Final Report Summary - REDD-ALERT (Reducing Emissions from Deforestation and Degradation through Alternative Landuses in Rainforests of the Tropics)
Project Context and Objectives:
Context
Climate change is widely recognised as the most serious environmental threat facing our planet today. The Intergovernmental Panel on Climate Change (IPCC) published its Fourth Assessment Report (AR4) in 2007 which concluded that warming of the earth's climate is now indisputable, and that it is very likely that this is due to emissions of greenhouse gases (GHGs) from human activities, particularly from the last half of the 20th century onwards. Atmospheric concentrations of the GHGs, which include carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O), are higher than any time over the last 650,000 years.
We urgently need, therefore, to find ways of reducing our emissions of these gases. The Kyoto Protocol, agreed in 1997 at the third session of the Conference of Parties (COP-3) to the United Nations Framework Convention on Climate Change (UNFCCC), and effective from February 16, 2005, obliged participating developed countries (Annex I countries) to collectively reduce their national GHG emissions by 5.2% below 1990 levels, averaged over the period 2008-2012. Developing countries (non-Annex I countries) do not have such obligations, but were able to participate in the Clean Development Mechanism (CDM), whereby Annex I countries were able to purchase credits for projects aimed at reducing GHG emissions in non-Annex I countries.
While fossil fuel use remains the dominant concern, conversion of forests into agricultural land is a major source of GHG emissions. Currently, a gross figure of 13 million ha of forests are lost annually (FAO, 2006), with net losses, allowing for afforestation and reforestation, at about 7.3 million ha y-1 (Nabuurs et al., 2007). Degradation, defined as decrease of density or increase of disturbance in forest classes, affected tropical regions at a rate of 2.4 million ha y-1 in the 1990s. Recent estimates are that deforestation released 3.0 Gt CO2 y-1 (range: 2.1-4.5 Gt CO2 y-1) into the atmosphere over the period 2000-2005 (Harris et al., 2012), both through the burning of the forest biomass, and from the oxidation of carbon stored in the soil under the trees during cultivation and in peatlands under drainage. Other GHGs, such as CH4 and N2O may also be emitted during slash-and-burn and subsequent land use. This represents an estimated 7 to 14% of total global anthropogenic CO2 emissions over the time period analysed (Harris et al., 2012). Brazil and Indonesia are the countries with the highest net deforestation rates, losing 3.3 and 0.7 Mha of forest annually (Hansen et al., 2010), and as such, total per capita CO2 emissions in Indonesia, for example, may be 30% above the average for Europe, despite a much lower energy use.
It was decided at the 7th COP in Marrakech in 2001 that CDM mechanisms of the Kyoto Protocol be largely restricted to the energy and industrial sectors. However, surrounded by many safeguards, afforestation and reforestation activities became eligible as Afforestation/Reforestation Clean Development Mechanisms (A/R CDM), although they have found little application as yet. Emission reduction by avoided deforestation was excluded because institutions were not ready to deal with the "additionality" (what would have occurred without intervention), "leakage" (emission reductions in one location causing emission increases in another) and "permanence" issues at either local or national scale. Apart from ambiguities in the definition of a forest (which also affect the A/R CDM rules), there are difficulties in accurate monitoring of the carbon stocks actually preserved. Part of forest conversion is due to planned development, part is linked to climatic extreme events, and part is beyond control of national governments. The costs of both applying and verifying rules more complex than those for A/R CDM rules, were expected to be too high, while there was uncertainty on the opportunity cost of clearing forests for land uses with higher economic returns.
However, there was a growing recognition that the contribution to global emissions from deforestation should not be neglected, particularly with the realization that "solutions" for Annex I countries, such as increased use of bio-fuels, had led to perverse incentives that were likely to have increased deforestation rates and global GHG emissions rather than reduced them. Stern (2007) concluded that reducing deforestation was a highly cost-effective way to reduce emissions relatively quickly, as well as providing co-benefits in terms of soils, water, climate protection, protection of biodiversity and livelihoods, rights of local communities, and sustainable forest management. Indeed, one study had suggested that with appropriate carbon pricing, emissions from deforestation could be stopped by 2020 (Rokityanskiy et al., 2007). Stern recommended that, with help from the international community, policies on deforestation should be shaped and led by the nations where the forests stand, and that compensation from the international community should be provided to take account of the opportunity costs of alternative uses of the land, the costs of administering and enforcing protection, and managing the transition.
At the 9th UNFCCC Conference of the Parties (COP-9), Santilli et al. (2005) presented a proposal for Compensated Reductions addressing some of these issues, in which non-Annex 1 countries such as Brazil or Indonesia could voluntarily choose to reduce their national emissions from deforestation. A certain annual rate of deforestation based on a historical period would be permitted and used as a baseline, but reductions in deforestation rate below this rate would then gain carbon credits at the end of a commitment period which could be sold on international carbon trading markets or to other governments, thus earning income for the country. As this was to be at the national level, it would help to address the issues of national leakage, as displacement of emissions to elsewhere in the country would still be accounted for in national inventories, although this would not be the case for international leakage (Mollicone et al., 2007). At a workshop in Bogor, Indonesia, in 2005, Schlamadinger et al. (2005) assessed the Santilli et al. (2005) proposal in more detail, and proposed improvements to some of the shortcomings that they perceived. These included the possible need for upfront financing to establish avoided deforestation schemes (with appropriate safeguards built in to ensure that emission reductions were actually delivered), the appropriate setting of baselines, and ways in which revenues generated from avoided deforestation could be actually used to address the drivers of deforestation at the local level (i.e. to the landowners who would have to change their behaviour). Revenues from avoided deforestation would have to be set against other income possibilities for the land. Possible mechanisms mentioned included a carbon tax if landowners wished to deforest for other land uses, payments to the landowners not to deforest, or investments in improving neighbouring agricultural productivity so that deforestation was not required.
While these proposals marked significant progress in the thinking, there were still many issues to resolve. These included how appropriate baseline deforestation rates were determined, how differences from these baselines would be measured, and how the uncertainties in these differences would be quantified. The sensitivity of these uncertainties of subsequent credits generated (particularly in relation to degradation as opposed to deforestation), and the implications of this for carbon markets and efficiency of emissions reduction, needed to be examined. Techniques for monitoring the change in national and regional carbon stocks caused by deforestation and degradation needed to be improved. Mixed land-use mosaics presented a specific challenge in this regard. Concerns were also expressed regarding potential loss of national or provincial sovereignty over natural resources, and that avoided deforestation schemes could create so many carbon credits that they would flood global carbon markets and trigger a price collapse that would undermine the market incentives for reducing emissions in other sectors (e.g. energy).
The Kyoto Protocol expires in 2012, and it was intended that a successor agreement should be finalized at the 15th COP in Copenhagen (2009). At the 11th COP in 2005, it was agreed that there should be a two-year period of discussion about Reduced Emissions from Deforestation and Forest Degradation (REDD), focusing on ß??relevant scientific, technical and methodological issues, and the exchange of relevant information and experiences, including policy approaches and positive incentives". At the COP-13 meeting in Bali in December 2007, REDD was a key agenda item, and it was decided that all countries should work towards improving data collection, estimation of emissions from deforestation and degradation, monitoring and reporting, and addressing institutional needs of developing countries. The drivers of deforestation should also be addressed, with a view to reducing emissions from this source through a range of policy approaches and positive incentives. Under the auspices of the UNFCCC Subsidiary Body for Scientific and Technological Advice (SBSTA), countries were invited to submit their views by 21 March 2008 on outstanding methodological issues such as assessment of forest cover and the associated carbon stocks and GHG emissions, how these are affected by sustainable forest management (SFM), and how reductions in emissions from deforestation and degradation below reference baselines (without displacement) could be demonstrated (see Herold & Johns, 2007). The coming two years would thus shape the post-2012 regime and the rules by which common but differentiated responsibility and accountability for climate change would be linked to global commitments to support development (Millennium Development Goals).
However, what had been largely missing from the debate at that stage, which Schlamadinger et al. (2005) touched upon but did not go into detail, was how emission reduction targets agreed at the national scale would be translated into changes of behaviour relating to land use at the local scale, while not affecting the rights of minority and vulnerable social groups, or the provision of ecosystem services in general. New ways to link the technical and institutional advances on REDD to local stakeholders were needed so that the various scenarios considered reflect local ambitions and response options. Social justice questions such as who will be the winners and losers (particularly in relation to common land), and how to avoid rewarding the "bad guys" while forest dwellers are evicted from their homes or forced to abandon their agricultural activities, would all need to be addressed. Active "negotiation support" was needed to achieve the "free and prior informed consent" status that is seen as a moral imperative to agreements that potentially affect livelihoods of people outside of the centres of political power. Moreover, a landscape perspective was required due to the spatial interdependencies of forest, agriculture and other land uses leading to deforestation and degradation at the forest margins ß?? it is not a "forest problem" alone. Solutions are likely to be quite context-specific and to vary from one country to the next, but without these crucial links between global and local, the climate policy discussions ran the risk of divorcing themselves from reality. The aim of the REDD-ALERT project, therefore, was to make a contribution to the evaluation of mechanisms that translate international-level agreements into instruments that would help change the behaviour of the people at the "coal-face" while minimising adverse repercussions on their livelihoods.
Project Results:
Objective 1: Documenting the socio-economic drivers of deforestation and degradation
This section presents the main science / technology results from work carried out under WP1, by discussing (i) the current state of global land use and expected trends over the period 2000-2030, (ii) the knowledge on causes and impacts of past and ongoing forest transitions, and (iii) prospects and policy options for a global forest transition, and (iv) results from the study sites.
Forest transitions
Our work showed that a handful of developing, tropical countries have recently been through a forest transition, thus shifting from shrinking to expanding forests at a national scale (Meyfroidt & Lambin, 2011). Different views exist on de?¬ning secondary forests and on whether reforestation should include some or all forms of tree plantations, or only naturally regenerating forests, which adds to the technical difficulties in measuring and characterizing reforestation. A few generic processes of forest transition were identified, as well as the broad geographical patterns of reforestation (Rudel et al., 2005; Lambin & Meyfroidt, 2010; Meyfroidt & Lambin, 2011). Countries do not necessarily follow a regular pattern of forest cover changes, and the causes and outcomes of forest transitions vary, so that forest transition is to be seen as a contingent process. Under certain conditions, environmental degradation can be perceived by land managers and, through such social-ecological feedback, can in turn become a driver of subsequent land use changes and of forest transition (Meyfroidt et al., in preparation). Restoring forests in one country is generally associated with a significant outsourcing of forest exploitation to neighbouring countries via increased imports of wood and sometimes agricultural products (Meyfroidt & Lambin, 2010). In Vietnam, the combination of policies restricting forest exploitation, rapid development of the wood processing industry and of exports of wood products, led to an increase in legal and illegal imports of timber and a displacement of forest extraction to neighbouring countries, such as Laos and Cambodia, equivalent to 39% of the regrowth in Vietnam's forests from 1987 to 2006 (Meyfroidt & Lambin, 2010). Policies to protect forests and promote reforestation therefore need to control this displacement of land use and channel it toward areas where the impacts are minimal (or beneficial). The ecological effects of reforestation are highly dependent on the residual deforestation of old-growth forests, the proportions of natural regeneration of forests and tree plantations, and the location and spatial patterns of the different types of forests (Meyfroidt & Lambin, 2011). Furthermore, net reforestation can conceal a continuing degradation or clearance of partly irreplaceable old-growth natural forests (Echeverria et al., 2006; Meyfroidt & Lambin, 2008). Policies supporting afforestation and reforestation should not assume that it will lead indiscriminately to environmental gains.
Prospects and policies for sustainable forest transitions
Although in theory the trade-offs between conserving forests and feeding the world's population could be minimized (Angelsen, 2010), a decrease in the availability of productive land and competition with other land uses will make a global forest transition difficult to achieve over the coming decades (see Table 1). Policies to achieve a forest transition include approaches to improve the supply of land-demanding products, and to control the demand for them (Meyfroidt & Lambin, 2011). The following factors hold the potential to significantly affect the supply of and demand for wood and agricultural products, and therefore contribute to control deforestation by addressing its drivers: (i) technological innovations and more efficient land-use practices to intensify agricultural and forestry production and reduce its environmental impacts; (ii) sound land management policies to control for rebound-effects; and (iii) changes in consumption patterns especially reduction of wastes and decreasing demand of the most land-demanding products ß?? e.g. meat. The combination of, first, promoting nature-friendly farming in areas with biophysical and social conditions unsuitable for large-scale intensive farming; and second, sparing land for forests through agricultural intensi?¬cation in high-potential agricultural areas might control the expansion of competing agricultural land uses (Fischer et al., 2008; Lindenmayer & Cunningham, 2012). Increasing off-farm economy, especially in the most marginal rural areas, can contribute to reducing the dependency on local natural resources. For commodities with high elasticity such as biofuels, meat, and luxury goods like coffee or exotic timber, agricultural intensi?¬cation and out-migration are unlikely to reduce the overall demand for agricultural land unless combined with policies to control rebound effects, e.g. by land-use zoning and demand-side interventions. Addressing final consumption and future global demand for wood and agricultural products is a critical aspect for any potential global forest transition (Meyfroidt et al., 2010; Lambin & Meyfroidt, 2011; Meyfroidt & Lambin, 2011). Across all these approaches, for the poorest fraction of humanity, the issues of equity, and access and sharing of food and land resources are crucial to ensure food security (Godfray et al., 2010). These approaches may rely on various tools, including state-level command-and-control, regulatory tools; rural and agricultural development policies; and emerging market-based instruments. REDD+ could support most of these strategies, and thus should not be considered only as a program of Payments for Environmental Services for agents of deforestation and forest degradation.
Objective 2: Quantifying rates of forest conversion and change in forest carbon stocks using improved methods
The objectives of this work-package were to 1) quantify current land use change and deforestation, and 2) improve monitoring tools. Data collection in the context of national greenhouse gas inventories following IPCC guidelines has to balance the errors (uncertainty) involved in 1) the use of the classification system (both at the "ground truth" and "remote sensed image" level), 2) the uncertainty in the properties (C stock per unit area) of each class and 3) the eligibility of different land cover types with respect to the scope of the policy.
Deforestation rates depend on forest definition
While the legends we use for land cover classification include a range of systems with various quality and quantity of tree cover, the concept of "deforestation" splits it into a "forest" and "non-forest" part. However, different stakeholders have different operational forest concepts ß?? ranging from the interest in untouched old-growth forest of strict conservation agents, to the very "weak" forest concept that was agreed as part of the Afforestation/Reforestation Clean Development Mechanism (A/R-CDM), referring to the FAO definition of minimum size, potential to reach a minimum tree cover and clarification what is meant by tree. Across such range of definitions we found the deforestation rate for Indonesia as a whole to be 3-5%/year without clear upward or downward trend within the 1990-2010 time frame, or shifting from +0.5%/year to -0.5%/year in the same time period (Figure 2). As the international negotiation arenas have not been able to clarify the forest concept that is to be used in REDD, there is far too much scope for picking the operational definition that matches the point one wants to make. Deforestation as such cannot be used to predict "emissions from deforestation", and the second D of REDD is essential to quantify emissions; the results of net emissions are much less sensitive to forest definition ß?? as earlier shown for Vietnam where an increase in net forest area was found to coincide with a continued loss of forest carbon stock, as densely stocked forest continued to be lost and the forest gained had much lower carbon stocks.
Reformulating allometric equations may help judge need for tree-specific equations
Part of the uncertainty in carbon stock data for land use systems is the use of "generic" allometric equations that relate tree biomass to stem diameter, while details of tree architecture differ between trees. A common form of such equation is Biomass = a(Diameter)b. Empiricial data sets suggest that both the a and b parameter differ considerably between tree species, and that thus tree-specific equations are needed. We noticed, however, that the a and b parameters of fitted equations are strongly linked, and that for a tree of 20-30 cm diameter the variation in predicted biomass is small. Reformulating the allometric equation to Biomass = a2(Diameter/DiamStandard)b2, with a standard tree diameter (DiamStandard) of about 20 cm leads to much lower variation in a and b, but a variation in a and b that is independent of each other (so it becomes a clean two parameter equation). At stand level, where there typically is a range of tree diameters, the choice of equation matters much less than the tree level differences suggest, as there is strong compensation. These results will help to improve C accounting methods, and justify the use of generic allometric models as default.
Simplifying from IPCC 5-pool data for carbon stocks
The IPCC prescribes that five pools of terrestrial carbon need to be assessed: trees, understory vegetation, roots of all plants, litter plus necromass, and soil organic carbon. Based on over 700 sample points we found that the data collection can be replaced by the use of defaults, as variation between land use types in soil carbon (in top 30 cm of the profile) plus litter layer plus necromass plus understorey vegetation represents only 6% of variation in tree biomass, while the belowground root biomass is estimated to be 25% (with little opportunity to verify this estimate). Under specific conditions of recent forest damage, however, the necromass pool can be more than 50% of the total carbon stock, and cannot be ignored in data collection.
Error propagation in C stock estimates
Errors in land use classification (with typically 85% accuracy at pixel level for a 25-point legend) combine with uncertainty around the typical C stock data used. We analysed how the combined errors relate to scale. Typically, aggregation across pixels reduces the random element in the error, while not reducing any bias that may be there. For REDD applications we found that aggregating up to a 1 km2 scale reduces uncertainty in C stock change to below 5%, while estimates for smaller pixel sizes may have a higher chance of containing error. The acceptable error level for a government-based incentive system on performance in C stock change is not empirically known, so our 5% threshold may have to be revised in future.
Opportunity costs curves
The monitoring of land use change needs to relate the economic benefits that (local) actors derive, to the global consequences for emissions. Initial estimates by the ASB consortium in 2007 in Indonesia, Cameroon and Peru suggested that 85% of emissions yielded net economic benefits of less than 4ß?¬/t CO2 at current prices. Further analysis of the spatial determinants of such abatement costs focused on soil type (peat versus mineral soils), accessibility (rivers versus roads), actor (large scale operations versus smallholders) and policy domain (existing land use restrictions and forest protection rules). We refined the calculation procedure in the ABACUS tool, which was endorsed by the World Bank in training at national level, and was refined based on user feedback.
Recommendations on the design of national monitoring systems
Reflecting on the cost of data collection, the marginal reduction in net uncertainty of estimates of terrestrial carbon stock change, we recommend a national monitoring system that pays specific attention to the low frequency of high-diameter trees that contain a disproportionally high part of total carbon stocks, while simplifying data collection on pools that were found not to vary much. Soil sampling for typical land use types in a national accounting system may be worthwhile, but for smaller accounting areas the costs of data collection probably exceed any benefits that might be derived through carbon markets.
Objective 3: Improving GHG accounting methods and default values of land use change in tropical forests
The overarching objectives of this multi-partner work-package were twofold: (1) to quantify the magnitude of GHG emissions and changes in belowground soil carbon stocks resulting from the land-use change at tropical forest margins, and (2) to identify the controlling mechanisms underlying the GHG emissions in both deeply weathered mineral soils and peat soils. To address these objectives, the project used a comprehensive approach consisting of literature reviews, field experiments, regional sampling campaigns, and modelling.
Field measurements on mineral soils
We investigated the impact land-use change has on SOC stocks in deeply weathered mineral soils of three REDD-ALERT study regions: (a) Jambi province, Sumatra, Indonesia, (b) southern Cameroon and (c) Ucayali province, Peru. Using a space-for-time substitution sampling approach, we quantified SOC stocks in the top 3 m of soil and compared converted land-uses with adjacent reference forest plots. In each country, we investigated the most predominant land-use trajectories relevant for the respective region. These included conversions from forest to cash-crop tree plantations (rubber, oil palm and cacao), conversions from forest to cattle pastures, and the conversion from forest to shifting cultivation systems (a rotation of crop-fallow-crop). In total, 157 plots were established where soil samples were taken to a depth of 3 m from a central soil pit and from the topsoil from 12 pooled composite samples. All soil samples were analyzed in the laboratory at the University of Goettingen for C, N, pH and effective cation exchange capacity (ECEC).
This study found that the conversion of forests to intensively managed land-uses such as oil palm plantations, cacao plantations, rubber plantations and cattle pastures caused significant SOC stock losses in the topsoil (Table 2). In contrast, the less intensively managed shifting cultivation land-use exhibited minimal change in SOC stocks in the soil profile.
Despite the fact that most of the SOC from the 3 m profile is found below 1 m depth (50-60% of total SOC stock), this subsoil C stock remained relatively unchanged by the land-use conversion. The only exception was in cacao plantations where decreases in SOC stock were measured down to 3 m. These plantations were generally much older and it is speculated they had the time to reach a steady state condition throughout the soil profile. The results on the magnitude of SOC losses in the soil profile and the constraints regulating C concentrations and changes therein is expected to be published in early to mid 2013.
Measurements of GHG emissions from mineral soils took place in the province of Jambi, Sumatra, Indonesia. Along a gradient of forest disturbance including rubber and oil palm plantations on mineral soils, measurements of C stocks and soil trace gas emissions were conducted during the period July 2010-August 2011. Soil N2O emissions were slightly reduced after forest degradation and conversion to rubber plantations but increased after conversion to oil palm plantation due to nitrogen fertilizer application. Soil CH4 fluxes changed from sink to source with forest degradation and conversion to oil palm plantation because of increased termite nests presence in these land-uses. An analysis on how termite species differently influence soil fluxes of CH4 is being conducted. Fluxes of N2O were significantly correlated with CH4 fluxes; mineralisation and nitrification potentials of the soil mineral N content and with the soil water-filled pore space during wet months. Soil respiration was not significantly affected by land-use change. The results on soil and vegetation C stocks are still being processed, and several publications are expected from the research.
Field measurements on peat soils
Field research in Jambi was also implemented in peatlands. Carbon stocks in living and dead vegetation, above- and below-ground, were measured in a primary peat swamp forest, a logged and drained peat forest, and a seven-year old oil palm plantation on peat. Total biomass C stocks decreased from 220 ?± 7 Mg ha-1 in the primary forest to 100 ?± 2 Mg ha-1 in the logged drained forest, and 32 ?± 4 Mg ha-1 in the oil palm plantation. Woody debris contributed 7% to total C stock in the primary forest, and as much as 43% and 31% in the logged drained forest and the oil palm plantation, respectively. Coarse root biomass represented 14%, 10% and 21% of total C stock in the primary forest, logged drained forest, and oil palm plantation, respectively. In the same land-use treatments, the root:shoot ratios of trees/palms were 0.19 ?± 0.01 0.21 ?± 0.04 and 0.60 ?± 0.42. The research will be submitted to the journal Forest Ecology and Management in 2013. Soil respiration was also measured in these systems. Preliminary results of measurements conducted between January and September 2012 indicate emission rates of 16 ?± 1.2 18.5 ?± 0.7 and 28.4 ?± 1.2 Mg CO2-C ha-1 yr-1 in the primary forest, logged drained forest, and oil palm plantation, respectively.
A further five sets of field experiments to measure GHG emissions were conducted in five land-use types located in Jambi and Riau, Sumatra Island. The first study aimed to evaluate total soil respiration as a function of distance from oil palm trees in relation to root distribution. The study was conducted on peatland under 6 and 15 year old oil palm plantations in Jambi Province, Indonesia from June 2011 to April 2012 and from January to April 2012, respectively. CO2 emissions were measured using an infrared gas analyser (IRGA), Li-COR 820. Total respiration was two to three times higher at points 1 m compared to that at points ß?¥3 m and linearly decreased with distances from the trees. At distances of 3.0-4.5 m from the trees the emission no longer increased with distance, indicating negligible influence of root-related respiration. The emission values at this distance for the 6 and 15 year old oil palm plantations were 38 and 34 Mg CO2 ha-1 yr-1 respectively. CO2 flux measurements in oil palm plantations would adequately represent the heterotrophic respiration if made at a distance of 3 m or more from the tree centre.
The second study was focused on the variation of CO2 emissions at five land-use types including oil palm, acacia forest, rubber, secondary forest and bareland in Riau, Jambi and Aceh, Indonesia. We conducted a detailed study of CO2 emissions using an infrared gas analyzer (IRGA LI-COR 820 model). CO2 emissions (Mg ha-1 yr-1) under oil palm plantations in Riau, Jambi and Aceh ranged from 18 ?±13 to 66 ?±24 Mg ha-1 yr-1, with the average of about 39 ?±19 Mg ha-1 yr-1. For adjacent plots with maximum distance between plots of 3.2 km, in Kampar Peninsula, Riau, CO2 emissions from secondary forest, acacia forest, bareland, rubber and oil palm plantation were 60 ?±25, 60 ?±19, 56 ?±26, 52 ?±17, 66 ?±24 Mg ha-1 yr-1, respectively. Our results showed that CO2 emissions increased with the increase of water table depth. Our finding is comparable with other CO2 flux measurement and can contribute to reducing uncertainty of peat CO2 emission estimates.
In the third study, we examined effects of various levels of water contents and laterite application on microbial (heterotrophic) respiration of peat soil. Bulk samples of surface (0 20 cm depth) and subsurface (30-50 cm depth) layers were collected from an oil palm plantation in Riau Province, Indonesia. Peat water contents were adjusted at 20, 40, 60, 80, and 100% water-filled pore space (WFPS). Laterite soil (clay containing high Al and Fe oxides) was applied at 3, 6 and 12 mg g-1 dry weight (1.2 2.4 and 4.8 Mg ha-1) for peat samples at 60% and 100% WFPS. Peat respiration differed between the soil layers, and was distinctly affected by water content, but less affected by laterite application. Peat respiration increased sharply from wet (ß?¥80% WFPS) to moist soil (60 to 40% WFPS) and decreased when the soil became dry (ß?¤40% WFPS). Maximum peat respiration for surface and sub-surface layers occurred around 50% and 30% WFPS, respectively. Laterite as a peat ameliorant accelerates rather than reduces peat respiration and thus it cannot be used for CO2 emission reduction.
The fourth study related to the spatial variation of CO2 flux with water table depth, soil moisture, and temperature under oil palm plantation in Jambi Province, Sumatera, Indonesia. A total of 480 CO2 flux measurements were made using an infrared gas analyzer (IRGA) every three months, at six different time intervals in a day. The results showed that the average CO2 flux was 46 ?±30 t CO2 ha-1 yr-1 (n=480). Within the one year study, the average CO2 flux did not show a clear relationship with instantaneous water table depth, soil moisture, and temperature. A positive correlation between these variables and the CO2 flux only occurred in October 2010, coinciding with the beginning of the wet season. The distance measurement points from the edge of the canal showed a positive correlation between CO2 flux (R2=0.6) water table depth (R2=0.6) and soil moisture (R2=0.5). The factors driving the CO2 flux at the study site was very complex, each affecting each other and working simultaneously.
In the fifth study, CO2 flux measurements were made in intact forest, logged forest, and oil palm plantation sites using four methods ß?? sampling at specified distances from palms, ??13C isotope analysis, random collar total efflux sampling, and simulation modelling. Results indicated interference in the ??13C results ß?? likely from methane oxidation, which has never been considered in this method before. Average total CO2 emissions from sites were 29 t CO2 ha-1 yr-1 in the wet season in the intact forest and 71 t CO2 ha-1 yr-1 in the dry season. Logged forest emissions were double in the deeper peat sites (109 t CO2 ha-1 yr-1) than the shallow peat sites (57 t CO2 ha-1 yr-1). Across the different oil palm sites total emissions ranged between 33-98 t CO2 ha-1 yr-1 on the deep and shallow peats. Heterotrophic emissions (excluding rhizosphere respiration) in the oil palm sites averaged 15-38 t CO2 ha-1 yr-1. Modelled heterotrophic emissions were higher at 37 55 t CO2 ha-1 yr-1 (higher due to the inclusion of rhizosphere respiration), and modelled net emissions at 44-60 t CO2 ha-1 yr-1. Soil C sampling established peat depths and C stocks of all sites ß?? peat depths varied between 3-7 m across all sites. Final analysis on the intact sites peat stocks is still ongoing, but in the oil palm sites total C stocks ranged between 171-199 t C ha-1 m-1 depth.
Effects of fire on the organic matter composition of tropical peat
This part of the work was based on investigation of how fire modifies peat organic matter chemistry with particular regard to shifts in carbon pools. The project was conducted in the tropical peatlands of Kalimantan, Indonesia. The research began with initially assessing the organic matter composition of recently burnt peat using a variety of methods including Pyrolysis-GC/MS (Py-GC/MS). This qualitative method provides a chemical fingerprint of the organic matter composition.
The first section of work assessed the short term effects of fire by assessing organic matter composition one month post-fire. Structural differences between burnt and unburned (inundated and drained) peat samples suggest that a combination of both fire and drainage causes alteration of the OM composition that is evident shortly after fire. The main observations are summarised as follows: (i) Unburned, inundated peat pyrolysates contain contributions from all compound classes. Lignin products such as guaiacol, methyl guaiacol and ethyl guaiacol are dominant for unburned samples. (ii) Long term drainage conditions induce oxic conditions in the upper peat layers causing a reduction in OM diversity particularly below 30 cm from the surface. Drained peat pyrolysates are dominated by aliphatic components. (iii) Surface, recently burnt and drained sample pyrolysates are, however, composed predominantly of aromatic and aliphatic compounds and are significantly reduced or depleted in all other compound classes, including lignin and polysaccharide derived compounds. A high aromatic and aliphatic content, including a large contribution from n-alkene/alkane doublets, suggests that the burnt peat is highly refractory and that much of the labile component has been lost or converted to other C forms. (iv) Subsurface burnt and drained sample pyrolysates are associated with aliphatic moieties with considerable variance from surface samples. Differences in surface vs. subsurface burnt and drained samples are attributed to differences in the impact of fire and decomposition. (v) Although fires in tropical peatlands are often considered to be smouldering, thereby penetrating the peat substrate, data in this analysis suggest that effects on the OM were predominately experienced in the surface peat layer (0-5 cm). Any differences between intact inundated peat and subsurface peat in burnt drained sites are likely to be caused by decomposition processes.
The work proceeded to investigate the impacts of different fire regimes on the peat organic matter including the longer term effects of fire on organic matter chemistry. Data suggests that there are significant differences in organic matter composition between recently burnt peat and peat collected approximately 1.5 years post-fire; however, these differences were only significant between surface samples. Below 5 cm organic matter composition does not significantly vary. Shortly after fire (November 2009), sample pyrolysates of the upper 5 cm of peat are significantly different from those collected approximately 1.5 years later (April 2011). PCA analysis differentiates between recently burnt peat and peat which had not been burnt for several years. Initially burning had transformed the peat OM composition so that it became highly alipahtic with considerable contributions from aromatic compounds as well as pyrolysis products thought to indicate the presence of charcoal e.g. naphthalene. 1.5 years later sample pyrolysates remain highly aliphatic, with considerable contributions from alkenes and alkanes. Dimethylnaphthalene and trimethylnaphthalene (biomarkers of charred material) both plot in similar factor space when assessing variance using PCA, the presence of such compounds 1.5 years after fire suggests charred organic matter is still present but less dominant than in those samples analysed shortly after the fire was extinguished. However the overall OM diversity (i.e. number of pyrolysis products identified, (111 Vs 72-75) has increased and there is evidence for some labile components such as cellulose (as indicated by levoglucosan) and furans which were depleted immediately after the fire. The proportion of oxygen in surface sample pyrolysates increases during the 1.5 year recovery period (3.9 % Vs 0.75 ?± 0.35%) which is due to the increased abundance of oxygen containing functional groups in polysaccharide and lignin derived compounds. In summary, peat fires cause considerable alteration to peat organic matter however these predominately impact the upper 5 cm of peat ß?? with less significant change at depth. The peat organic matter composition becomes more diverse 1.5 years after fire but even 14.5 years post-fire the organic matter does not recover back to its natural undisturbed state. Therefore it is suggested that fire has long term effects on peat organic matter however much of this change, albeit on a shorter temporal scale, mirrors compositional changes caused by long term peat drainage.
Understanding how tropical peatland fires modify peat OM composition can contribute to an improved knowledge of post-fire carbon cycling and nutrient cycling. Thermally induced alteration of peat OM may be inhibiting ecosystem recovery, ultimately influencing the global carbon balance. The results of this work also demonstrates that fire has caused the most labile OM components to be removed from the upper peat layer as well as the neoformation of aromatic structures, thus increasing the recalcitrance of the peat substrate. Changes in such carbon pools are likely linked to peat soil respiration rates as a recalcitrant peat effects by fire is more resistant to microbial decomposition than a labile peat- not affected by fire. Alteration of the OM composition may also have implications for physical properties on peat soil, for example enhanced water repellence, which would impact upon surface runoff and solute transport.
Objective 4: Identifying and assessing viable policy options addressing the drivers of deforestation
Drivers
The research in WP4 has shown that most instruments in the forest sector deal only with proximate drivers, and often in an inequitable manner. Very few deal with underlying drivers, and to the extent that they do so their effect is limited (e.g. debt for nature swaps, certification). However, the very nature of the underlying drivers (e.g. demographic, economic, technological, political or cultural trends) is that they are often slow processes operating at national or global levels resulting from the aggregated behaviour of many regional, national, subnational and individual entities (in some cases), referred to in the Panarchy literature as slow variables. Effecting change in these slow variables is either difficult due to their inertia, or unpredictable due to chance interactions with faster changing lower scale variables, and may therefore be beyond the power of any one of these entities to address. Collective action at the global level is clearly required, but there are often conflicting national interests (usually economic) that weaken the international resolve to find solutions to global environmental problems.
One way, however, to address underlying drivers of deforestation at the national level may be to mainstream forest protection into development paths. Forest policy needs to be integrated into sustainable development. This may not be easy in early stages of development or in early stages of the forest transition curve ß?? as the motivation to develop often comes at the cost of resource extractions from forests and forest land. The implication of this is that without substantial financial and institutional support from external actors, mainstreaming forests into forest governance in countries where a large percentage of land is under forests is unlikely.
Systems approaches
Regardless of the degree of deforestation present in a country, for lasting solutions to be developed, it is essential to see forests as components of larger systems of land use, which also include arable agriculture, grasslands, wetlands, and human settlements. Deforestation is only one of several major problems that humans need to grapple with in the next century - together with concomitant increases in demand for food, water and energy against a backdrop of climate change, urbanisation, and limited land resources, this has been referred to as the "perfect storm". Dealing with any one land use component (such as forests) in isolation is likely to result in partial solutions at best as the Law of Unintended Consequences starts to operate. Providing alternative employment opportunities to reduce dependence of people on forests for their livelihoods, for example, may result in increased GHG emissions from the industrial sector.
Although REDD began as an initiative to reduce deforestation, it is becoming apparent that if the global community is serious about reducing net GHG emissions, a systems approach is needed. This systems approach should take into considerations the many and complex interactions between forestry, agriculture and urban sectors, so as to optimise the fluxes of carbon and other nutrients, as well as social and economic flows, between different parts of the landscape. In this context, some even ask whether REDD funding might be effectively used for funding agricultural research to increase production on existing agricultural land, reduce losses in the food chain, and return nutrients from urban to rural areas, thereby reducing the need to clear further forests. Investing in agricultural research is a mitigation strategy in its own right ß?? it has been estimated that the avoided emissions through agricultural intensification since the 1960s have only cost around $4/tCO2e saved, cheaper than many other abatement measures.
Measures to ensure REDD+ success
To the extent that REDD can provide temporary relief to the climate change problem, we have proposed twelve generic measures that can help to embed REDD within the domestic forest institutions and have shown how these generic measures have contextual implications in Vietnam, Indonesia, Cameroon and Peru.
Future challenges
Countries are gradually investing in becoming ready for REDD projects in the hope that it will develop into a win-win situation. However, there are indications that if REDD is poorly designed and/ or implemented, it may turn into a lose-lose situation. But knowledge is power, and awareness of the risks may help countries and social actors mobilize themselves to prevent such risks. Furthermore, there is a possibility that the awareness of the actual costs of implementing forest conservation equitably may lead many to see this as less than cost-effective, and thus reduce the incentive to invest in. REDD faces considerable challenges; if these challenges are not rapidly addressed, REDD may disappear from the global agenda in its current form as suggested by the issue attention cycle.
Its enduring legacy, however, will be that it has mobilized global attention and social actors on the need to understand forests and human-forest interactions. There is far greater understanding of the drivers of deforestation in different countries and the limits of instruments in dealing with these forests. There is a realization that there is need to take a systemic view and see if mainstreaming forests in national development, agricultural, energy and mining policy leads to understanding how best to deal with forests. The next step thus is to understand how to mainstream forests and see if countries can reconcile their need and right to develop with the need to protect forests not just for themselves, but for the myriad local ecosystem services that they provide as well as for the global ecosystem services that they provide. This knowledge and the mobilization of a large number of actors may be in itself critical for creating a more comprehensive, equitable, legitimate, effective, efficient and enduring system for dealing with forests. As in the myth of Sisyphus, it will remain an uphill struggle.
Objective 5: Analysing policy scenarios on GHG emission reductions, land use and livelihoods
Assessing agent-based models
Agent-based simulation modelling has recently been arousing interest, due to its ability to model individual decision-making entities and their interactions, to incorporate social processes and non-monetary influences on decision-making, to conceptually reproduce non-linearities ("tipping points") often observed in space-time processes of innovation and change, and to dynamically link social and environmental processes.
As a starting point, Matthews & Dyer (2011) discussed whether traditional neo-classical economics models were adequate to analyse all policy instruments that might be deployed in a REDD+ context. It is now recognised that decision-making is influenced by many other factors beyond those considered by traditional economics (Geist & Lambin, 2002). Akerlof & Schiller (2009), for example, list a number of "animal spirits" that include confidence, fairness, corruption, pride, shame, neighbour influence, and social pressure, that are also influential, all of which will be relevant to a household's willingness to engage with REDD schemes. For example, even if it is economically rational, people might be reluctant to participate if their previous experience with local government has given them little confidence in whether they will actually receive compensation for not deforesting, despite alluring promises from officials. To account for these non-economic factors, van Vugt (2009) has proposed a 4Is framework of Incentives, Information, Identity, and Institutions, that influence people's decisions. Incentives in this framework include any incentives that enhance a household's assets.
Gathering data on these decision-influencing factors, however, is a significant challenge, particularly in developing countries. However, Matthews & Dyer suggested that each aspect in itself needs to be explored in depth to provide greater insight into decision-making processes and their implications than we have at present. At some time in the future it might be necessary to combine the important factors, particularly if interactions between them are being explored. A first step would be to include these factors into the models that we have already, and use sensitivity analysis to determine the likely influence they have on the overall dynamics of the system and the conclusions reached. Efforts could then be focused on obtaining information only for those factors that produce qualitatively different conclusions from traditional approaches.
Villamor et al. (2011) then reviewed different agent-based modelling approaches and the extent to which they consider the diversity factors inherent to landscape mosaics. Diversity challenges start from the representation of agents (i.e. aggregate or individual, and types), the agent decision making rule (i.e. social to less social factors approach), the agents' interaction medium (i.e. agent-to-agent, and/or agent with environment), and the integration of biophysical processes. They then discussed the challenges in integrating the non-economic motivations discussed above into the decision making of human agents, concluding that no comprehensive methodology is currently available for modelling of this type, and suggest that objective and interdisciplinary criteria should be used when scoping and constructing a model, and that the model should be based on the ultimate purpose of the outputs for planners or local people.
Modelling ecosystem carbon stocks
Most of the literature review and meta-analysis in WP3 (Powers et al., 2011) related to carbon stock changes associated with land-use changes in the tropical forest margins were partial with the analysis of either only NPP productivity or aboveground biomass or soil changes. To understand the effect of land-use change on the overall carbon stock changes, we need a whole system approach in which carbon in soils, roots, detritus, and other aboveground components are estimated following the pattern of land-use change. No review studies to our knowledge have been done to establish such relationships between the ecosystem carbon (sum of all organically derived carbon in various pools: both aboveground and belowground including soil) and various climatic or soil factors. We decided, therefore to extend this to also include above-ground carbon pools, using both a review of the literature and models for simulating the forest transition, to derive simple relationships between environmental variables such as temperature, water and soil variables such as clay and SOC content with the ecosystem carbon stock under a land-use or land-use change following deforestation in tropical forest regions.
The results of the study show only limited studies or data are still available that can estimate the carbon pools at an ecosystem level and we need more field studies to fill the gap. A simple regression model was derived using a small set of parameters, and was found to be useful to estimate the C stock changes associated with the land-use changes in the tropical forest margins. The results of the regression analysis showed that mean annual temperature is the most important single variable which explains much of the variability in carbon pools and pool changes under various landuse and landuse changes. However, there is a need to validate the model with a wider range of environmental and soil conditions before they can be used with confidence. Above- and below-ground measurements of carbon stocks were subsequently made(within WP2 and WP3) in three of the study sites ß?? Peru, Cameroon and Indonesia ß?? and at the time of writing are still being analysed. A draft paper is in preparation and is expected to be completed early in 2013.
Objective 6: Developing and using new negotiation support tools
The objective of this work-package was to seek the direct engagement of local, national and international stakeholders in the emerging mechanisms to reduce emissions from deforestation and degradation (REDD), by 1) synthesizing their perspectives, and 2) exploring with the stakeholders how the various REDD+ mechanisms (combining positive and negative incentives) could work out in practice. Role-play versions of process-based simulation models allowed for active engagement with "agents of change" that provide incentives to increase or decrease emissions. In parallel, 3) agent-based modelling and simulations explored the efficiency/fairness trade-off in REDD+ incentives and sought to establish benefit-sharing solutions that are acceptable to all.
Emission reduction primarily requires a shift in development trajectory for tropical forest margins, dealing with multiple actors, multiple incentives and multiple knowledge types. Actual economic benefits derived from land use changes that cause emissions have been low for forest-to-pasture conversion in the Amazon and for most types of land use change on peatlands in SE Asia. Even where such "opportunity costs" (net economic benefits of land use change foregone) are below a feasible C price, actual shifts in development trajectory require incentives that work for all, are within acceptable concepts of fairness and efficiency for all who might otherwise sabotage the process.
Four approaches to opportunity cost estimates
We compared the methods for four approaches to opportunity cost curves and advised where in the process of negotiating actual contracts that include Free and Prior Informed Consent, information needs to shift from (I) retrospective costs of system level change, and (II) retrospective analysis at pixel level, to forward looking scenario comparisons at landscape (III) or agent (IV) level.
Main dissemination activities and exploitation of results
The work carried out in the project is being disseminated through various peer-reviewed and other publications, as well as various activities, including broad audience publications, policy briefs, conferences and other presentations (see Tables A1 and A2). Here we present just some of the major activities which included presentation of project results at side events of the United Nations Framework Convention on Climate Change (UNFCCC) Conference of Parties (CoP) meetings to which many negotiators attended and contributed to the discussions.
UNFCCC CoP-14, December 2008, Poznan, Poland: This was before the project commenced, but the following talks were given in anticipation of the project:
1. "Introducing new initiatives to support implementation of REDD". Talk by Robin Matthews at ASB/MLURI side event at UNFCCC CoP-14 Forest Day, Dec 6, 2008.
2. "REDD-ALERT: Investigating ways of linking global climate policies to local behaviour change in tropical rainforest countries". Invited talk at EU side-event "Climate change research and observations beyond Europe", Dec 10, 2008.
Copenhagen Climate Change conference Mar 10-12, 2009: "REDD-ALERT: Evaluating global-level climate policy options and their local level implementation". Poster presented by R B Matthews at Copenhagen Climate Change conference.
UNFCCC CoP-15, December 2009, Copenhagen, Denmark: A side-event was organised by the EC DG-RTD entitled "Deforestation, Forest Conservation and the Climate Change Challenge" exploring the social, economic and environmental drivers of deforestation, their interdependence and consequences in South-East Asia, Africa and Latin America. Presentations included:
1. Welcome and introduction by Dr. A. Kentarchos (Climate Change and Environmental Risks Unit, DG-Research, European Commission)
2. "Solving the climate challenge within the planetary boundaries" by Prof. J. RockstrG¶m (Director of Stockholm Resilience Centre at Stockholm University, Director of Stockholm Environment institute)
3. "From deforestation to reforestation: conditions for sustainable land use" by Prof. E. Lambin (UniversitG© Catholique de Louvain, Belgium)
4. "REDD-ALERT: linking global climate arrangements to local land-use behaviour" by Dr. R. B. Matthews (Macaulay Institute, Aberdeen, UK)
5. "Opportunity costs of carbon emissions from land-use change: need to broaden scope of REDD" by Dr. M. van Noordwijk (ICRAF - regional coordinator for SE Asia)
6. Panel discussion ß?? debate. Moderator: Dr. E. Lipiatou, (Head of Unit, Climate Change and Environmental Risks Unit, DG-Research, European Commission)
A video-clip of this side-event can be viewed at http://ec.europa.eu/avservices/video/player.cfm?sitelang=en&ref=66868
UNFCCC CoP-18, December 2012: A side-event was organised by Dr Robin Matthews at the EU Pavilion at CoP-18 in Doha, Qatar on 29 November 2012 entitled "Is the window of opportunity for REDD+ closing?". Presentations included:
1. Welcome and introduction by Dr Luca Perez (Climate Change and Environmental Risks Unit, DG-Research, European Commission)
2. "Linking global climate arrangements to local land-use behaviour" by Dr Robin Matthews (Theme Leader, James Hutton Institute, Aberdeen, UK)
3. "Understanding, measuring and governing changes in forest carbon stocks in complex landscapes" by Dr Ole Mertz (I-REDD+ Coordinator, University of Copenhagen, Denmark)
4. "Diversity of land use trajectories and implications for REDD+" by Dr Daniel MG?ller (Senior Scientist, Humboldt UniversitG¤t zu Berlin and IAMO)
5. "Progress in determining reference emissions levels" by Dr Lou Verchot (Programme Director, CIFOR)
6. "Nationally Appropriate Mitigation Actions for the forests and other land uses of Indonesia: complementarity of policy instruments, funding streams and motivation" by Dr Meine van Noordwijk (Chief Scientific Advisor, ICRAF)
An article was prepared and published in the December 2011 issue of the stakeholder magazine International Innovation (Environment): Matthews, R.B. 2011. Evaluating local impacts from global GHG policies. In: International Innovation (Environment: December 2011 issue), pp. 64-66.
CNN Television broadcast a brief documentary on REDD-ALERT project activities by the Centre for International Forestry Research (CIFOR) and the World Agroforestry Center (ICRAF) in Indonesia. The programme was aired with the title 'Indonesia aims to halt deforestation' on 27 November 2010, which can be viewed using the following link:
http://www.cnn.com/2010/WORLD/asiapcf/11/22/indonesia.halt.deforestation/index.html
It is planned to publish a selection of papers from the REDD-ALERT project in a Special Issue of Mitigation and Adaptation Strategies for Global Change.
A full list of dissemination outputs is given in the Use and dissemination of foreground ß?? Section A below.
List of Websites:
Project website: http://www.redd-alert.eu/
List of contacts
Beneficiary Number Beneficiary name Beneficiary short name Country Contact
1 Macaulay Land Use Research Institute MLURI UK robin.matthews@hutton.ac.uk
2 UniversitG© Catholique de Louvain UCL BELGIUM eric.lambin@uclouvain.be
3 Vrije Universiteit Amsterdam VU NETHERLANDS onno.kuik@ivm.vu.nl
4 Georg August University of GG¶ttingen UGOE GERMANY eveldka@gwdg.de
5 World Agroforestry Centre ICRAF KENYA m.vannoordwijk@cgiar.org
6 Centre for International Forestry Research CIFOR INDONESIA l.verchot@cgiar.org
7 International Institute of Tropical Agriculture IITA NIGERIA j.gockowski@cgiar.org
8 Centro Internacional de Agricultura Tropical CIAT COLUMBIA g.hyman@cgiar.org
9 Indonesian Soils Research Institute ISRI INDONESIA fahmuddin_agus@yahoo.com
10 Research Centre for Forest Ecology and Environment RCFEE VIETNAM phuong.vt@rcfee.org.vn
11 Institut de Recherche Agricole pour le DG©veloppement IRAD CAMEROON mtchienko@yahoo.com
12 Instituto Nacional de Investigacion y Extension Agraria INIA PERU jcuellar@inia.gob.pe
Context
Climate change is widely recognised as the most serious environmental threat facing our planet today. The Intergovernmental Panel on Climate Change (IPCC) published its Fourth Assessment Report (AR4) in 2007 which concluded that warming of the earth's climate is now indisputable, and that it is very likely that this is due to emissions of greenhouse gases (GHGs) from human activities, particularly from the last half of the 20th century onwards. Atmospheric concentrations of the GHGs, which include carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O), are higher than any time over the last 650,000 years.
We urgently need, therefore, to find ways of reducing our emissions of these gases. The Kyoto Protocol, agreed in 1997 at the third session of the Conference of Parties (COP-3) to the United Nations Framework Convention on Climate Change (UNFCCC), and effective from February 16, 2005, obliged participating developed countries (Annex I countries) to collectively reduce their national GHG emissions by 5.2% below 1990 levels, averaged over the period 2008-2012. Developing countries (non-Annex I countries) do not have such obligations, but were able to participate in the Clean Development Mechanism (CDM), whereby Annex I countries were able to purchase credits for projects aimed at reducing GHG emissions in non-Annex I countries.
While fossil fuel use remains the dominant concern, conversion of forests into agricultural land is a major source of GHG emissions. Currently, a gross figure of 13 million ha of forests are lost annually (FAO, 2006), with net losses, allowing for afforestation and reforestation, at about 7.3 million ha y-1 (Nabuurs et al., 2007). Degradation, defined as decrease of density or increase of disturbance in forest classes, affected tropical regions at a rate of 2.4 million ha y-1 in the 1990s. Recent estimates are that deforestation released 3.0 Gt CO2 y-1 (range: 2.1-4.5 Gt CO2 y-1) into the atmosphere over the period 2000-2005 (Harris et al., 2012), both through the burning of the forest biomass, and from the oxidation of carbon stored in the soil under the trees during cultivation and in peatlands under drainage. Other GHGs, such as CH4 and N2O may also be emitted during slash-and-burn and subsequent land use. This represents an estimated 7 to 14% of total global anthropogenic CO2 emissions over the time period analysed (Harris et al., 2012). Brazil and Indonesia are the countries with the highest net deforestation rates, losing 3.3 and 0.7 Mha of forest annually (Hansen et al., 2010), and as such, total per capita CO2 emissions in Indonesia, for example, may be 30% above the average for Europe, despite a much lower energy use.
It was decided at the 7th COP in Marrakech in 2001 that CDM mechanisms of the Kyoto Protocol be largely restricted to the energy and industrial sectors. However, surrounded by many safeguards, afforestation and reforestation activities became eligible as Afforestation/Reforestation Clean Development Mechanisms (A/R CDM), although they have found little application as yet. Emission reduction by avoided deforestation was excluded because institutions were not ready to deal with the "additionality" (what would have occurred without intervention), "leakage" (emission reductions in one location causing emission increases in another) and "permanence" issues at either local or national scale. Apart from ambiguities in the definition of a forest (which also affect the A/R CDM rules), there are difficulties in accurate monitoring of the carbon stocks actually preserved. Part of forest conversion is due to planned development, part is linked to climatic extreme events, and part is beyond control of national governments. The costs of both applying and verifying rules more complex than those for A/R CDM rules, were expected to be too high, while there was uncertainty on the opportunity cost of clearing forests for land uses with higher economic returns.
However, there was a growing recognition that the contribution to global emissions from deforestation should not be neglected, particularly with the realization that "solutions" for Annex I countries, such as increased use of bio-fuels, had led to perverse incentives that were likely to have increased deforestation rates and global GHG emissions rather than reduced them. Stern (2007) concluded that reducing deforestation was a highly cost-effective way to reduce emissions relatively quickly, as well as providing co-benefits in terms of soils, water, climate protection, protection of biodiversity and livelihoods, rights of local communities, and sustainable forest management. Indeed, one study had suggested that with appropriate carbon pricing, emissions from deforestation could be stopped by 2020 (Rokityanskiy et al., 2007). Stern recommended that, with help from the international community, policies on deforestation should be shaped and led by the nations where the forests stand, and that compensation from the international community should be provided to take account of the opportunity costs of alternative uses of the land, the costs of administering and enforcing protection, and managing the transition.
At the 9th UNFCCC Conference of the Parties (COP-9), Santilli et al. (2005) presented a proposal for Compensated Reductions addressing some of these issues, in which non-Annex 1 countries such as Brazil or Indonesia could voluntarily choose to reduce their national emissions from deforestation. A certain annual rate of deforestation based on a historical period would be permitted and used as a baseline, but reductions in deforestation rate below this rate would then gain carbon credits at the end of a commitment period which could be sold on international carbon trading markets or to other governments, thus earning income for the country. As this was to be at the national level, it would help to address the issues of national leakage, as displacement of emissions to elsewhere in the country would still be accounted for in national inventories, although this would not be the case for international leakage (Mollicone et al., 2007). At a workshop in Bogor, Indonesia, in 2005, Schlamadinger et al. (2005) assessed the Santilli et al. (2005) proposal in more detail, and proposed improvements to some of the shortcomings that they perceived. These included the possible need for upfront financing to establish avoided deforestation schemes (with appropriate safeguards built in to ensure that emission reductions were actually delivered), the appropriate setting of baselines, and ways in which revenues generated from avoided deforestation could be actually used to address the drivers of deforestation at the local level (i.e. to the landowners who would have to change their behaviour). Revenues from avoided deforestation would have to be set against other income possibilities for the land. Possible mechanisms mentioned included a carbon tax if landowners wished to deforest for other land uses, payments to the landowners not to deforest, or investments in improving neighbouring agricultural productivity so that deforestation was not required.
While these proposals marked significant progress in the thinking, there were still many issues to resolve. These included how appropriate baseline deforestation rates were determined, how differences from these baselines would be measured, and how the uncertainties in these differences would be quantified. The sensitivity of these uncertainties of subsequent credits generated (particularly in relation to degradation as opposed to deforestation), and the implications of this for carbon markets and efficiency of emissions reduction, needed to be examined. Techniques for monitoring the change in national and regional carbon stocks caused by deforestation and degradation needed to be improved. Mixed land-use mosaics presented a specific challenge in this regard. Concerns were also expressed regarding potential loss of national or provincial sovereignty over natural resources, and that avoided deforestation schemes could create so many carbon credits that they would flood global carbon markets and trigger a price collapse that would undermine the market incentives for reducing emissions in other sectors (e.g. energy).
The Kyoto Protocol expires in 2012, and it was intended that a successor agreement should be finalized at the 15th COP in Copenhagen (2009). At the 11th COP in 2005, it was agreed that there should be a two-year period of discussion about Reduced Emissions from Deforestation and Forest Degradation (REDD), focusing on ß??relevant scientific, technical and methodological issues, and the exchange of relevant information and experiences, including policy approaches and positive incentives". At the COP-13 meeting in Bali in December 2007, REDD was a key agenda item, and it was decided that all countries should work towards improving data collection, estimation of emissions from deforestation and degradation, monitoring and reporting, and addressing institutional needs of developing countries. The drivers of deforestation should also be addressed, with a view to reducing emissions from this source through a range of policy approaches and positive incentives. Under the auspices of the UNFCCC Subsidiary Body for Scientific and Technological Advice (SBSTA), countries were invited to submit their views by 21 March 2008 on outstanding methodological issues such as assessment of forest cover and the associated carbon stocks and GHG emissions, how these are affected by sustainable forest management (SFM), and how reductions in emissions from deforestation and degradation below reference baselines (without displacement) could be demonstrated (see Herold & Johns, 2007). The coming two years would thus shape the post-2012 regime and the rules by which common but differentiated responsibility and accountability for climate change would be linked to global commitments to support development (Millennium Development Goals).
However, what had been largely missing from the debate at that stage, which Schlamadinger et al. (2005) touched upon but did not go into detail, was how emission reduction targets agreed at the national scale would be translated into changes of behaviour relating to land use at the local scale, while not affecting the rights of minority and vulnerable social groups, or the provision of ecosystem services in general. New ways to link the technical and institutional advances on REDD to local stakeholders were needed so that the various scenarios considered reflect local ambitions and response options. Social justice questions such as who will be the winners and losers (particularly in relation to common land), and how to avoid rewarding the "bad guys" while forest dwellers are evicted from their homes or forced to abandon their agricultural activities, would all need to be addressed. Active "negotiation support" was needed to achieve the "free and prior informed consent" status that is seen as a moral imperative to agreements that potentially affect livelihoods of people outside of the centres of political power. Moreover, a landscape perspective was required due to the spatial interdependencies of forest, agriculture and other land uses leading to deforestation and degradation at the forest margins ß?? it is not a "forest problem" alone. Solutions are likely to be quite context-specific and to vary from one country to the next, but without these crucial links between global and local, the climate policy discussions ran the risk of divorcing themselves from reality. The aim of the REDD-ALERT project, therefore, was to make a contribution to the evaluation of mechanisms that translate international-level agreements into instruments that would help change the behaviour of the people at the "coal-face" while minimising adverse repercussions on their livelihoods.
Project Results:
Objective 1: Documenting the socio-economic drivers of deforestation and degradation
This section presents the main science / technology results from work carried out under WP1, by discussing (i) the current state of global land use and expected trends over the period 2000-2030, (ii) the knowledge on causes and impacts of past and ongoing forest transitions, and (iii) prospects and policy options for a global forest transition, and (iv) results from the study sites.
Forest transitions
Our work showed that a handful of developing, tropical countries have recently been through a forest transition, thus shifting from shrinking to expanding forests at a national scale (Meyfroidt & Lambin, 2011). Different views exist on de?¬ning secondary forests and on whether reforestation should include some or all forms of tree plantations, or only naturally regenerating forests, which adds to the technical difficulties in measuring and characterizing reforestation. A few generic processes of forest transition were identified, as well as the broad geographical patterns of reforestation (Rudel et al., 2005; Lambin & Meyfroidt, 2010; Meyfroidt & Lambin, 2011). Countries do not necessarily follow a regular pattern of forest cover changes, and the causes and outcomes of forest transitions vary, so that forest transition is to be seen as a contingent process. Under certain conditions, environmental degradation can be perceived by land managers and, through such social-ecological feedback, can in turn become a driver of subsequent land use changes and of forest transition (Meyfroidt et al., in preparation). Restoring forests in one country is generally associated with a significant outsourcing of forest exploitation to neighbouring countries via increased imports of wood and sometimes agricultural products (Meyfroidt & Lambin, 2010). In Vietnam, the combination of policies restricting forest exploitation, rapid development of the wood processing industry and of exports of wood products, led to an increase in legal and illegal imports of timber and a displacement of forest extraction to neighbouring countries, such as Laos and Cambodia, equivalent to 39% of the regrowth in Vietnam's forests from 1987 to 2006 (Meyfroidt & Lambin, 2010). Policies to protect forests and promote reforestation therefore need to control this displacement of land use and channel it toward areas where the impacts are minimal (or beneficial). The ecological effects of reforestation are highly dependent on the residual deforestation of old-growth forests, the proportions of natural regeneration of forests and tree plantations, and the location and spatial patterns of the different types of forests (Meyfroidt & Lambin, 2011). Furthermore, net reforestation can conceal a continuing degradation or clearance of partly irreplaceable old-growth natural forests (Echeverria et al., 2006; Meyfroidt & Lambin, 2008). Policies supporting afforestation and reforestation should not assume that it will lead indiscriminately to environmental gains.
Prospects and policies for sustainable forest transitions
Although in theory the trade-offs between conserving forests and feeding the world's population could be minimized (Angelsen, 2010), a decrease in the availability of productive land and competition with other land uses will make a global forest transition difficult to achieve over the coming decades (see Table 1). Policies to achieve a forest transition include approaches to improve the supply of land-demanding products, and to control the demand for them (Meyfroidt & Lambin, 2011). The following factors hold the potential to significantly affect the supply of and demand for wood and agricultural products, and therefore contribute to control deforestation by addressing its drivers: (i) technological innovations and more efficient land-use practices to intensify agricultural and forestry production and reduce its environmental impacts; (ii) sound land management policies to control for rebound-effects; and (iii) changes in consumption patterns especially reduction of wastes and decreasing demand of the most land-demanding products ß?? e.g. meat. The combination of, first, promoting nature-friendly farming in areas with biophysical and social conditions unsuitable for large-scale intensive farming; and second, sparing land for forests through agricultural intensi?¬cation in high-potential agricultural areas might control the expansion of competing agricultural land uses (Fischer et al., 2008; Lindenmayer & Cunningham, 2012). Increasing off-farm economy, especially in the most marginal rural areas, can contribute to reducing the dependency on local natural resources. For commodities with high elasticity such as biofuels, meat, and luxury goods like coffee or exotic timber, agricultural intensi?¬cation and out-migration are unlikely to reduce the overall demand for agricultural land unless combined with policies to control rebound effects, e.g. by land-use zoning and demand-side interventions. Addressing final consumption and future global demand for wood and agricultural products is a critical aspect for any potential global forest transition (Meyfroidt et al., 2010; Lambin & Meyfroidt, 2011; Meyfroidt & Lambin, 2011). Across all these approaches, for the poorest fraction of humanity, the issues of equity, and access and sharing of food and land resources are crucial to ensure food security (Godfray et al., 2010). These approaches may rely on various tools, including state-level command-and-control, regulatory tools; rural and agricultural development policies; and emerging market-based instruments. REDD+ could support most of these strategies, and thus should not be considered only as a program of Payments for Environmental Services for agents of deforestation and forest degradation.
Objective 2: Quantifying rates of forest conversion and change in forest carbon stocks using improved methods
The objectives of this work-package were to 1) quantify current land use change and deforestation, and 2) improve monitoring tools. Data collection in the context of national greenhouse gas inventories following IPCC guidelines has to balance the errors (uncertainty) involved in 1) the use of the classification system (both at the "ground truth" and "remote sensed image" level), 2) the uncertainty in the properties (C stock per unit area) of each class and 3) the eligibility of different land cover types with respect to the scope of the policy.
Deforestation rates depend on forest definition
While the legends we use for land cover classification include a range of systems with various quality and quantity of tree cover, the concept of "deforestation" splits it into a "forest" and "non-forest" part. However, different stakeholders have different operational forest concepts ß?? ranging from the interest in untouched old-growth forest of strict conservation agents, to the very "weak" forest concept that was agreed as part of the Afforestation/Reforestation Clean Development Mechanism (A/R-CDM), referring to the FAO definition of minimum size, potential to reach a minimum tree cover and clarification what is meant by tree. Across such range of definitions we found the deforestation rate for Indonesia as a whole to be 3-5%/year without clear upward or downward trend within the 1990-2010 time frame, or shifting from +0.5%/year to -0.5%/year in the same time period (Figure 2). As the international negotiation arenas have not been able to clarify the forest concept that is to be used in REDD, there is far too much scope for picking the operational definition that matches the point one wants to make. Deforestation as such cannot be used to predict "emissions from deforestation", and the second D of REDD is essential to quantify emissions; the results of net emissions are much less sensitive to forest definition ß?? as earlier shown for Vietnam where an increase in net forest area was found to coincide with a continued loss of forest carbon stock, as densely stocked forest continued to be lost and the forest gained had much lower carbon stocks.
Reformulating allometric equations may help judge need for tree-specific equations
Part of the uncertainty in carbon stock data for land use systems is the use of "generic" allometric equations that relate tree biomass to stem diameter, while details of tree architecture differ between trees. A common form of such equation is Biomass = a(Diameter)b. Empiricial data sets suggest that both the a and b parameter differ considerably between tree species, and that thus tree-specific equations are needed. We noticed, however, that the a and b parameters of fitted equations are strongly linked, and that for a tree of 20-30 cm diameter the variation in predicted biomass is small. Reformulating the allometric equation to Biomass = a2(Diameter/DiamStandard)b2, with a standard tree diameter (DiamStandard) of about 20 cm leads to much lower variation in a and b, but a variation in a and b that is independent of each other (so it becomes a clean two parameter equation). At stand level, where there typically is a range of tree diameters, the choice of equation matters much less than the tree level differences suggest, as there is strong compensation. These results will help to improve C accounting methods, and justify the use of generic allometric models as default.
Simplifying from IPCC 5-pool data for carbon stocks
The IPCC prescribes that five pools of terrestrial carbon need to be assessed: trees, understory vegetation, roots of all plants, litter plus necromass, and soil organic carbon. Based on over 700 sample points we found that the data collection can be replaced by the use of defaults, as variation between land use types in soil carbon (in top 30 cm of the profile) plus litter layer plus necromass plus understorey vegetation represents only 6% of variation in tree biomass, while the belowground root biomass is estimated to be 25% (with little opportunity to verify this estimate). Under specific conditions of recent forest damage, however, the necromass pool can be more than 50% of the total carbon stock, and cannot be ignored in data collection.
Error propagation in C stock estimates
Errors in land use classification (with typically 85% accuracy at pixel level for a 25-point legend) combine with uncertainty around the typical C stock data used. We analysed how the combined errors relate to scale. Typically, aggregation across pixels reduces the random element in the error, while not reducing any bias that may be there. For REDD applications we found that aggregating up to a 1 km2 scale reduces uncertainty in C stock change to below 5%, while estimates for smaller pixel sizes may have a higher chance of containing error. The acceptable error level for a government-based incentive system on performance in C stock change is not empirically known, so our 5% threshold may have to be revised in future.
Opportunity costs curves
The monitoring of land use change needs to relate the economic benefits that (local) actors derive, to the global consequences for emissions. Initial estimates by the ASB consortium in 2007 in Indonesia, Cameroon and Peru suggested that 85% of emissions yielded net economic benefits of less than 4ß?¬/t CO2 at current prices. Further analysis of the spatial determinants of such abatement costs focused on soil type (peat versus mineral soils), accessibility (rivers versus roads), actor (large scale operations versus smallholders) and policy domain (existing land use restrictions and forest protection rules). We refined the calculation procedure in the ABACUS tool, which was endorsed by the World Bank in training at national level, and was refined based on user feedback.
Recommendations on the design of national monitoring systems
Reflecting on the cost of data collection, the marginal reduction in net uncertainty of estimates of terrestrial carbon stock change, we recommend a national monitoring system that pays specific attention to the low frequency of high-diameter trees that contain a disproportionally high part of total carbon stocks, while simplifying data collection on pools that were found not to vary much. Soil sampling for typical land use types in a national accounting system may be worthwhile, but for smaller accounting areas the costs of data collection probably exceed any benefits that might be derived through carbon markets.
Objective 3: Improving GHG accounting methods and default values of land use change in tropical forests
The overarching objectives of this multi-partner work-package were twofold: (1) to quantify the magnitude of GHG emissions and changes in belowground soil carbon stocks resulting from the land-use change at tropical forest margins, and (2) to identify the controlling mechanisms underlying the GHG emissions in both deeply weathered mineral soils and peat soils. To address these objectives, the project used a comprehensive approach consisting of literature reviews, field experiments, regional sampling campaigns, and modelling.
Field measurements on mineral soils
We investigated the impact land-use change has on SOC stocks in deeply weathered mineral soils of three REDD-ALERT study regions: (a) Jambi province, Sumatra, Indonesia, (b) southern Cameroon and (c) Ucayali province, Peru. Using a space-for-time substitution sampling approach, we quantified SOC stocks in the top 3 m of soil and compared converted land-uses with adjacent reference forest plots. In each country, we investigated the most predominant land-use trajectories relevant for the respective region. These included conversions from forest to cash-crop tree plantations (rubber, oil palm and cacao), conversions from forest to cattle pastures, and the conversion from forest to shifting cultivation systems (a rotation of crop-fallow-crop). In total, 157 plots were established where soil samples were taken to a depth of 3 m from a central soil pit and from the topsoil from 12 pooled composite samples. All soil samples were analyzed in the laboratory at the University of Goettingen for C, N, pH and effective cation exchange capacity (ECEC).
This study found that the conversion of forests to intensively managed land-uses such as oil palm plantations, cacao plantations, rubber plantations and cattle pastures caused significant SOC stock losses in the topsoil (Table 2). In contrast, the less intensively managed shifting cultivation land-use exhibited minimal change in SOC stocks in the soil profile.
Despite the fact that most of the SOC from the 3 m profile is found below 1 m depth (50-60% of total SOC stock), this subsoil C stock remained relatively unchanged by the land-use conversion. The only exception was in cacao plantations where decreases in SOC stock were measured down to 3 m. These plantations were generally much older and it is speculated they had the time to reach a steady state condition throughout the soil profile. The results on the magnitude of SOC losses in the soil profile and the constraints regulating C concentrations and changes therein is expected to be published in early to mid 2013.
Measurements of GHG emissions from mineral soils took place in the province of Jambi, Sumatra, Indonesia. Along a gradient of forest disturbance including rubber and oil palm plantations on mineral soils, measurements of C stocks and soil trace gas emissions were conducted during the period July 2010-August 2011. Soil N2O emissions were slightly reduced after forest degradation and conversion to rubber plantations but increased after conversion to oil palm plantation due to nitrogen fertilizer application. Soil CH4 fluxes changed from sink to source with forest degradation and conversion to oil palm plantation because of increased termite nests presence in these land-uses. An analysis on how termite species differently influence soil fluxes of CH4 is being conducted. Fluxes of N2O were significantly correlated with CH4 fluxes; mineralisation and nitrification potentials of the soil mineral N content and with the soil water-filled pore space during wet months. Soil respiration was not significantly affected by land-use change. The results on soil and vegetation C stocks are still being processed, and several publications are expected from the research.
Field measurements on peat soils
Field research in Jambi was also implemented in peatlands. Carbon stocks in living and dead vegetation, above- and below-ground, were measured in a primary peat swamp forest, a logged and drained peat forest, and a seven-year old oil palm plantation on peat. Total biomass C stocks decreased from 220 ?± 7 Mg ha-1 in the primary forest to 100 ?± 2 Mg ha-1 in the logged drained forest, and 32 ?± 4 Mg ha-1 in the oil palm plantation. Woody debris contributed 7% to total C stock in the primary forest, and as much as 43% and 31% in the logged drained forest and the oil palm plantation, respectively. Coarse root biomass represented 14%, 10% and 21% of total C stock in the primary forest, logged drained forest, and oil palm plantation, respectively. In the same land-use treatments, the root:shoot ratios of trees/palms were 0.19 ?± 0.01 0.21 ?± 0.04 and 0.60 ?± 0.42. The research will be submitted to the journal Forest Ecology and Management in 2013. Soil respiration was also measured in these systems. Preliminary results of measurements conducted between January and September 2012 indicate emission rates of 16 ?± 1.2 18.5 ?± 0.7 and 28.4 ?± 1.2 Mg CO2-C ha-1 yr-1 in the primary forest, logged drained forest, and oil palm plantation, respectively.
A further five sets of field experiments to measure GHG emissions were conducted in five land-use types located in Jambi and Riau, Sumatra Island. The first study aimed to evaluate total soil respiration as a function of distance from oil palm trees in relation to root distribution. The study was conducted on peatland under 6 and 15 year old oil palm plantations in Jambi Province, Indonesia from June 2011 to April 2012 and from January to April 2012, respectively. CO2 emissions were measured using an infrared gas analyser (IRGA), Li-COR 820. Total respiration was two to three times higher at points 1 m compared to that at points ß?¥3 m and linearly decreased with distances from the trees. At distances of 3.0-4.5 m from the trees the emission no longer increased with distance, indicating negligible influence of root-related respiration. The emission values at this distance for the 6 and 15 year old oil palm plantations were 38 and 34 Mg CO2 ha-1 yr-1 respectively. CO2 flux measurements in oil palm plantations would adequately represent the heterotrophic respiration if made at a distance of 3 m or more from the tree centre.
The second study was focused on the variation of CO2 emissions at five land-use types including oil palm, acacia forest, rubber, secondary forest and bareland in Riau, Jambi and Aceh, Indonesia. We conducted a detailed study of CO2 emissions using an infrared gas analyzer (IRGA LI-COR 820 model). CO2 emissions (Mg ha-1 yr-1) under oil palm plantations in Riau, Jambi and Aceh ranged from 18 ?±13 to 66 ?±24 Mg ha-1 yr-1, with the average of about 39 ?±19 Mg ha-1 yr-1. For adjacent plots with maximum distance between plots of 3.2 km, in Kampar Peninsula, Riau, CO2 emissions from secondary forest, acacia forest, bareland, rubber and oil palm plantation were 60 ?±25, 60 ?±19, 56 ?±26, 52 ?±17, 66 ?±24 Mg ha-1 yr-1, respectively. Our results showed that CO2 emissions increased with the increase of water table depth. Our finding is comparable with other CO2 flux measurement and can contribute to reducing uncertainty of peat CO2 emission estimates.
In the third study, we examined effects of various levels of water contents and laterite application on microbial (heterotrophic) respiration of peat soil. Bulk samples of surface (0 20 cm depth) and subsurface (30-50 cm depth) layers were collected from an oil palm plantation in Riau Province, Indonesia. Peat water contents were adjusted at 20, 40, 60, 80, and 100% water-filled pore space (WFPS). Laterite soil (clay containing high Al and Fe oxides) was applied at 3, 6 and 12 mg g-1 dry weight (1.2 2.4 and 4.8 Mg ha-1) for peat samples at 60% and 100% WFPS. Peat respiration differed between the soil layers, and was distinctly affected by water content, but less affected by laterite application. Peat respiration increased sharply from wet (ß?¥80% WFPS) to moist soil (60 to 40% WFPS) and decreased when the soil became dry (ß?¤40% WFPS). Maximum peat respiration for surface and sub-surface layers occurred around 50% and 30% WFPS, respectively. Laterite as a peat ameliorant accelerates rather than reduces peat respiration and thus it cannot be used for CO2 emission reduction.
The fourth study related to the spatial variation of CO2 flux with water table depth, soil moisture, and temperature under oil palm plantation in Jambi Province, Sumatera, Indonesia. A total of 480 CO2 flux measurements were made using an infrared gas analyzer (IRGA) every three months, at six different time intervals in a day. The results showed that the average CO2 flux was 46 ?±30 t CO2 ha-1 yr-1 (n=480). Within the one year study, the average CO2 flux did not show a clear relationship with instantaneous water table depth, soil moisture, and temperature. A positive correlation between these variables and the CO2 flux only occurred in October 2010, coinciding with the beginning of the wet season. The distance measurement points from the edge of the canal showed a positive correlation between CO2 flux (R2=0.6) water table depth (R2=0.6) and soil moisture (R2=0.5). The factors driving the CO2 flux at the study site was very complex, each affecting each other and working simultaneously.
In the fifth study, CO2 flux measurements were made in intact forest, logged forest, and oil palm plantation sites using four methods ß?? sampling at specified distances from palms, ??13C isotope analysis, random collar total efflux sampling, and simulation modelling. Results indicated interference in the ??13C results ß?? likely from methane oxidation, which has never been considered in this method before. Average total CO2 emissions from sites were 29 t CO2 ha-1 yr-1 in the wet season in the intact forest and 71 t CO2 ha-1 yr-1 in the dry season. Logged forest emissions were double in the deeper peat sites (109 t CO2 ha-1 yr-1) than the shallow peat sites (57 t CO2 ha-1 yr-1). Across the different oil palm sites total emissions ranged between 33-98 t CO2 ha-1 yr-1 on the deep and shallow peats. Heterotrophic emissions (excluding rhizosphere respiration) in the oil palm sites averaged 15-38 t CO2 ha-1 yr-1. Modelled heterotrophic emissions were higher at 37 55 t CO2 ha-1 yr-1 (higher due to the inclusion of rhizosphere respiration), and modelled net emissions at 44-60 t CO2 ha-1 yr-1. Soil C sampling established peat depths and C stocks of all sites ß?? peat depths varied between 3-7 m across all sites. Final analysis on the intact sites peat stocks is still ongoing, but in the oil palm sites total C stocks ranged between 171-199 t C ha-1 m-1 depth.
Effects of fire on the organic matter composition of tropical peat
This part of the work was based on investigation of how fire modifies peat organic matter chemistry with particular regard to shifts in carbon pools. The project was conducted in the tropical peatlands of Kalimantan, Indonesia. The research began with initially assessing the organic matter composition of recently burnt peat using a variety of methods including Pyrolysis-GC/MS (Py-GC/MS). This qualitative method provides a chemical fingerprint of the organic matter composition.
The first section of work assessed the short term effects of fire by assessing organic matter composition one month post-fire. Structural differences between burnt and unburned (inundated and drained) peat samples suggest that a combination of both fire and drainage causes alteration of the OM composition that is evident shortly after fire. The main observations are summarised as follows: (i) Unburned, inundated peat pyrolysates contain contributions from all compound classes. Lignin products such as guaiacol, methyl guaiacol and ethyl guaiacol are dominant for unburned samples. (ii) Long term drainage conditions induce oxic conditions in the upper peat layers causing a reduction in OM diversity particularly below 30 cm from the surface. Drained peat pyrolysates are dominated by aliphatic components. (iii) Surface, recently burnt and drained sample pyrolysates are, however, composed predominantly of aromatic and aliphatic compounds and are significantly reduced or depleted in all other compound classes, including lignin and polysaccharide derived compounds. A high aromatic and aliphatic content, including a large contribution from n-alkene/alkane doublets, suggests that the burnt peat is highly refractory and that much of the labile component has been lost or converted to other C forms. (iv) Subsurface burnt and drained sample pyrolysates are associated with aliphatic moieties with considerable variance from surface samples. Differences in surface vs. subsurface burnt and drained samples are attributed to differences in the impact of fire and decomposition. (v) Although fires in tropical peatlands are often considered to be smouldering, thereby penetrating the peat substrate, data in this analysis suggest that effects on the OM were predominately experienced in the surface peat layer (0-5 cm). Any differences between intact inundated peat and subsurface peat in burnt drained sites are likely to be caused by decomposition processes.
The work proceeded to investigate the impacts of different fire regimes on the peat organic matter including the longer term effects of fire on organic matter chemistry. Data suggests that there are significant differences in organic matter composition between recently burnt peat and peat collected approximately 1.5 years post-fire; however, these differences were only significant between surface samples. Below 5 cm organic matter composition does not significantly vary. Shortly after fire (November 2009), sample pyrolysates of the upper 5 cm of peat are significantly different from those collected approximately 1.5 years later (April 2011). PCA analysis differentiates between recently burnt peat and peat which had not been burnt for several years. Initially burning had transformed the peat OM composition so that it became highly alipahtic with considerable contributions from aromatic compounds as well as pyrolysis products thought to indicate the presence of charcoal e.g. naphthalene. 1.5 years later sample pyrolysates remain highly aliphatic, with considerable contributions from alkenes and alkanes. Dimethylnaphthalene and trimethylnaphthalene (biomarkers of charred material) both plot in similar factor space when assessing variance using PCA, the presence of such compounds 1.5 years after fire suggests charred organic matter is still present but less dominant than in those samples analysed shortly after the fire was extinguished. However the overall OM diversity (i.e. number of pyrolysis products identified, (111 Vs 72-75) has increased and there is evidence for some labile components such as cellulose (as indicated by levoglucosan) and furans which were depleted immediately after the fire. The proportion of oxygen in surface sample pyrolysates increases during the 1.5 year recovery period (3.9 % Vs 0.75 ?± 0.35%) which is due to the increased abundance of oxygen containing functional groups in polysaccharide and lignin derived compounds. In summary, peat fires cause considerable alteration to peat organic matter however these predominately impact the upper 5 cm of peat ß?? with less significant change at depth. The peat organic matter composition becomes more diverse 1.5 years after fire but even 14.5 years post-fire the organic matter does not recover back to its natural undisturbed state. Therefore it is suggested that fire has long term effects on peat organic matter however much of this change, albeit on a shorter temporal scale, mirrors compositional changes caused by long term peat drainage.
Understanding how tropical peatland fires modify peat OM composition can contribute to an improved knowledge of post-fire carbon cycling and nutrient cycling. Thermally induced alteration of peat OM may be inhibiting ecosystem recovery, ultimately influencing the global carbon balance. The results of this work also demonstrates that fire has caused the most labile OM components to be removed from the upper peat layer as well as the neoformation of aromatic structures, thus increasing the recalcitrance of the peat substrate. Changes in such carbon pools are likely linked to peat soil respiration rates as a recalcitrant peat effects by fire is more resistant to microbial decomposition than a labile peat- not affected by fire. Alteration of the OM composition may also have implications for physical properties on peat soil, for example enhanced water repellence, which would impact upon surface runoff and solute transport.
Objective 4: Identifying and assessing viable policy options addressing the drivers of deforestation
Drivers
The research in WP4 has shown that most instruments in the forest sector deal only with proximate drivers, and often in an inequitable manner. Very few deal with underlying drivers, and to the extent that they do so their effect is limited (e.g. debt for nature swaps, certification). However, the very nature of the underlying drivers (e.g. demographic, economic, technological, political or cultural trends) is that they are often slow processes operating at national or global levels resulting from the aggregated behaviour of many regional, national, subnational and individual entities (in some cases), referred to in the Panarchy literature as slow variables. Effecting change in these slow variables is either difficult due to their inertia, or unpredictable due to chance interactions with faster changing lower scale variables, and may therefore be beyond the power of any one of these entities to address. Collective action at the global level is clearly required, but there are often conflicting national interests (usually economic) that weaken the international resolve to find solutions to global environmental problems.
One way, however, to address underlying drivers of deforestation at the national level may be to mainstream forest protection into development paths. Forest policy needs to be integrated into sustainable development. This may not be easy in early stages of development or in early stages of the forest transition curve ß?? as the motivation to develop often comes at the cost of resource extractions from forests and forest land. The implication of this is that without substantial financial and institutional support from external actors, mainstreaming forests into forest governance in countries where a large percentage of land is under forests is unlikely.
Systems approaches
Regardless of the degree of deforestation present in a country, for lasting solutions to be developed, it is essential to see forests as components of larger systems of land use, which also include arable agriculture, grasslands, wetlands, and human settlements. Deforestation is only one of several major problems that humans need to grapple with in the next century - together with concomitant increases in demand for food, water and energy against a backdrop of climate change, urbanisation, and limited land resources, this has been referred to as the "perfect storm". Dealing with any one land use component (such as forests) in isolation is likely to result in partial solutions at best as the Law of Unintended Consequences starts to operate. Providing alternative employment opportunities to reduce dependence of people on forests for their livelihoods, for example, may result in increased GHG emissions from the industrial sector.
Although REDD began as an initiative to reduce deforestation, it is becoming apparent that if the global community is serious about reducing net GHG emissions, a systems approach is needed. This systems approach should take into considerations the many and complex interactions between forestry, agriculture and urban sectors, so as to optimise the fluxes of carbon and other nutrients, as well as social and economic flows, between different parts of the landscape. In this context, some even ask whether REDD funding might be effectively used for funding agricultural research to increase production on existing agricultural land, reduce losses in the food chain, and return nutrients from urban to rural areas, thereby reducing the need to clear further forests. Investing in agricultural research is a mitigation strategy in its own right ß?? it has been estimated that the avoided emissions through agricultural intensification since the 1960s have only cost around $4/tCO2e saved, cheaper than many other abatement measures.
Measures to ensure REDD+ success
To the extent that REDD can provide temporary relief to the climate change problem, we have proposed twelve generic measures that can help to embed REDD within the domestic forest institutions and have shown how these generic measures have contextual implications in Vietnam, Indonesia, Cameroon and Peru.
Future challenges
Countries are gradually investing in becoming ready for REDD projects in the hope that it will develop into a win-win situation. However, there are indications that if REDD is poorly designed and/ or implemented, it may turn into a lose-lose situation. But knowledge is power, and awareness of the risks may help countries and social actors mobilize themselves to prevent such risks. Furthermore, there is a possibility that the awareness of the actual costs of implementing forest conservation equitably may lead many to see this as less than cost-effective, and thus reduce the incentive to invest in. REDD faces considerable challenges; if these challenges are not rapidly addressed, REDD may disappear from the global agenda in its current form as suggested by the issue attention cycle.
Its enduring legacy, however, will be that it has mobilized global attention and social actors on the need to understand forests and human-forest interactions. There is far greater understanding of the drivers of deforestation in different countries and the limits of instruments in dealing with these forests. There is a realization that there is need to take a systemic view and see if mainstreaming forests in national development, agricultural, energy and mining policy leads to understanding how best to deal with forests. The next step thus is to understand how to mainstream forests and see if countries can reconcile their need and right to develop with the need to protect forests not just for themselves, but for the myriad local ecosystem services that they provide as well as for the global ecosystem services that they provide. This knowledge and the mobilization of a large number of actors may be in itself critical for creating a more comprehensive, equitable, legitimate, effective, efficient and enduring system for dealing with forests. As in the myth of Sisyphus, it will remain an uphill struggle.
Objective 5: Analysing policy scenarios on GHG emission reductions, land use and livelihoods
Assessing agent-based models
Agent-based simulation modelling has recently been arousing interest, due to its ability to model individual decision-making entities and their interactions, to incorporate social processes and non-monetary influences on decision-making, to conceptually reproduce non-linearities ("tipping points") often observed in space-time processes of innovation and change, and to dynamically link social and environmental processes.
As a starting point, Matthews & Dyer (2011) discussed whether traditional neo-classical economics models were adequate to analyse all policy instruments that might be deployed in a REDD+ context. It is now recognised that decision-making is influenced by many other factors beyond those considered by traditional economics (Geist & Lambin, 2002). Akerlof & Schiller (2009), for example, list a number of "animal spirits" that include confidence, fairness, corruption, pride, shame, neighbour influence, and social pressure, that are also influential, all of which will be relevant to a household's willingness to engage with REDD schemes. For example, even if it is economically rational, people might be reluctant to participate if their previous experience with local government has given them little confidence in whether they will actually receive compensation for not deforesting, despite alluring promises from officials. To account for these non-economic factors, van Vugt (2009) has proposed a 4Is framework of Incentives, Information, Identity, and Institutions, that influence people's decisions. Incentives in this framework include any incentives that enhance a household's assets.
Gathering data on these decision-influencing factors, however, is a significant challenge, particularly in developing countries. However, Matthews & Dyer suggested that each aspect in itself needs to be explored in depth to provide greater insight into decision-making processes and their implications than we have at present. At some time in the future it might be necessary to combine the important factors, particularly if interactions between them are being explored. A first step would be to include these factors into the models that we have already, and use sensitivity analysis to determine the likely influence they have on the overall dynamics of the system and the conclusions reached. Efforts could then be focused on obtaining information only for those factors that produce qualitatively different conclusions from traditional approaches.
Villamor et al. (2011) then reviewed different agent-based modelling approaches and the extent to which they consider the diversity factors inherent to landscape mosaics. Diversity challenges start from the representation of agents (i.e. aggregate or individual, and types), the agent decision making rule (i.e. social to less social factors approach), the agents' interaction medium (i.e. agent-to-agent, and/or agent with environment), and the integration of biophysical processes. They then discussed the challenges in integrating the non-economic motivations discussed above into the decision making of human agents, concluding that no comprehensive methodology is currently available for modelling of this type, and suggest that objective and interdisciplinary criteria should be used when scoping and constructing a model, and that the model should be based on the ultimate purpose of the outputs for planners or local people.
Modelling ecosystem carbon stocks
Most of the literature review and meta-analysis in WP3 (Powers et al., 2011) related to carbon stock changes associated with land-use changes in the tropical forest margins were partial with the analysis of either only NPP productivity or aboveground biomass or soil changes. To understand the effect of land-use change on the overall carbon stock changes, we need a whole system approach in which carbon in soils, roots, detritus, and other aboveground components are estimated following the pattern of land-use change. No review studies to our knowledge have been done to establish such relationships between the ecosystem carbon (sum of all organically derived carbon in various pools: both aboveground and belowground including soil) and various climatic or soil factors. We decided, therefore to extend this to also include above-ground carbon pools, using both a review of the literature and models for simulating the forest transition, to derive simple relationships between environmental variables such as temperature, water and soil variables such as clay and SOC content with the ecosystem carbon stock under a land-use or land-use change following deforestation in tropical forest regions.
The results of the study show only limited studies or data are still available that can estimate the carbon pools at an ecosystem level and we need more field studies to fill the gap. A simple regression model was derived using a small set of parameters, and was found to be useful to estimate the C stock changes associated with the land-use changes in the tropical forest margins. The results of the regression analysis showed that mean annual temperature is the most important single variable which explains much of the variability in carbon pools and pool changes under various landuse and landuse changes. However, there is a need to validate the model with a wider range of environmental and soil conditions before they can be used with confidence. Above- and below-ground measurements of carbon stocks were subsequently made(within WP2 and WP3) in three of the study sites ß?? Peru, Cameroon and Indonesia ß?? and at the time of writing are still being analysed. A draft paper is in preparation and is expected to be completed early in 2013.
Objective 6: Developing and using new negotiation support tools
The objective of this work-package was to seek the direct engagement of local, national and international stakeholders in the emerging mechanisms to reduce emissions from deforestation and degradation (REDD), by 1) synthesizing their perspectives, and 2) exploring with the stakeholders how the various REDD+ mechanisms (combining positive and negative incentives) could work out in practice. Role-play versions of process-based simulation models allowed for active engagement with "agents of change" that provide incentives to increase or decrease emissions. In parallel, 3) agent-based modelling and simulations explored the efficiency/fairness trade-off in REDD+ incentives and sought to establish benefit-sharing solutions that are acceptable to all.
Emission reduction primarily requires a shift in development trajectory for tropical forest margins, dealing with multiple actors, multiple incentives and multiple knowledge types. Actual economic benefits derived from land use changes that cause emissions have been low for forest-to-pasture conversion in the Amazon and for most types of land use change on peatlands in SE Asia. Even where such "opportunity costs" (net economic benefits of land use change foregone) are below a feasible C price, actual shifts in development trajectory require incentives that work for all, are within acceptable concepts of fairness and efficiency for all who might otherwise sabotage the process.
Four approaches to opportunity cost estimates
We compared the methods for four approaches to opportunity cost curves and advised where in the process of negotiating actual contracts that include Free and Prior Informed Consent, information needs to shift from (I) retrospective costs of system level change, and (II) retrospective analysis at pixel level, to forward looking scenario comparisons at landscape (III) or agent (IV) level.
Main dissemination activities and exploitation of results
The work carried out in the project is being disseminated through various peer-reviewed and other publications, as well as various activities, including broad audience publications, policy briefs, conferences and other presentations (see Tables A1 and A2). Here we present just some of the major activities which included presentation of project results at side events of the United Nations Framework Convention on Climate Change (UNFCCC) Conference of Parties (CoP) meetings to which many negotiators attended and contributed to the discussions.
UNFCCC CoP-14, December 2008, Poznan, Poland: This was before the project commenced, but the following talks were given in anticipation of the project:
1. "Introducing new initiatives to support implementation of REDD". Talk by Robin Matthews at ASB/MLURI side event at UNFCCC CoP-14 Forest Day, Dec 6, 2008.
2. "REDD-ALERT: Investigating ways of linking global climate policies to local behaviour change in tropical rainforest countries". Invited talk at EU side-event "Climate change research and observations beyond Europe", Dec 10, 2008.
Copenhagen Climate Change conference Mar 10-12, 2009: "REDD-ALERT: Evaluating global-level climate policy options and their local level implementation". Poster presented by R B Matthews at Copenhagen Climate Change conference.
UNFCCC CoP-15, December 2009, Copenhagen, Denmark: A side-event was organised by the EC DG-RTD entitled "Deforestation, Forest Conservation and the Climate Change Challenge" exploring the social, economic and environmental drivers of deforestation, their interdependence and consequences in South-East Asia, Africa and Latin America. Presentations included:
1. Welcome and introduction by Dr. A. Kentarchos (Climate Change and Environmental Risks Unit, DG-Research, European Commission)
2. "Solving the climate challenge within the planetary boundaries" by Prof. J. RockstrG¶m (Director of Stockholm Resilience Centre at Stockholm University, Director of Stockholm Environment institute)
3. "From deforestation to reforestation: conditions for sustainable land use" by Prof. E. Lambin (UniversitG© Catholique de Louvain, Belgium)
4. "REDD-ALERT: linking global climate arrangements to local land-use behaviour" by Dr. R. B. Matthews (Macaulay Institute, Aberdeen, UK)
5. "Opportunity costs of carbon emissions from land-use change: need to broaden scope of REDD" by Dr. M. van Noordwijk (ICRAF - regional coordinator for SE Asia)
6. Panel discussion ß?? debate. Moderator: Dr. E. Lipiatou, (Head of Unit, Climate Change and Environmental Risks Unit, DG-Research, European Commission)
A video-clip of this side-event can be viewed at http://ec.europa.eu/avservices/video/player.cfm?sitelang=en&ref=66868
UNFCCC CoP-18, December 2012: A side-event was organised by Dr Robin Matthews at the EU Pavilion at CoP-18 in Doha, Qatar on 29 November 2012 entitled "Is the window of opportunity for REDD+ closing?". Presentations included:
1. Welcome and introduction by Dr Luca Perez (Climate Change and Environmental Risks Unit, DG-Research, European Commission)
2. "Linking global climate arrangements to local land-use behaviour" by Dr Robin Matthews (Theme Leader, James Hutton Institute, Aberdeen, UK)
3. "Understanding, measuring and governing changes in forest carbon stocks in complex landscapes" by Dr Ole Mertz (I-REDD+ Coordinator, University of Copenhagen, Denmark)
4. "Diversity of land use trajectories and implications for REDD+" by Dr Daniel MG?ller (Senior Scientist, Humboldt UniversitG¤t zu Berlin and IAMO)
5. "Progress in determining reference emissions levels" by Dr Lou Verchot (Programme Director, CIFOR)
6. "Nationally Appropriate Mitigation Actions for the forests and other land uses of Indonesia: complementarity of policy instruments, funding streams and motivation" by Dr Meine van Noordwijk (Chief Scientific Advisor, ICRAF)
An article was prepared and published in the December 2011 issue of the stakeholder magazine International Innovation (Environment): Matthews, R.B. 2011. Evaluating local impacts from global GHG policies. In: International Innovation (Environment: December 2011 issue), pp. 64-66.
CNN Television broadcast a brief documentary on REDD-ALERT project activities by the Centre for International Forestry Research (CIFOR) and the World Agroforestry Center (ICRAF) in Indonesia. The programme was aired with the title 'Indonesia aims to halt deforestation' on 27 November 2010, which can be viewed using the following link:
http://www.cnn.com/2010/WORLD/asiapcf/11/22/indonesia.halt.deforestation/index.html
It is planned to publish a selection of papers from the REDD-ALERT project in a Special Issue of Mitigation and Adaptation Strategies for Global Change.
A full list of dissemination outputs is given in the Use and dissemination of foreground ß?? Section A below.
List of Websites:
Project website: http://www.redd-alert.eu/
List of contacts
Beneficiary Number Beneficiary name Beneficiary short name Country Contact
1 Macaulay Land Use Research Institute MLURI UK robin.matthews@hutton.ac.uk
2 UniversitG© Catholique de Louvain UCL BELGIUM eric.lambin@uclouvain.be
3 Vrije Universiteit Amsterdam VU NETHERLANDS onno.kuik@ivm.vu.nl
4 Georg August University of GG¶ttingen UGOE GERMANY eveldka@gwdg.de
5 World Agroforestry Centre ICRAF KENYA m.vannoordwijk@cgiar.org
6 Centre for International Forestry Research CIFOR INDONESIA l.verchot@cgiar.org
7 International Institute of Tropical Agriculture IITA NIGERIA j.gockowski@cgiar.org
8 Centro Internacional de Agricultura Tropical CIAT COLUMBIA g.hyman@cgiar.org
9 Indonesian Soils Research Institute ISRI INDONESIA fahmuddin_agus@yahoo.com
10 Research Centre for Forest Ecology and Environment RCFEE VIETNAM phuong.vt@rcfee.org.vn
11 Institut de Recherche Agricole pour le DG©veloppement IRAD CAMEROON mtchienko@yahoo.com
12 Instituto Nacional de Investigacion y Extension Agraria INIA PERU jcuellar@inia.gob.pe