Mapping and characterizing inland valley agro-ecosystems: The case of West Africa

Abstract

Inland valley agro-ecosystems are ecosystems with an agricultural production component which occur in the upper reaches of river systems that comprise lands with impeded drainage. In West Africa, the latter lands (i.e. the valley bottoms and their hydromorphic fringes) are estimated to occupy about 22-52 million hectares of land. In spite of their potential, these lands are underutilized and with limited success only. Some 10 to 25% of their total area is being cultivated, mainly by smallholders who grow rice at yields of around 1 ton/ha (paddy). Constraints to cultivation include extensive weed growth, lack of appropriate water management technologies, labour shortage, prevalence of water-related diseases, and unfavourable socio-economic conditions.

In the framework of a concerted regional approach to study these constraints, a multiple-scale method for the characterization of these agro-ecosystems was developed by partners in the West African Inland Valley Consortium. Methodological, organizational and institutional aspects of the approach are discussed and recommendations formulated for possible application or adaptation of the method in comparable environments in eastern and southern Africa.

Keywords: Inland valleys, agro-ecological characterization, agro-ecosystems, multiple scales, West Africa, common methodology, regional approaches.

Background and introduction

West Africa's inland valleys can be compared with the dambos, mbugas, matoros and vleis occurring in eastern and southern Africa. They are all a category of wetlands in the agro-ecological sense: land subject to excessive wetness to the extent that the wet conditions influence the choice of possible land uses. Wetlands comprise soils with impeded drainage which can be due either to flooding or to periodic high groundwater levels. Other broad categories of wetlands are coastal plains, inland basins and river floodplains. Table 1 shows the extent of these categories of wetlands, including inland valleys, in sub-Saharan Africa. It should be noted that the acreage of inland valleys was obtained through a rough and conservative extrapolation of data for West Africa and hence, they should be treated with care. Another, and apparently less conservative, estimate is by Raunet (1982, 1985) who arrived at a total of some 1 300 000 km². To a large extent, the wide range in area estimates must be due to differences in definitions and in map accuracy. For the same reasons, our own area estimates for West Africa, based on secondary sources, were in the range of 22 to 52 million ha of land, or some 7 to 16% of the total area (Windmeijer and Andriesse, 1993).

Wim Andriesse 
DLO Winand Staring Centre for Integrated Land, Soil and Water Research 
Wageningen, The Netherlands 

TABLE 1
Categories and areas of wetlands in tropical sub-Saharan Africa (total area 12.2 million km²; Source: Andriesse, 1986)

In spite of their potential, the inland valleys are only marginally used in West Africa and with limited success: only 10 to 25% of their total area is in cultivation (mainly rice), and yields are generally low (0.8 - 1.2 ton paddy/ha). Among the constraints to cultivation are extensive weed growth, lack of

appropriate water management technologies, labour shortage, prevalence of water-related diseases, and unfavourable socio-economic conditions.

The categories of coastal wetlands, inland basins and river floodplains form large and distinct physiographic units in the landscape. Inland valleys however, are small and relatively inconspicuous. Inland valleys also are varied: internally, where they comprise such different elements as valley bottoms, slopes and crests, as well as externally where they have a characteristically high spatial variability due to differences in parent material, physiography and climate and, as a resultant thereof, hydrology and soils. Inland valleys, therefore, do have very high variability in actual and potential uses.

Given this extreme variability, �nd because of our primary interest in their sustainable agricultural development, characterization of inland valley agro-ecosystems needs to be holistic in nature. N�t soils alone, topography, hydrology or land use are to be described, but their interrelations need to be characterized. Also, there is a need for characterization at different scales, implying description at different levels of detail, in order to suit different purposes and user groups, and to deal adequately with the inherent differences in variation of individual valley characteristics at different scales.

In this paper a multiple-scale system for the characterization and mapping of inland valley agro-ecosystems is described, providing increasing amount of detail at four consecutive levels. From high to low these are: macro, reconnaissance, semi-detailed, and detailed level. This system allows: (i) systematic description of inland valley agro-ecosystems, (ii) identification of constraints to sustainable agricultural development and use, (iii) targeting and implementation of research and (iv) extrapolation of research results and transfer of newly developed technologies to other areas with similar agro-ecological conditions.

The multi-scale characterization system has been developed in the framework of research activities by the West African Inland Valley Consortium (IVC; WARDA, 1993), for adoption as a common method by member states and organizations. It is based largely on work of the `Wageningen Group' in the IVC, being the DLO Winand Staring Centre for Integrated Land, Soil and Water Research, and the Department of Agronomy of the Wageningen Agricultural University, both in The Netherlands. An attempt is being made here however, to incorporate methods and concepts by other consortium partners, in particular those by the International Institute of Tropical Agriculture (IITA) in Nigeria, and the Centre for International Cooperation and Agricultural Research (CIRAD) of France.

In the chapters below, first a number of conceptual issues will be discussed. Next, methods of characterization at each of the levels indicated above are highlighted and thereafter some results of actual applications are given. Emphasis in these examples is on the semi-detailed characterization. Lastly, a number of pending research issues relevant to the method and its possible application in eastern and southern Africa is discussed.

Conceptual framework

Agro-ecosystems

Agro-ecosystems are ecosystems with an agricultural component in their primary production compartment. The concept is scale-neutral and may be used in reference to complete valley systems, individual valleys or elements thereof. Within the major agro-ecological zones of West Africa, inland valley agro-ecosystems are highly complex and heterogeneous, due to the high spatial variability of both physical and biological factors along the toposequences, and because of the temporal and spatial changes in cropping systems involving crop choice, crop rotations, intercropping and crop husbandry. The cropping systems as actually applied by farmers are also determined by endogenous factors such as household composition, assets, and labour availability.

Agro-ecological characterization

Agro-ecological characterization is the comprehensive description of agro-ecosystems on the basis of physical (climate, lithology, landform, soils and hydrology) and biotic parameters (vegetation and land use; Bunting, 1987). Land use (primary and secundary production) is described inclusive of its socio-economic identifiers (labour, capital input and management). The degree of detail of the information gathered in agro-ecological characterization is strongly related to the scale of characterization.

Multi-scale approach

A multi-scale approach has been developed because data collection, interpretation and use always take place at specific levels depending on the purpose (Andriesse et al, 1994; Izac et al, 1990). Scaling down to lower levels of characterization (i.e. desaggregating), implies greater detail of increasingly dynamic parameters, while at the same time certain macro-level parameters remain more or less static (for example climate and parent material at detailed level). On the other hand, while scaling up (aggregating), details distinguished for variables at low levels of characterization (for example specific crop rotations or nutrient contents in the soil) are generalized or even completely disregarded at higher levels.

Our multi-scale characterization system comprises four levels with their respective scales: macro level (scales between 1 : 1 000 000 and 1 : 5 000 000), reconnaissance level (1 : 100 000 - 1 : 250 000), semi-detailed level (1 : 25 000 - 1 : 50 000) and detailed level (1 : 5 000 - 10 000). They are presented schematically in Table 2 and they are described in more detail in the chapter on methods below.

Table 2
Levels and scales in agro-ecological characterization of inland valleys, units of analysis, objectives and main information sources (Source: Andriesse et al., 1994).

Inland valleys

Inland valleys are defined as the upper reaches of river systems, inland in respect of the main rivers and tributaries, in which river alluvial sedimentation processes are absent or of minor importance: they have only a minor floodplain and levee system (Windmeijer and Andriesse, 1993). Inland valleys comprise the toposequence, or continuum, of valley bottoms and minor floodplains which may be submerged for part of the year, their hydromorphic fringes, and contiguous upland slopes and crests, extending over an area that contributes runoff and seepage to the valley bottom (Figure 1). Longitudinally, inland valleys may be continuous and smooth, or interrupted and stepped, depending on the underlying rock formations. Individual inland valleys and their elements (i.e. valley bottoms, hydromorphic fringes, slopes and crests) are part of differentiated valley systems of first and higher-order streams which form the main unit of analysis.

Land cover and land use

Land cover is defined as the vegetation (natural or planted) or the human constructions covering the surface of the land, including water bodies and bare soil. Land use are the human activities that are directly related to land, making use of its resources or having an impact on it. Land cover is the reflection of land use and it may change with time. On top of the highly variable physical conditions in inland valleys (e.g. soils and hydrology), land cover and land use involve extremely dynamic variables, spatially (land elements, rotations, intercropping) as well as temporally (seasonal changes in crop rotations, applied inputs, cultivation practices, labour).

Methods

Macro characterization

Macro characterization, involving scales between 1 : 1 000 000 and 1 : 5 000 000, aims at the distinction of major agro-ecological zones, based on secondary information sources. Following the approach developed by FAO (1978), the agro-ecological zones are defined on the basis of the length of the growing period, i.e. the period in which the precipitation is higher than the half of the potential evapotranspiration plus an additional period to evaporate a amount of 100 mm water stored in the soil. For West Africa a further combination was made with existing vegetation-ecological zones (Aubreville, 1949). Additional data on lithology and landforms were collected from national and regional inventories. Based on this information the agro-ecological zones are subdivided into agro-ecological units, each with their specific physical characteristics. The description of the units includes broad estimates of the inland-valley density and general information on the prevalent production limitations.

Reconnaissance characterization

Characterization at reconnaissance level (scale 1 : 100 000 to 1 : 250 000) involves the further subdivision of agro-ecological units into sub-units. As reconnaissance-level characterizations are generally performed at country level, national sources of secondary information on landform, lithology, soils, climate, hydrology, major land use, morphology, population density and socio-economics are consulted. Information on the distribution and extent of inland valleys in (parts of) the country is obtained from satellite imagery, if available. Additional information on main land use is collected by means of rapid rural inventories. Within relevant agro-ecological sub-units, key areas are selected for further, more-detailed, characterization. By means of the concept of key areas, the characterization can be implemented in confined and representative areas and, thus, in a restricted period of time. The size of the key areas has been set at about 2 500 km², thus covering, at least in the West African setting, a number of valley systems (2-4 approximately) in each sub-unit.

In West Africa, IITA in particular has been instrumental in developing remote sensing technology linked with GIS-databases beyond collection of existing data and full field inventories (Thenkabail and Nolte, 1995). LANDSAT Thematic Mapper (TM) and SPOT high resolution imagery are used and processing is through ERDAS (Earth Resources Digital Analysis System). Satellite imagery being almost scale-independent, IITA's characterization methodology has the advantage of allowing to gradually merge from one scale into the next. Both wet- and dry-season near-real-time images are used for delimitation of land cover and and land use types, and for the analysis of occurrence and distribution of inland valleys. During ground-truthing a global position system (GPS) is used for geo-referencing of the data. Images, classifications, and inferred as well as field data are stored and processed in a digital database. This further facilitates statistical analysis.

Semi-detailed characterization

In semi-detailed characterization (scale 1 : 25 000 to 1 : 50 000) most data and information on the physical environment and on land cover and land use in the inland valley systems is collected through field surveys, by means of the Integrated Transect Method (Van Duivenbooden and Windmeijer, 1995; Van Duivenbooden et al, 1996). The inventory starts with the interpretation of recent aerial photographs of the key area. The airphotos are preferably at a scale of 1 : 25 000. Emphasis in the interpretation is on the shape and distribution of the inland valleys, the extent of the watersheds, categories of land cover, and on infrastructural features. The interpretation is used to select a number of representative valleys in each of the key areas. In these valleys some 4 to 5 transects are selected for field description. The transects should cut across different segments of the valleys, from one top of the crest to the other.

Along with the various land elements encountered in the transect (i.e. crests, slopes, hydromorphic fringes, valley bottoms), physical characteristics including slope gradient, soils and hydrology are described and the soils are sampled for analysis in the laboratory.

Land cover and use are described for a strip of land of 200 to 400 m width along the transect line. The actual width of this strip depends on the purpose of the survey, the scale of mapping, and on the heterogeneity of the land cover. The average plot size may be taken as the guiding criterion. For practical reasons however, the strip-width may have to be adjusted to suit local conditions, such as the accessibility of the vegetation cover. The minimum size of individual land cover/use units to be described, is set at 250 m². Crops, fallow and natural vegetation are described in terms of type, species, land and crop management and structure. Farmer interviews provide additional information on flooding in the valley bottoms, crops and crop rotations, land tenure, inputs (labour, fertilizer, etc.) and constraints.

The descriptions obtained are processed into so-called `agro-ecosystem diagrams' which show the physical parameters in a cross-section, in combination with a map (at scale 1 : 5 000) of the land use. Accompanying descriptions highlight the variability of valley characteristics within agro-ecological sub-units, as well as the differences between inland valleys in different agro-ecological sub-units, if applicable.

At this level of characterization a number of valley characteristics is quantified, serving objective comparison between entire valleys or between land elements within valleys. Physical as well as biotic characteristics are quantified (Table 3). The Valley Bottom Ratio (VBR, no dimension) for example, is the ratio of the area occupied by the higher parts of the valley (crests, slopes and fringes) over the area of the valley bottom. It is a possible measure of the amount of water that may accumulate, either as runoff, or as lateral flow from the higher parts of the valley into the valley bottom: a high VBR indicates a relatively narrow valley bottom and a potentially large accumulation of water in it. Land use is quantified by means of six parameters. The Land Use Class Ratio, as an example, is the relative area of the individual land use classes over the total area of the land element. From the individual Land Use Class Ratios, the generalized Land Use Ratio is calculated as the sum of cropped area, prepared land, grazing land and young fallow, over the total area of the land element.

TABLE 3
Quantifiable inland valley agro-ecosystem characteristics (Semi-detailed characterization; Source: Van Duivenbooden et al, 1996)

TVW: Total Valley Width [m]; VBW: Valley Bottom Width [m]; xx: area of land use subclass XX (see below); TA: Total Area of land subelement in the transect; AC: Annual Cropland; PC: Perennial Cropland; PL: Prepared Land; YF: Young Fallow; OF: Old Fallow; GL: managed Grazing Land; LP: area that has been ploughed, ridged, mounded, furrowed or bedded [all in ha].

Land use subclasses (XX) are: Wasteland; Grass and Forbland; Savanna and Shrubland; Woodland and Forest; Grazing Land; Young Fallow (less than 10 years old); Old Fallow (10-30 years old); Prepared Land (i.e. cleared, ploughed or sown); Annual Cropland (by species); Perennial Cropland (by species); Infrastructure; and Mined land.

Detailed characterization

Detailed characterization of valleys and land elements forms the last step of the multi-scale approach, and is performed at scales between 1 : 5 000 and 1 : 10 000. It is carried out in one or two representative valleys selected in the preceding characterization phase, and aims at the classification of the inland valleys and their land use, including the quantification and ranking of the constraints to sustainable production. The information required is collected in detailed field surveys and through farmer interviews. Field surveys start with the further interpretation of aerial photographs, linking the transect information on physical and agronomic characteristics, as gathered during semi-detailed fieldwork, to geographically defined map units. Temporal variability is assessed through monitoring programmes and, it is noted here specifically, research and experimentation form an integral part of the detailed characterization phase, as is simulation modeling.

For the development of improved management technologies, there is a need to assess the agricultural potential for the particular agro-ecosystem under study. One way of doing this is through modeling. Crop growth simulation models, for example, are important not only to understand and predict productivity of crops, but also to identify characteristics of crops (and crop varieties) for each of the land elements in inland valleys. Socio-economic models will allow insight into the probable response of farmers to new technologies, changes in prices and markets, etc. Modeling also provides a way through which the vast, but unrelated research results from the past decades may be made useful. Observations during the detailed characterization, can be used to validate these models.

Also at this level of characterization, additional parameters are defined. Total Land Use Intensity (TLI), for example, describes the intensity of agricultural exploitation in the last 30 years (prepared, cropped or fallow land) relative to the area that can be cropped. The Actual Land Use Intensity (ALI), expresses that exploitation in the last 10 years, thus excluding the area under old fallow. The relative importance of soil preparation can be expressed in the Soil Preparation Intensity (SPI), as this activity has effects on the land and the production level of the crop. The three parameters are defined as follows (abbreviations are explained in footnote to Table 3):

TLI: (AC + PC + PL + YF + GL + OF) * 100 / (TA - IN - WL)

ALI: (AC + PC + PL + YF + GL) * 100 / (TA - IN - WL)

SPI: LP * 100 / (AC + PC + PL)

A particular type of detailed characterization is the `diagnostic rapide hydrologique' - rapid hydrological diagnosis - as developed by CIRAD for inland valleys in southern Mali and northern Ghana (Legoupil and Lidon, 1996). The method is based on the measurement and interpretation of a number of physical characteristics ('indicators') for the determination of the method of water management best suited to the local conditions. The indicators are: soil permeability, depth to the impermeable subsoil, width and slope of the valley bottom, and cross-sectional wet profile. These characteristics are all directly measurable in the field. A number of compound indicators, based on decennial flooding, base flow and water table dynamics, are either collected through monitoring programmes, or inferred from secondary sources. By means of a user key, the relevant water management technique is identified.

Results

Macro characterization

In our area of interest, West Africa, three major agro-ecological zones are distinguished: the Equatorial Forest Zone, the Guinea Savanna Zone, and the Sudan Savanna Zone. Their main characteristics are summarized in Table 4. Further details are reported elsewhere (Windmeijer and Andriesse, 1993; Andriesse et al., 1994; Thenkabail and Nolte, 1995).

TABLE 4
Main characteristics and extent of the major agro-ecological zones in West Africa (Macro characterization. Sources: Andriesse and Fresco, 1991; Windmeijer and Andriesse, 1993).

^* Estimates based on maps at scale 1:5,000,000

Reconnaissance characterization

The major agro-ecological units in C�te d'Ivoire are the Plateaux, formed over Basement Complex formations in the Guinea Savanna Zone, and the Interior Plains, also formed over Basement Complex formations, in the Equatorial Forest Zone. Using information from secondary sources and data from the rapid inventory of the main land use in inland valleys in this country (Becker and Diallo, 1992), four agro-ecological sub-units were distinguished within the two major agro-ecological units.

In these two units key areas were selected, near Boundiali in the northern part of the country (in the Guinea Savanna Zone, on Plateaux), and near Gagnoa in the south (in the Equatorial Forest Zone, on Interior Plains). The key areas were selected in such a way, that each of them covers two different agro-ecological sub-units. The latter were distinguished mainly on the basis of a further differentiation of the prevailing lithology (i.e. Granites and Schists in Boundiali, and Migmatites and Schists in Gagnoa), and the (related) drainage density. Further details are shown in Table 5.

TABLE 5
Characteristics of the agro-ecological sub-units covered by the Boundiali and Gagnoa key areas in C�te d'Ivoire (Reconnaissance-level characterization; Source: Andriesse et al, 1994).

* In both Boundiali and Gagnoa about 80% of the rice is grown on the uplands.

Semi-detailed characterization

In the semi-detailed characterization that was carried out in C�te d'Ivoire (January-March 1993), 14 valleys have been studied. Eight valleys were in the Boundiali key area and six in Gagnoa. Figure 2 shows, as an example, an agro-ecosystem diagram, being a cross-section of the valley and the land use map along the transect in a valley formed in migmatite near Gagnoa. In Table 6, the main physical and land use characteristics are summarized, of the valleys studied in C�te d'Ivoire.

The table shows clear differences between the four valley systems. The valleys in Boundiali are much wider than those in Gagnoa. This is in accordance with the drainage densities as observed in the reconnaissance characterization. The valley bottoms however, are much wider in Gagnoa than in Boundiali: Valley Bottom Ratios (VBR) for Gagnoa are 5 (schists) and 13 (migmatites) respectively,

TABLE 6
Selected physical and land use characteristics of valley systems in the Boundiali and Gagnoa key areas, C�te d'Ivoire (Semi-detailed level characterization; Source: Van Duivenbooden et al, 1996).

whereas in Boundiali the VBRs are 64 (schists) and 83 (granite). The high standard deviations for the various land elements in the valleys indicate highly variable conditions along the slopes in all four valley systems.

Also in terms of land use a number of differences between the valleys stands out. Land use in Gagnoa, for example, is much more intensive than in Boundiali: Average Land Use Ratios (LUR) in Gagnoa are 76% and 65% on schists and migmatites respectively; in Boundiali LURs are 28% (on schists) and 33% (on granites). Due to this higher land pressure, fallows are shorter in Gagnoa (between 3 and 10 years) than in Boundiali (7 to 30 years). Also, the Fallow Indexes are much smaller in Gagnoa (0.30-0.51) than in Boundiali (0.71-0.75).

A considerable difference exists in soil preparation between the two areas. In Boundiali, most of the land is being prepared (ridging or ploughing mainly; SPIs are 62 and 75 respectively), while in Gagnoa (SPIs: 0 and 5) almost no land preparation takes place. Soil preparation is essential in Boundiali because of prevailing very high soil compaction.

Detailed characterization

Results of the detailed characterization are not yet available as most of the detailed characterization activities in West Africa are still on-going. Full inventories of physical and biotic characteristics in inland valleys will allow their further quantification. The increase in the level and amount of collected detail allows the quantification of the variability (spatially and temporarily) of physical, agronomic and socio-economic variables. For instance, the dynamics of farming and cropping systems within valleys are studied in relation to the dynamics of the physical and socio-economic conditions. Sub-surface water flow, water availability in the valley fringes and flooding regimes in the valley bottoms, their effect on nutrient availability (N, P, K) and (iron) toxicity, as well as nutrient balances of cropping systems and soil erosion processes are also subject of study.

Of course at certain points, results from previous research can be incorporated in the work to be carried out. The Inland Valley Consortium is making efforts to compile and synthesize this on-the-shelve information and technologies. This will assist in developing improved management packages adapted to the local agro-ecological conditions. Upon on-farm testing, these technologies may be extrapolated to other areas with similar agr-ecological conditions.

Figure 3 shows CIRAD's user key. It sets out, from top to bottom, the respective indicators (permeability, depth to impermeable layer, longitudinal slope, streambed prominence, discharge, watertable fluctuations and baseflow. Fitting the measured values for actual conditions into this key results in identification of the best suited water management system.

Pending research issues

Throughout the preceding text, and in supporting literature and manuals, a number of methodological issues has been raised which, in all frankness, need further scientific elaboration for the fine-tuning of the multi-scale system of agro-ecological characterization. First of all this concerns the issue of representativity. At several points in the multi-scale approach, so-called `representative units' are selected for further characterization at the next-lower level of detail: agro-ecological sub-units (at macro level), key areas (at reconnaissance level), and valley systems (semi-detailed level).

Representativity here is generally taken as `most common', and possible variation between representative units is taken into account merely by duplication, or multiplication, of the characterization activities (i.e. selection of 2 key areas per agro-ecological sub-unit, 2 to 4 valleys per key area, etc. In view of the large amount of (field) work involved in characterization, particularly at semi-detailed and detailed level, further scrutiny of this issue is required. Powerful support on the issue of selecting representative units is expected from remote sensing technology.

A similar issue concerns the number of transects and their exact location in the valley system, during semi-detailed characterization. The present prescription relies largely on the surveyors' craftsmanship in locating representative transects, cutting across as many land segments as possible in the inland valleys. Research is being undertaken by the `Wageningen Group' in the IVC to determine more-objective ways of locating the transects (random?, random-stratified?) and determining their required numbers to allow further statistical processing.

Also, the differentiating characteristics used in the characterization method are subject to scrutiny. For example, rather than determining the lateral extent of the hydromorphic fringes by means of time-consuming observations in soils, which moreover reveal their hydromorphy through very static indicators (soil colours, mottles, etc.), the search is for possible more- obvious and dynamic indicators, such as vegetation complexes or types.

With the renewed interest in the small wetlands (i.e. the inland valleys, fadamas, valley swamps, bas-fonds in West Africa, and the dambos, mbugas, matoros, and vleis of eastern and southern Africa) the issue of their classification re-emerges as well. Many researchers have attempted to clasify these intriguing environments from as many points of view. So far however, a comprehensive classification system is lacking. Therefore, to give just one example, we are not sure whether results obtained in research and experimentation in West Africa, will be applicable in eastern and southern African conditions as well. In this respect it is remarkable that Thomas and Gouldie (1985) in a standard volume on small wetlands define Dambos as `small channelless valleys in the tropics', whereas most of the contributors to the same book discuss environments with distinct channels or streambeds as can be judged from the many drawings and photographs provided.

Much of the research effort in inland valleys is geared towards alleviating productivity constraints in order to enable their 'sustainable' development. Admittedly, so far, in West Africa little attention has been paid to establishing the ecological value of inland valley agro-ecosystems, compartments thereof, or processes therein. Development of ways to 'value' (patches of) natural vegetation, or ecological processes, would provide means to balance productivity-driven decision making as against conservation and sustainability objectives.

So far, the multi-scale agro-ecological characterization method puts emphasis on characterization of physical and biotic (including agronomic) parameters at the various scale-levels. Socio-economic parameters are included only to some extent at the semi-detailed level, and, more profoundly, at the detailed level of characterization. Development of a system of scale-parallel socio-economic parameters for all the characterization levels, would further broaden the scope of the system and enhance its applicability in planning exercises at all scales.

Lastly, as has been stated earlier, the multi-scale agro-ecological characterization method provides a mechanism for the extrapolation of site-specific results of research to other areas with similar, or slightly different, conditions. Transfer mechanisms to areas other than such recommendation domains are still subject to research, as is the possibility to use agro-ecological information for the determination of priority issues in agricultural, or rather environmental, research.

Bibliography

Andriesse, W. 1986. Area and Distribution. In: Juo and Lowe (editors): The Wetlands and Rice in Sub-Saharan Africa. Proceedings of an International Conference. Ibadan, November 1985. IITA, Ibadan, Nigeria. p. 15-30.

Andriesse, W. and Fresco, L.O. 1991. A characterization of rice-growing environments in West Africa. Agriculture, Ecosystems and Environment, 33:377-395.

Andriesse, W., Fresco, L.O., van Duivenbooden, N., Windmeijer, P.N. 1994. Multi-scale characterization of inland valley agro-ecosystems in West Africa. Netherlands Journal of Agricultural Science 42(2):159-179.

Aubreville, A. 1949. Climat, for�ts et d�sertification de l'Afrique tropicale. Soci�t� des Editions G�ographiques Maritimes et Coloniales. Paris, France.

Becker, L. and Diallo, R. 1992. Characterization and classification of rice agro-ecosystems in C�te d'Ivoire. West Africa Rice Development Association, Bouak�, C�te d'Ivoire. 252 p. + Annexes.

Bunting, A.H. (Editor). 1987. Agricultural Environments: Characterization, Classification & Mapping. Proceedings of the Workshop on Agro-ecological Characterization, Classification and Mapping. Rome, Italy, April 1986. CAB International, Wallingford, UK. 335 p.

FAO. 1978. Report on the Agro-ecological Zones Project. Vol.I: Methodology and Results for Africa. FAO World Soil Resources Report 48. Food and Agriculture Organization, Rome, Italy. 158 p.

Izac, A-M.N., Swift, M.J., Andriesse, W. 1990. A strategy for Inland Valley Agro-ecosystems Research in West and Central Africa. RCMP Research Monograph No. 5. Resource and Crop Management Program, International Institute of Tropical Agriculture, Ibadan, Nigeria. 24 p.

Legoupil, J-C. and Lidon, B. 1996. Le diagnostic rapide hydraulique des bas-fonds. The Inland Valley Newsletter, 1:20-23.

Raunet, M. 1985. Bas-fonds et riziculture en Afrique. Approche structurale et comparative. L'Agronomie Tropicale, 40(3):181-201.

Raunet, M. 1982. Les bas-fonds en Afrique et � Madagascar. Institut de Recherches Agronomiques Tropicales (IRAT), Service de Pedologie, Montpellier, France. 56 p.

Thenkabail, P.S. and Nolte, C. 1995. Mapping and characterizing inland valley agro-ecosystems of West and Central Africa. A methodology integrating remote sensing, global positioning system and ground-truth data in a geographic information systems framework. RCMP Monograph No. 16. Resource and Crop Management Program, International Institute of Tropical Agriculture, Ibadan, Nigeria. 52 p.

Thomas, M.F. and Gouldie, A.S. (editors). 1985. Dambos: Small channelless valleys in the tropics. Characteristics, formation, utilisation. Zeitschrift f�r Geomorphologie Suppl. Band 52. 222 p.

Van Duivenbooden, N. and Windmeijer, P.N. 1995. Manual for semi-detailed characterization of inland valley agro-ecosystems. Characterization of Rice-growing Agro-ecosystems in West Africa, Technical Report No. 4. Technical Report No. 4. DLO Winand Staring Centre, Wageningen, The Netherlands. 85 p.

Van Duivenbooden, N., Windmeijer, P.N., Andriesse, W. and Fresco, L.O. 1996. The integrated transect method as a tool for land use characterization, with special reference to inland valley agro-ecosystems in West Africa. Landscape and Urban Planning 34:143-160.

WARDA. 1993. Sustainable Use of Inland Valley Agro-ecosystems in Sub-Saharan Africa. A Proposal for a Consortium Programme. West Africa Rice Development Association, Bouak�, C�te d'Ivoire. 79 p.

Windmeijer, P.N. and Andriesse, W. (Editors). 1993. Inland Valleys in West Africa: An Agro-ecological Characterization of Rice-Growing Environments. ILRI Publication 52. International Institute for Land Reclamation and Improvement, Wageningen, The Netherlands. 160 p.

Windmeijer, P.N., van Duivenbooden, N. and Andriesse, W. 1994. Semi-detailed Characterization of Inland Valleys in C�te d'Ivoire. Characterization of Rice-growing Agro-ecosystems in West Africa. Technical Report 3 (2 Vols.). DLO Winand Staring Centre, Wageningen, The Netherlands. 129 p.