Christian Valentin, ORSTOM, Bondy, France
Crusts are thin soil surface layers more compact and hard, when dry, than the material directly beneath. They hamper seedling emergence, reduce infiltration and favour runoff and erosion. Seal is generally the term given to a wet crust. Both crusts and seals are formed in the same way so no further distinction will be made in this paper. The term crust is preferred and is used from here on.
The first works mentioning soil crusts in West Africa mainly refer to dry areas: Mauritania (Audry and Rossetti 1962), Senegal (Aubert and Maignien 1948), Mali (Leprun 1978), Burkina Faso (Roose 1973), Niger (Boulet 1966), northern Nigeria (Sombroek and Zonneveld 1971), Tchad (Bocquier 1971) and Cameroon (Martin 1960). In these regions crusts affect most of the grazed and cropped land. Due to the drought combined with animal and human pressure, surface crusts have extended considerably in the last twenty years (Gavaud 1990), being at the same time a factor in, a result of, and indicator of desertification (Valentin and Casenave 1990). Studies in the wet savannah and in the forest zone are more recent. Here they are more strictly associated with agricultural activities. The experience gained in the Sahel (Collinet and Lafforgue 1978; Casenave and Valentin 1989) facilitated the identification of crust problems in cropped land in the wet savannah (in Cote d'Ivoire, Camara 1989, and in Togo Poss et al. 1990)) and in the rainforest (Collinet 1988a; Hartmann 1991).
The main objectives of this paper are:
Generally, two main types of crust are distinguished by their mode of formation (Chen et al. 1980): structural crusts and depositional crusts. The former develop in situ whereas the latter are formed of particles which have been transported from their original location. Detailed studies in West Africa (Casenave and Valentin 1989; Bresson and Valentin 1990; Valentin 1991) have led to a more comprehensive classification, which, though it was first developed for the arid and semi-arid zones, has been then applied satisfactorily in wetter climates in cropped land of the wet savannah zone (Valentin and Janeau 1990; Poss et al. 1990) and rainforest areas (Hartmann 1991).
In West Africa, two main types of structural crusts commonly occur depending on the texture of the top layer.
These crusts consist of a layer made of fine particles with rougher patches of partly broken down clods (Plate 7). These are marked by a generally higher surface and are more porous than the surrounding depressions where depositional crusts develop (Falayi and Bouma 1975; Valentin 1981; Levy et al. 1988; Pleuvret 1988; Valentin 1991). Slaking crusts form when soils contain enough clay (> 15-20%) to entrap and compress air during wetting so that aggregates break down. The process can also involve swelling and infilling. Since slaking crusts are formed primarily by wetting, they can develop even when the soil is protected from rainfall impact (Valentin and Ruiz Figueroa 1987).
Sieving crusts are formed of a layer of loose sand overlaying a thin layer of finer material (Plate 8). In the most developed form, the crust can consist of three well-sorted layers. The uppermost is composed of loose, coarse sand, the middle one consists of fine densely packed gains with vesicular pores, and the lower layer shows a higher content of fine particles with reduced porosity (Plate 9). This lower layer causes the low infiltrability (0-15 mm h-1) of the crust (Casenave and Valentin 1989 and 1992). The textural differentiation results from a sieving process. Water drop impact forms micro-craters, the walls of which present a clear, vertical sorting of particles (Valentin 1986a). Moreover, percolating water may enhance the downward movement of clay through the upper sandy layers. Clay particles can accumulate due to entrapped air within the underlying layers during infiltration (Collinet 1988b). This type of crusting mainly affects sandy and sandy-loam soils. Sieving crusts, also referred to as "filtration pavements" or "layered structural crusts", have been recognised in untilled soils in northern Niger (Valentin 1981), and in tilled soils in southern Togo (Poss et al. 1990) and the southern Côte d'Ivoire (Hartmann 1991).
A particular form of sieving crust is the pavement crust (Plate 10) where coarse rock fragments are embedded in the crust, the microstructure of which is very similar to the sieving crust with three layers. Vesicular porosity is very pronounced especially underneath the incorporated coarse fragments. Infiltrability is extremely poor (0-2 mm h-1) (Casenave and Valentin 1989 and 1992). These crusts range from the Sahara where this surface condition is named desert pavement or "reg", to the dry savannah zone.
Erosion crusts form barren patches of land (Plate 11), well-known by West African farmers or pastoralists, who give them various names: "zipelle" or "vuigo" in Central Burkina Faso, "white soil" or "glade" in More, "harde" in Foulani in northern Cameroon, and "naga" in Chad (Pias 1970).
These crusts are built up as a smooth and hard layer made of fine particles. Porosity is restricted to a few cracks and vesicles. Infiltrability is very poor (0-2 mm h-1) (Casenave and Valentin 1989 and 1992). Erosion crusts develop from both forms of structural crusts. They are formed from slaking structural crusts which have been smoothed and enriched in fine particles, and from sieving crusts where the loose sandy layers have been removed by overland flow or wind (Valentin 1985).
As observed in Mauritania (Barbey and Couté 1976), in Mali (Rietveld 1978), in Niger (Valentin 1981) and in Chad (Dulieu et al. 1977), erosion crusts can locally be strengthened by algae (Cyanophyceae). They can then form pedestal features where the surrounding weaker uncolonized erosion crust has been eroded (Casenave and Valentin 1989).
These algae are thought to be responsible for the low infiltrability of crusted sandy soils because they are hydrophobic. However, the physical binding effect of hyphae is of greater importance to infiltration. It has been observed that the hydrophobic tendency often disappears following rainfall: a few millimetres in the Sahelian zone (Rietveld 1978), a few tenths of millimetres in the rainforest zone.
Most often depositional crusts are built up with a combination of runoff and "still" depositional crusts.
These are characterized by alternate very thin layers contrasting in texture. These crusts can be up to a few centimetres thick, especially when they develop between two ridges or under furrow irrigation. Almost invariably, they overlie structural crusts with which they have sharp boundaries (Bresson and Boiffin 1990; Boiffin and Bresson 1987. Their infiltrability is much lower than that in the "mounds" of slaking crusts (Falayi and Bouma 1975; Valentin 1981; Levy et al. 1988; Pleuvret 1988; Valentin 1991). Runoff depositional crusts are formed by sediments deposited in overland flow.
These consist of densely packed and well-sorted particles, the size of which progressively increases with depth. The vertical particle size distribution, with coarser particles at the bottom and finer particles at the top, is the reverse of that observed in the sieving crusts. When dry, these crusts often break into curled-up plates. Still depositional crusts form in standing water where the larger particles sink rapidly to form the bottom layer, and the finer particles are deposited at the top. Infiltrability of still depositional crusts is poor (0-7 mm h-1) (Casenave and Valentin 1989 and 1992). The frequent, wide cracking common in these crusts favours seedling emergence. Algae may develop on these crusts (Plate 12).
Salty crusts are not extensive in West Africa and occur mainly near the sea or in the vicinity of lakes. They have some economic importance as a source of salt (Paradis 1980). In the Sahel, animals lick the ground in specific locations. For a few days, they stay near these so-called "salt licks" to take up their annual supply of minerals (Leprun 1978).
A range of minerals is found in such crusts. White saline crusts mainly consist of sodium and magnesium chlorates and sulphates. They are usually affected by wind erosion as they dry. Black saline crusts contain sodium carbonate combined with organic matter (Aubert 1976). Some yellow saline crusts may develop on very acid sulphate soils, like those studied by Le Brusq et al. (1987) in the Sine Saloum, Senegal. They are composed of aluminium, iron and magnesium sulphates. Sulphate crusts are generally the least porous to evaporating water and sodium chloride crusts are the most porous (Galizzi and Peinemann 1989).
White saline crusts are found in Mauritania (Audry and Rossetti 1962); in Gourma, Mali (Leprun 1978); in the vicinity of Lake Chad, namely in southern Niger (Gavaud 1975); in northern Cameroon (Brabant and Gavaud 1985) and in Chad (Pias 1970). Since most of these soils are clayey and sedimentary, these crusts are rather similar to sedimentation crusts, with polygonally shaped cracks. However, they can evolve in the dry season into a loose, powdery layer, as observed in northern Senegal (Mougenot 1983).
Crust formation markedly reduces the macroporosity of the soil surface layer. Since water infiltration into soil varies as the fourth power of the diameter of the pores, the impact can be very strong. The first study of the effects of crust formation on infiltration was presented in the USA by Duley (1939).
The double-ring infiltrometer was used in West Africa by Wilkinson and Aina (1976) to study the effect of surface crusting on water intake. Most authors consider this well-known method as unsatisfactory (Lafforgue and Naah 1976; Valentin 1981; Poss and Valentin 1983; Stroosnijder and Hoogmoed 1984; Valentin 1988). These workers note that:
Influence of surface conditions on hydrologic parameters on a dry soil in northern Burkina Faso (organic matter: clay 0.3%; clay: 7%; silt: 4%; sand: 89%) (adapted from Chevallier and Valentin 1984)
|Surface conditions||Residue cover (%)||Pr1 (mm)||F2 (mm h-1)||Rc3 (%)|
Tilled crusted patch
Tilled sandy micromound
1 Amount of rainfall necessary for runoff to occur on a dry soil
submitted to a simulated rainfall of 60 mm h-1.
2 Infiltration rate under saturated conditions.
3 Runoff coefficient measured for a cumulative rainfall depth of 360 mm, namely amount the whole amount falling during the rainy season.
4 Erosion crust
5 OM: 0.4%; clay: 4%; silt: 4%; sand: 92%.
These limitations encouraged researchers to design simulators which reproduce realistic patterns of rainfall in terms of depth, duration, intensity, drop size distribution and kinetic energy. A boom-type rain simulator, similar to that developed by Swanson (1965) was used in Côte d'Ivoire, Burkina Faso and Niger (Lafforgue and Naah 1976; Collinet and Valentin 1985; Collinet 1988a). It irrigates 200 m2 including two plots of 50 m2 each. A smaller rainfall simulator is described by Asseline and Valentin (1978). This readily-transportable equipment uses an oscillating nozzle which enables a change of rainfall intensity without stopping the experiment. This type of rainfall simulator has been intensively employed in the field in Côte d'Ivoire, Burkina Faso, Niger, Senegal, Togo, Cameroon and Congo.
Numerous studies have demonstrated the effects of crusts on infiltration in West Africa in all climatic zones:
Infiltration can be reduced even in very sandy soils due to the formation of a crust as observed in Niger by Valentin (1981) and in Mali by Hoogmoed and Stroosnijder (1984). In southern Togo, Poss et al. (1990) reported that under natural forest, water infiltrates into the soil whatever the amount and the intensity of the rain. By contrast, in cropped soil, infiltration can be as little as 20% of the heaviest rainstorm. These authors ascribed the low infiltrability of sandy soils in the region to the presence of a thin surface crust which develops at the onset of the rains.
Statistical analysis of data collected from 48 simulated rainfall plots in Burkina Faso showed that the presence of a crust and its type were the dominant factors influencing infiltration on bare soils (Albergel et al. 1986).
Infiltrability of crusted soils ranges from 0 to 25 mm h-1 depending upon the crust type, the cover and the soil moisture and the soil depth (Casenave and Valentin 1989). These infiltration rates are frequently exceeded by rainfall intensities under West African conditions so that runoff occurs under almost every rainstorm (Plate 13). Many studies (e.g. Valentin 1981; Albergel 1987; Casenave and Valentin 1992) have shown that the time to ponding and runoff is extremely reduced on crusted soils even when dry. In northern Burkina Faso, Chevallier and Valentin (1984) observed runoff after only 1 mm of rainfall at 60 mm h-1 on a dry sandy soil (Table 30). Consequently, heavy runoff is commonly observed on bare soils at the beginning of the rainy season.
The hydrological characteristics of surface crusts has been used by Albergel (1987) to predict water flow from small watersheds in Burkina Faso. This approach was extended to other countries (Côte d'Ivoire, Niger, Togo, Cameroon, Senegal and Congo). It is based upon a classification system of surface conditions, including surface crusts, micro-topography, vegetative cover and faunal activity (Casenave and Valentin 1989 and 1992). This method requires maps of specific surface conditions (Valentin 1986b).
Soil crusting can have some beneficial effects. In the northern fringe of the Sahel, even where rainfall water infiltrates totally, it is still insufficient to grow any crop. Here, runoff from crusted basins is collected naturally or artificially and concentrated over a smaller surface, or in a river bed where conditions are generally more favourable to infiltration. Without runoff from crusted soils, there would be many fewer temporary ponds and soils with water tables which are so crucial for cattle breeding and crop production in the arid and semi-arid regions.
The influence of crusts on water erosion is two-fold. On the one hand, crusts reduce infiltration and encourage runoff and erosion. On the other hand, they limit detachment because of their strength (Kowal 1974; Roose 1973; Remley and Bradford 1989) and further impoverishment in clay particles (Valentin 1981 and 1991). To make this influence clearer, one must consider the temporal and spatial scales of observations (Figure 34).
Where crusts develop, several processes interact to destroy aggregates (Boiffin 1984; Bresson and Valentin 1990). In the first stage, structural crusts are formed by in situ rearrangement of loosened particles without lateral movement. Wetting and subsequent slaking produce large amounts of dispersed clay and loose particles. The size distribution of these particles depends on the antecedent soil moisture and the characteristics of the rain (Loch 1989; Le Bissonnais 1990). These sediments are at risk of removal by overland flow. But at that stage, even though detachment may be high on the exposed clods, possibility of transport is small due to the surface roughness and high surface storage capacity. Moreover, soil in ponded areas is protected from further rainfall impacts.
In the following stage, microtopography formed by the remaining clods decreases. The smoother surface favours higher velocity runoff. Since the crust is already formed and compacted, detachment is less but the high carrying capacity of overland flow is capable of removing and transporting the particles detached earlier over great distances. Consequently erosion from the fields may be very heavy.
In the last stage, detachment is low and removal of particles from the fields can be limited even if runoff is large. At the field scale, crusts can protect soil from water erosion. At larger scale, heavy runoff produced by crusts upslope generally fosters more severe rill and gully erosion downslope (Planchon et al. 1987).
Soil crusts influence the eroding impact of wind which sorts particles, removes and deposits them. Loose particles of the sandy sieving crusts can be readily removed by wind leaving an erosion crust which is more resistant to wind erosion. Wind-drifted sands are entrapped by surrounding vegetation and these may evolve in turn into sieving crusts if the vegetation dies from drought or overgrazing (Valentin 1985). Barren spots capped with an erosion crust are therefore commonly surrounded by small mounds of windblown sand covered by grass vegetation (Aubert and Maignien 1949; Valentin 1985). On some recent dunes, almost devoid of fine particles, the sands form laminated crusts, known as aeolian crusts (Casenave and Valentin 1989).
Soil crusts reduce infiltration of water into soil, thus depleting the possible water storage and the chance of germination. Moreover, they form a mechanical obstacle for seedling emergence. Combined with the lack of water, mechanical impedance is often the limiting factor in stand establishment (Plate 14). Although crops commonly need replanting due to poor stand, especially in dry areas, there is less available data on the crusting effects on emergency than on infiltration and runoff. In West Africa, the most serious seedling emergence problems have been reported for sorghum (Charreau and Nicou 1971), pearl millet (Valentin 1981; Joshi 1987), upland rice (Ruiz-Figueroa 1983), and cowpea (Fapohunda 1986). Difficulties in stand establishment due to the presence of soil crusts have been also reported for cotton in USA (Bennett et al. 1964).
Crust strength and seedling emergence depend upon various factors:
As just mentioned, soil crusts may adversely affect stand establishment, infiltration and water storage. They also foster runoff and erosion, with losses of clay particles, organic matter and nutrients. Soil crusting is thus detrimental to the successful production of crops, and to the overall sustainabilty of the farming system. The loss of yield attributable to crust formation alone is however difficult to determine as it is one of several factors affecting crop yields.
Only statistical analysis can indicate some tendency. A significant relationship has been found between aggregate stability and cotton production Combeau (et al. 1961). Similarly, an index based also upon aggregate stability and porosity (Dabin 1962) was highly correlated with the yields of some tropical crops (rice, cocoa, banana). More recently, in Togo, the production of cotton was related to the soil surface conditions including the types of crust (Audebert and Blavet, personal communication).
Tentative correlation between susceptibility to soil crusting and certain soil groupings in West Africa. Note that these relationships are approximate and may be wrong in detail
|Soil group||Susceptibility to soil crusting|
Sodic, Gleyic, Haplic
General distribution of soil crusting and related aspects in West Africa
|Mean annual rainfall (mm)||>1600||1200-1600||800-1200||400-800||200-400 <200|
|Size of crusted patches||-||*||**||**||***||***|
|Severity of crusting||-||*||**||***||***||***|
|Severity of crsting||***||***||***||***||***||+|
* = low; ** = medium; *** = high; - = none or very little; + = generally not cultivated
As observed by several authors (Collinet 1988a; Mietton 1988; Casenave and Valentin 1989), no satisfactory correlation can be established at detailed map scales (up to 1:50 000 between soil type and its susceptibility to surface crusting. Therefore, detailed maps of specific surface conditions need to be prepared independently from existing soil maps (Valentin 1986b). At smaller scales, and particularly at continental scale, some broad relationships can be proposed, however (Table 31).
Many old soils of West Africa have been so rejuvenated by erosion that the nature of the parent material may influence soil surface conditions. Almost all the soils developed on the Precambrian Basement Complex (including gneiss, granite and schists) are prone to crusting when exposed to raindrops. Soils derived from schists, richer in silt, generally present more severe crusting problems (Poss and Valentin 1983) than those from gneiss or granite (Valentin et al. 1990). Moreover, the stability of kaolinitic soils to crust formation increases as weathering intensifies (Smith 1990).
Soils from basic rocks show higher resistance to crust formation than soils formed on granite (Smith, 1990). This has also been observed in West Africa, where soils on green basic rocks have a stable structure and associated high infiltrability (Casenave and Valentin 1989 and 1991). Soils derived from alluvial materials are very prone to crusting because particles have been already sorted (Valentin 1981; Ruiz Figueroa 1983; Valentin and Ruiz Figueroa 1987). Susceptibility to crusting of soils developed on aeolian sand (dunes) depends on the age of the deposits. Soil crusts develop mainly on old fixed dunes enriched in fine particles (clay and silt > 5%).
In typical self-mulching vertisols, crusts tend not to form as gilgai micro-relief and cracking favours infiltration at the beginning of the rainy season. Soil crusting can, however, affect vertisols containing sodium, or degraded by cultivation. Such a degradation sequence has been recently studied by Seyni-Boukar (1990) in northern Cameroon. Salic soils and solonetz are almost invariably affected by severe crusting.
Only 22 maps of surface conditions have been drawn in West Africa at 1:50 000, and one is being prepared at 1:200 000 scale in the region of Niamey. These maps are insufficient to prepare an overall map of severity of soil crusting in West Africa. Data collected during rainfall simulation experiments (Collinet 1988a; Casenave and Valentin 1989) provide, however, valuable information on the general distribution of soil crusting in the whole region (Table 32). Crusting can be roughly related to climatic zonation, and locally modified according to population density, farming system and soils conditions.
Under natural conditions, no serious crusting has been observed. This is mainly because the soil is protected from raindrop impact by the forest understorey and to the high organic matter content of the soils. Furthermore, the intense faunal activity tends to disturb the surface layer continuously (Collinet 1985).
When the forest is cleared and replaced by tree-plantations and cover crops, good surface conditions generally persist. Surface crusts only develop under certain limited conditions (Plate 15; Hartmann 1991). Conversely, on tilled and bare plots, marked crusting generally occurs causing severe water erosion (Roose 1973).
Under the wet savannah vegetation, soil crusting is a seasonal process resulting from complex interactions between rainfall, vegetation, bush-fires and faunal activity (Valentin et al. 1990). Good correlations between the soil properties and the surface conditions can be established. On higher slopes, Rhodic Ferralsols are usually covered by a sufficiently dense vegetation and soil fauna is active enough to prevent permanent crusting. Erosion soil crusts occur mainly on midslopes and lower ground on leached soils and physically degraded Xanthic Ferralsols (Poss and Valentin 1983; Planchon et al. 1987; Fritsch et al. 1990). These erosion crusts seal small patches of land between the grass-tufts.
When soil is cropped, severe crusting is observed, especially where clearing and tillage are performed with heavy equipment which compacts the soil and reduces faunal activity (Kooistra et al. 1990; Mitja et al. 1990). Even manual cultivation can lead to pronounced soil crusting, especially when cassava and corn are grown, as in southern Togo (Poss et al. 1990) or in Central Cote d'Ivoire (Camara 1989). Surface degradation can be hampered however by the presence of a gravelly surface layer. In West Africa where kaolinite prevails, gravels are generally not included in the crust (Collinet and Lafforgue 1979; Collinet and Valentin 1979; Casta et al. 1989).
In this climatic zone, the relationships between soil types and severity of crusting are clear under natural conditions. Patches of crusted soils are larger than in the wet savannah.
In this densely populated region, the natural environment has much suffered from clearance. Furthermore, mechanization has been used for a rather long time, especially for groundnut and cotton production. Because of the increase of population and the decline of rainfall farmers have extended their cropped areas and reduced the fallow period drastically. As a result, the nutrient reserves of some soils are seriously depleted so they cannot sustain vegetative cover any longer. Erosion crusts develop, hampering infiltration, initiating a vicious spiral evolution which has been termed "sahelization" (Plate 16), by Albergel and Valentin (1990). Typical Sahelian surface conditions may be now encountered in degraded zones in the Sudan.
Under this drier climate, no clear relation can be established between soil types, with inherited characters, and surface conditions which reflect present drier climatic conditions (Collinet and Lafforgue 1979). Regardless of human influence, the same type of crust may occur on very different soils whereas various crusts can develop on the same soil depending on local conditions such as relief, and vegetation. Earthworms generally do not survive in such a dry environment (Menaut et al. 1985) where soil crusts are mainly permanent features.
The old dune belt of this zone is cropped with millet and the lower land with sorghum. This region has suffered recent severe droughts. This climatic change has induced an extension of cropped areas with the clearing of marginal land such as the cuirassed plateaux. Serious degradation has occurred with the development of coarse pavements, generating heavy runoff and severe erosion, leading in some cases to desertification. Pastoralists have also been forced into these less favourable areas, so that large areas have been degraded by overgrazing.
In this climate, wind and water erosion interact to favour the formation of large patches of crusted land which are one of the characteristics of this region. As in the southern Sahel, crusts develop irrespective of soil type. Soil fauna is generally restricted to termites.
Apart from some limited areas where water is augmented by runoff from crusted hills (Ader Doutchi in Niger), agricultural activities are few and marginal. Most serious surface degradation is found near rivers (Niger, Senegal), permanent ponds or drilled water holes. On the old fixed dunes, trampling enhances wind erosion which in turn favours the formation of new dunes. On shallow soils on cuirassed plateaux, dramatic degradation can occur as they are readily overgrazed. In this case, the development of coarse pavements ruins the environment with no hope of foreseeable restoration (Valentin 1985).
This region is the most exposed to the encroachment of the Sahara which progresses in a dune front, as in Mauritania, or more insidiously as elsewhere by the coalescing of desertified pieces of land. The two main types of crust are erosion crusts and pavement crusts.
Soil crusts develop on soils of almost any texture. It is difficult to determine the texture which is most likely to crust because the detailed particle size distribution of sands is rarely taken into account. Organic matter contents and clay mineralogy can inhibit any attempt at generalization. However, results obtained in Côte d'Ivoire (Lafforgue and Naah 1976), Mali (Hoogmoed 1986), Niger (Hoogmoed 1986; Valentin 1986a), and Nigeria (Boers et al. 1988) tend to indicate that the texture most prone to crusting contains about 90% sand and 10% silt or clay. These field results were corroborated in the laboratory by Poesen (1986). With higher sand contents, the amount of fine particles is apparently not sufficient to clog the pores. As a result, the youngest dunes generally do not present crusting problems. Older dunes, gradually enriched in clay particles either by termite activity or by dust fall, are prone to severe crusting. Moberg et al. (1991) report that dustfall exceeds 600 kg ha-1 in northern Nigeria and contains 25% clay and 57% silt, the rest being very fine sand.
Studies on the influence of coarse fragments upon crusting give contradictory results depending on the climatic zone. As already mentioned, in the rainforest zone and in the wet savannah zone, where kaolinitic clay predominates, gravel formed from fragments of broken ironpans remains free on the surface of uncrusted topsoil (Plate 17). It plays the role of a mulch, limiting runoff and erosion (Collinet and Valentin 1979; Casta et al. 1989). Conversely, soils containing coarse fragments are severely crusted under dryer climates, except for the soils developed on basic rocks (Casenave and Valentin 1989).
In West Africa, crusting problems increase exponentially with increasing clay content up to about 18 % clay where the clay consists of 2:1 clay minerals (Nicou and Charreau, 1980). This effect is, however, influenced by the organic matter content. In Niger, Valentin (1991) described severe crusting on a soil containing 34% clay (mostly montmorillonite) and only 0.7% organic matter. Collinet (1988b) also observed that smectite and illite favour crusting while kaolinite makes the soils more stable. Negatively charged montmorillonite can neutralize the positively charged edges of kaolinite. The presence of small amounts of smectite and/or micaceous minerals in the soil can therefore drastically reduce their aggregate stability (Miller 1987; Stern et al. 1991).
Numerous studies have shown that as organic matter content increases, most physical properties are improved. The positive effects of organic carbon on aggregate stability and resistance to crusting have been clearly shown in the Central African Republic (Quantin and Combeau 1962), Senegal (Charreau and Nicou 1971), Nigeria (De Vleeschauwer et al. 1978) and in Côte d'Ivoire (Valentin and Janeau 1989).
A recent example from southern Togo (Poss et al. 1990) illustrates the beneficial role of organic matter in protecting soil from structural degradation. These authors observed that no runoff occurs whatever the intensity of rainfall under natural forest, even through the understorey and the litter have been removed. The highest organic matter content is in the 0-4 cm layer (4.2% OM), so aggregates are much more resistant than in a cropped field (0.7 per cent OM) and in a two-year old fallow (0.9% OM) where heavy runoff is observed.
Hydrophoby of organic compounds reduces the rapid wetting of dry aggregates, hence the shattering process and the formation of slaking structural crust (Le Bissonais 1988). The hydrophoby increases in the dry periods (Sebillotte 1968; Boiffin 1976). Aggregate stability is therefore subject to seasonal variation and reaches a maximum at the end of the dry season (Quantin and Combeau 1962).
Carbon content should not be considered separately from the texture. Minor increases of carbon content may have a more beneficial effect upon the structural stability of sandy soils than a higher increase in finer-textured soils. Considering the ratio:
S = Organic matter content (%) x 100
Clay (%) + Silt (%)
for numerous savannah soils of West Africa, Pieri (1989) distinguished the following critical values of S:
S < 5 severe physical degradation
5 < S < 7 high hazards of physical degradation
7 < S < 9 low hazards of physical degradation
9 < S no physical degradation
Not all organic compounds in soils are favourable to soil structure. Close relationships exist between aggregate stability and humic substances of high molecular weight (Piccolo and Mbagwu 1990). Small quantities of such compounds can stimulate the production of dispersible clay (Reid and Goss 1982).
Chemical fertility may interact with physical properties. When nutrients are exhausted, vegetative cover is weakened and with this the protection of the soil surface from raindrop impact.
The nature of the exchangeable cations can also influence crust susceptibility. Aggregate stability is positively correlated to Ca content and Ca/Mg ratio, and negatively to Na content (Kijne and Bishay 1974). Magnesium can also foster the dispersion of fine particles. West African soils are more prone to crusting when Mg/CEC is greater than 50% (Collinet 1988a). Similarly, Keren (1990) observed that the infiltration rate of two montmorillonitic Na-Mg saturated soils decreased more than that of Na-Ca soils.
High amounts of iron are generally associated with greater resistance to slaking under rainfall (Farres 1987). Very stable aggregates can therefore be found despite low organic matter contents, as shown in southeastern Nigeria by Obi et al. (1989). Chroma values from the Munsell chart seem to be an interesting indicator of susceptibility to crusting (Valentin and Janeau 1990) suggesting that haematite has a more pronounced stabilizing effect that goethite.
Earthworms promote the formation of relatively stable soil aggregates (Shipitalo and Protz 1988; Blanchard 1990) and they perforate already-formed surface crusts. Earthworm casts at the soil surface are a good indicator of a permeable soil (Casenave and Valentin, 1989 and 1992). The role of termites is less clear. Foraging termites like Trinervitermes harvest grass vegetation around their nests, leaving the soil bare and severely crusted (Janeau and Valentin 1987). These termites predominate during the first ten years of the fallow period, as shown in wet savannah in Côte d'Ivoire (Mitja et al. 1990). Later fungus-growing termites like Cubitermes develop which do not have much impact on crusting. Humivorous species like Macrotermes prevail after a fallow period of about 40 years. Erosion of their cathedral-shaped nests may cause crusting of a limited area around the termite-mounds. But these termites also build tunnels at the soil surface as protection against light, thereby perforating the surface and reducing crusting. The presence of such tunnels at the soil surface indicates high infiltrability (Casenave and Valentin 1989 and 1992).
As noted above the clearance of marginal grazing land causes severe surface degradation. The means employed to clear the land has a pronounced effect. Hand-felling followed by moderate burning of the residues does not induce surface alteration, especially in the rainforest area (Wilkinson and Aina 1976; Collinet 1988a; De Rouw 1991). Clearing by mechanical means (bulldozers and large tractors) tends to compact the soil (Lal 1987) and to scrape away most of the fertile top layer. Thus mechanical clearance generally hastens the process of soil surface deterioration and limits the possibility of restoration after cropping (Wilkinson and Aina 1976; Mitja 1990).
Tillage ranges from manual operations to land preparation with heavy machinery. Tillage properly executed, appropriate to the conditions (soil texture, soil moisture, energy and time) can destroy the existing crusts but at the same time it offers the best conditions for the formation of new ones.
Firstly, it is difficult to obtain optimal clod size distribution during the seedbed preparation. If the clods are large, crusting is limited but the contact between the soil and the seeds does not allow proper germination. If they are too small, they are readily slaked (Johnson et al. 1979). Secondly, repeated vehicular traffic by powered equipment (seedbed preparation, seeding, weeding and harvesting) generally compacts the soil, favouring the formation of a plough pan. Infiltration is then reduced and waterlogging of the ploughed layer may accelerate the formation of a surface crust (Boiffin 1984). Large clods may slake readily when the soil is very wet and unstable (Valentin and Ruiz-Figueroa 1987). Thirdly, tillage operations and cropping cause a decline in the meso-fauna because of mechanical effects, decreased organic-matter contents, negligible return of crop residues which limit nutrient supplies, and higher temperatures (Wilkinson and Aina 1976). This leads to low aggregate stability and the formation of crusts (Kooistra et al. 1990).
The impact of agriculture on surface deterioration differs with the cropping system depending on the climatic zone. Three main systems can be distinguished:
Traditional systems include many that are not stagnant but have evolved or are evolving gradually. They can integrate new crops and new technologies. Their main characteristic is shifting cultivation based upon a long fallow period. Such systems survive in sparsely populated regions of West Africa.
Organic matter contents drop dramatically under cultivation. For example, Aina (1979) reported that the aggregate stability of two cultivated loam sands from Ife, Nigeria, ranged from one-fifth to one-third of soils under secondary (15 to 25 years old) fallow, which contained about four times more organic matter than the cultivated soils (0.8%). Using statistical analysis Feller et al. (in press) show the decrease in carbon content is greater for fine-textured soils than for sandy soils. This loss of organic matter not only leads to less stable clods, but also to a significant increase in dispersible clay. This might be favoured by the increased ration of fulvic to humic acids (Oades 1984).
Fallowing restores soil organic matter contents, thus improving physical properties. This process requires at least ten years in the savannah zone (Charreau and Nicou 1971; Valentin and Janeau 1989), probably more in the rainforest area. The restoration of physical properties may be hampered where chemical fertility has been depleted too much by cropping or where the soil has been compacted (Mitja 1990; Seyni-Boukar 1990).
These systems have developed where density of population has increased land pressure. Fallowing is shortened or abandoned. Two contrasting situations exist in West Africa. In the most favourable, in terms of surface protection, farmers have a long history of land shortage (for instance, in the Korhogo region in northern Côte d'Ivoire, or in southeastern Nigeria). They have developed conservative technologies using multiple-species cropping and mulching. These systems may gradually integrate low-inputs (fertilizers and herbicides).
In other areas the high population is more recent. Fallow is shortened but without the application of alternative technology. The situation is even worse under monoculture or double cropping (groundnut-millet, cassava-maize, for instance). In such systems, chemical fertility is extremely depleted, surface crusting, erosion and wood infestation are major problems.
Three aspects of the modern farming systems in West Africa need to be discussed in terms of soil crusting: plantation systems, mostly in the coastal zone, irrigation systems, mainly in the dry zones, and the intensive use of fertilizers.
The farmers or the companies which manage commercial plantations are generally aware of soil degradation hazards. Cover crops or abundant leaf litter prevent the soil under perennial crops (oil palm, rubber, coffee or cocoa) from direct raindrop impact. The main risk of degradation occurs during clearing the forest or when replanting (Hartmann 1991).
More serious crusting problems occur under irrigation. Intensity and kinetic energy produced by sprinkle irrigation can be too high and cause severe crusting (Valentin and Ruiz-Figueroa 1987). Avoiding irrigation when the soils are dry reduces slaking and favours seedling emergence through weaker crusts (Fapohunda 1986). The kind of water used for irrigation in West Africa is rarely suitable for the maintenance of a surface structure. Irrigated arid soils, which may contain originally a high percentage of exchange sodium, are all the more susceptible to crusting when the electrolyte concentration of the irrigation water is high (Valet 1990).
The impact of fertilizers on surface structure is not much documented and the results of the few studies available are contradictory. Fertilizers are thought to protect soil surface indirectly by enhancing vegetation cover (Sajjapongse 1991). But Ma et al. (1991) showed experimentally that the use of NH4-fertilizer enhances swelling and dispersion of loess soils. In the southern Ivory Coast, Hartmann (1991) assumes that the repeated use of KCl-fertilizer may be partly responsible for crust formation in oil-palm plantations. Aina (1979), however, did not observe lower aggregate stability in fertilized soil than in unfertilized soil after ten years of continuous cropping in southern Nigeria.
Farmers' knowledge has developed into practices which tend to prevent or limit surface crusting or to restore degraded topsoil.
From the rainforest zone (De Rouw 1991) to the Sahelian region (Valentin 1981), dibbling is a well-known technique. It consists of seeding several seeds of the same crop or of different crops in the same manually opened hole. This practice aims at not only limiting some risks of germination failure but also at allowing the combined strength of the seeds to break up the possible surface crust, generally a depositional crust weaker than other types of crust (Plate 18).
Some tillage practices limit the formation of widespread erosion crust. Mounds and ridging in the contour maintain a high storage capacity and reduce the velocity of overland flow. They favour the formation of depositional crusts in the furrows (Figure 35). But weaker and more permeable crusts (structural crusts) cap the ridges or the mounds where crops are grown.
When weeding, farmers not only intend to remove weeds but also to destroy the soil crusts. Several authors highlighted that the beneficial effect of crust destruction to prevent runoff is very sort in time since a new structural crust generally forms after 20-30 mm of rain (Valentin 1981; Hoogmoed and Strosjnider 1984). Besides the fact that even limited gain in water intake may be crucial in the arid zone, one must consider that the beneficial effect of crust destruction can still be experienced after 100 mm of rainfall (Serpantié 1990; Lamachère 1991). This is approximately the amount of rainfall necessary to form a crust similar to that prior to the weeding (Figure 36).
Many traditional practices to limit soil and water losses also reduce surface crusting. Some others take benefit of runon from crusted watersheds to increase available water for the crops (runoff farming and water harvesting). Presenting all these technologies would be beyond the scope of this paper. Excellent review papers have been recently published on this subject (Hudson 1987; Reij et al. 1988). In addition to terracing in mountainous regions (northern Cameroon, Central Togo), some other practices on low slope gradient must be mentioned. In particular, stone contour lines (Plate 19), or contour hedgerow (Plate 20) are used in Niger and in Burkina Faso not to stop overland flow but to reduce its velocity. Such permeable structures diminish the hazards of formation of large patches of barren erosion crusts and favour depositional crusts, less harmful for seedling emergence and water intake.
Besides the already-mentioned mixed cropping, agroforestry systems in which trees and shrubs are grown together with other crops are part of many traditional systems including some in the semi-arid zone. This is illustrated, for example, by the important role played by the soil-improving tree Acacia albida in semi-arid Africa (Baumer 1987; Young 1989). The traditional practice of the Serere people in Senegal to reduce surface crust formation and wind erosion includes the preservation of trees in their fields (Acacia albida, Borassus aethiopium), enclosing them with these trees, and allowing the cattle of the herdsmen to graze on the fields during the dry season (Aubert and Maignien 1948).
Mulching is a common practice in some cropping systems (yam mounds are often capped with residues in the wet savannah zone - Plate 21). Less common is the practice used to restore productivity of barren patches of land, capped with an erosion crust, in semi-arid regions. Farmers here accumulate residues (mainly branches) on these crusted spots (Plate 22). Wind-drifted sands, leaves and seeds accumulate on the branches (Plate 23) and the debris attracts termites which perforate the crust, thus enhancing porosity (Casenave and Valentin 1989). Experiments conducted near Niamey (Chase and Boudouresque 1989) demonstrated that branch mulches encourage water to move deeper into the soil than tillage with a traditional hoe.
A traditional technique to rehabilitate barren crusted soils, often called "zai" in Burkina Faso, consists of digging holes 5-15 cm deep and 10-30 cm across (Plate 24). Some manure and debris are mixed with earth and put into the hole. Due to manure application, termite activity is enhanced and soil structure improved. Millet and sorghum yields are increased (Reij et al. 1988).
Cattle droppings tend to limit surface crusting in sandy soils (Valentin 1985) so long as the soil is not trampled heavily. As a result, integrating livestock with crop production can be considered as a means of preserving surface structure and enhancing fertility. In the semi-arid zones farmers allowed nomadic cattle to graze on their fields, benefiting thus from the manure. However, this one common system tends to break down. Although such practices are still used in some parts of the savannah zone, they have limits since 15 ha of land are needed to produce sufficient manure for one cropped hectare in semi-arid zones. This means that only 6 to 7% of the land could be cultivated, which is far from realistic (Breman, personal communication).
Since a large array of intrinsic and external factors are involved in soil crusting, simple and unique rules cannot be put forward. In particular, technical options should be adapted to environmental and human conditions. In addition, land management must be regarded as a whole. Specific practices to combat soil crusting cannot be considered in isolation from those adopted to meet other constraints. Furthermore, most of these practices do not greatly differ from those usually proposed for soil and water conservation or for enhancing the organic carbon content of the topsoil. Rather than attempting a thorough review of possible improved management practices, it seems more sensible to comment on some of them, using the ecological zones as a framework.
Improved fallow and fodder crops
In densely populated regions, the fallow period tends to become too short for soil physical properties to recover. Under such conditions, it can be profitable to improve the fallow vegetation. High-biomass crops not only act as a protective cover but also favour soil structure through the effects of roots and organic matter. In this respect, Damour and Kilian (1967) found in Malagazy that better aggregate stability was obtained with grasses than with plants with taproots and rhizomes. Furthermore, experiments conducted on fodder crops in Central African Republic indicate that under Pennisetum purpureum or Panicum maximum the aggregate stability recovers to the same level as under natural savannah within four years. Conversely, no improvement was observed with Stylosanthes gracilis and Pueraria javanica even after six years (Morel and Quantin 1964; 1972).
In Western Nigeria, Aina (1979) found that soil deterioration due to tillage was in decreasing order: plough-disk-harrow, plough and no-tillage. However, Casta et al. (1989) observed that ploughing lifts stones to the surface, thus inhibiting crust formation on the gravelly kaolinitic soils of the central Côte d'Ivoire. Under no-tillage systems, the soil is only disturbed during sowing. Specific hand tools for direct-drilling through residues have been developed. They use a coulter to open a narrow slit and press wheels to close the slit. The main limitation of this practice is the need of herbicides. These are expensive and very sensitive to the correct time and conditions of application. No-tillage may restrict rooting, so it must be recognised that no-tillage alone without mulching does not prevent surface crusting (Kooistra et al. 1990).
The availability of residues in the wet tropics facilitates the common use of residue mulches. These protect soil from raindrop impact, slowly decompose at the surface and raise the organic carbon content of the topsoil, promoting the formation of water-stable aggregates. The well-established role of residue mulching in enhancing soil structure, soil conservation, soil temperature, organic matter status, faunal activity, is however, moderated by the detrimental effects such as pest enhancement (Valentin and Roose 1980).
In dry areas of West Africa, weeding has a two-fold effect on soil crusting. On the one hand, it breaks the crusts and is thus in part a remedial measure. On the other, by removing the weeds, soil cover is reduced, accelerating the forming of new crusts. Arid soils, however, need to be tilled for crop production since they tend to harden (Charreau and Nicou 1971). A major choice is between inverting and non-inverting tillage. Non-inverting tillage is promoted because it is faster and less energy consuming. Its much poorer weed control, however, may greatly reduce the net effect. Tillage practices which enhance soil roughness must also be considered. The water infiltration rate in tied-ridge plots is lower than that of flat planting or open ridging, due to the formation of sedimentation crusts. With tied-ridging, however, rainwater is retained on site by the ties, in contrast with open ridging or flat planting, where it is lost as runoff, as illustrated by the results of Hulugalle (1990) in Burkina Faso. In Mali Stroosnijder and Hoogmoed (1984) showed that tied-ridges could prevent runoff and gave a surface storage of 20-30 mm of rainfall. However, as mentioned by Collinet and Valentin (1985), ridges (Plate 25), tied-ridges and tied-mounds may pond water and then collapse, resulting in soil and water losses greater than for flat planting. Hudson (1987) reported that the inconsistency and unreliability of results obtained with tied-ridging prevent it becoming more widely adopted.
In dry regions, the lack of water and nutrients limits vegetative production, and the amount of residue available to be returned to the soil. Furthermore, crop residues have an economic value, for example as animal fodder, straw for roofing, so that the use of mulch is often impracticable. It is also very difficult to till and sow when crop residue is left on the soil surface. Incorporated straw residues in sandy soils tend to decrease soil organic matter contents, as shown in Senegal (Geye and Ganry 1978). This must be due to increased consumption of carbon by micro-organisms (Pieri 1989).
Manure and fertilizer
Addition of organic material containing constituents of high molecular weight like cattle slurry is a useful management practice to improve aggregate stability. The favourable effect is enhanced with nitrogen fertilizer added to loamy soils (Pieri 1989). This illustrates the positive interactions which generally occur where several techniques are used simultaneously. In the Sahelian zone, low levels of phosphorus limit the optimal use of available water (Breman and Uithol 1987), restricting vegetation production, hence surface protection and organic matter content. A moderate use of fertilizers may therefore help to control the development of soil crusts. Conversely, intensive use is suspected to enhance it (Ma et al. 1991; Hartmann 1991).
Additional investigation of soil crusting in West Africa is needed. Three main approaches can be identified:
Priorities should be established by politicians, sponsors and scientists, and also by the farming communities involved.
It is difficult to assess the economic and environmental impact of soil crusting. In particular, little research has been done on the quantification of crusting effects on crop emergence.
The hydrologic properties of surface seals are now well documented, but more research is needed on the mechanical properties of soil crusts.
Apart from a few studies (for example, Casenave and Valentin 1989), research on the variations in space and time of soil crusts remains scarce. Most experiments are conducted on microplots in the laboratory rather than in the field over a long term. In this respect, remote sensing provides an opportunity to assess the spatial and temporal degradation of the land surface. Archives of existing satellite or aerial photographs should be used for retrospective studies on the rate and extension of crusted soils in relation to changes in agricultural systems. The objectives would be not merely to describe changes but to understand them fully so that predictive models can be used to explore possible future scenarios using climatic, edaphic and/or socio-economical hypotheses. This approach could help to delineate the critical limits of climatic change or population density beyond which land is irretrievably damaged.
A more detailed inventory of indigenous techniques used to prevent soil crusting or to rehabilitate deteriorated land should enable evaluation of the applicability of such local knowledge to other socio-economic and environmental conditions.
Since the beneficial effect on soil structure of surface cover is well established, but difficult to manage, especially in the drier zones, research should focus on the enhancement of aggregate stability. In this respect, two main lines of research should be followed: the management of organic matter status and the possible use of conditioners. In particular, the effects of improved fallow and crop rotation are worthy of more attention. For example, Valet (1990) recently showed that regeneration of surface structure could be achieved to some extent in degraded irrigated soils of the river Niger terraces by the use of forage crops (Echinochloa stagnina, Panicum maximum).
More than thirty years ago, Riquier (1955) considered that an efficient and inexpensive soil conditioner was essential. He suggested that research should be carried out with this end in view. He considered the task well worth the trouble, for it would remove the main obstacle checking the productivity of tropical soils and would preserve their fertility by protecting them against both wind and water erosion. Since then, soil conditioners, including krilium, gypsum and phosphogypsum have been proposed repeatedly as means of enhancing aggregate stability. Their use has generally given disappointing results because tested conditioners were effective for only a limited time or were not economically viable. Such frustrating results should not preclude further research since technological and economical conditions may evolve rapidly and reduce the cost of production and transport of conditioners.
In dry regions, surface crusting is not invariably detrimental. As already mentioned, runoff farming and water harvesting techniques benefit from the overland flow produced on crusted areas. Although such practices are popular among extension officers and widely recommended, they still lack a scientific basis. Empirical procedures cannot generally be applied to other environmental conditions. More research is needed. Virtually nothing is known in West Africa on the long-term effects of these technologies on crop production, on their influence on nutrient leaching, on their effects on water-table levels, and on water fluxes at the watershed scale. Finally, more research is needed on their acceptability to farming communities.
In conclusion, a few points can be highlighted: