Contents - Previous - Next

Analysis of local conditions

The two zones selected vary considerably in terms of erosion risks.

The low-lying plains (900 to 1500 m altitude), covered in shrub savannah, receive 800 to 1000 mm/yr of rain over two rainy seasons. This area is less rugged (slopes of less than 15%), less well-watered, and less densely populated than the rest of the country as a result of malaria and various tropical ailments, and consequently less exposed to the risk of erosion. Most of the eastern part of this area is presently given over to extensive livestock production, even though the soil is often quite fertile. The ferralitic or ferruginous soils here are less acid and less desaturated than elsewhere, but water runs off more easily (a result of the formation of slaking crusts) and crops suffer each year from rainfall irregularity and stress. Managing surface water is probably the main problem for agricultural development in the area, while losses through erosion and drainage are moderate.

The central plateau (1500 to 2000 m altitude) receives between 1200 and 1500 mm of rain, spread over ten months. Rainfall erosivity is significant (250 to 500 RUSA units), and the very dense farming population (250 to 800 per km²) has to cultivate every piece of land, including slopes of over 40% on the sides of convex hills.

During the first season (September to December), the rain is fine and only one quarter as forceful as in West Africa. It falls on dry, well-drained, and manually well-tilled soil, and does not cause much damage. However, in the second season (February to June) there are several larger, heavy storms (60 to 100 mm/day). If they fall on moist soil, on steeply sloping slopes or soil fine-tilled for sowing, the water forms rills, which then scour the full depth of the tilled soil down the whole length of the plot. All this soil easily blocks the erosion control ditches, which overflow, so that the runoff accumulated in them then cuts gullies which will wreck erosion control measures all the way down to the foot of the hill.

The topsoil horizon is quickly scoured, not only by rill erosion but also by dry mechanical erosion following multiple tillage procedures: deep-ploughing the land twice (to dig in weeds) and hoeing twice in each cropping season causes 30 to 60 tonnes of earth to move down the slope as far as the next obstacle, so that banks rise by 15 to 30 cm per year.

At this rate, the soil cover on the hilltops is soon stripped, uncovering alterites and blocks of rock. Much less water is now absorbed and retained, which means that during periods of heavy rainfall, large amounts of water gush down from the degraded hilltops, gullying the slopes, changing the flow rate of rivers, increasing peak flows, attacking river banks, and washing away the gravel from river beds. The delicate balance in these mountains is disrupted by uncontrolled clearing, overgrazing, growing crops that provide little cover on very steep slopes, and the removal of stones that protect river beds for building.

Ferralitic soils are generally very desaturated, acidic, (frequent pH of 5 - 4), deficient in P and N. and poor in bases. They seem on the whole very permeable, except where they have been compacted (tracks, cattle trails, paths to dwellings) or pounded by rain. They retain little water (1 mm of available water per centimetre of soil) or nutrients (1 to 5 meq/100 g of fine soil), so that it is important to maintain an adequate level of organic matter. They are often rejuvenated by erosion, with a layer of rubble or ferruginous gravel at a depth of between 30 and 100 cm. Soil erodibility ranges from low to medium on schist, and Wischmeier's K factor is generally under 0.20 (Roose and Sarrailh 1989, Ndayizigiyé 1993).

Erosion (t/ha/yr) and runoff (% of annual rainfall) on small plots ( 5 × 20 m) on steeply sloping (25-60%) ferralitic soils in Rwanda and Burundi

Plant cover


E t/ha/yr

Runoff Kaar %

Bare soil

tilled parallel with the slope

300 to 550

10 to 40%

Manioc or potato, maize/bean or pea-sorghum, as companion crops

traditional hoe tillage

50 to 150

10 to 37%

Crops + idem + 200 trees/ha

litter 50 kg/tree/yr

30 to 50

5 to 7 %

Idem + trees + hedges every 5 to 10 m


year 1: 7 to 16

10 to 15%

3 to 6 kg/m²/yr

year 4: 1 to 3

1 to 3%

Idem + trees + hedges

± covered ridges every 5 m

1 to 4

0.1 to 2%

Banana plantation

open, mulch removed (10 t/ha/yr) or

20 to 60

5 to 10% (45)

complete, mulch spread out or in lines

1 to 5

0 to 2%

Coffee plantation or manioc

thick mulch (20 t/ha/yr)

0 to 1

0.1 to 10%

Pinus forest, pasture, old fallow

(5-15 t/yr of litter)

0 to 1

1 to 10%

( ) = maximum levels recorded

Except for the two planting periods the countryside is green, for annual rainfall is good if irregularly distributed. Erosion risk would therefore be moderate if the cultivated slopes were not so steep (Berding 1992). Two country-wide surveys indicated that 50% of the cultivated land is on slopes exceeding 18%, 20% on slopes exceeding 40%, 5 to 6% on slopes exceeding 65% (the limit for terracing), and 1% on slopes exceeding 84%.

Erosion risks are aggravated locally by two phenomena.

Land tenure problems. The concern for equality in inheritance means that each heir receives an equal share of each section of land, which means in turn splitting the original plot into as many vertical strips as there are heirs. The result is that on densely populated hills (those farmed for a long time) very long, narrow plots are put under crops at the same time, which seriously increases the risk that sheet erosion will scour the soil right to the bottom of the slope. Once such scouring starts, it happens again each year in the same spots, because it is difficult to prevent runoff from flowing toward the lowest points in a field. The land is quickly ruined. Land tenure laws should be changed.

Landslides. If erosion control on a hill calls for digging total absorption ditches on slopes of over 40% or on shallow soils on a sliding alterite (schist, gneiss, micaceous rock or volcanic ash on granite domes), the slope is thrown out of balance. If a long series of storms waterlogs the soil cover (and especially if this is compounded by earthquakes), it can start sliding from one of these ditches, and continue down to the river, which can then be temporarily blocked by this mass of earth.

Experiments show how urgent it is to combine all available erosion control techniques in order to stabilize sloping land while also substantially increasing its productivity (see Table 40).

There are about 250 reliable measurements of annual erosion on plots of 100 m² (20 m in length) fairly similar to farmers' fields, on steep slopes (25 to 60%, except for the IRAZ banana plantations, where S = 8%), on ferralitic soils that have been somewhat rejuvenated or received colluvial deposits and are very desaturated and acid, but also very resistant to rainfall aggressiveness (K < 0.2 to 0.1). The results of these experiments indicate that:

• the risks of sheet and then rill erosion are very high on bare soil, varying from 300 to 550 t/ha/yr, depending much more on rainstorms than on slope; it would take only 5 to 10 years to remove the whole topsoil horizon (20 cm) at this rate;

• the risks of runoff (Kaar = 10 to 40%) can be serious on such steep slopes when they are poorly covered (as with degraded soil);

• traditional farming methods and intercropping do considerably lessen risks (C = 0.2 to 0.5), but not enough, since the tolerance threshold is no more than 1 to 12 t/ha/yr depending on soil depth;

• trees dotted among the crops do little to improve soil conservation;

• hedges of grass or bushes every 10 metres, plus large ridges covered with pulses or sweet potatoes every 5 metres, do constitute a valid preliminary solution;

• mulching (tested under banana, coffee or cassava) is a second solution which is immediately effective even on steep slopes;

• reforestation with pines (needle litter being very effective) or other species allowing an under-storey quickly reduces runoff and erosion to acceptable levels (Roose, Ndayizigiyé and Sekayange 1992).

Blind ditches and bench terraces cannot be studied effectively on these small plots (5 m wide). On land managed under erosion control projects, it has been seen that these methods can increase risks of gullying and landslides where the soil cover is thin or the slope too steep (> 40%)

Farming methods - not just erosion control structures - play the major role in stabilizing slopes.

In conclusion, these verdant landscapes can give an impression of stability to busy experts who are used to the gullied, bare land of semi-arid regions. In reality, however, the soil is very poor, very steep slopes of 60 to 100% are cultivated out of necessity as land is short, rain is excessive at some periods and scant at others, and the cover provided by crops on the most degraded land is too light to protect the soil from the various erosive processes in the Rwandan hills (see Figure 71).

FIGURE 71 Six processes leading to rural environmental degradation (cf. Roose 1992a, b) Quartzite/schist hill

Traditional techniques

The crops are planted in dispersed fashion around the habitat in direct relation to soil fertilization. When a young family sets up home on a levelled platform cut into the hill, it plants its banana plantation around it, and this will receive most of the available nutrients (domestic waste, crop residues, ash, peelings and latrine waste). Companion food crops are grown between the bananas: maize, beans, cush-cush, potato and herbs. A small field of maize intercropped with beans receives a little manure/compost, and broadcast-sown sorghum is grown in the second season.

The only plots not eroded are those that are mulched and under coffee trees: in order to avoid the penalties conscientiously imposed by Ministry of Agriculture field staff, coffee plots (100 to 200 m²) are copiously mulched with cassava and sorghum stalks, various types of grass pulled up from the banks, and banana leaves. The remaining land (two-thirds) receives no manure or fertilizer, and inevitably degrades under such frugal crops as cassava and sweet potato.

Weeds are carefully pulled up, either - depending on need and season - to feed stabled animals, or to cover furrows and reduce erosion, or to be piled up in large heaps, covered with earth and immediately planted with sweet potato cuttings. In any case, vegetation is very quickly recycled.

Plots are sometimes scattered several kilometres away from dwellings (rented fields). Despite the many disadvantages (time spent travelling back and forth, difficulty in guarding and manuring plots), scattered fields do allow farmers to cope with climate-related risks (localized storms and hail, damage from animals and disease). Young technocrats dream of dwellings concentrated in villages and consolidation of landholdings so as to promote intensive, modern, mechanized farming. This is a serious mistake in a country with no alternative way to feed a very large rural population forced off the land (no industry, no international waterway, no trade). Furthermore, the land is too steep to risk the introduction of tractors (little likelihood of cost effectiveness, and risks of compaction), and an element that now enriches the land (domestic waste) would become a pollutant hard to control within a village.

Present farming techniques take a great deal of work, which is often performed by groups of neighbours using two rudimentary implements, the machete (sometimes curved like a sickle) and a long-handled hoe. Following a short fallow (from a few months to one or two years), the soil surface is cleared of infesting weeds and then deep-ploughed to turn in the weeds (30 cm and more). Stolons and other persistent roots are dried in heaps, and composted or burnt. A month later the plot is fine-tilled for drill sowing (maize) or broadcast sowing (second-season sorghum); an intercrop may be sown after the first hoeing to fill empty seed holes and cover the whole area.

All tillage is manual, using hoes. Animal traction is difficult on steep slopes and is never even considered, for there is no tradition of draught animals. There is no mechanization (far too expensive at such a distance from the sea), which means that there is little compaction of deep horizons, and drainage seems normal. Deep drainage would be needed only in the vicinity of springs.

Ridging or larger mounding is confined to tuber crops and digging in weeds. On the other hand, crops are usually grown on raised beds or large mounds in the valleys and marshlands in order to ensure good drainage.

Apart from spreading manure on fields near dwellings, soil fertility is maintained by intercropping, rotation, digging in weeds, and a short fallow. However, there is an erosion control technique traditionally used on steep slopes, especially for growing peas on schist and in the highlands in the north and on the Zaire-Nile Divide (Nyamulinda 1989). It consists of micro-step terraces 1 metre wide, cut into the slope, preserving the root systems of clumps of grass. This allows space for a double row of maize/beans or peas. The risers (0.5 to 1 m high) are kept firmly in place by the root networks. The main concern with these narrow terraces is to keep the cultivated beds within the topsoil horizon, for the wider the terrace the more the soil structure is disturbed and the more the sterile deeper horizons are exposed (Roose et al. 1992). The traditional technique is to turn half a bed on to the one below in the second year - thereby mechanically shifting the surface layer of soil right along the slope. Trials on erosion plots have shown that with an improved version of this method (placing beds strictly along the contour and using the grass from the risers) all erosion can be stopped and rainwater better managed, even on schist soils on 60% slopes.

Lastly, there is a local technique of managing runoff on tracks, which consists of digging a pit in the upper slope, in which runoff and its load of sediment are directed. When it is half-full of sediment, a clump of banana trees is planted in it, to benefit from the additional water and nutrients. When the first pit is almost full, another is dug lower down (= Rudumburi).

In conclusion, traditional methods allowed maintenance of the stability of the landscape and a modest production level. Now that the population has become too numerous to keep enough land under fallow, something has to be done to keep the soil in place, but also to bring about a rapid increase in soil productivity for both food and fuelwood crops).

Suggestions for managing surface water


In semi-arid regions (especially the eastern savannah), placing land under cultivation brings a major increase in runoff and a reduction in evapotranspiration, and thus in the production of biomass. Runoff control measures (improvements in infiltration and localized storage) can therefore have a considerable impact on yields of crops that suffer as much from drought as from mineral deficiency. Farmers will quickly become interested in runoff management techniques.

In humid regions (R > 1000 mm), clearing land and putting it under cultivation bring an increase in the risks of runoff, in peak flows of rivers, and therefore in the risk of erosion of banks. There is a consequent reduction in drainage, the leaching of fertilizers, and the dry season flow of springs and rivers. Runoff (and erosion) control will thus have relatively little effect on crop yields, unless there are periods of drought during vulnerable phases in the growth cycle. This is one reason why erosion control has had little impact on yields in the humid hills of Rwanda, the other causes being the chemical poverty and acidity of the soil.

In conclusion, if runoff is reduced by farming techniques and/or suitable erosion control structures, plant production must be intensified to avoid increased risk of nutrient leaching by drainage water and landslides on steep slopes: hence the attraction of intercropping, fertilization and agroforestry.


Four approaches to surface water management can be identified, depending on climate and soil permeability, with corresponding erosion control structures and farming techniques (Roose, Ndayizingiyé and Sekayange 1992). Here only the most appropriate are described.

Cisterns of drinking water collecting 10 to 50 m³ of clean water from roofs considerably alleviate the work of carrying water, improve hygiene, and allow for a few penned animals, the production of manure and a very intensive multi-storey garden around dwellings.

Cisterns or pools collecting runoff water (100 to 500 m³) on tracks or rocky or overgrazed slopes allow livestock watering and supplementary irrigation of short-season vegetable and fruit crops (see Haiti).

Total absorption ditches encourage infiltration of runoff water on slopes of less than 20%, on deep, permeable soil. Unfortunately, they require a lot of work (200 to 350 days to dig, plus 20 to 50 days per year for upkeep), and hardly improve crop yields at all (which is why farmers abandon them). Their main attraction is in the gradual transformation of the landscape into very gently sloping terraces. Diversion ditches are unadvisable for mountainous regions, as gullying is bound to set in at their outlets.

Stop-wash lines or semi-pervious microdams (lines of grass, stones, hedges, grassed banks) do not stop runoff, but do slow down water, dissipate its energy, and spread it into sheets, thereby encouraging sedimentation. A bank quickly forms (20 to 30 cm/yr), with a gradual terrace which can then be transformed into two horizontal terraces, one enriched (reserved for intensive cropping), and the other poorer (frugal crops such as cassava and sweet potatoes), so that fertility must be gradually restored (see Figure 72). The demand for labour is more occasional (50 days to build, plus 10 days per year for upkeep), as are fertilizer requirements.

Horizontal or bench terraces allow all water (rain + runoff between terraces) to be absorbed, and make the most of manure inputs. Clearly, however, bench terracing requires a huge investment in terms of labour (500 to 1000 days/ha to build) and inputs (10 t/ha of manure, 1 to 5 t/ha of lime, plus the fertilizer for each crop) before the natural fertility of the soil is restored. This method should be chosen only if there are both inputs and the markets and praticable roads to capitalize on surplus production. There must be no risk of landslides.

Micro-step terraces (cultivated width about 1 metre) on permanent grassed risers (maximum 50 to 100 cm) require much less work and stabilize steep slopes very well under manual intercropping, since the crop roots remain in the original topsoil.


Tillage techniques that modify the state of the soil surface, roughness, plant cover, the activity of mesofauna and/or infiltration capacity are often very effective in reducing the volume of runoff and dissipating its energy.

Flat tillage with large clods is essential on soils that are too compacted. It temporarily increases infiltration, improves water storage and helps to dig in crop residues and combat weeds. Unfortunately, it inhibits earthworm activity, reduces soil cohesion, and increases erodibility by runoff water, especially if seeds are sown on a bed of very fine aggregates.

Mounding and ridging, parallel with the slope, gather together good topsoil so that large tubers can be grown, but these practices are dangerous on steep slopes since they concentrate runoff into trickles that can dig rills and gullies, and detach gravel and other stones that protect the soil from rainsplash.

FIGURE 72 Development of gradual terraces into horizontal terraces: a CIGAND project proposal

(1) At present the hills hold many gradual terraces which are too wide between banks which are too steep or even undermined at their base. After 5 to 7 years, the lower part of the terrace has filled out with fine soil, while the upper part is scoured and tends to become sterile: intervention is required.

(2) Fine the line in the terrace where sterile subsoil or rock appears at about 1.3 m (50 cm + ½ the height of the bank), plant a new line of grasses, and build a bank at a 40% slope, covered with grass + fodder legumes.

(3) To prevent the terrace above from becoming sterile after levelling work, the wall of the uphill bank can be knocked down one last time to correct its slope, and a good layer of topsoil spread on the terrace above.

Tied ridging perpendicular to the slope improves water storage under small rainstorms, but can lead to gullying or landslides under heavy storms. Only large ridges (H = width ³ 40 cm) permanently protected by creeping plants (e.g. sweet potato or forage pulses) and at intervals of under 5 metres, can break the force of runoff on slopes. Combined with hedges, they can quickly stabilize steep slopes (20 to 60%).

Contents - Previous - Next