This article appeared in a journal published by Elsevier.

The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elseviers archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy

Journal of Environmental Management 92 (2011) 1938e1949

Contents lists available at ScienceDirect

Journal of Environmental Management

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

Assessing the impact of sheries co-management interventions in developing countries: A meta-analysis

Louisa Evans a, b, Nia Cherrett a, Diemuth Pemsl a, *
a b

The WorldFish Center, P.O. Box 500 GPO, 10670 Penang, Malaysia The ARC Centre of Excellence for Coral Reef Studies, James Cook University, Townsville, Queensland 4811, Australia

a r t i c l e i n f o
Article history: Received 13 July 2010 Received in revised form 24 January 2011 Accepted 8 March 2011 Available online 29 April 2011 Keywords: Small-scale sheries Co-management Governance Environment Livelihoods Impact assessment

a b s t r a c t
Co-management is now established as a mainstream approach to small-scale sheries management across the developing world. A comprehensive review of 204 potential cases reveals a lack of impact assessments of sheries co-management. This study reports on a meta-analysis of the impact of sheries co-management in developing countries in 90 sites across 29 case-studies. The top ve most frequently measured process indicators are participation, in uence, rule compliance, control over resources, and con ict. The top ve most frequently measured outcome indicators are access to resources, resource well-being, shery yield, household well-being, and household income. To deal with the diversity of the 52 indicators measured and the different ways these data are collected and analysed, we apply a coding system to capture change over time. The results of the meta-analysis suggest that, overall sheries comanagement delivers bene ts to end-users through improvements in key process and outcome indicators. However, the dataset as a whole is constituted primarily of data from the Philippines. When we exclude this body of work, few generalisations can be made about the impact of sheries co-management. The lack of comparative data suitable for impact assessment and the dif culties in comparing data and generalising across countries and regions reiterates calls in other elds for more systematic approaches to understanding and evaluating governance frameworks. 2011 Elsevier Ltd. All rights reserved.

1. Introduction Small-scale sheries (SSF) provide essential ecosystem services to large numbers of people in the developing world. The importance of this sector for the local and national economies of many developing countries and the well-being of their people is gaining recognition (see Bn et al., 2007; FAO and The WorldFish Center, 2008; World Bank, 2004; World Resource Institute, 2005). However, sustainable development of SSF remains a considerable challenge, particularly in contexts of poverty, high dependence of increasing populations on ecosystem services, economic globalization and climate change (Andrew et al., 2007). This trend is often considered a crisis of governance (Acheson, 2006; Mahon et al., 2010; Young et al., 2007). Co-management is an alternative approach to conventional management of natural resources, which has found considerable traction within the sheries sector. Co-management is de ned as the sharing of responsibility and authority between the state and

* Corresponding author. Tel.: 1 (206) 388 3643. E-mail addresses: d.pemsl@cgiar.org, dpemsl@gmx.net (D. Pemsl). 0301-4797/$ e see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2011.03.010

resource-users but often involves collaboration between a variety of stakeholders, including different government agencies, nongovernmental organizations, research organizations, private enterprises and civil society more generally (Carlsson and Berkes, 2005). It aims to strengthen the potential of customary management to protect both the local environment and the stakes and rights of resource-users while improving the legitimacy of state involvement in sheries management through more inclusive and transparent decision-making processes. Inclusion of relevant and entitled shery stakeholders can lead to more effective collective action (at appropriate scales), con ict resolution, and higher compliance to management institutions (and therefore reduced costs of enforcement) as well as to integration of more diverse knowledge and value systems on which to base decisions, leading to better problem de nition, social learning, and innovation (Berkes, 2009; Kuperan et al., 2008; Pomeroy et al., 2007). The co-management and participatory management paradigms have been institutionalised through national environmental and development policy and legislation in many countries. Often these efforts are closely associated with other decentralisation processes, for example in the Philippines, Indonesia and Cambodia. In other cases, the co-management approach is simply promoted within

Author's personal copy

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949

1939

particular natural resource management sectors, including wildlife management, sheries, forestry and water. In the SSF sector, co-management arrangements are now the dominant approach to management. Our review identi ed 221 examples of co-management implementation in over 50 countries in the developing world. With approximately 40, 46, and 32 programmes in Africa, Latin America and the Caribbean, and the Paci c, respectively, and over 100 programmes in Asia (Appendix A, Supplementary online material). Many programmes have been funded through relatively long-term (ten years plus) or on-going investments by donor agencies including DANIDA (Denmark), SIDA (Sweden), IDRC (Canada), NZAID (New Zealand), DFID (United Kingdom), and USAID (United States of America) in coastal and/or inshore sheries in developing countries. Recent examples of the level of nancial commitment are the pledges by the EU and Swiss government to invest USD 25.5 million in sustainable coastal sheries and by The World Bank to provide USD 9.5 million in loans and grants to sheries comanagement in Senegal (http://go.worldbank.org/EERDE2GR80). Co-management is also the foundation of other alternative management approaches, including integrated area management, ecosystem-based management, protected area management, and adaptive management. After 20 years of investment in comanagement of SSF what is the impact on resource sustainability, poverty and food security in the developing world? Implementation of co-management is dif cult and measuring impacts is challenging, particularly in terms of poverty reduction and environmental sustainability (Bn and Neiland, 2006). These types of impacts, if achieved, are only realized in the medium to long term. This delay between intervention and outcome and the potential for indirect causal linkages between inputs and outputs complicates evaluation. There is also evidence that the transition to comanagement can result in negative impacts by allowing space for new leaders, interest groups, and processes of power to emerge (Bn et al., 2010; Blaikie, 2006). Whether co-management in practice delivers on its promises as a normative concept is yet to be systematically evaluated. To date, global assessments include efforts to evaluate general sheries management effectiveness (Bn and Neiland, 2003; Mora et al., 2009), compliance to the Code of Conduct for Responsible Fisheries (Pitcher et al., 2009), the potential of catch shares or private property regimes to prevent sheries collapse (Costello et al., 2008), and the performance of marine protected areas (MPAs) as a sheries management tool (Pollnac et al., 2001; Pomeroy et al., 2005a) and their impact on human well-being (Mascia et al., 2010). Assessments relevant to sheries co-management include those that focus on the conditions for successful implementation of sheries co-management, including Napier et al. (2005), Gelcich et al. (2006), Chuenpagdee and Jentoft (2007), and Gutirrez et al. (2011) and more qualitative reviews of impact, including, Sverdrup-Jensen and Raakjr Nielsen (1997) and Wilson et al. (2003, 2006). Only two previous studies quantitatively evaluate the outcomes of sheries co-management arrangements directly. Maliao et al. (2009) build on previous reviews (n 10) of community-based coastal resource management (CBCRM) projects (Pomeroy and Carlos, 1996), in order to assess the performance of a total sample of 16 CBCRM projects in the Philippines through a meta-analysis. Allison and Badjeck (2004) review sheries co-management experiences in tropical inland sheries. Their analysis is based on a sample of 19 case-studies, but primarily discusses conceptual and analytical aspects of co-management. Our review takes these efforts further through the compilation of a more extensive sample of case-studies covering a range of aquatic ecosystems across developing countries and focuses only on the assessment of impact. A number of challenges with respect

to data availability, data quality and variability, and potential publication biases were encountered. Systematic methods are used to overcome these challenges and to present a meta-analysis of SSF co-management assessments across the developing world in order to consolidate the empirical evidence of the impact of SSF co-management. 2. Meta-analysis methodology 2.1. Impact assessment methods and challenges Demonstrating impact has long been recognised as necessary in natural resource management and development research and practice (Baker, 2000; CGIAR, n.d). Typically, monitoring and evaluation describes the process of assessing a project or interventions performance against pre-stated objectives and goals, usually during its implementation. Impact assessments (IA) measure the delivery of bene ts, such as improved resource management, better livelihoods and poverty reduction (as well as possible costs and negative impacts of the intervention). An ex ante IA may be undertaken before the project begins and attempts to quantify the expected consequences of the intervention. An ex post IA evaluates impacts beyond the life cycle of the project (typically 1e5 years after the project has ended), assuming that it takes more time than the project duration for some bene ts (and costs) to materialize. In many cases co-management is introduced as a project of de ned length in which case impact assessment is relatively straightforward. However, in other instances co-management is introduced as a process and institutionalised through policy and legislation, for example, the co-managed Management and Exploitation Areas for Benthic Resources (MEABR) of Chile, which are initiatives that have been ongoing for more than a decade (Gelcich et al., 2010). In these circumstances the timing of an ex post IA can be problematic. Nevertheless, it is important to try and gauge the impact and sustainability of co-management interventions in the long term. Different methodologies are available to do this depending on the type of intervention being assessed (see Walker et al., 2008 for IA guidelines). Yet, assessments of natural resource management initiatives are especially dif cult because of issues of attribution, indirect impact pathways, and the time-lag between intervention and realization of bene ts (Waibel and Zilberman, 2007). Our review of available case-studies in SSF co-management did not identify any examples of rigorous ex post IA. We therefore analyze co-management assessments, reviews or evaluations that consider different impacts (captured by a range of process and outcome indicators) over time. Most of these studies measure a number of indicators at the beginning and towards the end of a project or reconstruct time series encompassing pre, during and post-project status, using perception analysis. 2.2. De ning co-management A de nition of co-management is needed to delineate which case-studies to include in the meta-analysis. For this study, we adapt a de nition of the term, which adequately distinguishes this type of management from customary or traditional community-based management and from state-led management. Co-management is a relationship between a resource-user group and another organization or entity (usually a government agency) for the purposes of sheries management in which some degree of responsibility and/or authority is conferred to both parties. This conceptualization amends the classic de nition proposed by Pomeroy and Berkes (1997) in two ways. First, the traditional de nition speci es that responsibility and authority are shared, and that both conditions are necessary. However, in reviewing the

Author's personal copy

1940

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949

literature, it is often not possible to determine with certitude that this double condition is met. It is only possible to assess whether a potential case describes a situation in which tasks or responsibilities are shared to some extent. The condition that responsibility and authority, not just one or the other, are shared cannot, therefore, be used as a means to exclude case-studies. We have therefore amended the de nition to acknowledge that responsibilities and/or authority are shared. Second, while co-management is traditionally de ned as a relationship between resource-users and the state (with or without the inclusion of other third parties), some co-management projects involve support by independent organizations or non-governmental organizations, in place of the state, particularly in regions such as the Paci c. Indeed, the cases from the Paci c included in this meta-analysis describe community-based management in which responsibilities are shared with non-governmental organizations. Therefore, the second amendment to the classic de nition describes co-management as a relationship between resource-users and another organization to include experiences of co-management with both state and other agencies. 2.3. Data collection A multi-step procedure was used to compile a global list of past and present sheries co-management initiatives implemented in developing countries, along with any corresponding information on evaluations or assessments of the initiatives. The procedure involved two main phases: i) a literature search including a) an electronic search for published and grey literature and b) expert elicitation to identify other potential case-studies and available data and; ii) the systematic selection of case-studies with available and appropriate data for analysis. This systematic method minimises publication bias by including unpublished assessments. 2.3.1. Literature search The rst step in the literature search involved a comprehensive search of various electronic library databases (including Aquatic Sciences and Fisheries Abstracts, CAB Direct, CSA Sociological Abstracts, EconLit and SA, Web of Science, WorldFish Library Catalogue and Scopus) using key words (different combinations of: sher*, communit*, collaborat*, cooperative*, participatory, small scale, coastal, artisan*, stakeholder*, manag*, impact, outcome*, result*, livelihood*, bene t*, socio*, econom*, poverty, food security and income*) and following up references therein. Second, a wider search of grey literature (using search engines on the worldwide web), was conducted to locate other potential casestudies not documented in academic literature. This nal list combining both published and grey literature was then presented to a group of scientists with expertise in sheries management and governance. The consultation provided us with additions to the list of cases and details of key informants at the regional level who could provide further information. This, along with our own research on the co-management community enabled us to contact over 112 key informants including researchers, project managers and sheries experts worldwide. Through the electronic literature search and expert elicitation process (which had a 75 percent response rate) a total of 637 sheries co-management articles were retrieved. From this sample, 433 cases were excluded because they made no reference to qualitative or quantitative impacts of co-management related to sheries or aquatic systems in developing countries. This left 204 articles from which to identify potential case-studies with IA or evaluation data for use in the meta-analysis (Appendix A, Supplementary online material for a global distribution of potential cases).

2.3.2. Case-study selection Case-studies for meta-analysis were selected from the sample of 204 potential case-studies referring to sheries co-management IA or evaluation in developing countries. Decisions of whether a casestudy was relevant to this investigation were based on a review of the abstract, methods and results sections of the associated documents. We performed a careful quality control of the 39 case-studies identi ed. Case-studies were excluded if they demonstrated one or more of the following characteristics: i) reporting of only secondary results; ii) no reference to methodology or basis for ndings and; iii) indications of a awed methodology for collection or analysis of data. The review process resulted in the exclusion of 10 case-studies. The 29 case-studies included in the meta-analysis represent only 14 percent of the original 204 potential cases. The six grounds for exclusion of the other 175 case-studies are presented in Table 1. The majority of cases, n 165, were excluded for lack of data or unavailability of data. For some other cases, the existence of relevant data was unknown and could not be resolved during our expert consultation. Finally, there were a few cases in which the data presented were not directly related to the co-management intervention. For example, there were cases reporting on indicators related to interventions other than co-management, such as the closure of protected areas. A very small percentage of cases were excluded because they did not meet the quality standard. 2.4. Meta-analysis Once we had identi ed all available cases for our review it became evident that, due to the variation in available data (wide range of different indicators and units of measurement) across the 29 case-studies, a number of organizational and standardizing steps were necessary. 2.4.1. Extracting and categorising indicators A total of 52 different indicators were used in the 29 casestudies. A systematic process of grouping overlapping indicators (i.e. pulling together indicators with different names, but measuring the same outcome) reduced the number of indicators to 40. None of the studies included in the review actually measured all these indicators. Case-studies in our sample covered between 1 and 18 indicators. To clarify the themes addressed by these indicators they were grouped into three categories: Natural Systems for ecological factors; People and Livelihoods for social aspects; and Institutions and Governance for indicators re ecting the governance process, policies and institutions involved in co-management arrangements (Evans and Andrews, 2009). These categories re ect the major types of co-management impacts within critical dimensions of a shery system. 2.4.2. Data coding Data were both qualitative and quantitative (Table 2, Section 3.1.1), and were collected through a number of different methods.
Table 1 Reasons for excluding co-management case-studies from the meta-analysis. Reason for exclusion De nition Cases from total sample (percent)

No data No data assessing impacts available 70 No information Existence of data assessing impacts 18 unknown Not relevant Data not related to co-management impacts 8 No access Data assessing impacts exist but 2 access was denied Quality issues Methods used to collect or analyse data did 2 not meet pre-determined quality standards

Author's personal copy

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949 Table 2 Co-management case-studies selected for meta-analysis. Case-study Country B1 B2 B3 B4 B5 C1 C2 C3 CL1 CL2 CL3 I1 K1 M1 PI1 P1 Project and water body or shery and project or implementation period Sites (n) Ecosystem type Data type Year of evaluation 1 6 1a 3 1 1 1 1 18 27 1 1 1 1 1c 6 Inland Inland Inland Inland Inland Inland Inland Inland Marine Marine Marine Marine Marine Marine Marine Marine Quan. Quan. Quan. Quan.d Quan. Qual. Qual. Qual. Quan. Quan. Qual.d Qual.d Quan. Quan. Qual.d Qual.d 2005 2006 1997 2006 2004 2004 2008 2005 2007 2007 2004 References

1941

Bangladesh Community Based Fisheries -2 (CBFM-2): 2001e2006 Bangladesh Community Based Fisheries Management eSouth and South East Asia (CBFM-SSEA): 2002e2007 Bangladesh Oxbow Lakes Small Scale Fishermen Project: 1991e1997 Bangladesh Management of Aquatic Ecosystems through Community Husbandry (MACH): 1998-2006 Bangladesh Mymensingh Aquaculture Extension Project Cambodia Tblong Kla Community Fishery: 1998 - ongoing Cambodia Tonle Sap Environmental Management Project (TSEMP): 2005e2008 Cambodia Policy Reform Impact Assessment: 2000 - ongoing Chile Chile Chile Indonesia Kenya Malaysia Paci c Islands Philippines Benthic Resource Management Areas e Region III (AMERB e III): 1992 - ongoing Benthic Resource Management Areas e Region IV (AMERB e IV): 1992 - ongoing Benthic Resource Management Areas e Region IV (AMERB e VI): 2002 - ongoing Sasi Fishery in Central Malukab DianieChale Management Area: 1995-ongoing Integrated Coastal Resource Management - Pulau Langkawi (ICRM ePL): 2003e2007 A Comparative Study of Coastal Resource management in the Paci c Islands Community Based Coastal Resource Management Central Visayas Regional Project-1 (CVRP-1): 1984e1996 Coastal Environment Programme (CEP): 1994e1996 Marine Conservation Project for San Salvador Island: 1989e1996 Sagay Marine Reserve, Negros Occidental: 1995e2002 Honda Bay Resource Management Project (HBRMP): 1989e1996 Cogtong Bay Mangrove Project: 1989e2004 Fisheries Sector Programme (FSP): 1990e2002 Sustainable Rural District Development Programme (SRDDP): 1994e1998 Community Fisheries Resource Management Project (CFRMP): 1991e1996 Local Government Support Programme: 1996e2002 Bais Bay Coastal Resource Management Projects: 1984-2005 Anilao Area Coastal Resource Management Projects: 1990-2005 Tanga Coastal Zone Conservation and Development Programme (TCZCDP) : 1994 - ongoing Integrated Coastal Resource Management - Pathew District (ICRM-PD): 2001e2006

Halls and Mustafa, 2006 The WorldFish Center, 2007 IFAD, 1998 Halder and Thompson, 2006 Winrock International, 2004 Viner et al., 2006 Wildblood, 2008 Cambodian Dept. of Fisheries & IMM Ltd., 2006 Stotz, 2008 Stotz, 2008 Gelcich et al., 2006

1997 Novaczek et al., 2001 1995e2007 McClanahan, 2010; McClanahan et al., 2008 2007 bin Saleh, 2008 1998e1999 World Bank, 2000 1996 Pomeroy et al., 1996

P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 TZ1 T1

a

Philippines Philippines Philippines Philippines Philippines Philippines Philippines Philippines Philippines Philippines Philippines Tanzania Thailand

2 1 1 1 1 1 1 1 1 1 1 6 1

Marine Marine Marine Marine Marine Marine Marine Marine Marine Marine Marine Marine Marine

Qual.d Both.d Qual.d Qual.d Qual.d Qual.d Qual.d Qual.d Qual.d Qual.d Qual.d Quan. Both

1996 1996 2002 1996 2004 2002 1998 1996

Pomeroy et al., 1996 Katon et al., 1999 Webb et al., 2004 Pomeroy et al., 1996 Maliao and Polohan, 2008 Israel et al., 2004 van Mulekom and Tria, 1999

Baticados and Agbayani, 2000 2002 Baylon, 2002 2005 Pomeroy et al., 2005b 2005 Pomeroy et al., 2005b 1998-2004 Wells et al., 2007 2006 Boromthanarat, 2007

A T-test was performed on the raw data (household income gures pre and post-project) provided in project reports to generate a p-value. The Sasi Fishery is the only case where a co-management customary sheries management arrangement exists. One data point was generated from an analysis of 31 co-management case-studies from the Paci c Islands of Fiji (6), Tonga (6), Samoa (6), Solomon Islands (7) and Pulau (6). No project implementation dates available. d Data with signi cance values (p-value).
b c

Although most of the data re ect change over time, they were measured, analysed and presented in different forms (e.g. ranked score, percentage, monetary value, number of species). The diversity in how data were collected, analyzed, recorded, and documented, alongside the relatively small sample of case-studies, rendered a statistical meta-analysis of co-management impact across countries and regions impossible (see Maliao et al., 2009). A coding system was, therefore, devised and used to combine the different types of data. We rst examined the trend of change. A positive trend over time was assigned a positive code (1), where there was no change, data were coded as 0, and for negative trends a negative code (1) is assigned. In addition, signi cance values were noted. If the casestudy results were signi cant (p < 0.05), the code was modi ed to 2 if the trend was positive and 2 is the trend was negative. Where the signi cance of the trend is less than 95 percent signi cant (p > 0.05) or where there is no signi cance value attributed to the

data, the coding remained as 1 or 1. Indicators are de ned in such a way that an increase (code 1 or 2) always equals a (positive) bene t i.e. an improvement of the situation while a decrease in any indicator (1 or 2) highlights a deterioration of the situation (for example a decrease in con ict is a positive impact and coded as 1 or 2 depending on the level of signi cance). 3. Results 3.1. Composition of data 3.1.1. Case-study characteristics Fig. 1 shows the regional distribution of the full set of potential co-management case-studies identi ed from the literature search and expert consultation (n 204). The largest percentage of casestudies are from Asia (47 percent), followed by Latin America and the Caribbean (21 percent), Africa (18 percent) and the Paci c

Author's personal copy

1942

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949

Pacific 14% Africa 18% Latin America & Caribbean 21%

Asia 47%

22% 7% 4% 4% 3% 2% 2% 3%

Philippines Bangladesh Sri Lanka Cambodia Viet Nam Malaysia Lao Others

Fig. 1. Total number of case-studies by region, with country breakdown for Asia.

region (14 percent). The country breakdown for Asia shows that the majority of case-studies within this region are from the Philippines with the remaining coming from Bangladesh, Sri Lanka, Cambodia, Vietnam, Malaysia and Lao PDR. The other category groups three cases from Indonesia and Thailand and one case-study each from China, India, and Myanmar (Appendix A, Supplementary online material). An overview of the 29 case-studies selected for analysis is presented in Table 2. The majority of these case-studies are located in Asia (79 percent), with the largest proportion situated in the Philippines (41 percent of all cases). These alone represent 52 percent of the Asian case-studies, with the remaining 48 percent located in Bangladesh (16 percent of total cases), Cambodia (10 percent), Indonesia, Malaysia and Thailand (4 percent each). The Latin America and the Caribbean region is represented by three case-studies (10 percent), Africa by two case-studies (7 percent) and the Paci c by only one case-study (4 percent). Within our sample seven case-studies (B2, B4, CL1, CL2, P1, P2 and TZ1) present data from more than one site (Table 3). Where data are available for separate sites they were counted individually, resulting in the inclusion of 90 sites from 29 case-studies. The sites include marine and inland aquatic habitats (83 and 17 percent of cases, respectively) ranging from coral reefs and mangrove forests to seasonal wetlands and lakes. 19 case-studies have IA data in qualitative form while ten report quantitative results. All qualitative data were generated through stakeholder perception analyses resulting in time series and treatment-control comparison data (13 case-studies followed the sheries co-management research framework methodology developed by the World Wide Collaborative Research Project on Fisheries Co-Management outlined in Pomeroy et al., 1996). The quantitative results were mostly time series and report on indicators such as household income, sh catch and biodiversity. The projects were implemented, evaluated and funded by a range of different government agencies, nongovernmental organisations, and (international) donors. There are also differences in how case-studies de ne and present treatment and control data. Technically, for a meaningful counterfactual, data from a treatment and a control group are compared (Walker et al., 2008). The terminology used in most

co-management evaluations is member (treatment) and nonmember (control). Members are individuals or households who were engaged in the shery co-management initiative, while the non-member population did not participate in these initiatives. However, in four case-studies (P7, P9, P11 and P12) member and non-member data were combined and only one set of results was reported. We did not use these data when comparing indicators, leaving a total set of 25 viable case-studies and 86 sites. In six casestudies (P1, P2, P3, P5, P8 and P10) the non-member data did not re ect an actual control group as the non-member population was located in the same area accessing the resource that the project was targeting. Therefore, non-members may have (indirectly) bene ted from the co-management intervention, thus a comparison of members and non-members would not reveal the impact of the intervention. We were left with too small a sample (n 7) of cases with separate member and non-member data to conduct any statistically meaningful comparison of the two groups. Thus, for the 25 case-studies used for meta-analysis only trends over time for member data are analyzed. 3.1.2. Impact assessment indicators Table 4 shows the 40 different indicators extracted from the 29 case-studies and the split of indicators across the categories chosen as a framework for this study: Natural Systems (8 indicators), Governance and Institutions, and People and Livelihoods (17 and 15, respectively). We further differentiate between process and outcome indicators. Process indicators re ect progress in one of a series of actions (e.g. collaboration, learning, communication) that are considered important for the legitimate and successful implementation of co-management. Outcome indicators correspond to goals, such as improved resource abundance, ecosystem health, livelihoods and other facets of human well-being, which re ect the overall objective of SSF co-management to achieve sustainable development. The ve most frequently utilised outcome indicators are household income (used 32 times), shery yield (30), resource well-being (19), household well-being (13) and resource access (12). These outcome indicators are distributed between the Natural System and the People and Livelihoods domains of the shery. The indicators shown below are not de ned by the authors but represent the terminology used in the studies from which they are extracted (for de nitions of these indicators see Appendix B in the Supplementary online material). In some cases the terminology used is technically incorrect, for example resource well-being (referring to resource health or status), but is nevertheless continued here to maintain clear links to the data source from which it is derived. The ve most frequently measured process indicators are participation (used 19 times), rule compliance (16), in uence (15), resource control (14) and con ict (13). All these fall within the Governance and Institutions domain of the shery system. Indeed, almost all indicators in this category are process indicators (15 out of 17). The relatively more frequent use of process indicators to measure SSF co-management impact points to the emphasis in theory and practice on co-management as a process of governance

Table 3 Summary of data used in the meta-analysis. Number of included case-studies ALL Asia Africa Paci c Latin America 29 23 2 1 3 Member only data (per case-study) 17 6 7 1 3 Member and non-member data (per case-study) 7 6 1 / / Combined data only (per case-study) 5 5 / / / Total number of case-study sites 90 36 7 1 46

Author's personal copy

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949 Table 4 Type and frequency of indicators used in the case-studies to assess co-management projects (n 40). Process indicators (frequency) Natural systems Outcome indicators (frequency) Fishery yield or resource harvest (30) Resource well-being (19) Coral cover (8) Fish density (7) Biodiversity trend or species richness (6) Fish diversity (3) Threat to resource (3) Bird diversity (1) Household income (32) Household well-being (13) Fish consumption (3) Food security (2)

1943

(Carlsson and Berkes, 2005). However, empowerment and improved mechanisms of collective action, while arguably important in their own right, do not guarantee other SSF management objectives of importance in developing countries, including improved resource status, better livelihoods and poverty reduction. In order to build a body of work that provides a comprehensive evaluation of the impact of co-management, it is necessary to include metrics of both process and outcome indicators. 3.2. Meta-analysis of data The results for the ve most frequently measured outcome and process indicators are presented in Figs. 2 and 3. While the trend in the process indicators is positive overall (Fig. 3), this is not the case for the outcome indicators (Fig. 2), which show more variability in the direction of change over time. Within the outcome indicators, only resource access has an overall negative trend (58 percent), with 33 percent signi cant negative change over time (Fig. 2). Common property regimes, including co-management, are often associated with changes in property rights and institutions, or rules (Ostrom, 1990). Access rights are typically designated to members of co-management institutions, such as sher associations, who then manage the institutions, or rules, that govern resource access. Interestingly, the results show an overall decline in resource access for members. As elaborated above (Section 3.1.1), we do not compare data of trends for members and nonmembers but analyze member only data over time. Declining access for members suggests that stricter access rules have been imposed. Declining access can be a positive or negative outcome for the shery overall, depending on the associated impacts on people and the shery resource. Resource access must, therefore, be analysed in the context of trends in shery sustainability and human well-being, both of which we address below. Fishery yield shows an increasing trend over time (77 percent of the total), although the signi cant trends are split fairly evenly between increases (10 percent) and decreases (7 percent). The resource well-being indicator shows more of a signi cant positive trend than a negative trend across cases (47 percent and 16 percent, respectively). The yield indicator is constructed from quantitative and qualitative data while the resource well-being indicator consists of perception analysis (qualitative) data. We discuss the

People and livelihoods

Satisfaction with reserve & sanctuary management (6) Resource knowledge (5) Communication (5) Community harmony (3) Marine reserve bene ts (2) Satisfaction with mangrove management (2) Community development (1) Economic equality (1) Self-esteem (1) Fair distribution of shing gears (1) Role in sheries management (1) Institutions and Participation (19) Access to resource (12) governance Rule compliance (16) Fisher violations (1) In uence (15) Resource control (14) Con ict (13) Destructive shing practices (9) Con ict management (6) Fair allocation of access rights (5) Collective decision-making (4) Fishere sher cooperation (3) Quality of resource management (2) Government- sher cooperation (2) Law enforcement (1) Distribution of government resources (1) Tradition of collective action (1)

Fig. 2. Trends in the ve most frequently measured outcome indicators.

Author's personal copy

1944

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949

Fig. 3. Trends in the ve most frequently measured process indicators.

implications of this below. Nevertheless, in combination, these indicators suggest tentative improvements to resource status as a result of co-management interventions. However, it is important to note, that the determinants of yield are often varied and can include environmental factors and other interventions, such as stock enhancement, associated with co-management, but not indicative of improved management. For instance, in case-studies B1, B2, B3 and B5, co-management arrangements were associated with enhanced sheries programmes involving the release of ngerlings into the water bodies. Such interventions can increase sh yields substantially without necessarily re ecting sustainable resource use. The indicators household well-being and income have a higher frequency of signi cant positive trends than negative change with 62 and 25 percent of the indicator trends being signi cantly positive, respectively. The household well-being indicator is a subjective indicator constructed from perception data. It is likely that positive trends in process indicators (Fig. 3) contribute as much to an improved perception of household well-being as for example, income. In general, this re ects well on co-management and suggests a level of buy-in to the approach (that does not appear to be undermined overall by declines in resource access). Testimony from those involved in co-management projects has often suggested that, to them, increased participation, inclusion, voice and empowerment are the most important contributions of the co-management idea (Jentoft, 2005). Household income also shows a positive trend, which is rather surprising considering the relatively mixed data on shery yield. However, it is likely that some of this income is derived from non- sheries sources either encouraged and facilitated within co-management projects (e.g., microcredit ventures implemented in case-study B1) or increasingly, from other global, regional or local trends unrelated to the co-management interventions, for instance remittances sent from household members who have migrated to urban areas to nd work (e.g. highlighted in case-studies C3, P1, P2 and P5). Relatively more positive trends in both resource and human well-being indicators suggests that the negative trend in resource access can be viewed overall as a neutral or positive outcome for comanagement members. The signi cant positive trends for the process indicators - ranging from 46 percent for con ict to 73 percent for in uence - imply that, generally, the governance and community empowerment goals of co-management are realised (Fig. 3). None of the case-studies

reported a negative trend for participation, with only 5 percent of the total indicator frequency indicating no change. Rule compliance is the only process indicator for which some case-studies found signi cant negative trends (6 percent). It is dif cult to draw out any trend in the data from an ecosystem or regional perspective. In Africa, Latin America and the Paci c the ndings are mixed with no real trend in either direction (Table 5). Few studies note signi cant increase or decrease in indicators, with the exception of the Gelcich et al. (2006) Chilean case-study which observes signi cant declines in three indicators following the implementation of the MEABR co-management system in a context of effective customary management. With so few case-studies in these regions it is not possible to generalize ndings. However, there is a positive trend across the Asian studies (55 and 16 percent of all data for marine and inland, respectively) with high signi cance levels predominantly in the marine cases (38 percent compared to 3 percent of data in inland cases). Rather than re ecting any regional or ecosystem dependent relationship these outcomes are likely to be associated with trends in natural resource management assessment (see sources of bias in Section 4) or with the dominance of perception data on process indicators in the dataset, particularly in the Philippines cases. To investigate how robust our ndings are, we have conducted two further analyses. We analyzed the top ten indicators separately for data derived from perception analysis (qualitative data, n 19) and from other methods (non-perception data, n 12). Since some case-studies use both data types, the combined sample size is larger

Table 5 Impact of co-management (frequency) by ecosystem and region. Positive Ecosystem Signi cant Not signi cant 11 48 0 2 39 100 36 0 3 1 0 7 11 4 No Change Negative Not Signi cant signi cant 12 15 3 2 6 38 14 0 14 3 0 0 17 6

Marine Africa (n 2) 0 Asia (n 15) 106 Latin America (n 3) 0 Paci c (n 1) 0 Inland Asia (n 8) 7 Total (Frequency) Total (Percentage) 113 40

Author's personal copy

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949

1945

than the number of case-studies (Fig. 4). In the perception analysis method, introduced by Pomeroy et al. (1996), interview respondents are asked to assign a score to each indicator for the past (10 or 15 years prior to evaluation), present (evaluation date) and future (10 or 15 years after evaluation). The signi cance values created from the difference between the past and present scores were used in the analysis. Second, in order to evaluate whether the dominance of cases from the Philippines (n 12) in uences ndings, we also analyze data for the top ten indicators excluding the Philippine case-studies (Fig. 5). Fig. 4 shows that household income increases over time, but signi cant changes are mainly found in data derived from perception analysis. The majority of non-perception data that report increased household income but do not nd signi cant change come from Bangladesh (n 4). Similar ndings for household income from perception and non-perception data suggest that ndings for this indicator are robust. This is further strengthened by the analysis excluding Philippines cases (Fig. 5): the household income indicator is used across a range of cases and shows overall positive change with a small portion being signi cant. However, as mentioned earlier, several case-studies, particularly those in Bangladesh, note that increases in incomes cannot always be attributed to co-management. Higher income can emerge from complementary project activities, such as introducing microcredits or providing skills training related to alternative income earning opportunities (unrelated to shing) or from completely independent trends such as migration and remittances. Stakeholders perceptions of income change are unlikely to distinguish

the determining drivers of change, and therefore, in the absence of comparative member and non-member data, even signi cant income trends, cannot be solely attributed to the successful implementation of co-management. The dominating positive trend in household well-being is found only in perception data with trends that mirror perception data on household income. It is not possible to validate this data by excluding the Philippines cases as the remaining sample, despite a positive trend, is too small to draw de nite conclusions. Fish yield data are split fairly evenly between perception and non-perception data. Non-perception sh yield trends show an overall increase over time (50 percent of the total), though these are not signi cant. Perception sh yield data also show an overall increase over time (27 percent) but reveal both signi cant positive (ten percent) and negative trends (seven percent). The majority of data points showing non-signi cant increases in yield over time come from Bangladesh. Again, changes in resource yield are not always directly related to improved resource management, but could be caused by other (external) factors. The resource well-being indicator is derived primarily from perception data and reports positive and negative signi cant change over time. Resource access is generated solely from perception data, but as for resource wellbeing, the fact that respondents willingly report on negative trends suggests that data re ect members actual views and dont seem skewed towards the positive because respondents feel they need to give positive feedback. We nd signi cant positive trends for process indicators, mainly derived from perception analysis. However, if we exclude the

Fig. 4. Comparison of indicator trends generated from perception (P) analysis and non-perception (NP) analysis methods.

Author's personal copy

1946

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949

Fig. 5. Data for the top outcome and process indicators excluding the Philippines cases.

Philippines cases, we are left with very few data points, from which to draw inferences. Indeed, what these analyses have shown is that in the absence of Philippine data the sample is too small to draw any robust conclusions or generalisations. Only the results for sheries yield and household income are spatially robust as they are supported by a diverse set of case-studies. However, without a control group, it is not possible to attribute trends in these indicators directly or solely to co-management arrangements. Finally, we examine all indicator trends (Table 4) categorised into Natural System, People and Livelihoods and Institutions and Governance (Fig. 6) with the Philippines dataset removed. Overall, there is very little global data showing signi cant trends in either a positive or negative direction. Within the aggregated Natural System category, we nd mixed outcomes. Under People and Livelihoods there is an overall positive trend, with a comparatively

larger proportion of signi cant positive than negative trends. For Institutions and Governance indicators we nd an overall positive trend of the impact of sheries co-management. However, only a fraction of these positive results is statistically signi cant while case-studies report a larger number of signi cant negative trends. 4. Discussion A recent paper on the evolution of co-management, reviews the many dimensions of this paradigm (Berkes, 2009). These include comanagement as a mechanism for power sharing, institution building, enhanced trust and social capital, problem solving and, more recently, knowledge-sharing and social learning. Co-management is seen by many as a normative process to improve the legitimacy and effectiveness of resource management. Our analysis draws

Fig. 6. Results for grouped categories of co-management success indicators (excluding Philippines cases).

Author's personal copy

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949

1947

out the process indicators that are emphasised and most commonly evaluated in practice. The most frequently used are participation, in uence, rule compliance, resource control and con ict. These do not directly address issues of power-sharing or trust-building, for example, but as a whole suggest improved inclusion of stakeholders in governance processes, improved capacity to control or in uence decision-making, and improved compliance to management rules over time. All of these process indicators exhibited a positive trend, a result supported by Mascia et al., 2010 in their investigation of MPA impacts on shing communities, Maliao et al. (2009) in their metaanalysis of Philippine cases, and Pomeroy et al. (2007) in their study of con ict and collaboration in SSF arrangements in Southeast Asia. Positive trends in process indicators are expected to lead to improved management outcomes, although literature shows that this is not guaranteed (Bn and Neiland, 2004, 2006). The key outcome indicators assessed in this meta-analysis include household income, household well-being, resource well-being, shery yield, and resource access. Overall, trends were proportionally more positive than negative over time, however, results are more varied than for process indicators and we found negative trends in a number of case-studies. This reiterates the importance of considering both process and outcome indicators to meaningfully evaluate and improve the impact of co-management as an approach to SSF management in developing countries. Separately analyzing perception and non-perception data to test the rigor of these ndings showed that results are relatively robust. Where both types of data are present within one indicator, the trends are generally similar, although non-perception data is less likely to nd or report signi cant change. Further, the perception data reports both negative and positive trends, which suggests that the data relatively accurately re ect opinion. However, when excluding the data from the Philippines we are left with virtually no data on process indicators, so we cannot generalize ndings. The only two indicators that are well represented across our global dataset are household income and shery yield, although very little of these data show signi cant change. Further, both indicators are likely in uenced by external factors such as micro-credit schemes, sheries enhancement, remittances or other income sources unrelated to sheries or general changes in the economic or ecological environment. Without a control group (non-member) comparison, the observed changes cannot be attributed to co-management with any level of certainty. Improved yield and household income are key desirable outcomes for SSF co-management in developing country contexts and we, therefore advocate for the continued emphasis on these two outcome indicators. Nevertheless, more widespread assessment of the performance of co-management, as re ected in process indicators is clearly necessary to examine whether the empowerment and governance aims of co-management are effective in a broader context. In conducting this meta-analysis we faced a number of challenges related to the availability of data. The systematic methodology applied aimed to ensure that data were representative and from reliable sources. A number of biases, common in assessments of natural resource management interventions (see Meinzen-Dick, 2007 for a discussion of bias in the agricultural water-use sector), however, remain. First, interventions are generally implemented in suitable or favourable places; this may explain the dominance of case-studies from the Philippines in our global sample of sheries co-management assessments in the developing world. The Philippines are renowned for community-based MPA management and a national commitment to co-management of natural resources more, including relevant institutional reform. Second, more successful projects are more likely to be evaluated. In particular, projects that fail and are discontinued are rarely documented in published or grey literature. Third, assessments reporting positive

results are more likely to be published (we tried to minimise this bias by including grey literature and unpublished cases). These sources of bias may be exacerbated as interventions are frequently evaluated by the very organisations (and sometimes the same individuals) that initiate and co-ordinate the intervention. Further, many of the casestudies included in our meta-analysis document trends within a project life cycle. The results are likely to look different if these projects are evaluated in a real ex post sense i.e. a few years after the project activities have ended. Indeed, Pomeroy and Carlos (1997) report that 81 percent of CBCRM programs in the Philippines are unsustainable in that they fail once external support is withdrawn. 5. Conclusion Assessments of natural resource management interventions are notoriously dif cult as re ected by the scarcity of impact assessment data on sheries co-management in developing countries. These challenges are not unique to sheries management but are common in other sectors. Poteete and Ostrom (2008) discuss the dif culties in collecting conclusive evidence on natural resource management interventions and in trading-off between context rich case-studies and large-N studies that evaluate a smaller set of variables. Most existing assessments present a diverse array of indicators and data types that are dif cult to compare across cases, countries and regions. To counter this, there is need for: i) comprehensive analytical frameworks, ii) meta-databases to document qualitative data and data derived from small samples of case-study sites and, iii) large-N eld-based studies (Meinzen-Dick, 2007; Ostrom, 2007; Poteete and Ostrom, 2008). An exploratory network for impact assessment of governance interventions in the forestry sector e the International Forestry Resource and Institutions (IFRI) network e does, however, demonstrate that substantial nancial, human and institutional investment is needed to conduct rigorous assessments of natural resource management interventions (Wollenberg et al., 2007). Yet, investments in management processes including monitoring and evaluation and/or ex post impact assessment should be commensurate with the value of the system being managed (Garcia et al., 2008). For SSF in developing country contexts this is problematic in that the economic value of these sheries is often unknown and other bene ts are intangible (ibid). Nevertheless, the nancial investment in sheries co-management as the primary management approach in SSF continues to be substantial, thereby justifying a greater investment in comprehensive and independent impact assessment. Conducting rigorous IA studies will help to better channel scarce (research) resources and promote approaches which have been shown to be successful. Acknowledgements A special thanks to William Koh for his tireless assistance in tracking down key references for this review and to the members of the sheries co-management community who responded to our requests for information. For helpful comments and suggestions on the methodology and earlier drafts of the manuscript we thank Edward Allison, Christophe Bn, Robert Pomeroy, Blake Ratner and Nireka Weeratunge. We are grateful for the constructive comments and suggestions of four anonymous reviewers. This work was carried out with funding provided by the WorldFish Center. Appendix. Supplementary material Supplementary data associated with the article can be found in online version, at doi:10.1016/j.jenvman.2011.03.010.

Author's personal copy

1948

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949 IFAD, 1998. Oxbow lakes small scale shermen project (Loan: 237-BD). Project completion & impact assessment, IFAD. Israel, D., Adan, E., Lopez, N., de Castro, J., 2004. Household perceptions of the longterm impact of coastal resources management in Panguil Bay. Philippines Institute for Development Studies, Philippines. Jentoft, S., 2005. Fisheries co-management as empowerment. Marine Policy 29 (1), 1e7. Katon, B., Pomeroy, R., Garces, L., Salamanca, A., 1999. Fisheries management of San Salvador Island, Philippines: A shared responsibility. Society and Natural Resources 12, 777e795. Kuperan, K., Abdullah, N.M.R., Pomeroy, R.S., Genio, E.L., Salamanca, A.M., 2008. Measuring transaction costs of sheries co-management. Coastal Management 36 (3), 225e240. Mahon, R., Fanning, L., McConney, P., Pollnac, R., 2010. Governance characteristics of large marine ecosystems. Marine Policy 34, 919e927. Maliao, R., Polohan, B., 2008. Evaluating the impacts of mangrove rehabilitation in Cogtong Bay, Philippines. Environmental management 41 (3), 414e424. Maliao, R., Pomeroy, R., Turingan, R., 2009. Performance of community-based coastal resource management (CBCRM) programs in the Philippines: a metaanalysis. Marine Policy 33 (5), 818e825. Mascia, M., Claus, C., Naidoo, R., 2010. Impacts of marine protected areas on shing communities. Conservation Biology 24 (5), 1424e1429. McClanahan, T., 2010. Effects of sheries closures and gear restrictions on shing income in a kenyan coral reef. Conservation Biology 24 (6), 1519e1528. McClanahan, T., Hicks, C., Darling, E., 2008. Malthusian over shing and efforts to overcome it on Kenyan coral reefs. Ecological Applications 18 (6), 1516e1529. Meinzen-Dick, R., 2007. Beyond panaceas in water institutions. Proceedings of the National Academy of Sciences of the United States of America 104 (39), 15200e15205. Mora, C., Myers, R.A., Coll, M., Libraloto, S., Pitcher, T.J., Sumaila, R.U., Zellers, D., Watson, R., Gaston, K.J., Worm, B., 2009. Management effectiveness of the worlds marine sheries. PLoS Biology 7 (6), e1000131. Napier, V.R., Branch, G.M., Harris, J.M., 2005. Evaluating conditions for successful comanagement of subsistence sheries in KwaZulu-Natal, South Africa. Environmental Conservation 32 (2), 165e177. Novaczek, I., Harkes, I., Sopacua, J., Tatuhey, M., 2001. An Institutional Analysis of Sasi Laut in Maluku. ICLARM/The WorldFish Center, Indonesia. Penang, Malaysia. Ostrom, E., 1990. Governing the Commons: The Evolution of Institutions for Collective Action. CUP. Ostrom, E., 2007. A diagnostic approach for going beyond panaceas. Proceedings of the National Academy of Sciences 104 (39), 15181e15187. Pitcher, T., Kalikoski, D., Pramod, G., Short, K., 2009. Not honouring the code. Nature 457, 658e659. Pollnac, R.B., Crawford, B.R., Gorospe, M.L.G., 2001. Discovering factors that in uence the success of community-based marine protected areas in the Visayas, Philippines. Ocean & Coastal Management 44, 683e710. Pomeroy, R., Berkes, F., 1997. Two to tango: the role of government in sheries comanagement. Marine Policy 21 (5), 465e480. Pomeroy, R., Carlos, M., 1996. A Review and Evaluation of Community-Based Coastal Resources Management Projects in the Philippines, 1984e1994. ICLARM/The WorldFish Center, Makati City, Philippines. Pomeroy, R., Carlos, M., 1997. Community-based coastal resource management in the Philippines: a review and evaluation of programs and projects, 1984e1994. Marine Policy 21 (5), 445e464. Pomeroy, R., Pollnac, R., Predo, C., Katon, B., 1996. Impact Evaluation of CommunityBased Coastal Resource Management Projects in the Philippines. ICLARM/The WorldFish Center & North Sea Center, Makati City, Philippines. Pomeroy, R., Watson, L., Parks, J., Cid, G., 2005a. How is your MPA doing? A methodology for evaluating the management effectiveness of marine protected areas. Ocean & Coastal Management 48 (7e8), 485e502. Pomeroy, R., Oracion, E., Pollnac, R., Caballes, D., 2005b. Perceived economic factors in uencing the sustainability of integrated coastal management projects in the Philippines. Ocean and Coastal Management 48 (3e6), 360e377. Pomeroy, R., Parks, J., Pollnac, R., Campson, T., Genio, E., Marlessy, C., Holle, E., Pido, M., Nissapa, A., Boromthanarat, S., Hue, N.T., 2007. Fish wars: con ict and collaboration in sheries management in Southeast Asia. Marine Policy 31 (6), 645e656. Poteete, A., Ostrom, E., 2008. Fifteen years of empirical research on collective action in natural resource management: struggling to build large-N databases based on qualitative research. World Development 36 (1), 176e195. Stotz, W., 2008. Evaluacion del proceso de implementacion de la medida de administracion areas de manejo y explotacion de recursos bentonicos (AMERB) en las regiones III y IV y elaboracion de una propuesta de mejoramiento de la medida. Universidad Catolica del Norte. Sverdrup-Jensen, S., Raakjr Nielsen, J., 1997. Co-Management in Small-Scale Fisheries: a Synthesis of Southern and West African Experiences. Institute for Fisheries Management and Coastal Community Development (IFM). Available online at URL. http://academic.research.microsoft.com/Paper/5010491.aspx. The WorldFish Center, 2007. Project Completion Report: Community Based Fisheries Management in South and South East Asia (CBFM-SSEA) Project. The WorldFish Center. van Mulekom, L., Tria, E., 1999. Community-based Coastal Resource Management in Orion, Bataan, Philippines: a Case Study on the Development of a Municipal Wide Community-Based Fisheries Co-Management System. SNV-Philippines & PRRM-Bataan, Manila, Philippines.

References
Acheson, J.M., 2006. Institutional failure in resource management. Annual Review of Anthropology 35, 117e134. Allison, E., Badjeck, M., 2004. Fisheries Co-management in Inland Waters: a Review of International Experience. Sustainable Fisheries Livelihood Programme. FAO & DFID. Andrew, N.L., Bn, C., Hall, S.J., Allison, E.H., Heck, S., Ratner, B.D., 2007. Diagnosis and management of small-scale sheries in developing countries. Fish and Fisheries 8, 227e240. Baker, J., 2000. Evaluating the Impact of Development Projects on Poverty: a Handbook for Practitioners. The World Bank, Washington D.C., USA. Baticados, D., Agbayani, R., 2000. Co-management in marine sheries in Malalison Island, Central Philippines. International Journal of Sustainable Development and World Ecology 7, 343e355. Baylon, C., 2002. Evaluation of the integrated municipal council as an institution for co-management in the coastal zone. WorldFish Center, Penang, Malaysia. Bn, C., Neiland, A., 2003. A Review of Fisheries Management Performance in Developing Countries with Particular Reference to Issues of Policy and Governance. Support Unit for International Fisheries & Aquatic Research (SIFAR) Department of Fisheries Food and Agriculture Organization. Bn, C., Neiland, A., 2004. Empowerment reform, yes. but empowerment of whom? Fisheries decentralization reforms in developing countries: a critical assessment with speci c reference to poverty reduction. Aquatic Resources, Culture and Development 1 (1), 35e49. Bn, C., Neiland, A., 2006. From Participation to Governance: a Critical Review of Governance, Co-Management and Participation in Natural Resources Management, with Particular Reference to Inland Fisheries in Developing Countries, Penang, Malaysia and Colombo, Sri Lanka. World Fish Center and CGIAR Challenge Program on Water and Food. Bn, C., Macfadyen, G., Allison, E., 2007. Increasing the Contribution of Small-Scale Fisheries to Poverty Alleviation and Food Security. FAO, Rome, Italy. Bn, C., Belal, E., Ousman Baba, M., Ovie, S., Raji, A., Malasha, I., Njaya, F., Na Andi, M., Russel, A., Neiland, A., 2010. Struggle, dispute and alliance over the new power: analyzing democratic decentralization of natural resources through the lenses of Africa inland sheries. World Development 37 (12), 1935e1950. Berkes, F., 2009. Evolution of co-management: role of knowledge generation, bridging organizations and social learning. Journal of Environmental Management 90 (5), 1692e1702. Blaikie, P., 2006. Is small really beautiful? Community-based natural resource management in Malawi and Botswana. World Development 34 (11), 1942e1957. Boromthanarat, S., 2007. Final project evaluation: Integrated coastal resource management in Pathew District (ICRM-PD), Songkhla, Thailand: Coastal Resources Institute (CORIN). Prince of Songkla University, Thailand. bin Saleh, I., 2008. Final Project Evaluation: Integrated Coastal Resource Management - Pulau Langkawi (ICRM-PL). Malaysia, Kuala Lumpur. Cambodian Dept. of Fisheries and IMM Ltd., 2006. Policy reform impact assessment: Impacts of the sheries policy reforms in Kampong Thom & Prey Veng Provinces. Cambodia Community Fisheries Development Of ce of the Department of Fisheries of the Royal Government of Cambodia, Cambodia. Carlsson, L., Berkes, F., 2005. Co-management: concepts and methodological implications. Journal of Environmental Management 75 (1), 65e76. CGIAR, n.d. The Gateway to Evidence that International Agricultural Research is Making a Difference. CGIAR Impact. Available at: http://impact.cgiar.org/ (Accessed 02.02.10.). Chuenpagdee, R., Jentoft, Svein, 2007. Step zero for sheries co-management: what precedes implementation. Marine Policy 31, 657e668. Costello, C., Gaines, S., Lynham, J., 2008. Can catch shares prevent sheries collapse? Science 321 (5896), 1678e1681. Evans, L., Andrews, N., 2009. Diagnosis and the Management of Constituency of Small-scale Fisheries. The World Fish Center, Penang, Malaysia. FAO and The WorldFish Center, 2008. Small-scale Capture Fisheries e a Global Overview with Emphasis on Developing Countries. A Preliminary Report of the Big Numbers Project (Sponsored by PROFISH World Bank). Garcia, S., Allison, E.H., Andrew, N.L., Bene, C., Bianchi, G., de Graaf, G., Kalikoski, G.J., Mahon, R., Orensanz, J.M., 2008. Towards Integrated Assessment and Advice in Small-Scale Fisheries: Principles and Processes. FAO, Rome, Italy. Gelcich, S., Edwards-Jones, G., Kaiser, M.J., Castilla, J.C., 2006. Co-management policy can reduce resilience in traditionally managed marine ecosystems. Ecosystems 9 (6), 951e966. Gelcich, S., Hughes, T.P., Olsson, P., Folke, C., Defeo, O., Fernandeza, M., Foale, S., Gunderson, L.H., Rodrigues-Sickerty, C., Scheffer, M., Steneck, R.S., Castilla, J.C., 2010. Navigating Transformations in Governance of Chilean Marine Coastal Resources. Gutirrez, N.L., Hilborn, R., Defeo, O., 2011. Leadership, social capital and incentives promote successful sheries. Nature 470 (7334), 386e389 [Epub 2011 Jan 5]. Halder, S., Thompson, P., 2006. Restoring Wetlands Through Improved Governance: Community Based Co-management in Bangladesh. The Management of Aquatic Ecosystems through Community Husbandry (MACH) Experience. Winrock International, Bangladesh Center for Advance Studies, Center for Natural Resource Studies, CARITAS Bangladesh, Dhaka, Bangladesh. Halls, A., Mustafa, M., 2006. Final assessment of the impact of the CBFM project on community-managed sheries in Bangladesh, Report to the WorldFish Center, Bangladesh.

Author's personal copy

L. Evans et al. / Journal of Environmental Management 92 (2011) 1938e1949 Viner, K., Ahmed, M., Bjorndal, T., Lorenzen, K., 2006. Development of Fisheries Co-Management in Cambodia: a Case Study and Its Implications. WorldFish Center. Waibel, H., Zilberman, D., 2007. International Research on Natural Resources Management: Advances in Impact Assessment. CAB International. Walker, T., Maredia, M., Kelley, T., La Rovere, R., Templeton, D., Thiele, G., Douthwaite, B., 2008. Strategic Guidance for Ex Post Impact Assessment of Agricultural Research. Science Council Secretariat, Rome, Italy. Webb, E., Maliao, R., Siar, S., 2004. Using local user perceptions to evaluate outcomes of protected area management in the Sagay Marine Reserve, Philippines. Environmental Conservation 31 (2), 138e148. Wells, S., Makoloweka, S., Samoilys, M., 2007. Putting Adaptive Management into Practice: Collaborative Coastal Management in Tanga, Northern Tanzania. IUCN and Irish Aid, Nairobi, Kenya. Wildblood, R., 2008. Tonle Sap Environmental Management Project (TSEMP) Component 2: Evaluating the Impact of TSEMP on Livelihoods. Conservation & Community Fisheries Development. Wilson, D.C., Raakjr Nielsen, J., Degnbol, P. (Eds.), 2003. The Fisheries Comanagement Experience: Accomplishments, Challenges and Prospects. Kluwer Academic Publishers, Dordrecht, The Netherlands.

1949

Wilson, D.C., Ahmed, M., Siarb, S.V., Kanagaratnam, U., 2006. Cross-scale linkages and adaptive management: sheries co-management in Asia. Marine Policy 30, 523e533. Winrock International, 2004. Mymensingh Aquaculture Extension Project: 1989e2003. Winrock International & DANIDA, Bangladesh. Wollenberg, E., Merino, L., Agrawal, A., Ostrom, E., 2007. Fourteen years of monitoring community-managed forests: learning from IFRIs experience. International Forestry Review 9 (2), 670e684. World Resource Institute, 2005. World Resources 2005: The Wealth of the Poor Managing Ecosystems to Fight Poverty. WRI, Washington D.C. World Bank, 2000. Voices from the Village - A Comparative Study of Coastal Resource Management in the Paci c Islands (Vol. 2 of 2), East Asia and Paci c Region. Papua New Guinea and Paci c Islands Country Management Unit: World Bank. World Bank, 2004. Saving Fish and Fishers: Toward Sustainable and Equitable Governance of the Global Fishing Sector. World Bank, Agriculture and Rural Development Department, Washington D.C. Young, O.R., Osherenko, G., Ogden, J., Crowder, L.B., Ogden, J., Wilson, J.A., Day, J.C., Douvere, F., Ehler, C.N., McLeod, K.L., Halperin, B.S., Peach, R., 2007. Solving the crisis in ocean governance: place-based management of marine ecosystems. Environment 49 (4), 20e32.