Agricultural sinks in the developing world: Different disciplines and different perspectives

Abstract

Agriculture is historically a source of atmospheric carbon. With the best management practices, this could be partly reversed, with a potential increase in size of the soil carbon pool that would have significant impact as a measure to mitigate climate change. This might benefit farmers twice over, as the carbon has a cash value, while the soil organic matter that contains it will assist productivity. There is therefore a business opportunity in carbon payments for farmers in the developed world. However, this article considers whether such payments might alleviate rural poverty and fund agricultural development in low-income countries. They might do so, but badly designed projects could lead to emission rather than sequestration, accompanied by productivity losses. Prevention of such perverse outcomes requires a multidisciplinary approach and an awareness of the political ecology context of sinks projects. An appropriate methodology is needed for the evaluation of soil carbon projects in developing countries.

Keywords:

• Agriculture
• carbon
• climate change
• efficiency
• equity
• soil organic matter

1. Introduction

Agriculture is a significant historical source of CO₂. If the processes that have caused this could be reversed, it would have the potential to be a useful carbon sink. This can be established by comparing the increase in atmospheric carbon since the Industrial Revolution to the amount emitted by agriculture since cultivation began. The former is about 167 petagrams (Pg) C, bringing the total atmospheric carbon pool to about 750 – 780 Pg C (Lal 1997, CDIAC 2003). The estimated amount emitted by agriculture has been 55 Pg C,1 of which perhaps 40 Pg C could be recovered through better management practices and the restoration of ecosystems (Lal & Bruce 1999). This is about a quarter of the anthropogenic emissions since 1850.

Farmers could benefit from this in two key ways. First, as countries enter into commitments to balance their emissions, this carbon could acquire cash value. During 2003, 78 million tons of carbon were traded in transactions that probably totalled about $330 million (Lecocq 2004). This is still a very small fraction of the global carbon pool. But the amounts are rising, with 64 million tons traded in January – May 2004.

The market is still dominated by a few participants, and their policy on acquisition of emissions credits must be subject to change. Prices are anyway highly variable, as Lecocq (2004) explains; although maximum prices exceed $6 per ton CO₂, the carbon bought may not eventually be certifiable, and the price depends in part on who agrees to bear that risk. Moreover, where projects are not Kyoto-compliant, buyers pay less; and although agricultural sinks may be eligible under Kyoto from 2012, this is uncertain. But even non-compliant projects average $1.34 per ton of CO₂; and it is likely that compliant and non-compliant trading systems will eventually become linked (Lecocq 2004), adding value even to non-compliant emissions reductions. This is prevalent more now that the EU has adopted a linking directive between the European carbon market and the Kyoto flexible instruments.

The second way in which farmers could benefit is that the practices recommended for sequestration – for example, soil-erosion control – should also enhance the productivity and long-term viability of agriculture, especially in fragile, marginal environments. This should have benefits for food security. Yet funding for agricultural research, development and investment in the developing world has been in decline since the 1980s. It would be useful if the market value of soil carbon can be harnessed to reverse this.

This paper concentrates on the developing world. Carbon is obviously also an opportunity for farmers in the developed world as well; but the issue the author is interested in is poverty alleviation, less of an issue in the developed world. In any case, the issues raised are not always the same. Farmers in the North already benefit from a subsidy regime, and some of its components could be modified (Renwick et al. 2003); the set-aside arrangements in the EU and the Conservation Reserve Program in the USA provide templates for devoting land to carbon sequestration, while producer or export subsidies can be redirected towards biofuels or crops that build up soil organic matter (Robbins 2004). There is scope for modifying such subsidies in areas where they currently cause emissions, for example in the livestock sector (Adger & Brown 1994). There is also greater capacity to operate trading mechanisms – and more experience, for example through the SO₂ (sulphur dioxide) market in the USA (Sandor et al. 2003).

Farmers in the developing world have fewer such mechanisms, and are more likely to trade through the flexible mechanisms of the Kyoto Protocol, in particular the Clean Development Mechanism (CDM); this is largely untested, but it already looks likely to raise issues of equity (Brown & Corbera 2003, Brown et al. 2004). They are likely to have smaller plot sizes, raising transaction costs,2 and their carbon abatement costs per ton will be high. This will make it hard for them to compete with large, economic projects, such as those in the power sector in India and China, which are expected to obtain much of the CDM funding (Austin et al. 1999). Last, but not least, attempting carbon sequestration in agriculture can have perverse outcomes; this is also true in the developed world, but the livelihoods consequences in low-income countries could be much graver, both because of the greater percentage of the population that relies on agriculture, and because their margin of existence is narrower.

By the same token, however, the gains may be greater; as Olsson and Ardö (2002) point out in their discussion of Sudan, a subsistence household will benefit proportionally more from external financial benefits from carbon. This is clearly worthwhile; FAO (2003a) estimates that 842 million people were undernourished in 1999 – 2001, mostly rural. This paper therefore focuses on the challenges and opportunities of agricultural carbon sequestration in developing countries, its pitfalls, and how they might be avoided through project design.

First, the position of agriculture in the carbon cycle should be considered.

2. Agriculture and the carbon cycle

Put crudely, plants are constructed in part from the conversion of CO₂ into organic matter. Some of this remains to augment the soil organic matter (SOM) content of the soil. If particles of SOM that are normally protected are exposed to the air and micro-organisms, their mineralization is accelerated, releasing CO₂ back into the atmosphere. Ploughing, wind and water erosion and any other process that disrupts the soil will increase this effect.

However, this mineralization must take place to some extent, to release nutrients to the crop. One could compare this process to the consumption of fossil fuel (which was once soil organic carbon (SOC)) by any other manufacturing process. Thus Albrecht (1938) pointed out that an acre (0.4 ha) of good soil in the USA's Midwest would burn carbon at an equivalent rate to 1.6 lb (just over 0.7 kg) an hour and that the heat thus produced would convert over 17 lb (about 7.7 kg) of water to steam at 100 lb (45.35 kg) pressure. On a warm July day, he suggested, every acre ‘may be roughly pictured as a factory using the equivalent of 1 horsepower.’

Whether or not one accepts this analogy, SOM is a raw material for food production. As McDonagh et al. (2001) have pointed out, there is currently much interest in management options that maintain SOM: ‘Levels of SOM have been shown to be particularly important in cultivated tropical soils where the high temperatures lead to rapid SOM breakdown, organic matter reserves are often low and the use of other inputs rare.’ Soil degradation and erosion cause substantial loss of SOM/SOC; their prevention and reversal are potentially important for the Kyoto Protocol (IPCC 2000), but also for food production.

The practices recommended for building up soil carbon are therefore both those that increase input of organic matter into the soil, and those that slow its decomposition (Batjes 2001). Some practices – for example, crop residue management – may do both. The Intergovernmental Panel on Climate Change (IPCC 2000) has given broad definitions of the management practices that could be used to sequester and/or retain soil carbon. There are three main categories: agricultural intensification; conservation tillage; and erosion control.

Agricultural intensification should sequester more carbon in theory by producing and conserving more carbon through better management of biomass. It may also be an alternative to extensification, whereby new and/or marginal land is brought into production; much of the historic loss of carbon from agriculture has happened in this way, and Vlek et al. (2004) have argued forcefully that the best option for carbon is to raise yields on existing cropland, and thus prevent deforestation.3 The distinction between intensification and extensification, and their respective effect on carbon pools, in Senegal have recently been highlighted by Liu et al. (2004), Tschakert and Tappan (2004) and Wood et al. (2004).

The second practice, conservation tillage, means less disturbance during ground preparation so that SOM should not be mineralized as quickly. Thirdly, erosion control, by reducing soil loss, also reduces mineralization, and has great theoretical potential for reducing emissions.

These IPCC definitions are not exhaustive. Follett (2001) cites increase of land cover through winter cover crops, nutrient inputs and supplemental irrigation. In his discussion of potential within the EU, Smith (2004) lists organic amendments (such as animal manure, sewage sludge, cereal straw and compost), improved rotations, deep-rooting crops and more. However, the IPCC strategies are very broad and cover most of these (rotations come under intensification, for example, while use of cereal straw is part of conservation tillage (CT); these will be discussed below).

There are some areas that are not covered – in particular, rangeland; but the IPCC has examined this elsewhere (Allen-Diaz et al. 1996), and its potential is discussed later in this article. Conversion to grassland or set-aside is not; although it can sequester a great deal of carbon, it is assumed that it would rarely be an option for resource-poor farmers. But there would be exceptions. Also omitted is the great potential in rehabilitation of degraded land, highlighted by Lal (1997). This has huge potential for carbon sequestration, funding for which could (for example) enable restoration of huge areas of the world's cropland, especially irrigated areas, that are affected by soil salinity (Lal & Bruce 1999). So although these three areas provide a useful working framework, they should be taken as a guide only.

Table I gives a rough guide as to the amount of carbon these three areas might sequester a year over the period taken to reach equilibrium, normally about 20 – 50 years (the amounts would be greater in the early years).

Table I. Carbon sequestration potential of strategies for arable land (Pg C/year).

Practice	Potential
Soil erosion control*	0.08 – 0.12
Soil restoration†	0.02 – 0.03
Conservation tillage and residue management	0.150 – 0.175
Improved farming/cropping systems (intensification)	0.18 – 0.24
Total	0.43 – 0.57

*However, this figure should be compared with Lal's later (2003) estimates of CO₂ lost through erosion. †This estimate includes the use of degraded soils for biofuels production.Source: adapted from Lal and Bruce (1999, p. 182).

These are global estimates. Regional analyses are harder to come by, and it is not easy to estimate how much of the global carbon sequestration potential in agriculture lies in the developing world. Bloomfield and Pearson (2000) say that there is a relative lack of regional analyses for developing countries, but suggest that the sequestration potential of cultivated tropical agricultural soils would be 8.9 – 11.8 Pg C by 2050, or about 0.178 – 0.236 Pg C/year – less than half the total of agriculture worldwide, but not insignificant. However, comparing different sources is problematic, as the assumptions are almost never the same, not least on how much land is likely to be managed in a carbon-friendly manner (Robbins 2004). A greater measure of agreement is needed on these assumptions. In the meantime, figures for sequestration potential, like carbon prices, should be taken as indicative only.

The remainder of this paper discusses some of those practices, and why they would not always have a ‘win – win’ outcome. It finishes by considering potential methodologies for the evaluation of sequestration projects that would avoid such dangers.

3. Agricultural intensification

The production of more biomass per hectare converts more CO₂ into plant material. So raising productivity should be a climate mitigation strategy as well as producing more food and income. If carbon credits can be used to finance inputs such as fertilizer, pesticide and improved cultivars, there would appear to be a ‘win – win’ scenario. But a carbon budget for such an initiative would have to consider the manufacture and transport of any inputs – and any on-farm emissions, such as methane from additional livestock.

Also, experience during the Green Revolution suggests that intensification can be of greater benefit to farmers at higher income levels, who are better able to use the extra inputs. In an environmental project, sustainability depends in part on equity between stakeholders (Adger et al. 2002); indeed Brown et al. (2004) have argued this point with specific reference to the Kyoto mechanisms, and there is already some evidence that it will be an issue with forestry projects (Brown & Corbera 2003, Brown et al. 2004). Tschakert (2004) has carried out a detailed cost – benefit analysis of carbon-friendly management practices for Senegal; her evidence suggests that it is those with greater resources who are most likely to benefit. It has also been suggested that poorer farmers will be less able and willing to assume the risks involved in any form of carbon contract (FAO 2002b).

The experience of the Green Revolution also highlights the distinction between simply increasing biomass, and managing it sustainably. Ali and Byerlee (2000) related production and yield trends to indicators for the natural-resource base in the Pakistan Punjab in the years following the Green Revolution. SOM had declined by 33% over the study period. Intensification for carbon would aim at building up SOM and would clearly demand more sustainable management of biomass.

This underlines the need for well-planned inputs. Irrigation, for example, carries risks; about 8% of the world's irrigated land is salinized to some extent, rising to 25% in semi-arid regions (FAO 2002a). Moreover, agriculture is responsible for about 69% of all freshwater withdrawals, and one developing country in five is expected to face water shortages by 2030 (Gregory et al. 2002). Thirty-three countries are now dependent on external sources for over half their renewable water resources (FAO 2003b).

But 40% of the world's food comes from the 17% of the cropland that is irrigated (Borlaug 2001, FAO 2002a), and it cannot be dispensed with. Besides, it is poor irrigation management, not irrigation as such, that has salinized large areas. And in some regions, water is still available; FAO (2002b, 2003c) has pointed out that sub-Saharan Africa has large untapped reserves of groundwater. Tanzania's Ministry of Water and Livestock Development has estimated that as little as 15% of the country's irrigation potential is being used (Lankford 2003). Carbon funding to increase biomass could release some of this potential.

Other forms of intensification through good management might also be used sustainably for carbon sequestration. An example is the ley farming developed by the International Center for Agricultural Research in the Dry Areas (ICARDA) in the 1980s and 1990s. This was in fact designed to address nutrient mining through cereals monocropping in West Asia (Christiansen & Manners 1995). Cereals are rotated with feed legumes, building up organic matter and nutrients that have been depleted by wheat or barley cultivation. Although not designed specifically to sequester carbon, it does do so (Ryan & Pala 2002). It does require investment in machinery (Christiansen & Manners 1995, Bahhady & Robbins 1998, Haidar et al. 1998); this might be funded with carbon credits. True, a commercial concern might hesitate to put money up front, but a body such as Global Environment Facility (GEF) or the World Bank's BioCarbon Fund might do so. However, there would be pitfalls: increased methane emissions from livestock, for example, and the opportunity costs of not monocropping cereals would vary with local wheat prices, which might themselves be affected by the project. Any carbon sequestration proposal on these lines would need multidisciplinary review.

4. No-till and low-till

As stated above, ploughing accelerates mineralization of SOM. This also increases erodibility, as soils become relatively more compacted, losing infiltration capacity; water is more likely to run off, and is also less likely to reach the crop's root zone. This applies especially in arid areas, where SOM is relatively low, and where rainfall can be concentrated in time and space.4

But it has been argued that most tropical soils do not need to be tilled (FAO 1998). Not tilling can be accompanied by leaving crop residues in the field; this reduces erodibility and encourages biotic activity. This is the basis of conservation agriculture, which normally involves no tillage at all, and reduced or conservation tillage (CT); the latter permits specific types of low-intensity tillage, and specifies that a minimum of 30% of crop residues be left in the field (IPCC 2000). These techniques originated in the USA in the second half of the 20th century as a response to land degradation, and have spread – in particular to southern Brazil, where they have been adopted over the last 30 years to combat declining productivity (FAO 2001, Sisti et al. 2004).

CT can also sequester carbon; this has already been widely appreciated in the USA (USDA-ARS 1997, IPCC 2000, Robbins 2004). CO₂ emissions are also cut on mechanized farms through fuel savings, which can reach 66% (Evers & Agostini 2001). There can also be productivity gains, with dramatic increases in maize and wheat yields reported in Brazil – although other evidence is more complex (Evers & Agostini 2001, Pretty & Ball 2001). These gains come from a higher level of soil nutrients, but also because microaggregates form around decomposing SOM, forming a better-aggregated soil that permits greater water infiltration (Lal 1997).

This seems to be a win – win strategy, but CT does have drawbacks. Farmers till to prepare a seedbed; that can be done through the crop residues using a seed drill instead, but tillage also exposes and controls pests and diseases. Arguably carbon funding for CT could deal with this by providing pesticides, but as Lal (1998) found in Western Nigeria, this can result in a loss of soil biota and a consequent collapse of soil structure, which would defeat the object of CT. Such inputs would need to be used with care. So appropriate rotations are needed to counter pests and diseases. They may also define how much carbon is really sequestered. For example, Sisti et al. (2004), in a long-term experiment in southern Brazil, found that a green-manure legume had to be included in a rotation to increase carbon stocks. And adequate nutrient cycling also needs to be ensured through appropriate practices. Mrabet et al. (2001), have also pointed out that nitrogen, phosphorus and other nutrients are supplied via mineralization, and that not tilling may affect this.

There are other questions. No- or low-till and crop residues can lower soil temperature (Lal et al. 1989) – usually a good thing in the tropics as it will slow SOM breakdown, but in some climates it may delay planting dates (Zeleke et al. 2004). Soils can become anaerobic under no-till, resulting in N₂O emissions; one estimate (European Commission 2000) has suggested this could offset 50 – 60% of gains in carbon.5 Faced with this complexity, farmers would need to develop flexible combinations of cropping sequences, tillage regimes and input use. It would be difficult to match payment for carbon-friendly practices with this sort of dynamic management. Moreover there would usually be opportunity costs from the retention of crop residues. In crop/livestock systems, crop residues are grazed, or used for fodder. This could even lead to carbon leakage; where such systems border on rangeland, such as large areas of the Middle East, the residues are traded with pastoralists, and their withdrawal could lead to overgrazing and loss of carbon from marginal grasslands, negating or reversing any carbon accumulation (Robbins 2004).

But CT may be increasingly needed to keep degraded arable land productive, and it may be necessary to meet these challenges even if carbon sequestration is not an objective. Given that adoption is often constrained by a lack of knowledge (IPCC 2000, Díaz-Zorita et al. 2002), carbon credits could fund technology transfer. Again, however, a carbon payments scheme for CT would be best planned with input from extension, agronomic, agricultural economics and farming systems perspectives.

5. Soil conservation, grassland management: Unintended consequences?

A further sequestration tactic is to prevent emissions from soil erosion; disturbance of soil by wind and water, as well as tillage, accelerates mineralization. The effect of this is not easy to quantify. Lal (2003) suggests that 4 – 6 Pg C is being translocated by erosion every year, but argues that much is redistributed, ends up in depressional sites or remains sequestered through aquatic ecosystems. The amount that is mineralized is controversial, with estimates varying between 20 and 70% (Jacinthe & Lal 2001); Lal inclines towards the lower figure and thinks the total is perhaps 0.8 – 1.2 Pg C a year (Lal 2003). But that is still enough to offset perhaps a third of the net annual increase in the atmospheric carbon pool – which is thought to be about 3.3 Pg C a year. Given that mechanical soil-conservation measures such as terracing and bunds often require more resources, especially labour inputs, than the farmer has to hand, providing such resources might seem a strong contender for carbon funding.

But no-one really knows how much soil is being lost by erosion; Lal himself has suggested that most available statistics are ‘subjective, qualitative, obsolete, crude and unreliable’ (Lal 2003). Moreover, prevention of soil loss may not itself have much effect on productive capacity; for, as Lu and Stocking (2000) demonstrate in their analysis of China's Loess Plateau, not all soils are worth saving. It may be that off-farm activities or even migration make more sense than prevention of soil loss even if such prevention is possible (Blaikie 1989). So farmers may hesitate to use scarce resources for the construction of terraces and bunds.

Offering carbon credits could persuade them in some cases. But experience suggests that distorting farmers' priorities like this can be counterproductive, even where farmers do regard soil erosion as a problem. Thus Bewket and Sterk (2002) found that farmers in East Gojjam, Ethiopia, acknowledged the need to conserve soil but had participated unwillingly in soil conservation as the prescribed bunds had caused more erosion than they prevented. Similar failures in soil-conservation programmes have been reported elsewhere. Hellin and Haigh (2002) list a number of reasons why farmers reject soil conservation, including inappropriate measures, non-acceptance of erosion as a problem, labour costs and lack of productivity benefits. There is a large reservoir of available literature on this.

It would be easy to conclude that carbon credits should not fund soil conservation, as the potential for perverse outcomes is just too great. However, soil erosion is a problem and so is climate change, even if their extent and consequences are still uncertain. Any model which finds synthesis between the mitigation of the two would be welcome.

The evidence suggests a need for programmes to be designed in concert with farmers, who can build their own priorities into the programme (Zöbisch et al. 1997). It also indicates that soil conservation, like other forms of agricultural technology, is best applied through an iterative process in which farmers accept, reject and modify measures as they go along – implying a need for process - rather than target-led programmes (McDonald & Brown 2000, Hellin & Haigh 2002). Attaching carbon funding to specific packages of technology could reproduce exactly the approach soil conservation that has failed in the past. CT and intensification raise the same issue; although they have different apparent drawbacks (Table II), they generally arise from their site-specificity in both cases.

Table II. Potential pitfalls of carbon sequestration strategies in agriculture.

	Conservation tillage
Intensification	(including residue management)	Soil erosion control
Increased CO2, N2O and CH4 from manufacture, transport, and use of organic and inorganic fertilizer	Increased incidence of pests and diseases	Confusion as to extent of erosion
	Opportunity costs of retaining crop residues in the field
Equity issues through differing capacities to use inputs	No carbon sequestration without appropriate rotations	Some soils with low organic matter are not worth saving, either for productivity or carbon sequestration
	Changes in soil temperature may affect planting dates
Salinization and groundwater mining through poor irrigation management	Risk, especially with true no-till, of anaerobic soils, reducing net mitigation through N2O emissions	Incentive payments can lead to conservation measures that are not maintained later
Opportunity costs of abandoning monocropping for sustainable rotations		High opportunity cost of scarce labour

This might be accommodated by actually measuring farmers' soil carbon rather than extrapolating it from the practices adopted. Once baseline data is available, estimates can then be based on remote sensing, which can indicate biomass. Such measurements could certainly be practical in areas with large farm sizes, but monitoring fragmented holdings of a hectare or two could present difficulties (Robbins 2004). Although these may not be insuperable (for example, farmers might trade as groups), this would also lower transaction costs, and certain land-tenure systems might require such an approach anyway. The choice between extrapolating sequestration rates from agreed packages of practices, and actual measurement of biomass, is a major challenge for the monetization of agricultural carbon pools.6

Rangeland also demonstrates the dangers of targets based on agreed land-management practices. It is a crucial carbon sink, but can be a tricky resource to define. Rango et al. (2002) quote Holechek et al.'s (1995) definition of it as uncultivated land that supports grazing and browsing animals; they argue that it therefore includes deserts, forests and all natural grasslands, but most definitions would not include forests. Sathaye and Meyers (1995) refer to ‘grasslands, savannas, and deserts’. Allen-Diaz et al. (1996, p. 134), in their contribution to Climate Change 1995, state that rangelands:

“[O]ccupy approximately 51 percent of the terrestrial surface of the Earth, or 68.5 million km²… [And] include unimproved grasslands, shrublands, savannas, and hot and cold deserts.”

It is as hard to be precise about the potential for grassland carbon as it is to estimate that for arable farming. Ni (2002) highlights the wide variation in estimates of China's grassland carbon storage, and ascribes them to differences in classification and estimation. But Allen-Diaz et al. (1996) states that rangeland contains about 36% of the world's total carbon in above- and below-ground biomass; Sathaye and Meyers (1995) quote estimates of 417 Pg C, mostly below ground. FAO (2000) suggest a figure of anywhere between 200 and 420 Pg C – again, mostly below ground and stable; it adds that grassland contains about 70 Mg/ha of soil carbon, similar to the content of forest soils. But FAO also suggests that perhaps 70% is degraded (FAO 2000). Again, while views on this differ, there is a general consensus that it is a problem. Batjes (2004) quotes Oldeman's estimate that about 31% of the permanent pastureland in Africa is affected by anthropogenic soil degradation. Gintzburger (1996) reports that about 9 million km² of the world's drylands have been rendered unproductive in the last 50 years.

The grave consequences for animal production make restoration and protection of grasslands a pressing concern. Again, carbon may provide resources for this. Sathaye and Meyers (1995) quote UNEP's 1991 estimate that rehabilitation of the world's rangelands would cost about $5 – 8 billion a year over 20 years. They report that, using the CENTURY model, sustainable rangeland management could sequester about 0.7 billion tons of carbon a year at the cost of $10 per ton, ‘comparable or superior to the estimates in the forestry sector’ (Sathaye and Meyers 1995). The World Bank and GEF are aware of this potential and have incorporated either approximate market or shadow prices for carbon in their analyses of pasture and grassland projects.

However, the challenge will again be to find sustainable management practices that do not impose inflexible management. The classical answer to steppe, or grassland, degradation is to rigidly control grazing. But in the semi-arid steppes of the Middle East and Central Asia, pastoralists need more, not less, freedom of movement because of the non-equilibrial nature of such fragile environments. As Sathaye and Meyers (1995) explain, rangeland response to grazing and rainfall has normally assumed a linear response to grazing pressure and that ultimate productivity potential is a constant. But it is now understood that the number of animals can be only loosely related to vegetation dynamics (World Bank 2003). So rangeland ecology now recognizes management strategies in which pastoralists will track the availability of grazing, often over quite large areas (Bruce & Mearns 2002).

Historically, pastoralists have functioned that way. But this has led outside observers to assume an open-access system – a ‘tragedy of the commons’ – implying an urgent need to impose controls (Rae et al. 2001). There are thus few countries in which the legislative framework permits such mobility (Bruce & Mearns 2002). Ironically, such regulations may displace unseen but functioning traditional mechanisms without providing a satisfactory substitute (Rae & Arab 1996).

The problems this presents for the monetization of grassland carbon are serious, but probably not insuperable. They do suggest that paying groups for managing carbon stocks over a defined area might not work. However, evidence from Mongolia suggests that degradation since 1991 has been in part caused by declining herd mobility due to a need to be near facilities (schools, hospitals, veterinary services, markets). Could their provision be funded by carbon-related payments? Thus, rather than being used to fund measures that limit rangeland use, the CDM or other transfers from developed countries could enhance sinks by helping people to move around. This does involve standing traditional ideas on their heads, but such lateral thinking may be needed if payment for sinks is to work.

6. Political ecology – A paradigm for sinks management?

The evidence so far presented might be summarized as follows: There is great potential to sequester carbon and increase productivity, but poor project design could have the opposite effect, and the difference between the two may lie in an ability to interpret the farming system from the farmers' point of view.

There is also a need for a multidisciplinary approach to soil carbon. Besides links between upstream and downstream research, horizontal linkages are also needed between agricultural economics, soil chemistry, agronomy, livestock management and crop science. For example, the literature on perverse outcomes in soil conservation is well known to soil scientists. It is less well known to the environmental scientists and international civil servants who are likely to end up administering the Kyoto flexible mechanisms, such as the CDM, that may be used to implement funding for carbon sinks.

However, there is another reason why the potential for perverse outcomes might be overlooked. Adams (2001) refers to “the view of sustainable development as essentially a managerial process, where reform of procedures will ensure some ‘optimal’ outcome”. Such a reform of procedure could include a nominal commitment to multidisciplinarity in sinks policy, with an inbuilt assumption that – provided this is ensured – a sinks project will not misfire. That may not be good enough. Stocking (1996) warns that responses to soil erosion in Africa have been coloured by the funding needs and political aspirations of those concerned: ‘Where environmental policy is to be affected and development action is to be proposed, erosion assessments should themselves be audited as to who has made them, and why’, he comments.

This suggests that sinks proposals should be viewed within a framework of political ecology. Tschakert (2001), in her discussion of her work in Senegal's Old Peanut Basin, argues for a political ecology approach, stating that: ‘Social scientists, accustomed to the crosscurrents of complexity and diversity, can help to bridge the disciplinary gap that has been developing in this field of climate change research.’ However, this could be taken further. At its most radical, political ecology combines with post-structural world views in which environmental challenges imply, in effect, that there is no single reality. As Blaikie (1996) explains, ‘there is an infinitely variable and subjective set of accounts from actors who reflect upon and try to make sense of their worlds’.

On the face of it, this will not help unravel the complexities of (for example) managing rangeland carbon. However, Forsyth (2003) argues that acknowledging the social construction of science need not imply rejection of the orthodox scientific study of biophysical phenomena; rather, he advocates ‘greater participation in the formulation of environmental science’. Arguably such participatory science does exist – farmer-participatory crop breeding, for example.

In the case of soil carbon, this seems to require an actor-oriented approach in order to incorporate multiple perspectives and values into research and/or project planning. As Brown (1998) has pointed out, conventional economics is ill-equipped to value biodiversity in a rainforest because it has a high existence value for people in the developed world, but a high use value for local people. Soil organic matter is the same. Development interventions designed to reconcile the different values attached to it must incorporate an understanding of why those values are assigned by different actors. Thus a farmer may be persuaded, by carbon funding, to leave his cereal straw in the field as part of a no-till system. But these residues will then not be available to farmers elsewhere in the farming system, who may deplete their own reserves of organic matter as a result. Moreover, as Long (1984) has said, development interventions interact with existing social phenomena. Carbon credits are no exception; land tenure and rent-seeking by corrupt officials are examples. An actor-oriented approach will be needed to uncover such issues. The task will be to see who benefits from a soil carbon project, and who doesn't, and why.

7. Choosing a methodology

The most obvious methodology for establishing this is through cost – benefit analysis (CBA). The price, or shadow price, of carbon is open to discussion, but in principle it can be quantified, as can other indicators for the rural or pastoral economy (inputs, opportunity costs for labour, farm-gate prices, etc.).

However, CBA has drawbacks in environmental management. To measure the ‘greatest good of the greatest number’ can emphasize or neglect groups of stakeholders. There is a special danger of this with carbon projects in that a major stakeholder is the rest of the world – that is to say, anyone that benefits from the removal of a ton of CO₂ from the atmosphere. As Brown and Corbera (2003) put it: ‘The development of carbon markets may privilege global claims over those of other users and scales’ – usually local land-users. So a straightforward CBA could lead to farmers/pastoralists implementing a project that was against their interests, and they would have to be persuaded to cooperate through cash incentives. As discussed above, this initiative has not normally been sustainable.

An alternative might be CBA of the management practices recommended for carbon sequestration without including external funding as an element. This establishes the extent to which such practices are congruent with farmers' long-term interests. It may also uncover equity or other issues, as Tschakert (2004) found in Senegal. This approach has the important advantage of assessing the long-term prospects of any carbon pools after external funding has ceased. From an actor-oriented perspective, however, it raises the opposite question: instead of privileging global claims, it privileges those of the farmers over anyone else who bears the external cost of carbon emission.

Such considerations have led to the development of alternatives such as stakeholder analysis, offering the ‘explicit consideration of potential trade-offs between different stakeholders’ (Grimble 1998). In its simplest form, the output from stakeholder analysis could consist of a matrix in which the same list of stakeholders form both axes, and boxes are ticked or crossed to indicate a complementarity or conflict of interests. However, this assists in project design only in a very broad sense. It might be found, for example, that ‘carbon sequestration’ conflicts with the ‘mobility of herds’, but there would be no indication as to why, or to which trade-offs might be acceptable in compensation. Neither is there any quantitative indication of the gravity – or otherwise – of any conflict of interest.

A stakeholder Multi-Criteria Analysis (MCA) does fundamentally the same thing but in a more sophisticated form, so that these factors can be made explicit. A simple MCA7 would begin with a list of objectives (for example, sequestration and livelihoods objectives). Different options are then identified for attaining those objectives (for example, use of crop residues, or management options for rangeland). Next a set of criteria is devised by which to measure attainment of those objectives (again, they might be the extent of carbon sequestration in the soil, health of rangeland, farming incomes or other outputs). The management options would then be ranked against their probable score on each criterion. Different weights can be attached to each criterion. It is this that makes MCA suitable for environmental decisions, in that the stakeholders themselves can be asked to weight the criteria. Stakeholder MCA is already an established tool for environmental planning, both in the developed world and elsewhere (Dodgson et al. 2000, Brown et al. 2002). Its use for the assessment of carbon sequestration is more recent, but has been applied to existing carbon projects in forestry in Mexico (Brown & Corbera 2003, Brown et al. 2004).

There are drawbacks. Stakeholder preferences must be elicited in such a way as to reflect the weight the respondent really attaches to them; this means that they should, ideally, be asked to mark each preference out of 100. Yet it is also necessary to avoid imposing too much cognitive complexity on respondents (Tompkins 2003). Arguably MCA is best used in semi-structured interviews, supported by key informant interviews in which interviewees can raise their own concerns. But it does offer a possible methodology for the evaluation of sinks proposals in agriculture, and others will emerge if carbon farming proves to be an attractive option.

8. Conclusions

This paper has not attempted to cover all the challenges posed by agricultural sinks. It has, for example, said little about monitoring and verification, or about the potential trading framework and the problems of transaction costs that would have to be overcome if resource-poor farmers were to benefit. These issues have been discussed elsewhere.8 Neither has there been any discussion of the fundamental equity or otherwise of selling carbon to developed countries. That is not because the author does not wish to enter that debate, but because it is beyond the scope of the paper.

What this paper has done, is to examine some of the practical pitfalls of agricultural sinks, and to suggest why these might be overlooked. Two basic reasons have emerged. One is the wide range of disciplines involved in the farming system. This can only be addressed through the multidisciplinary assessment of sinks projects. It has been argued that this must involve linkages between upstream and downstream research. Moreover, there is also a need for horizontal linkages across the range of disciplines at every level. The multiple roles of crop residues in some farming systems are a good example of why this is so.

But this paper has also argued for a broader appreciation of the political ecology context in which sinks projects should be assessed. As Stocking comments, ‘funding needs and political aspirations’ can colour perceptions, and this is as true of sinks as it is of soil erosion. It has therefore been argued that an actor-oriented approach is necessary for an assessment of a proposed agricultural sink project. Finally, a methodology has been identified that may answer that need. In the end, however, only practical experience with agricultural sinks projects will reveal whether the ‘win – win’ potential they seem to hold will be realized.

Acknowledgements

The author gratefully acknowledges the support of the UK's Natural Environment Research Council and Economic and Social Research Council.

Notes

Other sources put it rather higher – see, for example, Lal (2004).

Although these may be addressed by imaginative project design within the CDM; Baumert et al. (2000) describe how this might be done.

The authors suggest using carbon credits to finance greater fertilizer use.

This may not apply on slopes above a certain percentage, where the ground may need to be tilled to prevent runoff; Prinz et al. (1994) describe a case of this sort in Northern Algeria. Such cases would be relatively rare, but do underline the site-specificity of tillage regimes.

But this may not occur under reduced, rather than zero, tillage regimes; the definition of CT quoted above does accept these.

There is some debate about monitoring and verification. Lal (2004) argues that there is no fundamental problem and that the challenge is one of standardization. Robbins (2004) and Smith (2004) raise more difficulties, although neither suggests that it cannot be done.

The description of MCA is based in part on Dodgson et al. (2000).

See, for example, Baumert et al. (2000), FAO (2002b), Post et al. (2001) and Robbins (2004).