Scholarly article on topic 'Questioning the use of ‘degradation’ in climate mitigation: A case study of a forest carbon CDM project in Uganda'

Questioning the use of ‘degradation’ in climate mitigation: A case study of a forest carbon CDM project in Uganda Academic research paper on "Social and economic geography"

CC BY-NC-ND
0
0
Share paper
Academic journal
Land Use Policy
OECD Field of science
Keywords
{Degradation / Narrative / Deforestation / "Climate change" / CDM / "Remote sensing"}

Abstract of research paper on Social and economic geography, author of scientific article — Flora Hajdu, Oskar Penje, Klara Fischer

Abstract An urgent need to stop degradation is frequently cited as support for climate mitigation efforts involving forests. However, lessons learnt from social science research on degradation narratives are not taken into consideration. This creates a risk of problematic degradation narratives being used to legitimise forest carbon projects. This study examined a Clean Development Mechanism (CDM) forest plantation in Uganda, where incomplete and partly contradictory evidence on land use change was interpreted in a way that overemphasised degradation. This interpretation was in line with the interests of the forestry company proposing the CDM activity and with national interests in Uganda, and was stimulated by CDM guidelines and regulations. Our investigation revealed a more complex picture of land cover change in the area that did not support the narrative of an area undergoing continuous degradation. We therefore recommend that close scrutiny of the degradation narrative presented be included in every type of forest carbon project.

Academic research paper on topic "Questioning the use of ‘degradation’ in climate mitigation: A case study of a forest carbon CDM project in Uganda"

Contents lists available at ScienceDirect

Land Use Policy

journal homepage www.elsevier.com/locate/landusepol

Land Use Policy

Questioning the use of'degradation' in climate mitigation: A case study of a forest carbon CDM project in Uganda

Flora Hajdu (Researcher)*, Oskar Penje (Research Assistant), Klara Fischer (Researcher)

Department of Urban and Rural Development, P.O. Box 7012, SE-750 07 Uppsala, Sweden

ABSTRACT

An urgent need to stop degradation is frequently cited as support for climate mitigation efforts involving forests. However, lessons learnt from social science research on degradation narratives are not taken into consideration. This creates a risk of problematic degradation narratives being used to legitimise forest carbon projects. This study examined a Clean Development Mechanism (CDM) forest plantation in Uganda, where incomplete and partly contradictory evidence on land use change was interpreted in a way that overemphasised degradation. This interpretation was in line with the interests of the forestry company proposing the CDM activity and with national interests in Uganda, and was stimulated by CDM guidelines and regulations. Our investigation revealed a more complex picture of land cover change in the area that did not support the narrative of an area undergoing continuous degradation. We therefore recommend that close scrutiny of the degradation narrative presented be included in every type of forest carbon project.

© 2016 Published by Elsevier Ltd.

CrossMark

ARTICLE INFO

Article history: Received 21 June 2016 Received in revised form 14 September 2016 Accepted 19 September 2016

Keywords:

Degradation

Narrative

Deforestation

Climate change

Remote sensing

1. Introduction

Climate mitigation efforts are increasingly involving forests. In particular, forests in developing countries are seen as key for both sequestering carbon and halting increased emissions from deforestation that threaten to undo achievements in emission reductions (Pistorius, 2012; Holmgren, 2015). Various climate-forestry initiatives are therefore proliferating in global climate governance schemes (Leach and Scoones, 2015), as is evident in the focus on forests in the Voluntary Carbon Market (VCM) and in various REDD+ (Reducing Emissions from Deforestation and forest Degradation) initiatives (Holmgren, 2015). Within Clean Development Mechanism (CDM),1 which is currently the most widely tested scheme for global cooperation to mitigate climate change, there are few forest-related projects at present (most projects concern transfer of clean technology). However, lessons learnt from afforestation/reforestation CDM projects could prove important when other forest-carbon projects expand, as planned in the future.

* Corresponding author.

E-mail address: flora.hajdu@slu.se (F. Hajdu).

1 CDM projects allow rich countries to invest in carbon emissions reductions in developing countries and count these reductions towards their own reduction quota under the Kyoto protocol.

http://dx.doi.org/10.1016/j.landusepol.2016.09.016 0264-8377/© 2016 Published by Elsevier Ltd.

Degradation is a term frequently occurring in global climate mitigation talks, but seldom defined. An important social science body of literature criticises loosely defined 'degradation narratives' and their problematic consequences in various parts of the world (e.g. Fairhead and Leach, 1996,1998,2003; Fairhead and Scoones, 2005; Keeley and Scoones, 2003; Kull, 2004; Lambin et al., 2001; Leach and Scoones, 2015; Rohde et al., 2006; Stringer, 2009; Reenberg, 2013). However, this literature is not cited in UNFCCC reports on climate mitigation or in the various documents dealing specifically with CDM. Instead, it is frequently assumed that reducing degradation automatically leads to benefits for all involved. For example, tree-planting projects with the primary aim of mitigating climate change frequently claim that they also benefit conservation and the rural livelihoods of local communities (Cavanagh and Benjaminsen, 2014). However, various problems with this conceptualisation of only positive benefits and no sacrifices have been highlighted (Leach and Scoones, 2015; Lyons and Westoby, 2014; Nel, 2015).

In this paper, we consider the risk of problematic degradation narratives2 being constructed and used to legitimise CDM reforestation/afforestation projects and possibly other forest carbon

2 Narratives are inherently subjective and thus cannot be labelled 'true or false'. It is perhaps more helpful to discuss whether a narrative is based on unsubstantiated

projects. This could occur as a result of loosely defined degradation narratives, combined with the fact that describing an area as 'degraded' is usually in line with the interests of several actors in the CDM process. Furthermore, current CDM regulations do not sufficiently ensure that claims of degradation are empirically grounded.

After a brief introduction to the literature on degradation narratives in Section 2, we describe our case study - Kachung plantation in Uganda - in Section 3. We go on to discuss methods used for document analysis, interviews and the GIS control study in Section 4. The following findings in Section 5 are divided into five parts. We begin by showing how 'degradation' in the area has been used as a key motivation for the project (5.1), and how this seems to have influenced the empirical investigations into degradation conducted by the company proposing the plantation (5.2). We then discuss how these results were never questioned in external reviews of the project or by the buyer of the carbon credits (5.3). We go on to report on our own GIS control study, based on similar land cover data as used by the company, does not support the company's conclusions (5.4). Finally, we discuss our information about local contexts from documents and interviews through which a much more complex picture of land cover change in the area emerged (5.5). Based on these findings, we question the use of the degradation argument in afforestation/reforestation CDM projects and identify some key problems in the CDM process that permit it to be negatively influenced by degradation narratives in the discussion (Section 6). We recommend that the lessons learnt from the degradation narratives literature be taken into consideration in every type of forest carbon project and also make a few more concrete policy recommendations (Section 7).

2. Brief introduction to the literature on degradation narratives

A significant number of studies on 'degradation narratives' in Africa have shown that in various cases and locations, accepted knowledge about degradation and deforestation is not based on context-specific empirical evidence (e.g. Fairhead and Leach, 1996, 1998; Kull, 2004; Hajdu, 2009; Stringer, 2009; Rohde et al., 2006; Reenberg, 2013).3 Instead, old ideas about ecosystem stability, unsubstantiated claims about the inferiority of African natural resource use and a strong movement for nature conservation are often combined into a so-called 'degradation narrative' that serves the interests of those promoting various conservation measures. Subsequent scrutiny of these narratives by researchers studying human-nature interactions in African environments often reveals that the claimed deterioration of the local environment is not actually occurring, is significantly overestimated or wrongly attributed to local causes (Fairhead and Leach, 2003; Fairhead and Scoones, 2005).

The literature criticising 'degradation narratives' shows that, in a tradition dating back to colonial times, local people's practices of cultivation and use of fire have been seen as the key cause of degradation, while industrial-scale use of the land is not seen in the same way (Kull, 2004; Maddox, 2002). There are now indications that unsubstantiated degradation narratives are again influencing African land use investments. Recent private large-scale investments in forest plantations have been increasingly portrayed as not only contributing to halting degradation and sequestering carbon, but also benefiting local communities, who are portrayed as suffer-

opinions or some more concrete evidence, as we suggest in (Hajdu and Fischer, 2016).

3 In this paper we only give a very brief summary of the major findings of the degradation narrative literature. We discuss this literature and its links tothe current climate mitigation discourse more at length elsewhere (Hajdu and Fischer, 2016).

ing from degradation that they themselves are causing (Leach and Scoones, 2015).

However, investigations into local effects of projects have shown that consequences for local people can be 'ambivalent' (Locher and Muller-Boker, 2014) or downright negative (Cavanagh and Benjaminsen, 2014; Lyons and Westoby, 2014; Nel, 2015). Empirical investigations showing that local practices have varying effects on the local environment, depending on context, have also led researchers to question the assumptions about the connections between degradation and local land use (Fairhead and Scoones, 2005; Lambin et al., 2001). The belief that local practices are the main cause of degradation has led to the development of rules to regulate such practices, frequently resulting in negative effects on local livelihoods (Barrett et al., 2013; Rohde et al., 2006). It has also been shown that the choice of methods used in scientific investigations into e.g. land degradation, and interpretation of the data obtained, may have been influenced by unreliable degradation narratives (Fairhead and Scoones, 2005; Keeley and Scoones, 2003). As discussed by e.g. Frewer and Chan (2014), GIS methodology used to detect forest cover change is often described as a neutral tool for investigation, but in fact the choice of e.g. places and time frames to measure and interpretation of the data can be highly affected by preconceived ideas about human-nature interactions. Despite this, GIS investigations are often seen as more 'neutral' and evidence-based than e.g. interviews.

It has been suggested that the risk of degradation narratives influencing science and policy is particularly great if the narrative serves the interest of influential actors (Fairhead and Leach, 2003; Keeley and Scoones, 2003). It is therefore important to question degradation claims, e.g. in climate mitigation projects (Leach and Scoones, 2015), not least because actors that have much to gain from a certain area being portrayed as degraded may also have much influence over work to verify degradation status in that area. We do not claim that these actors necessarily have bad intentions, but rather that they may be unaware of the problems associated with degradation narratives and may believe they are acting in the best interests of local communities.

In the UN system surrounding CDM afforestation/reforestation (CDM A/R), the issue of 'restoring' degraded land remains important, but a curious conflation of plantation forest with natural forest means that this 'reforestation' often refers to plantations of exotic species (Sasaki and Putz, 2009). In the case reported in this paper, the CDM regulations required the Norwegian forestry company involved to demonstrate that the "lands to be afforested or reforested are degraded and the lands are still degrading or remain in a low carbon steady state" (UNFCCC, AR-AM0004/Version 04, page 1) in order to be eligible as a CDM project. However, these investigations into degradation were mainly conducted by the forestry company, which thus stood to gain from demonstrating degradation. We argue that this set-up risks allowing opinions and strong interests to influence investigations into degradation and the ensuing narrative, despite the seemingly rigorous regulations surrounding its verification.

3. The case context

The way in which degradation is discussed in CDM was examined through a case study of a project in Uganda that was mainly designed and implemented by the Norwegian forestry company Green Resources Ltd. (GR). Within the "Kachung Forest Project: Afforestation on Degraded Lands", the company planted trees (mainly pine and eucalyptus) in the Kachung Central Forest Reserve, which according to the Project Design Document (PDD) version 8 (2012) submitted in the UNFCCC system by GR consisted to a large degree of a "degraded savannah environment" (PDD,

2012:9). The Ugandan government approved the project, in line with CDM requirements, and leased the land to GR through its National Forest Authority (NFA), but otherwise was not actively involved as a partner. In 2011, the Swedish Energy Agency (SEA) committed to buying the carbon emissions reductions generated by the plantation through the CDM system. GR started planting in Kachung in 2006 and completed the planting phase in 2012.

This case was selected because it was one of the few operational carbon forestry projects in Africa at the time of the present study and because the terms degradation and deforestation were used as key motivations for the project in the document produced by GR for CDM certification (i.e. PDD, 2012) and by SEA when promoting the project to the Swedish public (e.g. SEA 2012:38-42). GR obviously had much to gain from the area being defined as degraded, since this enabled it to secure CDM funding. At the same time, the company was mainly responsible for investigating the degradation status of the area. We wanted to study whether this situation influenced how degradation was investigated.

Before the first plantation initiatives in the 1930s, the Kachung area was described as a vast savannah, used mainly for communal grazing (Chaudhry and Silim, 1980). The area was gazetted as a Central Forest Reserve (CFR) earmarked for plantation in 1952. Its management is entrusted to the Ugandan NFA. However, according to our interviews with NFA and GR staff, the reserve was never fully planted in the past. During the 1960s-1970s, people living close to many forest reserves were encouraged by government to take CFR land into use for farming. This included Kachung, where the Forest Department encouraged villagers to take land inside the reserve in 1966/67. The soil in the reserve proved to have good fertility, which attracted more smallholders in the early 1970s (Chaudhry and Silim, 1980).

The forestry context in Uganda is complex and contested. In many cases, the establishment of reserves and plantations has been associated with widespread violence, with people being forcibly removed with little or no prior warning (Cavanagh and Benjaminsen, 2014; Lyons and Westoby, 2014). Furthermore, as Turyahabwe and Banana (2008) show, the Ugandan government's capacity to implement, monitor and evaluate its forest policies has been consistently inadequate. As a solution to the lack of capacity of the NFA to manage the CFRs, Uganda changed its National Tree Planting Act in 2003 and opened up for private foreign investments in plantation forestry (Namanya, 2008; Lyons and Westoby, 2014). This made it possible for GR to lease land from the Ugandan state for Kachung plantation.

Our interviews with SEA and information given in the due diligence report ordered by SEA revealed that when GR received its licence to plant in Kachung, there were still many people farming and some people living inside the reserve. There are reports that before GR initiated their plantation, the NFA forcedly removed people from agricultural land inside the reserve, which led to resentment amongst the people affected. After this initial turbulence at establishment of the plantation, GR sought to avoid conflict and hired a local community officer to maintain continuous dialogue with local smallholders about phasing out agriculture in the reserve areas that were due to be planted and about possible compensation measures. Our interviews indicated that most smallholders did not experience violent removals, but still resent having lost farmland.

4. Material and methods

Relevant documents were analysed and interviews were conducted to identify the empirical evidence on which GR based its argument about the area being subjected to degradation. This was combined with new remote sensing-based investigations of the

land cover conditions during the period in question. A field visit to the plantation in Uganda was also made, where staff working for GR and local residents in the area were interviewed.

4.1. Interviews and document analysis

Information was gathered from documents and interviews with all the key actors involved in the Kachung project, i.e. SEA, GR (in Uganda and London), the NFA in Kampala and the Ugandan Designated National Authority (DNA) for CDM at the Ministry of Water and Environment, Climate Change Unit. The ways in which forest, savannah, deforestation, degradation and the role of the local people were discussed in documents and interviews were analysed with the aid of NVivo software. The Project Design Document (PDD) submitted by GR to the UNFCCC was key to the analysis, as it was the document in which GR had to justify the project, including describing original land use and how the project would impact on the current land use pattern. The content in the PDD was cross-checked in interviews with key GR staff locally in Uganda and with the individual at GR who had worked with the CDM application process for Kachung. We also submitted questions about unclear statements in the PDD by email to GR officials.

During a field visit to the Kachung plantation, we visited the nursery and the plantation and interviewed the GR community development officer for Kachung, the acting plantation manager and workers at the plantation nursery. We also visited eight of the 17 villages defined by GR as close enough to be affected by the plantation and performed eight interviews on an individual or group basis with smallholders living around the plantation. These interviews were not recorded, but detailed field notes were taken. Local interviews were held mainly for the purpose of data triangulation and to get a local perspective, and we did not seek to draw any major conclusions from the limited information obtained. The effects of the plantation on local livelihoods were not the main focus of this study.

4.2. GIS-based control study

We also performed a GIS-based control study, examining land cover change in the project area. The analysis was based on two separate sets of data products derived from the NASA/USGS Global Land Survey (GLS) collection of Landsat imagery. All input datasets are produced by the Global Land Cover Facility at the University of Maryland. The images used were from the Vegetation Continuous Fields Tree Cover layer (GLCF, 2011) and the Forest Cover Change layer (GLCF, 2014). The various properties of these layers are discussed further in Section 5.4, since this has implications for the findings.

5. Findings

The findings are organised into five sub-sections, as already described in more detail at the end of the introduction in Section 1.

5.1. Problematic degradation narratives used as key justification for the project

GR relied heavily on claims of degradation to justify the plantation. The quote below is an example of how the company described degradation in the Kachung area and its connection to local human land use in the PDD:

.. .the reserve has been subjected to continued degradation, especially over the last few decades where a significant increase in subsistence activities, such as shifting cultivation and grazing activities, fuel-wood collection, and charcoal production have

been witnessed, reducing a denser woodland savannah to a landscape with scarce pockets of trees. (PDD, 2012:9)

The quote above is typical for the PDD, in which local activities that have led to degradation are listed in detail several times. The terms 'degradation' and 'degraded' are used a total of 54 times and 'deforestation' and 'deforested' are used seven times in the PDD. On 51% of the occasions that degradation or deforestation is mentioned, it is in close connection with discussions about the negative effects of local people's land use. Moreover, apart from use by local people, the document never mentions any other human impact, such as commercial use or climate change. The connections to needs and pressures from outside the local area (such as a large urban market for charcoal) are only made briefly in the PDD, and do not influence the overall narrative of local people being solely responsible for local degradation.

The quote above, as well as statements made in our interviews with GR officials, indicated that degradation was indirectly defined only as a change towards less tree density in the landscape. This way of framing degradation only accounts for reduction in biomass and the land's capability to sequester carbon, while ignoring e.g. the biodiversity loss that otherwise features centrally in many definitions of forest degradation (Sasaki and Putz, 2009). The view also resonates with a still widespread perspective (although contested in current ecology; Holling, 1973; Waide, 1988) that a mature forest is the final end-state of succession that all ecosystems, irrespective of geographical context, will reach if left 'undisturbed'. This tendency to view forest as the most natural of landscapes is reflected in the PDD, which states that:

The Landsat image from 1989 together with the current land-use map shown in A.4.2 illustrates that the few remaining pockets of natural forest from 1989 were deforested within this interval. (PDD, 2012:27)

While GR acknowledges that imagery from before 1989 was not analysed, the wording 'remaining pockets of natural forest' reveals the underlying assumption that the area had been covered with forest at some earlier stage and that this is a more 'natural' condition in this environment than the savannah woodland that was present in 1989. Apart from not being well anchored in scientific evidence about the earlier state of the environment around Kachung, such perspectives become problematic when they are connected with value judgments about land use, which is common in degradation narratives (see Section 2). This was exactly the case in the PDD, which suggested that local grazing activities were the reason the area was deforested:

The reserve has been of a savanna [sic] land-class even before it was initially gazetted (FAO, 2009). At this time, the land was used for grazing activities, suggesting an absence of forest due to the prevalence of grass being used as fodder. (PDD, 2012:23)

The link made between the area being used for grazing and therefore having no forest is in fact not supported by the document cited as '(FAO, 2009)' in the PDD quote above,4. This document actually describes the land as being a "vast savanna [sic] area" at least since the 1930s, with no claims about this savannah ever having been forest. In another report referred to in the PDD (Kamanyire, 2000), the region where Kachung forest reserve is situated is broadly classed as a savannah environment where the grassland is "ideal for grazing" (2000:10). However, that report was mainly used by GR for its references to national statistics on deforesta-

4 This document is actually is not an FAO document from 2009, but ajournal article from 1980 (Chaudhry and Silim, 1980) available at the FAO corporate document repository.

tion rates in Uganda. The way in which GR selectively interpreted these reports led to unsubstantiated conclusions in the PDD about how grazing causes degradation. The Ugandan NFA, which leased the land in Kachung to GR, similarly framed degradation as caused by local residents and reversible through plantation forestry. The alignment of visions between the Norwegian company GR and the NFA could have been influenced by the fact that Norway was instrumental in the creation of the NFA in 2003 (Nel, 2014). However, this view of degradation is not unique to those actors, but indeed fits with a much wider 'African degradation narrative' (Hajdu and Fischer, 2016).

5.2. Degradation narratives seemed to influence GR investigations into degradation

For land to be eligible for a CDM A/R project, the participants must provide transparent and complete information that the vegetation within the project boundary is below the forest threshold (as defined by the host country), both at the time of the project start and at a cut-off date set at 31 December 1989.5 This requirement was met by GR using a "supervised classification of Landsat imagery from 1989" (PDD, 2012:24), which indeed seemed to prove that in 1989 the land was savannah used for communal grazing as discussed above.

However, the Landsat image from 1989 was also used by GR to demonstrate degradation in the reserve, in order to meet the requirement that the project activity complied with the applicability conditions that "lands to be afforested or reforested are degraded and the lands are still degrading or remain in a low carbon steady state" (cited in the PDD, 2012:27). In the PDD, GR claimed that the land had been degraded by comparing the 1989 Landsat image against a "current land-use map" (Figure A.4 .1.3.1 in the PDD)6 without explaining further about how this map had been produced or by whom. This figure is in fact a product of GR as well, a Landsat image from 2006, verified on the ground in 2009. It might show a fair representation of the ground conditions in 2006, but this is difficult to verify since polygons from other sources have been added, overlaying the original classified raster image. Furthermore, there is no description of how the classification was carried out or how the team ensured that their methods allowed fair comparisons between the two periods.

Using satellite images to determine a process of gradual land degradation, as GR chose to do in this case, requires consistent classification methodology. This is especially true when conducting studies in areas of sparse vegetation (such as woodland savannah). In order to validate the accuracy of automated image classification, ground truth samples need to be collated and compared against the computed land cover classes. When performing a supervised classification, the user generally chooses representative samples beforehand for each land cover class in the remotely sensed data. This area is often referred to as the training site. Programming can then translate the composition of the neighbouring pixels to an algorithm, which is used to detect similar land cover types in the entire image (Lillesand et al., 2015).

Detecting changes in vegetation between two dates through two different images is even more complex. In addition to choosing products with similar spatial resolution and technical qualities,

5 This date was chosen as sufficiently far back to ensure that no one could have predicted the future potential to earn money through the UNFCCC system for afforestation/reforestation, and at the same time as sufficiently recent to ensure the existance of reliable remote sensing data.

6 The PDD actually states "the map in Figure A.4.2." (page 27), which is a legal

map of the borders of areas eligible for CDM. We wrote to GR to ask them about this reference and they answered that this was a mistake and it should be Figure A.4.1.3.1.

seasonal changes and atmospheric conditions (affecting light reflection) also need to be taken into account in selection of the images. If no land stratification sample has been extracted and used as an input to image processing, the classification has to be based on clustering of pixels with similar values. This method would not meet the definition of supervised classification and would thus be less suitable when attempting to detect land cover changes, especially in savannah-type vegetation as in the case of Kachung.

GR claimed to have performed a supervised classification, but does not provide a clear description of the process, simply claiming that the analysis showed land class change, which it took to mean degradation:

The time series from 1995 to 2005, based on the NFA maps, shows how that the land class changed from an area of woodland vegetation to bush (synonymous with shrubland) and subsistence agricultural land. This land-class change clearly demonstrates how the vegetation has been degraded over this time period. (PDD, 2012:23)

The statements made in this quote and in that above from the PDD (2012:27) are very difficult to double-check, since the time series and NFA maps mentioned are not presented in the PDD. In most of the maps presented by GR there is also an absence of time/date indicators, which makes it difficult for any validator to follow the argumentation in the text.7 Another problem is that there is no real discussion on the quality and reliability of the data. As pointed out by e.g. Frewer and Chan (2014), the technical limitations of GIS are such that the analyst's own desires can influence outcomes in various ways, a risk that at least needs to be discussed.

In the absence of convincing empirical data, general statistics on deforestation in Uganda were used by GR as proof of degradation in Kachung, despite, as we have shown above, the low likelihood that the specific area around Kachung was forested in the past. The quote below illustrates the logical leap made from deforestation reported at national level to it being "not surprising" that degradation of savannah woodland had occurred at local level in Kachung. Subsistence activities of local people were used in this context as proof of deforestation:

As shown in the FAO's Global Forest Resources Assessment 2005, since 1990 Uganda's forests and wooded lands have decreased from approximately 6.3 million to 4.7 million hectares, which presents one of the highest deforestation rates in the world over the last decade. [...] In light of this, it is not surprising that deforestation, or more specifically degradation of savannah woodland, has been present at [Kachung] over the last century, principally due to the prevailing land-use of subsistence agriculture, fuel-wood collection, charcoal production and grazing activities. (PDD, 2012:33)

It should be noted here that there were other assessments of global forest cover change available apart from that by FAO (2005), which was the only one referred to by GR. For example, Sexton et al. (2013, Fig. 8 in that article) show a map of global forest cover change ca. 1990-2000 based on Global Land Cover Facility classification, where Uganda seems to have more forest gain than loss between these years and the particular region around Kachung has very little forest loss at all. Naturally, these data can also be questioned, but they show that there are no simple truths about land cover change. Rather, investigations into land cover change will always depend on the many choices made when deciding on baselines, interpret-

ing variations as temporary fluctuations or permanent changes, defining changes as good or bad, and attributing them to different causes. Such investigations are thus highly vulnerable to influence from degradation narratives and other preconceived notions held by actors in charge of such investigations. This vulnerability must be acknowledged and handled, rather than ignored and downplayed.

5.3. Degradation claims not sufficiently verified in the CDM process

Taken together, the issues discussed above could indicate that degradation was not sufficiently investigated by GR in the process leading up to drafting the PDD, but rather inferred from preconceived notions about local practices, with remote sensing imagery and maps interpreted in this light. The CDM process in the UNFCCC ought to have set up safeguards against potential gaps in knowledge and/or interests in profit skewing the analysis of local degradation. Instead, it was set up so that 'finding' degradation allowed GR to go forward with the project. Awareness of that degradation narratives can influence investigations, including GIS studies, seems to be lacking within UNFCCC in general, as reflected e.g. in the Reference Manual for Afforestation and Reforestation Projects under CDM implying that reforestation of degraded areas is of particular benefit to local people:

Successful implementation of well-designed afforestation and reforestation projects can potentially generate a number of co-benefits that promote sustainable development in the host Party. A larger part of these benefits is likely to flow directly to the local communities, although regional and national benefits can also be significant (UNFCCC, 2013: 13)

In fact, such benefits are difficult to ensure and many projects have mainly had negative local effects (Cavanagh and Benjaminsen, 2014; Leach and Scoones, 2015; Lyons and Westoby, 2014; Locher and Muller-Boker, 2014; Nel, 2015). At times the CDM guidelines even encourage degradation narratives. For example, the guidelines that GR followed when establishing Kachung as a CDM project state that if "the baseline scenario is degrading land involving decline in woody vegetation cover", then the "GHG emissions from felling, clearance, decay or burning of existing woody biomass during site preparation are insignificant" (UNFCCC, EB 50, Annex 21, page1). In essence, this means that it is profitable for the forestry company to show that the land is being deforested.

The UNFCCC system is otherwise full of cross-checks and verifications, where independent consultancy firms are required to validate and verify CDM projects. However, in these external controls there is no demand for independent evaluations of degradation claims. As a result, neither of the two external consultancy reports on Kachung contained any independent evaluation of the degradation claims made by GR. The Swedish Energy Agency (SEA), which purchased the CDM credits, also made its own field visits and commissioned a separate consultancy firm to perform a due diligence assessment on the plantation but, as with the consultancy firms validating and verifying the plantation, no independent evaluation of the degradation claims was made. Instead, SEA reiterated that the land in Kachung was being degraded by local people and that the plantation would therefore be an improvement both for the environment and for rural livelihoods in the area. Under the heading "Forest plantation leads to sustainable life in Uganda", SEA stated in an information brochure8 to the Swedish public that "a

7 When we asked to have more information about these maps and how they were generated GR answered that they had given more information to the validating agencies than what was written in the PDD, but that they currently could not get hold of this information.

8 After a very critical documentary about land conflicts in the plantation broadcasted on Swedish television in November 2015, the SEA removed this brochure from their website and published a statement where they acknowledged that the land was not unused before GR established the plantation. There has however not

Fig. 1. Forest cover change in the Kachung Central Forest Reserve and CDM project area. The layer shows a merge of Forest Cover Change layers from two periods, 1990-2000 and 2000-2005 (GLCF). The colours represent the net change in forest coverage per pixel with a spatial resolution of 30 m. Absence of colour represents non-forested areas (less than 30% tree cover) or water.

new healthy forest is emerging where there previously was only unused bushland". The area was described as "not having natural forests anymore" and it was stated that ".. .before the project started there had not been forest here for a long time. Bushland had completely taken over, but now forest is on its way back" (SEA, 2012 :39, authors' translation from Swedish). Our interviews with SEA officials showed that it never really occurred to them to question the interpretation of the area as degraded, and indeed there are no policies or regulations that would encourage them to do so.

5.4. Degradation claims seem unfounded according to our control

In order to further investigate the degradation claims made by GR in the PDD, in a control study we compared remote sensing data available for Kachung between 1990 and 2005. The data we were able to find were on forest cover change from 1990 to 2000 and 2000 to 2005. Between 2000 and 2005 (before GR started planting), there are also classified images available on tree cover.

The Forest Cover Change (FCC) layer is a post-classification change detection product showing changes in forest cover at a resolution of 30 m, for which classified data from GLS epochs91990, 2000 and 2005 have been used to generate pixels of forest loss; for-

been any acknowledgement from SEA, or any other actors involved in Kachung, about the lack of empirical basis for the description of degradation.

9 The term epoch refers to the nominal year that is represented by the image.

est gain and persistentforest. The nominal year is a global target year. This means that if no cloud-free image from the growing season can be obtained from the target year at a specific scene, an image from a previous or subsequent year will be used instead. In the case of the Kachung scene, the nominal year of 1990 is in fact represented by an image from 1986.

The Tree Cover (TC) layer contains estimates of the percentage of horizontal ground in each 30-m pixel covered by woody vegetation greater than 5 m in height. The data represent two nominal GLS epochs, 2000 and 2005. The layer is useful in monitoring site-specific changes, especially in areas of sparse vegetation. Comparing these two epochs indicated trends in woodland coverage at the time of the project start in 2006 (see Sexton et al. (2013) for complete methodology).

The definition of forest used by GLCF in the FCC and TC layers is: locations larger than 1 hectare in area with more than 30% tree cover higher than 5 m. This is the same definition as used by the NFA and also by GR in the Kachung case. Tree height is generally determined using Lidar techniques. In the control study, basic raster dataset overlaying and raster calculation were undertaken using the Esri software product ArcMap, which was also used for the final cartographic representation of the computed data.

Since the spectral signatures used (by GLCF) in both the FCC and TC datasets are generated to perform automated classifications on a global scale, there is a level of uncertainty when using them to establish conditions on local level. This is especially true for sparsely forested areas. Nevertheless, these are the most elaborate and com-

Fig. 2. Tree cover change in the Kachung Central Forest Reserve and CDM project area. The green colour represents percentage of tree canopy coverage for each 30-m pixel (GLCF). The blue polygon shows the area defined as woodland/forest remnant in the PDD (in the 1989 stratification map the area is classified as broadleaf plantation, p. 23). The brown polygons are areas defined as pockets of native forest in the 1989 stratification map. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3. Histogram of the pixels from the CDM project area in the two tree cover change maps in Fig. 2. The number of pixels representing a certain percentage tree cover is shown for each image.

prehensive of the classified forest cover datasets available to date, and were therefore considered to be suitable sources of information about historical land cover conditions in the area.

Fig. 1 shows forest cover change in Kachung between 1990 and 2005. As can be seen, the reserve as a whole experienced both loss and gain of forest in this period. These changes mainly took place close to the straight road that dissects the reserve north-to-south, in the area that is not within the CDM project area. Very few changes in forest cover can be detected within the boundaries of the CDM project area. The image in Fig. 1 cannot be the basis for any major conclusions about the area due to the problems we describe above, but nevertheless does not support a view of an area that has been significantly deforested.

The tree cover images from 2000 and 2005 (Fig. 2) also do not support the claim that there had been significant deforestation. On the contrary, they indicate that between 2000 and 2005 there was a notable increase in tree cover. It can be difficult to judge this by just looking at the images, and therefore in Fig. 3 we provide a histogram of the pixels from within the CDM project area in Fig. 2. The histogram shows that areas with 11-16% tree coverage dominated in 2000, while areas with a coverage of more than 18% were nearly twice as frequent five years later. A shift towards denser tree cover seems clear from the histogram.

Furthermore, the Landsat image from 1989 presented in the PDD by GR (image A.7.1, page 23) shows what is referred to as 'pockets of native forest' (claimed to have been deforested, as discussed above) and a section of'broadleaf plantation'. Neither of these is visible in any of the images from the control study. For increased clarity, in Fig. 2 we added two polygons showing the areas that were classed as 'native forest' and 'broadleaf plantation' in the PDD. It can be clearly seen that the areas within the polygons do not have visibly different tree cover than the surrounding areas. From these investigations, it therefore seems questionable whether these 'pockets of native forest' actually existed. Alternatively, the classifications may have been inaccurate and these areas should perhaps never have been called 'forests'. In the PDD (p.22), GR confirmed that it followed the definition of forest used by Uganda (as it was required

to do). The Global Land Cover Facility, from which we took our data, uses the same definition in its statistics on global changes in forest cover, as discussed above. The discrepancy between what we found in the control study and what GR reported can be taken as an example of the difficulties in tracking land cover change through remote sensing, and shows that GIS methodology on its own cannot provide any clear-cut evidence of historical changes. In interviews, however, SEA representatives mentioned that GIS investigations are seen as 'harder' evidence in the CDM process than evidence obtained through e.g. interviews with local residents. Our study indicates the extent to which GIS investigations are seen as hard evidence and taken as proof for degradation, without sufficient triangulation with interviews and other types of investigations.

We certainly do not claim, based on our data, that we have uncovered 'the truth' about land cover change in Kachung over the last 25 years. Using satellite imagery to determine discrete processes like the gradual thinning of tree cover in savannah requires fine resolution imagery and consistent classification methodology. A complete control study would demand classification using spectral signatures obtained at local training sites. However, our data do not at all support a picture of loss of trees in the Kachung area between 1989 and 2005. In fact, tree cover in the reserve seems to have increased between 2000 and 2005. We therefore consider it problematic that GR drew the opposite conclusions and presented these as 'truths'. The methods it used, the weight given to GIS over local interviews and the way it drew conclusions from rather unclear information on land cover are all obviously problematic in several regards. Furthermore, its conclusions were not investigated further in the CDM validation and verification process. As a result, GR was able to interpret scant and inherently conflictive data to fit with a pre-existing world view where degradation due to local practices was taken for granted. This description of historical land use also suited its intention of establishing a CDM plantation.

5.5. Local context highly relevant for explaining changes in vegetation

In contrast to the picture painted by GR in the PDD, interviews with local smallholders supported the findings from our GIS control study and indicated that Kachung has not been subjected to continuous linear degradation. Practices of taking land in the reserve have varied over time, as has the population pressure. For example, the pressure on the land in parts of the reserve seems to have increased significantly from around the late 1990s up until 2009, when some of the villages in the area hosted many refugees fleeing the Lord's Resistance Army (LRA). During this period, many people in the area rented out their own fields to the refugees, while they themselves opened new fields on reserve land. People also explained that the influx of refugees who did not have their own land from which to collect firewood created a temporary local market for charcoal, which led to charcoal production being much more common during this time than it is today around Kachung. As a result of this temporary influx of people and high wood consumption for charcoal production, it seems likely that the pressure on trees in the area was significantly higher during this period than before or after. Considering the fact that overall tree cover actually seems to have increased between 2000 and 2005, loss of trees is also likely to have been quite localised.

From the PDD, it is clear that GR knew about this historically turbulent situation (e.g. PDD, 2012:34). Despite this, it did not draw the conclusion that the situation preceding the establishment of Kachung might not be representative of the long-term land development pattern in the region. Rather, the reserve was portrayed as being in a state of constant degradation:

.. .the only identified realistic and credible land-use scenario that would have occurred on the land within the proposed project boundary in the absence of the reforestation project under the CDM is a continuation of the current land-use: degradation of the grass and shrubland stratum of the reserve to cropland or degraded grazing/grassland. (PDD, 2012:42)

Furthermore, local people were portrayed as "taking advantage" of a period of relaxed enforcement of forest laws in the early 2000s "to satisfy their subsistence and cash needs", rather than being victims of a civil war, and in fact also having been previously encouraged by government to take farm land in the reserves (as described in section 3):

... the local community took advantage of relaxed enforcement of forest law and regulations, during the transition from Forest Department to NFA (discussed in section C.5.1), to cultivate more land within the forest to satisfy their subsistence and cash needs. (PDD, 2012: 100)

As described above, local testimonies indicate that the area experienced more pressure on land between 1990 and 2009 than has been the case since then. In some villages, local residents stated that tree cover in their village (i.e. outside the boundaries of the plantation) has started to recover as the refugees have left the area. These testimonies indicate a possible alternative scenario to the continuous degradation described by GR, namely that some lost trees would have been regenerated without GR's intervention after 2009. Indeed it is likely that local practices of planting and conserving trees would have further stimulated re-growth. The work by Kavallin Giertta (2016) shows how women taking firewood in and around the Kachung plantation had well-established strategies for ensuring the local regeneration of trees. She also notes that according to the local women, it is actually difficult work to deforest an area in this tropical climate; tree trunks have to be burnt, for example, and keeping a garden open is labour-demanding. A notable finding in our interviews was that some local villagers stated that the restriction on using land inside the plantation has led to significantly increased pressure on village land. It is thus possible that the GR plantation, rather than reducing degradation, is in fact causing a reduction in tree numbers in some villages outside the reserve today.

Thus the historical and current socio-economic context in which the plantation was proposed is much more complex than described by GR in its PDD. Again, no cross-checks in the system seem to have been in place to establish whether the context had been investigated to an acceptable level.

6. Discussion

We demonstrated in this study that failure to include social science knowledge on degradation narratives within the current climate mitigation discourse means that degradation narratives can be constructed without sufficient empirical support and used to legitimise afforestation projects. We examined the case of a CDM A/R project in Uganda, where evidence indicated that the current CDM process allowed the creation of a narrative of degradation for the Kachung area that seems unfounded, or at least severely exaggerated.

The forestry company, GR, portrayed the situation in Kachung as one of continuous degradation due to loss of trees, mainly owing to local land use practices. As proof of this, it referred mainly to findings from remote sensing investigations. However, GR was not able to provide any data that showed this clearly. Rather, it seems that its interpretation of the data was coloured by a combination of several mutually reinforcing factors: their clear interests in demonstrating degradation in the area, preconceived notions about forests being

the ultimate end-state of all natural ecosystems, stories of degradation being caused by local practices in Africa, and general statistics on deforestation in Uganda. This interpretation was subsequently accepted by the actors within the UN system, as well as by the actor buying the credits (SEA).

Our investigations revealed a different picture, although we cannot claim that our methodologies were without flaws. We also reiterate that it is not possible to draw detailed conclusions about land cover change based solely on remote sensing data. However, we were able to identify weaknesses in the methodologies used by GR and our control study showed that GR's portrayal of degradation in the area is clearly questionable, e.g. the images presented in Figs. 1-3 do not support a picture of loss of trees in the area. In fact, tree cover in the reserve seems to have increased between 2000 and 2005 (i.e before GR started planting trees). This contradicts the statements made by GR in the PDD that the reserve area has experienced a significant and continuous reduction in tree numbers.

The situation in Kachung at the time when GR prepared the PDD was obviously complex and changing, with refugees causing a temporary increase in pressure on land and natural resources until most of them returned home in 2009. Trees lost in the reserve could well have been expected to regenerate in the years after that. However, GR interpreted the situation differently, even though it knew about the local history of the area. This shows the extent to which interpretations can be influenced by pre-existing perceptions and expectations if these are not questioned. Our interpretation of land cover change in the area, based on the GIS study and local interviews, suggested that there was no clear trend of deforestation and that local loss of trees was likely to be temporary.

The CDM regulations about afforestation/reforestation are very strict in demanding proof that proposed project areas have not been forested since 1989, and the methodology used when GR applied for Kachung to be CDM-certified also demands proof of degradation. These requirements are intended to guard against 'perverse incentives', e.g. cutting down a forest in order to be able to apply for money to replant it. However, in the Kachung case the demands for proof of degradation in fact created another 'perverse incentive' where the emphasis on degradation stimulated the reproduction of a degradation narrative about local land use. The demand to comply with the applicability conditions discussed above caused GR to argue that the area was in a state of degradation and that native forest had been deforested, drawing far-ranging conclusions from a methodology that could not really support this.

The way in which the CDM process for afforestation/reforestation is constructed at present thus seems to encourage companies applying for CDM accreditation to interpret the target area as being in a state of degradation. This is problematic, since the subsequent CDM process does not necessarily question or empirically scrutinise initial claims of degradation and since previous lessons about degradation narratives are not discussed. The current CDM system therefore does not sufficiently seek to ensure that degradation narratives are based on reliable evidence, despite its many cross-checks, validations and verifications to ensure thoroughness and reliability, e.g. the exact amount of CO2 sequestered.

On the whole, there seems to be a serious lack of knowledge at the UNFCCC about the past prevalence of degradation narratives without a sound empirical basis and the problems such narratives can cause. Wilson Rowe (2015) points out that the policy negotiations on mitigation mechanisms represent a clash between disciplines where a few succeed in getting their knowledge accepted and other important perspectives are side-lined, which she exemplifies by showing how REDD+ has neglected the knowledge and experiences of foresters in favour of market economists and remote sensing specialists. It indeed seems that

also social science knowledge about degradation narratives has been neglected in the UNFCCC processes. There is therefore an urgent need to integrate critical awareness of problematic degradation narratives into the CDM, REDD+ and other mitigation efforts, and indeed into every type of forest carbon project.

7. Policy recommendations

Based on our findings, we suggest the following more explicit policy recommendations:

• Actors who stand to gain from an area being classified as degraded should not be responsible for performing, nor be able to influence, the investigation into land use and land cover history in the area.

• It is difficult to rely solely on cut-off dates for land cover assessments. The situation in 1989 in a specific local context might be part of a fluctuating and dynamic process. Local contextual analysis, including e.g. interviews with local residents, and studies of archival data on land use need to be given more importance in the overall assessment of land use and land cover history.

• The findings of social science research on degradation narratives and colonial representations of local land use practices need to inform investigations into degradation within climate mitigation to a much higher degree. Care should be taken not to encourage degradation narratives without an empirical basis. An analytical framework that can be used to facilitate this has been published (Hajdu and Fischer, 2016)

Lastly, we question the need for demonstrating that degradation is the problem when applying to join various climate mitigation schemes. It should be possible to create a plantation solely for timber production and carbon sequestration, without having to contribute to afforestation or 'halting degradation'. The current preoccupation with halting degradation often serves to hide other strong economic interests (in plantations and carbon sequestration). Rather than claiming that local people need these plantations, it should be acknowledged that the actors which need them most are the companies profiting directly from the plantations and the companies and countries emittingtoo much CO2 and needing cheap ways of offsetting this.

Acknowledgements

The authors wish to thank The Swedish Research Council for funding this project (SWE:2012-083), and two anonymous reviewers for providing very helpful comments.

References

Barrett, G., Brooks, S., Josefsson, J., Zulu, N., 2013. Starting the conversation: land issues and critical conservation studies in post-colonial Africa. J. Contemp. Afr. Stud. 31 (3), 336-344. Cavanagh, C., Benjaminsen, T.A., 2014. Virtual nature: violent accumulation: the 'spectacular failure'of carbon offsetting at a Ugandan National Park. Geoforum 56,55-65.

Chaudhry, M.A., Silim, S., 1980. Agri-silviculture in Uganda: a case for kachung forest. Unasylva32 (21-25) (accessed 09.07.16) http://www.fao.org/docrep/ n8595e/n8595e05.htm. Fairhead, J., Leach, M., 1996. Misreading the African Landscape: Society and Ecology in a Forest-Savanna Mosaic, vol 90. Cambridge University Press. Fairhead, J., Leach, M., 1998. Reframing Deforestation: Global Analyses and Local

Realities with Studies in West Africa. Psychology Press. Fairhead, J., Leach, M., 2003. Science, Society and Power: Environmental Knowledge

and Policy in West Africa and the Caribbean. Cambridge University Press. Fairhead, J., Scoones, I., 2005. Local knowledge and the social shaping of soil investments: critical perspectives on the assessment of soil degradation in Africa. Land Use Policy 22 (1), 33-41. Frewer, T., Chan, S., 2014. GIS and the 'Usual suspects'-[Mis] understanding land use change in Cambodia. Hum. Ecol. 42 (2), 267-281.

Global Land Cover Facility (GLCF) and Goddard Space Flight Center, (GSFC), 2014. GLCF Forest Cover Change 2000-2005. Global Land Cover Facility, University of Maryland, College Park.

Hajdu, F., 2009. Questioning homogenous degradation narratives in Transkei: Livelihoods and natural resource use in two Pondoland villages. In: Guyot, S., Dellier, J. (Eds.), Rethinking the Wild Coast, South Africa: Eco-frontiers vs livelihoods in Pondoland. VDM Verlag.

Hajdu, F., Fischer, K., 2016. Problems, causes and solutions in the forest carbon discourse: a framework for analysing degradation narratives. Clim. Dev., http://dx.doi.org/10.1080/17565529.2016.1174663.

Holling, C.S., 1973. Resilience and stability of ecological systems. Annu. Rev. Ecol. Syst. 4,1-23.

Holmgren, S., 2015. Governing Forests in a Changing Climate: Exploring Patterns of Thought at the Climate Change-forest Policy Intersection. PhD Theis. Swedish University of Agricultural Sciences, Acta Universitatis agriculturae Sueciae, pp. 1652-6880 (61).

Kamanyire, M., 2000. Sustainability indicators for natural resource management & policy. In: Working Paper 3, Natural Resource Management and Policy in Uganda: Overview Paper. Economic Policy Research Centre, Kampala ftp://ftp.fao.org/agl/agll/kageradocs/08case studies/ug nrm overview paper.pdf.

Kavallin Giertta, F., 2016. What Are Trees For? An Ethnographic Study of Local Firewood Practices in Uganda in the Context of Deforestation and Climate Change Discourses MSc Thesis. Department of Urban and Rural Development, Swedish University of Agricultural Sciences.

Keeley, J., Scoones, I., 2003. Understanding Environmental Policy Processes: Cases from Africa. Earthscan.

Kull, C., 2004. Isle of Fire: The Political Ecology of Landscape Burning in Madagascar. California University Press Berkeley.

Lambin, E.F., Turner, B.L., Geist, H.J., Agbola, S.B., Angelsen, A., Bruce, J.W., Coomes, O.T., Dirzo, R., Fischer, G., Folke, C., George, P.S., Homewood, K., Imbernon, J., Leemans, R., Xiubin, L., Moran, E.F., Mortimore, M., Ramakrishnan, P.S., Richards, J.F., Skanes, H., Steffen, W., Svedin, U., Veldkamp, T.A., Vogel, C., Xu, J., 2001. The causes of land-use and land-cover change: moving beyond the myths. Global Environ. Change 11 (4), 261-269.

Leach, M., Scoones, I., 2015. Carbon Conflicts and Forest Landscapes in Africa. Routledge.

Lillesand, T., Kiefer, R.W., Chipman, J., 2015. Remote Sensing and Image Interpretation, 7th ed. Wiley (March 2015).

Locher, M., Müller-Böker, U., 2014. Investors are good, if they follow the rules-power relations and local perceptions in the case of two European forestry companies in Tanzania. Geogr. Helv. 69, 249-258.

Lyons, K., Westoby, P., 2014. Carbon colonialism and the new land grab: plantation forestry in Uganda and its livelihood impacts. J. Rural Stud. 36,13-21.

Maddox, G., 2002. 'Degradation narratives' and 'population time bombs': Myths and realities about African environments In: Dovers, S., Edgecombe, R., Guest, B. (Eds.), South Africa's Environmental History: Cases and Comparisons. Ohio University Press, David Philip Publishers, Athens, Cape Town, pp. 250-259.

Namanya, B., 2008. Challenges to CDM implementation in Uganda: a critical analysis of legal and policy barriers. Int. J. Green Energy 5, 255-267.

Nel, A., 2014. Sequestering Market Environmentalism: Geographies of Carbon Forestry and Unevenness in Uganda Phd Thesis. University of Otago.

Nel, A., 2015. 'Zones of awkward engagement' in Ugandan carbon forestry. In: Leach, M., Scoones I. (Eds.), Carbon Conflicts and Forest Landscapes in Africa. Routledge, London and New York (94-107).

PDD, Project Design Document Form for Afforestation and Reforestation Project Activities (CDM-AR-PDD). Version 8,4th September 2012. Kachung Forest Project: Afforestation on Degraded Lands. UNFCCC/CCNUCC. Available online: https://cdm.unfccc.int/Projects/DB/TUEV-SUED1301918616.32/view.

Pistorius, T., 2012. From RED to REDD+: the evolution of a forest-based mitigation approach for developing countries. Curr. Opin. Environ. Sustain. 4 (6), 638-645.

Reenberg, A., 2013. Insistent dryland narratives: portraits of knowledge about human-environmental interactions in Sahelian environment policy documents. West Afr. J. Appl. Ecol. 20 (1), 97-111.

Rohde, R.F., Moleele, N.M., Mphale, M., Allsopp, N., Chanda, R., Hoffman, M.T., Magole, L., Young, E., 2006. Dynamics of grazing policy and practice: environmental and social impacts in three communal areas of southern Africa. Environ. Sci. Policy 9 (3), 302-316, http://dx.doi.org/10.1016/j.envsci.2005.11. 009.

SEA, Swedish Energy Authority, 2012. Sveriges CDM och Jl-program. Stöd till

internationella klimatprojekt. Available for free at www.energimyndigheten.se, ET2012:13.

Sasaki, N., Putz, F.E., 2009. Critical need for new definitions of forest and forest degradation in global climate change agreements. Conserv. Lett. 2 (5), 226-232.

Sexton, J.O., Song, X.-P., Feng, M., Noojipady, P., Anand, A., Huang, C., Kim, D.-H., Collins, K.M., Channan, S., DiMiceli, C., Townshend, J.R.G., 2013. Global, 30-m resolution continuous fields of tree cover: landsat-based rescaling of MODIS Vegetation Continuous Fields with lidar-based estimates of error. Int. J. Digital Earth, http://dx.doi.org/10.1080/17538947.2013.786146,130321031236007.

Stringer, L.C., 2009. Testing the orthodoxies of land degradation policy in Swaziland. Land Use Policy 26 (2), 157-168, http://dx.doi.org/10.1016Zj. landusepol.2008.01.008.

Turyahabwe, N., Banana, A.Y., 2008. An overview of history and development of forest policy and legislation in Uganda. Int. For. Rev. 10 (4), 641-656 (16).

UNFCCC, 2013. In: UNFCCC (Ed.), Afforestation and Reforestation Projects Under the Clean Development: A Reference Manual Mechanism. Bonn: United Nations Framework Convention on Climate Change.

UNFCCC. (AR-AM0004/Version 04) Approved afforestation and reforestation baseline and monitoring methodology AR-AM0004 Reforestation or afforestation of land currently under agricultural use. UNFCCC, EB 50 Report, Annex 21 version 01. Available online: https://cdm.unfccc. int/EB/050/eb50_repan21. pdf.

Waide, J., 1988. Forest ecosystem stability: revision of the resistance-resilience model in relation to observable macroscopic properties of ecosystems. In: Forest Hydrology and Ecology at Coweeta. Springer, pp. 383-405. Wilson Rowe, E., 2015. Locating international REDD+ power relations debating forests and trees in international climate negotiations. Geoforum 66, 64-74.