A pilot project to store carbon as biomass in African woodlands

Abstract

Capturing carbon by planting trees or avoiding deforestation is thought to be a cost-effective way to reduce the inexorable rise in CO₂ in the atmosphere. We describe a way to motivate African farmers to plant trees and protect woodland, based on a Mozambican pilot project in the voluntary carbon market. By late 2009, 1510 farmers were enrolled. Between 2003 and 2009, the project was able to sell carbon credits totaling approximately US$1.3 million on the voluntary carbon market, corresponding to 156,000 tCO₂, at a price that averaged US$9.0 per ton. Moreover, the effect of the carbon project was to increase rural employment from 8.6 to 32%, whilst 73% of households raised commercial crops compared with 23% previously. There was also a notable development of social capital, with a measurable increase in literacy and the development of a business ethos with associated practical skills.

Figure 1.  Generalized diagram to show the flow of money (solid line) in a project to incentivize tree planting and forest protection.

Dashed lines signify activity of external agents.

Figure 2.  Measured carbon stocks in the Miombo woodlands.

Trees are defined as >5 cm DBH, saplings as 5 cm DBH but >0.3 cm diameter at 10 cm above the ground. Stocks are for annual maxima of grass and leaf biomass. All stocks are in tC ha^-1 ± standard errors of the means. Stem and root biomass were calculated based on the inventory of 58 plots (total area: 27.2 ha). On conversion to farmland not all carbon is removed, and we estimate the carbon lost to be 22.5 tC ha^-1.

DBH: Diameter at breast height (130cm).

Figure 3.  Trends in market prices over the course of the project.

The solid line shows market prices for the Clean Development Mechanism of the Kyoto Protocol, the dotted line shows the average for the voluntary sector (data from the World Bank archives), and the dashed line shows prices of the sales pertaining to the current project.

Data from [106].

Figure 4.  Business plan for revenues from carbon sales in the Sofala carbon livelihoods project.

Forests and woodlands contain vast stores of carbon, and it has always been clear that the planting of trees on nonforested land, and the prevention of deforestation are likely to be one of the most cost-effective ways to reduce the build-up of carbon dioxide in the atmosphere [1]. Indeed, the 1997 Kyoto Protocol recognizes this in its Article 3, although no policy directly related to preventing deforestation is included. In the case of the tropics, the recent deforestation rate accounts for 1–2 billion tons of carbon transferred to the atmosphere each year [2], enough to substantially increase rates of global warming. In 2005, at the conference of the UN Framework Convention on Climate Change (UNFCCC), the Coalition of Rainforest Nations initiated a request to consider ‘reducing emissions from deforestation in developing countries’, and by 2009 the UNFCCC reached an accord on a series of measures collectively known as reduced emissions from degradation and deforestation (REDD). Since then, there has been ongoing discussion on the best way to achieve REDD, with emphasis on securing finance from international agencies including the World Bank, the United Nations Development Programme (UNDP), the United Nations Environment Programme (UNEP) and the Food and Agriculture Organisation (FAO), and from national governments.

Meanwhile, a number of projects designed to curb deforestation and degradation in Africa and elsewhere have been active as part of the voluntary carbon market (VCM) [3]. The VCM developed independently of the Kyoto Protocol and is a means whereby organizations or individuals can offset their carbon emissions and receive certificates of voluntary emissions reductions (VERs) [4]. Only a few of these carbon projects have run for a sufficiently long period to be able to learn lessons that may be valuable as funding and planning for REDD develops. With this in mind, the first objective of the paper is to define the methodology and business model that evolved during a specific 5-year pilot project in Mozambique. From this development we aim to:

	▪ Provide new data on the costs of creating carbon offsets in tropical landscapes;
	▪ Identify the major difficulties and challenges that arise in such a project;
	▪ Outline the co-benefits of such a project, especially as they impact on living conditions and sustainable development in southern Africa;
	▪ Discuss the general prospects for this approach to climate abatement.

The carbon project was originally conceived to test the applicability of Plan Vivo to the African situation. Plan Vivo is a community-based carbon crediting system that rewards farmers for planting trees and conserving woodland with a view to assisting sustainable development [101], devised several years ago in quite different conditions in Mexico [102]. The criteria for success of the current project included the following: carbon stocks of Miombo woodland should be measured, agroforestry systems should be established, carbon-stock and socio-economic baselines defined, carbon sales should be at least US$200,000 and there should be significant co-benefits to the community.

▪ Project area

The area of the carbon project is the Sofala province of Mozambique, an area known as the Comunidade do Regulo Chicale. The project was initiated in Nhambita village. The study area of 558 km² is a rectangle: to the north-west 18.82°S, 33.90°E; to the south-east 19.21°S, 34.26°E. Of this, the carbon project currently occupies 1500 farms and 100 km² of forest, the rest being available for further development and to evaluate any project leakage.

The climate falls into Köppen class Aw (tropical savanna), being sub-tropical with alternating cool-dry winters (April–October) and hot-wet summers (November–March). The rainfall measured 25 km away at Chitengo for the years 1956–1969 and 1998–2007 was 850 ± 269 mm/year (mean and standard deviation).

The vegetation consists of tropical woodland, savanna, secondary woodland, riverine forest and cultivated plots of subsistence agriculture called machambas, with each machamba being 1–3 ha. The woodlands are classified as Miombo, a type of savanna woodland occupying 2.7 million km² in southern Africa. Miombo woodlands are subject to frequent burning associated with slash and burn agriculture, and their carbon stocks are believed to be lower than the biological potential for the region. After fire, most of the tree species resprout, but the rate of regrowth is inherently low, and is limited by the poor soils and the low and unreliable rainfall of the region [5].

The human population occupies widely scattered homesteads, each with several buildings made of bamboo or poles, grass and mud, usually with livestock (e.g., chickens, ducks, goats and pigs), bananas and fruit trees (e.g., mango and papaya) with a central area for cooking. The main activities are subsistence farming, wood-gathering, hunting and preparing food. Until recently there were no shops, and school houses were primitive – sometimes with no roof. Wood is collected for firewood by a human population that has grown recently, particularly following the end of the civil war in 1992. Most of the woodland is burned every 1–3 years, which prevents the woodland reaching its maximum biomass.

Methods

Methods are provided in full on the Plan Vivo website under ‘Sofala Community Carbon Project Project Design Document According to CCB [Climate, Community & Biodiversity] and Plan Vivo Standards’. In this article we present an outline of the structure and protocols [101].

▪ Carbon project principles & Plan Vivo

The Mozambican carbon project was the initiative of a commercial company, Envirotrade Ltd, which took the role of project developer [103]. The developer’s management role is to:

	▪ Define and delineate areas of land suitable for carbon sales;
	▪ Alert potential carbon buyers of the offsetting opportunity that the project presents;
	▪ Gain the trust of the recipient communities;
	▪ Negotiate contracts with local farmers;
	▪ Train the farmers to carry out specific management practices that are defined as technical specifications.

The project developer also carries out day-to-day business management through its Mozambican subsidiary company. It keeps records of activity, which form the basis of payments; these records may be interrogated later by the carbon buyers enabling them to personally verify that carbon stocks are being protected and enhanced. To satisfy requirements of the certification agency, there is a comprehensive Project Development Document (PDD) to enable the Certifier to issue certificates as VERs units that subsequently might be traded as part of a carbon trading system. The voluntary carbon standard chosen when the project started in 2003 was the Plan Vivo standard, one of the first in the field, which has developed a reputation for carbon projects with a strong component of community development. After the selection of Plan Vivo, it was necessary to engage external technical experts to establish carbon baselines (a carbon baseline is the expected carbon sequestration potential without the project), and write technical specifications to define procedures [104]. Baseline data collection included estimation of deforestation rates, analysis of concentrations of carbon and nutrients in the soil and socio-economic status. Following the initial data collection, there must be a plan for periodic monitoring of the area covered with woodland, and the carbon stocks per area of land.

The interaction between these components may be hypothetically represented by a block diagram (Figure 1). The income stream from carbon buyers is paid into a trust fund, which is fully audited and consists of representatives of a local community association and respected NGOs as well as a representative on the project developer. The carbon trust fund disburses funds to the farmers, and makes funds available for local community projects, including the building of schoolhouses and a clinic, as well as funds for the development of micro-industries including tree nurseries, bee-keeping, a saw mill and carpentry shop (these activities stimulated business skills in the community). The project developer needs to cover costs, which include: staff salaries, vehicles, commissioned research and training and third party audits.

▪ Contracts with farmers

The basis of contracts was a set of technical specifications [104]. These are land management instructions with an associated table to show how much carbon is expected to be sequestered as a result of following the protocol. The first seven Technical Specifications in the project were for tree planting aimed at increasing carbon stocks in the land used for growing crops (e.g., agro-forestry with an emphasis on N-fixing trees, or orchards and woodlots). One further technical specification then followed: reducing GHG emissions by avoiding deforestation. To incentivize farmers, payments were ex-ante, based on previous experience of the Plan Vivo Foundation where ex-ante crediting is widely used. It means immediate reward to the farmer; after planting, the farmer receives 30% of the payment, then 12% per year for 5 years, then a final payment of 10% in the seventh year. Thereafter, it is assumed that the benefits (e.g., fruits and crop improvements) of the plantings are clear to see, and the farmer will be unlikely to cut the trees.

▪ Biomass & deforestation rate

Stem and root biomass was calculated from an inventory of 60 plots using a new allometric model, obtained from the destructive harvest of 29 trees, which relates diameter to stem and root biomass. The 60 plots were recorded between 2004 and 2007. The growth rate was obtained from the repeated measurement of trees at 15 1-ha sample plots. Originally we planned to use LANDSAT to estimate current deforestation rates but following the failure of its sensor in 2003 we used SPOT 4 images at 20 m resolution for 1999 and 2007. At the end of the project, we began to use Synthetic Aperture Radar (ALOS, JAXA, Japan) to attempt to measure biomass from space [6].

▪ Baselines

We considered three candidate methods of arriving at a carbon-stock baseline for woodland. The first involves tracking the deforestation over the last few years using remote sensing and assuming that ‘without a project’ the rate would continue linearly. The second involves modeling the per capita destructive exploitation of woodland, and linking it to a model of how the population might grow over the coming decades in the absence of a project. The third method assumes that if the land is accessible, cultivatable, has extractive value and is unprotected (ACEU), then it will be deforested unless there is a specific intervention for conservation [103].

Socio-economic baselines were established by interviews with householders in 2004.

Results

▪ Carbon-stock baselines

The carbon stocks as above-ground biomass in sample plots from woodlands ranged from 5 to 60 tC ha^-1, with a mean of 29.9 tC ha^-1(Figure 2). This is much less than for the typical carbon stocks of rainforest (∼200 tC ha^-1) [7], or for the uplands of Mexico where the original Plan Vivo project was established [102]. The permanent sample plots showed a current annual increment of between 1 and 3 m³ ha^-1 annum^-1, comparable to the data available for this region from the National Forest Inventory, which is 1.2 m³ ha^-1 annum^-1. The latter implies a carbon sink of about 0.3 ton ha^-1 annum^-1, provided the site is not burned. When burnt to convert to machambas, the trees are not entirely destroyed and the mean residual biomass is 2.8 tC ha^-1.

The carbon stocks likely to accrue over 100 years as biomass on machambas, estimated from current knowledge of the planted species, range from 10 tC ha^-1 for Gliricidia interplanting, to 50 tC ha^-1 for a managed woodlot.

▪ Areas of land with project activities

The area of land occupied by project activities increased rapidly over the period of the project. By late 2008, the area delimited and enumerated for avoided deforestation was approximately 10,000 ha; it provided a resource of saleable carbon that greatly exceeded the current sales (Table 1). The up-take of agroforestry options was good, although initially many farmers opted for the least intensive technical specification, namely boundary planting, which provides relatively little carbon sequestration (Table 2). Later, many of these farmers added further Plan Vivos, in order to extend their involvement, therefore to increasing their carbon sequestration and income.

▪ Co-benefits & community involvement

By late 2009, there were 1422 villagers with over 2500 ha of land devoted to agroforestry practices carried out according to the technical specifications (Table 2). Participating farmers were trained in tree planting, harvesting, irrigation, composting and the community technicians were trained in record-keeping. Tree nurseries were established to on a commercial basis to provide planting stock at a rate of 180,000 per year. The 17 workers were trained in seed collection, soil preparation and general nursery techniques. They were initially formally employed and later they formed micro-enterprise companies. An office was established to deal with personnel issues, monitoring and record keeping for Plan Vivo; this employed 17 people. Fire management operations required 27 people in part-time employment. Some technicians were trained and employed to collect data from sample plots and to carry out basic science activities. Various micro-enterprises were successfully established from carbon revenues, as a means to increase and diversify the community’s income. As well as the nurseries, these included: saw mill work; carpentry; bee-keeping; keeping guinea fowl; turkeys and ducks; and the production of saleable crops. Socio-economic indicators were obtained by interviews with occupants of 245 households in 2004 and repeated survey in 2008. Cultivation of commercial crops increased from occuring in 23% of households to occuring in 73%; the number of households with livestock increased from 58 to 89%; employment of householders rose from 9 to 32%; and there was a small increase in literacy. Almost all of this benefit came from project activity (there were a few community members who made charcoal and sold it by the roadside but most charcoal makers were itinerant). We note that there are substantial differences in the initial starting conditions between the Plan Vivo project in Mexico [101] and those in the present project in which there is a much lower level of employment and literacy, a much poorer standard of health and a lower and less predictable level of rainfall.

▪ Remote sensing of deforestation rate & baselines

The deforestation rate over the period 1999–2007 was assessed as 2.4% per year, although there is a high degree of uncertainty in this estimate. If this rate were to continue linearly, there would be no remaining woodland in 42 years. An analysis of the rate of use of woodlands by a rapidly growing population revealed that the assumption of linearity, implied by this method of calculating a baseline, was not realistic. The baseline approach chosen was therefore the ACEU method, since the landscape is indeed accessible (>8 km from a main road), cultivable, extractable and, in the absence of a project, unprotected. Thus, for carbon sales relating to avoiding deforestation, the assumption made is that the carbon would all be lost in the absence of a project, except for the few trees remaining after clearance for cultivation.

▪ Cost of carbon capture

Saleable carbon is expressed as VERs in units of tCO₂. By July 2009, they were 979,788 tCO₂ available, of which 156,085 tons were already sold at an average price of US$8.62 per tCO₂, yielding US$1,331,620. The buyers came from several countries, and included private individuals, consortia, and businesses. An additional research-oriented income of US$2.02 million came from the European Union, and so the total input to the project was US$3.35 million. Such a high cost would not be applicable when a project such as this one is extended to other areas of the host country, since at that (later) stage certain lessons would have been learned and economies of scale apply. The cost of capturing carbon so far may thus be estimated as US$3.42 per tCO₂ sequestered, with the expectation that in future costs will reduce. This may be compared to market prices (Figure 3).

▪ A business model

In this section we show how the income stream from carbon sales is allocated (Figure 4).

The Envirotrade model functions on the basis that two thirds of all carbon finance is returned directly into the local economy of the country. Money from the remaining third will pay verification, validation and business costs.

A third to fund farmers & community grants

To incentivize farmers, payments need to include an ‘up-front’ installment, followed by further installments that are conditional on progress (e.g., were the trees planted properly; have they survived?). Farmers are paid on the basis of inspections, irrespective of whether the carbon offsets have actually been sold, an important feature that reflects the poverty alleviation objectives of the model.

A third to fund operations costs

A third is paid as operations costs to the project developer’s nonprofit local subsidiary, which is staffed and run by Mozambicans. This includes the day-to-day running of the carbon project, vehicles, materials and employment. If the sales exceed the money needed for contractual and operations costs then the balance is directed to the community trust fund, which may provide grants for other community-related activities such as construction of new school buildings, health clinics and micro-enterprises. The greater the sales, the greater the funds available to the community trust fund since the operations budget will remain the same while the money available is larger. This brings a tangible effect from the buyer to the community.

A third to the project developer

This covers all off-site and international administrative, research, project development and marketing costs and potentially provides the company with a profit. In this project the parent company took on the responsibility of covering any shortfalls at the community trust by extending interest-free loans. These short-falls arise where carbon is sequestered by project activities and is not immediately sold. Any profits retained by Envirotrade were a small proportion of total offset credit sales, and were used (along with other financing sources) to cover the development costs of new projects. A maximum of 8% is retained from total carbon sales as a management fee.

▪ Difficulties encountered

The main difficulties were as follows:

	▪ There were no appropriate (site-specific) allometric models to enable biomass to be estimated, so we embarked on the lengthy program of sampling trees to determine the relationship between basal area and biomass [8].
	▪ The landscape in degraded Miombo is highly heterogeneous and the average for the region as a whole was not the same as that of the parcels of land chosen for carbon sales. Specific enumeration was therefore required for each delineated parcel of land.
	▪  Available optical satellite data was judged to be inadequate, and so, at the end of the project, we began to work with radar satellite data as it became available from the Japanese ALOS sensor. The advantage of this approach is that radar penetrates the vegetation and gives completely different signals for woodland versus grass, and can be used to estimate biomass [6].
	▪ The project operated in peculiar circumstances: rural communities in central Mozambique were in the aftermath of a bitter Civil War (1977–1992) and, consequently, there was an acute poverty in social capital. This meant that the challenges to governance and community management were acute and delays occurred.

▪ Success as measured against initial criteria

All the initial criteria of success were achieved (i.e., carbon stocks of Miombo woodland were measured, agroforestry systems were established, baselines were defined as far as possible, carbon sales exceeded US$200,000 and there were significant co-benefits to the community). More offsets were sold for avoided deforestation than for tree planting. One may argue that to avoid deforestation now is more valuable to the sequestration of carbon over decades to come. However, in this project the selling price of carbon is the same irrespective of the system used (Table 1). Following evaluation in 2010, the project received Gold Standard certification in all three categories of the Climate, Community and Biodiversity Alliance.

Discussion

▪ Costs of creating carbon offsets

The cost of creating offsets, some $3.42 per tCO₂ is lower than the price of carbon on the VCM (Figure 3), leaving a profit margin. However, actual revenues have been less than this because not all the CO₂ that has been sequestered (in business terms, the ‘stock’) has yet been sold. As in any business, the saleability of the product in a free market depends on economic conditions, and in this case perceptions of the quality of certification may be critical. As international policies and carbon markets develop, it seems likely that the price and saleability of offsets will increase.

▪ The major challenges of business development

In the Results section we outlined the challenges. All new types of business face a similar set of challenges, some of which relate to technological limits and others relate to market readiness. In this case, the technological limit is defined by the measurement of carbon stocks. Reliable and long term satellite data are required to monitor biomass stocks. It is clear that in the future this role will be taken up by a new generation of synthetic aperture radar sensors. As for market readiness, carbon markets are in their infancy and are expected to grow.

▪ Co-benefits of the project

From the outset, the project was conceived as a means of contributing not only to reducing deforestation and trapping carbon but also to protect biodiversity and improve the living standards of the community. In fact, these goals are entirely compatible with each other. The development of a sustainable form of agroforestry has reduced the rate of encroachment onto the savanna woodland; moreover, protection of the woodland from fire has helped to conserve the rich flora and fauna. All of this has created employment in an otherwise very poor rural community. It has also incidentally developed social capital, since villagers have learned technical skills and elements of business management.

▪ General prospects for this approach to climate abatement

The potential to use savanna woodlands as a means to protect the climate system opens many possibilities. For Miombo alone, the biomass carbon stocks of the 2.7 million km² throughout southern Africa may be conservatively estimated as 40 billion tons, more than ten times the annual emission of fossil-derived carbon for the entire planet. Moreover, the value of this carbon, assuming US$10 per tCO₂ (or US$37 per tC), is US$1.45 trillion. To this further carbon stocks of other types of African savanna may be added, of which mopane is the most significant. As we have seen, the benefits extend to nature conservation and the sustainable development of the region.

Is this a cheap means of protecting the climate? The cost of producing carbon offsets by this method, estimated above to be US$3.4 per tCO₂ by the time all currently available stocks have been sold, may be compared with the US$1–5 per tCO₂ for a project in Nepal Himalaya [9] and a range of US$0–200 per tCO₂ reported in a review of many forestry projects [10]. Generally, it is cheaper to protect existing forests and woodlands than it is to plant new ones, since the latter incurs a relatively high labor cost. More importantly, the capacity for reducing the rise in CO₂ is greater for avoided deforestation, and the co-benefits in the form of ecosystem services are greater.

Comparison is sometimes made between the costs of climate mitigation by forestry measures versus other geoengineering solutions. Recent reports have found afforestation and avoided deforestation to be the most affordable technologies [11], much cheaper than the nearest geoengineering rival, which is the use of stratospheric aerosols [12]. For most geoengineering approaches, the exact costs are not clear because all are in the pilot phase. Likewise, in avoided deforestation there are many uncertainties that will take decades to resolve [13].

Meanwhile, we see the development of another source of funding to avoid deforestation, based not upon project level funding but on reduced deforestation at national level. The current global model of REDD is currently being advanced by a group of countries who have together pledged more than US$3.5 billion over the next 2 years towards REDD. Southern Africa, and especially the Miombo region, might be one of the recipients of such funding [14]. REDD funding would go to the relevant government ministries, and it would be up to them to decide whether to implement REDD through their own agencies or to subcontract to companies that have developed expertise through specific pilot projects. There will be major governance challenges of any future REDD to ensure that deforestation is actually avoided and to see that payments do, at least in part, benefit local communities.

Future perspective

The complexity we have seen in this single case study, involving issues of governance and equity as well as practical and technical matters, emphasizes the challenge for REDD. Rich countries have pledged US$5 billion for REDD projects with US$30 billion on the horizon. It seems a substantial sum, especially since the world economy is still recovering from a global recession, yet it is tiny in relation to the global gross domestic product, which is approximately US$60 trillion. However, US$30 billion could pay for the protection of approximately 3 billion tCO₂ (0.82 billion tC, perhaps half of all the annual deforestation). In order to substantially reduce deforestation, paying governments to pay farmers and other land managers is therefore affordable, and, in principle, it is a cheap way to mitigate climate change. The issues to be addressed over the next 5–10 years are more related to conditions of politics, governance and the means of verification. The politics of the wealthy democratic countries requires political will, which so far has been forthcoming from Europe but less so from the USA [105]. The governance of tropical countries needs to be able, from the beginning, to channel large sums of money more efficiently and fairly than has hitherto been possible in most aid programs, thus demonstrating to donor countries that money has been properly allocated. As for verification, space agencies need to speed up plans for detecting and measuring biomass changes. It is expected that all of these conditions will occur in the next 5–10 years.

Table 1.  Carbon stocks and their value in the two land uses of the project. Carbon stock per ha has been multiplied by 3.66 to convert to CO2.

Land use	Area delineated (ha)	CO2 ha^-1	Saleable CO2 (tons)	Sold CO2 (tons)	Value of saleable CO2 (million US$ at 2009)
Planting	2877	73.2	210,596	54,629	2.1
Avoided deforestation	9301	82.7	769,192	101,455	7.7

Value estimated at market price of US$10 per tCO₂. The analysis covers the period 2003–2009.

Table 2.  Areas under tree planting and agroforestry, and the financial barriers (start-up costs and maintenance costs in the early stages of the plantation).

Type of tree planting	Hectares under contract	Households signed up	Start-up costs	Annual maintenance costs
Boundary planting	1721	1428	25 US$/100m	10 US$/100m
Dispersed inter-planting	997	886	145 US$/ha	62.5 US$/ha
Fruit orchard with cashew	64	95	480 US$/ha	200 US$/ha
Fruit orchard with mango	55	57	520 US$/ha	200 US$/ha
Homestead planting	57	330	480 US$/ha	200 US$/ha
Woodlot	103	103	1100 US$/ha	430 US$/ha

Data from Envirotrade’s Project Development Document [101].

Executive summary

	▪ The pilot project took place in the Sofala province of Mozambique, one of the poorest regions of the world, with the aims of avoiding CO2 emissions, protecting biodiversity and contributing to sustainable development.
	▪ The project used the Plan Vivo approach, whereby the farmers sign contracts to establish agroforestry systems in place of slash and burn, and to protect woodlands.
	▪ The native woodlands have been lost at a rate of 2.4% per year. Their average carbon stocks are 29.9 tC per ha-1 whilst the farm plots that replace them hold just a few tons per ha.
	▪ It has been possible to sell carbon credits on the voluntary carbon market at an average price of US$8.6 per ton. The cost of establishing the resource of saleable carbon is approximately US$3 per ton. Part of the income is used for helping to establish small industries and to develop community projects.
	▪ The project serves as an example of how reduced emissions from degradation and deforestation might be developed.

Acknowledgements

The authors wish to thank all our colleagues who participated in the project. At the University of Edinburgh: Luke Spaddevechia, Gudrun Wallentin, Silvia Flaherty, Sarah Carter and Jim Wright; at the Edinburgh Centre for Carbon Management: Alex Smith, Willie McGhee, Will Garrett and Jessica Orrego; at Envirotrade: Robin Birley, Piet van Zyl, Antonio Serra and the team of field technicians; Masters students: Rohit Jindal, Evelina Sambane, Joao Fernando, Roberto Zolho, Alastair Herd and Claire Ghee; External Evaluators: Taco Koistra, Jan Wolf and Antonio Marzoli. We wish especially to thank Joanna Pennie for her excellent work and patience as the project administrator.

Financial & competing interests disclosure

We acknowledge financial support from the European Union, contract Contract No. B7-6200/2002/063-241/MZ, and the helpful comments of the referees. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.