%

Treatments Allokoto (1966-71) Rainfall: 437 mm Groundnut, millet, sorghum, cotton Slope 4%

Traditional Haoussa control strip

50 cm buffer strips dh = 40 cm °tillage + ridging °frequent hoeing

low stone walls dh = 80 cm °tilling + ridging °frequent hoeing

protected ridges dh = 80 cm °tilling + ridging °frequent hoeing








E t/ha/yr






dH = difference in level

Effect of barriers of grasses (Setaria) or shrubs (Leucaena or Calliandra) on erosion: Ruhande hill near Butaré in Rwanda (cf. König 1991)


E t/ha/yr

International control

- bare tilled fallow


Regional control

- traditional cassava crop


- idem + Grevillea robusta trees + combination crops


- alley cropping (Calliandra every 5 m) + cassava


- idem + companion crops


- Calliandra planted every 10 m


- Calliandra every 10 m


- Leucaena planted every 10 m


- idem + ridges every 5 m


- Setaria hedge grown from cuttings


° hedges = three staggered lines on a horizontal terrace 1 m wide, every 10 m
° alley cropping = two lines of shrubs every 5 m
° if the hedges are to be effective, a filtering layer - a mulch of brushwood, weeds or prunings from the hedges - must be placed on the ground

° Note:

- the ineffectiveness of free-standing trees
- the reasonable effect of alley cropping
- herbaceous hedges are more effective than shrub hedges, which may be supplemented by adding covered ridges every 5 m

Erosion reaches 300 to 500 t/ha/yr on the bare control plot, and 200 to 250 t/ha/yr under traditional cropping, but falls toward 1 t/ha/yr on plots protected by grass or shrubs, clearly indicating that treated plots are stable after two to three years. So SWC has been successful, but what of crop yields?

Tests show that the biomass produced by shrubs increases gradually from 1.5 to 3.2 kg per linear metre of hedge, that the presence of grass (Setaria) at the foot of the shrubs gives good results in the first year but that after this Calliandra alone produces more than Leucaena, and lastly that annual nutrient intake by the deep-rooted shrubs can be as high as 78 kg of nitrogen, 10 kg of phosphorus, and 17 kg of potassium, which are then redistributed in the form of mulch during the crop growth cycle.

It had been hoped that a combination of prunings from hedges and 10 t/ha/yr of manure (dried paddock dung) would gradually improve crop yields. In fact, however, levels remained only fair (500 to 250 kg/ha of maize, plus 500 to 800 kg/ha of beans in the first season, and 450 to 650 kg/ha of sorghum in the second season). In the fourth year it eventually became clear that soil fertility had to be restored (a supplement of 2.5 t/ha of lime, 10 t/ha of paddock dung, and 300 kg/ha of NPK - 17.17.17) if crop yields were to increase significantly (more than 2 t/ha of beans and 1.4 t/ha of sorghum).

Complex grassed risers. After some years the microdams (grass strips, hedges, etc.) have produced gradual terraces divided up by steep-sided banks protected by a grass sward.

Figure 47 shows the development of a grassed riser on a 20 to 60% slope over a five-to ten-year period, during which it reaches a maximum height of 1 to 1.5 metres, after which there is an increased risk of destruction from animal burrows, gullying, and landslides, and farming problems as well (Roose 1990).

Since any break in the line of the riser across the hill compromises these measures, it is recommended that the whole length be pegged out with the rural community, and the first line of grass indicating the basic contour line planted across the whole slope or hill, leaving the individual farmers to complete the work during the five successive years, at their own pace on their own land. Imposing treatment of the whole hillside on the rural community can undermine the farmers' sense of responsibility, leading to neglect of the necessary upkeep (Ngarambé 1991). If the grassed strip is not continuous, runoff water will rush into the breach, carving out a gully and undermining the whole system.

Lines of Stones, Stakes, Grass or Straw (Figure 48a and b). This heading covers single barrier lines established along contour lines and permeable to sheet runoff. Many examples can be seen in Sudano-Sahelian semi-arid areas of Mali, Burkina Faso and Niger. Stone lines slow down the runoff so that it spreads out in sheets several centimetres deep, thus causing sedimentation first of sand particles, and then of finer particles which tend to clog the surface. The lines filter the water, trapping straw, animal faeces dropped during the dry season on rangelands, and various types of floating organic residue. This creates a localized deposit of fertilizer in the sedimentation area and in the watered area. And if these lines are laid out in the right direction, in the dry season they also trap sand carried off by wind erosion. They enhance production potential by concentrating both water and nutrients over an area 2 to 6 metres broad above the barrier, and redistributing it below it when there is excess water.

FIGURE 47 Development of a semi-pervious erosion control structure to dissipate runoff energy (cf. Roose 1991)



1. Dissipating runoff energy by spreading the sheet over the soil surface, which is protected by a thick sward of grass.

2. Transforming the soil cover into a series of gently sloped, cropped areas and protected banks (0.5 to 2 m high depending on soil depth and resistance).


1st year Mark out the contour line (every 5 to 20 m depending on slope and soil depth) by planting perennial grasses. Tillage will add 10 cm of soil above, and remove 10 cm of soil below.

2nd year Plant 2 additional rows of grass × pulses.

3rd year Plant another 2 additional rows, then plant quick-growing trees at the foot of the bank to reduce risks of landslides.

Since the soil cover will now vary in depth, more demanding plants can be grown in the zone above the bank where soil, water and fertile elements accumulate, and hardier plants below it in the scoured area to be restored.


Terraces are formed progressively through erosion, but primarily through tillage ( + 20 cm in height per year). The farmers can do all the work themselves.

Such a system requires neither major investment nor major upkeep, and pegging out the contour line does not require such precise surveying as do drains.

There is no loss (as against 6 to 10% for ditches) in crop area: (wood, fruit and forage crops will grow particularly well on the banks).

Effectiveness is maintained even in the case of very heavy rainstorms, and increases with time.

Examples of semi-pervious microdams in a semi-arid area (cf. Roose 1989)

Stone line or line of grass, straw, stakes

• a single row of semi-pervious obstacles
• slows down and spreads runoff
• traps wind-blown sand and fine particles in runoff
• delicate: knocked over by livestock and rills, buried by runoff

Figure A

Honeycomb network

• tying (r) reduction in lateral flow
• used to restore soil at tier, foot of hills for water harvesting

Figure B

Low stone bund

• at least 2 to 3 layers of well-fitted stones
• held together by:

- grass > < sideways movements
- hedge + trees > < livestock (downstream)

• traps 5 to 15 cm of sand + OM + silt
• filters floating organic matter
• slows and spreads flow over more time and a larger area

Figure C

Stone walls

• carefully stacked flat stones
• wall + upstream sand draining filter beneath the wall
• develops into gradual terraces

Figure D

Semi-filtering dam

• large stone dam across the head of the valley
• reinforced horizontal crest
• slows flow
• core of finer packed material if a sheet of water is to be held back

Figure E

Hazards. When sheet runoff builds up before the barriers, it eventually finds a way out, and at this point it speeds up (the Venturi effect), carving out a channel below the line, then under the stones, thus burying them.

As it spreads out in front of the barrier, the sheet creates a lateral movement which can lead to a local concentration of runoff and the formation of a more aggressive drainage line capable of carving out channels and shifting gravel and pebbles.

These lines are fragile: a single kick from the hoof of a passing animal can shift a stone, causing a breach through which the water will pour. The resulting rills then develop into gullies.

The lines have a limited life-span. The stakes and straw rot and are attacked by termites. The clumps of grass are choked in the centre, so that they thin out and leave pernicious breaches. The stones are overturned by livestock or buried under sand.

However, the organic matter accumulated in front of the barriers will have drawn termites, which often improve macroporosity and absorption capacity at this point, so that abundant grass and shrubs, and sometimes even trees, will grow there. In order to counter the lateral movement of water that leads to soil erosion, lateral ties can be introduced either by making small ridges perpendicular to the barrier, or by tying the field laterally in a beehive pattern - as is often done by the Mossi in north-western Burkina Faso to restore soil at the foot of hills. Another option is to wait a year until the drainage lines become clear and then use large stones to reinforce the weaker points where water collects. The soil must also be kept rough through regular tillage in order to break up the slaking or sedimentation crust which seals the soil surface. Hoeing the earth up into tied ridges lowers the risk of water concentration. Lastly, a filter of straw or a hedge above the stone or grass lines will make them much more effective.

Low stone bunds (Figure 48c).

Two or three layers of stones are so arranged along the contour line as to reinforce one another, a technique widely used in Yatenga Province of Burkina Faso. Lining one hectare with these stone bunds (400 m) takes about 30 to 60 days' work, plus transport between quarry and field (1 day's use of a truck). The bunds slow down runoff, spreading it in sheets so that it is absorbed in less than one hour, thus causing successive sedimentation of sand, aggregates, and fine, humus-bearing particles, which will then form a sedimentation crust. Only the excess water flows over the first layer of stones. More water is stored than in the case of stone lines, and the sheet of water often covers 5 to 8 metres in front of the permeable barrier. The bund filters straw, animal faeces and various kinds of floating, organic matter, so that farmers see one of its most useful functions as that of maintaining soil fertility.

Theoretically, these stone bunds are laid perpendicular to the direction of the water flow, but not necessarily to winds. They thus do not always trap sand blown across the ground during the dry season.

The bottom of the first layer of stones is planted several centimetres deep in the ground, with soil between them, so that 5 to 15 cm of filtering, organic, sandy soil collects above the bund, improving soil storage capacity and forming new topsoil.

The second and third layers downhill (made of smaller stones or grass) divide the excess flow, absorbing the runoff energy and preventing channels from being dug out downstream of the barrier during major rainstorms. Tillage of the cropped strip and erosion will soon cause an embankment to form. It must be stabilized with grass, for example Andropogon and Pennisetum.

Hazards. If the top of the bund is not strictly horizontal (as in the case of smoothed contour lines), the sheet runoff flows toward the lower points and forms drainage lines, speeding up and carving out channels, which then develop into gullies and drain water from the whole slope. However, a strictly followed contour line will produce fields of widely varying sizes and shapes (a deviation of 10 metres for the slightest termite hill on 2% slopes), which causes problems for mechanized cropping. Even then, small drainage lines will form, but these can be treated in various ways:

• using large stones to reinforce the points where the water collects, and gradually levelling these areas;

• increasing the roughness of the soil through coarse tillage with toothed implements, repeated hoeing, and tied ridging;

• sowing grass (Andropogon) above and around the bund to brake sheet runoff;

• tying the field for 5 metres above the bund with earth ridges or stone bunds - although the latter can cause problems for mechanized cropping (Lamachère and Serpantié 1991).

Damage to the bund by passing livestock can be reduced by planting grass (to cover the stones), and a hedge below the bund, as well as trees which will eventually "tie" the landscape, creating an area dominated by hedges. In areas where stones are scarce, the same effect can be achieved by sowing a 50-cm strip of grass (Andropogon) or a hedge (at least three lines staggered on alternate rows) between two contour ridges (20 cm high) (Roose and Rodriguez 1990). In mountainous areas farmers often collect stones that come to the surface and heap them up on the edges of their fields (particularly on banks around the fields). If these heaps of stones are arranged along contour lines, they act as stone bunds.

A team of ORSTOM scientists (Serpantié, Lamachère, Martinelli and others 1986-1992) have studied the combined effects of stone bunds and tillage, comparing this with a control plot in a water-harvesting area in the Bidi region near Ouahigouya in Yatenga, in northwestern Burkina Faso (Table 37, Figure 49) (Serpantié and Martinelli 1987; Lamachère and Serpantié 1991).

Stone walls (Figure 48d).

These walls are carefully built by piling up flat stones, wedged with small rock chips. They are often found in sandstone hills, for example near Bamako in Mali. The first step is to dig a trench along the contour line, digging down to a coherent horizon, and build a drainage filter made of a layer of sand and gravel on the sides and bed of the trench.

On a medium to steep slope, gradual terraces are quickly formed by throwing the earth from the trench uphill and by water erosion, but above all by dry mechanical erosion during tillage.

Hazards. The direct pressure exercised on the wall by the soil cover and ground water can push it out of true so that it will bulge and, eventually collapsing unless there is good drainage above the wall and a bed of gravel below.

As time passes, the foot of the wall will be eaten away by water erosion or tillage in the field below it. This process can be slowed by planting grass or fruit trees on the bank that develops at the foot of the wall, which will prevent soil from creeping downhill.

Pervious and semi-pervious check dams (Figures 48e and 52).

This approach entails heaping up a line of large stones with the crest following the contour line, in order to dam the head of a valley, thus slowing flow and facilitating groundwater recharge. It can take 300 to 600 days' work to build these enormous stone bunds which are 200 to 300 metres long and 1 to 2 metres high. The length of the dam depends on its height and the depth of the gullying at each point. The crest of the dam must follow the contour line strictly and the dam itself must be set in a foundation trench 20 to 40 cm deep, which must also be covered by a pervious bed. This barrier slows down runoff to the valley bottom, although it passes quickly through the large stones - unless the dam has been built with a pervious core of finer gravel, which can hold back water for several days.

If a sheet of water is to be kept upstream of the dam (in other words, if it is to be a semi-pervious rather than a pervious dam), for example to grow rice, a clayey core behind the gravel filter will be necessary.

Sedimentation upstream of the stone dam is fine and slow (1 mm per year) in a fairly undegraded, undulating landscape, but can be fast (10 to 50 cm per year) in a gullied, hilly landscape (e.g. the area north of Ouagadougou - Dezilleau and Minoza 1988).

Hazards. If water filters too fast through the dam, erosion can burrow through it and below it. If there is too much pressure beneath, the water carves out a gully which will eventually breach the dam through headward erosion. A gravel and sand filter must therefore be built to slow down runoff: this material is poured into the foundation trench under the structure, as well as between the large stones that make up its core (Figure 52).

If the sheet of water flowing over the dam is over 20 cm, the structure may well be swept away by the speed of the current. This can be prevented by placing large stones, or better a small gabion, on the brow of the dam, and also on the slope of its upstream side (2/1 slope).

The water normally drains away fast after the rains stop, although the land immediately above is left waterlogged, so that the sorghum traditionally grown in flat valley bottoms rots and suffers from waterlogging, while water is not available long enough to grow rice - which is prized for feast-days. Only off-season gardens and fruit trees located around the valley in fact profit from the increased groundwater recharge from a check dam. It is therefore important to have a clear idea of objectives. Check dams recharge groundwater but do not hold back much water - in any case not enough to produce rice four years out of five in the Ouahigouya region. If the aim is to create a rice field, it is best to choose an impervious soil-saving earth dam, which will hold back a large enough sheet of water for this crop.

Tests on the effect of stone bunds and tillage on runoff, erosion and millet yields at the Bidi Station, Samniweogo, northern Yatenga, Burkina Faso (cf. Lamachère and Serpantié 1991)

Tropical, sandy-clayey, well-drained, ferruginous veil, 25 to 220 cm deep over ironstone. Slope 2.5%. Plots of 3100 m³, receiving runoff from a 1250 m² catchment area until the end of 1987.


Rainfall mm + 2



Erosion t/ha/yr

Yields kg/ha

Kaar %

Smooth moist soil

Max KR %



smooth dry


1985 + harvesting runoff


1. Traditional








2. T + stone bund








1986 + harvesting runoff


1. T








2. T + SB






3010 (+ 19%)

406 (+4%)

3. T + SB + Tillage

4640 (+ 54%)

837 (+ 106%)

1987 + harvesting runoff


1. T








2. T + SB






2330 (+ 32%)

443 (+ 28%)

3. T + SB + Ti






3140 (+ 35 %)

679 (+ 53 %)

1988 + ties - catchment runoff


1. T








2. T + SB






2090 (+ 11%)

362 (-6%)

3. T + SB + Ti






2290 (+ 10%)

438 (+ 21 %)

Average effect over 2, 3, 4 yrs





bund - control

+ 4%

- 34%

+ 22%

+ 22 %/4 yrs

bund + tillage - bund

+ 4.3%

+ 52%

+ 48%

+ 61 %/3 yrs

35 = Markedly different from control treatments

Notes on Table 37

Treatments are as follows:


1. T = traditional millet crop + light manuring: 7N + 10P + 7K sod seeding of 45000 plants then 2 hoeings, loosening around roots.

2. T + SB = idem 1 + stone bunds on contours at 40 kg/m², i.e. 2 lines of ironstone blocks every 20 m + lateral ties as from 1980.

3. T + SB + Ti = idem 2 + ox-drawn ploughing between 15 June and 15 July and sowing on the same day.

b. The stone bunds have little effect on overall runoff or on runoff during rainfall on rough soil (Kaar - see page 46), but much more on peak flows, the curbing and spreading of flow, and risks of erosion. The hydrodynamics depend on interaction between the bunds and the roughness of the soil. The impact on yields depends on the volume of rainfall and its distribution at flowering and earing.

c. Tillage, here combined with erosion control, can significantly improve infiltration of the first 100 mm of rain. After this, slaking and sedimentation crusts are so marked that gains in infiltration due to stone bunds become negligible.

d. The contribution of runoff harvesting is noticeable primarily on the uphill plots as long as the soil is absorbent enough; it increases the risks of drainage on the uphill side of the bunds.

e. Improving the water supply to plants raises the problem of replenishing available soil nutrients: increasing production of unrestored biomass hastens soil impoverishment.

f. The positive effect of tillage on improved infiltration and yields decreases from one year to the next, while losses through erosion increase. It seems that tillage weakens the soil, making it more erosion-prone, while hastening the mineralization of humus in the soil.

These results prompt certain observations:

1. The gain in infiltration when the flow of sheet runoff is slowed down by the stone bunds is relatively meagre if the ground is smooth, but much more substantial when the ground is kept rough by tilling, hoeing and ridging. There is thus a very positive interaction between tillage, soil roughness, and the effectiveness of stone bunds (Figures 50 and 51).

2. Since the main effect of stone bunds is that of curbing the flow of water by spreading it before these obstacles, peak runoff flows near the foot of the slope are reduced, flow persists longer after rainfall, and the speed of sheet runoff is reduced - which means that the sediment load is reduced by 50%.

3. Normally more biomass is produced with stone bunds except when waterlogging hinders the growth of millet. Millet grain production increases by 15 to 60% on fields with stone bunds, and by 50 to 80% when tillage is combined with stone bunds. However, during the driest years (e.g. 1985), yields are as low as on untreated plots. The use of such lines therefore does not reduce famine risk in years of serious drought. Millet yields can also suffer from waterlogging when monthly rainfall is over 200 mm.

4. The effect of a stone bund is felt primarily uphill, and secondarily downhill:

- uphill, sedimentation of sand and organic matter over 2 to 6 m, due to the spreading of sheet runoff, the stone bund becomes blocked to a height of 5 to 15 cm, which improves the topsoil; this area then receives an extra amount of water, which can improve the crop water supply but may lead to waterlogging and to leaching of nutrients through drainage (sorghum is less prone to this than millet);

- downhill, water trickles through between the top stones, and may either spread out in a sheet and irrigate the whole area once more, or concentrate into rills and dig paths for itself (gullying sheet erosion) which will drain, dig into and dry out the downhill area, and possibly destroy the system of stone bunds at some points through headward cutting; since the positive effects are confined to the neighbourhood of the bunds, it would be better to set up small lines every 20 m (25% of the area affected by treatment) rather than large ones every 50 m (only 10% of the area affected by treatment), particularly since large stones tend to funnel the flow into larger streams and hence increase the risks of heavy concentrations of water.

5. When sheet runoff is slowed down, pools accumulate along the line. This water will then move sideways to exit at a low point in the bund, and can thus cause lateral erosion and a dangerous concentration of water. Many solutions have been suggested to this problem of sideways movement: tied ridging or loosening around the roots to give a very rough soil surface, setting up ties perpendicular to the bund, or planting grass (Andropogon) to stabilize the immediate area of the bund.

FIGURE 49 Diagram of the experimental layout at Bidi in Burkina Faso

FIGURE 50 Comparison of runoff: control plot/treated plot. Average rainfall

FIGURE 51 Comparison of runoff: control plots/treated plots as a function of rainfall aggressiveness

FIGURE 52 Pervious or semi-pervious check dam


• sorghum yields
• security
• slowed runoff
• gully erosion


• cost: c. 500.000 CFA francs + collective labour profiting 1 or 2 landowners
• zones

± humid
± waterlogged
± silted

Problems connected with land ownership should also be borne in mind, for building a check dam of this type requires considerable work on the part of the local community (15 people for 30 days) and the transport of a huge volume of stones (from 100 to 500 m³, at an average cost of 4 to 40000 FF). And although the structure takes all this effort on the part of the community, it in fact allows improvement of a mere 0.5 to 1 hectare, probably belonging to a single family. In order to avoid wrangling, advance plans should be made to redistribute the improved land to those taking part - not always a possible solution. Another solution is to negotiate a form of payment by the beneficiary to the neighbours who help in the work, by creating a bank which advances money and recovers it on improved harvests in the following years. Food for work, payment in kind, or an undertaking to work for the others for some days, are other possibilities.

The same amount of effort and the same volume of stones could go to improve 10 to 20 hectares of sloping land belonging to twenty families, who could easily build the necessary stone bunds on their own. It has not yet been ascertained whether this form of watershed management can offer the same production security as management of the valley bottoms, bearing in mind that although certain bottoms are completely flooded in the rainy season, in particularly dry years they are the only places where villagers can be sure of a certain level of productivity.

The importance of ensuring household food security makes it vital to improve both the land in the valley bottoms, which will produce even in dry years, and the land on the hillsides, which will produce best in wet years.

However, it must always be borne in mind that, like other methods, this is not the universal solution. It is valid for certain gullied valley bottoms, but much less so for flat bottoms, where there is very little sedimentation.

In conclusion, there is a wide variety of semi-pervious microdams. They have the advantages of being easy for villagers to build, and of changing topographical conditions (the gradient of the slope). However, they do lead to the loss of some runoff as well as the nutrients and colloids that represent the wealth of these soils. This water could be recovered further downstream by irrigation infrastructure.

Contents - Previous - Next