Strategic issues for watershed development in Asia

m% J0OL)?23o THE WORLD BANK POLICY PLANNING AND RESEARCH STAFF Environment Department Strategic Issues for Watershed Development in Asia " William B. Magrath and John B. Doolette May 1990 Environment Working Paper No. 30 This paper has been prepared for internal use. The views and interpretations herein are those of the author(s) and should not be attributed to the World Bank, to its affiliated organizations or to any individual acting on their behalf. The authors are, respectively Land Resource Economist and Senior Agriculturalist in the World Bank, Asia Technical Department, Agriculture Division. This paper was prepared as part of a program of work on watershed management that was conducted jointly by the Asia Technical Department and the Environment Policy and Research Division. That study, which will be issued in full separately, included case studies of technical, economic, social and management issues and the authors acknowledge suggestions from the authors of those studies: Jim Smyle, Augusta Molnar, Ajit Banerjee, Glenn Morgan and Ronald Ng. An important source of material for this paper was provided by discussion at a World Bank sponsored seminar "Watershed Development in Asia: Current Knowledge and Problems" held in October 1988. The input of participants in that seminar, as well as Richard Grimshaw, Gloria Davis, John Spears, Jchn English, Dirk Leeuvrik, Lawrence Hamilton, Thomas Dunne and Jack Ives, is also acknowledged. Any remaining errors or omissions are solely the fault of the authors. Environment Working papers are not formal publications of the World Bank. They present preliminary and unpolished results of country analysis or research that is circulated to encourage discussion and comment; citation and the use of such a paper should take account of its provisional character. The findings, interpretations, and conclusions expressed in this paper -re entirely those of the author and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Because of the informality and to present the results of research with the least possible delay, the typescript has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. ABSTRACT This review of watershed development arose from the realization that a number of current and planned Bank supported projects in the Asia Region deal with the linked issues of upland productivity and environment and to a lesser extent are motivated by concern with downstream impacts such as flooding and sedimentation. The data reviewed make clear that there is no single watershed problems in the Asia Region. Rather there is a complex of issues related to increasing soil loss, rainfall runoff, land degradation, irregular stream flows and poverty that can best be understood in the frameaw rk of watersheds. Four problems stand out as being central, upland poverty, low and falling agricultural productivity, high rates of erosion and sedimentation of downstream infrastructure. The report - develops the rationale for a watershed management approach by discussing the nature and significance of some of the interconnections between upland and lowland areas. Focussing on the promotion of environmentally sound on- site development the report examines constraints and opportunities for expanding output while reducing erosion. The report concludes proposing specific actions and approaches, centered on small farm development and common property management, that development agencies should pursue in their operations. STRATEGIC ISSUES FOR WATERSHED DEVELOPMENT IN ASIA Table of Contents Page No. I. INTRODUCTION............................................. 1 II. MAJOR WATERSHED PROBLEMS................................ 2 III. POLICY RESPONSES......................................... 8 IV. INTERSECTORAL LINKAGES................................... 10 V. INVESTING TO INFLUENCE LINKAGES AND SOLVE PROBLEMS....... 17 VI. INCENTIVE FOR PARTICIPATION IN WATERSHED DEVELOPMENT..... 26 VII. RECOMMENDED APPROACHES TO WATERSHED DEVELOPMENT.......... 29 REFERENCES............................................... 32 Tables 1 Extent of Sloping Land in the Asia Region by Soil........ Type and Slope Class..................................... 3 2 Nepal - Annual Growth Rate of Cereal Crop Area, Production and Yield, by Region 1970-71-1980/81.......... 4 3 Area Covered by Improved Varieties in Nepal.............. 5 4 Upland Populations of the Asia Region.................... 6 5 Total Estimated Annual Costs of Soil Erosion on Java..... 7 6 Siltation of Selected Indian Reservoirs.................. 7 7 World Bank Assisted Wateshed Management Rel.tcd Projects Under Implementation............................ 9 Boxes 1 Environmental Guidelines for Mountain Roads.............. 16 2 Terraces................................................. 20 3 Vetiver Grass-Contour System for Soil and Moisture Conservation............................................. 22 4 Problem Definition in a Micro-watershed.................. 27 Annex 1 - List of Guidelines for Watershed Development or Management........................................ 34 Figure 1: Storm Flow of the Brahmaputra ..,.................. 12a - 1 - STRATEGIC ISSUES FOR WATERSHED DEVELOPMENT IN ASIA I. INTRODUCTION 1.1 Land that can be defined as watershed I/ in the Asia region is a very significant proportion of the total. Out of a gross area of 1,700 million ha around 236 million (14%) has slopes exceeding 30% (upper watershed) and another 664 million (39%) in the slope category 8-30% (lower watershed). This 900 million hectares covering more than half the region constitutes what is conventionally accepted as the watershed area. However, it is difficult to ignore any non-irrigated land below 8% because most of the strategies dis- cussed for the steeper lands are relevant and they can usefully be treated on a watershed basis. Around 65% of the region's rural population of 1,600 million are associated with these rainfed, watershed areas. Despite the existence of soil conservation agencies and watershed management authorities in these countries, the real managers of these lands are these local farmers and villagers. Constrained by poverty and technology, their pursuit of arable land, food and fodder has profound effects on the land and water resources of both upland and lowland areas. Mounting pressure on scarce land and forest resources, stemming from rising huiman and animal populations is leading to severe environmental degradation throughout the region. 1.2 The exact dimensions of this degradation have not been accurately measured but is manifest in a number of ways. These include high rates of soil erosion and declining yield on large areas of agricultural land, reduced livestock, carrying capacity sedimentation of dams reservoirs and irrigation systems and clearance of forests with consequent loss of biological diversity and forest products (deforestation). Together these trends threaten the ability of upland people to sustain an already precarious existence. 1.3 It is misleading to speak of the watershed problem of the Asia zegion. Therc are, in fact, multiple problems, some directly amenable to solution through physical actions requiring investment, some requiring policy reform and research first, and some, principally the consequence of geology and climate, which require continued adaptation and accommodation. These watershed problems are, however, connected by the fact that they can best be understood and dealt with in the context of physical planning units defined by &M flow of water. This chapter summarizes the main themes that emerge from an analysis of watershed problems in the region and proposes a strategy based heavily on on-site development as the most technically and economically effi- cient approach to the most critical problems. It is important to recognize significant physical differences between watersheds of the large river systems of the Hindu-Kush-Himalaya region, characterizec by high rates of erosion 1/ The pristine definition of watershed acknowledges the hydrologic concept of a watershed being the dividing line between two catchments: this has devolved to include the area of land contained within a drainage divide above a certain specified point on a stream. In the context of this report this latter is intended and the land area above about 30% slope is regarded as upper watershed and that between about 8% slope and 30% as lower watershed. - 2 - linked to ongoing processes of tectonic uplift and the smaller steeper watersheds of insular Southeast Asia. The latter, which result from quite different geologic processes including volcanic activity and upheavals of the ocean floor, offer considerably different response to human activity. The review attempts to distinguish between these differences where appropriate and avoid inappropriate generalizations. 1.4 The next section discusses four general problems that stand out as being the most significant watershed problems in the region; upland poverty, low and falling agricultural productivity, deforestation, high rates of erosion, and to the extent it is relevant, sedimentation of dams, reservoirs and irrigation systems. It also assesses the government and donors adequacy of response. The following section develops in greater detail the rationale for a watershed management approach by discussing the nature and significance of some of the interconnections between upland and lowland areas. Focussing on the promotion of environmentally sound on-site development, the next section examines constraints and opportunities for expanding output while reducing erosion. The concluding section proposes specific actions and approaches that development agencies should pursue in their operations and discussions with governments. II. MAJOR WATERSHED PROBLEMS 1.5 Low and Falling Agrigultural Productivity. The uplands of Asia include a great diversity of land forms. The Hindu-Kush-Himalaya region extends from Pakistan across northern India, Nepal, China and Bhutan and includes the worlds highest mountains and poorest countries. The range is in a continuous state of formation as tectonic drift drives the Indian plate under the Eurasian Plate at a rate of 5 cm per year, lifting the Himalayas 1 cm per year in altitude. Volcanic activity, on the other hand, is responsible for the original formation and continuous change of much of insular Southeast Asia. These geologic processes, combined with intense tropical rainfall, are also responsible for the formation and high agricultural productivity of the alluvtal plains of the region. Uplifted marine sediments provide yet another facet of the Asian uplands, fragile calcareous soils. 1.6 The estimated distribution of the major soils of the region are summarized in Table 1. The diversity of soil types is often just as pronounced at the local scale as for the region as a whole. An important common characteristic of the soils of the region is their susceptibility to productivity loss due to erosion. These soils, principally the Luvisols, Acrisols, Nitrosols, Lithosols and Ferrasols, constitute neariy 75% of the arable upland area. Of the remainder, volcanics (Andosols) deep loess deposits and some alluvial soils (Fluvisols), which together account for only about 10% of arable area, have deep effective rooting depth and are highly insensitive to productivity loss from erosion. 1.7 A study of the costs to the economy of soil erosion on the uplands of Java illustrates the magnitude of these damages. Based on analysis of factors causing erosion, the impact of erosion on the productivity of different soils, and the economics of alternative cropping systems, it was estimated that erosion cost the economy $315 annually. In Nepal, overall yields of cereal fell by over 1 % per year from 1970/71- 1980/81 in the Hills (Yadav 1987) (Table 2). The contribution of erosion to this decline is not known but in the Terai where erosion is less significant, yields were essentially constant. Table .t EX79~ OF SLDPIN2 LAM D4 THE ASIA RErION SY SOL TYPE 4D SLOPE CLASS (000 ha) Papua Ne Sangtadesh Bhutan Sur~ China India Indoneaia Kapucha Korea Lao 1 Malayia N1al Ouinea Philippinen Sr; Lanka Tha land ¥.etNa TOta Peceont of tota - --- Rol l ng to H I ly (8 -30Sl Ac,1a0a 20.516 7,022 10,734 2.643 3,310 44.225 2.4 Cambioola 13,986 13,614 979 496 24 443 581 30.12 1.8 Chernoz*a 36 36 - Perrasola 10 10 - Lilhoadl 971 6.199 578 7.748 0.4 haatand:e.. 27,453 27.453 1.6 LuviMola 786 40 35.830 3,58e 3.433 452 44.363 2.6 Nitro~ola 4,964 12,224 1.144 10,170 28,502 1.7 Podzola 2,104 2.104 0.1 Arenoola 14,808 440 15.248 0.9 R~*ooi 917 * 917- Andosl a 567 1,298 3,865 0.1 Vertibola 5.993 525 6.518 0.4 X*rosolb 11.591 11,591 0.7 Yoroaola 19,330 19.330 1.1 Subtota l Rolling to hilly 5.70 92,953 91.889 26.278 8.139 24 1.741 3,88 13,603 1.033 236 240,033 13.8 Ste~ply Di~actod to Nountainous (Slopa > Acrisola 2,665 34.79 108.38 7,123 23,673 448 1, 938 168.520 9.8 Camiaola 1,863 524 9.256 4.470 14,198 13,894 1,205 5,976 20,256 154 71,796 3.0 Rendzinas 2 112 4,004 6.116 0.1 errela * $:0 5,008 0.3 Phadza 1,775 1.775 0.1 o Lithoaoa 1,138 1.610 357.244 29,042 6,349 42 47 400 3,103 691 10 1,3%5 907 402.035 23.4 Luv;eoIa 317 2.95 2,427 8,003 64 1,160 443 63 408 16.155 0.9 Oreyza 2,343 2.343 0.1 Nitroaole 12 7,684 7,777 0.5 Reagosol a 491 314 805 - Andoolö 6.379 132 1.735 1.248 9,494 0.4 Subtota l St~ply diasectad to uo~nainous 1.863 4.450 45,562 472.447 62,710 55.732 42 1,384 64 2,008 9.079 27.942 3,591 1.641 1,994 1.315 691,824 38.8 Total aloping 51,312 565,399 154,099 82,010 4,523 2,032 10,820 31,80 17.194 2.674 1.994 1.551 931.857 52.6 Total area 13,400 4,700 65,800 932.500 297,300 181,200 17,700 9,800 23.100 32,900 13,700 45,000 29,800 6,500 51.200 32.500 1.727.600 Percent Percent 8.7 10.0 30.7 i4.5 32.0 12.7 8.6 45.6 15.9 0.7 13.8 Rölling to hlly Porcent 13.9 94.7 69.? 50.7 21.1 30.8 0.2 14.1 0.3 6.1 66.3 62.1 12.1 25.2 3.9 4.0 40.0 Steoply disacted lo muntainoun Perc*nt atoping 13.9 94.7 78.0 60.6 51.8 45.3 0.2 46.2 0.3 6.2 79.0 70.8 57.7 41.1 3.9 4.8 53.9 Sourcon. PA1/t3EC0, *World SOl Ma * Total land areaa f roa World Roaources Inatituto, World Resourcsa Repot (1986). -4- jable 2: NEPAL - ANNUAL GROWTH RATE OF CEREAL CROP AREA, PRODUCTION, AND YIELD, BY RECION, 1970/71-1980/81 Cereal crop/ Hills Terai Total.Nepal- growth segment Percent t-statistic Percent t-statistic Percent t-statistic Paddy Area 1.98 8.93 0.71 2.96 0.93 4.10 Production 0.72 1.45 0.73 0.87 0.73 1.03 Yield -1.24 -3.25 0.02 0.02 -0.20 -0.39 Maize Area 0.72 4.94 -0.86 -3.99 0.24 1.81 Production -1.14 -2.30 -0.88 -1.11 -1.06 -2.62 Yield -1.84 -4.70 -0.02 -0.02 -1.30 -3.68 Wheat Area 2.08 3.41 8.50 6.60 5.82 10.18 Production 3.86 6.05 12.79 5.17 8.64 7,54 Yield 1.74 3.51 3.96 2.13 2.67 2.74 Barley Area 0.24 1.00 -3.35 -1.89 -0.71 -2.35 Production -1.61 -3.48 -0.20 -0.08 -1.19 -1.83 Yield -1.85 -4.16 3.25 2.61 -0.49 -1.26 Millet Area 0.69 2.07 -0.24 -0.56 0.54 1.78 Production -0.54 -0.86 0.16 0.31 -0.44 -0.81 Yield -1.22 -3.54 0.41 2.41 -0.97 -3.32 Total cereals Area 1.27 8.77 1.50 5.90 1.42 8.13 Production 0.12 0.38 1.44 2.30 0.95 2.10 Yield -1.14 -3.87 -0.06 -0.12 -0.46 -1.39 Notes: Computed from unpublished data on area, production, and yield from the nApartmAnt of Pond and Agricultural Mrketig ScrvicC5, Napal. 1979/80 is omitted because it was a drought year. 1.8 Investments in agricultural research and extension oriented toward upland crops has also lagged behind that for lowland areas. This is evidenced by considerably lower rates of penetration of high yielding varieties. For example, in Nepal while in the Terai HYV's account for 33% of rice, 91% of wheat, and 30% of maize, in the Hills these figures are respectively only 21, 87 and 16 and in the mountains only 6, 72 and 7 (see Table 3). - 5 - Table 3 AREA COVERED BY IMPROVED VARIETIES IN NEPAL (Percent of Total Cropped Area) (1985) Crop Mountain Hill Terai Nepal Paddy 6.0 21.0 33.0 20.0 Maize 7.0 16.0 30.0 18.0 Wheat 72.0 87.00 91.0 84.0 Potato 6.0 5.0 5.0 5.0 Source: R.M. Joshi and M.K. Khatiwad "Agriculture Handbook Nepal" 1986. 1.9 Deforestation. Approximately 19% of the region is under closed forest. It is estimated that forest cover is receding at around one percent per year and that, in addition, degradation through overcutting and grazing is reducing productivity on much of the remaining stand. In much of the region forest resources are integral to the agricultural system as sources of fodder and minor products. In India, for example, fodder available from forest and wasteland is estimated to be almost 30% of the total availability (600-620 million tons, dry matter). 1.10 It is important to understand the dimension and degeneration of the forest areas. Mostly the steeper upper watershed areas were naturally forested in the Asia region and characteristically have been over-exploited for timber, fuel and fodder. Many of the forest areas are no longer forested and in others the forest in extremely degraded. From the run-off/erosion point of view evidence shows that trees themselves have little soil conservation benefit except to the extent that they foster the understory and litter.2/ It is the forest understorey of shrubs, herbs and litter that protects the soil surface maintaining the natural higher rates of infiltration. Without this, even dense forests, as seems to be the case with some tree species, have high rates of erosion even when reasonably managed. On the other hand degraded forest with dense undergrowth of grass and shrubs with intact litter have low erosion rates. This is well illustrated in data from a volcanic soil with 17-18% slope in Indonesian where the estimated rates of surface erosion in tons/ha/year were 5 or less for degraded torest with dense undergrowth and a pristine forest with litter intact, and greater than 75 for a 40 year old teak plantation and a forest with all litter removed (Carson, 1989). In discussing the ambivalent effects of vegetative cover for soil erosion control, Stocking, 1988, points out that even at slopes of 55%, rates of erosion from undisturbed forest are usually less than 0.5 tons/ha/year whereas planting a monoculture of Eucalyptus species as an erosion control measure stifled ground cover and accelerated sheet erosion. 2/ This discussion pertains to rill and interrill erosion. At the same time it is readily acknowledged that the root system of trees are probably the best of all kinds of vegetation in providing slope stability in land-slip prone areas. -6- 1.11 Fopulation ad Poverty. The population of the upland areas is not known with precision, but it can be estimated that approximately 27 million people live in the Himalayan region, 50 million in the steep uplands of China and about 30 million in the uplands of insular Southeast Asia. Estimated population on a country basis is given in Table 4. Table- 4: UPLAND POPULATIONS OF THE ASIA REGION Watershed Population Country Rural Population Slope Class >30% Slope Class 8-30% -(------------------------- ---. -- (Millions) --------------------------------- Bangladesh 84.62 0.58 Bhutan 1.25 1.25 Burma 28.88 China 577.10 50.00 India 586.05 17.50 Indonesia 124.80 12.00 Kampuchea Korea 14.94 6.55 Laos 3.15 Malaysia 9.98 Nepal 15.81 8.46 Papua New Guinea 2.92 1.34 Philippines 41.50 18.00 Sri Lanka 12.72 2.92 Thailand 43.13 8.42 Vietnam 50.64 Total 1,598.19 127.72 Source: Bank Staff Estimates. 1.12 Again there are few reliable data to indicate whether poverty and landlessness are more acute in upland areas then elsewhere in the region. Data from Indonesia indicate that landlessness is more common in lowland villages than in the uplands of Java, but concludes that many households in the uplands are among the poorest in Java (World Bank, 1989). In the Philippines it is estimated that recent immigrants to the uplands have an average per capita income of P 2,168, well below the official poverty line (World Bank, 1989). 1.13 Downstream Sedimentation. The deposition of eroded material in reservoirs and irrigation systems are considered by many to be major watershed management problems throughout the region yet there is only a relatively small number of watersheds out of the total with these infrastructures. It is clear that sedimentation imposes a high cost in terms of shortened investment life, high maintenance requirements and reduced services. It has been estimated that sedimentation on Java alone costs the economy of Java about $24-173 - 7 - million per year (World Bank, 1989) (Table 5). Comparisons of the design and currently estimated lives of reservoirs in India have been interpreted to show that erosion and sedimentation are not only severe and costly but also accelerating (Table 6). It is generally believed that the original estimates of expected sedimentation rates were faulty based on too few reliable data over too short a period. Table 5: TOTAL ESTIMATED ANNUAL COSTS OF SOIL EROSION ON JAVA ($ 000,000) West Java Central Java Jogyakarta East Java Java Off Site Irrigation System 1.7-5.7 0.8-2.7 0.1-0.5 1.2-4.0 7.9-12. Siltation Harbor Dredging (1984/85) 0.4-0.9 0.1-0.3 - 0.9-2.2 1,4-3.4 Reservoir 9.0-41.3 3.5-16.3 - 3.8-17.3 16.3-74.9 Sedimentation Total 11.1-47.9 4.4-19.3 0.1-0.5 5.9-23.5 21.5-91.2 Table 6: SILTATION OF SELECTED INDIAN RESERVOIRS Expected Life as Percent of Reservoir Assumed Rate Observed Rate Design Life (acre-feet per annum) Dhakar 23,000 33,475 68 Maithon 684 5,980 11 Mavurakshi 538 2,080 27 Nizam Sugar 530 8,725 6 Panchet 1,982 9,533 21 Ramgange 1,089 4,366 25 Tungabhadia 9,796 41,058 24 Ukai 7,448 21,758 34 Source: Brown and Wolf (1984) - 8 - 1.14 Flooding. / Although floods are a natural feature of the lowland areas of the region, they nonetheless impose severe hardship on local population and national economies. In India between 1953 and 1976, 1,240 lives were lost, 77,000 cattle destroyed and property damage in excess of Rs 2 billion annually due to floods. In 1988, the flood in Bangladesh claimed some 1,500 lives. Flood disasters in lower basins are caused by two much rainfall (or snow melt) in too short a time for the soil and the channels to handle, and are determined more by basin characteristics, river constrictions by bridges and roads, large amounts of totally non-absorbing surface in cities, glacial outbursts, landslides, inadequate levees and increasing flood plain occupancy, rather than by forest use or even forest conversion per se. 1.15 There are no reliable estimates of the economic damages caused by flooding. In addition to direct damages, floods, or more properly, the expectation of flooding, reduce perceived returns to investment and probably slows growth to a significant but unmeasurable degree. While it is likely that the floods of 1988 were the worst on record from an economic perspective, there is evidence that the physical severity of flooding has noz worsened, 1.16 An often neglected positive effect of flooding downstream is the delivery of nutrients to agricultural land. Soil moisture stored when flood waters recede also contributes to dry season yields For example, in the dry season following the 1988 Bangladesh floods production reached near iecord levels. 1.17 Dry Season Stream Flows. A direct consequence of excessive surface run-off that contribute to flash flooding is the reduced temporary storage of water in the soil profile and groundwater aquifers. Some of this stored water would normally have rejoined the surface water and contributed to stream flow in the dry season. Reduced dry season stream flow has serious consequences on downstream uses for power, irrigation and municipal supplies. III. POLICY RESPONSES 1.18 Governments'. The governments of the region and multilateral and bilateral donors have attempted to respond to the various watershed management problems of the region in a number of ways. Responsibility for watershed management activities is typically disbursed among a number of government agencies, including agricultural and forestry line agencies and in some cases free standing watershed development agencies and soil conservation **rvices. The common feature of these administrative approaches is that they focus on the implementrtion of physical investments on public and private land quite commonly with a predominant single technical solution and on encouraging the adoption of conservation oriented farming practices on private land. Traditional low technology farming systems have frequently been, not only risk minimizing, but also soil conserving whereas cash crops are much riskier in both respects. To push cash cropping has sometimes exacerbated erosion problems. 1.19 Donor Response. Recent interest in environmental issues on the part of donor agencies has led to an increase in the level of support for watershed management projects and programs. World Bank involvement in watershed ,2/ This discussion draws heavily on Rogers, Lydon, and Seckler (1989). - 9 - management has, over the past decade, primarily been through forestry, agriculture and integrated rural development projects. To the extent that there has been a strategy towards watershed development, it has focused on attempting to improve the productivity of smallholder agriculture thereby leading to a reduction in environmental deterioration. Table 7 lists Bank financed projects now under implementation in the Region. Table 7: WORLD BANK ASSISTED WATERSHED MANAGEMENT RELATED PROJECTS UNDER IMPLEMENTATION China Red Soils Area Development Gansu Provincial Development Indonesia Yogjakarta Rural Development Upland Agriculture and Conservation Forestry InstitutioAs and Conservation Thailand Northern Agricultural Development La Philippines Central Visayas Regional Watershed Management & Erosion Control Bhutan Forestry Development India Kandi Watershed and Area Development /- Himalayan Watershed Management Haryana and Jammu Kashmir Social Forertry National Social Forestry Pilot Project for Watershed Development Nepal Second Rural Development Mahakali Hills Community Forestry Development and Training Second Forestry Marsyandi Hydroelectric LA Recently completed. 1.20 The Evolution of Conservation Tec ilogy. The pattern of investment and organizational design followed by governments and donors has failed to keep pace with the rapid evolution of soil and moisture conservation technology for tropical situations that has occurred during the last 20 years. Early approaches to soil conservation were developed for large landholdings in temperate regions and were based on structural and engineering treatments. Attempts to apply these approaches in developing country agriculture, characterized by small holdings, diverse cropping systems, extremes of topography and climate and severe limits on financial resources and skills, have proven disappointing. With rigorous attention to appropriate design criteria, engineered systems can, under some circumstances, function in tropical environments. However, experience has shown that these are usually lacking. Moreover, the high unit costs of these technologies and their indivisibility implies that these approaches are beyond the means of all but the most favored farmers and communities unless heavily subsidized. Even with - 10 - government subsidies for initial investment costs, recurrent costs in the form of maintenance and land taken out of production seem, judging by farmer response, to make these approaches uneconomic. 1.21 While recognition of the technical shortcomings of traditional approaches to conservation is growing, alternative technical and institutional approaches are still emerging. For example, although it has long been known that maintaining continuous vegetative cover is the most effective way of reducing sheet erosion, it has been difficult to promote heavy applications of mulch and retention of significant areas under permanent cover on small farms. Typically, the pressure on smallholders to cultivate all available land and to utilize all available fodder militate against maintenance of permanent vegetation. Even when individuals farmers are inclined toward such practices, the widespread practice of free grazing makes such a strategy impractical. 1.22 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 and opportunities for expanding output. The second, possible because of the uniquely nonsite specific characteristics of vetiver grass, Vetiveria zizanioides, is the widespread promotion of this grass for use as a contour hedgerow. Properly balanced these two strategies can form the basis for a comprehensive approach to agricultural technology for upland areas in the Asia region. A focus on smallfarmer development does not deny the seriousness of downstream watershed problems in the region. It is, however, a recognition of both the difficulties of reducing downstream problems and of the complementarities between agricultural development and environmental improvement. The next section elaborates on the difficulties of dealing with downstream damages through landuse changes in upper watersheds and the following section returns to the theme of strategies for small farmer development and common property management. IV. INTERSECTORAL LINKAGES 1.23 Discussions of watershed management are generally dominated by concern over a number physical linkages having to do with the movement of soil and water within drainage basins. These issues form the basis for watershed management and the significance of the hydrologic cycle for water resource planning can not be overstated. Research and project experience, however, demonstrate that there are severe limits to the ability of policymakers to alter these connections via conventional approaches to watershed management. 1.24 Often neglected in analyses of watershed management are political, economic and social linkages. While the power of these connections to alter en.ironmental quality and to minimize the often tragic downstream consequences of the physical links is unproven, it is clear that they provide an underexploited avenue to improve productivity and the quality of life of upland populations. 1.25 This section analyzes these linkages. Among the issues of concern are the extent to which changes in land use in upper watersheds can contribute to either worsening or improving damages due to flooding, sedimentation, or interruption of dry season streamflows. From the point of view of investment policy, the relevant issue is the extent to which degraded lands can be restored in order to reverse any negative downstream impacts in an - 11 - economically relevant period. The section provides an economic description of watershed management highlighting the role of externalities, technical change and its multipurpose nature. It continues with an analysis of the potential of non-physical linkages and concludes by assessing the rigidities and asymmetries inherent in watershed development. 1.26 Physical Linkages. The need for watershed management arises from the interconnected nature of soil, water and landuse systems. The watershed manager explicitly recognizes the fact that upstream landuse generates not only direct outputs such as timber, crops and fodder, but also downstream outputs such as water yield and sedimentation. In the absence of unified management, or some comparable arrangement, upstream users will adopt practices without regard for impacts on downstream residents.A/ 1.27 Types of Physical Linkages. The physical connections between upland and lowlands are manifested in two specific ways: (a) Sedimentation, the delivery of eroded material into or adjacent to waterways and infrastructure; (b) Streamflows, the quantity, distribution and timing of flows from upper catchment to lower channels. The two are intimately connected by the fact that a major source of sediment movement is provided by raindrops as they strike, and flow across, the soil surface. In addition, the ability of soil to permit infiltration and (at least temporarily) retain moisture tends to be associated with its ability to withstand detachment and transport. Another major source of sediment when it occurs mostly in young active mountain systems, subject to high rainfall, is mass wasting (syn. mass movement, landslip). 1.28 The initial movement of soil particles is termed detachment and ultimately all eroded material is deposited downhill and then downstream. However, the processes of delivery are highly dependent on the local environment including catchment size and topography, levels and intensity of rainfall, slope, soil type, vegetations and land use. The time that may elapse between the initial detachment of the soil and its final flushing out from the system is frequently measured in decades for watersheds larger than 100 km2 1.29 Sedimentation of reservoirs, irrigation systems and other infrastructure imposes significant costs through diminished investment life, reduction in output, and increased maintenance costs. In the case of Java it is estimated that the various downstream costs of erosion amount to between $24 and $173 million. 1.30 Flows of water are much faster but no less complex. The movement of water from its original landfall, to minor and major channels and ultimately A/ An additional consequence of this lack of markets is that the party that undertakes watershed management will not be able to charge for all the services it provides. Charges are feasible fcr certain products, such as timber, but experience with cost recovery through charges for reductions in sedimentation and flood control are not encouraging. - 12 - to the sea, is influenced by antecedent moisture conditions, the inherent infiltration and storage capacity of the soil, vegetation and land use. Typically, in a mild rain and during the initial phases of more intense storms, water that reaches the ground through any vegetative canopy, first enteis into sLorage in the upper layers ot the soil. A portion of this stored water will evaporate, some will be taken up by plants and transpired and the remainder will percolate to the groundwater from which some portion will be returned as surface water. As the storage capacity of the soil diminishes, water accumulates on the land surface and moves downslope through various processes of inter- or overland flow. Depending on the length and intensity of the storm, topography and other factors, this inter- or overland flow may continue to a watercourse or may end in percolation into the downhill soil. If a storm of sufficient duration occurs and soils begin to saturate or a storm of sufficient intensity occurs such that soil infiltration capacity is exceeded then several processes (surface and subsurface) deliver water to the drainage lines and streams. Vegetation affects these processes by storing water on leaf and stem surfaces, increasing surface roughness (thus slowing runoff and affording greater opportunities for infiltration), adding to the storage capacity of the soil by the presence of roots, reducing the initial intensity of raindrop impact, and to a minor extent, by immediately absorbing moisture through the root system. 1.31 The two aspects of streamflow of greatest interest are the quantity and distribution over time. Quantity of flow has obvious implications for the viability of downstream investments in power, irrigation and municipal water. Alternative landuses can have significant impact on water yields principally through the substitution of more or less moisture using vegetation. Hamilton and King, (1983) reviewed the literature on the impact of forests on streamflow. They found that forests are heavy water users and that conversion of forested watersheds to agricultural or other uses tends to increase total yield, but increase peak and streamflow for any storm event. 1.32 The distribution of flow throughout the year also has obvious implications for downstream investments and stream bank erosion, but peak flows are more importantly related to the danger of flooding. There is some uncertainty as to the effects of environmental conditions in upper watersheds on the frequency and severity of flooding, because the effects diminish as distance down the watershed from the treated area increases. As indicated above, the movement of water off upper slopes is a function of a number of parameters. Of paramount importance is the intensity of the storm event and its aeographic spread. 1.33 Disentangling the impact of upland land use from other factors in determining the frequenc-, and severity of downstream flooding has proven an elusive task. Both the quality and quantity of historical data on landuse changes and flood occurrence is so poor as to make statistical analysis impossible. The large number of variables involved similarly makes theoretical analysis speculative and somewhat dangerous. A consensus among policy analysts is now starting to emerge, however, that suggests that agriculture and forestry in upper watersheds play a relatively minor role in exacerbating the effects of major catastrophic flood events (Hamilton, 1987, Ives and Messerli, 1989, Rogers, Lydon and Seckler, 1989). 1.34 For example, Figure 1 summarizes data on annual runoff, sediment load and high- and low-flows for the Brahmaputra for the period 1955-1979. It -12а- ...�..._. ю iw �� . 1• i1! : � , �• в i i � tя яМ у {г � 3: � �. � и а� � ��, � 8 11 10{1 ! ♦ � �� ♦ �$ и вМ � ��. tt {1 1о0 ! '�i 1wd вивlя � и �а _.. я � я.вир им I1яr1 r.• � � � 1вЧввМ01вМ � 1 1t 10я F 4 � �вд в..,.i 1У • �..��...� --- lяsо {яяв гоео lягв ию и1s lяао р NlGN ►twMyY�11 ю � да �ю � -� =а ♦ ь ю �. .._._�.._.._._..� _._.__.._ ...._ �... , lяаs txo 1яа говs и{о 1яа, tв» я �а nда {nа{оымРм :� ' � ' �`�-%� �-,. а � '� .-+ а �/ � � " в - - - ------ -- ' ' ' 1 иео {sчб �яа �яs �яю �ям �мя Figure 1; 5Lок�а F7.aws of the Втаhтариtэга (,1.955-19b91 Source: ives and Mesaerli . 13 - is clear that if the period 1955-1969 is excluded, the data may indicate a deterioration in environmental quality.V However, considering the entire period shows no definitive trend. Moreover, the data on high and low-flows are not consistent with the generally accepted expectation of land degradation causing higher peak flows and lower dry season flows (for details see Ives and Kesserli, pp. 136-137). Data on the incidence and physical severity of floods in the Ganges-Brahmaputra delta do not support the hypothesis of a trend toward worsening floods. It is likely however, that concern over flood damages is growing as a function of greater economic activity in flood prone areas..k/ 1.35 Elosion-Strgamflow Interaction. As noted the tendency of some soils to resist erosion is often associated with their infiltration and storage capacities. In addition, land management practices to reduce erosion frequently have the effect of reducing runoff (see Chapter II). Another way in which the two may interact to exacerbate the danger of flooding is via the accretion of riverbottoms. Diminished channel capacity lowers the magnitude of flow required to cause flooding and may contribute to the frequency and severity of flood damage. 1.36 Waters_bgd Mamagement Investments. In practice, in addition to environmental protection, most watershed management projects have multiple objectives. In many circumstances it is possible to improve the environment and increase the output of goods and services at the same time. This is particularly true when a project also averts long term losses due to produczivity declines. This aspect of tradeoffs in watershed management is frequently misunderstood. 1.37 The fact that watershed management projects can accomplish more than one objective has important policy implications. Generally, to resolve an externality problem tax or subsidy schemes are recommended to "internalize" the externality. Farmers might be given subsidies to adopt soil conservation practices, or taxed if they fail to adopt. Subsidies are the more frequently used approach and their budgetary cost is justified on the basis of reduced downstream damages. However, when watershed managemAnt practices !ake both upstream and downstream residents better off and leave no one worse off (are said to be Pareto improving) this simple justification no longer holds. Some subsidies way, however, be justified on other grounds, such as an inability on the part of poor farmers to finance purchased inputs, or wait for the maturation of tree crops (i.e., capital market imperfections)..Z/ 5J The increase in run-off could be attributed to changes in weather or more glacial melting. See for example Kumra and Rao (1985) who found that the value of agricultural flood damage has been falling in Bangladesh while non- agricultural damage has been increasing. .Z/ It is conceivable that situations will arise where there are real and significant conflicts between alternative uses. This situation was first analyzed by Gregory (1955) and (1957) such conflict, and the tax- cum-subsidy schemes it suggests, are more relevant to developed country, low population density watersheds. - 14 - 1.38 In order to have this property of being both multipurpose and Pareto improving, watershed management projects must either be able to exploit one other property of unimproved watersheds, namely, technical inefficiency, or must introduce technological change. Typically projects do both, promoting technologies that are already practiced on a small scale, introducing technologies used in other places and developing entirely new technologies. 1.39 Economic Linkages. While physical interconnections are the basis for concern with watershed management, a strategy that takes advantage of social, economic and institutional linkages provides the greatest opportunity for success. Upland areas have such critical connections with national economies as being: (a) Sources of Raw Materials. Despite often difficult conditions upland areas often possess a comparative advantage in the production of certain commodities. In much of the Asia region timber and grazing represent the primary resources with potential for export to lowland areas. However, in the case of Indonesia (Roche, 1987), upland areas may have advantages in specialty crops such as clove and high value horticultural crops such as fruits and vegetables. A significant feature of many of the agricultural outputs for which upland areas are well suited is that they often possess a high income elasticity of demand. Consequently, income growth in lower areas can contribute to strengthening linkages between uplands and lowlands while at the same time providing opportunities for expanding sustainable production. (b) Sources and Sinks for labor. Upland areas in the region were historically, with some exceptions,j/ relatively sparsely populated. Population pressure in more favored environments has resulted in increased migration to the uplands. Countering this has been growth in nonfarm employment opportunities in urban areas. Seasonal employment opportunities in lowland agriculture and urban areas are increasingly important contributors to upland income.2/ (c) Markets for Downstream Production. Because of low incomes and high transport costs upland areas have generally not provided major markets for goods produced in lowland areas. In those areas where incomes have grown and where infrastructure investments have reduced transport costs upland areas do constitute significant markets. 1.40 Political Linkages. Upper watersheds, in addition to being physically remote, are often politically remote as well. The attention of national policymakers is naturally drawn to the concerns of urban and more affluent lowland agricultural populations. To the extent that developments in upper watersheds are a major item on the national agenda, it is because of their impact, via the physical linkages mentioned above, on the well being of V/ Most notably Nepal where the prevalence of malaria inhibited settlement of the Terrai. 2/ Shifts in the opportunity cost of labor brought about by changing opportunities off the farm also have important implications for farmer interest in adopting more intensive soil conservation measures. . 15 - downstream grouips. Political forces further bias policy and investment against upland areas in other ways. Watersheds are physical units that frequently do not conveniently overlap with administrative boundaries. Although the downward flow of the physical consequences of landuse do not observe these boundaries, limits on the ability of government agencies to transcend them are severe. 1.41 Asymmetry and Rigidities of Linkages. From the perspective of investment analysis and policymaking the most important aspect of the various linkages described in this chapter is their susceptibility to influence. It is clear that there are significant rigidities and asymmetries that mark the response of linkages within watersheds and which consequently constrain the strategies and objectives of watershed management. 1.42 One Way Flows of Physical Linkages. The most obvious illustration of these asymmetries is the one way flow of soil and moisture in watersheds. This implies a role for public involvement. 1.43 Temporal Asymmetry. As noted the movement of sediment in watersheds may involve extended periods of time. A corollary to this is that remediation works may also involve long lags. Pearce (1986) estimates that the time required for sediment to be flushed from drainage systems of large basins may exceed 40 years. In a present value sense, any reductions possible from the adoption of watershed management practices will be of minimal economic significance. However there are downstream benefits to be achieved within reasonable time periods from conservation land treatments as the focus moves upstream to smaller watersheds. These relate to both sediment and flooding in non-major storm events (Hamilton, 1987). 1.44 Geologic Erosion While in principle, it is possible to totally stop erosion, the underlying potential of soil to move is frequently so great as to preclude stopping erosion at any reasonable economic cost. This underlying rate of soil loss is termed geologic or natural erosion and provides a partial but useful guide to the design of soil conservation policies. Much of the steep uplands of the Asia region are naturally prone to erosion due to their geology. One particular source of sediment that is largely subject to control is roadside erosion (see Box 1). - 16 - Ao.l I ENVIRONMENTAL GUIDELINES FOR MOUNTAIN ROADS Rural development program will often include provision for the development of roads in mountainous areas, for general access into an inaccessible area, uu V&uvlde links across the mountain, or as part of mining and forestryllogging activities. General access: These roads are built to cross the mountain, so as to access the region beyond, provide a link between two regions, or access mountain communities and resources. Such roads are generally used by heavy traffic as well as passenger vehicles, usually buses. They are often built for the long term, with the expectation of future traffic growth. Mining/Logging access: These road are built for specific forms of vehicles, particularly heavy machinery. They are also built for access into surveying and/or new work areas. Usually they will be built with a finite working lifespan and should be developed so that as each road/tract falls into disuse, it can be reseeded, or reclaimed by the natural environment. 1. Erosion Erosion is one of the major problems faced in mountain road building. Cutting of land for road penetration helps not only trigger off landslides, but also accentuates problems of erosion, through changing the profile of the mountain and removing natural, protective vegetation cover. As angles of repose are accentuated, soil and rook slipping, or water erosion can become a problem, leading to siltation of waterways, and deterioration of the mountain environment. Unless preventive measures are built into the project, it is also possible that the road or track may be rendered unusable because of landslide damage. Erosion prevention measures should be part of road design and must be implemented at every stage of road building. They may include embankment reinforcement, plantation of appropriate trees on road banks and the use of steppes to maintain angles of repose. Measures chosen must be appropriate to local constraints and should be easily maintainable. It will be necessary to include upkeep as part of road maintenance training for local crews. As possible, the road bed should be cut in areas having least potential for erosion or for landslide damage. However, roads should not be built in river channels, as they will then not only exacerbate siltation problems in the river, but help form conduits for torrents, which in turn will accelerate erosion, increase siltation, reduce the usable lifespan of the roadway and further destroy local bio-systems and eco-niches. The negative effects of erosion, landslides, siltation, from the building of roads in or near mountain river channels, can be minimised by early recognition and awareness of the potential for problems. It is important to maintain flexibility as to the location of the road(s), and determine through on-site exploration whether other locations may pose more opportunities and less constraints for road building. 2. Torrents Another factor to be considered in the building of mountain roads is the likelihood of the chosen roadbed becoming a route for torrents which could cause severe or bank erosion, heavy siltation in downstream river beds, and possibly destroy the road as well. If a road must be cut through a torrent channel, reducing the channel's width, damage potential will be increased. Embankments will need to be built to protect the road and the river. They will have to be Seared for maximum expected torrent capacity. Culverts and other drainage systems will need to be built to lessen the possibility of surface or base erosion caused by overflowing waters. In general, it is considered sound policy to build roads as far removed from river beds as possible. Source Excerpted from Enironmental Guidelines Rural Roads, United Nations Environment Programe, Nairobi 1986. - 17 - 1.45 Asymmetry of Policy Linkages. Policy linkages between uplands and lowlands have not received the attention they deserve. Unlike physical linkages there is likely to be greater two way influence in this area. Theoretical analysis of farmers incentives for soil conservation, however, have been largely unable to demonstrate how these linkages operate. Barbier (1988), for example, modeled farmers decisions to implement soil conservation using an optimal control framework and showed that price and other agricultural policies could play a significant role in determining privately profitable soil management strategies. Barrett (1988) using a similar approach found price to be unrelated to soil management.1Q/ 1.46 Roche (1987) taking a different approach analyzed the impact of policies and growth patterns on upland land use in Indonesia. He notes that rapid lowland income growth, due in part to the successful intensification of irrigated rice production and industrialization, operating through high income elasticities of demand for vegetables and fruits, has created an incentive for upland farmers to shift to less erosion intensive cropping patterns. Observations in upland areas provide further indications that in areas with access to good markets and particularly demand for meat and livestock products the incentive for establishing and maintaining permanent vegetation is strong. The aggregate environmental impact of these incentives has never been assessed. 1.47 Hyde (1988) explored the consequences of unemployment in lowland areas of the Philippines on the environment of the uplands through a general equilibrium model. The model which allowed migration as an equilibrating process showed that tax policies, which in the aggregate favor capital have a significant influence on migration to upland areas. Similarly rice subsidies were found to increase both the agricultural labor force and the upland population. Trade policies that would encourage exports were found to have a positive impact on the uplands via an expansion of lowland industrial employment. V. INVESTING TO INFLUENCE LINKAGES AND SOLVE PROBLEMS 1.48 Taken together the data reviewed on watershed linkages suggest that, although physical connections shape the environment for investment planning and policy interventions, there are marked rigiditit-a in hnth time and space. These rigidities limit the scope for economically viable investments aimed primarily at resolving off-site and downstream problems. Fortunately, as discussed in this section, there is ample opportunity for directly productive investment in upland areas. Most of the approaches that would fit into such a strategy are also consistent with the long term objective of preventing or ameliorating the downstream consequences of watershed deterioration. 1.49 The essential elements of a strategy for upland development are the same as would apply in low land areas and include the need for a DQ/ Related models have been presented by McConnell (1983), and Bhide, Pope and Heady (1982). - 18 - positive incentive framework and availability of appropriate technical innovations. In contrast to lowlands, upland areas are characterized by much greater diversity, less amenable to large scale investments (especially irrigation), and generally face runoff and soil erosion problems. Accordingly, strategies for upland areas require greater emphasis on generating a capacity for site specific recommendations, and specifically a focus on improving rainfed agriculture through low cost methods of soil and moisture conservation. Due to the greater reliance of upland farm households on non-arable land such as forest and communal grazing land farm development strategies in the uplands also need a broader focus than the strict commodity cropping focus of lowland programs. Technologies and Techniques for Improving Upland Agriculture 1.50 Although investments in the development of agricultural technology for upland areas lags far behind that for lowlands, the general principals for increasing yields are well known and there a large number of specific interventions which can now be recommended for specific applications. The two key constraints to improving agriculture in upland areas relate to soil and moisture conservation. 1.51 In practice productivity decline due to soil erosion is related to the following soil characteristics (which are discussed in more detail in Chapter II): rooting depth, water reserves available to the plant, distribution of plant nutrients in the soil profile and the chemical/physical properties of the subsoil. Of primary importance for design of an upland strategy is the connection between soil moisture and erosion. In tropical soils, water use efficiency (kg dry matter produced/liter of water use) can be reduced to less than half that of uneroded soils with progressive topsoil removal up to 35 cm despite high rates of fertilizer application. General Aproaches to Enhancing Upland Agriculture 1.52 Basic improvement in agronomic techniques, improved varieties, higher quality seeds, improved pest management and tillage practices, often provide the best opportunities for increasing agricultural output. A key technique that is integral to rainfed agriculture is contour cultivation. Contour cultivation and ridging across the slope have prndiuad yield increases from 6-66% on 3-32% slopes compared with the traditional up and down slope cultivation with further increases if other treatments such as mulching are combined. The evidence to support the general recommendation that all rainfed cropping activities, annual or perennial, be carried out on the contour is overwhelming. 1.53 Recent and less frequently the subject of published research is the application of cropping/farming system technology to on-farm soil conservation. It is motivated by the acknowledgement that social, economic and tenurial factors influence farmer's ability and willingness to adopt and maintain soil conservation measures. Farmers, especially poor small holders, need direct short-term benefits from any innovation they introduce into their farming systems. Soil conservation measures have apparently been seen as investments that do not do this and indeed frequently take away from the farmer's limited resource base by - 19 - demanding space for banks and water disposal structures. By contrast their neighbors respond rapidly to investments that have immediate benefit such as the case with levees required from rice paddies. 1.54 Farming SyStems isCtsea:0-h hat, uuL n u"UUmly Lwn apple.%J Lu suil conservation. Yet attempts to implement physical conservation works and land use planning are usually frustrate4 by lack of acceptability. It becomes important, therefore, to find points within the farming operations where practices that meet soil conservation objectives and increase incomes without unduly increasing risks can be introduced. Specific Techniques 1.55 While there is clearly a need to design a package of conservation and yield increasing interventions to be consistent with the needs of a specific site, there are a number of generic approaches which have widespread usefulness as well as the potential for misapplication. These can be grouped as vegetative/cultural and structural. Current conservation practices in the Asia region focus on structural approaches and there is a need to assess the potential for fuller utilization of alternatives. Chapter II provides a fuller discussion of the evidence on these groups of techniques but there are a number of aspects which are relevant to their place in a strategy for upland developments. 1.56 Structural Treatments. Structural treatments, earthbanks, land leveling, and terracing have been applied extensively in watershed projects throughout the region. Experimental and project level result with structural measures have been mixed but generally poor. These observations may seem inconsistent with the fact that terraces, in particular, are a widespread and integral part of the agricultural landscaping of the region (see Box 2). Hcwever, several features of structural approaches may account for their generally poor performance: (a) A major obstacle to the widespread use of structural approaches to soil conservation is their high unit costs. Costs for terracing in Indonesia are estimated to range from $400 to $1,000/ha. Construction of earth bunds in India is estimated to cost between $23 and $150/ha depending on soil type and slope. Aside from high initial costs and the financing burden they impose, structural techniques inevitability require high levels of maintenance. Failure to properly maintain structures can lead to their total failure and can actually accelerate soil loss. (b) one source of difficulties with structural measures is the widespread failure to adjust designs and standards to accommodate the engineering properties of particular soils and local rainfall patterns. The inherent instability of some soils, can result in massive failure of structures. For example, saturation of the topsoil over relatively impervious subsoil results in slumping. (c) Structural measures generally operate on the principal of slowing water. Water is usually directed along field boundaries toward natural drainage ways. Drainage ways need to be large enough to accommodate peak flows, otherwise they will be overtopped damaging the adjacent field or the drains themselves will fail. Moreover, the . 20 - Bpx Z: TERRACES Bench terraces are part of the landscape of the Asia region especially S.E. Asian countries, China and the Philippines. This discussion concerns reverse sloped bench terraces and outward Sloped bench terraces used in steep uplands of humid and semi-arid regions respectively to change the slope of the land in order to increase the area that can be cultivated "safely" and "control" run-off. Level bench terraces (irrigation type for rice paddy) do not concern us nor do conservation bench terraces used in arid regions to harvest rainfall on part of the slope and direct it to a level bench. Whether or not these bench terraces represent the best treatment for the respective locations in terms of land use capability criteria or meeting the needs of the population is essentially academic since they exist and future land development programs should start from this reality. The review of research into the effectiveness of terraces in respect of sediment yield, run-off and productivity (Chapter II) highlights widely divergent results. This corroborates consistent observations. Although terrace technology is well understood and engineering design readily available it is frequently poorly applied. There are three common shortcomings. The first is failure to relate soil type - rainfall characteristics-cropping pattern to design. The next shortcoming has to do with recognizing that a water disposal component of terrace system is integral; farmers are reluctant to lose the 3-5% land required, and compromise in design increases run-off and damage. Thirdly, farmer operation and maintenance is generally shoddy compared with level benches used for paddy which implies a questionable economic situation. Developing a program for correcting poor terrace starts by defining the treatment options. Carson (1989) discussed the limitations of terraces in several agroecological zones in Indonesia, and suggests some appropriate soil conservation strategies. The key issue most likely is finding a farm production system attractive enough to induce the majority of occupants to become involved and finding a way for them to convert to it. This indeed could correct the disparate performance of upland terrace farmer compared with those managing paddy terraces. After this the treatment options can be laid out with priority going to vegetative - cultural measures which should be cheaper to implement and maintain. For example, if the horizontal grade exceeds 1% instead of further earth moving with it inherent costs judicious use of a vegetative barrier such as vetiver grass would induce natural and rapid levelling and at the same time control the effluent point which in turn would change the dimension of waterway rehabilitation. Only then would structural treatments be considered for problems still without solution. Although sensitive to soil type and rainfall characteristics it is generally accepted that 60% is the upper limit of slope for any sort of terrace. Beyond this riser height and width are too great, bench width to narrow and the net arable area down to about 50%. - 21 - natural drains may receive more runoff than they are capable of safely handling, resulting in a danger of gully erosion. The planners incentive to design drains to accommodate peak flows runs counter to farmers' desire to minimize land take out of production. This usually results in no drains or undersized drains prone to failure. (d) Construction of soil conservation structures usually entails earth movement that exposes infertile subsoil. This reduces yields in early years of the structure and amounts to an additional construction costs. 1.57 Research has added very little to traditional farmers understanding of the potential use of structural measures. Detailed analysis of terrace designs and maintenance in Nepal, for example, has shown that use of outward sloping terraces is an effective means of allowing surplus water to move off the terrace while causing minimal surface erosion. Attempts to reduce runoff by introducing backsloping terraces resulted in collapse of the "improved" terrace due to concentration of water. Similarly the epparently low levels of maintenance and poor quality construction of terraces supplied by projects in Indonesia may reflect the interaction of farmer perception of the dubious value of terracing and the attractiveness of the assorted subsidies and incentives provided. In the absence of compelling evidence that a significantly new and attractive on-farm structural technology can be suggested to farmers, there seems limited justification for the central role they now play in watershed development projects. 1.58 Vegetativefultural. Vegetative/cultural measures to improve upland agriculture include contour cultivation, techniques to reduce tillage, addition of new crops and changes in timing or cropping pattern (inter- cropping, etc.) or stand architecture to provide for more continuous and effective soil cover. In some cases the use of vegetative treatments is intimately mixed with cultural practices, such as with contour cultivation with grass strips, while in other cases vegetative measures stand alone as in the establishment of permanent cover. 1.59 Vegetative measures have been shown to be highly effective in minimizing erosion by reducing the impact of raindrops as they strike the soil. Mulches, certain agroforestry options and permanent cover crops can perform this function. 1.60 Plants can also be used to form a physical barrier to slow runoff and assist already moving soil. Workers have, for a long time, been looking for suitable species and a number of species have been proposed for use in this manner including napier grass, vetiver grass, and the tree species Laucaena. The utility of different species in this capacity will vary, depending on circumstances. However, the particular features of vetiver grass, discussed in detail in Box 3, make it particularly well suited for this application. - 22 - Box 8: VETIVER GRASS - CONTOUR SYSTEM FOR SOIL AND MOISTURE CONSERVATION The notion of carrying out all farming operatons especially cultivation and planting on the contour in any rainfed situation, on any slope, for any crop is overwhelming. Customarily, barriers are constructed on the contour at certain vertical intervals according to slop to break the length of the slope so as to check the velocity of run-off water and trap silt. These also serve as guidelines for contour cultivation. Graded earth banks (syn. bunde) usually with a horizontal gradient of up to 1X to feed excess water Into a prepared waterway have been employed extensively. These structures have severe limitation in the tropics and do not fit small holdings due to loss of arable area for the bank Itself and the waterway. Vegetative barriers on the contour have distinct advantages over earth banks, namely, a vegetative barrier require. about one tenth of the space and no water disposal system Is necessary, a vegetative barrier slows down surface run-off and causes It to deposit the silt load while the water seeps through spread out, with increased opportunity to infiltrate. This also avoids the problem of waterlogging common behind the bank. A plant suitable for a vegetative barrier requires particular morphological characteristics. Its root system should be aggressive and deep without rhizomes or stolona so as not to spread out of line; the crown should be below the surface for protection against fire and over grazing; the culms tough and unattractive to animals and pests; and the flowers if any essentially sterile so as notto permit spreading by seed. The plant should be a perennial and persistent, tiller fro*ly and intermingle with It. neighbors (some clump grase do not). To date the only plant know to meet these criteria Is vetiver grass, (Vetiveria sizanloides). It has an extremely wide range of climatic conditions over which It Is adapted and further exh ibits adequate growth over a wide range of soil types including those with what are highly unfavorable properties for many plants. Vetiver grase has been used for this purpose and as permanent field boundaries for a long time and hence it is know to persist Indefinitely, once established, without maintenance. It is propagated by root slips which the farmer may plant himself on a roughly surveyed contour lines. Given moderately favorable conditions the hedge would be complete after three growing seasons, less with high fertility, high rainfall and close planting. Apart from physical advantages, the system Is cheap to establish and maintain and can be managed entirely by the farmer. Compare this with the engineered system which requires earth moving equipment, complete cooperation among neighbors especially for water dispeal components that impose Intolerable burden on those further down the slope, and regular re-building every three to five years. Vetiver grass has other applications, due to its unique morphology. Among these are, protecting paddy banks, dam catchments and drainage lines from siltation, roadsides and stream Cnks from erosion and it performs the soil and moisture conservation function when planted in V-ditches with fruit and forest trees. The same pr!nciple applies to the stabilization of degraded non-arable lands which shrubs which might be coppiced for feed on fodder could be used as barriers on the contour. The search for suitable shrube continues. - 23 - 1.61 Vegetative systems, based on whatever species, have a number of advantages over structural systems: (a) Post. Vegetative measures for soil conservation generally can be promoted at low costs. In many cases the major cost of promoting these measures is for extension advice. Costs for vetiver grass hedgerows in India are estimated to be $18/ha. (b) &Atability. Unlike structural measures which require detailed engineering and site planning, vegetative approaches are relatively insensitive to issues such as proper alignment on contours, irregularities in field boundaries and minor errors in placement. (c) Farmer Controlled. Because vegetative methods are relatively inexpensive and do not require use of machinery or sophisticated surveying, individual farmers can take the initiative in adopting conservation measures. An indigenous system of contour alley cropping using bands of L&ucaena has been used widely in the steep lands of Cebu in the Philippines. A particular advantage is that the cropping area sacrificed to the conservation measure is considerably less than with the typical structural approach. This is especially true with respect to grass contour hedgerows. Farmer's willingness to devote arable land to essentially permanent cover is often largely dependent on the degree to which livestock are integrated to the farming system. Investing in Non-Arable Areas 1.62 A large proportion of a typical watershed anywhere in the region is non-arable in the sense of not having the capability for arable agriculture by virtue of soil or slope characteristics or for historical reasons. The consequences of run-off/erosion on these non-arable areas are significant in just the same way as the arable areas. Productivity is lost, sedimentation and local flash flooding are increased and dry season stream flows reduced. In terms of these latter three factors nonarable areas are the more significant contributor. 1.63 Whereas landuse and ownership are generally less complicated in the so-called arable areas, namely, crop based agriculture and mostly private, in the non-arable areas divided among forest, grazing and community lands and variously owned by government (mainly forest department), communally and privately. This two dimensional split on non-arable land is further complicated by dejure and defacto rights of access for both landed and landless rural families. This diversity of use and ownership has exacerbated the degradation effects and makes treatment more difficult. 1.64 From the perspective of restoring productivity, the significance of removing vegetative cover in terms of its radical effect on soil moisture status has to be recognized. Removing the understorey and litter results in reduced rates of infiltration. This is particularly significant in the wet- dry tropics and the semi-arid tropics where the effective rainfall may be no more than half the actual by virtue of run-off. In order to redevelop these areas something has to be done first to restore the soil moisture status. Contour vegetative hedge treatments improve infiltrations. Reduced livestock numbers reduces soil compaction as would reduced use of heavy equipment in - 24 - forest harvesting. Controlling fire, which induces water repellancy in some soils, can also assist. 1.65 Redefining land use become the important next step. While it may be the most advantageous use of land and the best soil conservation strategy to try to return bare land or degraded forest areas to the original mix of species, other options could include closed mixed species forest, single species plantations, fuelvood plantations, silvi-pastoral plantations or pasture. It is not within the scope of this study to develop guidelines for determining which option might be employed under any given situation. At the .ame time, the importance of doing so ought not to be underestimated since it takes account of the needs of the population and may ultimately determine the success of the investment. Treatment of Forest Areas 1.66 Stabili;ation. Denuded slopes in non-arable areas need to be stabilized first before any of the options are applied. This cuts down run- off, increase infiltration and trap erosion products simply by creating vegetative barriers on the contour at approximately three to four meter vertical intervals. This stabilization technology can be applied on village common land, grazing land and wasteland, as well as forest land. The vegetative barrier should ideally comprise indigenous, locally adapted shrubs that are unpalatable, deep rooting, easily propagated and capable of forming a dense hedge, planted into a V-ditch or trench. In the absence of suitable shrubs, vetiver grass would form a suitable hedge in most circumstances in the region. The cost of of establishing this treatment is likely to be on the order of $15 a hectare. It would be applied regardless of what the interhedge spaces might later be used for. 1.67 Revegetation. Artificial forestation is the option most commonly applied probably because most areas find themselves under the jurisdiction of forest departments whose mandate it is to plant trees and manage plantations. Success rate is low and the costs per hectare high which has brought the tech- nology into question. The questions are valid but only when the issues of stabilization/soil moisture status and land use are resolved. Chapter III reviews the methods of revegetation that are presently practiced in Asia. There are key shortcomings which have to do with selection of species and quality of planting materials, land preparation, methods of planting and planting geometry, protection and management. A serious non-technical shortcoming has been that forestation has been carried out without the support or agreement of local people who may customarily harvest some resources from these areas. The technical shortcomings are well understood and little or no additional research is required to be able to grow most tree species successfully. Addressing all these issues so as to do everything well, however, results in costs per hectare in the range of $500-1,000. This is generally too high to be replicable over wide areas. The Problem Solving Approach 1.68 It is know beyond doubt that removing the vegetative cover causes accelerated erosion so that if over time an undisturbed vegetative cover can be recreated then the problem is solved. Unfortunately, - 25 - (a) time may not be an option, in which case intervention is required to accelerate recreating the vegetative cover, (b) natural regeneration may not be an option, since the degradation process has changed the soil moisture and nutrient status so as to require intervention, and (c) an undisturbed vegetative cover may not be an option, as witi land required for arable agriculture, grazing, fuelwood or fodder pro,'ue- tion, in which case vegetative and cultural, farming systems, st ur- tural or a combination of all these interventions are required. 1.69 The issue is how to decide which approach to take and how to avoid the very common approach which consists of a single solution in search of a problem. What is required is information specific to the site and a clear understanding of what impact is required. There are two main questions. The first question is one of scale: the size and definition of the planning unit. The larger the planning unit the greater the heterogeneity in terms of actual land use, land capability, microclimate, soils, vegetation and people. The larger the size the more likely that one or two widely applied solutions will fail. The second question relates to what must be achieved. If the objective is simply to reduce downstream sedimentation, then this might be achieved by a simple technique such as a check dam, but this would have no effect on erosion induced productivity decline on the arable areas. The objectives are rarely simple and hence invariably require a set of solutions. 1.70 The folly of a Single Solution. Terracing as a treatment is a good example since it is a common choice in the tropics and done properly can be effective. Whether or not terracing is a sound proposal depends on the type of terrace in relation to rainfall, soil depth, drainage and structural strength, land ownership patterns, crops to be grown on the terrace, farming systems and quality of the extension services. Terraces of themselves do not necessarily conserve soil or moisture, improve productivity or decrease sedimentation. Poorly farmed terraces may result in greater degradation, inappropriate types of terraces or terracing inappropriate soil types may result in reduction in productivity and accelerated soil loss. Even when properly used terrace technology is only part of a system. This is true of all soil conservation interventions, there is no one technology that applied in isolation will achieve a soil conservation benefit on anything but a very small scale. 1.71 The Importance of Planning at a Micro Level. Selecting appropriate on-site soil conservation practices is an area where it is possible to construct guidelines that can be easily followed at a local level. At a regional planning level it is more difficult because of the increased heter- ogeneity and the quality of the information available. Information on where and what circumstance various soil conservation technologies are applicable is available now. Whether or not this information can be used depends on the quality of the information from within the planning area. Given the import- ance of scale and in light of the linkages ascribed to economic functions it is clear that a planning unit has some optimum dimension. There are advantages and disadvantages in using a hydrological (physical) unit, an administrative (political) unit or a set of villages (social unit) and cases can be made for each of these. Since villages tend to be located close to drainage lines so as to be able to exploit lower arable and higher non-arable - 26 - lands, the boundaries of the village's area of influence frequently coincide roughly with watershed boundaries. As a general rule the hydrological unit is preferred. A typical watershed (100,000-200,000 ha) comprises a series of subwatershed (5,000-15,000 ha) which in turn are made up of five to ten microwatersheds (500-2,500 ha). Experience seems to suggest that the subwatershed as it is defined here is a convenient planning unit but that the plans constitute an aggregate of the information from the constituent microwatersheds. 1.72 Having avoided the single solution trap, the approach in developing the conservation strategy is to define the problems that are evident in the subwatershed and to define the objectives. Since for very large watersheds there is no immediate linkage between erosion in the upper catchment and downstream sedimentation, the most supportable programs are those whose objective is to raise farm incomes and increase sustainability on-site on both arable and non-arable land in the upper watershed. 1.73 Problem definition necessitates quick tabulation through intaractive planning with the population employing such techniques as rapid rural appraisal to tabulate the problems (see Box 4). In a short time it would be possible to deal with the population in terms of their dependencies, needs and aspirations, their perceptions of the dimension and causes of degradation and the microwatershed itself in terms of the land classes (arable, non-arable, private, village, public) and the respective needs for treatment and by this method to come up with objectives, a strategy and action plans. 1.74 The Menu of Solutions. Within any subwatershed there will be a number of treatments which are likely to be appropriate ways of addressing the problems that have been defined in the interactive planning process and which can address the objectives. These treatments have been discussed earlier. A list of eligible treatments can be assembled easily given what is known about their efficacy in the particular agroecological situations, their synergism and their cost. It remains then to match up the solutions with the problems. This has several dimensions. In so far as the private arable land is concerned, the farmer will undoubtedly choose from the list Che treatments that are income enhancing in the short term. He may or may not require support to apply them but little coercion would be needed. Other treatments such as stabilizing or revegetating non-arable areas or treating drainage lines will require the population's consent and cooperation but would mostly be implemented by some other agency or organization. VI. INCENTIVE FOR PARTICIPATION IN WATERSHED DEVELOPMENT 1.75 There are three well recognized areas which influence farmers' willingness to participate in watershed development programs, and more specifically to implement soil conservation treatments on their farms. These are: (a) land tenure, (b) profitability of the farming system and the scope to improve profit- ability, and - 27 - Bo 4: PROBLEM DEFINITION IN A MICRO WATERSHED Interactive Planning A number of approaches have been developed around the world for carrying out interactive, village planning in microwatersheds. These planning approaches have as their objective to put planners, agency staff, and villagers on a common ground for identifying key problems, analyzing their causes, and devising realistic, action plans that reflect local needs and the availability of government and local resources. Successful approaches are those which include techniques for collecting and discussing information in an open-ended way, which draw strongly upon indigenous, technical knowledge as well as professional expertise, and which are conducted in stages to allow villagers to participate in devising action plans, rather than simply reacting to plans drawnup by government extension agents or officials. The RRA Technique Rapid rural appraisal (RRA) is a technique which is often employed in interactive planning. RRA is not a methodology, but a set of investigative togls adapted to short-term analysis of particular sets of problems of natural resource management. It is often used to gain an initial understanding of problems on the basis of the analysis of secondary data combined with a struc- tured field investigation. In combination with other formal surveys, RRA can be used in monitoring program performance and evaluating program efficiency or program impact. Unlike traditional research, RRA teams include planners as well as researchers; and their investigative tools are designed to encourage as much interaction with villagers as possible. RRA tools include: (a) group and individual interviewing; (b) cross-checking information (triangulation); (c) direct observation; (d) use of sketch maps, diagrams, village transects; (e) sampling tailored to a shortened-time framed; and (f) redesign of plan as hypotheses change and new options emerge. A Sample Application to Watershed DevelouttuiL A watershed development program which includes soil and moisture conservation, forestry, on-farm tree planting, and pasture improvement, is being implemented by several government agencies in a subwatershed. A team of one or two RRA-trained persons and government extension agents would visit villages to analyze people's needs and individual inLerviews with different types of households, the team reviews environmental/economic problems with the villagers, adding to villager statements with their own direct observations, and plotting the information with villagers on sketch maps showing village areas of influence. Conflicts over use of the resources within the village or between villages and over government regulations or uses are important topics, so are the Institutional mechanisms for resolving these problems. Villagers discuss options which they feel will help to resolve their problems and with the team draw up an action plan, based on their own time and resources and the available government inputs, programs, and resources of the extension departments represented. - 28 - (c) farmers economic status which has to do with the proportion of farm income derived from the farm, and whether the farming system is a cash or subsistence one. 1.76 Land Inure. Tenurial arrangements on arable land are less than simple. Land tenure in so far as it affects implementation of soil conservation measures goes beyond the issue of guaranteed long term access to land, to include access to credit, ability to make decisions on land development, the proportion of returns that accrue to the user and the ability to transfer rights. In relation to these attributes there are some seven to eight tenure categories ranging from very secure for privately owned land with title to very insecure for some forms of sharecropping and for private cultivators on public lands. The attributes, the categories and the consequences for participation in development programs on arable land are discussed in Chapter VI. An important conclusion of that review is that tenurial arrangements do have an important influence on the land user's decision to participate or not to participate, but also that on private land low adoption rates have frequently been attributed to tenurial constraints where poor technology options are the real constraint. Two significant changes in recent Bank assisted projects are: (a) the move towards greater use of vegetative, cultural and farming systems related treatments which are, overall, more effective fcr soil and moisture conservation but are more amenable to a wider range of tenure categories, and (b) the presentation of a menu of technical options from which the farmer may choose according to his personal situation rather than a single package which he may find intimidating and reject. 1.77 Profit. Sustainability and Risk. Incentives to participate lie in offering an on-farm production system that is more lucrative than the current with the minimum increase in risk. Adequate evidence exists that in all but Le ubL uacglal situations (shallow stoney soils, steep slopes) total production can be improved by addressing soil moisture management and soil fertility restoration and that techniques for doing this generally mitigate run-off and soil erosion. Yield increases, for example, from contour cultiva- tion alone which involves very little cost can be 50% or more: the resulting improvement in soil moisture management then allows modest levels of applied to be effective. Improved plant density due to seeding rate and row spacing adjustment, improved fertilizer strategy or inter-cropping can improved yield and soil protection. 1.78 Suport for Participants. However attractive the propostl.s might be, most small-holders need support to implement new initiatives. The support is in several areas for example: (a) support to layout, establish and maintain hedges on the contour (for example with vetiver grass); (b) support to reshape inappropriately designed terraces, plant surface risers with suitable fodder plants; - 29 - (c) support with good quality planting materials of most suitable cultures; (d) with demonstrations of improved farming systems and cultural treat- ments; and (e) with credit and extension. VII. RECOMMENDED APPROACHES TO WATRSHED DEVELORMENT. 1.79 The data reviewed in this analysis makes clear that there is no single watershed management problem in the Asia region. Rather there is a complex of issues related to increasing soil loss, land degradation, irregular streamflows and poverty that can best be understood in the framework of watersheds as physical planning units. This analysis suggests that the donor community continue, and accumulate, their efforts on the development of environmentally sound and higher productivity, upland farming systems. Greater attention needs to be given to the diversity of upland agriculture and the need to develop a local capacity for diagnosis of constraints to productivity growth and design of site specific solutions. Forest land and livestock grazing systems similarly need to be addressed because on these lands few soil and water conservation measures are in place. Projects promoting single solution, structural approaches to soil conservation problems should be reduced and greater effort gives to use of vegetative techniques. Technology development projects, with provision for careful experimental design and rigorous testing of proposed techniques may be required before large scale projects can be expected to be viable. 1.80 Flooding and Sediment. The potential for reduction of flood and sediment damage far downstream from land use changes, rehabilitation, and reforestation in the upper watersheds of the Asian region appears limited. Catastrophic floods, such as that which devastated Bangladesh in the fall of 1988, seem to be the result of heavy rains largely falling on already saturated soils and non-absorbing surfaces in lower reaches. The nature of deltic regions, which in fact have been formed by the flood borne deposition of eroded material, is that they are subject to flooding. There seems to be no statistical evidence indicating a secular increase in the occurrence of flood events. Increases in flood damage are largely explained by increased floodplain occupancy and higher value land use. 1.81 Similarly, data on sedimentation supports the argument that in large river basins whose headwaters lie within the geologically young, unstable mountain ranges of Asia (for example, the Himalayas) a very large portion of sediment loads are the result of natural processes and therefore the damages caused by sedimentation are on the same order ascribable to such natural geological processes. Human induced sedimentation, while having severe impacts within smaller watersheds, will have much diminished impacts within the context of larger river basins. To speak of reducing geologic erosion is not practical and reducing human induced erosion in those areas where widespread disturbance has already taken place by alteration in landuse or, for example, reforestation, will not necessarily have the impact required. This is particularly true in light of the residence time of sediment within a catchment whereby sediments currently stored within the channels and floodplains will continue to move through the system for extended periods. Nonetheless, at some point watershed rehabilitation must be undertaken to - 30 - begin the process of reducing sediment transport and deposition, though for large basins the time lag for downstream reduction may be decades or centuries. While the problems caused by sedimentation are real and significant, they should be dealt with through other means such as operating practices, dredging and appropriate design. In particular, decisionmakers should give greater attention to assessing the validity of assumptions about sedimentation rates and predictions of investment life. In some cases it may be necessary to either accept or reject water resource developments depending on whether the expected life time to sedimentation is sufficiently long. However, the likelihood of eventual sedimentation must be recognized. 1.82 Similarly, recognition of the probability of flooding should be factored into policy making with respect to downstream areas. Decisions to site developments in floodplains, to promote settlement, and the actual design of infrastructure and buildings should recognize the certainty of an eventual flood. 1.83 Cost Sharing and Cost Recovery. A corollary of the limited impact of land use changes on downstream damages is that there is limited justification for schemes to compensate upland farmers and communities for adopting conser- vation practices. There is considerable scope for identifying techniques that are profitable from the perspective of upland farmers and which will reduce or at least not increase erosion and runoff. Various subsidies and compensation schemes may be required to bridge the gap between adoption of a conservation measure and the realization of a sustainable net return. However, such compensation should be seen as transitional and not as part of a policy of ongoing subsidy. 1.84 Experience with subsidy schemes for adopting conservation measures has not been encouraging. Unless carefully designed, subsidies can lead to an over emphasis on construction of structural measures, neglect of maintenance requirements and serve as a disincentive to use of less expensive measures that would otherwise be adopted by farmers on their own. An example of a promising strategy has been used in the Central Visayas Regional Project in the Philippines. To promote adoption of contour hedgerows the project lends breeding cattle conditional on establishment and maintenance of a hedgerow sufficient to support stall feeding of the offspring. This provides a power- ful incentive for both establishment and maintenance of the hedgerow and has been extremely effective. 1.85 Rural Infrastructure. Road and trail construntion in upland areas can contribute to either environmental improvement or deterioration and provides one of the few ways of significantly affecting downstream sedimenta- tion. The extension of road networks can lead to deterioration of watersheds by facilitating access to fragile remote areas. On the other hand, improved access to markets can improve incomes which can generally be expected to lead to adoption of more conservation oriented farming. Road and trail design and construction methods need to incorporate adequate safeguards to minimize erosion and sedimentation. As attention is shifted from use of structural conservation measures to agronomic approaches, the engineering expertise of soil conservation agencies can be reallocated to road construction and rehabilitation. 1.86 Analytic Methods. There are no special characteristics of watershed development projects, vis-a-vis other development projects that require a - 31 - fundamentally different approach for their economic analysis. Standard approaches to the analysis of agricultural projects, based on a with-without comparison will produce a satisfactory estimate of project worth. Special attention may be required to understanding incentives as perceived by farmers and communities given the importance of common property resources and the often precarious nature of land tenure systems. 1.87 While this review concludes that the downstream impacts of land use changes within large river basins will generally be of marginal importance over anytime period of economic interest, there are no conceptual constraints on taking any such benefits into account. Meaningful judgements on the physical impacts of specific intervention can be developed through approaches such as sediment budgeting as well as on the basis of long-term measurement and modelling studies. Well established approaches can then be used to translate physical impacts into economic terms. 1.88 More relevant to most watershed development projects is the need to integrate technical judgements on the impact of erosion and conversely conser- vation on crop yields with economic and financial analysis of cropping systems. Although data are seldom available for a particular project site, there is usually evidence from other places that can be used to guide economic analysis. Input from experienced agricultural specialists is needed to ensure the validity of the assumptions made in these calculations. Input from social scientists are also important in order to assess constraints to adoption especially on common property and public land. 1.89 Guidelines. There appears to be no need to prepare technical guidelines for watershed development projects. Annex 1 lists guidelines for various aspects of watershed development projects that have been issued by a variety of governments and international agencies. 1.90 Funding Procedures. New approaches to disbursing project funds are required given that: (a) overall project costs can be estimated by extrapolating from detailed plans developed at the microwatershed level, (b) a number of different government and semi-government agencies and non-government organizations can legitimately be involved in implementing watershed actions; and (c) not all activities need to be synchronized. For example, it is reasonable to envisage a watershed development fund from which approved agencies may be reimbursed for completing treatments eligible for reimbursement. This requires definition of the eligible treatments, definition of the organizations who may implement them and verification procedures relating to implementation. The Bank and other development agencies would be advised to explore more effective funding procedures. 1.91 The need for commitment. Watershed management projects are complex interventions that require effective multidisciplinary collaboration, commitment by governments and local communities, and sustained efforts. For development agencies to be effective partners in this area will require a recognition that watershed projects, while not necessarily large or expensive, will initially require heavy inputs of staff throughout the project cycle and particularly in preparation and supervision. Agencies also needs to recognize and act on the need for government commitment to resolving watershed problems. Without serious commitment by governments and their field staff investments are unlikely to succeed. - 32 - Barbier, Edward (1988), "The Economics of Farm-Level Adoption of Soil Conservation Measures in the Uplands of Java", World Bank Environment Department of Working Paper No. 11. Barrett, Scott (1988), "OphMal Soil Conservation and The Inadequacy of Atricultural Pricing Reforms", Unpublished draft, University of Cambridge. Bhide, Shashanka, C. Arden Pope and Earl 0. Heady (1982), "A Dynamic Analysis of Economics of Soil Conservation: An Anglication of Ophmal Control Mheoal", CARD Report 110, SWCP Series III, The Center for Agriculture and Rural Development, Iowa State University, Ames, Iowa. Brown, Lester and Eric Wolfe (1984), "Soil Erosion - Ouiet Crisis in-the World Economr", Woilfwatch Institute Paper 60, (Washington, D.C.), September. Carson, Brian (1989), "Soil Conservation Strategies for Unland Areas of Indonesia", Occasional Paper No. 9. Environment and Policy Institutes, East West Center, Honolulu. East - West Center, Honolulu FAO/UNBSCO (1982), Soil Map of the World, (Rome/Paris) Gilmour. Gregory, G. Robinson (1955), "An Economic Aproval to Multiple Use" Forest Science, Vol. 1, No. 1, March 1955. Gregory, G. Robinson (1957), "Economic Research Needs in Watershed Management", Proceedings, Society of American Forestus, Syracuse, New York. Hamilton, L.S. and P. King (1983), "Tropical Forested Watersheds; Hydrologic ana soils Response to HajoL Uueu ut Cuvsbions" (Westview: Boulder). Hamilton, Lawrence (1988), "The Recent Bangladsh Flood Disaster was not Caused by Deforestation Alone", Environmental Conservation, pp. 369-370. Hyde, William (1988), "General Public Policy Impacts on Upland Resources and the Environment in The Philippines". Draft rzport pre--red for the World Bank. Ives, J. D and B. Messerli: (1989), The HimalaYan Dilemma (Rutledge: New York). Joshi, R.M. and Khatiwada, M.K. (1986), "Agricultural Handbook Nepal", (Agri. Publication Series: Kathmandu). McConnell, Kenneth (1983), "American Journal of Agricultural Economics", February pp. 83-89. Pearce, Andrew J. (1986), "Erosion and Sedimentation" Paper Prepared for the Workshop on Ecological Principles for Watershed Management April 9-11, East-West Center Honolulu. . 33 - Roche, Frederick (1987), "Sustainable Farm Develoment in Java's Critical Lands: Is a Green Revolution Really Necessary? "(Unpublished Manuscript), Cornell University. Rogers, Peter, Peter Lydon and David Seckler (1989), Eastern Waters Study: Strategies to Manage Flood and Drougbt in-the Gorges - Brahmautren Basin, Report prepared for USAID April. Stocking, M.A. (1988), "Assessing Vegetative Cover and Management Effects" in R. Lal, ed., Soil Erosion Research Methods, Soil and Water Conversion Society, Ankery, Iowa. UNEP (1986), Environmental Guidelines for Rural oads, UNEP Environmental Management Guidelines No. 13, Nariol: World Bank (1989a), Indonesia: Forest, Land and Water: Issues in Sustainable De Ment, Report No. 7822-IND. World Bank (1989b), Philippines: Forestry. Fisheries and Agricultural Resource Manaemnt, (Farm) Study Report No. 7388-PH. Yadav, Ram P. (1987), OAgricultural Research in Nepal: Resources Allocation, Structure and Incentives", Research Report 62 International Food Policy Research Institute, Washington, D.C. ANNEX 1 - 34 - Guidelines for Watershed Management Dani, Anis, and J. Gabriel Campbell, Sustaining Upland Resources Peoples Participation in Watershed Management, ICIMOD Occasional Paper No. 3, International Centre for Integrated Mountain Development and FAQt Kathmandu July 1986. FAO, Strategies, Approaches and Systems in Integrated Watershed Management, FA0 Conservation Guide 14 FA0s Rome 1986. Gregersen, H.M., K.N. Brooks, J.A. Dixon, L.S. Hamilton, Guidelines for Economic Appraisal of Watershed Management Projects, FAO Conservation Guide 16 (East-West Center, SIDA, FAO) Rome 1987. Hamilton, L.S. and P.N. King, Tropical Forested Watershedes Hydrologic and Soils Response to Major Uses or Conversions, Westview: Boulder 1983. Hufachmidt, M.M., James, D.E., Meister, A.D., Bower, B.T. and Dixon, J.A., Environment, Natural Systems and Development - An Economic Valuation Guide, The Johns Hopkins University Press: Baltimore 1983. Runkle, S., et. al Monitoring Stream Winter for Land Use imputs: A Training Marval for Natural Resource Manaiement Specialists, Water Resources Division, U.S. National Park Service (Fort Collins, Colorado) 1987. Pearce. Andrew J., and Lawrence S. Hamilton, Water and Soil Conservation Guidelines for Land Use Planning Report of a Sex-nar Workshop, East-West Center: Honolulu May 1986. Spears, J.S. and R.D.H. Rove, wPreliminary Guidelines for Designing Watershed Rehabilitation Projects for Bank Financing' in World Bank, Proceedings of the Second Agricultural Sector Symposium, World Bank: Washington 1981, pp 408-423. United Nations Environment Program, Environmental Guidelines for Watershed Development, UNEP Environmental Management Guidelines No. 3, UNEPt Nairobi