Paper 9 Institutional aspects of watershed management

Anthony Bottrall

Independent Consultant

Summary: The focus of this paper is on the management of upland watersheds. Although conceptually a powerful case can be made for using watersheds as the basis for development planning within a region, the historic dominance of downstream interests has contributed to the neglect of upland development; and it is only recently that heightened concerns about poverty and the environment have encouraged the emergence of new approaches to the development and management of these complex, diverse and risk-prone areas. Evidence from recent experience in Africa and South Asia suggests that, both in contexts where large downstream projects such as hydro-electric schemes provide scope for revenue to be reinvested upstream and where there are no such opportunities, the only approach to development that is likely to achieve success is one that seeks to build from the bottom up. Active farmer participation in the planning and development process is essential, and this is most likely to be elicited where quickly realisable benefits can be obtained from water conservation. Choice of appropriate technologies and institutions will vary substantially, according to the physical and socio-economic characteristics of each locality. The paper concludes with some general propositions and a brief outline of some of the elements that might be needed to form a successful strategy for future development in this difficult field.

Introduction

'Watershed management' can mean many different things to different people. Perspectives may vary greatly, depending on a person's disciplinary background and field experience. Any attempt to generalise about the subject must therefore be prefaced by some discussion of definitions and subject-matter boundaries and by a statement of limitations and possible biases in the author's perceptions.

Even on its own, the word 'watershed' can be a source of confusion. It is used here in its American sense, as a synonym for what has been more commonly known in Britain as a 'catchment' - a usage that requires the traditional British watershed to be referred to as a 'watershed boundary'. Many may regret the change, but this is the terminology now most generally accepted internationally. When 'management' is added, we find that the term embraces a large number of possible themes and sub-themes. These are the product of different permutations of several factors, of which the most important are:

The physical unit to be managed, which can vary greatly in terms of:

- scale, from very large (whole drainage or river basin) to very small (micro watersheds of as little as 500 ha)

- location (the whole basin, or only its upper or lower sections)

The nature of the management task, which can range from water resources planning (usually at the level of the whole river basin or in its lower, downstream section) to the implementation of development activities involving the combined management of water, land and other resources (often within upper watershed areas only).

The themes of large-scale water resources policy and planning and of downstream water management are quite extensively addressed in other conference papers. This paper will therefore focus primarily on issues relating to the management of upland watersheds: "the area of land contained within a drainage divide above a certain specified point on a stream" (Doollette and Magrath, 1990) - a point often determined in practice by engineering works that control and divert surface water for downstream uses, including major irrigation or hydroelectric schemes. In this sense of the term, watershed management is often understood to mean an approach to managing agriculture in predominantly rainfed upland areas, in deliberate contradistinction to the management of irrigated lowland agriculture.

In the discussion that follows, it should be borne in mind that this author's direct perceptions and experiences of watershed management issues have been largely derived from South Asia. Though the paper attempts to balance this through references to experience elsewhere, many of its generalisations and hypotheses are unlikely to be valid universally and should be regarded as very much open to further modification and correction in the light of other regions' physical, demographic, social, economic and political conditions. This is a subject that has yet to be studied systematically across countries and regions. In the absence of such work, this is inevitably a preliminary and partial foray into seriously under-explored territory.

THE LARGER CONTEXT: UPSTREAM-DOWNSTREAM RELATIONSHIPS

As an entry into discussing the theme of upland watershed management, it is important to understand the larger context in which any programme or project action designed to promote such management is likely to be embedded. Conceptually, a powerful case can be made for using watersheds, in their largest sense as drainage basins, to provide the basic framework for development planning within a region; and for advocating the creation of institutions that will moderate between the interests of upstream and downstream users. Thus it has been argued that:

"where the land/water/climate/people interaction is a major focus for development (i.e. land and water use) this interaction is most strongly expressed in a watershed. Moreover, water seems increasingly to be the key physical resource limiting or triggering economic development in rural areas. The success or failure of a plan can lie in choosing a boundary which is relevant to the principal planning issues, covers sufficient natural and social linkages to operate as a functional unit, and allows effective plan implementation ... The watershed unit not only integrates natural systems, but ales;' many social processes and patterns as well ... " (Hamilton and King, 1984).

The same authors go on to observe that "ecosystem processes ... provide compelling reasons for using the drainage basin as an integrating planning unit". With respect to social processes, they see the need for integrated planning as equally compelling, while recognizing that "in many developing countries ... occupational, social, or ethnic differences between highlanders and lowlanders, hill tribes and valley tribes, upstream residents and downstream residents" tend to lead to conflict. While at the micro- or meso- (e.g. small upland valley) level, these differences may help to create a sense of local solidarity, there is usually a strong "polarization of interest and values between upstream and downstream inhabitants". At this macro- (drainage basin) level, the authors envisage a planning process, the institutions of which are not specified, that will recognize the likely existence of conflict, "with upstream residents often perceiving themselves as bearing the brunt of the costs to the benefit of the downstream people"; and will then seek to "internalize" the problem by introducing "some system of trade-off or compensation”.

This conceptual framework could be further elaborated to posit a hierarchy of institutions with responsibilities for rural development (or natural resources) planning and management, from micro-watershed/village level, through watershed/district level, to drainage basin/regional level. However, in practice, despite the framework's inherent attractiveness and logical appeal, the prevalent institutions in most developing countries bear little or no resemblance to it. Instead, their agencies of government tend to be organised on vertical, sectoral lines, with all departments, other than those concerned with large-scale surface irrigation, operating within non-hydrological boundaries; and, because historically the early priorities for water development have nearly always lain downstream, it serves the interests of the more politically powerful downstream users to retain the original institutions set up for the purpose rather than agree to their replacement by something that would be much more 'rational' and equitable from an ideal resource optimising viewpoint. In other words, the shape of water related planning and management institutions has so far been much more influenced by realpolitik (implicit in Hamilton and King's reference to the "polarization" of upstream /downstream interests) than by objective considerations of hydrological rationality.

Thus, in the formative stages of large-scale water development, where the principal objectives are to provide irrigation, flood control and/or hydroelectric power to downstream users, the leading government agencies concerned have tended to see the management of upstream watersheds as having the primary, if not sole, purpose of improving downstream sedimentation and runoff regimes. In such a perception, those watersheds are physical entities to be 'treated' through various technical interventions, such as tree-planting and soil conservation. To the extent that they are also perceived to be populated, the conventional agency view has been that the environment should be protected from degradation by the local people through enforced forest and soil conservation measures. Meanwhile, upland areas unadjacent to major water projects have been seen as low priority, low potential areas (unless well forested) and have experienced institutional neglect, with at best thin and fragmented coverage by many different line agencies.

To begin to reverse this process over time and give the upstream areas even a vestige of the bargaining power that Hamilton and King envisage, two conditions are necessary: increasing pressure of demand for the total water resources of a river basin; and the emergence of political (often allied to financial) pressures for institutional reform. In many parts of South Asia, which has a long history of water resources development, the first condition clearly applies; yet attempts in India to promote a National Water Policy on the basis of river basin planning continues to be effectively stalled by a powerful nexus of people whose interests are best served by the continuing dominance of decision-making by Irrigation Departments in charge of heavily subsidised, construction-oriented surface water projects. Similarly in Bangladesh an excellently conceived Master Water Plan has been supplanted by a Flood Action Plan that gives primacy to large surface water control measures. So long as such conditions prevail, upland areas will remain at a severe disadvantage (Bottrall, 1992).

The new interest in upland development

The degree of downstream bias in much of South Asia may be extreme, but it is illustrative of a broader general tendency. Nevertheless, recent years have seen a marked swing of the pendulum, especially among donor agencies, towards greater investment in previously neglected upland areas. This has been accompanied by a new concern to develop alternative approaches to the management of upland areas that would benefit the very large number of poor and disadvantaged people living there as well as those living downstream (Romm, 1981). Principal reasons for the change include rapidly increasing population densities in fragile upland environments; the persistent failure of enforced conservation; and the growing difficulty of finding economically attractive large water projects without major social and environmental costs.

In some cases, where upstream/downstream conflicts have become particularly acute (as e.g. between drought-affected uplanders and sugar-cane-irrigating lowlanders in Maharashtra; or between tribal 'oustees' and proposed lowland beneficiaries of the Narmada dams), a still more radical position has been taken by environmental and other groups in support of action that would benefit the former instead of (or even at the expense of) the latter. But most of the new initiatives are not being designed, at least overtly, as a direct challenge to downstream interests.

These new concerns have brought with them a wide range of suggested strategies for improving the conditions and livelihoods of poor upland farming communities through better resource management. Many of these strategies have been 'sectorally' conceived, in that they are designed to be undertaken through programmes directed by leading line department agencies. They include programmes of rainfed (or dryland) agricultural development, often using a farming systems research and extension (FSR/E) approach; social or community forestry programmes on government-owned land; farm forestry or agroforestry programmes on private land; and small-scale (surface and groundwater-based) irrigation development. Such programmes, although geographically located within upland watersheds, cannot be included within the category of 'watershed management programmes' precisely because of their sectoral/departmental conception and organisation, the lines of which run independently across the hydrological boundaries by which watersheds are defined.

Contrasting with these programmes are others that have been conceived from the outset within a watershed/land and water management framework. They include would-be holistic programmes with the terms watershed or micro-watershed development in their titles; and others, focused in a somewhat more disaggregated way, on soil and water conservation or on water harvesting. It is on these kinds of programme that the rest of this paper focuses.

Upland watershed management: in search of principles

It is a priori unlikely that many useful generalisations could be made about upland watershed management that were not to some degree contingent on local circumstances, simply because of the great variability of conditions among such watersheds, as well as within each one. While one would hope to identify a few basic principles of good management that were universally valid, others could be expected to be highly dependent on certain key variables, including:

· The nature of the physical environment (water regime, soils, topography, production potential)

· The nature of the socio-economic environment (demographic characteristics, social structure, land tenure, market linkages, etc.)

· The range of feasible new technology options, especially with respect to the capture and use of scarce water

· The availability of existing institutions, or the scope for creating new ones, with capacity to plan and manage those new technologies successfully.

A review of some recent literature has attempted to generalise about the subject and suggests that we are still a long way from having a coherent analytical framework that would enable questions of institutional choice to be systematically addressed. Some of those who have been prepared to offer technical prescriptions across a wide range of physical and social contexts are either unclear about the kind of management institutions that would effect the desired changes or give no explanation as to why the institutions they favour would be appropriate. Others have given much more thought to the interplay of physical, social and technical factors in determining appropriate forms of management, but their observations often relate to only a limited range of micro environments and many questions about higher-level organisation tend to be left inadequately answered.

Two views from the top

Within the first category fall recent publications by Sir Charles Pereira (1989) and the World Bank (Doollette and Magrath, 1990). Pereira's book calls for urgent investment on a massive scale to protect whole river basins, especially their downstream portions, from the mounting flood and other damage being caused by upstream population pressures. In place of "rural rehabilitation schemes [that] are treating many individual [small] watersheds ... a solution can be reached only by the governments administering these threatened river valleys".

The book has the merit of dealing with all aspects of upland land use, including forestry and livestock as well as rainfed cropping and of seeking to match choice of technology to a wide range of different agroclimatic conditions; and it rightly emphasises the critical importance of water conservation and sparing but timely local irrigation in drought-prone upland conditions (e.g. pp. 98, 169). But when it comes to how the proposed new investment programmes are to be planned and managed, no clear picture emerges other than that work should be directed by some kind of government authority with the necessary technical expertise; and that the role of local people is to learn and "co-operate" (e.g. pp. 170, 209, 213). No account is taken of the extremely limited financial and manpower resources available to many Third World governments, especially in Africa; nor is there any recognition of the need to win the support of local people as active participants in the planning and management process.

If there is any dominant management model in Sir Charles's mind, it is probably the Tennessee Valley Authority (TVA)(that "classical watershed development programme", pp. 198-9). The great virtue of the TVA approach, which is of course feasible only in the context of major dams, is that it is designed to generate substantial revenue (especially from hydroelectricity) for investment in upland development - something that is otherwise in scarce supply. Unfortunately, attempts to replicate the model in developing countries - e.g. through the Damodar Valley Corporation (DVC) in eastern India and the Athi River Authority in Kenya have not been particularly successful. An important reason in the DVC's case has been the dominance of downstream interests and a failure to take the interests of poor, backward, 'tribal' upstream inhabitants seriously. A more general problem with such 'Authorities' is that they have historically tended to take a top-down, blueprint approach to the development of upland areas that is quite inappropriate to the needs of the people who live in those 'complex, diverse and risk-prone' (CDR) environments.

The World Bank publication differs from the Pereira book in several important respects: it deals with watershed development in Asia only (in fact, most of the field evidence comes from parts of India and Java); it deliberately avoids discussion of downstream effects; the focus is on a relatively narrow range of technologies and farm management practices concerned with improving rainfed agriculture through better soil and water conservation techniques; and it recognises that, "despite the existence of soil conservation agencies and management authorities, the real managers of these lands are the local farmers and villagers" (p.1). Though a useful source of technical information and ideas, the most striking feature of a publication that purports to give advice on a continental scale is the curiously partial and limited view that it offers of development options. Manifestations of this are:

· A remarkable absence of detailed discussion about the influence of different physical factors, especially rainfall, on the choice of soil and water conservation technologies

· A bias against any kind of 'structural' intervention (e.g. contour bunding, checkdams, water harvesting structures) under any circumstances (e.g. pp. 17, 71ff., 97ff.)

· A very restricted interpretation of the meaning of 'watershed development', to the extent of making it synonymous with 'rainfed/dryland agricultural development'.

The chief explanation of this tendency towards a narrow reductionist solution, despite acknowledgement of the great variability of upland environments (e.g. pp. 2, 16), lies in the supposed universal applicability of vetiver grass - the World Bank's miracle cure for all upland environments, apparently. The Bank's excessive faith in this plant, and the distortions it has introduced to the rest of its strategic thinking, are graphically illustrated in the following passage from the Doollette and Magrath publication:

"Currently, two complementary strategies for the development of conservation-oriented upland farming are evolving. The first is the adoption of a problem-solving approach aimed at identifying, on a site-specific basis, the key constraints to and opportunities for expanding output. The second, possible because of the uniquely non-site specific characteristics of vetiver grass, ... is the widespread use of this grass as a contour hedgerow".

An additional factor contributing to the limited 'rainfed agriculture' definition of watershed management (and manifested in the avoidance both of larger river-basin issues and of upland water harvesting possibilities) appears to be a 'sectoral' perspective that subscribes to the view that upland watersheds should be the responsibility of agricultural development agencies, while anything to do with water development should be left to the water/irrigation engineers. This sharp and unhelpful dichotomy is very apparent in the way most watershed development programmes have been designed in India. Thus, in the State of Karnataka the leading agency for upland watershed development is called the Dryland Development Board; and it excludes from its consideration not only large river-basin management issues but also the improvement of the numerous tanks (small reservoirs) that lie within the upper catchments and form an integral part of their hydrology. It seems that aid agencies too find it difficult to break away from old habits of sectoral and departmental thinking.

Another weakness of the publication is its perfunctory treatment of organisation and management issues. Reading between the lines, one is led to understand that there should be some kind of project authority at the "typical" watershed level (of around 100 000 200 000 ha); that the sub-watershed (5000 15 000 ha) provides the basis for a "convenient planning unit"; and that, to ensure locally-acceptable solutions, consultation should be held with farmers at the micro watershed (500 2500 ha) level. But there is no discussion of how to involve farmers as active participants in the planning process or how to build on their own indigenous soil and water conservation practices.

Alternative views from below

A very different view of possible options and strategies is offered by a number of publications that argue the need to build the development process from the bottom up, on the basis of an intimate understanding of local people's conditions and needs and of their own indigenous knowledge and practices. Among these is an exceptionally well-argued report on soil and water conservation in sub-Saharan Africa from IFAD (1992) and several others from India (e.g. SPWD, 1989; ICRISAT, 1991) that come to similar, though not identical, conclusions.

The IFAD report starts with a critique of the "dismal record" of colonial and post-colonial efforts in Africa to combat soil erosion through conservation projects that have "relied on sanctions and penalties to achieve their targets". It then discusses some more recent approaches that have been based on the recognition that "technical remedies can only succeed if attuned to socio-economic constraints" and that the participation of the resource users themselves is vital to success. This implies making use of traditional skills, working through existing local institutions, and involving the intended beneficiaries in the process of project identification, design and implementation. Because of the wide range of different environmental and socioeconomic conditions within the region, it is only by these means that the appropriate location specific technical solutions can be found (ibid).

In elaboration of this general argument, the report makes many important contributions towards a better understanding of some of the basic principles that seem to underlie good upland watershed management. The following are among the most noteworthy:

(a) A wide range of indigenous "ethno-engineering" practices can be found across sub Saharan Africa, with technical designs varying greatly in response to physical conditions (rainfall, soils, topography) and socio-conomic conditions (including population density); these not only provide the starting-point for improvement plans in particular localities but provide scope for the development of a more general typology of "ethnological options likely to be suitable in different environments (see Reij, 1991).

(b) If they are to be sustainable and replicable, intervention programmes must be low-cost: this means going for vegetative methods of conservation wherever possible, but does not rule out "engineering structures" in some contexts.

(c) Interventions must also lead to short-term, sustainable yield increases: "Conserving soil for future generations is not an argument that will convince resource-poor land users to engage in soil and water conservation. Short-term yield increases of 15-20% in the first year may not be sufficiently attractive to farmers. Net increases of 50% will be more convincing. In arid and semi-arid regions, water harvesting and moisture conservation techniques often permit such increases" (see Tiffen 1992 for evidence from Machakos district, Kenya).

(d) If farmers are to be offered any subsidy, the levels should be carefully calculated on the basis of macro- and micro-economic considerations.

(e) Such programmes require long-term commitment on the part of the external support agency.

(f) The development process should "start from below and work upwards". The most effective entry-point is at the individual farm level and the village boundary provides a better basis for local planning decisions than a strictly-defined hydrological watershed. Although "from a hydrological point of view, the catchment presents the rational and technically appropriate unit of intervention for soil and water conservation", farmers and village communities rarely perceive it as such; and the fact that it has been used in the past as the basis for enforcing 'top-down' measures leads the authors to conclude that "any return to catchment planning or conventional watershed management should be viewed with caution".

(g) Soil and water conservation work is best integrated into the larger framework of agricultural extension programmes; new organisations and separate cadres should not be created for the purpose.

One point that does not emerge clearly from the IFAD report is the basis on which larger programmes should be organised. Points (f) and (g) suggest that there is no need for a larger watershed-based organisation - that it may indeed be positively undesirable; and the implication seems to be that a fairly scattered and 'atomised' village-by-village approach would be acceptable. Given that experience in the use of participatory methods is still quite limited, it is probably fair to conclude that most of the work described is still in a pilot phase and that further thought is still required about larger questions of programming and project organisation (Reij, 1991).

Recent experience with participatory approaches in South Asia come to strikingly similar conclusions, except with the respect to the unit of local organisation (village versus microwatershed) and the basis for organising at the project and programme level. Thus, in place of a long history of past failure on the part of government schemes that have attempted to impose certain standard soil conservation practices on unwilling farmers, it is argued that the only hope of achieving sustainable success can come from first understanding existing indigenous practices and then superimposing improvements on them in dose consultation with the farmers concerned (ICRISAT, 1991). While there is agreement with the IFAD report that the easiest basis on which to gain acceptance of improved technologies is through interventions at the individual farm level "that require minimal group action" (ibid), there is also evidence from different parts of the region that, in certain circumstances, farmers can and do work closely together to achieve common benefits. Thus, in the hills of Nepal, "farmers are capable of creating and maintaining large and complex multi-member systems to achieve mutually beneficial results", including shaping field ridges in long and complex patterns, shaping terraces to suit topographical differences, and using inverted siphons to transport irrigation water. Moreover, an increasing number of NGOs are finding ways of successfully combining a village-based approach to local organisation with a micro-watershed-based work programme, through the use of Participatory Rural Appraisal (PRA) techniques (see SPWD, 1989; Shah et al., 1991). In at least one case - a project in northern Karnataka, involving a partnership between a donor agency, the state government watershed development agency and an NGO - a project structure has been created whereby a 55 000 ha watershed is being developed through a management hierarchy that is ultimately based on 37 village/micro-watershed societies (pp. 35 ff.); and other NGO networks with similar potential higher-level linkages with government agencies are developing elsewhere in India (SPWD, 1989).

The differences between the African and South Asian perceptions about the scope and desirability of organising development on a hydrological watershed basis may stem from a number of factors. The principal reasons may be significant differences in local topography (typically large, flat river basins in Africa versus steeper upland watersheds in much of South Asia) and population density (relatively low in Africa versus much higher densities in South Asia, leading to a closer congruence between village and micro-watershed boundaries). Another factor may be the greater involvement of NGOs in this type of work in many parts of South Asia and their capacity, through PRA, (a) to help villagers address common property management issues that still appear problematic in African conditions (IFAD, 1992); and (b) to empower them to deal effectively with government agencies that would otherwise be inclined to impose standardised top-down solutions.

In South Asia, as in sub-Saharan Africa, the extent to which farmers are likely to find it necessary or attractive to undertake collective action, particularly with respect to water conservation and management, will vary greatly according to rainfall conditions (see, e.g. Ray, 1986 in conjunction with Reij, 1991). Thus, in areas of very low rainfall heavy investment in water harvesting structures is common; in the 400 700 mm annual rainfall range the main reliance is on in situ moisture conservation in individual fields, unless there is also scope for groundwater exploitation; and in higher-rainfall, higher-runoff conditions (rising up to 1500 mm/year in eastern India) major returns become possible from small-scale water storages, especially if the stored water is applied to high-value horticultural crops or agroforestry systems.

In all these conditions, and especially the last, there can be little doubt that, in principle at least, a soil-and-water conservation/micro-watershed approach offers far better opportunities for substantial increases in rural incomes than more narrowly defined sectoral programmes such those based on 'pure' rainfed FSR/E. Experience with the latter in high rainfall areas in eastern India suggests that they would be greatly strengthened if they were placed within a watershed management framework that allowed them to incorporate action in support of community-based water conservation and management (Bottrall and John, in press). The same argument could be extended to social or community forestry programmes.

Some concluding propositions

Clearly there is still much to be learnt about the principles on which appropriate choices should be made both with respect to the technologies and the institutions of upland watershed development. However, on the basis of the evidence presented in this paper, it seems permissible to conclude with the following broad propositions, some of which, however tentative, may serve as useful initial guidelines for future programme development:

1. Substantial public investment in upland watershed development appears strongly justified on both equity and sustainability grounds, particularly in view of many countries' long history of neglect of, and discrimination against, the interests of large numbers of poor upland people; and in some higher rainfall contexts, the economic returns to such investment could also be very high, exceeding returns to investment in conventional downstream irrigated agriculture.

2. For the same reasons, in any larger exercise aimed at achieving reforms in water resources policy through the use of market mechanisms to reallocate water among different uses, it would be justifiable to protect the interests of current and prospective upland water users by means of substantial cross-sectoral public subsidies (though there is no reason why government should bear 100% of the investment costs, given the evidence that farmers may often be willing to contribute a significant proportion themselves - see e.g. ICRISAT, 1991; IFAD, 1992.

3. Given that water is likely to be 'the key physical resource limiting or triggering economic development' within a watershed, farmers will be most responsive to other measures, including soil conservation, if they are also combined with quickly realizable benefits from water conservation; and strategies that do not necessarily include water conservation measures (such as those employing FSR/E approaches) will be enhanced by their inclusion - provided the extra benefits are not outweighed by the extra financial and management costs.

4. Given the wide range of possible rainfall and other physical determinants of local context, appropriateness of technology must be highly contingent on those factors; and any proposed strategy that offers relatively undifferentiated technical solutions irrespective of those factors must be open to serious question.

5. Conservation measures that can be carried out at the individual farm level are intrinsically easier to accomplish than those requiring community action. In most lower rainfall (<700 mm) conditions, the need and scope for larger, community-level investments is limited, but they and the potential benefits increase markedly in higher rainfall, higher run-off conditions.

6. The complex, diverse and risk-prone (CDR) nature of upland watersheds gives a high premium to local indigenous knowledge and to participatory methods of planning and management, especially at the micro-watershed/village level. Here, external agencies will be most effective if playing a responsive, enabling role. Conversely, they will be least effective if they use a top-down, undifferentiated package approach - especially if they are the same agencies that were previously involved in attempts to enforce conservation. Any strategy based on the assumption that government agencies can directly manage watersheds is doomed to fail.

7. Government agencies have historically been organized to address issues on a fragmented sectoral basis, with a consequent tendency to miss important ecosystem linkages and an inability to support an integrated approach to watershed development. However, if they are required to promote such an approach, the conventional bureaucratic response will tend to have high administrative costs (inter-departmental coordinating committees or a new multi-sectoral agency). By contrast, farmers find it relatively easy to think holistically about the multi-faceted development of the particular micro-watershed they live in and they are often capable of planning and managing a sequence of 'multi-sectoral' activities (such as soil and water conservation, followed by modifications in tree, crop and livestock management) without recourse to complex institutional arrangements. It may therefore be possible to keep the administrative costs of integrated watershed development within acceptable limits by delegating a large part of local management responsibilities to farmers' organisations and, wherever possible, NGO intermediaries.

8. Such a division of responsibilities would allow more government agency resources to be allocated to the predominantly planning functions at meso- and macro-watershed levels without which no watershed management programme would be complete. Key functions here would be the collation and analysis of available hydrological data and use of the results to establish water allocation principles and water rights at different levels within the watershed. This could eventually provide the basis for a larger watershed organisation capable of participating in the kind of upstream/downstream planning and adjudication process envisaged by Hamilton and King at the drainage basin level.

9. A possible strategy for a donor agency to consider might thus contain the following elements:

- support for NGOs to mobilise people at village/micro-watershed levels for local-level planning, using PRA methods

- support for R&D of new technologies - for low-cost water conservation, conveyance and application; and sustainable and sometimes potentially highreturn farming systems

- help in developing a viable, low-cost (government or non-government) technical support system at a higher 'watershed' level, to support the needs of microwatershed villages, as the programme builds up

- where appropriate, creation of linkages with a larger watershed inventory and planning exercise (and/or a larger rural/district development programme)

- provision of training in new methods and techniques, including PRA, it&D/Extension (in place of conventional FSR/E), and watershed inventory and planning.

- support for monitoring, not only of impact but also of processes, with a view to wider programme replication.

References

BOTTRALL, A. F. and JOHN, K. C. (In press) Adapting farming systems research to the needs of collective decision-making. Journal of the Asian Farming Systems Association, Manila.

BOTTRALL, A. F. (1992) Fits and misfits over time and space: technologies and institutions of water development for South Asian agriculture. Contemporary South Asia, 1(2), 227-247. Oxford.

DOOLLETTE, J. B and MAGRATH, W. B. (1990) Water development in Asia: strategies and techniques. World Bank Technical Paper 127, Washington DC, USA. World Bank.

HAMILTON, L. S. and KING, P. N. (1984) Watersheds and rural development planning. Environmentalist, 4, Supplement 7, 80 86.

ICRISAT (1991) Fanners' Practices and Soil and Water Conservation Programmes. Summary Proceedings of Workshop, 19-21 June.

IFAD (1992) Sod and Water Conservation in Sub-Saharan Africa.

PEREIRA, H.C. (1989) Policy and practice in management of tropical watersheds. Westview/ Belhaven.

RAY, D. (1986) Some agricultural policy effects of encouraging water harvesting in India. Agricultural Administration, 21, 4, 235-248.

REIJ, C. (1991) Indigenous Soil and Water Conservation in Africa. IIED, Gatekeeper Series 27.

ROMM, J. (1981) The uncultivated half of India. Indian Forester, 107 (1&2) 1-21 and 69-83.

SHAH, P., BHARADWAJ, G. and AMBASTHA, R. (1991) Participatory impact monitoring of a soil and water conservation programme by farmers, extension volunteers and AKRSP in Gujarat. In:ParticipatoryRuralAppraisal:ProceedingsoftheFebruaryl991 Bangalore PRA Trainers' Workshop. MASCARENHAS, J. et al. (eds) IIED, RRA Notes 13, 128-131.

SOCIETY FOR PROMOTION OF WASTELANDS DEVELOPMENT (SPWD) (1989) National Workshop on Small-scale Watershed Development, SPWD, Delhi, India.

TIFFEN, M. (1992) Environmental change and dryland management in Machakos district, Kenya 193090: Institutional profile. Overseas Development Institute Working Paper 62.

Discussion

It was suggested that there might be a danger of institutionalising a pattern of water management which could be inflexible and rigid. A 'bottom-up' approach was advocated in which links to higher levels should be constructed; new approaches should be embedded in community knowledge and involve NGOs. It was said that the case had been presented in the paper for upland watershed management but the same conditions do not apply in the lower catchments. In addition, there are great differences between the intensive and extensive farming systems.