Scholarly article on topic 'Boundary work: Knowledge co-production for negotiating payment for watershed services in Indonesia'

Boundary work: Knowledge co-production for negotiating payment for watershed services in Indonesia Academic research paper on "Earth and related environmental sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Ecosystem Services
Keywords
{"Watershed management" / "Payment for ecosystem services (PES)" / "Knowledge diversity" / "Knowledge co-production" / "Boundary work"}

Abstract of research paper on Earth and related environmental sciences, author of scientific article — Beria Leimona, Betha Lusiana, Meine van Noordwijk, Elok Mulyoutami, Andree Ekadinata, et al.

Abstract Boundary work has been proven effective in bridging research communities and the gap between action and policy-making in sustainable development. Applying this boundary-work framework, the manuscript examines the process of knowledge co-production and evaluates its effectiveness in supporting the negotiation process of four cases of payment for watershed services (PWS) in Indonesia. Our case studies reveal that local communities and policy-makers have a diverse range of knowledge regarding watershed functions and services. Recognizing this knowledge diversity, and combining it with scientific information, leads to (i) enlightenment, by engaging local stakeholders in more active roles for knowledge co-production thus setting realistic targets for ecosystem services’ interventions in the design of PWS schemes; (ii) decision-making support for stakeholders, by providing opportunities for collaborative learning; and (iii) effective negotiations, by providing salient and credible information. We recognize 10 different prototypes that lead to a better understanding of how payments can be channeled to enhance, or at least maintain, underlying hydrological functions. The case studies, in different landscape configurations and associated PWS prototype settings, show that knowledge interfacing and sharing towards co-producing collaborative products helps to clarify the performance-based indicators for effective PWS negotiation between potential sellers and buyers of ecosystem services.

Academic research paper on topic "Boundary work: Knowledge co-production for negotiating payment for watershed services in Indonesia"

Contents lists available at ScienceDirect

Ecosystem Services

journal homepage: www.elsevier.com/locate/ecoser

Boundary work: Knowledge co-production for negotiating payment for watershed services in Indonesia

Beria Leimona *, Betha Lusiana, Meine van Noordwijk, Elok Mulyoutami, Andree Ekadinata, Sacha Amaruzaman

World Agroforestry Centre (ICRAF), Southeast Asia Regional Office, Indonesia

ARTICLE INFO ABSTRACT

Boundary work has been proven effective in bridging research communities and the gap between action and policy-making in sustainable development. Applying this boundary-work framework, the manuscript examines the process of knowledge co-production and evaluates its effectiveness in supporting the negotiation process of four cases of payment for watershed services (PWS) in Indonesia. Our case studies reveal that local communities and policy-makers have a diverse range of knowledge regarding watershed functions and services. Recognizing this knowledge diversity, and combining it with scientific information, leads to (i) enlightenment, by engaging local stakeholders in more active roles for knowledge coproduction thus setting realistic targets for ecosystem services' interventions in the design of PWS schemes; (ii) decision-making support for stakeholders, by providing opportunities for collaborative learning; and (iii) effective negotiations, by providing salient and credible information. We recognize 10 different prototypes that lead to a better understanding of how payments can be channeled to enhance, or at least maintain, underlying hydrological functions. The case studies, in different landscape configurations and associated PWS prototype settings, show that knowledge interfacing and sharing towards co-producing collaborative products helps to clarify the performance-based indicators for effective PWS negotiation between potential sellers and buyers of ecosystem services. © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

CrossMark

Article history: Received 27 March 2014 Received in revised form 1 July 2015

Accepted 13 July 2015

Keywords:

Watershed management Payment for ecosystem services (PES) Knowledge diversity Knowledge co-production Boundary work

1. Introduction

Watershed degradation affects fresh water supply and quality, and increases the frequency of water-related disasters. It thus has a negative impact on human wellbeing. However, increased land-use intensity in upland areas also provides livelihood options for a growing population. Balancing the trade-off between the economic gains of more intensive land use and the insurance investment in watershed conservation working towards healthy watersheds is a decision-making challenge (Barbier and Burgess, 1997; MA, 2005). The short-term benefits of intensification commonly lead to increased exposure to climate risk and a possible downward spiral into land degradation. To achieve both livelihood and conservation goals, policy instruments, such as public investment and market-based instruments, can build enabling environments to manage this trade-off and shift land-management decisions (Braat and de Groot, 2012; Tomich et al., 2004).

Inspired by the way Costa Rica reformed its existing forest subsidy scheme into a Payments for Ecosystem Services system in the 1990s (Chomitz et al., 1999), the last decade has seen broader

* Corresponding author. Fax: +62 251 8625416. E-mail address: l.beria@cgiar.org (B. Leimona).

experimentation with markets and payments for watershed services and with policy and institutional options for watershed management elsewhere (e.g. in Asia and Africa) (Adhikari and Boag, 2013; Leimona et al., 2015; Namirembe et al., 2014). The process of design and negotiation required to establish a sustainable Payment for Watershed Services (PWS) scheme is knowledge-intensive, involving multiple actors and potentially conflicting objectives with diverse and dynamic multi-faceted knowledge systems.

One major challenge of the negotiation process is that key actors often propose and develop plans for watershed policy based on perceptions rather than scientific realities, local ecological knowledge acquired by direct contact with the environment (Chapman, 2002; Schalenbourg, 2004) and locally-evolved ecosystem management practices (Berkes, 1999; Berkes et al., 2000)1. According to Maiello et al. (2013), public services managers often rely purely on expert and

1 In this paper, the level of analysis of 'local ecological knowledge' is part of the broader concept of 'Traditional Ecological Knowledge' or TEK (Berkes, 1999, 2000). The TEK encompassed management practices based on ecological knowledge, and social, historical, cultural and institutional mechanisms behind management practices. 'Local Ecological Knowledge' focuses on management practices based on local communities' ecological knowledge. These practices could integrate both conventional resource management, and local and traditional society's ones with various degrees of combination.

http://dx.doi.org/10.1016/j.ecoser.2015.07.002

2212-0416/© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

administrative knowledge and do not integrate political, scientific and community perspectives. From the perspective of scientific knowledge, the predictive skill of hydrological models that are not calibrated using local data is still disappointingly low, despite major international efforts to remedy this shortcoming (Hrachowitz et al., 2013). On the other hand, local data may not be readily available or, when they are available, of uncertain reliability. Nonetheless, watershed rehabilitation efforts, including those involving a PWS scheme, mostly neglect local farming practices and wisdom (Joshi et al., 2004b). Many concentrate on large-scale tree planting as a 'one-size-fits-all' solution, tackling environmental issues as if they were mere technicalities (Maiello et al., 2013).

Given the complexities involved, an operational model that enables knowledge interfacing and sharing, to facilitate and support the complex negotiations, is essential for developing a sustainable flow of incentives for watershed provision. The principles of knowledge interfacing and sharing assume that knowledge is produced jointly (co-production of knowledge) through collaborative learning between 'experts' (i.e. scientists) and 'users' (i.e. managers and decision-makers) (Roux et al., 2006). The central challenges for knowledge co-production are to respect the complementarity of knowledge systems, to integrate multidisciplinary collaboration within science, and to enhance the growth of relevant, legitimate and credible evidence-based input from any of the contributing knowledge systems. Indeed, scientific inquiry cannot thrive without a safe space, protected from political correctness, stakeholder vested interests and existing policy frameworks. Thus, the (knowledge) boundary between science and action needs to be semipermeable (Van Noordwijk et al., 2009a).

The conceptual framework for analysis of ecosystem services that was successful in science-policy arenas (Reid et al., 2006) does not necessarily match local knowledge systems and conceptualizations (Tomich et al., 2004). Knowledge in this context can be defined (Joshi et al., 2004a) as a logical interpretation of qualitative or quantitative observations ('data'), acquired directly or indirectly from other sources, used to convey understanding that can be articulated and recorded independently of the interpreter and used for predictions and decisions. Knowledge systems include the way knowledge changes by various modes of learning. An effective method for representing the apparent logic of local or public/policy knowledge consists of dissecting statements into their unitary elements that describe relations (of a correlative or causal nature) of many contextual entities (Dixon et al., 2001). It builds up a local vocabulary that may or may not have equivalents in other languages, and is careful in retaining context to statements made. A description of a knowledge system then includes a dictionary of terms (often in a generic-specific hierarchical relationship) and a set of relationships between these terms. The degree to which correlative and causal relations are differentiated, as well as the number and types of 'causes' that is invoked can vary between knowledge systems. Knowledge systems of a similar 'domain' can differ in the entities (vocabulary), types of relationships, context-specificity of relationships and the types of new data (observations) that is needed to modify established interpretation. Scientific knowledge has generally started off as a subset of public knowledge, but it is stricter on the types of evidence it allows, keen on reducing context-specificity of explanations (seeking generalizations that are robust) and restrictive on the types of forces it invokes as explanatory factor.

Where knowledge systems differ, firstly, the practical implications for what to do or not to do may still match. In a well-studied example, (Lansing, 2012) documented how traditional subak institutions for management of irrigated rice-fields on Bali were in fact ecologically superior to 'modern' technically derived irrigation systems. The local knowledge systems, expressed in procedures that lead to synchronicity in the start of a new growing season, used a

different rationale than ecological analysis of pest pressure and water availability, but the resulting practice was aligned. Secondly, it is a judgement call whether or not any difference in rationales matters and is an obstacle to communication and negotiation. As long as they support similar decisions and indicative value systems, differences can be accepted as a mutually enriching diversity, but where they lead to contradictory outcomes (e.g. 'trees increase water flows' versus 'trees decrease water flows'), exploration of context and observational roots of the statements may be needed before progress can be made (Van Noordwijk et al., 2009b).

Boundary work, the analysis of boundaries in a knowledge-action system, is defined as 'the process through which the research community organises its relations with the worlds of action and policy making' (Cash et al., 2002; Clark et al., 2010). Boundary-work studies undertaken in the context of developing economies have explored how knowledge generated by (a) a single discipline; or (b) multiple disciplines and knowledge systems, can be used for (i) general enlightenment (contextual clarification); (ii) decision-making support for stakeholders; and (iii) negotiations among multiple stakeholders who have and selectively use multiple knowledge claims (Clark et al., 2011). The use of knowledge for negotiation support is the most complex form of boundary work and, as emphasized by Clark et al. (2011), the information used during this process needs to be salient (i.e. information about providing ecosystem services and joining a PWS programme, in terms relevant to the local watershed context), credible (i.e. technically adequate according to ecosystem services' measurements), and legitimate (i.e. 'fair, unbiased and respectful of all stakeholders'). There are manifold challenges to the effective integration of these types of knowledge into negotiations and decision-making: salience requires that the knowledge shared is contextualized and participative; the credibility of multiple knowledge systems must be demonstrable; and it is crucial that knowledge claims can be proven as legitimate, as conditions of such negotiations are often politicized and contested.

This paper investigates the lessons learned about boundary work for payment for watershed functions in Indonesia drawn from the Rewarding Upland Poor for Environmental Services (RUPES) 2 action-research network, coordinated by the World Agroforestry Centre. Previous research on boundary work in integrated natural resource management carried out in Indonesia for the Alternative to Slash and Burn (ASB) programme of the CGIAR, a global research partnership for a food-secure future, concluded that the ASB-RUPES project had succeeded in creating multiple forms of boundary work as the basis for conflict negotiation and the key 'win/win' option (Clark et al., 2011). Our current research sought to extend this work by further exploring, comparing and evaluating the effectiveness of boundary work in four PWS cases in Indonesia. Using Rapid Hydrological Appraisal (RHA), a tool for scoping PWS schemes (Jeanes et al., 2006), we firstly systematized the captured diversity of knowledge regarding landscape characteristics, problems, and related land-management issues among local watershed managers. These managers are usually farmers making decisions about their land practices and government officers acting as policy providers. Secondly, we highlighted concordances between these perceptions and knowledge claims regarding the cause and effects of watershed problems by checking hydrological modeling produced for each RHA report. We inquired, in the context of boundary work, whether this process could potentially expose logical contradictions and enhance the credibility

2 RUPES is an action-research network on payment for ecosystem services in Asia that was initially funded by the International Fund for Agricultural Development , 2002-2006 (Phase I) and 2007-2011 (Phase II). Research continues by the World Agroforestry Centre as part of the CGIAR Research Program on Forests, Trees and Agroforestry.

of the existing information and knowledge on watershed function and management (enlightenment), support more salient decision-making by stakeholders in watershed management and enhance more legitimate negotiation for any proposed PWS schemes. In the discussion, we reviewed factors that could support or obstruct the effectiveness of boundary work for negotiating a PWS scheme. Four PWS case study sites in Indonesia were selected from a wider set of RUPES sites to represent landscape configurations in different climatic zones. Finally, we recommended 10 prototypes for PWS and estimated the effectiveness of each configuration for solving underlying watershed problems (i.e. increased water yield, better water quality, and others) and negotiation under PWS schemes. Methods applied in the RHA tool for identifying knowledge diversity among local communities and policy-makers, and for developing hydrological modeling, are presented in Appendix 1, and published elsewhere (Jeanes et al., 2006; Van Noordwijk et al., 2013).

2. Case studies from watersheds of various landscape configurations in Indonesia

The case studies examined in this paper are Kapuas Hulu in West Kalimantan, Sumberjaya in Lampung, Singkarak in West Sumatra, and Talau in East Nusa Tenggara (Fig. 1). The sites represent substantially different human and landscape characteristics found across Indonesia (Table 1 ). Kapuas Hulu is dominated by a tropical forest landscape with very low human population density; the remaining sites have medium-to-high population densities and are dominated by agricultural landscapes that range from complex tree crops and horticulture to paddy fields. Recently, four configurations of the forest-agriculture interface were distinguished in relation to food security (Van Noordwijk et al., 2014a) and ecosystem services (Leimona et al., 2015): (I) predominantly forested landscapes that involve swiddening as a form of agriculture, well integrated with forest regeneration; (II) landscapes with forest segregated from, and in a zero-sum game competing with, an intensifying form of agriculture; (III) landscapes with a forest-agroforestry-agriculture continuum with intermediate-intensity land uses at the interface; and (IV) landscapes, often with high population density and use intensity, with a segregated, protected forest component focussed on non-provisioning services, as well as a productive agroforestry-agriculture gradient. Configurations II and III can both evolve from I and into IV. In applying this scheme to our sites, a fifth configuration

emerged, which can be seen as a low-forest version of IV. This Configuration V landscape is represented here by Talau in East Nusa Tenggara, where grassland for foraging is the most common land cover, because it is located in the driest part of Indonesia. The main site characteristics and their configurations are summarized in Table 2 and Fig. 2.

Complete RHA studies at each site were coordinated by the authors of this paper and supported by scientists of the World Agroforestry Centre (ICRAF) in Kapuas Hulu (Lusiana et al., 2008a), Singkarak (Farida et al., 2005), and Talau (Lusiana et al., 2008b). For the Sumberjaya case, the knowledge systems on watershed management were authored separately, i.e., hydrological modeling (Verbist, 2008; Verbist et al., 2010) and local ecological knowledge (Agus et al., 2002; Chapman, 2002; Schalenbourg, 2004). PWS efforts at two sites (Singkarak in West Sumatra and Sumberjaya in Lampung) were action-research sites from the RUPES (Phase 2) project coordinated by ICRAF in collaboration with local government and NGOs during 2002-2011. Research at the other two sites (Kapuas Hulu in West Kalimantan and Talau in East Nusa Teng-gara) was coordinated by the World Wildlife Fund (WWF) Indonesia in collaboration with the Equitable Payment for Watershed Services (EPWS) consortium of CARE International and the International Institute for Environment and Development (IIED) during 2008-2011.

2.1. Kapuas Hulu in West Kalimantan

As part of WWF in Indonesia's EPWS project, the interventions in Kapuas Hulu aimed towards preparing and establishing a solid, verifiable business model for equitable PWS. The next step involved building up the business models and establishing PWS mechanisms at the selected sites. This project was based on the assumption that the high rate of deforestation was the primary cause of the watershed problems in Kapuas Hulu, particularly affecting a downstream regional drinking water company (known in Indonesian as a PDAM or Perusahaan Daerah Air Minum).

Forest is the dominant land cover over 90% of the total watershed (Lusiana et al., 2008a). A national park marks a hot-spot biodiversity area. The Kapuas Hulu Basin is home to several indigenous Dayak tribes: the Iban, Kantu', Tamanbaloh, Kayan, Bukat, and Punan. Farmers cultivate their lands intensively in the Sibau and Mendalam sub-catchment. However, the main livelihoods of the local stakeholders are based on gathering forest products and extensive local agroforestry practices (tembawang or mixed planting).

Fig. 1. Location of the case study.

Table 1

Characteristics of landscape configuration

Configuration

Ecosystem condition

Typical inhabitant

Environmental problem

Example of case study

Natural forest dominates, swidden-fallow with intact functions

Short-fallow rotations and permanent agriculture spatially segregated from forests

Agroforests and changes in integrated agriculture-agro-forestry-forest transition

Mosaic of remnant forests, permanent (intensive) agriculture and agroforests

Native sub-humid ecosystem of savanna and patches of agriculture

'Forest' people, low population density

Localized and temporary effects of shifting cultivation swiddens

Local people and/or migrants starting market integration and commercialized agriculture; forest management institutions

Local and migrant people with tree-based system as primary source of income with high potential commercialization

Local and migrant people with intensive and commercialized commodities; forest institutions

Local people, subsistence farming system

Classic agriculture problems of over intensified slash and burn, soil compaction and erosion, potential deforestation

Pressures to alter to monoculture system and forest conversion in nearby protected area

Continued degradation and further deforestation, high potential of environmental disasters

Degraded native ecosystem due to intensive farming

Kapuas Hulu, Kalimantan (Lusiana et al., 2008c)

Mae Chaem, Northern Thailand* (Van Noordwijk et al., 2014a)

Sumberjaya, Lampung (Verbist, 2008)

Singkarak, West Sumatra (Farida et al., 2005)

Talau, East Nusa Tenggara (Lusiana et al., 2008b)

Note: * not available as a case in this study but Configuration II is discussed in Section 4 to provide a more comprehensive perspective on PWS prototypes. F = forest; af = agroforest; A = agriculture; G= grassland

Table 2

Main characteristics of study sites

Configurationa Kapuas Hulu Sumberjaya Singkarak Talau

I III IV V

Province West Kalimantan Lampung West Sumatra East Nusa T

Regency Kapuas Hulu West Lampung Upper: Solok Lower: Tanah Datar Belu

Population densityb (person.km-2) 8 86 95 (upper) 151

257 (lower)

Catchment area (hectares) 9 80,000 54,190 10,780 72,000

Climatic conditions Humid tropics Humid tropics Humid tropics Sub-humid

No. of wet months 10-12 7 5 4

Total annual rainfall mm year-1 4100 2500 2760 1605

Human population density, km2 4 130 90 140c

Human Development Index (HDI) 0.71 0.70 0.75 0.64

Forest fractiond 0.91 0.25 0.17 0.01

(ha per capita) (19.9) (0.19) (0.22) (0.00)

Agroforestry and tree crop area fraction 0.03 0.46 0.17 0.07

(ha per capita) (0.63) (0.35) (0.22) (0.04)

Rice field fraction 0.00 0.13 0.17 0.16

(ha per capita) (0.00) (0.10) (0.23) (0.10)

a The typology of configuration is discussed in the text.

b Based on population density of regency in 2012 (BPS, 2013). The average population density for Indonesia in 2013 was 130.

c There was a significant increased of population in Belu due to migration of Indonesian refugees from Timor Leste when the country had its independency. d Forest cover was based on the analysis of Landsat-TM acquisition in 2005-2008, in a similar timescale to the RHA studies.

2.1.1. Local and policy-maker viewpoints

In Kapuas Hulu, different ethnic communities and livelihood options strongly influence the land-use pattern along the river. Upstream residents tend to have less permanent dwellings, less technology and use subsistence practices; they are mostly hunters and subsistence food-gatherers with high levels of income uncertainty. These people believed that erosion and landslides caused by logging activities in the upstream areas and riparian zones led to heavy economic losses. In Sibau and Mendalam, people blamed the establishment of shortcuts across riverbanks to hasten water transport as a cause of sedimentation. The Mendalam people were also concerned about the recent establishment of a forest concession company in the area, as well as illegal mining

and small-scale logging.

The Dayak people in Kapuas Hulu use their own customary law (adat) for forest management. The adat limits forest provisioning services solely to domestic use for specific activities, such as timber and animal harvesting, with permission granted by the adat leaders. The laws also define protected areas, including forested areas and lakes, and include rules governing fishing practices, such as banning fishing using electric shock and poison. The Melayu Sambus community agreed to avoid the use of pesticides and insecticides when opening new land, and have prevented outsiders from opening and exploiting new land. The Putusibau PDAM in the district capital of Kapuas Hulu indicated that turbidity was problematic and resulted in decreased water quality for domestic use.

Furthermore, gold-mining activities had the potential to increase water pollution due to toxic mercury usage. The local community and policy-makers claimed that the key environmental problems in this area were forest degradation, river siltation, lack of fresh water, and substantial water pollution. Moreover, river siltation leads to river shallowness, which could disturb river traffic; and boats are the main mode of transport in this area.

2.1.2. Hydrological modeling results

An analysis of the forest conversion effects on the water balance in a simple water balance model (Van Noordwijk et al., 2011) revealed that reducing the forest cover in this area would increase surface runoff and reduce soil quick-flow. Therefore, these effects must be anticipated in healthy riparian zones to avoid increased river sedimentation. The landscape water balance analysis also showed that, before 2004, the runoff fraction in the Kapuas Hulu Basin was low, demonstrating the ability of the Kapuas Hulu basin to maintain its watershed function, especially river flow (Fig. 3). However, there were already signs of smaller catchment-scale degradation. The estimated smaller catchment-level landscape water balance in the Mendalam sub-catchment indicated that the runoff fraction was six times the overall basin fraction (Lusiana et al., 2008a).

Between 2001 and 2004, the forest area in the Kapuas Hulu basin decreased by approximately 130 km2 and the total farmer-managed area increased by approximately 42 km2. These changes were insignificant in the context of the total basin area. However, the changes represented a substantial relative increase in agricultural land. Moreover, settlement had more than doubled within this period. These changes mostly occurred in the provincial land area designated as a 'dry agricultural' zone. Most land changes occurred along the river outside the national park, where

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Reservoir/ Water body

Settlement

Other land

Crop (non-paddy) lands Rice field fraction Agroforest&Tree crops Forest

Kapuas Hulu Sumberjaya Singkarak Talau Fig. 2. Land use fraction representing the configuration of the watersheds.

Fig. 3. Water balance of Kapuas Hulu basin for different land-use scenarios.

agriculture was allowed under the law.

2.1.3. Implications for watershed management as collaborative products

Results from the hydrological assessment indicated that in the current situation, the landscape was sufficiently intact to maintain a low sedimentation rate that might potentially harm the operation of the downstream water company. The local perspectives on sedimentation varied, though some of them confirmed the hy-drological modeling results, i.e. that sedimentation was caused by modification of the river bank not by forest conversion. However, the project demanded a solid, verifiable business model for watershed functions as the basis for establishing an operational PWS mechanism on this site. To bridge the gaps between local perspectives and the results of the wider hydrological knowledge assessment, the project focused on rescaling the overall target area. In this case, rather than targeting the whole watershed, the project manager specified five villages within a sub-watershed of Kapuas Hulu, namely the Mendalam sub-watershed, for sedimentation reduction activities and monitoring. Moreover, the facilitators chose to focus on the Mendalam sub-catchment due to the development of intensified systems (i.e. vegetable plots and coffee systems) and the threat of a new forest concession in the area. Therefore, rehabilitation along the riparian zone and establishment of a local agroforestry system, or tembawang, became the favored solutions for this watershed, rather than a focus on forest conservation and restoration. At the district level, the managers were recommended to focus on regulating the negative environmental impacts of gold mining.

In financial terms, the district government had allocated an annual budget of approximately USD 20,000 to each village to be paid every year since 2012. The fund was to be made available for allocations to both individual participants of the PWS programme and to village revenue, with uncertain percentage allocations for each village. Land rehabilitation was organized on private land owned by local stakeholders on the assumption that the district government represented their target buyer, i.e. the local PDAM.

2.2. Sumberjaya in Lampung

The Sumberjaya watershed in Lampung, Sumatra is a benchmark of the ASB Negotiation Support Models for Integrated Natural Resources project, where upstream farmers were initially held largely responsible for uncontrolled deforestation causing heavy sedimentation to downstream beneficiaries (Verbist, 2008; Verbist et al., 2005).

The watershed, the primary contributor to the Way Besai River, is located near Bukit Rigis watershed protection forest. Downstream of the Way Besai River, a hydroelectric power (HEP) company produced approximately 480-2042 MWh of electricity daily, which was distributed to three provinces in Sumatra. Sumberjaya communities are typically multi-ethnic but consist primarily of people from the Semendo ethnic group and migrants from Java (Sundanese and Javanese). The Semendo people mostly practice slash and burn agriculture, while the migrants have permanent coffee-based plantations, both monoculture and multistrata systems, on the hill slopes and paddy fields along riparian strips. The multistrata systems are agroforestry coffee systems in which farmers plant various timber and fruit trees in their coffee gardens; these have been in use since the late 1980s.

2.2.1. Local and policy-maker viewpoints

In Sumberjaya, farmers cultivated coffee on steep erosion-prone land; paddy fields faced flooding problems and riverbank abrasion. Moreover, farmers converted primary and secondary forest to monoculture and multistrata or agroforestry coffee

gardens. Farmers had been aware of, and prepared to invest heavily in, artificial fertilisers to increase productivity and they have applied a range of erosion-restraint measures in their coffee gardens, such as terraces, trenches, ridges, and pits. Certain tree species, such as Gliricidia sepium, were selected and positioned; the plant components were manipulated based on soil-management issues. In the 1990s, the local government and its forestry department suggested that uncontrolled deforestation and coffee conversion on the slopes had led to a substantial increase in erosion and a reduction in Way Besai River discharge. Forest boundary enforcement led to the eviction of thousands of farmers between 1991 and 1996. Evicted farmers were resettled on infertile acid lowland peneplain or converted swamp forest in northeastern Lampung. From 1998, after political changes in Indonesia, farmers requiring a livelihood returned to the area, often with the silent approval of the local government, which in turn needed income and was interested in economic development.

2.2.2. Hydrological modeling results

ICRAF scientists tested the erosion rate under various land-use types (i.e. forest, bare soil, multistrata and monoculture coffee systems) in two plots during the period 2001-2005. The data revealed that soil properties have a greater effect on the erosion rate compared to tree-cover density (Verbist, 2008; Verbist et al., 2010). The erosion rate in the first plot was between 4 tonne/ha/ year for forest and 30 tonne/ha/year for bare soil, decreasing to between 0.1 (forest) and 4 tonne ha_ 1 year(bare soil) for the second plot with the same treatments . The coffee garden erosion rates fell between the bare soil and forest rates. The highest erosion rate occurred in a 3-year-old coffee garden, gradually decreasing as litter layers established soil cover. The Sumberjaya watershed has an old crater landscape with a high diversity of geological substrates. Even under dense forest cover, some pristine headwaters can become very turbid. Furthermore, the research showed that land use has a less important role than geological characteristics in levels of river sedimentation. The overall watershed-level sediment yield was caused by landslides, riverbank collapse, and dirt footpaths. The research also showed that catchments with relatively high forest cover (more than 30% coverage) also had the highest sediment yield (Fig. 4).

Daily rainfall and discharge (water flow) time series show that, although the average rainfall remained constant, the average discharge increased, likely because forest was converted to coffee gardens which reduced evapotranspiration. The low flows in the Way Besai during the dry season further decreased. However, the number of years with a prolonged dry season also decreased. An increase in El Niño years (1976 versus 1991,1994,1997, and 2004) caused the perception that dry season flows were reduced by local land-use changes rather than by global climate change.

2.23. Implications for watershed management as collaborative products

The multi-stakeholder assessment and negotiation provided directions for solving the watershed conflicts and to potentially contribute to ecosystem services' provisions by reducing sedimentation coming from upstream. In 2001, the Ministry of Forestry promulgated decree 31/2001 on community forestry, which provided an enabling policy environment for developing more specific guidelines for contracts at the district level. The scheme required farmers to form organizations and follow management guidelines approved by local forestry officials. The HKm (Hutan Kemasyarakatan, or community forest) permit in a protected forest area can be considered to be a payment for watershed services because farmers were obliged to plant at least 400 trees per hectare as a condition for joining the voluntary programme. Conditional tenure security to utilize forestland has a probationary

Fig. 4. Average of plot-level erosion in Sumberjaya for monoculture coffee and forest in three sub-catchments (Verbist, 2008). Note: Way Ringkih (WR), Way Tebu (WT) and Way Petai (WP); F: Forest; the numbers 1, 2, 3,4, 8,11 refer to the age of the sun coffee gardens.

period of five years and can be extended to 25 years if the HKm group accomplishes all its criteria and indicators.

In an effort to directly connect the landscape to the PWS scheme, ICRAF also coordinated a River Care scheme in 2007. This was a collective action programme organized by communities living along the riparian strip who undertook the responsibility to produce clean water by reducing soil sedimentation for the downstream HEP (Pasha et al., 2012). The payment was either made in cash (USD 2222 in Gunung Sari) or in the form of construction of a micro-hydroelectric power plant with a capacity of 5000 W (in Buluh Kapur) and a monetary value similar to that paid in Gunung Sari if the community could reduce the sediment by at least 30%. The payments would decrease—i.e. USD 833, USD 555, and USD 278—for sediment reductions of 21-29%, 10-20%, and less than 10%, respectively. An external stakeholder, the local Forestry Service, monitored the scheme every three months.

2.3. Singkarak in West Sumatra

The Singkarak Basin is categorised as one of the most critical watersheds in Indonesia (RUPES, 2010). Watershed rehabilitation in early 2000 was mostly organized by the national government: monoculture pine trees were planted in the upstream area. Continuous degradation, despite the mass 'greening' programme, provided valid rationales for other options, such as implementing a PWS scheme in this basin. Geographically, the watershed is a part of Bukit Barisan mountain range. It consists primarily of volcanoes, with Lake Singkarak situated in the middle of the basin. A hydroelectric project located in the downstream section has diverted most of the lake outflow from its natural outlet (the Om-bilin River) into the Anai River, which flows westward into the Indian Ocean. The dominant ethnic group in Singkarak is the Minangkabau, who have a distinctive matrilineal culture that governs and enforces its adat laws and conventions.

2.3.1. Local and public policy-maker viewpoints

The Singkarak communities observed that the overall water availability tended to be good in the Paninggahan area (one of the upstream nagari, a collection of villages under shared jurisdiction), and water only became slightly scarce during the dry season. They also observed that floodwaters entering the paddy fields around the lake had increased since the construction of the HEP dam. People living around the lake also had water-quality problems caused by domestic pollution, which decreased their fish harvest from the lake. They perceived that the HEP company was not able to provide as much electricity as expected because of large

fluctuations in the lake's water level.

The tree type (pine versus broadleaf) was perceived to have an effect on total evapotranspiration from foliage with a subsequent effect on the total soil water availability and levels of water flowing downstream. The local people claimed that soils 'dried up' after pines were planted in previously mixed forested areas. In recent years, pine has been extensively used in reforestation programmes in the area. As a solution to the negative effects of pines on water availability, local stakeholders suggested mahogany and teak as examples of species that required less water.

The villagers also valued their local biodiversity, such as a native fish species (ikan bilih) that is intensively harvested for commercial culinary purposes. The villagers developed a village regulation to protect the fish stock by allowing only fish of a certain size to be harvested. The district government bought these fish and released them into the lake.

Government officials also mentioned season, land coverage, soil type, and tree type as factors affecting water availability. An informal government group also agreed that trees helped to hold water in the ground, reducing run-off and soil erosion. Government officials said that forest clearing to the south of Singkarak Lake was causing most of the flood problems, primarily because there was less forest area that could hold water and reduce flooding. Farmers shared a similar opinion regarding the factors that affected flooding and subsequent effects on rice crops and damage to fields and irrigation channels.

2.3.2. Hydrological modeling results

The water balance model suggested that the overall water shortfall for electricity generation ws a problem of timing or lack of effective storage capacity. The lake storage capacity at the peak of the rainy season was insufficient to retain the water. Therefore, the water wa allowed to overflow into the Ombilin River. The main issue was whether the upstream watershed could retain enough water to provide stable flows during the dry season, which lasts approximately 2-3 months. Fig. 5 shows the modeling results for different land-use scenarios in the watershed: (1) all degraded lands are converted to natural forest; (2) business as usual, i.e. current mixed land use; and (3) all lands are severely degraded. The hydrological model revealed that the presumed positive relationship between reforestation efforts and water availability for HEP did not likely exist. Climatic variation affects the HEP company performance more than land-use changes in the basin.

Furthermore, decreasing water quality would trigger eu-trophication in the lake. Although this condition would not affect the overall lake quality, it would reduce the HEP's efficiency in producing hydroelectricity. Therefore, maintaining the lake's water quality was important for all stakeholders. Priority actions would

have to focus on the rivers and streams that carried the highest sediment, nutrient, and organic pollutant loads. Most notably, this applied to the Sumani River, which drains the largest area of intensive horticulture and passes a medium-sized town. Pollution control at the point-source level must complement efforts based on land cover.

2.3.3. Implications for watershed management as collaborative products

In contrast to a single solution of planting pine trees to rehabilitate the watershed function, the assessment proposed nested watershed management: at the upstream villages; at the villages surrounding the lake; and at the whole watershed level. At the upstream villages, where the streams had good water quality, the communities focused on maintaining the environment intact by conserving their local biodiversity, such as continuing the village regulation on native ikan bilih fish conservation. Another option was to rejuvenate their old coffee-agroforestry system for growing organic coffee rather than changing it to a monoculture system. Starting in 2010, at the nagari level, farmers managed 49 hectares of degraded land in a voluntary carbon market (VCM) scheme with a private company from the Netherlands. This scheme applied participatory tree selection to rehabilitate the degraded land, taking into consideration farmers' knowledge of tree species and market potential.

The management of the lake was initiated by 12 nagari leaders from villages surrounding the lake, who submitted a proposal to manage the lake to the Ministry of Environment. The management activities were specific to environmental problems faced by the nagari, and included constructing rubbish traps, and developing recycling units and water purification systems. To coordinate this effort, a nagari forum was established.

At the watershed level, rehabilitating degraded land was still the focus. Better tree selections were made, adopting the local agroforestry system. A local NGO called Yayasan Danau Singkarak negotiated the redistribution of the water tax and hydropower royalties to local communities through the nagari as PWS schemes, under the assumption that land rehabilitation through reforestation would increase the total lake water and result in a better water supply for commercial water uses in the long term. Previously, the existing local government regulations stated that 30%, 35%, and 35% of the surface and ground water tax was allocated to the province, district producing the water, and other districts in West Sumatra, respectively, without any share given to the local communities at nagari level.

Fig. 5. Water balance of Singkarak basin with different land-use scenarios (Farida et al., 2005). Note: A: Effect of land-use change scenarios on various terms of water balance averaged over a 20-year period; B: Effect of changes in mean annual rainfall on the performance of the HEP relative to its design capacity, in anaverage, worst and best year.

2.4. Talau in East Nusa Tenggara

The Talau watershed was the second site for the EPWS project targeting better water supply from springs for domestic users and the PDAM in Atambua. Talau is a trans-national watershed that is in both Indonesia and Timor Leste. Rivers from the watershed drain into the Ombai Strait in Timor Leste. Lahurus and Motabuik are two important sub-catchments in Talau, representing 2% and 15% of the total watershed area, respectively. Grassland was the dominant land cover in the area (66%); forest constituted only 1% of land cover.

The dominant ethnic groups in the Belu Regency were the Te-tun, Dawan, Bunak and Kemak. All these groups had strong cultural traditions that influenced their daily lives. Customary law (adat) influenced natural resource management. The Belu people recognized three strata of law: (i) Kneter/Neter or way of life; (ii) Ktaek/Taek or norms; and (iii) Ukun badu or taboos and restrictions. The last stratum set the rules for natural resource management, stating that natural resources (e.g. soil, water, large rocks, large trees, and mountains) were considered sacred but were also collectively owned.

2.41. Local and policy-maker viewpoints

Local knowledge regarding seasons and climate was tightly connected to knowledge of the planting calendar because of the long dry (8 months) and short rainy (3-4 months) seasons. The severe dry season affected local plant selection. Moreover, locals believed that the forest had an important role as a groundwater provider, regulator, and source of livelihoods. Local people also had a well-articulated understanding of the relationship between vegetation, soil, and water availability. According to them, species with deep roots that could hold water in the ground, such as betel nut, mahogany and candlenut, were suitable plants for the spring area. Local people said that planting teak near springs was not good because the species required a lot of water and did not maintain water in its roots or trunk, releasing water into the air. The forest was associated with the existence of springs. Tree density and tree species were largely connected to groundwater availability. Trees tend to hold rainwater, maintain groundwater and prevent erosion.

The local stakeholders had institutionalised the protection of water sources, access to water, and water allocation. Sub-ethnic groups, or clans, treated springs as sacred groves and controlled their use. The forest surrounding springs, known locally as mamar, was protected from livestock and loggers. People who belonged to the clan were allowed to use some economically commercial plants, such as sirih (Piper betle) and pinang (Areca catechu). In the past, only clan members were allowed to use water from springs. People from other clans had to ask permission and would be penalized if they refused to comply with the owner's rules. However, since the early 2000s, adat law had stopped applying such strict water regulations because of weak enforcement and a sudden increase in migration of refugees from Timor Leste in 2000. This triggered conflicts over water use in some parts, primarily due to water distribution to areas outside the village where the source was located.

2.4.2. Hydrological modeling results

The Talau watershed river flow was largely seasonal; the risk of flash floods was especially high in the latter part of the rainy season when the landscape's storage capacity was saturated and heavy rainfall passed to the river without much buffering. Buffering capacity was lower in years with high rainfall and consequently high total water discharge. The landscape water balance in both the Talau watershed and the Lahurus sub-catchment showed strong seasonal differences (Fig. 6). Actual evapotranspiration was much lower than the potential evapotranspiration due to the strong

seasonality of rainfall and limited storage capacity of soil water.

From an eco-hydrological perspective, it is likely that planting more trees in the area, as suggested by both locals and policymakers, will not substantially increase low flows, potentially even decreasing the current base flow (Fig. 7). Model estimates showed that converting non-productive land (defined as grassland and bush/shrub land) into agroforestry systems or forest did not change the annual low flow. Nevertheless, adding trees reduced surface runoff and increased soil quick-flow. This result implies that rainfall does not immediately reach the river, increasing the watershed buffering capacity. Consequently, flash flooding can be avoided. Assuming that runoff is highly correlated with soil erosion, a reduction in surface runoff also suggests a reduction in soil erosion and therefore improved water quality.

2.4.3. Implications for watershed management as collaborative products

Reviving local spring management is essential to solve internal water conflicts and establish better water distribution, as the water quantity problem faced by local communities is more sensitive to climate variability and the balance of water supply and

Fig. 6. Estimated annual water balance of Talau watershed and Lahurus sub-catchment during the rainy and dry seasons.

Fig. 7. Water balance of Talau watershed and Lahurus sub-catchment for different land-use scenarios (Lusiana et al., 2008d).

Table 3

Analysis of knowledge diversity and its management implications for sub-watershed (local) and watershed levels for each landscape configuration

Configuration

I. Kapuas Hulu

III. Sumberjaya

IV. Singkarak

V. Talau

Knowledge diversity

Initial perceived problem by local stakeholders

Perceived causes of degradation by local stakeholders

Hydrologist findings

Increased sediment yields, thus decreasing the drinking water quality.

Forest conversion to agriculture and illegal logging. Illegal mining.

Increased sediment yield, thus clogging the HEP electricity generator and causing low electricity production.

Forest conversion to agroforestry coffee gardens as the major cause of conflicts between Forestry Service and local communities.

Low runoff showed that the watershed continued Sedimentation was primarily caused by instable to function well with the current land practices watershed geological characteristics.

and changes.

Intensive use along the riparian zone causing riverbank collapse and river edge cutting for boat transportation were sediment yield sources.

Knowledge co-production and collaborative products

Management implications at Reviving the Tembawang traditional agroforestry sub-watershed or local system near the riparian zone to help reduce soil level erosion pressures.

Management implications for watershed level

Driver

NGO-driven: project started as part of the EPWS by WWF/CARE/IIE.

Supported by ICRAF for technical assessment.

Young coffee plantations (less than 3 years), landslides (occurred in the forested area), riverbank collapse, and dirt footpaths were sediment yield sources.

Simple sediment retention construction, including dirt path compaction, and planting deep root trees is useful to reduce surface erosion.

Conservation in the riparian zone involves village members along the river. Raising awareness about not cutting the river banks for boat transportation.

Maintenance intact upper watershed forests as a potential for REDD + type schemes. Law enforcement for illegal logging and mining, and limiting logging permits.

PWS and other instruments for provision of ecosystem services

Maturity of the initiative 2006-2008 (Phase 1)

Conservation in the riparian zone involves village members along the river. Managing coffee gardens by applying simple construction and multistrata tree-planting.

2002-2010 (RUPES Phase 1 and 2) 2011-ongoing (upscaled by local stakeholders)

• International research organization-driven: project started as part of RUPES by ICRAF and partners.

Floods and decreased lake water levels, thus disturbing the HEP operations.

Upstream watershed deforestation, while rehabilitation efforts both not improving watershed functions and not supported by local communities.

The decreased water level was caused by ineffective watershed buffering in retaining water during the rainy season. The downstream water quality was affected by high domestic and agricultural pollution. Floods were mostly caused by HEP river diversions.

Reforestation uses trees with low evapotranspiration.

Local wisdom maintains clean upstream water and conserves the native ikan bilih.

Upstream village level: maintaining current intact environment, i.e. biodiversity conservation, such as organic coffee. Land rehabilitation by maintaining multi-functionality of local tree species. Villages surrounding the lake: improving water quality in lake and connecting river.

2002-2010 (RUPES Phase 1 and 2) 2011-ongoing (upscaled by local stakeholders) • International research organization-driven: project started as part of RUPES by ICRAF and partners.

Decreased the spring water supply.

Deforestation surrounding the water spring.

Lack of spring water was predominantly caused by climatic changes and ineffective buffering in the watershed. Overconsumption and unwise spring water usage worsened water management and caused conflicts.

Reviving local spring water management wisdom to help solve internal conflicts.

Creation of embung (artificial reservoir) to collect rainfall water for domestic use. Promotion of tree-planting of selected species to increase watershed buffering. Spring water management with wise consumption and regulated extraction by PDAM.

2006-2012 (Phase 1 and 2)

NGO-driven: project started as part of the EPWS by WWF/CARE/IIED.

Supported by ICRAF for technical assessment.

Intermediaries/Facilitators

WWF Indonesia

ICRAF (initial)

FKKT HKM, a grass root institution (2nd and 3rd phase) and University of Lampung for monitoring

ICRAF (initial) • CARE Indonesia

CO2 BV (for voluntary carbon market) • WWF Indonesia

A lake management forum

Expected watershed function by users

Sedimentation reduction Regular flow

Sedimentation reduction

Rehabilitation of watershed protection forest

Rehabilitation of watershed protection forest

Good water quality of the lake

Regular flow

(Expected) Mechanism

Earmark payment from water bill

25-year permits for community forestry

Distribution of royalties from a state-owned HEP company

Earmark payment from water bill

ar a ec

demand, particularly in dry periods, than to land management activities (i.e. tree planting). At the wider scale, wise consumption and regulated extraction by the PDAM are important for protecting water availability for the villagers surrounding the springs. Since local communities have useful knowledge of selecting tree species that are suitable for soil and water conservation, collective tree-planting becomes one of the options to increase the watershed's buffering capacity.

In 2007, to ensure collaborative watershed management among local communities and government, a Memorandum of Understanding (MoU) was signed between the Belu district government and the Lasiolat community group, which represents seven villages in the Lasiolat sub-district of Belu. The MoU set out the general roles and responsibilities of both parties. As a result, the local government allocated funds from its annual budget through the relevant Belu district service (i.e. the Forestry and Plantation Services) for watershed conservation. In 2008, they allocated approximately USD 48,000, estimating a similar allocation for the years ahead.

3. Case comparison and analysis: knowledge diversity and coproduction

it th wn

For each case study, we identified knowledge diversity generated by stakeholders relevant to watershed management, from the perspectives of both local communities and government officers. In this section of the study, we analyzed the synergy and contrasting views from both viewpoints, then investigated how the scientific findings could be integrated to co-produce a collaborative product for better watershed management (Table 3). Collaborative products take the form of village and community agreements, and joint watershed management plans, as the basis for PWS contracts that include activities for reviving local wisdom. Further, we analyzed the effectiveness of this boundary work by highlighting its use of knowledge (enlightenment, decision-making and negotiation support) for development and implementation of PWS schemes as summarized on Table 4.

M w < X ■

3.1. Knowledge diversity among actors

Most upstream communities in the study sites use water for domestic consumption and smallholding agriculture, such as paddy fields, fishponds, and plantations. The communities recognised the importance of ecosystem functions, such as forests interacting with rivers, and utilized these functions as services to benefit their livelihoods, including cultural rituals. Therefore, perceptions of hydrological problems were directly related to processes influencing daily and cultural activities, and income. Additionally, communities perceived that they could endure watershed problems caused by external stakeholders, such as HEP construction that worsens flooding, a PDAM that reduces local water supply, or a concession company that performs extensive logging upstream.

All farmers were able to describe in detail different elements of water storage, quality and flow within their landscape, the interactions among them, and their cause-effect relationships. Reduced predictability of river flow and reduced water quality were quoted as major issues. Local communities tended to focus on plot-level solutions and had limited ability to formulate descriptions of large-scale ecological processes. The communities applied a variety of techniques to solve their watershed problems. The solutions were fairly consistent among the sites, even though the sites were geographically dispersed. For example, people at all sites consistently mentioned mahogany as an example of a tree species that retained water and pine trees as an example of a species with high

¿1 ZX T3

y O S c

v eh, Í3 6 3

te E tJ <u 5 <u

c 23 &

•2 S 4= 3 n <e

5 13 ^

ra ÍS 'D ra

■sac

c T o it tfig.

u ra "ra

¡S 73 S

<D "3 c

^ ra m +

■3 >-

^ OJ QJ ^

{3 V ro

•a &

• T3 _ co

ÍS^ 'D

6« 3 =

Mu g ! "P

.5 c <13 ■

l^jj c

'ÍS o

E 5 ™

T3 d g

'i: ^ &

a <u ^^

^ ^ "O

^ra ^ .s -c -

p^íá^ -

G ro £ ii

J-i -u ^ C E—fO C fO(U

i esí

ra r1-^

raJHti > fi

C-- £ -c

■S-6.& ^

iS « ^ ^^^

ra -n u ^

¡Ú QJ

u -3<U

water use that reduced stream flow. In contrast, forestry officials attributed positive effects to all trees regardless of species. More generally, knowledge in the public/policy arena seems to converge on, and justify, the way regulations have been framed, rather than being open to challenges that derive from actual observations.

In finding solutions, local community members sought location-specific solutions and the general public and policy stakeholders referred to generic solutions, as also highlighted by Verbist et al. (2005) and Schalenbourg (2004), i.e. large-scale forest protection and rehabilitation through reforestation, as important actions for responding to floods, soil erosion, and riverbank abrasion. These actions can be contradictory and, ultimately, the measures applied do not meet their conservation goals. For example, the Singkarak case showed that solutions preferred by policy-makers to solve watershed problems (mass reforestation by planting pines) could cause problems for other stakeholders (drying out water resources due to pine's high evapotranspiration rate). In Kapuas Hulu, watershed solutions focused more on the removal of inappropriate policies, such as granting permits to logging companies, rather than changing local-level land use or applying other environmental economic instruments.

3.2. Knowledge co-production of collaborative products

3.2.1. Enlightenment

The knowledge identified from the local and public domains was verified through hydrological modeling. Cross-site analysis showed that the reality check provided by this formal and systematic mode of knowledge creation (i.e. applied quantitative hydrological research) enriched information about causes of, and solutions for, watershed problems. This applied knowledge on hydrological functions of watersheds aims at being contextual and specifically packaged to address the needs of PWS design and negotiation.

In some cases, a supply-and-demand imbalance for provisioning ecosystem services (e.g. water allocation in Talau and over-fishing in Singkarak) and human-induced activities with no direct relations to land-use change (e.g. cutting the river bank in Kapuas Hulu) caused more watershed problems than local (upstream) land-use practices. In the Sumberjaya case, coffee plantations in the multistrata systems could produce litter layers that prevented soil erosion. This was different from the previous perception in which all coffee plantations were blamed for river sedimentation.

Furthermore, the simulation results helped to ascertain whether stakeholder perceptions, understanding the hydrological situation, and solutions to tackle emerging problems actually represented the status quo or provided information about future problems. For example, the landscape water balance analysis in Kapuas Hulu showed that the basin condition in the business-as-usual scenario was similar to the forested condition, suggesting that this scenario maintained its hydrological function because the tree cover remained 'pristine' with nearly 100% tree cover. As environmental management implications for PWS, any interventions to increase watershed functions become less significant and payments for watershed functions may not be relevant in this situation compared to maintenance of forest cover under a REDD + scheme, for example.

An example where local knowledge enriched scientific analysis can be seen in the introduction of a 'flow persistence' parameter in hydrological models (Van Noordwijk et al., 2011), based on a recurrent concern over loss of predictability of river flow during watershed degradation. This concept became a key to a recursive model of river flow that can efficiently describe the health of a watershed in the way it processes rainfall, without requiring knowledge of rainfall events.

3.2.2. Decision

A nested management decision to improve watershed functions and services was the most significant collaborative product to emerge from the RHA application. In each watershed, this was translated into two spatial scales: local and watershed. Local watershed management referred to a scale that was optimal for a cluster or landscape with specific and homogenous characteristics, inhabited by people in close proximity, as in sub-watersheds. Such a cluster usually consisted of several farmers' groups in one or two villages, but sometimes more, depending on the size of the watershed.

The amount of knowledge generated through the boundary work that is applied in decision-making varies. In Kapuas Hulu, the intermediary decided to concentrate on sub-watershed activities by increasing productivity of the local rubber-agroforestry system and riparian rehabilitation in some parts of the watershed. This knowledge was also used to establish water governance systems at different levels (i.e. village level, communities surrounding the lake, and the nagari forum in Singkarak; a community forestry scheme at the watershed level and conditional corporate social responsibility by the HEP company in the riparian zone in Sum-berjaya). In Talau, however, the NGO facilitator decided to run the business-as-usual activity of planting trees surrounding the spring.

3.2.3. Negotiation

The degree of negotiation that occurred is linked to the results of boundary work and aligns with the decisions made by the related stakeholders at each watershed. These decisions were clearly influenced by various factors. In Kapuas Hulu and Talau, as previously mentioned, the project objectives were to establish business models on quantifiable PWS at each site. Thus, although in Kapuas Hulu, for example, the hydrological assessment recommended a stronger business model for carbon-stock maintenance compared to quantifiable watershed function, the NGO, as the grantee, directed the project towards the latter: PWS. This led to challenges in providing credible options for the water investors, as the interventions under this project did not sufficiently increase and improve watershed functions. In both Kapuas Hulu and Talau, the projects targeted development funds from their respective local governments through MoUs containing broad indicators for watershed management at the village level, which were not performance-based schemes as suggested in the PWS. On the other hand, in Sumberjaya and Singkarak, results from boundary work significantly influenced the outputs of the negotiation process, particularly with regard to designing PWS mechanisms, establishing elements of the contract and its monitoring systems, and the overall structure of watershed governance. This was due to the consistent efforts of facilitators in applying results from boundary-work activities. Table 4 summarizes variations in the use of knowledge in each landscape configuration.

4. Discussion

The present study has analyzed the role of an action-research programme in stimulating a process of knowledge co-production through collaborative learning between 'experts' and 'users' involved in developing and implementing PWS in Indonesia. Our case studies showed that improving information availability, through recognizing the value of interfacing and sharing diverse knowledge, is a prerequisite for increasing the quality and sus-tainability of PWS programmes. The review found that the factors influencing the design and implementation of PWS programmes vary, and extend beyond multi-perception ecological knowledge and scientific data availability (Leimona et al., 2009).

4.1. Challenges and potentials for effective boundary work

4.1.1. Enlightenment

The basic use of knowledge engaged during boundary work is intended to advance general understanding that may influence the decisions of multiple stakeholders through knowledge sharing and diffusion (Clark et al., 2011). In the context of this study, the process of boundary-work enlightenment constructs a body of knowledge that is technically adequate for the handling of hy-drological evidence based on the compilation of stakeholder perceptions and interdisciplinary scientific research under the RHA. Our case studies showed that in all cases, the boundary work provided information on ecosystem services' functions, including their baseline and dynamics under different scenarios of land-use practices. The knowledge diversity contributed to enhancing the credibility of the boundary work, particularly when it was applied to the ungauged basins reflected in our case studies. The literature captures this finding and consistently suggests similar approaches, such as the integration of human-water systems deploying concepts such as socio-hydrology (Sivapalan et al., 2012), hydro-psychology (Sivakumar, 2011), and cross-disciplinary integration (Wagener et al., 2010). Further, a sound scientific knowledge alone may not be sufficient in providing the information needed to achieve socio-ecological sustainability for natural resource management (Puente-Rodríguez et al., 2014). Instead, the inclusion of different stakeholders is needed from the early stages of the project to structure the information and facilitate the integration of different types of knowledge from various points of view, to enhance the integrated natural resource management structure.

4.1.2. Decision-making

As previously mentioned, many conservation actions are only based on general beliefs (e.g. that planting trees in an upper watershed can increase the stored water volume of a downstream lake). Moreover, many people firmly believe that planting trees can solve all environmental problems. Potential buyers or consumers of ecosystem services may have motivations based on this common myth and assume that certain conservation practices are able to enhance provision of ecosystem services to their benefit. The scientific fact that planting trees can actually reduce the base flow due to increased evapotranspiration, as shown in the Sing-karak case study (Van Noordwijk et al., 2007), may reduce investor motivation to participate in any PWS scheme when the buyer's interest is increasing water quantity. Moreover, an incomplete understanding of forest versus watershed problems can produce undesired results, i.e. a misconception that reforestation is not important. Benevolent environmental agencies acting as intermediaries might not disclose these 'contradictory' facts; they may consider the strategic use of scientific information, selecting the most desirable results in order to dissuade buyers from withdrawing from the scheme. Asheim (2010) and Sivakumar (2011) presented empirical evidence of this effect. Therefore, boundary work may provide a salient basis for decision-making but many other factors contribute to final decisions, which in turn influence the negotiation process as described below. And, interestingly, Lusiana et al. (2011) pointed out that potential model users for a resource-management model prioritized salience as an important characteristic compared to credibility and legitimacy. Van Noord-wijk et al. (2014b) discussed the perception that watershed-management projects have shifted too far towards satisfying landuser needs and are not providing measurable improvement in hydrological services.

4.1.3. Negotiation

During the negotiation process, buyers and sellers of ecosystem services would have to strategically consider the various

Hydrological function and target of intervention and applicability of 10 PWS prototypes (expanded from (Van Noordwijk, 2006)) in the various landscape configurations, as represented in the case studies.

Primary watershed issue Indicative ranking on degree of reversibility by human intervention at each landscape configurationa PWS prototypes: plausible actions by land users to enhance hydrological function Business model: (insurance premium) for or on behalf of beneficiaries Potential basis for joint monitoring of conditional contracts

I II b III IV V

Water yield (Annual blue-water yield versus green-water use.) n + + + + + + + WY1: Restoring vegetation-level water use, and hence subsurface and surface flows to that of natural vegetation from values that are lower (less or more-open vegetation) or higher (fast-growing trees)c; WY2: Maintaining ecological flows that support aquatic life forms (and associated fisheries etc.) • Avoidance of changes in overall hydrology with associated risks • In specific landscapes: avoidance of recirculation of subsurface salinity • Continued aquatic systems with associated services • Total vegetation cover and/or specified tree frequency by size, location and type, linked to water balance models of usable blue water in relation to rainfall • River flow in specified locations, with disclaimers (force majeur) for extreme weather episodes • Native aquatic species diversity

Rainfall pattern elsewhere (rainbow water) + + + + + - WY3: Maintaining green water use as contribution to atmospheric recycling for downwind rainfall Increase of down-wind rainfall and buffered rainfall variability relative to ocean-temperature patterns • Vegetation type, linked to location-specific coupled vegetation-rainfall-climate models

Flow pattern of blue water n ++ +++ + + + WF4: Increasing rainfall infiltration, maximizing use of slow-release groundwater pathways, reducing flood volume and duration (increased flow persistence) WF5: Modifying operating rules for reservoirs and hydropower schemesc • To reduce flood disaster risks • To enhance dry season flows, extending time period for usable blue water • Regular water supply for run-of-the-river hydropower generators (including micro-hydro plants) • River flow persistence (predictability in time series) • Presence of vegetative cover and/or surface litter as influence on infiltration

Controlling sediment load of rivers n +++ +++ + + + + WS6: Enhancing sediment filter strips in fields and across landscape matrix WS7: Protecting river banks, riparian zones and landslide-prone slopes • To improve efficiency of reservoir-based and run-of-the-river hydropower operators and drinking water provisioning filters • Avoid mudflows and dam bursts • Avoid marine sedimentation on coral reefs • Sediment load of streams and rivers at specified observation points • Vegetative and litter layer sediment filter zones • Vegetation in riparian zones

Water quality n ++ + + + + + + + WQ8: Protecting springs and sources of domestic water use WQ9: Reducing point and distributed (nonpoint) sources of pollution WQ10: Waste-water treatment to match biological recovery from (organic) pollutants. • Water quality for domestic and industrial use • Reduced costs of water treatment • Aquatic ecosystem services and effects on coastal marine systems • Biological water quality indicators (biota-based) • Biological oxygen demand • Escherichia coli counts • Agreed measures to control point sources and reduce nonpoint sources (e.g. agrochemicals)

Note: WY=water yield, WF=water flow, WS=water sedimentation; WQ= water quality "Flooding, as occurred in Bangkok in 2011-2012 in the Chao Praya basin, can be due to end-of-season rains at a time all reservoirs are full, to maximize dry-season water availability.

a means prime watershed conditions conducive for all watershed services' provision; + indicates degree of reversibility in ecosystem services' provisions by certain human interventions in specific configurations: + =weak; + + =medium; + + + =strong.

b This configuration was not represented in the four Indonesian cases, but an example from Northern Thailand was discussed by van Noordwijk et al. (2014a). c Green-water use is directly related to local benefits through micro-climate effects on air humidity and temperature.

opportunities and threats to accomplishing both their own and others' goals (as well as potential losses). In an ideal situation, actions have to take place at a sizeable scale and timeframe for providing real impacts for provision of ecosystem services, targeting benefits for external stakeholders. In environmental-management negotiations, however, stakeholders tendentiously prefer 'starting with easy wins rather than the most urgent issues' (Van Noordwijk et al., 2004). This tendency also seems to apply in our case studies, where implementing agencies selected pilots with complete historical data, or strategically interposed alternative agendas, instead of establishing a PWS scheme. In addition, donor obligations have a strong influence on determining hotspots or a targeted pilot area in which the defined ecosystem service selection is made. Often these choices are not based fully on scientific facts but are purposively chosen as the option corresponding best with the project design documents, or the location nearest to the potential buyers. This is not uncommon because a donor indicator for a successful PWS scheme is often skewed towards achieving a successful transaction between sellers and buyers, with a contract, clear business model, or MoU signed by both parties. The case of community conservation concessions in Indonesia indicated that negotiation on project time horizon, budget amount, and the notion of performance-based payment with donors were all essential to the success of a payment for ecosystem services project (Wunder et al., 2008). In the domain of public policies, there is a long history of selective use of science (Galudra and Sirait, 2006), where watershed functions are barely monitored, quantified or analyzed scientifically.

The presence of facilitators during boundary work is needed to ensure that the PWS initiative functions with fairness, efficiency and legitimacy. Boundary organizations, such as NGOs and academic research institutions, may act not only as experts and disseminators of knowledge; they can also facilitate dialogue and negotiations between multiple stakeholders. Thus, there must be a watershed governance system that is able to bridge and accommodate different types of knowledge, that can use and integrate this combined knowledge-base to define real problems, and which can be used to facilitate dialogue to develop appropriate and legitimate PWS schemes.

4.2. Prototype of PWS for increasing negotiation effectiveness

This final section presents a synthesis of the preceding boundary-work discussion, and broadens it to include prototype PWS schemes that may apply across various different landscape configurations (Table 5). It has been observed that effective boundary work implies that stakeholders at each landscape configuration co-produce both scientific and participative options for managing watershed complexities and can consecutively govern several prototype PWS schemes in a single watershed (Table 3). Thus, to comprehensively target different aspects of the hydro-logical cycle — as represented by the blue, green, gray and rainbow waters of the 'colors of water' model (Van Noordwijk et al., 2015)3 — we expanded the sample range of landscape configurations by extracting the results for Configuration II from Van Noordwijk et al. (2014a), as previously introduced in Section 2 as an addition

3 Blue water in streams and rivers has been the traditional focus of hydrology and institutions for watershed management, but on average it only constitutes about 40% of rainfall. It is complemented by recycled gray water returning to rivers after human use. Water in the soil, used for evapotranspiration, is termed green water and has become part of the debate, especially where fast-growing trees use more water than average vegetation and reduce blue-water flows. More recently green-water use is recognized as part of an atmospheric recycling pathway, with globally about 40% of terrestrial evapotranspiration returning as rainfall over land (van der Ent et al., 2010; van Noordwijk et al., 2014b).

to our current configurations (I, III, IV, V). Table 5 presents the basic information for enabling more effective negotiation of PWS, drawn from the results of knowledge co-production in boundary work and refined the work from Van Noordwijk (2006). The information addresses questions, such as: Which prototypes of human efforts in each landscape configuration would most effectively lead to a positive ecosystem services' increment? To what degree is each landscape configuration responsive to such efforts? What business models can be built from each prototype? And, what are the local indicators for joint ecosystem services' monitoring under a performance-based scheme?

Configuration I is a prime landscape in watershed services' provision as shown in Table 1 on landscape configuration, so efforts to protect its vigorous condition are essential for maintaining the flow of services from this landscape. A small-scale rehabilitation might be possible as a PWS scheme, though its contribution to solving the wider problems of watershed management at this scale is negligible. Compared to the other configurations surveyed, this forested landscape of a sizable scale, such as a region or continent, may have a relatively important role for influencing the process of returning water molecules to the atmosphere and may affect rainfall elsewhere (WY3 on rainbow water) (Van Noordwijk et al., 2014c). For increment of total water yield, rehabilitation efforts such as planting trees with a low evapotranspiration rate on degraded or low tree coverage landscapes (e.g. Configurations III and V) might be compelling to the beneficiaries of the watershed services. When spatial and time-scale coverage are feasible, monitoring of the water balance (WY1 on annual blue-water yield), and native aquatic species diversity (WY2), might be included in the performance-based contractual agreement under PWS.

Human interventions on landscapes with a high mosaic of patches, as clearly represented by Configuration III and to a lesser extent by Configurations II and III, have an important role in regulating the water flow or continuation of flow during the dry season and for avoiding high peaks and floods in wet months (WF4). One plausible intervention is to maintain good soil structure to ensure a better rate of rain infiltration into the soil.

Soil erosion, a condition that is common to landscapes where human intervention has occurred, is represented by Configurations II and III. Soil erosion can be reduced by the presence of a litter layer on farmers' gardens and enhancing sediment filter strips (WS6), and by protecting fragile hot-spots by adding deep-rooted vegetative strips or other simple constructions (WS7). In addition, engineered infrastructure may also help to reduce the impact of erosion by improving the efficiency of hydropower operations and drinking-water provisioning filters.

Similarly, human efforts in degraded and mixed landscapes, as represented by Configuration III, can help to improve water quality for domestic users by reducing water pollution (WQ8-10). Practising organic agriculture decreases water pollution in streams and slowly reduces nitrogen and pesticide content that has been absorbed into deep soil layers and polluted ground water. Additionally, raising awareness among people residing around water bodies, such as streams, lakes, and springs, may help to reduce direct pollution from domestic waste and contamination by Es-cherichia coli due to poor sanitation.

In the context of knowledge diversity and co-production, landscapes with rich mixed natural and human patches, as seen in the middle-range landscapes of configurations II—IV, in contrast to landscapes with dominant land covers of forest (Configuration I) and grassland (Configuration V), tend to have a fertile combination of stakeholders (indigenous and migrant local communities) and more potential buyers of ecosystem services (such as water companies and domestic urban dwellers) because these landscapes are usually in the proximity of highly populated cities. Landscape and stakeholder characteristics (including ecosystem services' providers,

buyers and intermediaries) may influence the effectiveness of boundary work for PWS negotiation. Some configurations may provide more options for human interventions compared to others, since these landscapes are more sensitive to ecosystem services' reversibility towards its increment. On the other hand, the nature and level of interactions among such diverse actors may increase potential for abundant knowledge co-production (Edelenbos et al., 2011; Petts and Brooks, 2006), since knowledge can flow both downwards from land managers (as providers) to downstream stakeholders (as beneficiaries) and upwards from downstream stakeholders to upstream communities. This interaction can potentially enrich both, as knowledge is a non-consumable resource, which is not reduced by its use.

5. Conclusion

Early and thorough analysis of the knowledge diversity and coproduction of boundary work for designing and negotiating incentives for enhancing watershed services could increase the quality and sustainability of these emerging policies and schemes. This can be achieved by acknowledging common hydrological issues among diverse relevant actors and providing objective forecasts for how ecosystem services will respond to watershed management (enlightenment), enhancing the quality of decision-making, thus maintaining good social capital between stakeholders, and finally increasing levels of fair negotiation on PWS, an essential final result for any PWS in a developing country context.

With regards to knowledge diversity, we found that local perceptions of hydrological problems were primarily related to processes influencing daily activities and income. Watershed problems were perceived to be caused by external stakeholders. Descriptions of detailed elements within their landscape were shared, including the interactions among them and their cause-effect relationships. Local stakeholders tended to focus on plot-level solutions and had limited ability to formulate large-scale ecological processes. Policy makers referred to generic solutions, which often contradicted to more specific local knowledge. This led to measures that did not meet the agreed conservation goals, and which often focused on regulation of land use activities rather than specifying a desired and achievable hydrological service level. On the other hand, technical hydrological analysis of location-specific water balance models helped to ascertain whether stakeholder perceptions of the hy-drological situation, and solutions to tackle emerging problems, actually represented the status quo or provided information about future problems.

Design and negotiation of PWS will often require the assistance

of intermediaries who can integrate different knowledge systems and bridge between disparate stakeholder interests, offering options for integration into appropriate and legitimate schemes as a basis for negotiations. We recognize 10 different prototypes of PWS schemes that can be combined or used on a stand-alone basis in a watershed setting. This format leads to a better understanding of how payments can be channeled to effectively enhance, or at least maintain, the underlying hydrological functions at each landscape configuration.

Our case studies and other global experiences indicated that currently practiced PWS schemes have remained at relatively small scales and pilot levels (villages or sub-watershed levels) and their long-term sustainability of the emerging schemes cannot yet be empirically judged. Most schemes were donor-driven, with a limited budget and time frame, because this approach is relatively new and requires a large investment to mature. Nevertheless, the results showed that the recognition, appreciation, and use of effective boundary work in the early PWS scheme planning and designing stages has demonstrated the relevance and pertinence of this operational model. It has allowed for effective communication strategies and also permitted intermediaries and project managers to facilitate negotiations between providers and beneficiaries of ecosystem services, to work towards operational and sustainable payment systems and to some extent provide inputs to enhance PWS performances. A common knowledge base for interpreting measurements is a prerequisite for increased performance-based conditionality. For some of the PWS prototypes and in some landscape configurations this may be harder to achieve than elsewhere, and further empirical testing of our typology is needed.

Acknowledgment

We acknowledge the financial support from International Fund for Agricultural Development (IFAD), the European Union, and the Deutsche Gesellschaft für Technische Zusammenarbeit GmbH (GTZ) of the Federal Ministry for Economic Cooperation and Development (BMZ), and the CGIAR Research program on Forests, Trees and Agroforestry (FTA). The views expressed do not necessarily reflect those of the funding agencies. We also appreciate supports from our colleagues from the World Agroforestry Centre: Peter Minang, and Robert Finlayson for editing the manuscript, WWF Indonesia: Tri Agung Rooswiadji, Azhar, and Hermayani Putera during the preparation of this study.

Appendix 1. Method of applying the Rapid Hydrological Assessment (RHA)

The Rapid Hydrological Assessment for Payment for Watershed Services employed a combined qualitative and quantitative research methodology, encompassing both primary and secondary analysis of empirical evidence from the Indonesian cases (Table 1A). The four case studies were based on hydrological assessment (Jeanes et al., 2006) to gather information and synthesize the two key knowledge systems from the perspective of 'users/managers/decision makers', i.e. local, general public/policy-maker, and 'experts' i.e. modeler/hydrologist ecological knowledge.

The local and public/policy-maker knowledge acquisition method was modified from the knowledge based system approach (Dixon et al., 2001). First, stakeholder analysis was performed to define the actors and their roles in watershed management. Then, in the knowledge articulation step, the perception of local stakeholders and policy-makers on hydrological functions, water movement and the consequences of land use options on the landscape were determined by conducting a series key informant interviews and participatory focus group discussions targeting relevant stakeholders as the result of the stakeholder analysis. Local people are the actual land managers who work and interact with the watershed landscape on a day-to-day basis. Regency- and provincial-level policy-makers are actors who have been given a mandate to control and manage watershed areas, assuming that the policies they create will have a pronounced effect on future watershed conditions. In addition to that, the researchers conducted a series of participatory landscape appraisal (Hoang et al., 2013) as an early diagnostic tool of the issues in a landscape by doing a transect walks and semi-structured interviews with the farmers

Table 1A

Research components of local, policy-makers, and hydrologist ecological knowledge

Local ecological knowledge

Goal Local-specific analysis of the problem and its cause and effect.

Source of information Important informants and village members

Documents needed Base map as a foundation for participatory mapping

Questions asked and topics explored Where are the hot-spots within the watershed that cause degradation?

What are the existing land use patterns in the watershed? Who contributes to the current land use pattern? Why have these land-use patterns developed?

What are examples of areas that decrease or buffer watershed degradation? Do good practices for solving watershed problems exist? What are those practices? Public or Policy-maker Ecological Knowledge

Goal Analyse perceptions regarding watershed-level environmental and water resource problems and their causes and effects.

Source of information Government officers, community leaders, and the general public, including downstream stakeholders

Documents needed Base and thematic maps

Environmental reports and watershed profiles Questions asked and topics explored What and where do watershed problems occur? Who caused the watershed problems? What are the reasons?

What are the past and current (1) land use, (2) forest cover, (3) river flow, (3) water quality and use, (4) lake, and (5) river problems?

Are any developmental projects planned within the watershed? Will these projects cause environmental degradation? Modeler or hydrologist ecological knowledge

Goal Plausible land use change scenarios to analyse drivers and effects to watersheds

Source of information Land use modeler and hydrologist

Documents needed Spatial data: topographic, landform, geology, soil, natural vegetation, land use time series and administrative maps

Climatic data: daily rainfall Hydrological data: daily water level Questions asked and topics explored What changes have occurred in the watershed? What are the land use change drivers?

How do land use changes affect water balance and use within the watershed? What are main indicators affecting watershed water quantity and quality? What are land cover effects on watershed water balance and river flow?

and other relevant stakeholders in each watershed.

Existing publically available data were used for the general hydrological model characteristics, ensuring that the process is repeatable for sites in different climatic zones. The hydrological model activities (Jeanes et al., 2006) are as follows:

• Gathering and reviewing existing relevant information on climatic and hydrological watershed data, including rainfall and river flow data and land cover maps;

• Analysing land cover/land use change and its consequences on water balance, including the watershed river flow;

• Modeling the watershed water balance, including scenario analysis for plausible land cover changes and their likely effects on watershed functions.

The hydrological model recognizes multiple scales (De Groot et al., 2010; Ranieri et al., 2004) that range from the plot level, where infiltration is affected by the topsoil conditions; to the stream level, which generally involves multiple farms; to the river level, which is influenced by domestic water use, waste management and land use; and finally to the catchment level, which may include industrial and (semi) urban use. To model the effects of current and future land use change on watershed hydrological functions, we applied the GenRiver 2.0 computer software model (van Noordwijk, 2002; van Noordwijk et al., 2010) with a climatic and hydrological dataset for at least a 20-year time-series. GenRiver is a simple water balance model that simulates river flow. It was developed for data-scarce situations and is based on empirical equations. The model can be used to explore the basic changes in river flow characteristics across spatial scales from the patch level through the sub-catchment and catchment levels. Appendix 1 provides the equations derived from Van Noordwijk et al. (2006) for measuring watershed indicators used in the hydrological model.

To analyse the watershed landscape configuration and land use dynamics, spatial data were acquired from satellite imagery for land cover mapping, a digital elevation model for watershed characterization, and thematic maps for landscape configuration analysis. The next step was to process the terrain characteristics for watershed delineation and analysis of land use/cover changes and their trajectories.

Applied research should be transdiciplinary, heterogenous, and directed at solving practical problems. Accordingly, a series of interviews were organized with important stakeholders (mostly project managers) involved in the implementation of PWS schemes. The interviews provided information about the progress of the PWS schemes, scenario types that resulted from the scoping study applied to establish conservation contracts between ES providers and buyers, and strengths and weaknesses of the application of knowledge types for designing and planning a PWS scheme. There were two stages of methods developed for the local and public/policy knowledge. Initially, the knowledge systems were fully and explicitly mapped and analyzed diagrammatically as shown by Joshi et al. (2004b). Later, the method was simplified. Consecutive interviews are only administered when there are significantly important new elements disclosed within initial interviews.

Appendix 2. Six prototype situations for payment for watershed functions in upland agricultural systems

Environmental Service

Providers/Sellers

Users/Buyers

Main Issue

1. Total water yield for hydro-electricity via storage lake

2. Regular water supply for hy-droelectricity via runoff-the-river

3. Drinking water provision (surface or

groundwater)

4. Flood prevention

5. Landslide prevention

6. General watershed rehabilitation and erosion control

Impacts on total water yield small; reservoir sedimentation issue may dominate the debate; option for sediment traps and landscape filters

A change from soil quick flow (saturated forest soils) to overland flow will have some effect on buffering of river flows and hydroelectric operation time

Intensive agriculture and horticulture will cause rapid pollution of surface flows and slow but persistent pollution of groundwater flows with nitrogen and pesticides; people residing around streams cause pollution E.coli and diseases

Land use effects strongest for flow buffering of small-to medium sized events, with saturation dominating the large events

Mortality of deep-rooted trees ('anchors') causes temporary increase in landslide risk

Promoting tree cover and permanence of litter layer protecting the soil is a good precaution

Consumer satisfaction depends on continued functioning; high project investment costs, little subsequent management flexibility

Intercepting sediment flows rather than avoiding them is generally easier to accomplish; sediment flows out of well-managed upper catchments may still be high because of geological and geomorphological processes

Interventions influencing the speed of drainage (linked to paths, roads and drains) have the most direct effect on buffering at larger scales

Willingness to pay for drinking wa- Slow response of groundwater flows

ter depends on quality assurance to changes in the pollutant status

from medical perspective, as well as make 'regulation' a more effective

taste solution than results based markets

Relevance of upland land use depends on location (floodplains) and engineering solutions (dykes, storage reservoirs)

Relevance depends strongly on location in the flow paths

Risk avoidance for the rare category of large events

Deep landslides are little affected by land cover

Communication gap with scientists

'Holistic' perception of watershed functions survives despite the lack of who try to enhance clarity clear impacts on specifics

Source: Van Noordwijk, 2006

References

MA, 2005. The Millennium Ecosystem Assessment Series, Washington DC.

RUPES, 2010. Where we work: brief profile of RUPES action research sites. World Agroforestry Centre, Bogor, Indonesia, p. 54.

Adhikari, B., Boag, G., 2013. Designing payments for ecosystem services schemes: some considerations. Curr. Opin. Environ. Sustain. 5, 72-77.

Agus, F., Gintings, A.N., van Noordwijk, M., 2002. Pilihan teknologi agroforestri/ konservasi tanah untuk areal pertanian berbasis kopi di Sumberjaya, Lampung Barat World Agroforestry Centre Southeast Asia Regional Office, Bogor, Indonesia.

Asheim, G., 2010. Strategic use of environmental information. Environ. Resourc. Econ. 46, 207-216.

Barbier, E.B., Burgess, J.C., 1997. The economics of tropical forest land use options. Land Econ., 174-195.

Berkes, F., 1999. Sacred Ecology: Traditional Ecological Knowledge and Resource Management. Taylor & Francis.

Berkes, F., Colding, J., Folke, C., 2000. Rediscovery of traditional ecological knowledge as adaptive management. Ecol. Appl. 10, 1251-1262.

Braat, L.C., de Groot, R., 2012. The ecosystem services agenda:bridging the worlds of natural science and economics, conservation and development, and public and private policy. Ecosyst. Serv. 1, 4-15.

Cash, D., Clark, W.C., Alcock, F., Dickson, N.M., Eckley, N., Jäger, J., 2002. Salience, credibility, legitimacy and boundaries: linking research, assessment and decision making, KSG Working Papers Series RWP02-046.

Chapman, M.G., 2002. Local ecological knowledge of soil and water conservation in the coffee gardens of Sumberjaya, Sumatra, Forestry. The University of Wales Bangor, Bangor, p. 59.

Chomitz, K.M., Brenes, E., Constantino, L., 1999. Financing environmental services: the Costa Rican experience and its implications. Sci. Total Environ. 240, 157-169.

Clark, W.C., Tomich, T.P., Van Noordwijk M., Dickson, N.M., Catacutan, D., Guston D., McNie, E, 2010. Toward a general theory of boundary work: Insights from the CGIAR's natural resource management programs, CID Working Paper No.

199 Center for International Development, Harvard University Cambridge, MA.

Clark, W.C., Tomich, T.P., Noordwijk, M.V., Guston, D., Delia, C., Dickson, N.M., McNie, E., 2011. Boundary work for sustainable development: natural resource management at the Consultative Group on International Agricultural Research (CGIAR).

Dixon, H.J., Doores, J.W., Joshi, L., Sinclair, F.L., 2001. Agroecological Knowledge Toolkit for Windows: methodological guidelines, computer software and manual for AKT5, School of Agricultural and Forest Sciences. University of Wales, Bangor, p. 171.

Edelenbos, J., van Buuren, A., van Schie, N., 2011. Co-producing knowledge: joint knowledge production between experts, bureaucrats and stakeholders in Dutch water management projects. Environ. Sci. Policy 14, 675-684.

Farida, Jeanes, K., Kurniasari, D., Atiek Widayati, Ekadinata, A., Hadi, D.P., Laxman-Joshi,Suyamto, D., Van Noordwijk, M., 2005. Rapid Hydrological Appraisal (RHA) of Singkarak Lake in the context of Rewarding Upland Poor for Environmental Services (RUPES). World Agroforestry Centre - ICRAF SEA Regional Office Bogor, Indonesia, p. 113.

Galudra, G., Sirait, M., 2006. The Unfinished Debate: Socio-Legal and Science Discourses on Forest Land-Use and Tenure Policy in 20th Century Indonesia.11th Biennial Congress of the International Association for the Study of Common Property, Bali, Indonesia, pp. 1-17.

De Groot, R.S., Alkemade, R., Braat, L., Hein, L., Willemen, L., 2010. Challenges in integrating the concept of ecosystem services and values in landscape planning, management and decision making. Ecol. Complex. 7, 260-272.

Hoang, M.H., Joshi, L., Van Noordwijk, M., 2013. Participatory landscape appraisal. In: Van Noordwijk, M., Lusiana, B., Leimona, B., Dewi, S., Wulandari, D. (Eds.), Negotiation-support toolkit for learning landscapes. World Agroforestry Centre, Bogor, pp. 16-21.

Hrachowitz, M., Savenije, H., Bloschl, G., McDonnell, J., Sivapalan, M., Pomeroy, J., Arheimer, B., Blume, T., Clark, M., Ehret, U., 2013. A decade of Predictions in Ungauged Basins (PUB)—a review. Hydrol. Sci. J. 58, 1198-1255.

Jeanes, K., Van Noordwijk, M., Joshi, L., Widayati, A., Farida, Leimona, B., 2006. Rapid

Hydrological Appraisal in the context of environmental service rewards. World Agroforestry Centre - ICRAF SEA Regional Office, Bogor.

Joshi, L., Schalenbourg, W., Johansson, L., Khasanah, Nm, Stefanus, E., Fagerström, M. H., 2004a. Soil and water movement: combining Local Ecological Knowledge with that of modellers when scaling up from plot to landscape level. In: Van Noordwijk, M., Cadisch, G., Ong, C. (Eds.), Belowground interaction in tropical agroecosystem: concepts and models with multiple plant components. CAB International Wallingford (UK), pp. 349-364.

Joshi, L., Shrestha, P., Moss, C., Sinclair, F., 2004b. Locally derived knowledge of soil fertility and its emerging role in integrated natural resource management. In: Van Noordwijk, M., Cadisch, G., Ong, C. (Eds.), Belowground interaction in tropical agroecosystem: concepts and models with multiple plant components. CAB International Wallingford (UK), pp. 17-39.

Lansing, J.S., 2012. Perfect order: recognizing complexity in Bali. Princeton University Press.

Leimona, B., Joshi, L., Van Noordwijk, M., 2009. Can rewards for environmental services benefit the poor? Lessons from Asia. Int. J. Commons 3, 82-107.

Leimona, B., Van Noordwijk, M., de Groot, R., Leemans, R., 2015. Fairly efficient, efficiently fair: Lessons from designing and testing payment schemes for ecosystem services in Asia. Ecosyst. Serv. 12, 16-28.

Lusiana, B., van Noordwijk, M., Suyamto, D., Mulia, R., Joshi, L., Cadisch, G., 2011. Users' perspectives on validity of a simulation model for natural resource management. Int. J. Agric Sustain. 9, 364-378.

Lusiana, B., Widodo, R., Mulyoutami, E., 2008a. Assessing Hydrological Situation of Kapuas Hulu Basin, Kapuas Hulu Regency, West Kalimantan, Working Paper World Agroforestry Centre, Bogor.

Lusiana, B., Widodo, R., Mulyoutami, E., 2008b. Assessing Hydrological Situation of Talau Watershed, Belu Regency, East Nusa Tenggara, Working Paper. World Agroforestry Centre Bogor.

Lusiana, B., Widodo, R., Mulyoutami, E., Nugroho, D.A., Van Noordwijk, M. Assessing hydrological situation of Kapuas Hulu Basin, Kapuas Hulu Regency, West Kalimantan, Working Paper No. 57 2008c World Agroforestry Centre - ICRAF, SEA Regional Office Bogor, Indonesia.

Lusiana, B., Widodo, R., Mulyoutami, E., Nugroho, D.A., Van Noordwijk, M., 2008d. Assessing the hydrological situation of Talau Watershed, Belu Regency, East Nusa Tenggara, Working Paper No. 58. World Agroforestry Centre - ICRAF, SEA Regional Office Bogor, Indonesia 72.

Maiello, A., Viegas, C.V., Frey, M., Ribeiro, J.L., D., 2013. Public managers as catalysts of knowledge co-production? Investigating knowledge dynamics in local environmental policy. Environ. Sci. Policy 27, 141-150.

Namirembe, S., Leimona, B., Van Noordwijk, M., Bernard, F., Bacwayo, K.E., 2014. Co-investment paradigms as alternatives to payments for tree-based ecosystem services in Africa. Curr. Opin. Environ. Sustain. 6, 89-97.

van der Ent, R.J., Savenije, H.H., Schaefli, B., Steele-Dunne, S.C., 2010. Origin and fate of atmospheric moisture over continents. Water Resources Research 46.

Van Noordwijk, M., 2006. RUPES typology of environmental service worthy of reward, RUPES Working Paper. World Agroforestry Centre, Bogor, p. 53.

Van Noordwijk, M., Bizard, V., Wangpakapattanawong, P., Tata, H.L., Villamor, G.B., Leimona, B., 2014a. Tree cover transitions and food security in Southeast Asia. Global Food Security.

Van Noordwijk, M., Catacutan, D., Clark, W.C., 2009a. Linking Scientific Knowledge with Policy Action in Natural Resource Management, Policy Brief. ASB Partnership—World Agroforestry Centre, Nairobi.

Van Noordwijk, M., Farida, S.P., Agus, F., Hairiah, K., Suprayogo, D., Verbist, B., Garrity, D.P., Okono, A., Grayson, M., Parrott, S., 2006. Watershed functions in productive agricultural landscapes with trees. In: Garrity, D., Okono, A., Grayson, M., Parrott, S. (Eds.), World Agroforestry into the Future. World Agroforestry Centre, Nairobi, Kenya, pp. 103-112.

Van Noordwijk, M., Leimona, B., Emerton, L., Tomich, T.P., Velarde, S.J., Kallesoe, M., Sekher, M., Swallow, B., 2007. Criteria and indicators for environmental service compensation and reward mechanisms: realistic, voluntary, conditional and pro-poor. World Agroforestry Centre, Nairobi.

Van Noordwijk, M., Leimona, B., Xing, M., Tanika, L., Namirembe, S., Suprayogo, D., 2014b. Water-focused landscape management. In: Minang, P.A., Van Noordwijk, M., Freeman, O.E., Mbow, C., Leeuw, Jd, Catacutan, D. (Eds.), Climate-Smart Landscapes: Multifunctionality in Practice. ASB Partnership for The Tropical Forest margins—World Agroforestry Centre, Kenya, p. 179.

Van Noordwijk, M., Leimona, B., Xing, M., Tanika, L., Namirembe, S., Suprayogo, D.,

2015. Water-focused lanscape management. In: Minang, P.A., Van Noordwijk, M., Freeman, O.E., Mbow, C., Leeuw, Jd, Catacutan, D. (Eds.), Climate-smart Landscapes: Multifunctionality in Practice. World Agroforestry Centre, Nairobi, p. 405.

Van Noordwijk, M., Lusiana, B., Leimona, B., Dewi, S., Wulandari, D., 2013. Negotiation-Support Toolkit For Learning Landscapes. World Agroforestry Centre, Bogor.

Van Noordwijk, M., Namirembe, S., Catacutan, D., Williamson, D., Gebrekirstos, A., 2014c. Pricing rainbow, green, blue and grey water: tree cover and geopolitics of climatic teleconnections. Curr. Opin. Environ. Sustain. 6, 41-47.

Van Noordwijk, M., Tomich, T.P., Chandler, F., 2004. An introduction to the conceptual basis of RUPES. World Agroforestry Centre, Bogor, Indonesia.

Van Noordwijk, M., Widodo, R.H., Farida, A., Suyamto, D., Lusiana, B., Tanika, L., Khasanah, N., 2011. GenRiver and FlowPer: Generic River and Flow Persistence Models. User Manual Version 2.0 World Agroforestry Centre (ICRAF) Southeast Asia Programme, Bogor.

Van Noordwijk, M., Wulandari, M., Quan, N.H., Ha, H.M., 2009b. Trees in Multi-Use Landscapes in Southeast Asia (TULSEA). A negotiation support toolbox for Integrated Natural Resource Management (INRM). World Agroforestry Centre (ICRAF) Vietnam, Hanoi, Vietnam.

Pasha, R., Asmawan, T., Leimona, B., Setiawan, E., Irawadi, C., 2012. Commoditized or co-invested environmental services? Rewards for environmental services scheme: River Care program, Way Besai watershed, Lampung, Indonesia Working Paper No. 148. World Agroforestry Centre (ICRAF) Southeast Asia Regional Program Bogor, Indonesia 27.

Petts, J., Brooks, C., 2006. Expert conceptualisations of the role of lay knowledge in environmental decisionmaking: challenges for deliberative democracy. Environ. Plann. A 38, 1045.

Puente-Rodríguez, D., Giebels, D., de Jonge, V.N., 2014. Strengthening coastal zone management in the Wadden Sea by applying 'knowledge-practice interfaces. Ocean Coast. Manag.

Ranieri, S.B.L., Stirzaker, R., Suprayogo, D., Purwanto, E., De Willigen, P., Van

Noordwijk, M., 2004. Managing movements of water, solutes and soil: from plot to landscape scale. In: Van Noordwijk, M., Cadisch, G., Ong, C.K. (Eds.), Below-ground Interactions in Tropical Agroecosystems. CAB International, Wallingford, UK, pp. 329-347.

Reid, W., Berkes, F., Wilbanks, T., Capistrano, D., 2006. Bridging scales and knowledge systems, Millennium Ecosystem Assessment Washington, DC.

Roux, D.J., Rogers, K.H., Biggs, H., Ashton, P.J., Sergeant, A., 2006. Bridging the science-management divide: Moving from unidirectional knowledge transfer to knowledge interfacing and sharing.

Schalenbourg, W., 2004. Farmers' local ecological knowledge of soil and watershed functions in Sumberjaya, Sumatra, Indonesia. Faculteit Landbouwkundige en Toegepaste Biologische Wetenschappen, Katholieke Universiteit Leuven.

Sivakumar, B., 2011. Hydropsychology: the human side of water research. Hydrol. Sci. J. 56; , pp. 719-732.

Sivapalan, M., Savenije, H.H., Bloschl, G., 2012. Socio-hydrology: a new science of people and water. Hydrol. Process. 26, 1270-1276.

Tomich, T.P., Thomas, D.E., Van Noordwijk, M., 2004. Environmental services and land use change in Southeast Asia: from recognition to regulation or reward? Agric. Ecosyst. Environ. 104, 229-244.

Verbist, B., 2008. Assessment of watershed functions to support negotiations in a catchment under land use conflict in Sumberjaya, Indonesia, Department of Earth and environmental Sciences. Katholieke Universiteit Leuven, Belgium, p. 210.

Verbist, B., Van Noordwijk, M., Agus, F., Widianto, HartoWidodo, R., Purnomosidi, P., 2005. Not seeing the trees for the forest? From eviction to negotiation in Sumberjaya, Lampung, Sumatra, Indonesia. ETFRN News: Forests, Water and Livelihods 45/46.

Verbist, B., Poesen, J., Van Noordwijk, M., Suprayogo, D., Agus, F., Deckers, J., 2010. Factors affecting soil loss at plot scale and sediment yield at catchment scale in a tropical volcanic agroforestry landscape. Catena 80, 34-46.

Wagener, T., Sivapalan, M., Troch, P.A., McGlynn, B.L., Harman, C.J., Gupta, H.V., Kumar, P., Rao, P.S.C., Basu, N.B., Wilson, J.S., 2010. The future of hydrology: An evolving science for a changing world. Water Resourc. Res. 46.

Wunder, S., Campbell, B., Frost, P.G., Sayer, J.A., Iwan, R., Wollenberg, L., 2008. When donors get cold feet: the community conservation concession in Setulang (Kalimantan, Indonesia) that never happened. Ecol. Soc 13, 12.