**Contents**


### **About the Special Issue Editors**

**Bas Arts** (1961) is a Personal Professor of Forest Governance for the Forest and Nature Conservation Policy (FNP) group at Wageningen University, the Netherlands. Between January 2006 and July 2018, he was the Chair of the FNP. Arts holds a master's degree in biology (majoring in forest ecology and environmental sciences, 1987) and a PhD in policy sciences and public administration (1998). Currently, Arts' research and teaching focus on international forest, biodiversity and climate change governance, and local natural resource management and their interconnections (local/global nexus and multi-level governance). Arts was a Visiting Professor at the Europe College in Bruges, Belgium (2006–2010), at BOKU University of Life Sciences in Vienna, Austria (March–April 2010), at the Geography Department of Cambridge University, UK (October 2013), and at the School of Natural Resources and the Environment of the University of Michigan, USA (March–April 2014). He was also a Visiting Fellow at the Max Planck Institute on Common Goods in Bonn, Germany (2002). He has (co)produced over 150 academic publications, including journal articles, books and book chapters.

**Verina Ingram** is a social scientist focusing on the interactions between people, natural resources and markets. She has 20 years of experience working with governments, NGOs, businesses, researchers and communities in Africa, Europe and Asia. Verina currently works on tropical commodities, including cocoa, timber, and non-timber products, evaluating impacts on sustainability and livelihoods, and the telecoupled links between how landscapes and value chains are governed. Verina has worked with Wageningen University and Research since 2012, and currently also works with the Forest and Nature Conservation Policy (FNP) group and Wageningen Economic Research. She led the Central Africa Livelihoods Team at the Centre for International Forestry Research (CIFOR) from 2008 to 2010 and worked with SNV Cameroon on market access for the poor and natural resource management from 2004 to 2008. Her PhD from the University of Amsterdam concerned the impacts of governance arrangements on the sustainability of forest product value chains in the Congo Basin.

**Maria Brockhaus** is Professor and Chair of International Forest Policy at the University of Helsinki, and a member of Helsus, the University's new sustainability center. Her main research themes are forest policy and governance in climate change mitigation and adaptation, the political economy of tropical deforestation, policy and institutional change and policy network analysis. Before joining the University of Helsinki in 2016, she was a Senior Scientist at the Center for International Forestry Research (CIFOR), and had led the policy component in CIFOR's Global Comparative Study (GCS) on REDD+ since 2009. Understanding what enables (and hinders) transformational change in and beyond the forestry sector fascinates her—especially when dissecting politics and power in highly contested policy arenas. Maria has published extensively, and established and led large, impact-oriented research teams in the Global South over the past decades.

### *Editorial* **The Performance of REDD**+**: From Global Governance to Local Practices**

#### **Bas Arts 1,\*, Verina Ingram <sup>1</sup> and Maria Brockhaus 2,3**


Received: 29 July 2019; Accepted: 17 September 2019; Published: 24 September 2019

#### **1. Introducing REDD**+ **and the Need to Address Performance**

Whilst 'REDD' is the acronym for reducing emissions from deforestation and forest degradation, 'REDD+' refers to efforts to reduce emissions from deforestation and forest degradation, foster conservation, promote the sustainable management of forests, and enhance forest carbon stocks [1]. The basic idea behind REDD+ is that more carbon is sequestrated and stocked in tropical forests by improving their conservation, management, and sustainable use, thus contributing to mitigating climate change. Developing countries and relevant stakeholders will be financially compensated for these endeavors, either through public funds or carbon markets.

First known as "Avoided Deforestation" (AD) and discussed as a mitigation option at the United Nations Framework Convention on Climate Change (UNFCCC) in the early 2000s, AD subsequently became RED, REDD, and REDD+, with the concept expanding to incorporate forest conservation, management, and use [2]. Whereas RED was 'just' an innovative proposal tabled by Costa Rica and Papua New Guinea in 2005, REDD activities were outlined in the Bali Action Plan resulting from the 13th Conference of Parties to the UNFCCC (COP 13) in 2007, to become a legally-binding article in the 2015 Paris Agreement. In between these events, international bodies such as the World Bank, United Nations Development Program (UNDP), and United Nations Environment Program UNEP, and developed countries (notably Norway, Germany, the UK, Australia, and the US), started REDD+ programs and funds, and developing countries entered into so-called "readiness activities" to prepare for participation in REDD+. Currently, hundreds of REDD+ projects are being implemented around the world [3], whilst REDD+ has changed considerably over time, being previously described as "the world's largest experiment in Payments for Ecosystem Services" [4], and more recently, as results-based aid [5]. The performance of REDD+ has been largely measured in terms of emission reductions calculated on the basis of forest reference (emission) levels (National or subnational reference levels expressed as tonnes of CO2 equivalent per year for a reference period against which emissions and removals from a results period are compared, serving as benchmarks for assessing the national performance in terms of implementing REDD+ activities). This reduction of expected results is then assumed to have been compensated for by results-based finance (RBF), as was the case for Brazil, through payments from the Green Climate Fund in early 2019. The Bali Action Plan requires REDD+ projects to measure changes in net carbon emissions resulting from project activities. In 2013, the COP19 adopted the Warsaw Framework for REDD+, where results-based payments can be made once safeguards have been addressed and respected. The Measurement, Reporting, and Verification (MRV) of RBF were elaborated at COP24 in Katowice, Poland, December 2018, with MRV capacities being reported as increasing [6].

However, many observers, stakeholders, and scholars are critical. Carbon markets have not developed as was envisioned in 2005, and the efficiency and effectiveness of REDD+ to mitigate climate change have been questioned, with projects' impacts being unclear. As Angelsen et al. [7] state, "As an idea, REDD+ is a success story (...).Yet a decade after being launched in the Bali Action Plan (UNFCCC 2007), broad consensus is that—in practice—REDD+ has not met the world's high expectations". While some studies have reported positive impacts of REDD+ in terms of forest carbon [8–13], the expected transformational change away from business as usual of deforestation and forest degradation has not taken place thus far. Forest cover loss and tropical deforestation mainly driven by agricultural and forestry commodities trade continue [14–17]. The results of REDD+ in terms of sharing the burdens, costs, and benefits have been questioned [18]. The progress of policy reforms seen as underpinning REDD+, such as tenure reform, has also been questioned [19]. For some, the lack of performance is a result of major flaws in the design of the instrument itself (as a market-based payment-for-ecosystem-services scheme (PES) and has led to the claim that "REDD+ is dead" [20]. Others argue that rather than blaming the instrument itself, there is a need to consider the wider environment in which REDD+ was supposed to perform and deformations of the initial PES idea this environment created, with powerful actors interested in maintaining the status quo, leading to inefficiency [21]. However, as Angelsen et al. [7,21] suggest, REDD+, as both an innovative governance mechanism and a political process, will finally perform, as climate change becomes more pressing, carbon markets improve, and technical issues related to the efficiency and effectiveness of REDD+ are solved. Nonetheless, as Wong et al. [22] pointed out, a performance or results-based payment approach is no guarantee for an effective and transparent REDD+. Their analysis of narratives underlying REDD+ benefit sharing mechanisms highlights the importance of linking payments for performance to the contexts in which the results are defined and agreed upon, along with the conditions enabling social and political acceptance [22].

Given this context, this special issue is timely in addressing the need to assess the political and socio-economic dimensions of the performance of REDD+, which are of relevance for policy-makers, practitioners, and scholars. This implies taking into account the various levels (from the global to the local) and dimensions (e.g., results-based payments, MRV, co-benefits, and community engagement), as well as different (disciplinary) connotations of performance. We therefore pose the following question: what does performance mean? In answering it, we provide examples of assessments of performance. In our conclusion, we reflect on the representativeness of these examples and their limitations when looking at the current range of REDD initiatives, along with what is missing in terms of evaluating the performance of REDD+. We finish by concluding why performance assessment remains so relevant today.

#### **2. 'Performance' Means Di**ff**erent Things**

'Performance' refers to the act or process of executing a task or function. Hence, the concept of performance moves beyond an account of results and related payments, and includes the process leading to the result, even though performance and results are often used interchangeably. In this Special Issue, performance means whether a public policy, program, or project 'works'; delivers its promises; or achieves its previously set objectives [23]. As a wider concept, performance can be understood and evaluated differently [24]. As the following subsections—and the individual papers in this special issue—show in greater detail, it can refer to (i) the economic (cost)effectiveness of policies; (ii) multi-criteria achievements; or (iii) evaluation models based on the notions of output, outcome, and impact. Alternative ways of evaluating performance, presented in the last two subsections below, assess how (iv) impacts are 'performed' or 'staged' by stakeholders, independently of what's happening on the ground, or how (v) certain policy discourses 'produce' certain implementation and evaluation practices, and not others, which is referred to as 'performativity' [25].

#### *2.1. Policy E*ff*ectiveness*

Economics has dominated policy evaluation, where performance is equated with (cost) effectiveness, efficiency, relevance, and sustainability based on the objectives of a policy [26–28]. In the REDD+ literature, the distinction between cost-efficiency and outcome effectiveness is prominent (see, for example, [1,29,30]). Two evaluation methods are commonly used: ex-post, often qualitative or mixed methods assessments of the degree to which policy objectives are attained within the policy's timeframe, and ex-ante, often the quantitative, statistical analysis of future costs and benefits of policy measures, measured against current and future values, based on discount rates [31].

For REDD+, this implies comparing result-based payments from carbon funds and markets for forest managers at the end of the policy's time frame, compared to the costs incurred, by using a cost–benefit analysis of REDD+ initiatives. Correa et al. [32] detail the allocation of financial resources from the USD 667 million Brazilian Amazon fund across different stakeholders for scientific and technological development, sustainable production, monitoring and control, and land tenure regularization. However, in terms of the forest conservation effectiveness, as well as emissions reductions from deforestation, they found that deforestation rates had risen since 2013, despite increased fund disbursements. Samndong and Vatn [33] took a qualitative approach to show the lack of effectiveness and legitimacy of REDD+ projects in the Équateur province in the DR Congo. Umunay et al. [34] hypothesized ex-ante the run of three policy mechanisms to reduce commodity-driven deforestation. They examined 19 cases where REDD+ programs, jurisdictional approaches (JAs), and private sector commitments intersect, evaluating potential options against established criteria. They found that most were located in countries with high deforestation rates—attracting REDD+ program, JA, and private sector commitment activities. These policies alone did not appear effective in countering tree cover loss; however, when operating together, these efforts were seen to have potential to reduce commodity-driven deforestation, enhancing and complementing each other. Overman et al.'s [35] economic analysis of the impact of national REDD+ programs in Guyana suggests that indigenous communities with legal forest tenure benefited financially in terms of cash income. However, there were modest benefits from commercial forest uses, and extremely skewed private-public sharing of net revenue from forest-based resources and inhibitive forest damage costs at rising carbon prices. Moreover, carbon stocks were substantially lower in locally-managed forests.

#### *2.2. Multi-Criteria Policy Achievements*

Alongside the economic view on policy performance are juridical, administrative, and political perspectives, with scholars and practitioners increasingly using a combination of criteria and approaches from different traditions. An example is the JEP Triangle, which uses Juridical, Economic and Political indicators for policy performance [36]. Besides efficiency, criteria such as legitimacy, justice, legality, democracy, and participation, are also considered important. Ex-post and ex-ante evaluations and ex-nunc (often legal-based) approaches may be used. The latter evaluate policy processes "from now on" rather than (projected) impacts.

For REDD+, this, for example, implies the assessment of community participation in projects, not only to evaluate the degree of democracy in project implementation, but also the effectiveness of community achievements towards the 'triple-win' of climate change mitigation, biodiversity conservation, and community development. Millbank et al.'s [37] review of 25 sub-national REDD+ projects across the world that adhere to Climate, Community, and Biodiversity Alliance (CCB) REDD+ certification standards embraced a much broader set of objectives than just REDD+. However, a gap was found, with few projects actively monitoring the impact against these goals and progress, and only a third of the sustainable development goals targeted by REDD+ projects showing 'improvement'. Samndong and Vatn [33] examined the DR Congo REDD+ program in terms of its effectiveness in protecting forests for carbon storage and alleviating rural poverty, and its appropriateness as a governance tool. They found that the lack of harmonization between REDD+ as an instrument

promoted by the international community created a competition between state and customary authorities. This allowed powerful actors to 'shop' between systems and legitimize their increased use and control of forest resources. Weatherley-Singh and Gupta [38] discuss the implications of European Union policies on "embodied deforestation" targeting EU agricultural commodity imports as drivers of deforestation. They find that despite substantial debate, policy measures for reducing the impact of the EU's consumption of agricultural commodities associated with multiple drivers of tropical deforestation have not been developed. However, they see potential for a more integrated EU policy approach to tackling tropical deforestation.

#### *2.3. Policy Outputs, Outcomes, and Impacts*

Policy performance is also equated with 'impact' [39]. The concept of impact arguably derives from Easton's [40] political system approach, with the output-outcome-impact model common in impact evaluation and impact assessment widely used in organization studies, conflict studies, international relations, and development studies [41]. The model is based on the notion that policies lead to interventions (on a macro, meso, or micro level) that aim to have a positive societal impact. Interventions result in *outputs,* tangible actions, reports, products, and/or commitments expressed in projects, programs, law, funds, etc. Outputs ideally result in behavioural changes, in outcomes, manifested at a (usually short-term) temporal scale and at different spatial scales, such as on a national macro-economic, ecosystem or landscape, sector, firm, community, household, and individual level. Outcomes can be influenced by contextual factors, such as cultural, political, and the business environment factors. The ultimate effects of policies and outcomes are impacts: the actual contributions to intended problem solving and opportunity seeking. Often, evidence on the counterfactual (what would have happened without the intervention) is gathered to provide a comparison in evaluations. A theory of change is used to articulate assumptions about the process through which such impacts will occur and accompanying assumptions envisaged in an impact pathway, often with verifiable and measurable indicators of the output, outcome, and impact.

For REDD+, this implies the analysis of outputs (such as a single REDD+ project), outcomes (for example, behavioral changes of forest managers due to REDD+ initiatives), and/or impacts (such as an increase of carbon sequestration and carbon stocks in a forest) in a specific landscape. Correa et al. [32] attribute the apparent lack of effectiveness to the distribution mechanism of the Brazilian Amazon fund that prioritized diverse organizations rather than a strategic selection of projects due to its predetermined theory of change. Rosa da Conceição et al. [42] looked at REDD+ in Ecuador and Peru and the pathways of two government-led, incentive-based forest conservation and poverty reduction programs for forest-based populations. They found that political interests affected policy design, resulting in trade-offs for longer-term societal efficiency in favor of short-term administrative goals. Non-environmental outcomes were often prioritized, due to perceptions of political feasibility, the influence of non-environmental government agencies, and beliefs in specific government roles or public responses. Overman et al.'s [35] evaluation of the performance of REDD+ in Guyana map out an economic foundation for a national REDD+ low-emissions impact pathway towards changing forest governance. They suggest that REDD+ can provide the incentives needed for governments to counter the drivers that threaten forest-dependent people because concessions, or land grabs allowed by the government (as often observed in other countries), are costly. In this specific context, maintaining the forest, and hence its management by forest-dependent people, results in low emissions and incomes from REDD+, in contrast to higher emissions due to conversion or forest concessions.

#### *2.4. The Way to Success is 'Performed' by Stakeholders*

One of the main challenges in the evaluation of programs and projects relates to the credibility of what is reported, what was measured and how, and the overall policy relevance of the evaluation itself. Relevance here refers to the relevance of the overall goals of the policies vis a vis the key problems [28], and hence focuses on the assessment of policy documents and matches between the policy problem at

hand and the goals stated. The policy relevance of an evaluation itself can also refer to the relevance of the measures selected to assess impacts, asking if what is assessed is actually of policy relevance and useful for identifying appropriate policies and measures to tackle the overall problem. While methods used to assess and monitor impacts can increase rigor and robustness [26,43], some of the factors that can undermine the credibility and legitimacy in REDD+ assessments are the often cited issues of poor documentation, costs, time, and technical capacities [44,45]. Beyond these, credibility and legitimacy are jeopardized by a strong inclination to frame results as overly positive, whatever the 'real' effects on the ground [46]. Therefore, there are (implicit) biases in any evaluation method to produce positive results, which affects their credibility. In addition, many stakeholders engaged in evaluations may be biased: evaluators paid to assess projects; project participants hoping for an extension and new resources; politicians with political motives and careers; and scientists with a bias for publication impact. In other words, many interests converge towards performing or staging projects as 'successes'.

The papers in this issue highlight that REDD+ evaluations indeed encounter risks of positive framing in practice. This appears as not necessarily deliberate, but emerges from the interests and sympathies of those involved. For REDD+, this implies that a critical assessment is needed of how and why project evaluations are biased towards success stories, or how some stakeholders tell positive stories to evaluators, while the situation on the ground is rather different. Another emerging aspect is how local actors express agency in (re)framing and changing the narratives of performance according to their own interests, beyond those paying for a "good performance". These experiences highlight the importance of both local context and culture, as well as subnational, national, and international interests, in understanding performance narratives. Correa et al. [32] recount the significant investments in the Brazilian Amazon fund, widely seen as successful. However, when dissected, the funds' performance reflects the arbitrary support of different projects adopting highly different theories of change, many of which are not chiefly interested in reducing deforestation, but based on the stakeholder's own preferences and activities. In the DR Congo, a study on the performance of the REDD+ program [33] highlights the exclusion of some timber and charcoal actors, which has detrimental effects on the effectiveness and legitimacy of REDD+, but which is not reported back to the program [47]. Additionally, Millbank et al. [18] find a marked gap in 25 subnational, CCB-certified REDD+ projects between aspirations on paper and the monitoring of progress.

#### *2.5. The 'Performativity' of Policy Discourse*

Where classical evaluation studies examine how policy performs after implementation, they often take policy discourses for granted and use external yardsticks (such as effectiveness, efficiency, and/or legitimacy) to assess their performance. This approach is problematized by 'performativity' scholars [25,48], who argue that performance is not only a function of the (un)intended effects of implementing policy, but also of the policy discourse itself. Discourses 'make' certain objects and subjects, whilst (un)intentionally excluding others, and so constitute certain implementation and evaluation practices. This has implications for the representation and inclusiveness of such policies.

For REDD+, this implies the analysis of how certain types of stakeholders are pre-defined in REDD+ project narratives, while others are implicitly or explicitly excluded. It could also imply the problematization of the use of certain benchmarks to assess the success and failure of REDD+, because these produce their own assessment realities, while excluding others, for example, those of local practitioners. Skutsch and Turnhout [49] illustrate this performative aspect of REDD+ policies, which frame communities as both beneficiaries and implementation agents. They note that international policies generally do not clarify who or what communities actually are. Their analysis of international and national policy documents on REDD+ worldwide demonstrates the unreality and disconnection between scales and the heterogeneity of communities targeted in national REDD+ policies. They close on a note of warning (p. 13), expressing that the attachment within policy discourses to the "community myth" could catalyze positive change, but "may also cause blindness in terms of the practical implications for communities, that will ultimately do them a disservice". At the same time, communities

are not passive, docile receivers of REDD+. Den Besten and colleagues [50] portray the on-the ground practices between global policymaking and local implementation in Ghana. Whilst global actors led the implementation of REDD+ in a cocoa production landscape in the high forest area, they depended on local actors to make REDD+ work. Consequently, it was integrated into existing community-based conservation, forest restoration, and agro-forestry practices, thus transforming REDD+ to resemble these local practices.

#### **3. Conclusions**

This special issue aims to take stock of the current state of REDD+ performances. The papers reflect the diverse understandings of what constitutes performance and performativity. They also show a broad range of methods used to assess performance, from qualitative in-depth studies to quantified, statistically rigorous approaches, with most papers reflecting on a relatively small number of comparative cases; thus, the representativeness of these examples is limited. Systematic reviews and studies assessing large numbers of cases and employing a counterfactual approach are notably absent among the papers in this issue and can be seen as missing in the suite of approaches which can be used to evaluate performance. Taking all papers together, an overall conclusion about the performance of REDD+ remains challenging, with most papers providing a rather bleak outlook, and some being somewhat more optimistic in their assessment. REDD+ exhibits potential to bring about change: it can contribute to achieving international objectives and targets (Paris Agreement, SDGs); bring substantial income to communities who sustainably manage their forests against the trend of forest conversion; it can reduce deforestation once various approaches are smartly combined (private sector and juridical approach); and it appears to re-energize 'old-fashioned' forest management approaches (such as community forestry and forest restoration). At the same time, the papers also show that the rhetoric is stronger than the evidence of practices on the ground; that short-term, administrative interests overshadow long-term environmental ones; that REDD+ rules adversely interact with state and customary institutions; and that REDD+ lacks local legitimacy by excluding non-elites.

These different, partly contradicting conclusions can, in our view, be particularly attributed to the nature of this special issue: namely, that we invited scholars applying different interpretations and methods of performance assessment to contribute. Some of approaches are more critical than others towards policy evaluations in general and REDD+ evaluations in particular, for example, the 'staging performance' and 'performativity' approaches. It is thus no wonder that these come up with less favorable assessments of REDD+ performance; this can be attributed to the nature of these approaches. What we also see, however, is that mainstream performance evaluation methodologies also result in different outcomes; for example, economic versus public administration evaluations, due to data and assessment choices, as well as due to the different cases evaluated. Moreover, the papers highlight that the subject of performance itself is understudied: it is a difficult beast to measure, and as a research topic, is subject to selective narratives, values, and interests. The case studies in these papers also reflect the challenges of understanding, defining, and measuring performance, so the publication serves wider societal interests rather than selected ones.

Finally, REDD+ is still in an early phase of implementation, and one cannot expect its performance—the delivering of its perceived potential—to have fully materialized yet. Experiences of policy evaluations also highlight that we need to be realistic in our expectation that policies have not only positive, but also negative, direct and indirect, and intended and unintended, effects. In addition, we can neither expect that all policy evaluation traditions and approaches produce similar outcomes, nor that their conclusions will 'automatically' converge over time; that would be naïve. However, time is pressing, as impacts of REDD+ occur, whilst payments for results are being requested and issued. Therefore, we strongly encourage further REDD+ performance assessments and call for innovative yet rigorous analyses, and for comparative evaluations using different approaches to judge performance, in order to help develop our understanding of REDD+ performance.

**Conflicts of Interest:** The authors declare that they have no conflicts of interests.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Amazon Fund 10 Years Later: Lessons from the World's Largest REDD+ Program**

**Juliano Correa 1,\*, Richard van der Hoff 2,3 and Raoni Rajão <sup>3</sup>**


Received: 27 December 2018; Accepted: 26 February 2019; Published: 19 March 2019

**Abstract:** Results-Based Funding (RBF) for Reducing Emissions from Deforestation and Forest Degradation (REDD+) has become an important instrument for channeling financial resources to forest conservation activities. At the same time, much literature on conservation funding is ambiguous about the effectiveness of existing RBF schemes. Many effectiveness evaluations follow a simplified version of the principal-agent model, but in practice, the relation between aid providers and funding recipients is much more complex. As a consequence, intermediary steps of conservation funding are often not accounted for in effectiveness studies. This research paper aims to provide a nuanced understanding of conservation funding by analyzing the allocation of financial resources for one of the largest RBF schemes for REDD+ in the world: the Brazilian Amazon Fund. As part of this analysis, this study has built a dataset of information, with unprecedented detail, on Amazon Fund projects, in order to accurately reconstruct the allocation of financial resources across different stakeholders (i.e., governments, NGOs, research institutions), geographies, and activities. The results show that that the distribution of resources of the Amazon Fund lack a clear strategy that could maximize the results of the fund in terms of deforestation reduction. First, there are evidences that in some cases governmental organizations lack financial additionality for their projects, which renders the growing share of funding to this type of stakeholder particularly worrisome. Second, the Amazon Fund allocations did also not systematically have privileged the municipalities that showed the recent highest deforestation rates. rom the 10 municipalities with the higher deforestation rates in 2017, only 2 are amongst the top 100 receiving per/Ha considering the 775 municipalities from Legal Amazon. Third, the allocation of the financial resources from the Amazon Fund reflects the support of different projects that adopt significantly diverging theories of change, many of which are not primarily concerned with attaining further deforestation reductions. These results reflect the current approach adopted by the Amazon Fund, that do not actively seek areas for intervention, but instead wait for project submissions from proponents. As a consequence, project owners exert much influence on to the type of activities that they support how deforestation reduction is expected to be attained. The article concludes that the Amazon Fund as well as other RBF programs, should evolve over time in order to develop a more targeted funding strategy to maximize the long-term impact in reducing emissions from deforestation.

**Keywords:** REDD+; Amazon Fund; Results-Based Funding; benefit distribution; resource allocation; climate change funding; effectiveness; forest conservation funding

#### **1. Introduction**

The international allocation of funds to activities intended to funding forest conservation—directly or indirectly—is said to be a "highly cost-effective way of reducing greenhouse gas emissions on climate change" [1]. Among many types of financial mechanisms for pursuing this approach, Results-Based Funding (RBF) for Reducing Emissions from Deforestation and Forest Degradation (REDD, or REDD+ for a broader suite of activities) has become an important instrument for channeling financial resources to forest conservation activities [2,3]. RBF can be defined as the "transfer of money or material goods conditional upon taking a measurable action or achieving a predetermined performance target" [4–7]. The success of RBF instruments for REDD+ stems from political controversies related to initial REDD+ proposals that favored offset-based markets [8]. Brazilian government, in particular, has been known to challenge the use of markets on the basis of sovereignty concerns [6,9]. Instead, Brazil created the Amazon Fund in 2008 in order to receive results-based payments for achievements in deforestation reductions [10], which plummeted between 2004 and 2012 [11–13]. Similar developments have also occurred in international forest governance debates, as the Green Climate Fund became the central financial instrument for REDD+ [14], testifing the growing prevalence of RBF approaches in forest governance. Despite this dominance, the effectiveness of RBF has been challenged by scholars [5,7,15–18], while others have showed that donor and receiving countries and stakeholders often disagree on how to best evaluate these schemes and distribute the resources [19,20].

This research paper aims to enhance the understanding of intermediary stages of RBF for forest conservation by reconstructing the allocation of financial resources from the Brazilian Amazon Fund to individual projects and analyzing the underlying rationales behind this allocation. Between 2008 and 2017, the Amazon Fund has received more than USD 1.2 billion in donations, committed USD 667.3 million for the financial support of 96 approved projects, and thereby represents the largest and most longstanding RBF initiatives in forest governance worldwide [10,13,21]. An analysis of financial resource allocation could, therefore, provide important lessons on the intermediary stages of RBF (as Amazon Fund) to REDD+ and other conservation purposes. Our analysis exposes the underlying intervention logic (or 'theory of change') adopted for redistributing financial resources, which is useful for identifying the main factors for successful or failing forest conservation funding. The remainder of this paper proceeds as follows. Section 2 reviews the literature on related resource allocations, including the theories of change, criteria for resource allocation, benefit-sharing mechanisms, and impacts. Section 3 then outlines our approach and Section 4 presents data about the distribution of Amazon Fund resources. Section 5 concludes with our main findings and their implications for impact and policy making.

#### **2. Aid Effectiveness and the Complex Relations between Service Providers and Service Users**

Deforestation reduction [17,22] is a relatively recent trend in the broader context of development aid that has usually targeted health, education, or biodiversity conservation [16,23]. Although using the same model, the literature generally refers to aid as funding for REDD+ initiatives, since the former seems to be charity while the last is close to the climate change concepts, where developed countries should fund initiatives of forest conservation to offset their historical emissions [2].

Although this aid could come in many forms, RBF has become an increasingly appealing approach due to its simplicity from both the donor and receiver sides. On the donor side, the payments are done based on the measurement of a result already achieved, reducing substantially the transactional risk. On the receiver side, RBF promises the transfer of resources with "no strings attached" as countries are able to decide how best to invest the payments. Since receiving countries would want to receive an increasing volume of resources, they would be incentivized to invest the RBF proceedings in a way that reduces deforestation the most. A closer look, however, reveals that many of the issues that have plagued REDD+ and development aid more in general are still present in RBF, namely: benefit distribution, intervention design, and effectiveness.

One of the key design choices around REDD+ programs concerns the definition of "who needs to be involved, whose interests are at stake, and the expected co-benefits and required safeguards" [19]. Moreover, their discussion of approaches to reducing tropical forest degradation highlights the importance of contextualizing local realities, responding to new knowledge and experience, and incorporating the full complexity of forest loss and degradation, among others [24,25]. Many scholars have highlighted the issues of equitable sharing of net benefits from REDD+ projects (e.g., [26,27]). For instance, Luttrell, Loft, Fernanda Gebara, Kweka, Brockhaus, Angelsen, and Sunderlin [27] distinguish a number of possible rationales for the distribution of REDD+ benefits. They have emphasized: (1) actors with legal rights; (2) actors achieving reductions in emissions; (3) low-emitting forest stewards; (4) actors incurring the costs of REDD+ implementation; (5) effective facilitators of REDD+ implementation; and (6) the poorest actors. They note great variation in how implementing countries apply these rationales, implying that this is a function of context, project design and the beneficiaries (see also [8]). Some scholars find that "equity can have significant positive feedback on program outcomes and legitimacy over the longer term" [26,28,29]. According to Vatn and Vedeld [30], market-based approaches were found to be the most problematic among governance structures, since they do not address equity. These observations suggest a theme of providing equal opportunities to stakeholders. Yet rigorous analysis, and even comprehensive evaluations of net benefits and their distribution, are scarce, in part because of the way decisions are made about distributions of resources within and across REDD+ projects [19].

Another key aspect of RBF is the choice, by the receiving country, of the interventions that will be supported by the program. [27,31]. Weatherley-Singh and Gupta [32], for example, find that REDD+ activities must directly target the drivers of deforestation, such as forest fires and illegal logging, as well as structural drivers, such as changes in land tenure and land-use planning. However, they argue that not all drivers are considered, as most schemes do not address cattle ranching, corruption, roadbuilding, and/or commodity demands, among others (see also [9,33]). As important as the choice of the type of intervention is, the definition of the territories that will be prioritized by REDD+. Wolosin, Breitfeller, and Schaap [10] show that the geographical distribution of REDD+ finance can be largely explained by priorities on tree cover, tree-cover loss, and carbon emissions at national (70%–94%) and subnational (58%–72%) levels, though institutional capacity and political commitments have also been influential. Other work highlights significant gaps for specific priority areas. Some scholars point to areas in the Amazon region facing high deforestation pressure that are important for emissions and biodiversity [33–35]. Other scholars argue for additional investments in the network of protected areas, given their importance to date in curbing deforestation and the risks from deforestation dynamics [36,37]. Still others argue that support should also consolidate pristine or intact or stable forests to ensure long-term conservation (e.g., [35]). While the majority of available literature strongly emphasizes improved protection of high-risk areas, at the least for prioritizing additional impacts in the short run, various goals play parts within comprehensive approaches to forest conservation.

Finally, different studies have pointed out that it is not clear that RBF leads to the efficient use of resources, as initially assumed. The proponents of RBF expected that, since receiving countries have a direct financial incentive to reduce deforestation, they would strive to support actions on the ground that contribute directly to that aim. However, a closer look suggests that that empirical evidence on the effectiveness of RBF schemes is either lacking or points to contradictory effects [5], a problem already well known in relation to development aid [38]. On the one hand, authors such as Restivo, Shandra, and Sommer [17] argue that more bilateral aid from the United States Agency for International Development (USAID) has a lowering effect on forest loss. On the other hand, studies such as Hermanrud and de Soysa [22] report that forest conservation funding from Norway's International Forest and Climate Initiative (NICFI), one of the largest aid initiatives in the world and the main donor to the Amazon Fund, has had no effect in halting forest area loss. In a similar way, Bare, Kauffman, and Miller [18], for example, argue that forest conservation funding in sub-Saharan

Africa "is not associated with reduced deforestation rates at the national scale" and even claim that short-term impacts had negative effects. Both studies have strong limitations, since they do not control for other drivers of deforestation, such as agricultural prices, and they assume that relatively small-scale programs (as a percentage of the country's Gross National Product), are going to show effects at national level [39]. Nevertheless, these studies show that there is a growing concern with the effectiveness of RBF, in general and NICFI in particular.

The problem with evaluating the effectiveness of RBF initiatives is that the relations between service users (aid providers) and service providers (aid users) are much more complex than a simplified reading of the principal-agent model found in the studies cited above. According to Paul [7], the contracted agency relationship is often one between the donor organization and a recipient organization or ministry, whereas results may come from other organizations that ultimately spend the financial resources from these donations but have no direct relation with the donor organization (i.e., non-contracted agency relation). In this respect, for example, the UN-REDD+ programme from the United Nations Development Program (UNDP) supports 94 projects in Cambodia, Sri Lanka, Panama, Paraguay, Democratic Republic of the Congo, and Nigeria. However, UNDP are directly related only to the governmental focal point of each country, relating only indirectly with the local beneficiary [40].

According to Van der Hoff, Rajão and Leroy [19], the indirect relations between financial donations, 'project performance', and deforestation rates underlie discursive tensions between donor and recipient countries. While formally all parties agree that RPF should be based solely on deforestation reductions already achieved, donors are also increasingly concerned with the lack of evidence of efficiency of funded projects in driving additional reductions, and in this way fueling a virtuous circle. These tensions and conflicts suggest that the intermediary processes of forest conservation funding are poorly understood, particularly with respect to how they affect aid effectiveness. Some authors have suggested that addressing these conflicts requires new approaches to aid effectiveness evaluations that account for the complex relations of RBF for REDD+, particularly the intermediary stages of forest conservation funding. This could imply, for instance, that transfers should be conditional upon desired results, as within well-implemented payments for ecosystem services (PES) approaches [28]. Such conditions could also require environmental additionality—that is, providing more ecosystem services than they would provide in the non-existence of such payments [41,42]. In addition, REDD+ should be 'financially additional', beyond already planned funding [43]. While attractive, the idea of adding specific demands of additionality to RBF goes against the simplicity and 'hands off' approach that made RBF popular in the first place. Furthermore, this approach would entail a return of many elements of the project-based model defined by Verified Carbon Standard among others, which have also proven to be highly problematic [44]

The growing body of literature presented above presents valuable insights on how RBF should be designed and presents some of its dilemmas and contradictory results. But while allot has been said about how large RBF programs should look like, until recently we lacked a strong record of largescale schemes to look back and draw lessons from concrete experiences. This study provides the first comprehensive analysis of the first decade of the Amazon Fund, the world largest REDD+ RBF program [45,46]. Our study aims to reveal the design choices adopted by the Fund by analyzing its resource distribution across beneficiaries, activities, and geographies. While this study does not provide a quantitative impact analysis of the fund, it allows us to understand how the allocation of financial resources corresponds with various REDD+ design choices, as reflected in the available literature on REDD+, and the extent to which this may affect its long-term effectiveness. From this, this study draws lessons that could be used to improve the Amazon Fund in Brazil and other large RBF programs.

#### **3. Research Approach and Methodology**

This research paper conceptualizes the Amazon Fund as an intermediary organization that links the forest conservation funding provided by donor organizations to the individual projects (see Figure 1). Created in 2008, the Amazon Fund was the first large scale RBF program to be implemented. As such, the fund played an important role in shaping the discussions around REDD+ at the United Nations Framework Convention on Climate Change (UNFCCC). For this reason, the UNFCCC's Warsaw Framework for REDD+ adopted, to a large degree, the modus operandi pioneered by Brazil. Financial donations to the Amazon Fund mainly come from Norway's International Climate and Forest Initiative (NICFI) and the German Development Bank (KfW). The Amazon Fund consists of a steering committee (COFA), which is responsible for establishing allocation guidelines, and a technical committee (CTFA), which is responsible for approving results in terms of reducing emissions from deforestation. The managing organization of the Amazon Fund is the Brazilian Development Bank (BNDES) and is responsible for the approval (or rejection) of submitted project proposals according to predefined guidelines, as well as for the receipt and allocation of financial resources. Since 2015, BNDES has also become eligible to receive financial resources from the Green Climate Fund (decree 8.576/15), whereas other organizations like the government-owned bank Caixa Econômica Federal (CEF) and the Brazilian Biodiversity Fund (FUNBIO) may also become recipients. Financial resources are allocated to a wide variety of organizations. Federal government organizations include the Brazilian Agricultural Research Corporation (EMBRAPA), the Brazilian Institute for Space Research (INPE), the Brazilian Institute for the Environment and Renewable Natural Resources (IBAMA) and the National Police Force (FNSP). Non-governmental organizations also abound and include the Sustainable Amazon Foundation (FAS), the Amazon Institute for Human and Environment (IMAZON), Amazon Environmental Research Institute (IPAM), and The Nature Conservancy (TNC), among others. State government organizations are mostly represented by the environmental or agricultural secretariats of the nine Brazilian states in the Legal Amazon, while some state secretariats outside this region were also recipients. Finally, municipal government secretariats and federal universities were also supported financially by the Amazon Fund.

Understanding how forest conservation funding to the Amazon Fund contributes to the effective reduction of emissions from deforestation and forest degradation involves connecting the project activities (each with a specific shared benefit), geographies, and supported activities, to the overall objective of emissions reduction. The Amazon Fund already provides an annual report that divides the funding distribution according to four broad categories: (1) monitoring and control, (2) land tenure regularization, (3) sustainable production, and (4) scientific and technological development [13]. However, to understand the allocation of financial resources in light of the design outlined above, it is necessary to further refine the available information from the Amazon Fund. For this purpose, we have built a project database with detailed information on the beneficiaries, activities and geographies that received financial resources from the Amazon Fund (see Figure S8 in Supplementary Materials).

Our primary data source is the Amazon Fund´s website, as well as its annual activity reports (see Figures S3 and S6 in Supplementary Materials). We collected all data available on all of the 96 projects that received support between 2008 and 2017. This data includes project objectives, beneficiaries, implementing organization, territorial scope, committed and disbursed amounts, and activities conducted, among other information. Websites of project owners provided additional information. To refine the data for providing geographical information, we used the municipality as the entity. In Brazil, municipalities reflect the smallest geographical unit for monitoring deforestation, applying public policies, allocating government resources, and evaluating (see Tables S1 and S2 in Supplementary Materials).

One of the main challenges of generating data at the municipal level is the variation of project target areas, which may involve biomes, river basins, protected areas, or indigenous territories. Based on the available literature, we designed rules to determine the municipalities encompassed by each project (see Figure S5 and Tables S3 and S4 in Supplementary Materials). When project disbursements covered multiple municipalities, we used a weight factor in order to determine the share of financial support that each municipality received (see Figure S7 in Supplementary Materials). After the geographical allocation of financial resources, we further categorized the dataset

by main-component, which reflects the Amazon Fund´s theory of change. As projects may contribute to multiple main-components, we conducted one interview by email with a BNDES manager, the managing organization of the Amazon Fund, who replied with a spreadsheet including data dividing the investments of each Amazon Fund project by main-component. Finally, we further categorized the dataset by activity (also called specific-components). As a main-component can be composed by multiple activities, if more than one activity by main-component was verified, then the amounts were equally divided across them. The assumptions in response at divergences or limitations of data collected are presented at Figure S6 in Supplementary Materials. The final database contains 10,493 lines of information structured by project, location, main-component, and specific-component (see Figures S1, S4 and S8 in Supplementary Materials). The procedures for collecting and interpreting data, and constructing the database, are detailed in the supplements outcomes (see Figure S2 in Supplementary Materials). The Amazon Fund accountability is in Brazilian Reais currency. All financial data were converted from Brazilian reais to US dollars by using the rate for the day they were received, which corresponds with the methodology used for the English publications of the Amazon Fund. To evaluate the additionality of the Brazilian governmental agencies budgets (accountable in Brazilian reais) with the Amazon Fund disbursements, we used an average exchange rate between 2009 and 2017, in order to reduce the effects of exchange rate fluctuation.

**Figure 1.** The Flows of Amazon Fund. Redd, Reducing Emissions from Deforestation and Forest Degradation; CEF, Caixa Econômica Federal; FUNBIO, Brazilian Biodiversity Fund; IBAMA, Brazilian Institute for the Environment and Renewable Natural Resources; EMBRAPA, Brazilian Agricultural Research Corporation; FNSP, National Police Force; INPE, Brazilian Institute for Space Research; CENSIPAM, Center for the Management of the Amazon Protection System.

#### **4. Results: Resource Allocations by the Amazon Fund**

Currently, the approval of projects and disbursements are made on the basis of criteria and guidelines updated biannually by COFA. The 2017–2018 document lists 14 minimum requirements that potential projects must meet, some (i.e., items B4, B5, B6, B7, and B14) determining conceptual boundaries of project activities. Projects also must demonstrate coherence with environmental and forest policies, most notably the national Action Plan for the Prevention and Control of Deforestation in the Legal Amazon (PPCDAm), including its manifestations in state governments (PPCDs), and the national policy for Regenerating Native Vegetation (ProVeg) [13]. Projects are also evaluated with respect to coherence with Brazil's National REDD+ Strategy (ENREDD+), which in turn incorporates implementation of PPCDAm and compliance with the Brazilian Forest Code. Finally, projects are expected to be financially additional, i.e., to go beyond existing public environmental budgets and other forms of finance.

The Amazon Fund maintains an open channel for submissions indicating that 80% for the resources should be invested in the Amazon biome (an area that encompasses 40% of the country). In addition to that, the fund also has made public calls aiming at fostering specific activities, such as sustainable production, inclusive value chains and the management of indigenous lands. These calls account for 8.4% of the resources committed by the fund by December 2017. A recent call for forest restauration from 2017 added a spatial priority criteria that provides up to 12 points (from a total of 100) if the project is located in a high priority water basin within a municipality blacklisted as a top priority for deforestation control by the Ministry of Environment [47]. In both the calls and the open submission channel, however, the Amazon Fund adopts largely a passive approach, waiting for project owners to send proposals, rather than actively identifying areas under high risk of deforestation where the impact of the resources would be maximized in terms of deforestation reduction.

#### *4.1. Benefit Distribution Across Stakeholders*

The distribution of financial commitments across stakeholders shows some variation across years (Figure 2, left panel). In 2017, over 95% of a total of USD 667.3 million went to state governments (USD 256.6 million) or NGOs (USD 241.1 million) or federal governments (USD 140.6 million), with their shares varying considerably per year. Of a total of USD 140.4 million in 2013, about 70% (or USD 102.9 million) went to projects of state governments that received almost no such commitments either two years earlier or two years later. This peak took place as a consequence of a change in the rule of the Amazon Fund that allowed the approval of larger "structural projects", as the implementation of the Rural Environmental Register (CAR). By contrast, commitments to NGOs projects were relatively stable over time, averaging USD 22 million until 2016, though rising to USD 44.5 million in 2017 (implying variation in the NGOs' share). Commitments to federal government projects were also uneven, with slight peaks in 2012 and 2017 (USD 31.7 million, 41.2 million).

**Figure 2.** Annual committed (**L**) and disbursed (**R**) amounts per stakeholder (in million USD).

However, the ability of different stakeholders to approve projects with the Amazon Fund did not match their implementation capabilities. In the last decade, only USD 405.3 of 667.3 million (i.e., 60.7%) has been transferred to project owners. Average annual disbursements to state governments have hovered between USD 16 and 21 million in most years, with a sudden peak of USD 47.6 million in 2014 and then a sharp drop to USD 4.8 million in 2015. Disbursements to federal government increased exponentially from a small base of only USD 2.4 million even in 2014 to USD 37.7 million in 2017. Finally, disbursements to NGOs steadily increased from USD 6.4 million in 2010 to USD 30.7 million in 2017. From these three groups of beneficiaries, the Federal Government has been demonstrated the largest implementation gap, starting with a very low implementation rate and reaching the execution of only 47% of the committed values by 2017. This was followed by the State Governments, whose spending rates stayed below 50%. Municipalities, Universities, and NGOs, in contrast, presented a better implementation capacity, being able to invest most of the resources obtained from the Fund.

To understand these variations in disbursements, we must also consider the characteristics of the projects supported by the Fund. Federal government projects, for instance, were concentrated within eight projects involving six recipient agencies. Of the total amounts in this category, USD 64.3 million (i.e., 47.2%) went to organizations that develop satellite-based monitoring systems and provide information on deforestation trends, namely INPE and CENSIPAM. Another USD 35.9 million (i.e., 26.7%) went to organizations responsible for enforcing environmental laws and policies, namely IBAMA and FNSP. The remaining USD 40.5 million (i.e., 25.9%) went to EMBRAPA units to disseminate knowledge about sustainable production and the recovery of degraded areas throughout Brazil, and to the Brazilian Forest Service (SFB) for the collection of information aiming to increase the forest data available (see Section 4.3). While the IBAMA manage to invest 17.5% of the funds received, by 2017, INPE and CENSIPAM used only 58.6%, implying that the development of a radar-based monitoring system is lagging behind schedule.

The committed and disbursed peaks for state government projects in 2013 and 2014 (Figure 3) correspond with contextual factors as well, including a surge in state government projects toward development and implementation of the Rural Environmental Register (CAR). CAR is a federal policy instrument introduced in 2012, with the adoption of the new Forest Code (law 12.651/2012), to enhance law enforcement capacity. However, despite the federal law and a centralized national system, the registers must be executed at state or municipal level (art 29, §1). CAR implementation has, therefore, become a major concern for state governments, especially after the system went live in 2014 [48]. This can be seen in both spending and appeals to the Amazon Fund [13]. Within the 13 states that have approved projects, 85% of disbursements went to seven of the nine inside the Amazon Biome.

The linear increase in disbursements to NGOs reflects yet another set of contextual factors, in this case related to Amazon Fund process adjustments over time. Disbursements to projects were slow, to start, due to rigid assessment procedures intended to show professionalism; in the eyes of donor organizations and BNDES management, that slowness also reflected some lack of understanding of project owners [13,19]. Minutes of COFA meetings indicate that, in response to these challenges, the Amazon Fund adopted a number of measures in order to facilitate and accelerate the disbursement process, including public calls for submitting project proposals. While the consequences of these responses are reflected in the linear increase in approved projects and disbursements to NGOs, 80% of the financial resources were concentrated in half of the NGOs that received support from the Amazon Fund. While the Amazon Fund does include distributional equity amongst its performance criteria, this concentration reveals that usually only high-capacity and professional civil society organizations, such as FAS, IMAZON, and TNC, are able to access the fund (see Figure S9 in Supplementary Materials).

In addition to exposing the implementation capability of different governmental agencies, a comparison between the disbursement of the Amazon Fund with the yearly government budget also reveals the ability of the Fund to foster additional actions. One of the key principles of the first donation contract signed in 2008 between Norway and Brazil was the warrant that the Amazon Fund

would not replace but supplement tax payer funds [2,20,22,49]. However, it is possible to observe that the increases in disbursements to federal agencies coincided with their decreasing governmental budgets, particularly after 2014 (Figure 4). This suggests the occurrence of a partial substitution for the agency expenditure of taxpayer-funded budgets using the Amazon Fund. For instance, IBAMA's committed budgets to reduce deforestation, combat fires, and conduct environmental inspections have been reducing since 2012, with a strong reduction from USD 50.64 million in 2014 to USD 29.07 million in 2017. These reductions have been partially offset by Amazon Fund disbursements starting in 2015. Similarly, INPE´s budget fell from USD 84.5 million in 2010 to USD 43.63 million in 2017, 2017, and CENSIPAM has also lost more than 70% of its governmental funding from 2009 and 2017. In those three cases, the Amazon Fund played an important role offsetting those budgetary losses from 2015 onwards, in the case of CENSIPAM even outmatching governmental funds. Those trends include rising implementation rates for turning federal commitments into disbursements, which increased from 3.7% in 2014 to 26.8% in 2017.

**Figure 3.** Implementation rates as disbursed, divided by committed (consolidated amounts), by Stakeholder.

These observations cannot, by themselves, confirm a direct causal relationship between the increasing financial disbursements from the Amazon Fund and the decreasing budgets of the recipient federal agencies. Furthermore, it should be highlighted that the period following 2015 witnessed one of the worst political, economic, and fiscal crises in Brazil's history. At the same time, however, contextual factors seem to correspond with an interpretation that the forest conservation funding provided through the Amazon Fund lacks in some instances financial additionality, particularly considering the unfavorable political climate for environmental protection [50], the greater flexibility within forest legislation since 2012 [51], multiple bills for reducing environmental protection during election year 2018, and, as a consequence of all these factors, rising deforestation rates since 2012 [52].

**Figure 4.** Comparison of Federal Committed Budgets with the Amazon Fund disbursements for INPE, IBAMA and CENSIPAM (used average 2009–2017 exchange rate: 2.434). Committed amounts represent the Portuguese term, 'Empenhado', an act that guarantees that there is an amount necessary to pay for an assumed commitment and creates a payment obligation for the government.

#### *4.2. Geographical Distribution*

Spatially, Amazon Fund allocations display a large concentration (Figure 5a) in 64 municipalities along the (Figure 5a) region stretching from the southeast of Pará towards the western regions in the Mato Grosso, Rondônia and Acre states, municipalities that contain, since 2000, the highest consolidated deforestation rates in Brazil. NGO and state projects explain much of this concentration (Figure 5b,c), whereas federal projects had no significant contribution, mainly due to their nationwide focus (Figure 5c,d). Federal government projects are the most evenly distributed across the landscape, averaging below 26 USD/ha, which could be due to the all-encompassing nature of the geographic information systems (GIS—Geographic Information Systems) and remote sensing activities that these projects tend to promote. At the same time, disbursements to larger federal agencies, such as EMBRAPA, tend to concentrate in eight cities in the Legal Amazon, including Rio Branco, Manaus, Boa Vista and Macapá, where these agencies are located (Figure 5d). Finally, while municipalities benefit indirectly from various types of support, direct support only went to 6 of the 772 municipalities in the Legal Amazon and amounted to only USD 7.8 million. Most of these resources (65.2%) went to the municipal government of Alta Floresta, in northern Mato Grosso. In addition, the Amazon Fund had also financed research of the state universities of Pará (in Belem) and Amazonas (in Manaus) as well as to the development of satellite-based monitoring systems by INPE in Manaus (Figure 5g).

State government projects are mostly responsible for monitoring and control (Figure 5c), particularly through activities, as the structuring of environmental secretariats, CAR implementation, and training of firefighters (see Section 4.3 for details). State governments that more actively sought the support of Amazon Fund for monitoring and control were Acre, Maranhão, Tocantins, and Rondônia. Particularly, Acre has a strong presence in investments in sustainable production, spread throughout its territory (Figure 5e,f). The distribution of resources also portrays low intensity towards Land Tenure Regularization activities, independent of the region or stakeholder (Figure 5h), However, the Amazon Fund allocations did not systematically privilege the municipalities that showed the recent highest deforestation rates. For instance, from the 10 municipalities with the highest deforestation rates in 2017, only two were amongst the top 100 receiving per/Ha, considering the 775 municipalities from Legal Amazon. Furthermore, the support from the Amazon Fund tend to arrive in a context in which clearings have already been reduced substantially due to other factors or the depletion of forests (see Table S5 and Figure S10 in Supplementary Materials). This spatial pattern of project distribution

confirms the apparent lack of strategy of the Amazon Fund, as a consequence of a largely passive approach that waits for proposals rather than actively seeking opportunities for fostering projects in areas with high deforestation risk.

**Figure 5.** Spatial distribution of Amazon Fund investments per municipality by Stakeholder and by main-component.

#### *4.3. Distribution Across Activities*

Almost half of the total commitments (USD 667.3 million) has gone to monitoring and control (USD 326.7 million), while one third (USD 201.9 million) has gone to sustainable production (see Figure 6 and Table 1). The latter category has been relatively steady over time, as have the small land tenure commitments. By contrast, the large investment monitoring and control have been uneven over time: starting slow with an average of USD 20.3 million in the first four years, peaking in 2013 at USD 94.0 million, and then settling at an average of USD 30.6 million from 2015 on (Figure 6, left panel). Finally, nearly all commitments for scientific and technological development occurred in 2012 (USD 40.7 million).

**Figure 6.** Annual committed (**A**) and disbursed (**B**) amounts per main-component (in millions USD).

Although slightly slower than noted above, actual disbursements to individual projects have corresponded to commitments, with most disbursements going to monitoring and control (49.6%) and sustainable production (31.9%). Monitoring and Control was responsible for most of the variation (see right graph of Figure 6), peaking in 2014 (USD 43.1 million) and 2017 (USD 53.5 million). Notably, disbursements for scientific and technological development have never gotten much traction, slightly peaking only in 2013 and 2014, and also presents the lowest implementation rate up to 2017 (Figure 7).

Monitoring and control efforts involved mostly state and federal government projects (USD 187.1 million and USD 100.1 million, respectively). It was the only category, though, that included the unique international project supported by the Amazon Fund, aiming to help develop the capacity to monitor deforestation in eight neighboring countries that also contain the Amazon biome (USD 11.8 million). However, most of the monitoring and control investments (USD 113.0 million) was allocated to CAR implementation. A large share of the funds provided for this activity (USD 102.5 million) was used by state governments to acquire equipment (GPS, computers, software) and provide training for effective processing of CAR proposals. Another share (USD 52 million) was invested in the capacity-building of environmental secretariats for CAR implementation and other environmental policies, including the creation of municipal secretariats, the acquisition of cars and buildings, the hiring of employees and training in-monitoring deforestation, landscape analysis, sustainable supply chains, and measurement. In addition, some resources were used to promote CAR among landowners and to provide georeferencing services for landowners. A small amount went to development of a state system for granting environmental licensing to new businesses and companies. Therefore, in total, 18% of the resources committed by the fund have been invested in the implementation of CAR.


improving methods to estimate biomass and emissions; make available a solution for storing and processing a large volume of geospatial data, called "Brazil Data Cube", between others.

**Figure 7.** Implementation rates as disbursed divided by committed (consolidated amounts), by main-component.

Monitoring activities that were exclusively promoted by federal government organizations involved the improvement of satellite-based monitoring systems for fighting deforestation (PRODES— Annual Deforestation System and DETER—Real Time Deforestation System, USD 76.1 million) and forest fires (PREVFOGO-Fire Prevention System, USD 6.3 million). State governments also invested in forest fire combat (USD 32.5 million), but emphasized control activities (e.g., the creation of firefighter units), rather than monitoring activities. Other investments by federal government organizations targeted the strengthening of law enforcement (USD 29.6 million) in two projects by IBAMA and FNSP; this funding was mostly spent on the acquisition of vehicles, helicopters, equipment, and buildings. While NGOs received much financial support from the Amazon Fund (USD 241.1 million), their support for monitoring and control activities was relatively small (USD 11.6 million) and only involved CAR implementation.

In the category of sustainable production, resources mostly went to NGOs (USD 154.7 million) and state government organizations (USD 42.1 million) (see Table 1). Nearly all state government investments went to the promotion of sustainable forest activities, the acquisition of equipment (tanks, driers, processing units' machines, warehouses), and the provision of professional training and technical assistance (in pisciculture and aquaculture, nut and Açaí extraction, pasture management, as well as forestry and agroforestry systems). This result suggests that the social benefits from the Amazon Fund in terms of rural poverty reduction and sustainable farming were carried out mostly by NGOs and state governments.

Investments in regularizing land tenure, notably spending on territorial zoning and protected-area management and indigenous lands, came almost exclusively from state governments (USD 23.8 million) and NGOs (USD 46.6 million). This investment provides indirect benefits for indigenous peoples, quilombos (descendants from fugitive slaves), riverine people, smallholders, and settlements. No such investments were federal. Federal governments did invest substantially in scientific and technological

development, which involved field data collection by the Brazilian Forest Service (SFB) for building the National Forest Inventory (USD 31.7 million).

Universities, by contrast, invested the most financial resources in scientific research (USD 4.7 million) and development of the research infrastructure (USD 3.9 million). For instance, one project from the Federal University of Pará conducted research for the development of new products from bioactive compounds of plants typical of the Amazon Biome (USD 0.7 million), and invested in the development of new forest products, such as herbal medicines, cosmetics, and food products, among others. Natura, a private cosmetics company from Brazil, announced in 2016 an investment of more than USD 70 million in biodiversity inputs as part of its Amazon Program that aims to develop a new line of products with origins in Amazon Biodiversity.

#### **5. Amazon Fund Design Choices and Effectiveness**

The findings of our analysis of the recipient projects in the Brazilian Amazon Fund reflect a broad variety of stakeholders and activities. Following the categorization of Luttrell, Loft, Fernanda Gebara, Kweka, Brockhaus, Angelsen, and Sunderlin [27], the recipient projects of the financial resources from the Amazon Fund often involve the largely indirect contributions of effective facilitators, legal rights holders, cost-incurring groups, forest stewards, or poor communities. Moreover, the Amazon Fund's financial resources were channeled towards the direct and structural drivers of deforestation, but this distribution was not proportional to the importance of addressing these drivers, as argued by some scholars (e.g., [32]). Investment patterns tend to reflect specific relations between specific stakeholder groups and project activities. Although activities also vary considerably, there are some general patterns. Federal government organizations tend to invest in development of monitoring systems (45.7%) and inventory data (22.6%), which reflects a main concern with gaining control over deforestation dynamics. State government organizations tend to invest mostly in CAR implementation (40.1%) and capacity-building for state and municipal organizations (20.3%), thereby incurring many of the costs of federal policies. Finally, investments by NGOs have mainly benefited local communities who aim to adopt sustainable production activities (64.2%), but NGOs have also supported (more than federal or state government organizations) land tenure regularization projects (19.3%).

The geographical distribution of financial resources seemed to follow a more focused rationale. We found that many project organizations were located in municipalities with the highest consolidated deforestation rater of Brazil. For instance, NGO projects for territorial and ecological zoning, strengthening of PA and IT management, as well sustainable production, represent 30% of the total disbursements from the Amazon Fund and were largely located in this region. Disbursements from the Amazon Fund to the three main recipient categories have generally benefited municipalities located in areas where deforestation threats are highest [53]. This observation only partially corresponds with the findings by Wolosin, Breitfeller, and Schaap [10], as we found no evidence of substantial contributions to areas with high tree cover, which are more commonly found in remote areas of the Amazon biome [35].

Within the pre-established main-components of the Amazon Fund, we also found variation in the activities that compose these categories. For instance, while most financial resources were channeled to the strengthening of monitoring and control activities by federal and state governments (USD 287.2 million), their investments have focused on monitoring activities like satellite imaging (USD 70.6 million) and CAR implementation (USD 102.5 million). This result contrasts with the substantially smaller investments in control activities like combating forest fires (USD 32.5 million) or law enforcement (USD 29.6 million). This trend is representative of the broader resource allocation within the monitoring and control category. Similarly, investments in land regularization were mainly directed at indigenous territories and protected areas (USD 66.0 million), whereas smallholders (USD 4.3 million) received much less support.

Based on our findings on the variations in financial resource distribution, we argue that the project owners impose a substantial influence on the nature of activities that forest conservation funding ultimately supports. The current approach adopted by the fund incentivize project submissions in activities and geographies where they may be most successful in reducing deforestation had a limited effect. Corresponding with the study by Weatherley-Singh and Gupta [32], for example, the Amazon Fund restricts financial resource allocation to the four main-components of its theory of change, while not addressing alternative factors, such as the impacts of cattle ranching, road construction, international demand for agricultural products, or corruption. However, any project proposal that adheres to the project quality criteria and guidelines of the Amazon Fund [13] may become eligible for financial support. In other words, the Amazon Fund takes a more passive stance towards resource allocation after the criteria and guidelines are in place. This view accounts for the great variety of stakeholders, activities, and geographies, as described above, since each stakeholder category seems to prefer a different investment strategy. Such behavior may ultimately undermine the effectiveness of conservation funding provided by Norwegian and German donor organizations, at least in terms of emissions reductions.

As already argued in Section 2, the Amazon Fund´s theory of change is generally geared towards deforestation reduction, but the design choices of individual projects are primarily directed at contributing to one or more main components. The evaluation of a completed project in northern Mato Grosso [46], for instance, indicates that the project geared its intervention logic upon its contribution to the main-components "sustainable development" and "monitoring and control", and stated that the main contribution to emissions reductions came from "the restoration of native vegetation and pastures and the planting of native species in permanent protection areas". The extent to which such projects achieved emissions reductions was not stated in the report and would admittedly be a complex methodological endeavor. The leeway that projects have in contributing to these main-components, although important for attracting project proposals, accounts (at least partially) for the imbalanced allocation of financial resources discussed above and may, to some extent, undermine the Amazon Fund's contribution to deforestation reduction.

It is important to note that this undermining of the Amazon Fund´s overall contribution is by no means intentional. At the same time, there are also indications that some projects require a more in-depth evaluation and a longitudinal approach in order to observe their outcomes come to fruition. Particularly but not exclusively, projects from governmental organizations are under greater pressure from critical considerations of their contribution to emissions reductions. One may argue that investments in CAR implementation, for example, support more structural improvements of a nation-wide instrument to enhance monitoring capacity, but some studies point out that it is still unclear whether and to what extent this instrument, indeed, contributes to reducing deforestation [48,54]. In addition, our analysis indicates that federal government organizations (i.e., CENSIPAM, INPE and IBAMA) tend to lack financial additionality. Particularly, the substitutive nature of the Amazon Fund financial resources of IBAMA projects is worrying, because these investments often involve more direct contributions to reducing deforestation, most notably the enhancement of (the capacity for) environmental inspections and fire combat. While the lack of funding for law enforcement may have led to an even higher spike on deforestation rates, a country with a mature enough environmental governance should be able to grant a stable source of public funding by giving priority to this agenda.

#### **6. Conclusions**

Our analysis also helps to understand why empirical studies seem ambiguous about the effectiveness of forest conservation funding. As explained in Section 3, BNDES' approach to distributing financial resources from the Amazon Fund to individual projects occurs based on the evaluation of project proposals based on the funds widely encompassing guidelines rather than a strategic selection of projects based on the need to reduce deforestation in areas under threat. As a consequence, our findings show that disbursements by the Amazon Fund to individual projects adhere to very diverging theories of change within a broader REDD+ and RBF strategy. The contribution of each individual project for deforestation reduction are complex to be measured and require additional studies [17,18,22]. Nevertheless, our results suggest that the lack of strategic focus of disbursements may compromise the ability of the fund to obtain further deforestation reductions on the short term.

It is particularly concerning the observation that the resources provided by the Amazon Fund have offset budgetary losses from the Brazilian government in some areas, putting into question the financial additionality of the fund. At the same time, deforestation rates have been on the rise since 2012, the same period during which the fund has started to take place more steadily [19]. It should be emphasized that the fund is not expected to influence deforestation rates for the whole biome, and the lack of additionality in some years can be explained by the economic and fiscal crisis in Brazil. However, these trends taken jointly may weaken the credibility of financial support from the Amazon Fund and other RBF programs on the long term. The sustainable development activities in NGO projects seem to incite less critique, but these projects require much closer scrutiny in order to understand the extent to which they indeed reduce deforestation. Our analysis confirms the argument by Van der Hoff, Rajão, and Leroy [19] that the "demands for demonstrating the results of the Amazon Fund in a scientifically rigorous manner are likely to become an important topic for donor countries".

Alternatively, the Amazon Fund could adopt a more active approach to the allocation of financial resources, for example, by prioritizing a smaller set of activities, with a strong geographical focus. Most importantly, the Amazon Fund should actively identify potential locations and project owners and assist them in constructing high impact proposals. Likewise, the fund should also improve its impact monitoring capabilities and provide incentives to projects that deliver deforestation reductions within the timeframe of the project. This is especially important, as the political climate in Brazil, United States and other countries has become more hostile to environmental interests [52,53,55].

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4907/10/3/272/s1, Figure S1: Model for Database Structuration, Figure S2: Steps to collect the variables, Figure S3: Individual Project Page on Amazon Fund website, Figure S4: Database structured at Level I—Projects, Figure S5: Diagram of rules to determine municipalities encompassed by projects, Figure S6: Project Tree, Figure S7: Municipalities weighted by project, Figure S8: Final Database Structure, Figure S9: Pareto graft for NGO's and State projects (USD left bar and % of committed amounts right side, Figure S10: Deforestation in Legal Amazon, PRODES-INPE (2017), Table S1: Municipalities geospatial information sources, Table S2: Municipalities Data Source, Table S3: Variables included in the main-component level, Table S4: Weight calculations per main-component, Table S5: 10 Municipalities with the higher deforestation rates between 2016 to 2017. PRODES-INPE (2017), Table S6: Research assumptions in response at divergences/limitations of data collection.

**Author Contributions:** Conceptualization, J.C., R.v.d.H. and R.R.; Data curation, J.C.; Formal analysis, R.v.d.H. and R.R.; Investigation, J.C. and R.v.d.H.; Methodology: J.C. and R.R.; Visualization, R.R.; Writing—Review & Editing, J.C., R.v.d.H. and R.R.

**Funding:** This research was supported by the Universidade Federal de Minas Gerais, through the Pro-Rectory of Research-PRPq/UFMG. Also was financed by CAPES–Brazilian Federal Agency for Support and Evaluation of Graduate Education within the Ministry of Education of Brazil.

**Acknowledgments:** We thank Alexander Pfaff from Duke University for the inspiration and academic support, and Centro de Inteligência Territorial (CIT), PRPQ/UFMG and CAPES for providing funding for this research.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Spiders in the Web: Understanding the Evolution of REDD+ in Southwest Ghana**

#### **Jan Willem den Besten 1,2,\*, Bas Arts <sup>1</sup> and Jelle Behagel <sup>1</sup>**


Received: 22 November 2018; Accepted: 17 January 2019; Published: 2 February 2019

**Abstract:** The implementation of the global programme on Reducing Emissions from Deforestation and Forest Degradation in developing countries, and the role of Conservation, Sustainable Management of Forests and Enhancement of Forest Carbon Stocks (REDD+) is lacks a robust financial mechanism and is widely criticized for producing too little positive impact for climate, nature, and people. In many countries with tropical forests however, a variety of REDD+ projects continue to develop on the ground. This paper fills in some of the gaps in our understanding of the dynamic relation between global policymaking and implementation of REDD+ on the ground. Using the introduction of REDD+ in Southwest Ghana as an example, we apply a practice-based approach to analyze the different roles that local actors and global-local intermediaries played in the introduction of REDD+. Our results show a more balanced picture than polarized debates at the global levels suggest. The logic of practice explains how REDD+ was translated to the local situation. Global actors took a lead but depended on local actors to make REDD+ work. Together, they integrated elements of existing practices that helped REDD+ 'land' locally but also transformed REDD+ globally to resemble such local practices. REDD+ initiatives absorbed elements from established community-based conservation, forest restoration, and sustainable agro-forestry practices. The evolution of REDD+ in Ghana reflects global trends to integrate REDD+ with landscape approaches.

**Keywords:** REDD+; practice-based approach; global-local nexus; forest and climate policy; Ghana

#### **1. Introduction**

Over the last decade, Reducing Emissions from Deforestation and Forest Degradation in developing countries, and the role of Conservation, Sustainable Management of Forests and Enhancement of Forest Carbon Stocks (REDD+) emerged as an international effort to fight tropical deforestation and to mobilize finance for reducing CO2 emissions based on avoided deforestation and forest degradation. It developed from a daring proposal in 2005 at the 11th Conference of Parties (COP) at the United Nations Framework Convention on Climate Change (UNFCCC) to a key component of the 2015 Paris Agreement. The 2013 Warsaw Framework was particularly important because it provided guidelines on monitoring through remote sensing and ground-based observations. Subsequently, the Paris agreement created the basis for REDD+ countries to attract investments from donor nations and private sector actors through zero-deforestation policies and carbon markets linked to the national climate action plans or Intended Nationally Determined Contributions (INDC) of REDD+ countries [1–5].

REDD+ developed because it was seen to be a legitimate climate mitigation option that involved developing countries in climate mitigation with promises to benefit local communities and biodiversity conservation [6]. Over time, the introduction of REDD+ has resulted in a plethora of REDD+ initiatives. REDD+ has a strong market share in the voluntary carbon space and bilateral and multilateral

programs have pledged billions of USD to REDD+ preparations including 100 million to Ghana [7]. Given this flexibility, REDD+ is considered to take place in a 'global-local nexus' of forest governance [8]. The global-local nexus combines the qualities of multi-level governance—that highlights how policies are connected across global, national, and local levels—with an emphasis on how networks of actors shape policy more horizontally as well [9,10].

The global-local nexus has not received much academic attention yet in forest policy. Selected studies do study how global, national, and local levels together affect forest policy [9–11]. These studies highlight the specific interactions that occur between different levels of governance. In the case of REDD+, such studies often include calls for better inclusion of local actors and coordination across levels [10,11], amongst others. Other studies limit themselves to the interactions between global and domestic levels of governance more specifically [12–14], mostly focusing on the effectiveness of international efforts—including REDD+—to influence forest policies on the ground. However, studies that focus both explicitly on the global-local nexus and on how actors actively move from one level to another are hard to find. The few studies that do so emphasize that interactions amongst governance actors produce specific practices that are highly sensitive to social-ecological context [15–17].

Critics REDD+ argue that projects fail to be sensitive to both global aspirations and local needs. First, the demand for REDD+ projects that produce emission reductions is relatively small. Reasons include the slump in global carbon prices due to the post-2008 economic slowdown and the failure of the Parties to the UNFCCC to agree on a financial architecture for REDD+. Second, REDD+ is viewed by some as a false solution that deflects attention from the need for companies and governments in the industrial world to take a lead in decarbonizing the economy and places too much responsibility on local communities. Third, the inclusiveness and effectiveness of REDD+ on the ground is challenged by critics that fear for the exploitation and further marginalization of local communities in light of inadequate legal frameworks and a lack of transparency and rule of law [18–23]. Finally, Fletcher et al. [24] argue that conservation markets cannot compete with extractive markets and will always need additional support in terms of subsidies and regulation.

Responding to critiques of REDD+, Angelsen et al. and others [18,25] argue that REDD+ serves as a discursive resource that is constantly reproduced—and altered—in practice. Indeed, Turnhout et al. [26] highlight that REDD+ over time has moved from being a carbon-centered, market-based instrument, to include broader climate mitigation and nature conservation strategies that focus on co-benefits and landscape approaches. Den Besten et al. [27] illustrated that this 'evolution' of REDD+ takes place in the global-local nexus where actors and ideas travel across levels: the capacity building and learning activities of local REDD+ piloting and testing informed REDD+ policy development at the global level. This contributed to the prioritization of inclusive and extensive REDD+ preparations and governance development, ahead of possible future carbon finance mechanisms [6,28].

Ghana is a prominent REDD+ country that has witnessed one of the highest deforestation rates in the world [29]. In 2010, 21.7% of land or 4,940,000 hectares was covered by forest [29]. Deforestation is a critical environmental and economic issue and Ghana. Since 1990, the country has lost more than 33.7% of its forests [29], costing the forest sector an estimated USD \$500 million [30]. Subsequently, Ghana received a lot of attention from REDD+ donors and programs. At the same time, it is a country where challenges regarding land rights and inclusive governance by some were seen to complicate the implementation of REDD+ [31]. Despite pledges and commitments from REDD+ donors, relatively little of these funds are shown to reach actors at the lower levels of governance [7]. Early REDD+ actions in Ghana have nevertheless contributed to a strong REDD+ commitment as part of the country's INDC in the context of the Paris agreement [32].

In this article, we show that both global-local intermediaries and local actors have played a key role in the evolving meaning and practice of REDD+ in Ghana, often with limited funding on the ground and in areas where the challenges in terms of land and tenure rights were great. To better understand how the global idea of REDD+ was increasingly shaped by local REDD+ initiatives and

practices, this article focuses on the process of REDD+ introduction in Southwest (SW) Ghana between 2007 and 2017. We apply a practice-based approach [15,33] to analyze how actors across governance scales shape the meaning of REDD+ and how key elements of REDD+ were introduced in practice.

#### **2. Materials and Methods**

When analyzing global forest policy, many studies have generated useful insights into the roles that global actors—including governments, non-governmental organizations (NGOs), development agencies, and companies—play on higher levels of governance [6,33–36]. These studies discuss how actors may form new and sometimes unexpected alliances in diplomatic processes [37] and subsequently form coalitions that support new forest discourses, including REDD+. At the same time, these studies often fail to explain how local community members exactly respond to new ideas, policies, and programs in their daily activities and how they relate to the actors designing these policies and programs and how individual actors can operate at multiple governance levels.

To flesh out how local levels of forest governance contribute to global policy development, we need to better understand how the agency of actors is shaped in these contexts [15]. Practice-based approaches explore how actors shape their ideas, identities and behaviors in the context of social practices. In particular, they emphasize that reality—and thus also change—emerges from our practical engagement with it [15]. Practices can be broadly defined as routinized behaviors where meaning and action are entwined [33,38]. Actors are situated in these practices. Their ideas, identities, and behavior are shaped as they reproduce the practice of which they are part [39,40]. Practices are therefore not only just 'entities' but also 'performances' [15,33] of which actors are an active part: as they perform practices, they simultaneously change practices over time.

Practices are not restricted to a single place. They emerge in one place and can travel over time and space, including across a global-local nexus. To understand this mobility of practice, Shove et al. [33] identify three key elements that make up a practice: meaning, competences, and materiality. Meaning includes images and ideas that form the emotional, motivational and normative components of practice. Competences include technical knowledge, know-how and skills that people have or need to perform the practice. Materiality includes physical and technological attributes that are part of society, such as machines, cooking stoves, utensils [33]. These elements are thought to give unity to a practice: when their links are strengthened, the practice persists and when their links are broken the practice dies out. Change can mean that the link between two elements (e.g., meaning and competence) persists while the link with a third element (e.g., materiality) is broken. For our analysis, we use the operationalization of the conceptual framework of Shove et al. [33] by Arts et al. [41]. They interpret meaning as ideas and discourses; competences as standards and procedures; and materiality as technologies and resources.

When practices arrive in a certain place, people unpack them and have the agency to fit them to the local situation [33]. When actors do so, they can change elements and combine them in new ways with the elements of other practices. This ability of actors to change practices within specific situations is called 'situated agency' [40]. It means that actors have the freedom to change practices, not so much because they are autonomous agents, but more so because they are part of the practices that they find themselves in [40]. Accordingly, they have a certain freedom and agency to improvise on existing practices. This improvisation is often done by recombining existing elements of practice, also known as institutional 'bricolage': "a process through which people, consciously and un-consciously, assemble or reshape institutional arrangements, drawing on whatever materials and resources are available, regardless of their original purpose" ([42], p4).

We apply the practice-based approach [33,40] and the idea of 'bricolage' [42] to the case of the introduction of REDD+ in SW Ghana. Doing so, we explore how REDD+ 'traveled' to this area as a set of elements of practice. The practice-based approach makes it possible to analyze how actors received, interpreted, and adjusted REDD+ in SW Ghana to their local practice. By dissecting REDD+ into the materiality, competencies and meaning as elements, it was possible to assess how these elements

combined with elements of other, existing practices as part of the local uptake of REDD+. The actors reconfigured elements via their situated agency in existing practices in the unpacking, influencing, and re-aligning of the elements of REDD+. We focused in particular on how groups of actors sought to introduce new elements of practice in existing routines and practices of forest management and community-based conservation.

We collected data on the introduction of REDD+ in the cocoa growing landscape in the High Forest Zone in SW Ghana from 2009 to 2016. (see Figure 1). Ghana has both seen high rates of deforestation [29] and the development of a variety of REDD+ initiatives. Between 2009 and 2016, the first author helped organize several dialogues, multi-stakeholder consultations and knowledge sharing initiatives while working for the International Union for Conservation of Nature National (IUCN) and the IUCN National Committee in the Netherlands (IUCN-NL). This provided context and contacts with various actors with different views and interests in the introduction of REDD+ in Ghana. It also enabled us to observe interactions of actors with different roles, expertise and expectations with regard to the effectiveness and fairness REDD+. The 'web of actors' are described below following the observations of the first author as one of these actors (representing IUCN). A non-authoritative graphic representation is given in Figure 2.

**Figure 1.** Map with main research area in SW Ghana highlighted.

**Figure 2.** Graphic representation of the relations between various actors of the web of actors of Reducing Emissions from Deforestation and Forest Degradation in developing countries, and the role of Conservation, Sustainable Management of Forests and Enhancement of Forest Carbon Stocks' (REDD+) implementation of SW Ghana as observed by the first author. Actors are color-coded as follows: Red = spiders in the web; Yellow = REDD+ implementation working groups; Blue = global actor; Orange =-national actor; Green = local actor.

To understand the role of global-local intermediaries in Ghana, it was important for the authors to interact with actors working on REDD+ implementation at the national level. In Ghana, the Forestry Commission (FC) of the Ministry of Lands and Natural Resources (MLNR) functions as the focal point for REDD+. The Climate Change Unit functions as the secretariat of the multi-stakeholder REDD + Technical Working Group (NRTWG) that coordinates Ghana's overall REDD+ development process. Government, private sector, civil society and other organizations are represented in this group. The NRTWG gives advice and guidance on all REDD+ processes (FCPF, 2012). Parts of REDD+ are driven by the Natural Resources Advisory Council (ENRAC), the Natural Resources and Environmental Governance Technical Coordination Committee (NREG TCC) that is responsible for the Forest Investment Program and the coordination of the Forest Carbon Partnership Facility (FCPF). Actors in Ghana working with these institutions at the national level, as well as inter-governmental development organizations, conservation and more critical NGOs were interviewed as part of this research. These actors were directly involved in both the international REDD+ engagements and discussions, and the translation of these concepts to the local situation in Ghana. These actors were operating at the global-local nexus, but not in isolation. They worked directly or indirectly with actors at the local levels. Men and women were for example involved in the translation, interpretation and implementation of REDD+ at the local level in the Wassa-Amenfi-West District (WAW). As a case study, we interviewed such actors in WAW and asked them how they viewed, interacted and worked with national and international actors. Locally, members of local settler communities, the District Assembly, traditional authorities, the Forestry Services Division of the Forestry Commission, members of the Community Resource Management Areas (CREMA), companies, local entrepreneurs, representatives of youth and women groups and members of the local multi-stakeholder REDD+ working group were interviewed.

Data collection consisted of three stages. A total of 35 Interviews took place. First, 20 interviews were conducted with local residents of three villages in the WAW where stakeholder consultations on REDD+ had taken place. These interviews included several group interviews with local smallholder farmers with informal land tenure arrangements. Most of them had been exposed to REDD+ information and awareness raising activities. A few interviewees were local chiefs and two farmers owned larger plots of land. In addition to group and individual interviews in the villages, group interviews were also conducted with one of the CREMAs in WAW and the REDD+ multi-stakeholder platform in the District. The District head and a professional working for the District Assembly were also interviewed. The villages had in the past been involved with various community-based conservation initiatives, such as when Globally Significant Biodiversity Areas were identified. In these initiatives, as well as during the introduction of REDD+, the Forest Services Department of the Forestry Commission of the Government of Ghana played an important liaising role between global NGOs, local communities, and the District Assembly. Second, 15 interviews and various additional informal conversations took place with NGOs, government agencies, and companies that were involved in the introduction or awareness raising REDD+ in SW Ghana. Third, various REDD+ preparatory meetings, dialogues and discussions were observed, and meeting reports studied.

Transcripts of interviews and collected documents and field notes were first coded for categories such as prevailing ideas about forest dependency; indications of actors that presented information and concerns of local actors at the national and global levels; indications of practices and activities that helped local actors in socio-economic development and indications of uptake or change of elements of REDD+ introduced. In a second round of coding, data was analyzed for the three elements of practice, i.e., meaning, competences, and materiality. These initial rounds were then followed by several iterations to structure the results as they are presented below.

#### **3. Results**

#### *3.1. The Introduction of REDD+ Practice in SW Ghana*

#### 3.1.1. Meaning: Ideas and Discourse

The global meaning of REDD+ is underpinned by the idea that deforestation is a major contributor to climate change because of the CO2 emissions it causes. Governments, NGOs, Inter-Governmental Organizations (IGOs), companies, and research organizations that helped develop REDD+ at the global level believed that that it was imperative that tropical forests should be protected for climate but also for the biodiversity they harbored and for the local livelihoods they supported [4,5,27]. These ideas are supported by global discourses that assume that addressing deforestation will be an effective, fast, and financially efficient way to help global CO2 emissions to peak and subsequently fall [43,44].

In Ghana, organizations that were involved in early REDD+ introductory activities included the Nature Conservation Research Center (NCRC), Forest Trends, Rainforest Alliance, Katoomba Group, Conservation Alliance, SNV, Tropenbos Ghana, University of Ghana, World Bank Ghana, Price Waterhouse Coopers (PWC), Permean Global, the Rockefeller Foundation, and Form International. Government agencies such as the Ghana Forestry Commission (FC) that hosted the National REDD+ Secretariat and led the development of Ghana's National REDD+ Strategy played an important role [38,45]. These actors had in common that they worked across the local and global levels of REDD+ development. In particular the non-governmental actors were active in introducing the idea of REDD+ in SW Ghana. Where introductions took place, these organizations generally focused on people and organizations in locations where they were already involved in community-based conservation and sustainable agro-forestry practices.

#### 3.1.2. Materiality: Technology and Resources

Globally, the introduction of REDD+ was closely connected to the emergence and development of remote sensing equipment to monitor forest cover change in ever-greater detail, and information technology (IT) to quantify these changes in terms of loss of carbon to the atmosphere. The emergence of carbon markets after the Kyoto Protocol that went into effect in 2005 made it possible to credit reduced deforestation and sell the emission reduction permits. Advances in development of use of increasingly complex computer programs also made the presentation of data in maps possible. These made up the materiality of REDD+. The maps of forest cover change and fluxes in carbon stocks became powerful tools to promote the idea and meaning behind REDD+. The combining of these elements clearly strengthened the unity of REDD+ as a practice.

In Ghana, like in many countries, access to forest resources has been unequally distributed. This instilled widespread concern about the access to REDD+ benefits for different groups. REDD+ benefit sharing therefore became a key topic of debate when local actors were engaged in REDD+ preparations. In Ghana, it was mostly official institutions and prominent NGOs that used and adapted remote sensing technology for local biomass monitoring, Carbon stock mapping and the development of national and sub-national MRV systems [46]. Additionally, the economic aspect of REDD+ materiality did not materialize quickly at the local levels. Challenges around land and resource ownership rights and unequal access of communities to forest benefits meant that the translation of Carbon finance into concrete REDD+ projects with benefit sharing was slow. In the meantime, most public REDD+ funds flowed to large, government-led initiatives that deployed relatively few resources piloting and testing [7].

#### 3.1.3. Competences: Standards and Procedures

Global policy development of REDD+ introduced standards and procedures for establishing deforestation rates and baselines for measuring and accounting carbon. New systems for the monitoring, verification, and reporting (MRV) of reduced CO2 emissions through reduced deforestation, degradation and through forest restoration, made it possible to translate these data into carbon credits. In SW Ghana, competencies to use Geographic Information Systems (GIS), remote sensing, carbon and biomass mapping were mostly introduced and applied by global-local intermediaries such as IUCN, Forest Trends, Katoomba Group and Nature Conservation Research Centre (NCRC) [47]. They worked with the GFC and the Centre for Remote Sensing and Geographic Information Services of the University of Ghana. Local actors were only in some cases involved for on-the-ground verification and to achieve greater accuracy and inclusion [48]. Maps showing trends of deforestation and established baselines were important to prioritize REDD+ action, and to have a reference against which it could be established that REDD+ action would be additional [49]. The mastering of these skills and the resulting calculations and mappings helped organizations in Ghana to promote the argumentation behind REDD+ (meaning) but also made it possible to attract REDD+ finance from international donors. It illustrates that the possibilities for the elements of ideas, materiality, and competences to be linked, contributed to the creation of unity of REDD+ practice.

#### 3.1.4. The Role of Global-Local Intermediary Organziations

The introduction of REDD+ was dominated by organizations that served as intermediary between global and local levels of forest governance. These intermediaries were often NGOs that prioritized the introduction of REDD+ information and ideas to local actors. Technologies, resources, standards, and procedures were however more likely to be overlooked at lower levels of governance. An additional challenge for REDD+ was that REDD+ intermediaries initially did not connect to actors and spaces that represented the larger industrial complex behind deforestation. The dominating idea was that local communities had to be in charge of, and benefit from, REDD+, even when REDD+ introductions were not considered to be truly bottom-up. REDD+ intermediaries had an advantage over local actors because they had access to global ideas and knowledge about REDD+, they took a lead in the disbursement of information, and they dominated the recruitment of participants locally. The following sections will detail how these discrepancies were (partially) addressed while REDD+ evolved.

#### *3.2. The Unpacking and Evolution of REDD+ in Southwest Ghana*

In this section, we assess how the 'global-local intermediaries' introduced REDD+ and how REDD+ evolved by integrating elements of existing local practices. The results show how different practices link to and follow up on each other, as well as the role that various actors play in this ongoing process of change. We identified several practices that were empirically linked to REDD+ and to REDD+ working groups by the actors in the field. The most important types of practices thus found were (1) community-based conservation; (2) tree planting; (3) agroforestry and sustainable agriculture; and (4) integrated landscape approaches. Each type of practice is analyzed below.

#### 3.2.1. REDD+ and Community-Based Conservation

#### Meaning

The ideas behind REDD+ resonated well with local meanings in SW Ghana. A total of 18 out of 20 women and men that we interviewed in WAW confirmed that forests and trees were important for their socio-economic development. Many also believed that forests helped buffer against heat, droughts, floods, and disturbed seasons. The idea that protecting forests contributed to a healthy environmental change resonated well with prevailing, local, holistic views of forests as their social, cultural, spiritual, and physical domain [50]. Most respondents related their personal experiences with climate variability to climate change and believed that they contributed to extreme weather events through their own involvement in tree-cutting. This belief and understanding was probably partly a result of the awareness raising and information activities organized as part of REDD+ preparations and earlier conservation practices but some respondents also referred to information about climate change that they received through the media. Local people seemed perceptive even if they were not in a position to fully assess how their contributions stood in comparison to other global forces driving climate change. A local farmer in WAW for example said:

"In the dry season the trees preserve water. They give us strength because the trees give us clean air ... they prevent high temperatures and heavy rains, we need to protect them so we can develop the area well. Streams will (then) never dry up. ... If we cut the forest, rainfall will reduce so in order to get rainfall and promote agriculture, we need to keep forests intact."

The global meaning of REDD+ was somewhat less compatible with prevailing assumptions among local actors about the role that forests played in their socio-economic development. One of the underpinning ideas of REDD+ social and environmental safeguards is that local communities will prioritize protecting natural forests because they rely on these forests for food, medicinal plants, proteins, medicinal herbs, water, and fuel. Our analysis of local responses to REDD+ however suggests that there is a mismatch between the assumptions behind REDD+ and those of local communities. As we will see in the following sections, they are often inclined to give priority to development opportunities "outside" the forest.

#### Materiality

REDD+ initiatives in the study period saw little trickling down of forest and carbon monitoring technologies to the lower levels. Additionally, the introduction of REDD+ carbon payments hardly materialized locally while carbon finance from donors and multilateral initiatives mostly flowed to national programs and agencies. Yet, the prospect of possible economic benefits from REDD+ projects created opportunities for people to think about the role that the marked could have in the management of their forest landscapes. It put the spotlight on prevailing local issues of inequality and the experience of people that they had mostly been barred from protected forests. Local interviewees therefore, when asked about opportunities for economic development from forests, often mentioned

options "outside" the forest, such as agriculture, agro-forestry development, and tree-planting. In the words of a professional working for a local conservation NGO:

"People might rather not have forests because they are disillusioned with the way in which things used to be managed."

The Secretary of a Community Resource Management Area in WAW summed up that REDD+ sounded great but would not be able to materialize for some time due to persisting tenure insecurity and inadequate benefit sharing arrangements:

"Because of REDD+ we know what we can achieve one day."

The introduction of REDD+ forced actors engaged in community-based conservation to consider the challenges that local communities faced in receiving formal recognition of informal land tenure and ownership arrangements. Statutory and traditional land tenure arrangements in Ghana overlapped and often conflicted. They were virtually nowhere gazetted, and informal tenure arrangements of settler farmers in SW Ghana did not have any formal legal protection at all [13,31]. Even though the introduction of REDD+ benefits did not materialize, the introduction of REDD+ ideas and discourses helped create renewed urgency for the issues of tenure and rights.

#### Competences

REDD+ introduced competencies to use forest and carbon monitoring technology as well as standards and procedures for establishing deforestation rates and baselines, and for the measuring and accounting of forest carbon. As these technologies were mostly used by national-level agencies, these competencies did not play a direct role in the reconfiguration of REDD+ at the local levels. The focus on equity and the social and legal dynamics of REDD+ however did trigger a demand for an altogether different set of competencies. Standards and procedures had to be developed to ensure social fairness and avoid further exclusion and marginalization of local communities. Here, something interesting happened. The formulation, adoption and prioritization of environmental and social safeguards as part of international REDD+ program and policy development was a direct response to concerns voiced by local actors and their global-local intermediaries [27]. Support for a social and environmental focus was particularly urgent in SW Ghana where formal and traditional legal systems contradicted and where local tenure was considered insufficiently protected by law [13]. The very early REDD+ piloting that global REDD+ had made possible, helped change global REDD+ in such a way, that it created a demand for standards and procedures for the strengthening and safeguarding of social and environmental benefits [27]. In many cases, local actors were already acquainted with these kinds of standards, because they had been developed as part of earlier conservation practices.

The introduction of REDD+ created fresh "demand" for competencies and procedures to ensure social and environmental safeguards, and the deployment of these local competencies contributed to the development of REDD+ in new directions. In SW Ghana for example, existing community-based conservation practices put a lot of emphasis on the importance that forests played in the social, economic, and cultural lives of local people. Participatory research in WAW showed how cash and non-cash use of forest products helped forest-dependent communities to move out of poverty in the long term [51]. REDD+ consultations and interviews with local men and women in the same areas however suggested that socio-economic development could not come from protecting forests alone. Many respondents believed that the future lay in raising standards of agriculture and tree-planting as well as income generation activities "outside" the forest such as tourism and jobs in the city. In the words of local women living near the Globally Significant Biodiversity Area (GSBA) in WAW:

"We need knowledge for planting crops, for agricultural improvement, production and output of agriculture of our livelihoods for example not to burn our fields, we need the technological know how to improve that."

"I will tell that before we benefited from the forests and now we don't get benefits. So maybe we can get additional support from the government. It is best is for someone to come and teach us techniques to multiply agriculture production."

In short, the outcomes of REDD+ stakeholder consultations in areas where global-local intermediaries had worked on community-based conservation generally pointed at a local need for standards and procedures to redress unequal forest and land access rights, and for competencies to create economic value from the forest areas. The following sections will show how different practices of REDD+ integrated responses to these local demands for particular competencies by building on established practices such as tree planting and sustainable agro-forestry.

#### 3.2.2. REDD+ and Tree Planting Practices

#### Meaning

Local meanings behind tree planting resonated well with REDD+ meaning. People felt that bringing back trees on deforested and degraded lands were good for themselves and for nature. It also helped them buffer against the impacts of extreme weather. For local people, it was easier to see that they could reap benefits from planting trees than from not cutting naturally growing trees that someone else might cut anyway. In areas where deforestation took place and where individual land rights were not well-defined, it would be difficult to attribute reduced deforestation to individual action and individual people in order to distribute benefits as a reward [52]. The perception was also different because local people could claim rights over planted trees but not over naturally growing trees. Cutting trees was a way for farmers to affirm their right over land tenure and cocoa farmers were afraid to let trees mature in their cocoa farms because licensed loggers were allowed to harvest naturally growing trees, causing damage to crops in the process [28,31]. Reserves or forests nominally owned by traditional authorities were generally treated as common pool resources, resulting in widespread illegal deforestation and forest degradation [31].

#### Materiality

In Ghana, tree planting could more easily be translated into material gains, because each person that planted and took care of a certain number of trees, would be able to receive incentives or compensation, and could also use other products of the tree. As local farmers said:

"Local people don't want to invest in not using forest for 20 years. If they plant trees such as rubber, they will at least receive income at some point."

A local timber company explained:

"For example, in this REDD+ area, we can give them (local smallholders) incentives to plant trees and then maybe go further, to help them establish small scale saw mills where they process the wood for charcoal, preserve the wood for their daily lives, timber for their domestic use, and then to sell to the local market."

As REDD+ was negotiated at the global levels, many campaign NGOs had criticized this component of the scope of REDD+ out of concern that large scale plantation companies could access REDD+ finance [27]. The above quotes of local actors illustrate that the options for "forest restoration" could indeed be used by plantation companies, but that it also created opportunities to address the needs of local actors. This potential for tree planting and plantation management was strongly represented in early REDD+ proposals and plans in Ghana. The planting of community woodlots for charcoal and fuelwood production, for example, was a main focus area for the REDD+ Readiness Preparation Proposal of the Government of Ghana. The World Bank Forest Investment Program (FIP) and the Government of Ghana also looked into possibilities to support forest restoration and plantation management under REDD+ in Ghana [53]. The FIP was set up to bridge early phases of

REDD+ readiness preparations with future private sector investments [27]. In Ghana, it financed a timber plantation initiative by the Dutch company Form International, a company that had already independently produced Voluntary Carbon Standard (VCS)-certified REDD+ Carbon Credits in their inclusive and sustainable plantation program in Ghana.

#### Competences

Tree planting practices in Ghana also included important competencies for the clarification of rights that combined well with REDD+. The rights that local people held over planted trees were laid down in the constitution of Ghana [54,55]. In many areas in SW Ghana, local farmers did not know how to assert this right or were not even aware of it, but global-local intermediaries introduced standards and procedures to create awareness and recognition of these rights. Already before REDD+ introductions started earnestly, global-local intermediaries such as IUCN had worked with local actors on the clarification and formalization of this right. They had raised awareness among local bureaucrats and smallholders about existing regulations on the ownership of planted trees, and had developed tree registration procedures with local authorities [56]. The certainty over rights, the prospect of future returns in timber, and possible carbon payments triggered small NGOs and land owners to start small plantations, even though national and local REDD+ monitoring systems were not yet in place. One local farmer in WAW for example was not part of a formal REDD+ project and would alone never be able to produce verified carbon credits. The costs of most certification were so high that it was generally accepted that hundreds of smallholders would have to be aggregated to have the scale needed to carry such investments and generate enough carbon credits to be able to recover upfront investments. He had set up his own mini plantation anyway:

"I heard a lot on the radio about the money that we could receive for trees. When I decided to have my own little tree plantation early on, many people in my village thought I was crazy, but now that the trees are maturing, they see that I am already reaping benefits, for the community and for my family, regardless of any payments I might receive from carbon or not."

Additionally, larger-scale plantation initiatives and policies included provisions to protect the rights of local people and make the sharing of benefits possible. One such provision was the Modified Taungya System (MTS), a forestry system that involved inter-planting trees with agricultural crops. Respondents regularly suggested that the competences developed under MTS served as an example how benefit sharing could be possible under REDD+.

#### 3.2.3. REDD+, Agro-Forestry, and Sustainable Agriculture Practices

#### Meaning

As the development of REDD+ progressed in Ghana, its focus gradually shifted to sustainable agroforestry systems and tree planting. Smallholder agro-forestry cocoa production had been a main contributor to the gradual degradation and destruction of biodiversity-rich and high-carbon rainforest in SW Ghana [57]. With a main focus on carbon and avoiding deforestation, the global meaning of REDD+ nevertheless converged well with the meaning of sustainable agro-forestry practices. REDD+ proponents developed programs that focused on increasing per-hectare cocoa yields, improving input management and certified crops. These approaches shared ideas with REDD+ on reducing deforestation, reducing CO2 emissions, and creating outcomes for local communities.

#### Materiality

Creating economic return from sustainable agro-forestry presented REDD+ with an alternative for carbon finance. The one-million USD 'Climate Cocoa Partnership for REDD+ Preparation' of Rainforest Alliance International and trading company Olam International in the Juabeso-Bia District in SW Ghana for example made that shift. The project focused on generating increased income from sustainable cocoa production and biodiversity protection through bringing back shade trees and forest protection. A Landscape Management Board consisting of community representatives was set up. The project achieved positive outcomes for communities and biodiversity, but it did not produce verified CO2 emissions. In the end, the project was not a success because it cost more than the premium sale of sustainable cocoa could deliver. There was also some misunderstanding between the different parties whether REDD+ credits would be generated. The project however showed that the materiality of reducing Carbon emissions and deforestation in (agro-) forestry practices combined well with the idea of REDD+ to address social, environmental and climate objectives simultaneously. This new combination of elements developed also internationally under the term Climate Smart Agriculture (CSA). FAO [58] defines CSA as "an agriculture that sustainably increases productivity, resilience (adaptation), reduces/removes greenhouse gases (mitigation) while enhancing the achievement".

#### Competences

Sustainable agriculture and agro-forestry practices presented REDD+ with standards and procedures that helped farmers create more return from their fields and competencies to improve their rights. In established sustainable agro-forestry practices, smallholder cocoa farmers were trained in competencies to apply fertilizers and other farm inputs. These meant that these famers could increase their income and have more stable harvests. The planting of trees for shade was also encouraged, and this contributed to the halting of deforestation and the protection of biodiversity. Additionally, it could, if needed, be translated in reduced CO2 emissions and increased carbon sequestration. Sustainable cocoa agro-forestry presented REDD+ with standards and procedures to achieve social and environmental REDD+ objectives such as increased income, gender inclusiveness, improved input management and soil treatment. It fitted the need for REDD+ to diversify income streams.

#### 3.2.4. Integrated Landscape Approaches to REDD+ in SW Ghana

#### Meaning

After 2010 and 2013, global-local REDD+ intermediaries in Ghana collaborated to develop landscape approaches to REDD+. These approaches integrated ideas about conservation, inclusive governance, soil and water conservation, sustainable agricultural intensification and the re-planting of shade trees as a means to reduce CO2 emissions from the landscape and achieve additional social and environmental benefits. In Ghana, a civil society initiative of NGOs, traditional leaders, cocoa trading companies, farmer cooperatives, insurance companies, consultancy firms, certifying agencies and scientists worked with the country's REDD+ Secretariat and Cocoa Board to develop Ghana's 'Emission Reductions Program for the Cocoa Forest Mosaic Landscape' [59,60]. Leading actors in this movement were also actively involved in global trends to integrate elements of REDD+, sustainable agro-forestry, and Climate Smart Agriculture in landscape and jurisdictional approaches [61]. During that time, the landscape approach became a dominant REDD+ discourse globally. It added the idea of territoriality, which according to some actors is problematic because landscape boundaries do not correlate to administrative-political boundaries [12]. The landscape approach nevertheless provided a useful concept that helped in the linking of the elements of materiality and competencies and thus in reproducing REDD+ towards new directions.

#### Materiality

With the broadening of the scope of REDD+ landscape approaches, the element of carbon monitoring and accounting found new connection with other elements of REDD+ practice. Whereas project-based REDD+ did not "land" due to faltering carbon markets, the scale of large multilateral and donor REDD+ finance arrangements fitted well with landscape and jurisdictional approaches to REDD+. In Ghana, the Cocoa Forest Landscape Mosaic program successfully applied for funding from the World Bank Carbon Fund. On-the-ground preparations for REDD+ in the cocoa sector were a continuous process where focus gradually shifted from individual projects and carbon finance to creating economic value from climate smart agriculture in a landscape approach. In the words of a professional working on the program:

"Finally (after years of REDD+ preparations) we sat down with the community and looked at the landscape. And we decided there was no play for REDD+ because carbon finance could not come from reduced deforestation alone. We therefore decided to go for Climate Smart Cocoa."

In 2014, this cocoa landscape program managed to secure the commitment of at least 40 million USD in funding from the World Bank's Carbon Fund to demonstrate REDD+ on a landscape scale [62]. In the next section, we further explain the success of this particular program and the finance it was able to attract.

#### Competences

The Cocoa Forest Landscape Mosaic program in Ghana was built on a thorough analysis of climate vulnerability in the cocoa sector and the suitability of soils for cocoa production across the landscape. The introduction of procedures for better cocoa planting techniques, farm input use and the planting of shade trees would be crucial to maintain soil productivity and lift farmers out of poverty. Where soils were sub-optimal and climate models were unfavorable, cocoa would have to be taken out of production altogether. However, another set of procedures focused on social inclusion and the clarification of rights: the procedures around the rollout of the Community Resource Management Area (CREMA) model. The CREMA standards and procedures proved a good match with the materiality and meaning of REDD+ landscape approaches.

CREMA became a central component of the development of the Government of Ghana's REDD+ policies and they contributed to the enthusiasm of international donors in the Cocoa Forest Landscape Mosaic program [45,46]. CREMA represented a set of standards and procedures for the devolution of forest resource rights to local communities that had legal backing in Ghana's National Forest and Wildlife Policy [63,64]. NGOs, community groups, and government organizations had years of experience with CREMA as part of community-based conservation and wildlife management [28,65]. In the early 2010s, it became a central element of many REDD+ pilots. In some cases, such as in the Juabeso-Bia project, similar governance structures were designed, modelled on CREMA [66]. For the Cocoa Forest Landscape Mosaic program, CREMA could function as a mini landscape within which the merits of a landscape approach could be combined with the rigor of REDD+ monitoring and verification of reduced deforestation. The interventions and monitoring could subsequently be replicated in new CREMAs and eventually cover the larger landscape. In this way, CREMA was made instrumental in the development of standards and procedures needed to implement REDD+ in a landscape approach.

#### **4. Discussion**

The introduction of REDD+ in SW Ghana was not a clear-cut, linear process in which local actors simply implemented global ideas and aspirations. REDD+ was initiated by public, private and non-profit actors that worked across global and local governance scales. They chose sites and got involved in the introduction of REDD+ with local actors with whom they were already involved with in established community-conservation practices. Our practice-based analysis shows how these local actors and global-local intermediaries renegotiated REDD+ and helped shape the evolution of REDD+ at the local, and in return, at the global level. The CREMA model for example, with its procedures for multi-stakeholder collaboration, community consensus building, decision-making, and the formalization of rights, was integrated in various REDD+ initiatives and served to illustrate to international audiences how REDD+ could work at the implementation level [60,63,67].

Applying and refining the framework for studying practices of Shove et al. [33] helped us understand how actors involved in the local introduction of REDD+ combined REDD+ meaning, materiality, and competences with elements of local practices. First, REDD+ ideas about the importance to protect forests resonated well with local ideas and beliefs about the important role that forests played in livelihoods and cultural development of local actors. These ideas prevailed in established practices and, as a result, these practices provided entry points for the introduction of REDD+. Prevailing ideas about the importance of ecologically healthy forests for example helped people realize the assertion of REDD+ that forests also play important roles in reducing the impacts of extreme weather events and climate change. The introduction of REDD+ in Ghana however also exposed divergence between global and local understanding of the role that forests can play in the development of community benefits. Specifically, global ideas on REDD+ underplayed the trade-offs between social and environmental interests in local practice.

The introduction of REDD+ and concerns over social issues created a demand for competences for the organizing of multi-stakeholder engagement, participatory decision-making making, the formalization of rights, and the development of benefit distribution mechanisms. Global negotiations over REDD+ policies and programs started to prioritize social, environmental and governance issues over technical issues of monitoring and this led to the adoption of social and environmental safeguards at COP15 [27]. This shift in thinking at the global level was at least partly a result of early REDD+ introductions and testing at the local levels, such as in SW Ghana [27]. The reshaping of REDD+ on the ground happened in interaction with developments at the global levels, and it was the result of an enactment of REDD+ in the context of established practices in which global-local and local actors were jointly involved. Therefore, as carbon finance moved to the background, established practices such as the creation of economic value tree planting, sustainable agro-forestry, and CSA complemented or replaced this material element of REDD+ practice.

Third, when looking at the material element of REDD+ it transpired that local people welcomed the introduction of carbon finance or other forms of economic value for forests and that monitoring was less relevant in the absence of well-functioning global carbon markets. However, they expressed serious doubts that REDD+ benefits would be effectively and fairly shared among local communities. Their concern was based on past experiences of being denied their rights over natural resources. Neither local nor global-local intermediary actors that were involved in the introduction of REDD+ in SW Ghana considered this the end of REDD+. Instead, at REDD+ introductory and preparatory meetings, people that were involved in the introduction of REDD+ often stressed that REDD+ created opportunities to put concerns over rights and inclusive governance on national and local political agendas.

Our case of in SW Ghana illustrates that global-local intermediaries played an important role in the development of REDD+ pilots. Local community members, local government agencies, and local private sector actors depend on these global-local intermediary actors to provide ideas and information about new technologies, instructions, and skills. Many authors confirm that global policy processes are a source for ideas and discourses [11,12]. What our research shows is that these global processes equally depend on the materiality and competences that form the elements of local practice. The global-local intermediaries that led introductions of REDD+ at the local levels did not only shape the introduction of REDD+ locally, but also influenced the development of REDD+ at the global levels, together with actors involved in the introduction of REDD+ in other countries [27]. As early as 2008, and through 2012, lessons from REDD+ introductions, pilots, and dialogues in SW Ghana reached global REDD+ debates [28,31,68–71]. The mobility of REDD+ to move from the local to the international levels continues to be relevant today, as global negotiations over REDD+ policies and programs are ongoing along with critical debate about the effectiveness and inclusiveness of REDD+ [24,25].

This channeling of ideas, experiences, competences, information and alternatives up and down the global-local nexus make the global-local intermediaries look like "spiders" in a web. They take the lead in initiating REDD+ locally; in choosing sites for REDD+ introduction, and in the recruitment of local actors. The spiders dominate the process, manage to maintain their place in the web of networks and relations along which elements of REDD+ practices travel from the global to new places such as in SW Ghana. These actors enjoy a position of comparative advantage, having "access" both to local and global ideas, competencies and materialities. Local actors depended on the decisions and resources of these global-local intermediaries, who had comparative advantage over them. Global-local actors in turn also depend on local actors. It was in the joint enactment of established practices with local actors that they could renegotiate and reshape REDD+. By refining the meaning of REDD+ and by absorbing and combining materiality and competences from established practices, global-local actors showed international REDD+ policymakers how REDD+ could work on the ground.

#### **5. Conclusions**

Global-local and local actors in SW Ghana continue to develop local interpretations of REDD+ and attract international resources for their initiatives, even while REDD+ is declared "dead" by some [24]. This shows the importance of understanding of REDD+ as part of the global-local nexus of governance [7] By considering the connections between forest governance levels and the key role that both local and local-global actors played, REDD+ is shown to evolve in response to local needs, rather than simply failing to live up to the first ideas that were articulated at the global policy level. While none of the resulting REDD+ initiatives in SW Ghana are 'schoolbook' examples of reducing deforestation and forest degradation through carbon finance based on verified emissions reductions, they are actively changing local practice of tropical forest conservation. Moreover, even when it is too early to tell if those REDD+ projects or programs managed to create a breakthrough in entrenched issues of poverty, social exclusion and a lack of rights, we do see increased attention for these issues, also on a global level.

Ambiguous results do not mean that REDD+ merely reinforced a status quo or contributed to "business as usual". Confronted with REDD+, local actors used their situated agency to integrate the meaning, competences, and materiality of global REDD+ in locally established practices of conservation, tree planting, agro-forestry, and integrated landscape management. This enabled them to absorb elements of these practices in REDD+, as they re-negotiated and shaped REDD+ in SW Ghana. In turn, the intermediary global-local actors that initiated and led these REDD+ introductions acted like "spiders" in the web between the global and local levels. They channeled ideas, information and resources from the global to the local levels, chose sites for introduction of REDD+ and recruited local actors for implementation.

The case illustrates how global policy on REDD+ can reach local levels of governance but also what the challenges are. Global-local actors worked from an advantageous situation because they had access to REDD+ ideas and information, techniques and technologies and the necessary competencies to translate it to the local situation. They did, however, depend on the extent to which local actors could make REDD+ work on the ground. Local people had agency that was situated in practices that they are already involved in, and they were the ones to draw on elements which did or did not combine well with elements of REDD+ practice. One of the most important lessons from the example in SW Ghana is that global REDD+ design has underestimated the trade-offs between REDD+ effectiveness to mitigate climate change, to be fair and inclusive, and to halt deforestation and biodiversity loss. When REDD+ was enacted on the ground, this often translated to interesting initiatives with local communities that could combine goals of social development, conservation, and climate mitigation. Whether these local initiatives can in turn provide the basis for a more global answer to political and economic drivers of deforestation is still an open question.

**Author Contributions:** Writing—original draft, J.W.d.B.; Writing—review & editing, B.A. and J.B. B.A. and J.B. contributed to the conceptualization, writing, review, editing and supervision of this research and the writing of the paper.

**Funding:** This research received no external funding.

**Acknowledgments:** We would like to thank three anonymous reviewers for their useful input. We also would like to thank IUCN and IUCN NL to make some of the data collection possible.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **"Embodied Deforestation" as a New EU Policy Debate to Tackle Tropical Forest Loss: Assessing Implications for REDD+ Performance**

#### **Janice Weatherley-Singh 1,2,\* and Aarti Gupta <sup>1</sup>**


**\*** Correspondence: jweatherleysingh@wcs.org; Tel.: +32-(0)-2739-3008

Received: 12 September 2018; Accepted: 27 November 2018; Published: 1 December 2018

**Abstract:** The need to tackle international drivers of deforestation has long been acknowledged; but remains little addressed via policy measures. In the European Union (EU), a new policy debate is emerging around the concept of "embodied deforestation", which targets EU agricultural commodity imports as drivers of deforestation. The notion views deforestation as an externality generated by EU imports associated with tropical deforestation. Our article examines whether this concept represents a shift in tackling international-level drivers of tropical deforestation within EU policy. We also examine, from a networked governance perspective, whether this new debate fuels further fragmentation or rather a move towards a more integrated approach to combating tropical forest loss within EU policy, and what the implications are for other initiatives, such as the climate change related "reducing emissions from deforestation and forest degradation" (REDD+). Our analysis draws on an extensive analysis of EU policy documents and semi-structured interviews with stakeholders and EU decision-makers. We find that, despite growing debate around the concept of embodied deforestation, policy measures necessary to reduce the impact of EU consumption of agricultural commodities associated with tropical deforestation have not yet been developed. We conclude that "embodied deforestation" remains more an idea than reality within EU policy to date, with the burden of responsibility for addressing international deforestation drivers still largely remaining on developing countries. There is still potential, however, for this debate to lead to a more integrated approach to tackling tropical deforestation within EU policy, if it comes to be seen, together with REDD+, as one of a number of linked approaches to EU efforts to combat deforestation.

**Keywords:** REDD+; European Union; forest policy; deforestation drivers; tropical forests

#### **1. Introduction**

The need to tackle drivers of deforestation and forest degradation operating at the international level has long been recognized [1], but has thus far largely been overlooked by national environmental policy-makers [2]. However, the issue is now beginning to be actively taken up in policy debates, including within the European Union (EU), as global trade in agricultural products becomes a more prominent driver of deforestation [3]. More than half of all deforestation and forest degradation worldwide is now estimated to be due to the conversion of forestland for commercial agriculture to meet global demand for food, fuel, and fibre [4]. Global demand for commodities is a major driver of deforestation in Latin America and Asia in particular, and a significant component of this global demand originates from within the EU [5]. How developed countries contribute to deforestation in tropical countries, particularly through importing agricultural commodities, is thus garnering increased attention, including within the EU.

A study produced for the European Commission in 2013 [6] (p.iv) estimated, for example, that the EU is responsible for 10% of global "embodied deforestation", i.e., deforestation as an externality in the production, trade or consumption of a good, commodity, or service. Recent declarations at the international level to reduce or end deforestation, such as the 2014 United Nations (UN) New York Declaration on Forests, and the Amsterdam Declaration by a number of European governments, as well as commitments by multinational companies to promote deforestation-free supply chains and certification schemes, such as the Roundtable on Sustainable Palm Oil (RSPO), are giving further impetus to policymakers seeking to tackle international consumption patterns that fuel tropical deforestation. Arguably, one of the first attempts by the EU to decrease the negative impacts of its consumption on tropical forests was through its 2009 Renewable Energy Directive, which required the introduction of sustainability criteria [7] to ensure biofuel production did not impact biodiverse primary forests [8]. The development of EU sustainability criteria was influenced by voluntary certification schemes, such as the RSPO [9], but both have been criticised for their limited ability to achieve sustainability in practice [10,11]. These efforts are now increasing, most recently, with adoption of a European Parliament resolution in 2017, which advocated restrictions on palm oil imports because of their negative environmental impacts, including adverse impacts on forests [12].

While the sustainability of commodity supply chains is increasingly the focus of both academic and policy scrutiny, our article analyses how these issues land within an increasingly fragmented global and EU-level forest governance architecture, with a wide array of initiatives spearheaded by both state and non-state actors to address tropical deforestation. Another highly visible such initiative is REDD+ (REDD+ stands for: Reducing emissions from deforestation and forest degradation in developing countries; and the role of conservation, sustainable management of forests and enhancement of forest carbon stocks in developing countries) under the United Nations Framework Convention on Climate Change (UNFCCC), which financially compensates developing countries for reducing greenhouse gas emissions associated with deforestation and forest degradation [13]. REDD+ has been understood and implemented in a variety of ways by a range of state and non-state actors [14,15]. These include project-level initiatives, aiming to deliver co-benefits for biodiversity and communities in specific local contexts [16]; performance-based carbon payments [17,18]; and, more recently, sustainable landscape approaches that often involve the private sector [19,20].

REDD+ projects and sub-national initiatives have been shown to be responsive to some drivers operating at local and national levels, but are largely unable to tackle drivers of tropical forest loss operating at the international level [21,22], particularly those linked to agricultural production [23]. In addition to its focus on local as opposed to national or international drivers, REDD+ has also been criticised because of negative socio-economic impacts, such as fuelling inequality through restrictions on access to forests and the commodification of carbon [24,25]. In general, commentators see REDD+ as having failed to live up to the initial high expectations following its introduction in international climate change policy discussions in 2005, especially in terms of finance flowing to developing countries to combat tropical deforestation [15]. Notwithstanding such criticisms, recent research points to a reconceptualization of REDD+ in which it is viewed as a conservation and development measure, with more realistic expectations with regard to its performance, rather than being seen as "the" answer to tropical deforestation [26–28]. Furthermore, there is also expectation in both policy circles and academic literature that newer REDD+ landscape approaches, particularly those that involve the agri-business sector, will be better able to tackle drivers of deforestation linked to the expansion of large-scale agricultural commodities for export [29], even if their impact on drivers operating at the global level is limited [30]. The new interest in tackling international drivers linked to consumption in developed countries (i.e., the concept of embodied deforestation), combined with newly emerging REDD+ initiatives involving the private sector, may thus signal a new and more integrated approach to combating tropical forest loss, one that prioritizes tackling drivers of deforestation and forest degradation operating at the international level.

Yet whether this potential is being realized remains little analysed. The concept of embodied deforestation has not yet been discussed in the scholarly literature, given its very recent emergence within the EU policy arena. How the concept is understood, and the extent to which it is gaining traction from policy-makers, requires further examination. It is also unclear what role, if any, is foreseen for REDD+ by those engaged in this new discussion. Will a focus on embodied deforestation marginalise REDD+ or give it fresh relevance? How is REDD+ performance, particularly in relation to tackling international drivers, related to this new approach, if at all? And, finally, will tackling embodied deforestation become one of many initiatives aiming to address tropical deforestation within the EU, contributing to further fragmentation, or will it facilitate a more integrated approach?

This article analyses these timely questions. We proceed as follows: Section 2 reviews the concepts of fragmented and networked forest governance architectures, as the conceptual lens through which we analyse the nature and implications of the new notion of embodied deforestation. Section 3 outlines our methodology and methods of data generation and analysis. Section 4 maps the existing fragmented approach to forest policy-making in the EU, and how the embodied deforestation debate is emerging within this policy context. Section 5 presents our analysis of whether and how this new notion is being translated into EU policy. We conclude with considering whether it represents a move towards a more fragmented or integrated forest policy, and with what implications for REDD+.

#### **2. From Fragmented to Networked Forest Governance: A Conceptual Shift**

Forest governance arrangements, both globally and within the EU, have long been considered fragmented, insofar as multiple sites of governance authority co-exist. In recent years, there has been a shift from analysing causes and consequences of fragmentation to analysing the networked nature of forest governance [13,31]. Networked forest governance involves bringing together a wide range of actors representing different interests from the private and public spheres [32], with coordination and negotiation between independent stakeholders interacting horizontally (rather than vertically) [33]. In this context, forest governance by formal political administrative structures is replaced by diffuse and complex networks that involve a wide range of actors, [34], including from the private sector [35]. Although networked forest governance scenarios often describe a retreating of the state to make way for other actors, recent studies highlight that the state still imposes limits on the involvement of other actors [33] and often continues to exert considerable influence over the policy process [36].

As production and consumption of forest and agricultural products becomes increasingly globalised, with complex supply chains, it becomes increasingly difficult for states to regulate and govern the sustainability of production, leading to an emergence of hybrid forms of state and private governance [37]. Within globalised commodity chains, highly complex horizontal and vertical chains and networks emerge, with interactions between actors at different points of transactions. As such, networked governance can be " ... conceived as a mosaic of both formal and informal networks, interconnecting production practices in the space of place to the space of flows of global trade" [38]. In terms of action to tackle deforestation drivers linked to global commodity chains, the intervention points in such networked contexts are thus often located in developed countries (such as the EU), and therefore are outside the sphere of policy action that can be taken by governments in developing countries. The globalisation of commodity chains is increasingly impacting the effectiveness of REDD+ implementation, which is being adopted in a fragmented manner at sub-national or national levels [39]. This has meant, in some cases, a displacement of deforestation from early to late adopters of REDD+ [40].

It has been observed, however, that governance of some global value chains is also becoming more coordinated between concerned public and private actors at different scales [41]. Few studies, however, have analysed diverse EU forest policies from a networked forest governance perspective, with little analysis of the implications of emerging and new policy discourses for future EU action on, inter alia, REDD+. Some forest governance studies have been conducted in relation to the EU Forest Law Enforcement Governance and Trade (FLEGT) [42], but these have tended to focus on its impacts in

partner (i.e., developing) countries (for example, [43,44]). Studies have also been undertaken with regard to regional European forest policy, which indicate a long-recognised lack of policy integration within the EU itself [45], and an increasing trend of influence by economic state interests over environmental interests [46]. It is therefore timely to analyse whether the emerging EU "embodied deforestation" debate represents an actual shift in responsibility to address deforestation drivers from developing countries (via supply-side measures) to developed countries (via demand-side measures), whether it signals a shift from a fragmented to a more integrated approach, and what the implications for REDD+ are.

#### **3. Methodology and Methods of Analysis**

This analysis relies on qualitative methodologies of document analysis and interviews. Given that there is little published secondary literature on the concept of embodied deforestation, our sources of data have been almost exclusively primary documents, as well as semi-structured interviews with those involved in this very new, emerging debate. Our analysis is thus based on detailed primary document analysis of 55 recent policy documents (generated during the period of January 2014 to December 2017) developed by or for the EU, which we identified as being of relevance to EU tropical forest policy (for a complete list of analysed documents, see Appendix A). We selected these according to the following procedure: With regard to European Parliament documents, a search was conducted on the Parliament website for documents containing a reference to the word "forest". Any documents found were then included in the analysis, if they concerned global or tropical forests (as opposed to only being concerned with European forests). No equivalent search function exists on the European Commission website, so documents were instead searched for on the webpages of relevant Directorate-Generals covering policy areas considered to be relevant to the issue of international tropical forest policy, namely: climate change, energy, trade, sustainable development, agriculture, foreign policy, environment, development, and the general future direction of EU strategy and budget. Again, documents were only included in the analysis if they referred to forests globally or outside of the EU (with the exception being a few overarching documents that set out the general direction of future EU policies, which were included for their relevance to all EU policy areas).

Once the documents had been selected, the document analysis consisted of mining these documents to distil answers to a number of questions, through close reading of each. The questions related to, inter alia, what new policy measures were being proposed/advocated; what drivers of deforestation or forest degradation were sought to be addressed, if any; whether the emphasis was on tackling consumption in the EU, or on support for measures in developing countries; whether REDD+ was mentioned and if so, how; and whether new policies or funding for REDD+ implementation were being proposed, also in conjunction with efforts to tackle international drivers.

In addition to the document analysis, 15 semi-structured interviews were conducted with stakeholders and policy-makers involved in discussions on embodied deforestation in the EU, during the period of February–August 2018. These were intended to verify the findings of the document analysis. Stakeholders were mainly selected based on their participation in two conferences organised by the European Commission on tropical deforestation in 2014 and 2017 [47] and because they were known to be actively involved in EU tropical forest policy debates. A stakeholder mapping was undertaken to select a range of participants working in different organisations. There was, however, a lower response rate from those working in EU institutions and national governments than from those working for NGOs and research institutes. A list of interviewees is provided in Appendix B, and includes five policy-makers from EU institutions and national governments, five representatives of environmental NGOs, four independent experts and researchers, and one staff member from a UN agency. The questions posed included: Whether stakeholders saw the embodied deforestation concept as useful and why; whether they saw a change in the balance between EU support for demand-side measures (to be implemented by the EU) versus supply-side measures (to be undertaken by developing countries) to tackle tropical deforestation; what outcomes they hoped for; and what role they envisaged for REDD+ in evolving EU policies targeting deforestation.

#### **4. Mapping the Fragmented Evolution of EU Tropical Forest Policy**

Since there is no single international treaty dealing with tropical deforestation, the issue has been dealt with globally in a fragmented manner, through a range of diverse policy instruments and agreements on related topics, such as biodiversity or climate change [48]. EU policy instruments dealing with tropical forests have generally developed in response to participation in such international United Nations (UN) conventions and agreements. Thus, tropical forests have been dealt with across a range of EU policy instruments, mirroring the situation at the global-level where forests are addressed within international agreements on biodiversity, trade in endangered species, climate change, etc. EU forest policy is thus spread across various EU-level institutions, such as Directorates-General for climate change, development cooperation, environment, and trade. Each of these have their own perspectives on forest issues, with a similar situation prevailing at the level of individual EU Member States as well, where different dimensions of forest policy, whether climate, trade or biodiversity related, are usually addressed by different national-level ministries [49].

External EU policy-making on forests has been largely aligned with the goals of specific, existing UN Conventions and international agreements. Thus, policies on biodiversity have sought to find a balance between biodiversity conservation and its sustainable economic use [50], with a stated aim of achieving and supporting sustainable forest management in this context [49]. The EU Biodiversity Strategy [51] sets out actions to implement the UN Convention on Biological Diversity (CBD) and promote the conservation and sustainable use of biodiversity, including forests. Similarly, EU wildlife trade legislation [52] was adopted to implement rules under the Convention on the International Trade in Endangered Species (CITES) in order to protect species (including timber species) threatened by international trade. The EU also participates in the UN International Tropical Timber Organisation (ITTO), which was established from a commercial viewpoint to reconcile sustainable forest management with expansion of the tropical timber trade [53], and the UN Forum on Forests (UNFF), which emphasises the need to combat deforestation through expansion of sustainable forest management (SFM) [5]. Arguably, the EU policy instrument that has gained the most traction, political attention, and funding in relation to tropical forests is its Forest Law, Enforcement, Governance, and Trade (FLEGT) initiative developed in the context of the 2002 World Summit on Sustainable Development [54]. FLEGT aims to reduce imports of illegally logged timber into the EU, including by supporting action in developing countries to strengthen sustainable forest management and improve governance [55].

The EU and its Member States are also parties to the UN Framework Convention on Climate Change (UNFCCC) and have together contributed about 30% of global finance for REDD+. A large proportion of this has, however, been in the form of bilateral aid from Germany and the UK [56]. The EU established its REDD facility in 2010, but has also donated to existing multilateral initiatives, such as the World Bank Forest Carbon Partnership Facility (FCPF) and UN-REDD programme [57], rather than choosing to become a major player in its own right within global REDD+ discussions. The introduction of REDD+ in the UNFCCC discussions in 2005 represented a significant shift in the objectives of international forest policies, including at the EU level [58], with the focus on reducing carbon emissions from deforestation and forest degradation [59]. This new focus on carbon as the main service provided by forests contrasted to previous approaches [60] that aimed to find a balance between biodiversity conservation and logging for timber. The development of REDD+ was highly influenced by a growing narrative on "payments for ecosystem services" (PES) [61], which explored new funding sources for the conservation of ecosystems based on their utilitarian socio-economic values [62,63]. The PES approach was also promoted by EU policy-makers who, for example, funded ecosystem valuation studies such as the 2009 "The Economics of Ecosystems and Biodiversity" report [64]. Although widely endorsed within policy, the PES approach has also drawn criticism from those who highlighted concerns about the commodification of forests for their carbon values, and the potentially negative ecological [65,66] and social [67,68] impacts of this shift in focus.

Concerns have also been expressed that REDD+ unfairly burdens developing countries, with some suggesting it has been used as a distraction to cover up the lack of action by developed countries to tackle their own greenhouse gas emissions [69]. In terms of external forest policy, EU policy processes have, like other international policymaking fora, historically focused more on supply-side rather than demand-side measures. They have done so by supporting actions taking place in partner (developing) countries, including through REDD+. The evolution of FLEGT, however, did signal a recognition that the EU needed both supply and demand-side measures to tackle illegal timber imports. Thus, the EU has been working to reduce the negative impacts of its tropical timber imports through engagement in the ITTO and CITES, and the adoption of its FLEGT Action Plan on combating illegal logging. It is now also beginning to develop policies to decrease the wider environmental impacts of consumption patterns, and its greenhouse gas emissions, and has adopted a Circular Economy Action Plan in 2015 to help transition to a sustainable, low carbon, resource efficient, and competitive economy [70].

Similarly, there has been an apparent shift in the implementation of development aid policies and programmes, with developed countries moving away from simply acting as donors to also committing to change their own policies. The adoption of the UN Sustainable Development Goals (SDGs) in 2015, for example, represented a significant change in focus from their predecessor, the Millennium Development Goals (MDGs), due to their universal applicability, with goals applying to both developed and developing countries. The European Commission is also considering whether to propose an EU Action Plan to Combat Tropical Deforestation, and in 2018, published a study [71] outlining the feasibility of policy options to tackle the drivers of tropical deforestation linked to EU imports of so-called "forest-risk" commodities, a new term gaining traction within this policy debate, which appears to refer to globally traded agricultural commodities that are associated with significant tropical deforestation. However, the European Commission has not yet decided which, if any, option to pursue. We turn next to whether this emerging discussion represents a real shift in EU policies dealing with tropical forests, by presenting the findings from our document analysis and interviews.

#### **5. Tackling EU Embodied Deforestation: A New Approach to Addressing Drivers?**

This section addresses whether and how the new debate on embodied deforestation is poised to address demand-side, international deforestation drivers, through specific adjustments to EU forest policy. It does so by analysing three aspects of this broad question: First, how the notion of embodied deforestation is conceptualized, and what new policy measures, if any, are advocated by policy-makers and stakeholders to tackle it; second, whether these debates and developments signal a real shift towards tackling deforestation drivers linked to EU consumption (i.e., demand side drivers); and third, whether these new debates and developments signal a move towards a less fragmented and more integrated approach to EU forest policy, and what role remains for REDD+ herein.

#### *5.1. Conceptualising Embodied Deforestation: Emerging Policy Narratives*

Our study of EU policy documents reveals a high level of EU support for implementing the SDGs (a central focus of 20 of the policy documents we analysed) and, to a lesser extent, climate action (a central focus of 10 documents), as illustrated in Table 1. EU policies across a range of subject areas are being shaped to reflect the aims of the SDGs. Those relating to climate, energy, and environment are all being tailored to tackle climate change, relating both to the EU's own emissions and to supporting actions in developing countries. In accordance with a realignment of its development policies to reflect the SDGs, the EU increasingly views its role less as a donor and more as a partner with developing countries. For example, the European Consensus document sets out the EU's development aid priorities, but also includes action on EU consumption patterns. The EU is also now encouraging other countries to address their own consumption patterns to become more sustainable, for example, through the EU-funded Switch Asia programme (Switch Asia is an EU funding programme to support sustainable production and consumption in Asia, http://www.switch-asia.eu/).



Somewhat surprisingly, however, tackling global deforestation and REDD+ are low priorities within wider policy discussions on climate and the SDGs, and the relevance of forests and land use, in particular, to the climate debate is largely missed. Instead, EU climate action appears to focus more on tackling EU greenhouse gas emissions and how this can lead to innovation and new jobs within the EU, rather than on reducing negative environmental impacts elsewhere, caused by EU consumption. The Eurostat report on monitoring EU action to achieving the SDGs demonstrates the strong shift towards supporting action within the EU rather than in developing countries. Although the need to halt global deforestation is mentioned [72] (p. 299), the proposed indicators to monitor progress only cover forests in the EU. A key finding of our analysis is that REDD+, in particular, has a very low profile within analysed policy documents. It is the dominant theme of only two EU policy documents, both of which are reports on EU activities undertaken in the past. Of the 55 policy documents we analysed, only 16 mention REDD+ in passing. Even reports and documents that are very supportive of policy measures to conserve tropical forests, such as the European Parliament report calling for EU action for sustainability (which highlights the need to address deforestation drivers and expresses support for afforestation for mitigation), do not mention REDD+ specifically.

Whilst some support was expressed for "embodied deforestation" as a conceptual approach in our stakeholder interviews, it is unclear whether sufficient momentum is behind it to signal a change of approach in external EU forest policy. Around half of those interviewed consider the "embodied deforestation" concept to signal a useful approach, whilst identifying a number of limitations (Interviews with: EU policy-maker in Brussels, 1 February 2018; NGO representative by skype, 16 March 2018; EU agency representative by skype, 12 April 2018; independent expert by skype, 18 April 2018; researcher in Brussels, 26 April 2018; NGO representative by skype, 18 May 2018; EU policy-maker in Brussels, 24 May 2018; independent expert by skype, 17 August 2018). Interviewees noted, for example, that various terms are being used to express similar concepts, which is confusing. For example, the French government refers to "imported deforestation" [73], Sweden refers to reducing its ecological footprint (Interview with UN agency staff member by skype, 2 February 2018), and the recent European Commission feasibility study refers to "embedded deforestation" [71] (p. 31). Second, it was observed that the term has to be explained each time it is used, which limits its usage to policy-makers rather than being broadly understandable to a wider public. Others noted that all imports of a particular commodity are treated, within this simplified concept, as having the same deforestation impacts, which limits its accuracy. One interviewee commented that alternative concepts, such as "sustainable supply chains", are more likely to be understood and supported by a wider audience (Interview with EU policy-maker in Brussels, 1 February 2018). Despite the limitations of the approach, we next turn to considering whether the EU is moving towards a greater emphasis on demand-side measures to be adopted by developed countries, thereby sharing responsibility for tackling deforestation drivers more evenly than previous initiatives focusing on supply-side actions by developing countries.

#### *5.2. Shifting Responsibility from South to North: Targeting Demand?*

In terms of whether there has been a shift in the balance of responsibility for tackling deforestation drivers from developing to developing countries, with a corresponding shift in the balance between demand and supply-side measures in EU external forest policy, there appears to be a shift in rhetoric, at least. This is demonstrated, for example, by the European Parliament report on palm oil [12], which advocates restricting imports of palm oil to the EU to prevent deforestation. It should be noted, however, that this is just a policy recommendation by the European Parliament, with no proposals from the Commission to implement it, partly because this report was met with strong political opposition from Indonesia and Malaysia, who want to protect their export markets, as widely reported in the media, for example, [74] More generally, in our analysis of policy documents, we identified various supply-side and demand-side measures being advocated within EU policies relevant to forests (see Table 2 for the list of suggested new policy measures), with a higher number of actions relating to the

demand-side. As in the case of palm oil, however, very few of the suggested demand-side measures are mentioned in official policy documents. Instead, they are mainly mentioned in the draft of a feasibility study undertaken by consultants for the European Commission. If enacted, however, these suggested new policy measures would help to address a number of international drivers of deforestation and forest degradation. These include illegal logging and the international trade in timber (building on existing EU efforts under FLEGT and the EUTR), agricultural conversion linked to the global export of commodities and imports, as well as international financial transfers associated with deforestation.

Interviewees expressed strong support, for example, for the EU Timber Regulation (EUTR) as a model for tackling demand, despite well documented implementation challenges [49,75]. At least six interviewees noted that the EU needs to take a regulatory approach to tackling deforestation, along the lines of an EUTR for agricultural commodities (Interviews with: EU policy-maker in Brussels, 1 February 2018; NGO representative by skype, 23 February 2018; and NGO representative by skype, 16 March 2018; an EU agency member by skype, 12 April 2018; independent expert by skype 15 April 2018; independent expert by skype, 17 August 2018). Two of the interviewees also stated that the EU should develop a policy instrument that provides transparency in financial reporting by companies on deforestation risks (Interviews with an EU policy-maker in Brussels, 1 February 2018; an NGO representative by skype 12 April 2018). The only recent EU legislative proposal to actually tackle a driver of tropical deforestation or forest degradation is a proposal from the European Commission to decrease competition for land between biofuels, agriculture, and forests through changes to the Renewable Energy Directive. This was developed in response to considerable criticism of the EU's biofuels policy, see, for example [76]. As revealed in one of our interviews, however, the proposal was watered down considerably in early 2018 before being adopted (Interview with EU policy advisor in Brussels, 28 June 2018). Therefore, despite the rhetoric and the growing number of new suggested policy measures to tackle embodied deforestation, there is very little in the way of actual new legal or policy proposals to address EU consumption impacts. As summed up by one of our interviewees, in general, the European Commission is mostly interested in developing voluntary rather than regulatory measures to tackle demand and "expects more from partner countries than they are willing to do themselves" (Interview with an NGO representative by skype, 23 February 2018).

Our interviews with stakeholders involved in the EU policy debate also reveal differing views as to whether the EU should go beyond achieving legality to also mandating sustainability standards that imports should meet or aim for "zero deforestation" targets for commodity imports. This emerging discussion is building on experiences in relation to FLEGT, which currently only covers legality but could, in theory, be expanded to include sustainability criteria [42,44]. Four of the interviewees (Interviews with an EU policy-maker in Brussels, 1 February 2018; an NGO representative by skype, 16 March 2018; an NGO representative by skype, 12 April 2018; independent expert by skype, 17 August 2018) consider that a new policy tool to deal with deforestation should be based on sustainability rather than legality standards, although they recognised the challenges this posed in terms of gaining acceptance from partner developing countries. They also expressed concern that the EU would be imposing its sustainability standards and governing beyond its borders, although it should be noted that none of our interviewees recognised that in some cases, partner countries may have higher standards in place than the EU would like to impose for example [7]. Differing views were also expressed as to whether partner countries have adequate legal frameworks in place through which to implement a sustainability approach, or whether these would need to be updated or developed first.


**2.**NewsuggestedpolicymeasurestocombattropicaldeforestationwithindocumentspublishedbyorforEU

#### *5.3. From Fragmentation to Integration in Networked EU Forest Governance: What Role for REDD+?*

Currently, EU external forest policy-making is highly fragmented, inconsistently applied, and contains gaps, or as one interviewee put it "efforts are all over the place so an overarching approach is needed" (Interview with EU policy-maker in Brussels, 1 February 2018). Another interviewee outlined the challenges as including contradictory policies proposed by different Commission Directorate-Generals; a lack of coherent planning as to how the EU will achieve its international forest policy commitments; and no standardised EU definition or understanding of key concepts, such as Sustainable Forest Management (SFM), REDD+, or even forests (Interview with NGO representative by skype, 18 May 2018). As pointed out by one interviewee, "the EU needs a deforestation policy and a REDD+ policy as it's not clear what it's trying to achieve in either sphere" and "the Member States all have divergent views" (Interview with an EU agency staff member by skype, 12 April 2018), thereby demonstrating a challenge of multilevel governance.

These views are reflected in a report from the EU REDD+ facility that outlines the REDD+ activities the EU is now supporting to address tropical deforestation, which includes several initiatives that are not generally thought of as being "REDD+", such as, for example, demand-side measures. This could either be interpreted as REDD+ evolving into a new conceptual approach or becoming increasingly irrelevant or side-lined in funding priorities. Those we interviewed gave differing views regarding the potential of REDD+ and its performance, also in terms of addressing deforestation drivers. One interviewee noted that "REDD+ has got so complex it's stuck and should go back to being an offsetting mechanism" (Interview with NGO representative by skype, 15 February 2018), whereas another was of the opposite view that "REDD+ is an umbrella which also encompasses sustainable supply chain approaches" (Interview with EU agency staff member by skype, 12 April 2018).

Looking ahead, as shown in Table 2, very little is actually suggested within documents published by or for EU policy-makers in relation to new REDD+ policies or funding. Divergent views were expressed in our interviews regarding the continuing role of the EU as a donor to REDD+. In theory, the EU target of 20% of budgetary spending for climate objectives (with a proposed increase to 25% in the European Commission's proposal for the new EU budget post-2020) [77] should make way for significant funding to be made available for REDD+. Large-scale European Commission funding has not been forthcoming in practice, however, although some individual EU countries, such as Germany and the UK, have provided substantial bilateral support for REDD+ [57]. Three interviewees (Interview with UN agency staff member by skype, 2 February 2018; NGO representative by skype, 15 February 2018; EU agency staff member by skype, 12 April 2018) were of the view that REDD+ finance from donors, such as the EU, has been helpful in creating enabling conditions to combat deforestation, even as the readiness process has created expectations within partner (developing) countries of continued finance (as also mentioned by Hein et al [78]). This would need to be followed through, for example, by funding REDD+ landscape/jurisdictional approaches, yet our analysis suggests that such increasing funding is not necessarily envisioned.

Skepticism was also expressed regarding the proper establishment of a functioning carbon market and the lack of interest in this by the private sector, who seem to be more attracted to the idea of sustainable supply chains (Interviews with UN staff member by skype, 2 February 2018; NGO representative by skype, 15 February 2018; NGO representative by skype, on 16 March 2018; and EU agency staff member by skype, 12 April 2018), although it was also noted that international aviation carbon offsets schemes could provide a new way forward, as also commented by Golub [79]. A Commission-funded study published in March 2018 outlines potential EU policy options for tackling tropical deforestation and includes a recommendation for the EU to support "jurisdictional REDD+ projects to promote sustainable and deforestation-free agriculture production" [80] (p. 75), which, if implemented, could provide a new way forward for EU support for REDD+ initiatives that combines jurisdictional REDD+ with supply chain approaches. Furthermore, EU development aid policies are increasingly looking to partner with the private sector to deliver policy goals through public-private-partnerships (PPPs), which may open new opportunities and lead to novel networked

forest governance arrangements in partner developing countries [81], but also risks watering down forest conservation objectives.

Several interviewees mentioned the need for the EU to contribute to global dialogues on REDD+ or other multilateral processes that can link demand and supply-side measures, and suggested that the SDGs could provide a helpful framework for this process (potential synergies between REDD+ and the SDGs have also been noted by others, such as [82]). However, it was noted that there is no obvious existing UN fora where such a dialogue could take place, with little confidence expressed in the UNFF (in line with previous criticism, for example, [83]). It was therefore suggested that in practice it may be more workable for the EU to encourage further dialogue on this issue (Interviews with UN staff member by skype, 2 February 2018, NGO representative by skype, 15 February 2018, NGO representative by skype, 23 February 2018 and EU agency staff member by skype, 12 April 2018). One interviewee noted that the UNFCCC is now turning its attention to decreasing greenhouse gas emissions linked to agriculture. This could be an area in which the EU could envisage playing a leading role, both by tackling emissions linked to its own domestic agriculture as well as through promoting sustainable supply chains and responsible consumption, partially through existing REDD+ approaches. Others note, however, that there is limited potential to build synergies between LULUCF accounting and REDD+ (see, for example, [84]) and from the perspective of one interviewee, the EU's own rules on LULUCF accounting lack credibility and environmental integrity (Interview with an EU agency staff member by skype, 12 April 2018).

Our analysis reveals, furthermore, several calls from different stakeholders for an EU Action Plan to Combat Tropical Deforestation (mentioned in five documents from the European Parliament and one from a research agency), and in two interviews (Interview with NGO representative by skype, 16 March 2018 and independent expert by skype, 18 April 2018). An Action Plan could provide a new comprehensive approach to external EU forest policy, with an overarching policy framework that includes both demand-side and supply-side measures. It could also place current support for REDD+ within a more integrated approach, as one of a number of measures to tackle deforestation rather than as the sole solution to tackling tropical deforestation. EU action plans can be effective tools for increasing political and financial support to tackle an issue. The publication of the EU FLEGT Action Plan in 2003, for example, garnered significant support for preventing the imports of illegal logged timber into the EU [85], thereby tackling one important driver of forest degradation. The publication of an EU Action Plan against Wildlife Trafficking in 2016 similarly gave considerable impetus to addressing wildlife trafficking. Action plans can, however, also remain weak, non-binding documents if they do not include legislative proposals. Thus, they may fail to achieve inter-sectoral integration, as was reportedly the case for the EU's domestic forest strategy and associated action plan [45,86]. So far, the European Commission is yet to come forward with a proposal for an action plan in relation to tropical deforestation.

#### **6. Conclusions**

Our analysis has shown that there is growing momentum within the EU to implement the SDGs and act on climate change, but tackling tropical deforestation or supporting REDD+ remain relatively low priorities for policy-makers within these wider debates. There has been a genuine shift in emphasis within policy debates towards addressing greenhouse gas emissions and negative impacts of consumption at the EU level. However, a gap remains between such rhetoric and specific policy measures being proposed to tackle EU consumption in relation to tropical deforestation. Despite several calls for an EU Action Plan against Tropical Deforestation that contains regulatory measures on EU imports of forest-risk commodities, and suggested new policy measures included in several documents, the European Commission has not yet acted. This may be because economic interests are now dominating strongly over environmental ones, as has happened in the case of domestic EU forest policies [87], although there are indications that some private companies are actually supportive of greater EU action on tropical forests [71].

With regard to REDD+, very little is being proposed in terms of new policies or funding. This could be because of the perception that the private sector is more interested in "sustainable supply chain" approaches, despite limited evidence that they can deliver on social sustainability objectives and reduce deforestation [78,88]. It is currently unclear whether supply-chain approaches will give further impetus to jurisdictional REDD+, whether they will replace or be a substitute for REDD+ initiatives, or whether new networked governance arrangements will emerge with partner (developing) countries. If the European Commission does decide to develop an overarching action plan that includes both demand- and supply-side measures, this could indeed signal a new integrated EU approach to tackling deforestation, which replaces the current range of fragmented approaches. Under this scenario, REDD+ could become one of a number of linked policy approaches, rather than being burdened with the expectation that it can solve tropical deforestation (and all associated issues) on its own. This would also spread the burden of responsibility more evenly between developed and developing countries. Embedding REDD+ within the framework of an initiative, such as an EU Action Plan, could also contribute to overcoming two major challenges, namely the lack of REDD+ finance and the need to address international drivers of deforestation and forest degradation [89]. However, in the absence of actual EU legislative proposals to tackle drivers linked to global consumption and new commitments to REDD+ finance, the interest in tackling embodied deforestation is unlikely to signify major policy change, with the burden of responsibility to combat deforestation continuing to fall on developing countries.

**Author Contributions:** Conceptualization, J.W.-S. and A.G.; Data curation, J.W.-S.; Formal analysis, J.W.-S.; Investigation, J.W.-S.; Methodology, J.W.-S.; Supervision, A.G.; Writing—original draft, J.W.-S.; Writing—review & editing, A.G.

**Funding:** This research received no external funding. Wageningen University will cover the cost of publication (open access fee).

**Conflicts of Interest:** The authors declare no conflict of interest. The lead author is employed by the Wildlife Conservation Society (WCS) which works to advise on and advocate for EU policy action in favour of wildlife conservation. This research was undertaken independently in a personal capacity as part of a PhD programme at Wageningen University and was not funded by WCS.


**Appendix A. List of EU Policy Documents Included in the Analysis**



**Appendix B. List of** 

#### **References**


2016. Available online: https://ec.europa.eu/digital-single-market/en/news/communication-commissioneuropean-parliament-council-european-economic-and-social-committee-a-0 (accessed on 30 November 2018).


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
