Hydrological strategy to develop and manage African wetland resources for sustainable agricultural use

Introduction

The overall objective of this paper is to summarize work undertaken by the Institute of Hydrology (UK) in response to a request from the Land and Water Development Division of FAO to provide a baseline for a hydrological strategy for the development and management of African wetland resources for sustainable agricultural use (Bullock et al, in preparation).

In an attempt to improve agricultural production in Africa in the context of food security, FAO has identified four general problems that have to be addressed if programmes for agricultural development are to be successful: (i) limited scope for rainfed cultivation, (ii) rural labour shortages, (iii) limitations on the large-scale irrigation option as a solution to increased food production and (iv) highly unreliable rainfall hindering the adoption of "green revolution" technology.

In view of these considerations, abundant African wetlands appear to have high potential for the development of small-scale farming systems at village scales, without necessarily requiring major inputs. However, the potential of wetlands as agricultural resource have remained largely untapped. (Izac et al, 1990). Only between 10 and 25% of inland valleys in West Africa are used for agricultural production Within Zimbabwe, it is estimated that it would be possible to expand the informal cultivation of dambos by up to 25% of the total wetland area (Faulkner and Lambert, 1991) without significant deleterious impacts. This expansion would yield major benefits to local communities by contributing to food security, financial income and health, particularly amongst the female population.

Whatever the potential benefits of wetland agriculture, development will not be permitted to expand in an unrestrained manner. Seeking to stem the progressive encroachment onto, and loss of, wetlands, the international conservation organizations strongly promote the concept of "wise use" of wetlands. Although not necessarily against agricultural development, the Wise Use Working Group of Ramsar has issued "Guidelines for implementation of the wise use concept of the Convention" which call for the establishment of national wetland policies covering all problems and activities related to wetlands.

A thorough understanding of the hydrology of wetlands is central to the wise use concept, because water is intrinsic to the natural properties of a wetland ecosystem.However, while wetlands have been

Andrew Bullock, Kevin Gilman, Matthew McCartney, Dominic Waughray, 
Ken Blyth, Anthony Andrews 
Institute of Hydrology, Wallingford, United Kingdom  

the focus of many scientific and development disciplines, the hydrological knowledge of African wetlands remains in its infancy. There are indeed few African scientists who specialize in wetland hydrology, and the status of appropriate databases and analytical techniques is at best unconsolidated. Critically, in the context of this paper, the hydrological knowledge base is inadequate for formulation of sustainable agricultural development at a sub-continental scale.

In this regard, FAO supports the African Technical Cooperation Network in Wetland Development and Management (WEDEM). The general objectives of WEDEM are to promote the exchange of information on wetland development. As one contribution towards the overall FAO goal of the achievement of food security in Africa, FAO is seeking to revitalize WEDEM through two main areas of activity; first, the commissioning of country position papers on wetland development and management in African countries; second, the commissioning of a continent-wide perspective on a hydrological classification system of wetlands. Regarding the latter, in November 1994 FAO approached the Institute of Hydrology, Wallingford UK, expressing interest in the development of a classification system of wetlands that would allow distinction between (i) wetland types, (ii) hydrological function, (iii) ecological function, (iv) development options and constraints and (v) carrying capacity. In response, the Institute of Hydrology submitted a pre-proposal of the technical outputs which would move towards a basis upon which an achievable classification could be implemented, rather than the supply of an `operational' classification system. No detailed specification for the hydrological classification was defined by FAO, nor firm viewpoints expressed as regards the nature and implementation of any mechanisms or products associated with the classification. The initiation of a hydrological classification is viewed at this time as establishing the base line for wider discussion prior to the implementation of an operational technical programme. The Institute of Hydrology was contracted to implement a desk-study for an achievable classification scheme, with resources to access information at FAO (Rome) and IUCN (Switzerland) but without resources to access directly information within African countries.

In the absence of a detailed technical specification, there were however a number of components which FAO considered to be of importance:

1. the scope of the hydrological classification should be pan-African, although it is recognized at the outset that more information may be available for southern Africa;
2. the scope should be restricted to freshwater inland wetlands;
3. the hydrological classification should be capable of being implemented within a Geographical Information System (GIS);
4. the classification scheme should provide the means for assessing the suitability of wetlands for agricultural development;
5. the type of agricultural development remains unspecified; FAO wished to be advised of, and have the capability to assess, different development options;
6. the classification scheme should address agricultural development in a sustainable context;
7. agricultural development may involve manipulation of the wetlands; such manipulations should be considered in the context of sustainability;
8. FAO recognizes that agricultural development might change the function of wetlands; it wished to be advised upon the nature and magnitude of the impacts of different development options, and to move towards the means to assess the impacts of a particular development option;
9. FAO recognizes that there are a number of constraints upon wetland development. Consequently, it wished to be advised upon the nature of these constraints, and to move towards the means to assess the constraints upon a particular development option.
10. the approach for wetland classification should adopt similar principles to the Land Use Classification System being developed by the Institute of Terrestrial Ecology on behalf of FAO, being essentially a flexible attribute-based classification scheme.

In line with the above issues this paper, which is a summary of the main report (Bullock et al, in preparation), follows a sequence based upon:

• a review of base line information relating to wetland mapping and databases
• a review of existing/scheduled regional GIS coverages of environmental data
• an evaluation of existing wetland classification schemes
• a review of African wetland hydrology
• a review of the options for, and constraints upon, sustainable production in African wetlands
• a technical proposal for a hydrological classification for the development and management of African wetland resources for sustainable agricultural use
• a proposed implementation strategy, schedule and financial summary for a pilot implementation in southern Africa

Review of baseline information relating to wetland mapping and databases

There have been a series of key publications which review different aspects of African wetlands at the continental scale (Table 1), which yield valuable information and data. Many of these include descriptions of the geographical distribution and location of wetlands.

It is believed that there has been no consistent mapping policy for wetlands at the continental scale that includes all African wetlands. As a consequence, there is wide divergence amongst estimates of wetland extent in Africa (Table 2).

FAO's estimate of 1.25 million km² is based on mapping of naturally- and marginal naturally-flooded areas at a continental scale, using the Soil Map of the World. The estimate can be considered to be more objective than the other pan-African estimates because firstly, unlike the others, it does not consider only selected wetlands of international importance and secondly, a standard mapping scheme has been applied. However, because of the scale of mapping, only the larger flooded areas are inventoried, and many of the smaller wetlands are ignored. It is therefore likely that the FAO inventory estimate of 1.25 million km² is an underprediction. The University of Leiden has estimated that of the total wetland extent in Africa, approximately 50% is composed of larger floodplains (such as may be mapped by FAO) while the other 50% comprises smaller wetlands (such as may not be mapped by FAO). Applying such a general rule, and thereby doubling the FAO mapped extent of 1.25 million km², would yield an estimate of total wetland extent in Africa to be around 2.50 million km². This figure would suggest that previously published estimates could be an order of magnitude too low, and indeed, that Africa may have a larger extent of wetland than any of the other continents. In terms of the number of individual wetlands in Africa, Balek (1989) estimates there to be some 10 to 100 thousand individual wetland systems alone.Assuming an average areal extent of individual dambo wetlands in southern Africa to be 1 to 5 km², then 1.25 million km² of small unmapped wetlands would suggest that the number of individual wetland systems is more in the range of 100 thousand to 1 million, and hence an order of magnitude higher than previously estimated.

TABLE 1
Key publications reviewing African wetlands at the continental and regional scales

TABLE 2
Different estimates of total wetland extent in Africa

Consolidated continental-scale mapping of wetlands has been undertaken, but considering only the larger wetlands, and at least two initiatives provide GIS coverages of the larger wetlands; the FAO Soil Map of the World and the World Conservation Monitoring Centre. Africa's smaller wetlands have generally been excluded from these two initiatives for reasons of scale, but can not be ignored because of their extensive distribution and the cumulative coverage of a large number of small wetlands. In addition to basic mapping exercises, initiatives have been undertaken to establish databases of wetland attributes at the continental scale. These include the University of Leiden Wetlands database, the African Wetland database "CLASSIFY", established by Samara Consulting, and the Ramsar "List of wetlands of International Importance".

However, the mapping of wetland boundaries has been undertaken, albeit in a somewhat ad hoc manner, during the construction and publication of national cartographic map series by national mapping surveys. The issues of wetland mapping under these circumstances are complicated by matters of scale and cartographic interpretation. Throughout Africa, national map series have been published at different scales, and not surprisingly, there has been no international standardization of the definition to be used during the cartographic interpretation of wetlands. This situation is illustrated by Table 3, which reviews wetland mapping on national map series amongst 11 countries of southern Africa. It would be anticipated that this variability in polygon mapping illustrated in the southern African region would be repeated throughout the continent.

As discussed above the multitude of smaller headwater African wetlands are omitted by existing GIS coverages and consequently there are large discrepancies in the representation of the areal extent of wetlands. For example, the FAO 0.5E x 0.5E grid of wetland density in Africa identifies the major wetland systems, such as the Okavango and Zambezi floodplains but omits the many headwater dambo wetlands, thereby underestimating wetland coverage by at least 50%. The spatial distribution of different wetland types in Southern Africa is represented on national topographic maps, although no standard wetland nomenclature is adopted.

In order to overcome the deficiency in representing the many small wetlands of Southern Africa, a new wetland density grid was produced by the Southern FRIEND project (Unesco 1997). The data set was developed by sampling mapped wetlands (Table 3) using a 15 minute resolution grid square, adopting a 16-class wetland definition system to preserve these national systems. Certain wetland classes such as floodplains and dambos are represented by areal extent while others such as pans are represented by frequency within a grid cell.

At the regional scale, there have been several initiatives in wetland mapping and databases; in West Africa, these include initiatives by the WURP (Wetland Utilization Research Project), CORAF (Conference des responsables de la Rechereche Agronomique africains), the IUCN Task Force on Sahelian Floodplains and the Inland Valley Consortium (IVC) Research. In East Africa, IUCN activities are linked to the development and implementation of wetland biodiversity programmes, in association with the GEF Biodiversity Support Project, principally in Uganda, Tanzania and Kenya. In Uganda, a Wetland Biodiversity Inventory team has been assembled to prioritize wetlands for biodiversity conservation. In Tanzania, a project proposal to develop a National Wetlands Programme

TABLE 3
Review of wetland mapping on national map series within southern Africa

has been initiated, including the establishment of a Wetland Information Centre with a database and library, with support from the GEF Biodiversity Project. In Kenya, the National Wetlands Committee was expected to be ratified by the Inter-ministerial Committee on Environment during 1994, with an early activity being the establishment of a wetland information centre. Wetland mapping and inventory activities have commenced or are being proposed in Kenya and OSIENALA's "Lake Victoria Wetlands Inventory, Assessment and Mapping programme". The National Environment Management Council (NEMC) has begun to develop a Tanzanian Wetlands Relational Database. In Uganda, the first proposed product of the first immediate objective in the Prospectus of the Ugandan National Wetlands Conservation & Management Programme is wetland resource inventories. Under the Norwegian-funded Biomass Project within the Forestry Department of the Directorate of Environmental Protection, wetlands have been delimited by supervised classification on 1:50 000 SPOT images. Rwanda is well-advanced in the creation of an inventory of the country's wetlands with a computer database storing over 50 attributes for each individual wetland.

During the period 1994-1996, the objective of the IUCN Wetlands Programme in southern Africa is to develop the regional capacity to manage wetlands through the organization of technical workshops on wetland conservation and management issues, and through the further development of national wetlands programmes and training activities.

Remote sensing has very considerable potential in the mapping and characterization of wetlands. Wetlands occur over a wide range of spatial scales and the size of area to be monitored will strongly determine the choice of satellite and sensor. However, because wetlands may be very heterogeneous, but at the same time extensive, a stratified sampling scheme which makes use of both low and high resolution satellite data may be an appropriate general approach. Low spatial resolution generally equates with large swath width and consequently more frequent repeat coverage. As spatial resolution increases, swath width decreases and the interval between repeat passes increases. These basic geometric and temporal sampling functions must be considered alongside the spectral sensitivity of the available sensors and the measurement requirement. Five main application areas for remote sensing of wetlands can be identified and these are summarized in Table 4. Several of these applications have been implemented in Africa, and case studies are presented in Bullock et al. (in preparation).

Existing/scheduled regional GIS coverages of environmental data

Although wetland delimitation in Africa has been shown above to be at a low level at this time, but advancing at pace, the establishment of other thematic databases is already well advanced. Central to the collation and distribution of existing coverages is the Global Resource Information Database (GRID) of UNEP, which distributes existing coverages in ARCINFO format free-of-charge. From a hydrological perspective, both FAO, through its GIS-Hydrology in Africa programme and the Institute of Hydrology are advanced in the creation of hydrologically relevant gridded databases at least at a regional, if not yet a continental scale. The Unesco FRIEND initiative and the FAO AFRICOVER initiative are also in the process of developing new and regionally appropriate databases and there are many, many initiatives being implemented on a national scale. Thus, although wetland mapping is not in itself well-advanced, the background data for defining environmental variables associated with wetlands and their catchments are established. When using such environmental databases, and especially when working at a continental scale which may integrate separate regional initiatives, it is essential to bear in mind the existence of Standardization

TABLE 4
Summary of remote sensing options for wetland characterisation

Status A = currently operation. B = launch planned in 1995.

Note Repeat cycle is for exact repeat of orbit: orbit sub-cycles often allow more frequent observation of area of interest.

Commissions, which provide a mechanism for unifying different data sources. One such example is the UNEP/ITE Harmonization of nomenclature for recording land use and land cover globally.

Combining GIS coverages of wetland polygons with such extensive thematic environmental GIS coverages creates the technical feasibility to retrieve thematic data for wetlands and their upstream catchments. Such retrievals represent a standard facility within GIS software. Defining catchment boundaries upstream of individual wetland polygons is a critical state in defining the hydrological inputs into wetlands. There are basically two alternatives to automated catchment boundary definition: a two-dimensional method based on digitized river networks and a three-dimensional solution based on flowpaths generated within a Digital Elevation Model.

To assess the potential of wetlands for crop production or farming systems, one is interested in knowledge of indicative climatological, hydrological and soil requirements for selected crops. It is THE critical interface in the assessment of wetland suitability for agriculture between defined environmental factor ranges for wetland crops (or cropping systems) and the definition of the occurrence of the same environmental factors in wetlands. In principle, both are technically feasible with the investment of targeted effort to do so. This interface places a demand upon the agronomist community to define the key environmental factor ranges for the dominant wetland crops or cropping systems. It also places a demand upon the climatological, hydrological and pedological community to map, or predict (by simulation models) the occurrence of these key environmental factors. Field survey of every individual wetland is obviously impossible, and so an understanding of the causative processes which explain wetland properties is a requisite phase. The existence of thematic GIS coverages provides the geographical framework for assigning key properties to individual wetlands.

In seeking to define the environmental factors which determine suitability for cropping, the Booker Tropical Soil Manual sets out crop environmental requirements for selected individual crops and cultivars (Table 5) and the FAO ECOCROP database (Table 6) presents equivalent data for approximately 1200 crops within an easily accessible PC-based retrieval package.

In the characterization of wetlands as suitable for agricultural production, it is the key challenge to be able to estimate values of these crop requirements for individual wetlands.

Wetland classification schemes

Cowardin et al. (1979) identify four major objectives of a classification scheme:

1. to describe ecological units that have certain homogeneous natural attributes
2. to arrange these units in a system that will aid decisions about resource management
3. to identify classification units for inventory and mapping
4. to provide uniformity in concepts and terminology

However, wetlands do not fall easily into discrete classes, and there are many ways of setting up a classification scheme, not all of them useful. For the case of complex natural systems like wetlands, the principles of classification are:

TABLE 5
Crop environmental requirements (Source: Landon, 1991)

TABLE 6
Environmental factor values in the FAO ECOCROP database

1. the allocation of a member of a continuum to one of a finite number of distinct classes
2. the assumption, which needs to be justified, that members assigned to a given class will exhibit similar behaviour
3. perhaps most importantly, the scheme must be free from the subjectivity that could give rise to unproductive debate between experts.

Even classification according to an objective scheme is not a value-free activity, and this single fact has given rise to much debate and misunderstanding among wetland scientists. It is a prerequisite of the development of a classification scheme that the objective should be clearly stated at the beginning and kept in view throughout. But it must be appreciated that there is no classification scheme that will satisfy every member of the user community, and most schemes leave room for further development in the form of additional classes at any stage and the use of lower-order modifiers to refine the classification to take account of local or specialist needs.

There are six main criteria used in past wetland classifications on which partition have been based: (i) ecology, (ii) ontogeny, (iii) hydrochemistry, (iv) soils, (v) topography and (vi) hydrology and examples of criterion-specific and combined classifications are presented in Bullock et al. (in preparation).

Several factors have combined to put the United States into the leading position on the classification, mapping and inventorization of wetlands. The first major classification project in the US was the Circular 39 inventory prepared by the Fish & Wildlife Service (Shaw & Fredine 1956), drawing on funds largely derived from the sale of Federal Duck Stamps. The current US Fish & Wildlife Service classification gives a comprehensive coverage of US wetlands, and it gives a more balanced and ecologically better-founded approach than the wildfowl-biased Circular 39. Comprehensive as the US Fish & Wildlife Service scheme may appear, it is not entirely satisfactory for sites outside North America and the IUCN, faced with the need to prepare a classification scheme for the Ramsar Convention, modified the US scheme to simplify it and extend its scope.

It is important that the structure of a classification scheme should be logically consistent. At the site level, and only after the application of (sometimes sweeping) simplifying assumptions, it is possible to allocate wetlands to a class in the lower levels of a suitable classification scheme. However, this assignment may disguise the fact that the wetland site consists of a diversity of habitats, some of which, if treated individually, would have been assigned to other classes. Intra-site diversity is ultimately a problem that relates to perceived value and the allocation of resources: if the effective management or conservation of a site is perceived to outweigh the costs of detailed mapping or hydrological study, the classification can be backed up or extended to the required level of detail.

The imprecise and seasonal nature of wetland boundaries is a considerable difficulty to those concerned with the implementation of conservation and land use strategies such as the "no net loss" policy in force in the United States. As a result, a compound definition of wetland conditions has evolved, in which three elements, hydrology, soils and ecology, are combined to give a means by which boundaries can be located on the ground. To a certain extent, the integrating function of these three indices counteracts the diffuse and mobile character of the boundary.

There is no pre-existing wetland classification scheme appropriate for the FAO objective of hydrological classification for sustainable agricultural use of wetland in Africa. Relevant classification activity is underway, or planned, in several African countries including Sierra Leone, Côte d'Ivoire, Benin, Nigeria, Cameroon, Mali and Rwanda, although the distinction between mapping, inventory and classification is not always clear from available information. It might be an over-generalization, but perhaps one which is not too far from the truth, to state that wetland delimitation is better advanced in southern Africa but with inventory and classification studies so far attracting a low level of investment, while in West Africa inventory and classification activities and ambitions are more advanced than the basic wetland delimitation exercise. For this reason, it is valid to consider that the approaches to classification emanating from West Africa are likely to find wider application throughout Africa. In particular, the International Institute of Tropical Agriculture (IITA) has defined a strategy for inland valley agro-ecosystems in West and Central Africa (Izac, 1991) which could be considered as contributing components to an overall FAO strategy for wetland classification in Africa. Full details of this strategy have been presented and ideas originating from this strategy have been incorporated in the recommended strategy development in this report.

An alternative approach to rigid prior classification is the establishment of a flexible attribute system which allows different users to interrogate the same core data base of attribute values for each wetland polygon according to their own purposes. Within such an approach, the core data base should comprise an archive of pre-defined attribute headings, for which individual values for each attribute are archived for each wetland polygon. One key advantage of this approach is that the user can make as many different selections as is sought, not necessarily having to pre-define all attributes but only those with which the user is particularly concerned. Geographical Information Systems (GIS) are specifically designed for this purpose, being an organized collection of computer hardware, software, geographic data, and personnel designed to efficiently capture, store, update, manipulate, analyze and display all forms of geographically referenced information. It is envisaged that a GIS flexible attribute definition and selection approach is the most appropriate direction for African wetland classification to adopt, in preference to the more rigid prior classification approach discussed.

African wetland hydrology

The hydrology of African wetlands is frequently claimed to be hampered by the paucity of reliable hydrological records and to date there has been no attempts to collate and integrate findings of those few studies which have been undertaken in a single comprehensive manner. Bullock et al. (in preparation) presents a collation of some 17 studies of the hydrology of African wetlands, albeit one which is recognized as not being exhaustive. Generally, the incorporated studies can be categorized into three types of study: detailed investigations of individual wetlands, regional hydrological studies concerned with the relationships between wetland extent and river flow regimes, and monitoring and modeling of internal wetland hydrology for agricultural purposes.

Descriptions of the hydrological functions of wetlands has attracted much attention in wetland policy as complex hydrological processes become generalized to statements such as `wetlands reduce flooding' or `wetlands promote groundwater recharge'. While there has been a considerable research effort directed at the hydrological functioning of wetlands in temperate climates, particularly in the USA, much less has been conducted on tropical wetlands. Consequently, while there is considerable scientific literature elaborating upon conceptual viewpoints of the role of wetlands and their hydrological function these are rarely verified by direct evidence obtained from research studies. The Adamus and Stockwell assessment which generalizes functions, products and attributes for different types of wetlands has been widely disseminated throughout Africa within documentation of IUCN.

To the hydrologist, the three water quantity functions that are quoted can be considered as merely a subset of functions associated with the interactions between wetlands and groundwater/river systems. Thus, it is possible from the hydrological viewpoint to propose a different structure to the Table of functions. In doing so, the hydrologist may well consider ascribing wetland influence to more specific indices of hydrological process, such as:

• Evapotranspiration and total river flow contributions
• Groundwater recharge
• Base flow contributions, including annual base flow, dry season flow duration, dry season flow magnitude, dry season flow recession rates
• Flood characteristics, including flood volume generation, flood peak magnitudes and flood timing

The understanding that wetlands perform important water quantity functions, including flood flow modification, groundwater recharge and flow regulation, has the potential to significantly influenced wetland management policy. However, detailed review of the supporting scientific evidence suggests a lack of unanimity amongst experimental results. To redress the absence of readily-accessible scientific information, a database of published statements on the water quantity functions of wetlands from 145 studies worldwide has been established. This provides a benchmark of the accumulated knowledge of wetland influences upon downstream river flows and groundwater aquifers. A total of 393 functional statements were extracted from the scientific literature. These were structured according to wetland hydrological type, the manner in which functional conclusions have been drawn, and different hydrological measures, with emphasis on gross water balance terms, groundwater recharge, base flow and low flows, flood response and river flow variability. Geographically, the sample is dominated by 80 studies from North America (including 23 different U.S. States and six Canadian Provinces/ Territories), with additionally 34 studies from 14 countries in Europe (including the former Soviet Union), 24 studies from 10 countries in Africa and 7 from elsewhere (New Zealand (2), South America, Brazil, India, Indonesia and Malaysia). The types of wetlands amongst the studies reviewed are, in decreasing proportions, `groundwater slope' (38%), floodplains (20%), `surface water slope' (18%), `general and unspecified' (14%), and `surface and groundwater depression' types (5% each). This demonstrates an emphasis on the headwater types of wetland (80% of studies of known wetland type). It is important to note, of course, that there is no implied relationship between this sample distribution and that of the population of wetland types world wide. The total number of functional statements extracted from the 145 studies is 393. This represents, on average, close to three separate functional statements per study. Regarding different aspects of hydrological response, the functional statements concern, to an approximately equal degree, gross water balance terms (114), flood response (101) and base flow and low flows (98), and, to a lesser degree, groundwater recharge (53) and river flow variability (27).

Although comprehensive in terms of the breadth of coverage, the detailed content of the presented database is limited to a large degree, and paraphrased or quoted statements often omit that information on the rate or magnitude which underpin with the functional statements, the explanatory processes, details of specific experimental implementation and the geographical context of the specific wetland. Those interested in more substantial details are referred to the original papers and documents. In particular, from a hydrological perspective, the content of the database may be perceived to be limited because of its emphasis on functions rather than processes, given that the concept of functions is not well-established within the hydrological community. However, while more process information can be extracted from the set of publications, the target of this paper is the use of functional generalizations to represent wetland hydrology to the wetland management policy arena. Clearly, there is a strong case for the development of a deeper understanding of hydrological processes within wetlands to evolve. Wetland hydrology would be well-served by a process-based understanding which targets runoff generation and which is then either directly transmitted to the wetland management policy arena or transmitted through the concept of functions.

Table 7 collates the summary functional statements for different wetland types from the stated conservation perspective; that wetland promotion of flood control, groundwater recharge and flow regulation functions are beneficial. The Table demonstrates that only one third of all functional statements regarding the water quantity functions of wetlands can be interpreted as beneficial in conservation terms from the perspective of the receiving systems. Half of all statements can be interpreted as detrimental in conservation terms. This database provides the basis for considerable future research on the hydrological functions of wetlands.

TABLE 7
Collation of summary functional statements for different wetland types from a conservation perspective

On the basis of the preceding discussion a number of conclusions can be drawn, as follows:

• it is not possible to draw clear, unreserved conclusions regarding the influence of wetlands upon different processes within the hydrological cycle. The preceding review has illustrated that positive, neutral and negative impacts upon different streamflow components are equally in evidence
• The Table of wetland functions which has been widely distributed by IUCN (Table 5.2 in Bullock et al, in preparation) can be considered as an oversimplification of the evidence and one which does not give rightful cognizance to evidence available prior to the date of its publication.

There is consequently a broadly established ethos that regards wetlands as sacrosanct because they promote recharge, reduce seasonal variability and reduce flooding. As has been illustrated, such broad-brushed assignations of function are clearly lacking the empirical support from African hydrological studies. Yet concentration upon the need to simplify wetland hydrology in terms of expression of these valued functions has negated the intercomparison of the wide variability amongst component processes that is evident, and consideration of exceptions. The way forward, on the evidence of past studies at least, is to recognize that existing generalizations in terms of streamflow impact lack any rigorous scientific defense and indeed that any interested party can be selective in the evidence available to them to promote a particular standpoint. The key point is to recognize the existence of variability of function and to move towards an explanation, and ultimately a prediction, of that variability.

One upshot of the search for generalized "rules" is the absence of more appropriate detailed water balance or deterministic models which account for individual processes, and which are capable of simulating the diverse configuration of environments in which African wetlands occur. The development and application of such water balance models for African wetlands must be given a high priority by FAO to establish a more rigorous process-based approach that is needed to counter existing widely-held concepts. Field-based data are already available for perhaps 10-15 individual wetlands which would facilitate calibration, at least in the early stages by means of long-term averages of the key water balance components. The extensive GIS coverages of environmental variables discussed above are an obvious step in defining the environmental configurations and climatological inputs associated with these wetland studies and equally in providing a means for extrapolation to other wetlands. Modeling studies are required at a more advanced level when considering the time series element of wetland processes, and at this time there are few if any appropriate rainfall-runoff models available for African environments, with a distributed structure capable of simulating wetland components. With respect to the larger wetlands of the Zambezi, a grid-based macro-scale distributed model has been developed for that catchment, with algorithms for simulating wetland evaporation losses, and information on the duration of flooding (Vorosmarty et al, 1991). The limitation of clearly defined areas of floodplain inundation was identified as a limiting factor in water balance optimization, but clearly this macro-scale grid-based distributed modeling approach is an appropriate avenue for simulating water balances of the larger floodplain wetlands. For reasons of scale, such grid-scale models are less appropriate for smaller wetlands, for which smaller-scale deterministic modeling of hydrological inputs is required.

In addition, the present lack of a clear, and consistently implemented classification system hinders our ability to make comparisons and generalizations of the hydrological function of different African wetland "types". The discussion of this section has not sought to distinguish between variations in function and different types of wetlands, it is enough to recognize that even the "dambo" type alone displays sufficient variation in response. Water balance modeling studies would, at an early stage, do best to focus on very simplified wetland types - perhaps distinguishing between small-headwater and middle-reach floodplains - to investigate the key differences in function before moving to a more complicated classification variants. However, once underway, the potential to classify wetlands then would obviously provide the framework for broader extrapolation of functions.

In contrast to the significant number of studies dealing with water quantity aspects of wetland hydrology, no African-based studies can be traced that deal similarly with wetland functions with respect to sedimentation, or indeed on water quality impacts at the catchment scale.

Wetland development for agricultural purposes may seek to modify the wetland through the construction of structures or alternatively cultivate without any structural modification. The general principles of structural modification are to provide an appropriate drainage system to prevent floods and to allow excess water to drain away by gravity, to provide an irrigation structure to supply and maintain sufficient water in each plot, level the ground to maintain water at a suitable depth in each field. Alternatively, the natural structure of the wetland may be such that agriculture can proceed without the necessity for such constructions and modification, as occurs for example with horticultural production on dambo margins in much of southern Africa. In such cases, the principal modification is the replacement of natural wetland vegetation by agricultural crops. There is concern that the agricultural use of wetlands, both with and without structural modification, can detrimentally impact upon the properties and functioning of the natural wetland system, as well as impacting upon downstream "users", dependent upon certain wetland outputs. Thus, consideration of the impact of agricultural development must address the nature and magnitude of changes introduced into both the internal hydrology and the hydrological regime downstream of a wetland.

In summarizing impacts of modification, it is clear that, although there are isolated studies of agricultural modification of wetlands in Africa, there is insufficient evidence at this time to compile a useful summary of the impacts of different wetland modification practices upon water quantity, water quality and sedimentation. In particular, this search for simplification and generalization is restricted by the fact that the natural functions of wetlands, and specifically their determination of downstream river flow, are complex, as has been shown. It is not possible, as perhaps is being sought, for example to state that "wetland type A has the impact, in its natural state of reducing flood magnitude by x%, and that the introduction of a particular type of drainage will, by increasing the rate of flood removal, reverse the natural reduction impact to an increase in flood magnitude of y%". Such statements are clearly ambitious, and perhaps are the type of information that FAO should in the long term be seeking to reliably recommend agricultural development without significant environmental impacts. At this point in time, the level of knowledge available rarely allows us to say more than "wetland A (being an individual wetland rather than a type) has the impact in its current state of reducing flood magnitude", with variation in the sign of impact (positive or negative) between individual wetlands, very little indication of numerical magnitude of the impact, and even less regarding the impacts of man-induced modification. One approach towards summarizing such impacts is conceptualization, and conceptual modeling, of process and impact, but such an exercise should be conducted in association between agricultural engineers and hydrologists, and be based on a firmer body of scientific evidence than is collated here. It is recommended that FAO establish a specific working party, or organize a conference theme on this aspect as a means of advancing this area of concern.

Review of the options for, and constraints upon, sustainable production in African wetlands

The Lake Victoria Basin provides a useful example of wetland issues associated with agriculture, and these are fully described in Bullock et al., 1995. Geoff Howard (Pers. Comm.) of the East African Wetlands Programme draws a useful comparison between three countries of the Basin; in Kenya, the main issues are those of drainage of wetlands to increase the available land resource in the high rainfall areas, while concern lies with over-use of water and grazing resources in the wetlands of lower rainfall areas; in Uganda, the main issues are that of exploiting products of wetlands and the problems associated with definition of ownership; in Tanzania, the issues are much more diverse, reflecting the varied nature of wetlands in that country.

Wetlands associated with Lake Victoria have played a significant role in the daily lives of the people surrounding the lake. The wetlands traditionally have been used for the several purposes including fishing, hunting, salt making, ceramics, grazing and other material and cultural uses. Dugan (1990) provides a useful summary of the causes of wetland loss at a global scale. Table 8 is a first attempt at assessing the relative importance of causes of wetland change in the Lake Basin, adopting the broad headings used by Dugan.

TABLE 8
Motivating factors for wetland change in the Lake Victoria Basin

The swamps and wetlands around Lake Victoria are under increasing pressure from a number of motivating factors for change. As has been clearly illustrated, the driving force in the modification of wetlands around Lake Victoria is to acquire land for human settlement and agricultural production - for human survival. Riparian countries are at the stage of having identified future wetland developments, if not necessarily being in a position to implement them.

If the trends in wetland use of the last 50 years are continued, the Lake Victoria wetland resource will not endure long into the next century, with potentially severe consequences to the Lake Victoria system. It is of course due to the recognition of this threat that each of the riparian countries has established national wetland coordination structures as a first step towards the evolution of wetland policy. The institutional framework has evolved in concert with new thinking in relation to the sustainable utilization of wetlands. This involves management of something similar to the original swamp ecosystem by making some economic use of the swamp, thereby ensuring that the wetland is more or less conserved intact.

Sustainable use of wetlands does not necessarily rule out agricultural production, or indeed drainage, where such modification is appropriate. The past and considerable experience of wetland development now enables the identification of the constraints on future irrigation and drainage, as identified in Table 9. The recognition of such constraints can lead to the evolution of a more sustainable wetland development strategy.

TABLE 9
Constraints on future development for irrigation and development

The extended benefit-cost approach is a economic policy tool for analyzing the options and constraints of agricultural development on wetland resources in Africa. More work, following a benefit-cost framework and drawing upon the use of TEV, option and quasi-option values, shadow projects, sustainability constraints and the clearer incorporation of legislative constraints such as the wise use and precautionary principle is recommended. Research of this nature could prove especially useful for certain key wetland systems, particularly when examined as part of a regional agricultural development strategy.

However in order to present practical and sustainable management options for the agricultural development of African wetland resources, there remains a need to move towards a much more integrated assessment of the implications of wetland change. The development of a model that integrates the hydro-ecological , agricultural, economic and social processes and feedbacks within African wetland resources, would be a key step towards proposing such management solutions.

Technical proposal for a hydrological classification for the development and management of African wetland resources for sustainable agricultural use

In aiming to implement a continent-wide wetland classification for Africa, which establishes scientific and management-decision links between hydrology and agricultural use, it is important to recognize the low base from which this goal is to be launched. Without doubt, there are local cells of expertise within institutions in Africa who would feel comfortable in contributing parts of the solution within restricted geographical confines. The broader geographical scope and inter-disciplinary nature of the issue will require a solution which is ambitious in scientific, technical and institutional terms. A recommended strategy for implementing a hydrological strategy for the development and management of African wetlands for agricultural use is presented. The strategy draws on conclusions in this report based on a review of published literature, and technical capabilities in GIS and hydrology. Where necessary, the strategy recommends technical workshops or technical documents to establish the state-of-the-art of specific elements which have received little attention in the past, such as wetland water balance simulation or agriculturally-relevant properties of wetlands.

If implemented in full, the strategy will result in the agricultural development of wetlands, in a manner which has been shown to be sustainable in hydrological, economic and conservation terms. It is worth considering this agricultural strategy in relation to other wetland strategies to place matters in their correct context. The principal wetland management strategy which has been pursued over the past decade is one of wetland conservation, as driven by IUCN and the Ramsar Convention. An approximate estimate of the financial investment would be one which has exceeded 20 million US$ over ten years, essentially to conserve wetlands in their present state. The implementation of an agricultural development strategy in Africa might not differ substantially in either cost or time scale.

The strategy is structured under eleven tasks; each task is set out with an objective and a number of recommended activities. Such is the uneven nature of wetland activity in Africa that certain activities may already be comparatively well advanced in certain regions or countries, and cultural concepts may demand diverse approaches across the continent. For these and other reasons, it is perhaps best to consider that such a strategy should be implemented on a regional (multi-country) basis rather than as a pan-African approach. It is essential that the fullest involvement of the scientific, research and user communities are engaged and maintained from the earliest stage, if products of the strategy are to be relevant, acceptable and, ultimately, useful and productive.

TASK 1. Baseline study It has been identified that there exists extensive documentary evidence relating to wetlands, but much of this exists in the `grey' area of unpublished reports and papers, and that wetland computer-data base initiatives have been implemented within European institutions, but have received a low-level of uptake within Africa. Against this background, it is the objective of this task to implement activities which will improve the collation, accessibility and dissemination of existing wetland hydrological data research and management issues. It is recommended that this objective is achieved through three activities:

1.1. Establishment of a central wetland library facility.

1.2. Production and distribution of an annotated bibliography of wetland literature.

1.3. Improve accessibility (through dissemination and provision of services) to existing wetland computer data-base services, for example the EDWIN and Samara Publishing systems.

TASK 2. Wetland delimitation It has been identified that there has been no single comprehensive mapping exercise of African wetlands undertaken to-date, and information has been generated by a diverse series of mapping exercises and studies at different scales, existing GIS wetland coverages are inadequate for the purposes of applying a hydrological classification to all African wetlands, the capture of boundaries of individual wetland polygons is technically feasible, but no African country has yet successfully produced a national wetland GIS coverage, and that remote sensing displays clear potential for monitoring water extent and seasonal changes in vegetation, for the estimation of evaporation and soil moisture and for topographic information. Against this background, it is the objective of this task to implement activities which will establish wetland GIS coverages in Africa through wetland delimitation using GIS and remote sensing technology. It is recommended that this objective is achieved through five activities:

2.1. GIS capacity building

2.2. Sub-regional review of wetland polygon mapping

2.3. Development of wetland boundary capture strategy

2.4. Technical report on wetland applications of remote sensing

2.5. Capture of individual wetland polygons

TASK 3. Simulation of hydrological inputs into wetlands It has been identified that catchment boundaries (or watersheds) can be rapidly and automatically defined using structured digital river networks to the required degree of accuracy or from Digital Elevation models, that the association of a wetland polygon with a structured river network will permit the derivation of a boundary of the catchment draining into the wetland, that thematic data base initiatives within Africa have assembled substantial GIS coverages of hydrologically relevant data, describing both the physical and social environment, that there has been much effort directed at implementing regional hydrology within GIS environments, which establishes a solid foundation of knowledge and that remote sensing displays clear potential for the direct observation of wetland hydrological properties, including water extent, seasonal changes in vegetation, estimation of evaporation and soil moisture. Against this background, it is the objective of this task to implement activities which will establish methodologies for simulating the hydrological inputs to wetlands within a GIS environment through eight activities:

3.1 Association of wetland polygons with a structured digital river network or Digital Elevation Model.

3.2 Derivation of upstream catchment boundary for individual wetland polygons.

3.3 Assembly of GIS thematic coverages for catchment characterization, relevant to hydrological simulation.

3.4 Installation of relevant thematic coverages into GIS facilities.

3.5 Technical report/workshop on the regional simulation of wetland water balance and hydrological properties within a GIS environment.

3.6 Simulation of wetland hydrological inputs for individual wetland polygons.

3.7 Application of remote sensing for direct observation of wetland hydrological properties.

3.8 Validation and/or modification of simulation of wetland hydrological inputs.

TASK 4. Characterization of internal wetland attributes (hydrology, soils and vegetation) It has been identified that there is no classification scheme that will satisfy every member of the user community and information can be lost in the classification process, that there is no existing wetland classification scheme appropriate for the FAO objective of hydrological classification for sustainable agricultural use, that an alternative approach to rigid prior classification is the establishment of a more flexible attribute system which allows different users to interrogate the same core data base of attribute values for each wetland polygon according to their own purposes and that it is recommended that a GIS flexible attribute definition and selection approach is the most appropriate direction for African wetland `classification' to adopt. Against this background, it is the objective of this task to implement activities which will characterize the internal wetland attributes of hydrology, soils and vegetation within individual wetland systems through five activities:

4.1. Technical/report workshop on the definition of agriculturally-relevant wetland attributes.

4.2. A literature-based regional report on the hydrological, soil and vegetation characteristics of wetland systems.

4.3. Development of guidelines based on causative factors and wetland attributes, to "predict" wetland attributes within individual wetland systems.

4.4. Define agro-ecological zones on the basis of GIS coverages.

4.5. Characterize attributes of individual wetland polygons on the basis of causative factors and agro-ecological zones.

TASK 5. Hydrological outputs from wetland systems to the downstream environment, as determined by wetland functions It has been identified that while there has been a considerable research effort directed at the hydrological functioning of wetlands in temperate climates, particularly in the USA, much less has been conducted on tropical wetlands, that it is not possible to draw clear, unreserved conclusions from published literature regarding the influence of wetlands upon different processes within the hydrological cycle, that there is an absence of appropriately detailed water balance or deterministic models which explain individual processes within wetlands which are capable of simulating the diverse configuration of environments in which African wetlands occur, that the published literature is very sparse with respect to African-based studies which deal with wetland functions with respect to water quality or sedimentation impacted at the catchment scale and that although there are isolated studies of agricultural modification of wetlands, there is insufficient evidence to compile a useful summary, and certainly not to draw generalized conclusions of impact. One approach towards summarizing such impacts is conceptualization, and conceptual modeling, of process and impact by association between agricultural engineers and hydrologists.

Against this background, it is the objective of this task to implement activities which will simulate hydrological outputs from individual wetland systems into the downstream environment. This will enable quantitative assessments of the impact of potential wetland modification upon downstream developments and natural environments which could be affected in a detrimental or positive manner through six activities:

5.1. Evaluation of existing rainfall-models to assess suitability for simulation of wetland hydrological functions.

5.2. Development, if necessary, of a wetland hydrological function model that is sufficiently physically-realistic to quantify the impact of land-use change and/or water abstractions (both within and upstream of a wetland) on wetland hydrology and downstream river flows.

5.3. Incorporation of wetland hydrological function model within GIS.

5.4. Definition of key hydrological and water resources management indicators, which form the basis of the definition of sustainable development. These should be expressed in terms of both the internal wetland properties and the requirements of the downstream environment.

5.5. Technical report and/or workshop to quantify the impact of wetland utilization or structural modification on internal wetland properties (particularly structure and fertility).

5.6. Technical report and/or workshop to quantify the impact of wetland utilization or structural modification on downstream outputs of wetlands, in terms of key indicators of sustainable development.

TASK 6. Conservation and legal status of wetlands It has been identified that by the beginning of 1993, there were 18 African nations contracting to the Ramsar agreement, and thereby assigning conservation status to wetlands as being International Importance (especially as Waterfowl Habitat). African nations have designated a total of 53 wetlands to the list, ranging between 1 and 12 (South Africa) per country, covering a total of 42 288 km² and that certain African countries refer to wetlands in national legislation (for example, Zimbabwe) or even possess national wetland policies (Uganda). The nature of such legislation or policy may limit wetland development options. Against this background, it is the objective of this task to implement activities which will establish the conservation, legal and policy status of wetlands through two activities:

6.1. Regional review of the conservation, legal and policy status of wetlands, and the limits on wetland development.

6.2. Classification of wetlands according to conservation, legal and policy status.

TASK 7. Potential options and constraints for wetland agricultural development It has been identified that there are a large number of ways in which communities can benefit from wetlands and that there are numerous constraints, both technical and socio-economic, on the utilization of wetlands. Against this background, it is the objective of this task to implement activities which will establish potential options for, and constraints upon, wetland agricultural development, thereby enabling the identification of regionally-specific wetland development options, through five activities:

7.1. Regional workshop on regionally-appropriate potential wetland cropping systems, resulting in definition of development options.

7.2. Specification of a range of catchment and wetland attributes required to sustain each wetland cropping system.

7.3. Definition of wetland modifications associated with different cropping systems.

7.4. Development of practical guidelines for implementing individual wetland cropping systems, with regard to region-specific development constraints and likely yields.

7.5. Development of guidelines for application of an integrated hydrological/economic assessment.

TASK 8. Categorization of the suitability of individual wetlands for implementation of wetland cropping systems It is the objective of this task to implement activities which will permit the categorization of the suitability of individual wetlands for the implementation of selected wetland cropping systems. By categorizing individual wetlands into different levels of suitability, planners will be capable of assessing the potential of wetland agricultural development at district, catchment, sub-regional, national and regional scales through two activities:

8.1. Based on the range of catchment and wetland attributes required to sustain each wetland cropping system, and the attribute values calculated for each individual wetland, categorize individual wetlands in terms of their suitability for implementation of each cropping system.

8.2. Regional impact simulation of implementation of cropping systems on wetland attributes, functions and outputs.

TASK 9. Development of decision making software for wetland agricultural development It is the objective of this task to develop software that can be used in feasibility studies to assist in decision making for wetland agricultural use through nine activities:

9.1. Software scooping and design specification.

9.2. Development of software

9.3. Transfer of GIS databases into pc based software.

9.4. Writing of user manual.

9.5. Installation of pc hardware and peripherals.

9.6. Installation of decision making software.

9.7. Beta-testing of software.

9.8. Software and user manual modification.

9.9. Software release and training courses.

TASK 10. National or regional wetland development plans It is the objective of this task to implement activities which will generate national or regional wetland development plans, based on wetland suitability. Such plans should then be submitted to environmental impact assessments and environmental economic assessments prior to implementation. It is recommended that this objective is achieved through two activities:

10.1. Define national and or regional plans, based on suitability categories.

10.2. Application of an integrated hydrological/economic research assessment.

TASK 11. Implementation of wetland development It is the objective of this task to implement activities which will implement the agricultural development of wetlands in a sustainable manner. It is recommended that this objective is achieved through three activities:

11.1. Organizations responsible for implementation of agricultural development of wetlands.

11.2. Training of agricultural planners.

11.3. Demonstration projects.

Proposed implementation strategy, schedule and financial summary for a pilot implementation in Southern Africa

The overall objective of this early stage of the project is to define an implementation strategy for a pilot study for the technical proposal. It is recommended that the implementation strategy is applied to one pilot study region in southern Africa. For reasons of practicality, it is proposed that the pilot study area comprises no more than 5 countries, and preliminary recommendations for component countries within the pilot region could include Malawi, Zimbabwe, Zambia, Mozambique, Tanzania. The region displays a variety of wetland types and most countries, but not all, have been active in wetland research for agricultural purposes. Within each region, the status of existing wetland mapping and classification is known to be diverse amongst the selected countries.

The proposed implementation strategy comprises the eleven Tasks as specified above. Three additional tasks are introduced into the pilot implementation. Two non-technical activities have been introduced; first, a Steering Committee to guide the implementation of the strategy, and second, a Post-Project Evaluation to advance the utilization of results and the evaluation of the pilot study results for subsequent wider application. The third additional task is the development and distribution of decision-making software for wetland agricultural development.

The strategy could be implemented within a 3 year (36 month) period, with an overall budget estimate of approximately US$ 1 500 000. An additional US$ 120 000 may be required for the development of a wetland hydrological function model depending on the outcome of a needs assessment. A full breakdown of the schedule and costs of individual elements of the strategy is presented in Bullock et al. (in preparation).

Acknowledgements

During the writing of this report, the authors have benefited from the inputs and advice of several organizations and individuals, and would like to acknowledge the contributions of FAO, IUCN, ORSTOM, the World Conservation Monitoring Centre, NERC Institute of Terrestrial Ecology, SAMARA Publishing, University of Leiden.

Bibliography

Adams, W.M. 1993. Indigenous use of wetlands and sustainable development in West Africa. The Geographical Journal 159 (2), 209-218.

Balek, J. & Perry, J.E. 1973. Hydrology of seasonal inundated African headwater swamps. Journal of Hydrology 19, 227-249.

Bullock, A; Keya, S.O.; Muthuri, F.M.; Baily-Watts, A.; Waughray, D.K. 1995. Lake Victoria Environmental Management Programme. Report to FAO. Institute of Hydrology, Wallingford, UK.

Bullock, Andrew; Gilman, Kevin; McCartney, Matthew; Waughray, Dominic; Blyth, Ken; Andrew, Anthony. In preparation. Hydrological strategy to develop and manage African wetland resources for sustainable agricultural use. FAO, Rome. 173 p.

Cowardin, L.M.; Carter, V.; Golet, F.C.; LaRoe, E.T. 1979. Classification of wetlands and deepwater habits of the United States. ESDI Fish and Wildlife Service, Washington DC. FWS/OBS-79/31. 103 p.

Dudan, P. 1992. Research priorities in wetland science, wetland conservation and agricultural development. In: Maltby, E.; Dugan, P.J.; Lefeuve, J.C. (editors). Conservation and development: the sustainable use of wetland resources. Proceedings of the third international wetlands conference.

Dugan, P. (editor). 1993. Wetlands in danger. Mitchell Beazley, London.

Faulkner & Lambert. 1991. The effect of irrigation on dambo hydrology: a case study. Journal of Hydrology 123, 147-161.

Iza, A.M.N.; Swift, M.J.; Andriesse, W. 1991. A strategy for inland valley agro-ecosystem research in West and Central Africa. RCMP Research Monograph No. 5, Resource and Crop Management Programme, IITA, Ibadan, Nigeria.

Shaw, S.P.; Fredine, C.G. 1956. Wetlands of the United States: their extent and their value to waterfowl and other wildlife. US. Department of the Interior, Fish and Wildlife Service Circular No. 39.