doi:10.3763/ijas.2007.0322

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62© 2008 Earthscan. ISSN: 1473-5903 (print), 1747-762X (online). www.earthscanjournals.com

Multi-year assessment of Unilever’s progress towards agricultural sustainability I: indicators, methodology and pilot farm results

J. Pretty1*, G. Smith2, K.W.T. Goulding3, S.J. Groves4, I. Henderson5, R.E. Hine1, V. King2, J. van Oostrum2, D.J. Pendlington6, J.K. Vis6 and C. Walter6

1Department of Biological Sciences and Centre for Environment and Society, University of Essex, Colchester, UK; 2Unilever Research & Development, Colworth Laboratory, Sharnbrook, Bedford, UK; 3Department of Soil Science, Rothamsted Research, Harpenden, Herts, UK; 4ADAS, Gleadthorpe Grange, Meden Vale, Mansfi eld, Nottinghamshire, UK; 5British Trust for Ornithology, Thetford, Norfolk, UK; and 6Unilever Rotterdam, Weena 455, Rotterdam, Netherlands

This review describes the establishment in 1997 of an agricultural sustainability initiative by the foods, home and personal care company, Unilever. It analyses the development and testing of a system of indicators used over several years on the company’s model research farm at Colworth in the UK. The approach taken was fi rst to develop a sustainability audit, based around a common set of indicators, and then to support pilot projects for a select number of crops, with the aim of adapting parameters for each crop, establishing baselines, developing recommendations to increase agricultural sustainability, and holding fi eld trials to test these new practices and technologies.

The purpose of the initiative was the development of a system of agricultural assessment that would be practical and effective over short time scales so that changes in company policies and practice could be made. The indicator structure developed uses 10 clusters of indicators (later revised to 11). These had to be easily measurable, and so not costly; relatively non-contestable, and so convincing to internal and external stakeholders; responsive to management action; and lead to value creation for farmers, rural communities and businesses. It was found by Unilever that the main advantage of the audit was not in the emergence of a sustainability index (which was rejected), but in the development of increased knowledge and understanding of agricultural and environmental interactions that emerged during the discussion and assessment of the indicators. The process of its use was more important than any scores that emerged.

This paper summarizes the changes in selected indicators for each of fi ve novel management practices tested on the pilot farm (spring versus winter cropping; reduced nitrogen fertilizers; reduced pesticide applications; mixed rotation and cover crops; and fi eld margin management). A brief analysis of the agronomic conclusions is given for each. The overall conclusion for farm practices from this research is that an optimal rotation has both spring and winter crops, as this spreads labour costs on farm and environmental costs. The results of the Colworth project suggest that key components

*Corresponding author. Email: jpretty@essex.ac.ukAn Associate Editor of the International Journal of Agricultural Sustainability, Colin Sage, acted as the sole handling editor for this paper, which was peer-reviewed in the normal way. The lead author was neither involved in the selection of referees nor in the decision to accept the fi nal version of this paper.

38 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

of successful sustainable farming projects include management to create a more diverse landscape, and close attention to the timing and frequency of agrochemical applications.

Keywords: agricultural sustainability, indicators, private sector, policy, UK

Introduction

There have been remarkable rises in agricultural output since the advent and spread of modern meth-ods of agricultural production. Aggregate world food production has grown by 145% since the beginning of the 1960s. In Africa, growth was by 140%, in Latin America 200%, and in Asia 280%. In indus-trialized countries, though production started from a higher base, it still doubled in the USA, and rose by 68% in Western Europe (FAO, 2006). Over the same period, there have been shifts in consumer behaviour over food, and the wider political econ-omy of farming and food (Ferro-Luzzi & James, 2000; Frumkin, 2005; Goodman & Watts, 1997), and agricultural systems are now recognized to be a signifi cant source of environmental harm (Conway, 1997; MEA, 2005; Pretty et al., 2000; Tilman, 1999). Since the early 1960s, the total agricultural area has expanded by 11% from 4.5 to 5 billion hectares, and the arable area from 1.27 to 1.40 bil-lion ha. Livestock production has also risen, with a worldwide fourfold increase in numbers of chick-ens, twofold increase in pigs, and 40–50% increases in numbers of cattle, sheep and goats. The intensity of production on agricultural lands has also risen substantially. The area under irrigation and number of agricultural machines has grown by about two-fold, and the consumption of all fertilizers by four-fold (and nitrogen fertilizers by sevenfold). The use of pesticides in agriculture has also increased, and now amounts to some 2.56 billion kg per year (Hazell & Wood, 2007; Pretty, 2007).

The ineffi cient use of some of these inputs and factors of production, however, led to signifi cant change to local environments, with effects often extending further afi eld. Increased agricultural area contributes substantially to the loss of habitats, associated biodiversity and their valuable environ-mental services (McNeely & Scherr, 2003; MEA, 2005). Some 30–80% of nitrogen applied to farm-land escapes to contaminate water systems and the atmosphere, as well as increasing the incidence of

some disease vectors (Giles, 2005; Pretty et al., 2003a; Smil, 2001; Townsend et al., 2003; Victor & Reuben, 2002). Irrigation water is often used ineffi ciently, causing waterlogging and salinization, as well as diverting water from other domestic and industrial users, and agricultural machinery has increased the consumption of fossil fuels in food production (Leach, 1976; Stout, 1998). There is also growing evidence to suggest that the aggregate costs in terms of lost or foregone benefi ts from environ-mental services are too great for the world to bear (MEA, 2005; Ruttan, 1999). The costs of these environmental problems are often called externali-ties, as they do not appear in any formal account-ing systems (Baumol & Oates, 1988; Hanley et al., 1998). Yet many agricultural systems themselves are now threatened because key natural assets that provide important services are being undermined or diminished, for example, soil quality through erosion and salinization and water quantity and quality. Agricultural systems in all parts of the world will have to make improvements. In many agricultural systems, the challenge is to increase food production to solve immediate problems of hunger. In others, the focus will be more on adjust-ments that maintain food production whilst increas-ing the fl ow of environmental goods and services.

Today, concerns about sustainability centre on the need to develop agricultural technologies and practices that (1) do not have adverse effects on the environment (partly because the environment is an important asset for farming); and (2) are accessible to and effective for farmers, and lead both to improvements in food productivity and have positive side-effects on environmental goods and services. Sustainability in agricultural systems incorporates concepts of both resilience (the capac-ity of systems to buffer shocks and stresses) and persistence (the capacity of systems to continue over long periods), and addresses multifactoral economic, social and environmental outcomes.

This paper describes the establishment in 1997 of an Agricultural Sustainability Initiative by the

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 39

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

food and home care company, Unilever, analyses the development of a system of indicators, and evaluates their use over several years on the com-pany’s model research farm at Colworth in the UK. The second paper (Pretty et al., 2008) describes the use of these indicators on a variety of crops (peas, spinach, tomatoes, tea and oil palm) in both industrialized (Australia, Germany, Greece, Italy, UK and USA) and developing countries (Brazil, Ghana, India, Kenya and Tanzania).

Assessing agricultural sustainability

Many different expressions have come to be used to imply greater sustainability in some agricultural systems over prevailing ones (both pre-industrial and industrialized). These include biodynamic, community-based, ecoagriculture, ecological, environmentally sensitive, extensive, farm-fresh, free-range, low-input, organic, permaculture, sustainable and wise-use (Bawden, 2007; Clements & Shrestha, 2004; Conway, 1997; Cox et al., 2004; Gliessman, 2005; McNeely & Scherr, 2003; NRC, 2000; Pretty, 1995). There is continuing and intense debate about whether agricultural systems using some of these terms can qualify as sustainable (Altieri, 1995; Balfour, 1943; Lampkin & Padel, 1994; Trewevas, 2002). Sustainable systems can be taken as those that aim to make the best use of environmental goods and services whilst not damaging these assets (ACRE, 2007; Altieri, 1995; Conway, 1997; Gliessman, 2004, 2005; Hinchliffe et al., 1999; Jackson & Jackson, 2002; Li Wenhua, 2001; McNeely & Scherr, 2003; MEA, 2005, NRC, 2000; Pretty, 1995, 2005, 2007; Swift et al., 2004; Tilman et al., 2002; Tomich et al., 2004; Uphoff, 2002). The key principles of agricultural sustain-ability are to:

(1) integrate biological and ecological processes such as nutrient cycling, nitrogen fi xation, soil regeneration, competition, predation and para-sitism into food production processes;

(2) minimize the use of those non-renewable inputs and any inputs that cause harm to the environ-ment or to the health of farmers and consumers;

(3) make productive use of the knowledge and skills of farmers, farm workers and advisors, so improving their self-reliance and substitut-ing human capital for costly external inputs;

(4) make productive use of people’s collective capacities to work together to solve common agricultural and natural resource problems, such as for pest, watershed, irrigation, forest and credit management; and

(5) provide a living for the farm family (i.e. are economically sustainable).

The idea of agricultural sustainability, though, does not mean ruling out any technologies or practices on ideological grounds. If a technology works to improve productivity for farmers, and does not cause undue harm to the environment, then it is likely to have some positive sustainability outcomes. Agricultural systems emphasizing these principles also tend to be multi-functional within landscapes and economies (Dobbs & Pretty, 2004; MEA, 2005; Pretty et al., 2006). They jointly produce food and other goods for farmers and markets, but also contribute to a range of valued public goods, such as clean water, wildlife and habitats, carbon sequestration, fl ood protection, groundwater recharge, landscape amenity value, and leisure/tourism. In this way, sustainability can be seen as both relative and case-dependent, and implies a balance between a range of agricultural and environmental goods and services.

As a more sustainable agriculture seeks to make the best use of nature’s goods and services, so technologies and practices must be locally adapted (Conway, 1997; Costanza et al., 2007). These are most likely to emerge from new confi gurations of social capital, comprising relations of trust embo-died in new social organizations, and new horizon-tal and vertical partnerships between institutions, and human capital comprising leadership, ingenu-ity, management skills and capacity to innovate. Agricultural systems with high levels of social and human assets are more able to innovate in the face of uncertainty (Bunch & Lopez, 1999; Chambers et al., 1989; Olsson & Folke, 2001; Pretty & Ward, 2001; Uphoff, 1998). This suggests that there are likely to be many pathways towards agricultural sustainability, and further implies that no single confi guration of technologies, inputs and ecological management is more likely to be widely applicable than another. Agricultural sustainability implies the need to fi t these factors to the specifi c circum-stances of different agricultural systems.

A common, though erroneous, assumption about agricultural sustainability is that it implies a net

40 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

reduction in input use, so making such systems essentially extensive (they require more land to produce the same amount of food). Recent empiri-cal evidence shows that successful agricultural sus-tainability initiatives and projects arise from shifts in the factors of agricultural production (e.g. from fertilizers to nitrogen-fi xing legumes; from pesti-cides to emphasis on natural predation; from ploughing to zero- or minimum-tillage) (Gallagher et al., 2005; Pretty et al., 2006). A better concept than extensive is one that centres on making better use of existing resources (e.g. land, water, biodiver-sity) and technologies (Buttel, 2003; Conway & Pretty, 1991; Pretty et al., 2000; Tegtmeier & Duffy, 2004). The critical question focuses on the ‘type of intensifi cation’ – more intensive use of natural, social and human capital assets, combined with the best available technologies and inputs (best genotypes and best ecological management) that minimize or eliminate harm to the environ-ment, can be termed ‘sustainable intensifi cation’. Although there has been considerable progress towards understanding agricultural sustainability in both developing countries (Pretty et al., 2006) and industrialized countries (Firbank et al., 2005; Leake, 2000; Moss, 2007), there has been relatively little published evidence of this in the private food production, manufacturing and retail sectors.

Unilever and development of sustainability indicators

In 1997, Unilever began an internal process of adopting agricultural sustainability principles for its various food businesses. It had earlier supported the emergence of the Marine Stewardship Council as a system of certifying the sourcing of fi sh, and then began to focus on the long-term sustainability of its agricultural supply chains. Unilever relies heavily on natural raw materials for both food and home care products, notably vegetable oils, vegeta-bles, fi eld tomatoes, tea and fi sh. In 2006, the com-pany’s share of raw material crops as a percentage of world volume was 12% for peas (frozen), 12% for black tea, 28% for spinach (frozen), 7% for tomatoes and 4% for palm oil. Unilever is both a buyer of products on open markets, and directly involved in agricultural production, either on farms or plantations or through contract growers.

For business success, the company needed a sustained supply of these materials. However, the idea of sustainability implied a widening of its obligations beyond shareholders, customers and employees and opened new opportunities to make positive contributions to environments and com-munities across a range of countries and agroeco-logical zones.

The company thus came to recognize that it could not work alone on agricultural sustainability, and sought to adopt a partnership approach to support progress towards agricultural sustain-ability as well as the development of technologies and practices to transform production systems. An internal Sustainable Agriculture Steering Group was established in 1998 to coordinate activities, and working groups for each crop developed prin-ciples and practices for promoting sustainability. Review workshops with appropriate scientifi c advice were then held in the Netherlands, UK, Germany, Malaysia, Italy, Greece and the United States. An independent Sustainable Agriculture Advisory Board with members drawn from the international agricultural research, policy and NGO communities was appointed, and this held scientifi c and policy forums to explore how sus-tainability indictors could be addressed in the con-text of rapidly changing international and national policy environments.

The approach taken was fi rst to develop a sus-tainability audit, based around a common set of indicators, and then to support pilot projects for a select number of crops, with the aim of adapting parameters for each crop, establishing baseline measures, developing recommendations to increase agricultural sustainability, and holding fi eld trials to test these new practices and technologies. The fi ve crops, their locations for cultivation, and the project start dates are shown in Table 1.

It has long been argued that the complexity and multi-faceted nature of agricultural sustainability cannot be resolved to a single metric (ACRE, 2007; Dobson, 1999; Firbank et al., 2005; Naess, 1992). As a result, a wide range of traditions and approaches have been developed for the assessment of agriculture’s impact on the environmental. Some of these are applicable to the farm/fi eld scale while others require assessments based on land-scape-scale impacts of agriculture as a whole, and so are not sensitive to small-scale changes in

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 41

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

management. These approaches include (but are not limited to):

● energy accounting that compares inputs and out-puts according to energy content (Cormack & Metcalfe, 2000; Leach, 1976; Leake, 2000);

● environmental economics that puts a monetary cost on agriculture’s positive and negative exter-nalities (Daily, 1997; Norse et al., 2001: Pretty et al., 2000, 2006; Tegtmeier & Duffy, 2004);

● carbon accounting that measures the sinks and sources of greenhouse gases in agricultural systems (Pretty et al., 2002, Smith et al., 2000; Smith & Trines, 2007; Swingland, 2003);

● quantifi ed risk assessments and environmental management systems (Lewis et al., 1997);

● environmental harm indices and multi-criteria mapping (DETR, 1998; Stirling & Mayer, 2000);

● environmental audits for the development of management systems (such as BS14001), and standards-based approaches (RCEP, 1998);

● key species approaches (Sommerville & Walker, 1990);

● sustainability indicators as developed by Defra and the OECD (MAFF, 2000; OECD, 1998).

This wide variety of approaches suggests that no single system of assessment is likely to be able to give a comprehensive and uncontested view of the

status of any one agricultural system. Here, however, the company was concerned with the development of a system of agricultural assessment that would be practical, effi cient and effective over reasonably short time scales so that changes in company policies and practice could be made towards greater sustainability in its widest sense. A further consideration was that different individu-als and organizations, including farmers, landowners, non-government organizations, other companies, and interests within Unilever, would need to be able to use the assessment as a reference point for assess-ing existing impacts, and then track changes (hope-fully improvements) over time. For these reasons, an indicator-based system was developed. This was derived from a set of four sustainability principles adopted by the company (Table 2).

The Sustainability Indicator structure developed uses 10 clusters of indicators (later revised to include ‘animal welfare’ when the scope was increased from purely crop-based projects). These were selected on the basis that indicators must be easily measurable, and so not costly; relatively non-contestable, and so convincing to internal and external stakeholders; prone to management action; and lead to value creation for farmers, rural communities and busi-nesses. These 10 indicator clusters were applied to every crop regardless of agricultural system, so that

Table 1 Crops selected for sustainable agriculture initiative

Crop Location Start date for projects

Peas UK 1997

Tea Kenya 1999

Tea India 2000

Tea Tanzania 2001

Spinach Germany 1999

Spinach Italy 2000

Tomatoes Brazil 2000

Tomatoes Australia 2000

Tomatoes USA (California) 2001

Palm oil Malaysia 1999

Palm oil Ghana 2001

42 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

comparisons could be made. These covered biologi-cal measures (soils, nutrients, pest management, biodiversity), business measures (value chain), physical measures (energy, water) and social and economic measures (social and human capital, local economy). Several measurable parameters for each indicator cluster were then identifi ed for each crop, and systems put in place to assess and measure them within Lead Agriculture Programmes. These were developed to address a variety of inputs and resources, and impacts on the natural resources, social capital and rural economy. Although some parameters were applicable to all crops and locations, others were site-specifi c or crop-specifi c. This was because the state of natural resources and the goods and ser-vices used by agricultural systems differs according to local climatic and biophysical conditions. A key measure for biodiversity in a temperate agricultural system may, for example, differ greatly from one in the tropics. In this way, the ten indicator clusters were intended to remain consistent across the crop systems and countries, but the parameters chosen would be site specifi c (Table 3).

It was initially envisaged that the audit would be completed by measuring each selected parameter, and then converting this to a score on a scale of 1–10. Parameters would be equally weighted within each indicator cluster, and so a cluster with three parameters would have each parameter contribut-ing one-third to the total cluster score. All 10 (and later 11) indicator cluster scores could then be summed up to give an aggregate score or index for the agricultural system. The primary advantage of this was that it would allow a wide range of incommensurable data and parameters relating to sustainability and farm performance to be assessed and compared on a single metric. This could have

been useful for measuring progress over time, but not necessarily permit comparisons with agricul-tural systems not using the audit. This system would also allow for the score to be disaggregated and stakeholders interested in performance in any one area to monitor change over time.

Several disadvantages were soon obvious. The system required expert judgement on the essential parameters and score-conversion relationships, and so was only as good as the existing information base. For example, although there was general agreement that a score of ‘10’ for measured soil loss would be appropriate for a situation where soil was being generated at the same rate as lost, the appro-priate value was rarely available and so a score of 10 was assigned to an unachievable ‘no soil loss’. There was even less agreement on the value to assign to ‘0’ for this parameter, and whether a common value could be used for both temperate and tropical (where rainfall has higher erosivity) crops. It also treated each indicator cluster and parameter with equal weighting. Although equal weighting was eventually agreed, because weightings added an unnecessary level of complexity to a system in which implicit weighting is already a part (through the choices of parameters and indicators), it did mean that param-eters with little overall impact but high manageabil-ity (e.g. the source of soil for nurseries) were treated in the same way as parameters with high impact.

Although comparisons were initially made between Unilever farms and agricultural systems where the audit had been applied, it was soon found that there was little value in ascribing a single score to a whole agricultural system. The main advantages of the audit was thus not so much in the emergence of a fi nal sustainability index, but in the development of increased knowledge and

Table 2 Sustainability principles adopted by Unilever

Unilever believes that sustainable agriculture should support the following principles:

(1) It should produce crops with high yield and nutritional quality to meet existing and future needs, while keeping resource inputs as low as possible.

(2) It must ensure that any adverse effects on soil fertility, water and air quality, and biodiversity from agricultural activities are minimized, and positive contributions are made where possible.

(3) It should optimize the use of renewable resources while minimizing the use of non-renewable resources.

(4) It should enable local communities to protect and improve their well-being and environment.

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 43

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

Table 3 Indicator clusters and examples of measurable parameters for the sustainability audit

Indicator clusters and rationale Typical parameters to be measured

(1) Soil fertility and health

Soil is fundamental to agricultural systems, and a rich soil ecosystem contributes to crop and livestock performance. Sustainable practices can improve benefi cial components of the soil’s ecosystem.

Number of benefi cial organisms (e.g. earthworms per m2); number of predatory mites; number of benefi cial microorganisms; soil organic matter content (measure of healthy soil structure)

(2) Soil loss

Soil eroded by water and wind can lose both structure and organic matter, diminishing the assets of an agricultural system. Sustainable practices can reduce soil erosion.

Soil cover index (proportion of year soil is covered with crop); soil erosion (loss of top soil e.g. t ha�1 yr�1).

(3) Nutrients

Crops and livestock need a balance of nutrients. Some of these can be fi xed locally (e.g. nitrogen), and some must be imported. Nutrients are lost through cropping, erosion and gaseous emissions. Sustainable practices can both enhance local production of nutrients and reduce losses.

Amount of inorganic nitrogen (N), phosphorus (P) and potassium (K) applied (per ha or per tonne of product); proportion of N fi xed on site/imported; balance of outputs vs inputs of N/P/K over crop rotations; emissions of N-compounds to air.

(4) Pest management

When pesticides are applied to crops or livestock, a small but signifi cant proportion can escape to water and air, or accumulate in foods, thus potentially affecting human health and ecosystems. Sustainable agriculture practices can substitute natural controls for some pesticides, so reducing dependence on externally introduced substances. The aim is to develop Integrated Pest Management (IPM) strategies for all crops.

Level achieved of bringing crop under IPM (checklist approach); amount of pesticides (kg active ingredient) applied (per ha or per tonne of product); type of pesticide applied (toxicity assessed by profi ling, positive listing and/or use of weighting factors).

(5) Biodiversity

Agriculture has shaped most ecosystems in the world, and biodiversity can be improved or reduced by agricultural practices. Some biodiversity is highly benefi cial for agriculture. Sustainable practices can improve biodiversity – by ‘greening the middle’ of fi elds as well as ‘greening the edge’.

Level of biodiversity on site: number of species (e.g. birds, butterfl ies); farm landscape; habitat for natural predator systems (e.g. hedgerows, ponds, noncropped areas); level of biodiversity off-site – both within farm and in the wider landscape.

(6) Value chain

Value chain is the term for the total value-adding activities which lead to putting a product on the market. For food products, farm economics is an integral part of the value chain. Farmers need to know what infl uences the economics of their farm and what non-economic value they produce. Sustainable practices should be able to maintain or improve farm economics and add to environmental goods and services.

Total value of produce per ha; farm income trends; conformance to quality specifi cations (e.g. nutritional value, including minerals, pesticide residues, foreign bodies); ratio of solid waste re-used/recycled over solid waste disposed to landfi ll; marginal costs for various crops and various fi elds/plots; fi nancial risk management and solvency; ecosystem goods and service values created.

(Continued)

44 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

understanding of agricultural, environmental and human interactions and processes that emerge during the discussion and assessment of the indica-tors. The process of its use was thus more impor-tant than any scores that emerged. However, value would be created over time as measured parameters resulted in observed trends towards or away from more sustainability. This was particularly important where actions were identifi ed and implemented,

and then their effects monitored and assessed. This system was fi rst tested on the Unilever experimen-tal farm at Colworth, UK.

The Colworth experimental farm

The Colworth Farm Project, established in 1999, was one of the fi rst of the sustainable agriculture

Table 3 Continued

Indicator clusters and rationale Typical parameters to be measured

(7) Energy

Although the energy of sunlight is a fundamental input to agriculture, the energy balance of agricultural systems depends on the additional energy supplied from non-renewable sources to power machinery. Sustainable practices can improve the energy balance and ensure that it remains positive (more energy comes out than goes in).

Energy balance (total energy input/total energy output, including transport where relevant); ratio of renewable over non-renewable energy inputs; emissions to air (greenhouse and pollutant gases).

(8) Water

Some agricultural systems make use of water for irrigation, some pollute or contaminate ground or surface water with pesticides, nutrients or soil. Sustainable practices can make targeted use of water as an input, and reduce losses.

Amount of water used per ha or tonne of product for irrigation; leaching and runoff of pesticides to surface and ground water; leaching and runoff of N/P/K (nutrients) to surface and ground water.

(9) Social and human capital

The challenge of using natural resources sustainably is fundamentally a social one. It requires collective action, the sharing of new knowledge and continuous innovation. Sustainable agriculture practices can improve both social and human capital, as well as reply on it for improvements.

Group dynamics and organizational density (farmer groups); (rural) community awareness of relevance and benefi ts of sustainable practices; connectivity to society and institutions at large; rate of agricultural innovation.

(10) Local economy

Agricultural inputs (goods, labour, services) can be sourced from many places, but when they come from the local economy, expenditure helps to sustain local businesses and livelihoods. Sustainable agriculture practices can help to make the best use of local and available resources in order to increase overall effi ciency.

Amount of money/profi t reinvested locally; percentage of goods, labour and services sourced locally; employment level in local community.

(11) Animal welfare

Animal husbandry systems are becoming ever more specialized and further removed from the natural habitat the (farm) animal came from. Treatment of animals in these artifi cial environments is a major ethical concern. Care must be taken that the animals can live in harmony with their environment.

Feeding; housing; treatment of diseases; watering; freedom from abuse.

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 45

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

initiatives developed by Unilever (called Lead Agriculture Programmes). The farm was used to assess new agricultural methods and practices in a real commercial situation, but with a relatively risk-free environment, that is, with no commercial pressures. The research outcomes were to be rele-vant to the range of crops used by the company, including oilseed rape, linseed, cereals, vining peas and mustard. The Colworth Farm (map ref. SP 969617) is a 500-hectare estate, comprising 400 ha of arable land of predominantly heavy clay (Hanslope series; Gleysol in the FAO classifi ca-tion) and 100 ha of semi-ancient and natural woodland in Bedfordshire, England. Sixty hectares consisting of eight fi elds were dedicated to the project (Figure 1a, b).

The aim of the project was to provide represen-tative information over a sequence of years on two crops (peas and oilseed rape) within a typical cereal-dominated rotation. This would allow, at the whole farm level, an assessment of the practi-cal (and impractical) implications of alternative practices that could then be used to infl uence key stakeholders, internally and externally, on criti-cal issues required for the sustainable production of raw materials. The objectives of the project were to:

(1) measure the long-term impact of conventional farm practice compared to experimental and potentially more sustainable alternatives;

(2) assess and improve biodiversity in a modern conventional farm context;

(3) understand the fi nancial consequences of these experimental agricultural practices.

The multi-year project was designed to provide more than a snapshot view of ecological processes in arable farming. Short-term assessments of organisms responding to farm management and weather can be misleading, and the project was designed to provide data from several years of observations and of various stages of crop rota-tions, and on a variety of fi elds. This allowed an assessment of both the magnitude and speed of the response by organisms and processes to changed management practices. It was recognized from the beginning that the knowledge of response-times was critical for planning future sustainability sce-narios. The project also set out to elicit the response of animals and plants to changing management

practices by deliberately adopting drastic changes in management (e.g. reductions in input levels). These procedures resulted in potential risk to crop production, which could not be tested in a com-mercial environment. While some of the scenarios tested might not be practical for commercial farms,

Figure 1 (a) Location of Colworth farm; and (b) layout of experimental plots

46 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

they were designed to challenge and provide an insight into practical alternatives to conventional thinking and practice.

Neither the conventional nor the experimental practices were static. They evolved throughout the project. This allowed the project to refl ect current and emergent thinking and learning. The project investigated the impacts of the fi ve farm manage-ment scenarios on the ten indicators, together with an assessment of the wider fi nancial and economic outcomes:

(1) comparison of spring versus winter cropping;(2) effect of reduced nitrogen fertilizers;(3) effect of reduced pesticide (insecticide, herbi-

cide and fungicide) applications;(4) infl uence of mixed rotation and cover crops;(5) fi eld margin management;(6) fi nancial analysis.

Six of the eight fi elds (Figure 1b) were split into quarters, with two sets of paired treatments in each: for example, conventional versus reduced pesticides, and conventional versus reduced nitro-gen fertilizer. The other two fi elds were split in two, with all six experimental practices compared with all conventional practices. The effects of ploughing and changes in cultivation were also investigated. However, because the frequency of ploughing on the Colworth farm had already been reduced and non-inversion techniques adopted, conditions were considered equivalent to mini-mum tillage on similar soil types. Here we summa-rize the changes in selected indicators for each of the fi ve novel management practices tested on the pilot farm. A brief analysis of the agronomic conclusions is given for each. There is not the space to report on all ten indicators for each of the fi ve management practices tested, and so a repre-sentative sample of trends are chosen and discussed in this paper.

Comparison of spring and winter cropping

Indicator 3: Nutrient (N) leachingIn spring cropping, land can either be ploughed in autumn and left uncropped or left undisturbed until spring. Ploughing incorporates crop residues and aerates the soil, thereby enhancing the micro-biological processes by which N in the soil organic matter is mineralized into plant-available forms such as nitrate. During the fi rst three years of the project, spring-cropped land was left undisturbed in both the experimental and conventional treat-ments in winter prior to ploughing and sowing. Nitrogen fertilizer rates were reduced by over 50% in the experimental treatment yet the effect on nitrate leaching loss was minimal (mean nitrate concentration in drainage water 2000 to 2003; conventional 80, experimental 83 mg l�1). In 2003, autumn ploughing was introduced into the conven-tional treatment. Table 4 shows that when autumn ploughing was introduced, leaching losses were higher than when the land was left undisturbed. This suggests that autumn cultivation released nitrate which, in the absence of a growing crop, was leached during the winter months.

For winter-sown crops, a major contribution to nitrate leaching is the release of nitrate that follows autumn cultivations. Losses can be limited by ensuring that crops are established early so that released nitrate is captured rather than leached during the following winter. Nitrogen uptake by an autumn-sown crop can be 5–50kgNha�1 depending on crop type and density (typically �20 kg N ha�1). A loss of 17 kg N ha�1 leached from the soil during an average winter in eastern England, with 150 mm of drainage, would cause average nitrate concen-trations in drainage water to breach the 50 mg nitrate l�1 EC Directive limit. However, although leaving the land undisturbed in autumn prior to

Table 4 Effect of shift to autumn ploughing on the average concentration of nitrate in water leaching from the fi eld, 2003–2004

Before autumn ploughing, 2000–2003 (mg NO3 1−1)

After autumn ploughing, 2003–2004 (mg NO3 1−1)

Standard rates of N fertilizer applications

68.3 157.5

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 47

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

spring cropping would have reduced N losses, the heavy soil texture at Colworth presented many practical problems when trying to achieve a suit-able seedbed. Therefore the advantages in terms of reduced N loss from not ploughing in autumn have to be balanced against potential damage to soil structure associated with early spring cultivations on heavy wet soils. On these soils, compaction, cap-ping and soil erosion may cause greater environ-mental damage damage in terms of sediment and phosphate loss (which were not measured) than the extra nitrogen released from autumn ploughing.

Indicator 5: BiodiversitySpring-sown wheat crops supported a higher diver-sity of low-impact weeds such as speedwell (Veronica spp.), groundsel (Senecio vulgaris L.) and knotgrass (Polygonum vulgare L.) that are important as bird food. Winter-sown crops supported more pernicious weeds such as cleavers (Galium aparine L.) and grasses, in particular blackgrass. As a consequence, the spring-sown plots only required a single appli-cation of herbicide compared with several applica-tions for the winter-sown crop. Crop rotation played a part, as wheat crops following oil-seed rape tended to support high densities of ryegrass (Lolium perenne), chickweed (Stellaria media L.) and cleavers. Nonetheless, levels of pernicious weeds fell in both sowing regimes (see Figure 2a).

Differences between spring and winter-sown plots were especially pronounced in the plots where pesticides were reduced. In winter-sown plots, reduced pesticide applications led to greater popu-lations of pernicious weeds and a 47% reduction in yield (mean 4.5 t ha�1, compared with 8.5 t ha�1 in conventional winter wheat). In these plots mecha-nical weeding failed to reduce the impact of perni-cious weeds. On the other hand, pernicious weeds were largely absent in spring crops, where the only treatment required was an application of non- selective herbicide before drilling, and yield was only reduced by 18% (average 5.5 t ha�1 with reduced pesticides compared with 6.7 t ha�1 in conventional plots).

In some years, winter-sown crops supported greater populations of invertebrates than spring-sown, especially spiders and ground beetles. Ground-hunting wolf-spiders (Pardosa sp.) and two early-summer beetles (Nebria brevicollis and Agonum dorsale) were notably abundant. The abundance of

invertebrates may have been due to the presence of winter crop cover and the associated microclimate afforded by the grass cover in these plots. Differences in invertebrate populations in spring and winter-sown plots were more pronounced in the absence of pesticides, as grass cover increased in the latter. But there were no differences in inver-tebrate abundance under wheat crops following set-aside. This may have been a refl ection of the

Figure 2 (a) Levels of pernicious weeds in spring and winter sown wheat. (b) Density of ground beetles 25m into the crop. (c) Density of ground beetles 25m into the crop

48 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

relative cleanliness of the winter-sown plots fol-lowing set-aside, indicating that weed cover is important to invertebrates as well as crop cover. There were indications that greater numbers of invertebrates occurred further into the crop after 4 years of spring cropping, perhaps refl ecting the open structure of the crop earlier in the season, or the different weed fl ora (Figure 2b, c). Due to the relatively late development of spring crops, sky-larks continued to breed and forage during June and July, by which time winter-sown crops had become too tall and dense to allow access to the birds. Spring cereals, set-aside and peas allow sky-larks the opportunity to fulfi ll breeding potential, unlike winter-sown crops.

Indicator 1: Soil healthSeasonal timing of cultivation and the regularity and depth of ploughing are important for soil health. Most agriculturally relevant earthworm species are inactive during winter and high summer, as they seek to avoid freezing or desiccation by creating deep burrows, aestivating and/or producing cocoons. The least harmful time to plough is during winter or high summer (i.e. when soils are below 4°C or very dry). Therefore, spring cropping, where the soil is left undisturbed after autumn harvest and ploughed in spring, is likely to be harmful to some soil invertebrates. Winter cropping, where the soil is only disturbed in autumn (particularly when ploughing takes place soon after harvest) may be the least harmful (Figure 3).

Conclusions on agronomyPloughing and non-inversion tillage were equally effective in creating a good autumn seedbed for winter crops. Achieving good results for spring crops, however, still required a primary cultivation by ploughing in autumn, and subsequent cultiva-tion on the weathered surface in spring. Data from Colworth indicates that there is little advantage in ploughing on heavy land in spring. Over-wintered ploughing does provide the opportunity to estab-lish spring crops in good seedbeds after limited secondary cultivations. Spring cropping also allows farmers to spread labour requirements over the year, and increase market opportunities for farm busi-nesses. In addition, many of the newly emerging markets for farmers (e.g. energy, fuel, fi bre and pharmaceutical crops) are likely to stimulate the

inclusion of spring crops in the rotation (these novel crops were not included in the project rotation). At Colworth, spring wheat yielded better under the low pesticide regime than winter wheat (which normally exceeds yields in conventional spring wheat by 25%) and gave a higher gross margin by accessing premium markets.

Figure 3 (a) Total earthworm abundance, Colworth, 2003. (b) Adult earthworm biomass, Colworth, 2003. (c) Juvenile earthworm biomass, Colworth, 2003

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 49

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

Effect of reduced nitrogen fertilizer

Indicator 3: Nutrient (N) leachingNitrate losses by leaching were measured in selected fi elds, where either conventional N fertilizer rates were applied (as defi ned by good agricultural prac-tice), or where rates were substantially reduced (2000–2003: 33%; 2004: 66% of conventional rates). Autumn Soil Mineral Nitrogen (SMN) levels, as an indicator of potential winter leaching risk, were measured in all fi elds across all seasons. The purpose of these treatments was to study the infl uence of N rate on leaching losses, and its interaction with weed growth, canopy density and in-fi eld biodiversity. When considering the impact of N fertilizer rates on N losses, it is important to separate applications above the economic optimum rate for the crop from those below the optimum. In this study we only considered rates at or below the economic optimum. In reality, determining the economic optimum with precision is impossible as growth is also affected by other factors, such as weather, pests and diseases, which infl uence the crops after fertilizer has been applied. The inten-tion of reducing N application rates was to reduce the risk of overfertilisation and evaluate nitrate leaching losses.

The effect of N rate varied between different fi elds. Figure 4a and b show data from sections of the same fi eld. Figure 4a shows the effect of N rate on a winter cropping regime, and Figure 4b the effect on a spring cropping regime. The differences over three years between winter and spring crop-ping are shown in Table 5. Reduced N fertilizer had a positive effect on reducing nitrate concentra-tions and thus leaching losses under comparison a

but not b. The considerable variation across and within years is shown in the fi gures, implying that other management and fi eld factors were also affecting the release of leachable N. Predicting the effect of N rate on leaching losses on a fi eld-by-fi eld basis is therefore diffi cult. However, it is likely to be more important to assess leaching losses on a catchment scale, as environmental outcomes such as eutrophication operate at this level. Autumn SMN data are now, therefore, used to measure the

Table 5 Nitrate leaching losses from winter and spring cropping under different fertilizer regimes

Winter crop (mg NO3 1−1) Spring crop (NO3 1−1) Mean

Standard nitrogen fertilizer regime 110.5 93.2 102.7

Reduced nitrogen fertilizer regime 71.7 87.5 78.9

Mean 90.6 90.2 90.4

Note: winter/standard n � 52; winter/reduced n � 43; spring/standard n � 55; spring/reduced n � 46.

Figure 4 (a) Winter cropping: nitrate leaching under two N fertilizer regimes. (b) Spring cropping: nitrate leaching under two N fertilizer regimes

50 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

leaching potential in the catchment at Colworth. This is less precise than measuring actual leaching loss, but is a good indicator of leachable N in the soil at the start of winter and how much N is likely to be lost during winter drainage.

Indicator 5: BiodiversityThere were lower weed burdens in reduced N fer-tilizer plots in 2003, with fat hen (Chenopodium album) and cleavers abundant under high N despite the higher herbicide inputs (Figure 5). In other years, though, no differences were observed between the two treatments. In the absence of herbicides, both plots became very weedy, and, although the density and diversity of weeds were initially greater in con-ventional N fertilizer plots, reduced N fertilizer plots eventually became more weedy as the rotation pro-gressed. The most abundant weeds were pernicious blackgrass and cleavers, populations of which damaged crops after three years of the experiment. Populations of these weeds were suppressed by minimum tillage of the soil before drilling.

There were no clear responses of invertebrates to low fertilizer treatments, except where smaller her-bicide applications produced a greater diversity of weeds. Here, there were strong responses and larger invertebrate populations (see later). There was no clear response to reduced N fertilizer rates although the combined effect of low nutrients and low her-bicide inputs supported higher densities of inverte-brates. This combined effect was probably a response to higher weed diversity rather than abundance.

Conclusions on agronomyConventional farm practice at Colworth relies on synthetic nutrient inputs to improve the quantity and often the quality of the crops. As well as infl u-encing crop development, these inputs inadvertently increase the competitive ability of certain weeds, and foliar diseases are more prevalent due to increased gross leaf area. As the Colworth farm falls within a Nitrate Vulnerable Zone, timing and quantity of fertilizer use are subject to strict guidelines. The project reduced application rates of N fertilizer beyond these guidelines. Samples were taken from each fi eld in both autumn and spring, to provide an accurate picture of available SMN. This allowed subsequent applications of N fertilizer to be adjusted to match crop needs. SMN monitoring is also a useful tool for managing N inputs, and is now used routinely on the Colworth farm.

Effect of reduced pesticide applications

Indicator 4: Pesticide leachingBand spraying was used in one fi eld to investigate whether this would reduce pesticide leaching. In spring 2004, herbicide was applied at the standard concentration in the conventional treatment using a conventional sprayer. On the same day herbicide was applied at the same concentration to a differ-ent section of the same fi eld using a band sprayer in conjunction with mechanical weeding. This treat-ment received half the volume of herbicide per hectare than applied in the conventional treatment plot. During the following 13-day period 50 mm of rain were recorded at the site. Mean concentrations of herbicide were measured in the soil water close to fi eld drains on three occasions, and these indi-cate that there were lower levels following band spraying (Figure 6). Although the amount of active ingredient applied in the band spraying treatment was half that in the conventional section, the concentrations at drain depth were substantially less than half. This suggests that targeting the spray on to the crop and weed leaf surfaces, but away from the soil, may provide benefi ts in addition to reductions in active ingredient used.

Indicator 5: BiodiversityWeed build-up occurred in reduced pesticide plots. The use of mechanical weeding in 2002–2003 had only a limited impact on weed populations, and

Figure 5 Pernicious weeds under different fertilizer regimes (2003)

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 51

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

would need to be carried out much more frequently to have a pronounced effect (crops were weeded once in autumn and once in early spring). There were large populations of low-impact weeds that

are important as food to birds in all crops with reduced inputs, especially wheat. Small fl owering annual weeds such as fi eld speedwell (Veronica per-sica Poiret), fi eld forget-me-not (Myosotis arvensis L. Hill) and cut-leaved crane’s-bill (Geranium dissectum L.) provided diversity and cover for invertebrates close to the ground in most crops.

Predatory ground invertebrates were more abundant in wheat plots with reduced herbicides (Figure 7a). Consistent reductions of pesticides over the course of the rotation produced greater densities of invertebrates further into the crop (Figure 7b, c). It is possible that the winter ground cover afforded by these crops was a favourable habitat. The density and abundance of these inver-tebrate predators was strongly correlated with the weed diversity in the crop, not weed abundance (Figure 7d). Butterfl ies (common blue (Polyommatus icarus), skipper (Hesperiidae) and marbled white (Melanargia galathea), showed similar responses,

Figure 7 (a) Density of beetles in fi elds under different pesticide management regimes. (b) Effect of pesticide regime on Poecilus cupreus (ground beetle). (c) Effect of pesticide regime on Philonthus cognatus (rove beetle). (d) Weed density and ground beetle numbers

Figure 6 Effect of different application regimes on herbicide losses

52 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

with greater diversity in fi elds with more diverse weed populations.). This suggests that it is the presence of the additional low-impact weeds in the crop, not the abundance of pernicious weeds, which increases predatory invertebrate density – fi elds choked with blackgrass and cleavers did not promote greater invertebrate activity. Promoting populations of low-impact weeds in crops may thus promote large populations of benefi cial arthro-pods, enhancing both the natural control of crop pests and the availability of invertebrate prey for farmland birds.

One of the factors related to higher bird popu-lations was the tolerance of weedy fallows. These were cereal stubbles on which herbicide applica-tions were delayed from April until June, available to birds particularly in 2002 and 2004 (Figure 8a). By contrast conventional early (April) weed-control in cereal stubbles led to the lowest densities of birds of any fi eld type including winter wheat.

Data also show that some bird species forage extensively within crops where generally, densities of bird species were higher in low pesticide plots (Figure 8b). This relationship was especially sig-nifi cant for skylark and linnet whose abundance was related to invertebrate and weed populations respectively. There was consistent association with low-impact weed species (geranium, speedwell, fat hen, groundsel Sinapis, Myosostis species). By contrast, an abundance of pernicious weeds in some crops was detrimental to birds. For skylarks, dense patches of blackgrass and wild oats pre-vented access to the ground, and shortened the breeding season in 2003 compared to earlier years, thus preventing skylarks from producing late summer broods (Figure 9). Overall, a balanced and effective in-fi eld management of pesticides for retaining both yield and biodiversity was diffi -cult to achieve and it was far easier to include the low herbicide benefi ts as weedy fallows within a mixed rotation.

Conclusions on agronomyThe project showed that reducing herbicides cut variable costs by approximately 50% (saving £125 ha�1), but caused yield losses of around 65% (equivalent to £420 ha�1 gross margin). The cost/benefi t differential would therefore only be viable in a depressed market where the cost of yield losses was relatively low. In addition, pernicious weeds accumulated causing yield reductions in subse-quent crops. Data on accumulated impacts of low herbicide rates over four years showed that both crops and biodiversity were adversely affected.

Figure 8 (a) Farmland birds in different fi eld types over 5 years. (b) Bird densities in crops, by pesticide treatment

Figure 9 Effect of high weed densities in 2003 on skylark breeding density

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 53

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

Generating a high diversity of low-impact weeds in crops is possible using selective, although more expensive, herbicides. However, intermediate levels of pesticide use were not tested and may provide further opportunities. Benefi ts may also come from outcropped areas or the physical manipulation of crops to create structural variation (not investigated in this project) or from mixed cropping (see below). Since pressure to protect water resources from dif-fuse pollution sources may reduce the availability of some agrochemical products in the future, crops will have to be managed using a range of manage-ment methods.

Infl uence of mixed rotation and cover crops

Indicator 3: Nutrient (N) leachingThe diffuse pollution of rivers and groundwater by agriculture is a result of leaching losses accumu-lated over a wide geographical area or catchment (Moss, 2007). One of the factors infl uencing the scale of pollution is the mix of crops grown, as different crops are associated with different diffuse pollution risks. Although nitrate leaching from different crops was not measured in the Colworth project, Figure 10a shows data from the Unilever pea project that

Figure 10 (a) Nitrate leaching under different crops in pea rotation (sample size in parentheses). (b) The effect of cultivation timing and cover crops on nitrate concentration in drainage water (mg NO3 l) and total N leached (kg N ha�1) after vining peas. (c) Autumn soil mineral N concentrations, 2000–2004 (kg N ha�1). (d) Effect of soil organic matter under peas on nitrate leaching

54 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

measured nitrate leaching loss from a range of crops at a number of farms over the 5-year period. The results show that the three break crops (especially peas and potatoes) resulted in highest N losses. There are many factors that give rise to these differences: higher leaching losses occur where substantial N in crop residue is returned (as with peas), where the land is cultivated early, and where crop residues are incorporated before winter.

Analysis of all fi elds at Colworth also showed that SMN levels were 11% lower where reduced rates of N had been applied (Figure 10c). At low rates of N application the crop takes up most of the N present in the soil together with that applied in fertilizer. At higher application rates the effi ciency of uptake declines and at rates substantially above the economic optimum, virtually all the excess N is left unused in the soil at the end of the season. These data indicate that within the Colworth catch-ment, lower fertilizer rates reduced nitrate leaching potential by 11%.

A further important factor in nitrate leaching is the relationship with soil organic matter (SOM). In a study funded by Birds Eye Wall’s, nitrate leaching was measured in 10 fi elds, of differing soils types, after a vining pea crop. No N fertilizers were applied and measurements were taken during the same time interval. There was a strong relationship between soil organic matter content and the nitrate concen-trations measured in drainage water (Figure 10d).

This appears to challenge the common under-standing that high SOM levels are always benefi -cial, and this relationship should be taken into account when planning fertilizer strategies. It is also important not to ignore the potential effects of N rate reductions on crop growth. Where N supply is restricted, crops are vulnerable to poor establish-ment through nutrient defi ciency. This may cause other environmental problems, such as the need for additional pesticide applications to limit weed competition. In the event of crop failure, much of the N fertilizer applied may be leached the follow-ing winter.

The mixed rotation in the Colworth project may therefore have infl uenced nitrate leaching losses. Cover crops and cultivation timing are two tech-niques that can be used to reduce nitrate losses after some break crops. The vining pea study funded by Birds Eye Walls showed that delaying cultivation until just before the following wheat

crop was sown, rather than ploughing straight after the harvest, reduced the amount of N leaching loss by 26%. When a cover crop was established after the pea harvest, there was a further 44% reduction in losses (70% in total) (Figure 10b).

Indicator 5: BiodiversityThis project demonstrated a strong and surprisingly rapid response by a wide range of bird species to changed management practices. Figure 11a shows the annual change in breeding birds belonging to the Farmland Bird Indicator, where the rate of increase was faster than for woodland species at the same site. For example, grey partridge increased from one to three pairs, skylarks from 10 to 20 pairs, yellow wagtails began to breed in the fourth year, by which time even lapwings, which had not been seen as a breeding species on the site for many years, had attempted to breed. Success with all of these species was a result of improved conditions within the cropped environment. Over 50% of the foraging trips made by yellowhammers, a boundary species, were into crops, where a large proportion of their summer diet was obtained.

Figure 11 (a) Changes in breeding birds at Colworth. (b) Bird densities by fi eld content

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 55

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

Interestingly, some 70% of the farmland bird increase occurred within two years of the experi-ment starting, suggesting that some species are able to recover rapidly under appropriate circumstances (though we cannot be certain that they did not move from elsewhere). Seed-eating bird species, such as fi nches and buntings, were quicker to respond than insectivorous species, suggesting that seed resources in winter became more quickly available than summer invertebrate resources that are used by virtually all bird species. Structural components of the landscape (boundaries, crops and margins) also infl uenced bird densities and changes in bird popu-lations. The sustained increase in bird population size at Colworth was a consequence of two compo-nents: large scale habitat availability resulting from greater complexity in the crop mosaic, and changes to habitat quality due to herbicide restrictions on crops and fallows (Henderson et al., 2006). Greater crop complexity had three consequences:

(1) Habitat availability: preferred crops, such as oilseed rape, weedy fallows and peas were espe-cially important in providing food and breeding sites for birds, at up to fi ve times (in rape) the density of winter wheat (see Figure 11b). Oilseed rape (whitethroat and buntings), weedy set-aside (skylark and seed-eating bird species) and to a lesser extent peas (skylarks and thrushes) provided complementary opportunities for birds to forage and/or breed.

(2) Landscape variability: a mixed crop rotation provided options and opportunities for birds to forage and breed throughout the summer season (and in winter), due to the differential development of crop types.

(3) Coincidence of preferred conditions: the mixed rotation meant that the coincidence of pre-ferred fi eld content and preferred fi eld location occurred in at least two of the four experimen-tal years – 2002 and 2004. Between the peaks, sub-optimal combinations of crop type and crop location were still an improvement on blanket coverage by one crop, especially winter wheat.

Since high-quality habitats such as weedy set-aside rarely occupy more than 10% of the land area, and good margins no more than 5% of the land area, including crops such as oilseed rape and peas in the rotation can double or triple the potential habitat

available to some species. For example, pea crops can provide good breeding conditions for lapwings in otherwise unsuitable cereal landscapes (Henderson et al., 2004). Small changes in the mosaic pattern of landscapes can affect large num-bers of birds, as open monocultures only support a fraction of their potential biodiversity. This project has demonstrated that conventionally managed, viable crops can also be favourable habitats for birds. This, together with high quality habitats, such as boundaries and fi eld margins, will support larger bird populations with minimal negative consequences for crop management. The principle of creating a mosaic of cropped and out-cropped areas can be applied to most agricultural systems.

Observations on agronomyTraditionally a mixed rotation provides a break in the cycle of various weeds, pests and diseases, which if left unchecked would seriously damage following crops. This remains an important factor in the adop-tion of mixed rotations. Other factors which affect the adoption of a mixed rotation include the rela-tionship with other crops in the rotation, such as the transfer of diseases or pests between crops, or the timing of operations. In some cases, a late-harvested crop can affect the establishment of following crops. Capital requirements also need to be considered according to whether potential crops fi t with exist-ing machinery, processing and storage facilities.

Field margin management

Field margins (uncultivated areas around the edge of a fi eld) provide a valuable wildlife habitat when sown with an appropriate mixture of grasses, legumes and/or wildfl owers. Such margins can be created quickly and produce a rapid rise in bio diversity. Field margins can also boost natural pest control by attracting benefi cial insects that predate or parasitize pests. However, their composition and management is highly variable, and this project aimed to assess and optimize four types of fi eld margin management. A comparison of the annual costs for four different types of margins is shown in Table 6.

In response to the loss of biodiversity from agricultural landscapes over the last 50 years, a number of agri-environment schemes have been introduced (Dobbs & Pretty, 2004). These schemes have the potential to increase the provision of

56 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

habitats for farmland wildlife over a large scale. However, it will be important that the habitats created are both of high quality for biodiversity and practical for farmers to manage.

One sub-project (2000–2004) compared six margin options at six different sites. Margins were monitored for populations of bumblebees, butterfl ies, ground and canopy-dwelling invertebrates (also in the fi nal year small mammals, birds). Six habitats were compared:

(1) Crop (control): conventional arable crop management.

(2) Conservation headland: arable crop manage-ment with restricted herbicide and insecticide application.

(3) Natural regeneration: uncropped margin with annual autumn cultivation and no inputs.

(4) Tussocky grass: uncropped margin sown with fi ve grass species.

(5) Wildfl ower: uncropped margin sown with 21 species of native wildfl ower and four species of fi ne grass.

(6) Pollen and nectar: uncropped margin sown with four species of agricultural legume and four species of fi ne grass.

In addition, a plot of 0.3ha in the centre of each fi eld was sown annually with a wild bird seed mix-ture of four seed-bearing cover crops to provide food and cover for farmland birds throughout the winter. Over the course of the project, it was evident that habitats of high biodiversity value can be recre-ated on arable land using existing farming skills, provided a simple set of management prescriptions is followed. Abundance and diversity of invertebrate and vertebrate groups increased dramatically, even in the fi rst year of the project (bumblebees up 600-fold in margins compared with crop, butterfl ies up 75-fold, and spiders up threefold). These increases were maintained over the course of the project. No single treatment or habitat type was preferred by all the groups studied, suggesting that a diversity of high quality habitats is important for conserving all farmland wildlife. We conclude that fi eld margins are an important part of farm management for bio-diversity, and that a mix of approaches is required.

Financial analysis

Table 7 shows the gross margins obtained for different management regimes in winter wheat.

Table 7 Effects of four management regimes on gross margins of arable cropping (whole rotation)

Practice 2002 2003 2004 Mean

Conventional £570 £685 £605 £610

Reduced pesticides £515 £380 — £445

Reduced nitrogen fertilizer — £605 £340 £475

Spring vs. winter cropping — £635 £535 £585

Reduced pesticides and fertilizer £575 £335 £565 £490

Table 6 Field margin cost comparisons

Type of margin Lifespan of margin (years) Annual cost in £ ha−1 year−1

Grass and wildfl ower mix 10 70

Pollen and nectar mix 5 80

Tussocky grass 10 15

Natural vegetation 1 25

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 57

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

The relatively high gross margin achieved with spring cropping was maintained by selling spring wheat into milling markets, often receiving a £20 t�1 premium, and through the reduced input costs of spring wheat compared to winter wheat (£170 ha�1 compared with £230 ha�1). Pesticides usually account for half of variable costs in winter wheat production (£115 ha�1), and can be higher on farms such as Colworth where grass weed pres-sure is high. The removal of pesticides from the management system therefore reduced variable costs substantially. However, subsequent weed build-up dramatically reduced crop yield and qual-ity, and hence affected gross margin.

Fertilizer costs usually account for 30% of the total variable costs, so their removal or reduction results in smaller variable cost savings than for pesticides. Whilst crop yields were reduced and dif-fi culties experienced in reaching quality parameters for spring milling wheat crops, the overall effect on output was not as great as for pesticide reduction (see Table 7). Gross margins for winter wheat varied widely in 2003 and 2004, mainly because of signifi cant yield variation. In 2003, lowering fertilizer reduced yield from 9.8t ha�1 (£735 ha�1) to 7.5 t ha�1 (£565 ha�1), a 23% reduction in both yield and gross margin, whereas in 2004 lower fer-tilizer reduced yields from 8.2t ha�1 (£530 ha�1) to 4.9 t ha�1 (£370 ha�1), a 40% reduction in yield and 29% reduction in gross margin.

In general, reducing fertilizer inputs reduced crop yields, but it also reduced the abundance of pernicious weeds, meaning less competition for the crop and less reliance on herbicide inputs. Yields in low-fertilizer plots where pesticides were also reduced and were therefore higher than when pesticides alone were reduced. Meanwhile, greater variable cost savings were made than in the reduced N fertilizer scenario, and overall gross margin was higher than when either of the input types was reduced separately. It was also observed that, when the winter wheat commodity price was low, the combined lower input regime provided a more favourable gross margin than the conventional crop. In 2002 for example, conventional plots yielded 8.8tha�1 (gross margin £570ha�1), whereas com-bined treatment plots yielded 6.3tha�1 (gross mar-gin of £575ha�1) selling into the same market. This was important, especially as the low-input crop did not attract premium prices in this case.

Conclusions

Sustainability indicators

The discipline of assessing impacts of changes in farm practice using the 10 indicator structure (later 11) has been helpful in that it permitted science-based, quantitative and qualitative environmental, social, economic and commercial assessments to be made within a single structure. It further enabled differ-ent agricultural operations (e.g. spring vs. winter cropping) to be compared and contrasted, and real choices made for the farming practices with the best commercial and environmental profi les. However, it was clear that there was not enough time, even during this fairly lengthy experiment, to establish what were the best indicator parameters to moni-tor and then get both baseline and long-term data related to change linked to these parameters. Thus the initial plans to measure progress towards or away from sustainability were not achieved. It also proved diffi cult to separate the effects of the agricultural crop and management in any one year from long-term site effects (e.g. the spread of blackgrass) and wide variations in climate and bird populations.

The indicator framework did, though, have a positive impact on the various institutions involved. It was essential for Unilever’s understanding of the trade-offs between a variety of farm treatments. It also had a positive impact on all the project part-ners, whose views changed signifi cantly over time, because annual results were discussed together and co-invention of new treatment ideas was a joint activity. The Colworth project also contributed to the wider policy thinking. For example, the nutrient leaching work has directly contributed to the policy development on nitrate vulnerable zones (NVZs), and Colworth is now a permanent NVZ monitoring site. One problem that became clear was that ditch-side compliance with the Water Framework Directive is impossible, if farmers are to be able to produce crops and have a livelihood. The outreach work at Colworth with farmers locally has led to a variety of innovations, such as the controlled traffi c farming work, as well as helped to shift attitudes about sustainable agricul-ture. However, it was diffi cult (at the time) to trans-late the treatments involving extreme reductions in fertilizers and pesticides into marketable stories

58 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

for farmers, although some wheat was sold as conservation grade.

Implications for farm practices

The overall conclusion for farm practices from this research is that an optimal rotation has both spring and winter crops, as this spreads labour costs on farm and environmental costs (e.g. the trade-offs between nutrient leaching and farmland biodiver-sity). Herbicides can be targeted, particularly in spring crops, to give diverse undercover in the crop, which benefi ts biodiversity, but enhances eco- system services of invertebrates. These insights are refl ected in the company’s new guidelines on sustainable agricultural practices (Unilever, 2003). It was also clear that crop choice has an effect on biodiversity and ecosystem services, particularly if habitats are put in place to encourage these inter-actions and the habitat and crop are within forag-ing distances of key bird populations. Oilseed rape is good for invertebrates and birds, and spring crops in general are benefi cial as they lengthen the time for nesting opportunities for birds.

This project thus established the following key lessons:

(1) It is diffi cult to disaggregate the relative effects of pesticides and fertilizers on yields (e.g. high fertilizer use encourages weeds, thereby neces-sitating higher herbicide use).

(2) Ploughing has a greater infl uence on nitrate leaching than reducing fertilizer rates, particu-larly on the heavy soil at Colworth.

(3) Band spraying can signifi cantly reduce pesticide leaching.

(4) Bird diversity and numbers respond quickly to the introduction of winter oilseed rape, spring peas and set-aside in combination with estab-lished hedges and establishing certain types of fi eld margins (in what formerly was a block area of crops on a 1 in 3 rotation – winter wheat, winter beans, winter oilseed rape).

(5) Birds and invertebrates respond positively to reduced herbicide applications, particularly when this results in a diverse, low density fl ora in the crop; high densities of pernicious weeds do not benefi t birds or invertebrates and signif-icantly reduce yield.

(6) Diverse low density fl ora within the crop are only possible in spring crops.

(7) Spring ploughing is detrimental to earthworms compared with autumn ploughing.

(8) When commodity prices are low, the gross mar-gins for the extreme and conventional treatments are comparable.

Wider implications

The results of the Colworth project suggest that key components of successful sustainable farming projects include management to create a more diverse landscape, and close attention to the timing and frequency of agrochemical applications. Both factors were signifi cant in supporting higher levels of biodiversity on the farm, the former with less impact on crop yields. Landscape diversifi cation is a pragmatic option for delivering more sustainable practices on farms, allowing the incorporation of conventionally managed crops to increase the avail-ability of habitats for wildlife, without affecting crop yields. Market conditions and future legisla-tion for the control of diffuse water pollution may constrain farmer fl exibility in designing mixed rotations. However, the deregulation of crop subsi-dies and the move towards single farm payments with stronger environmental requirements may remove some of these constraints.

Future research should focus on the optimal arrangements of mixed rotations in winter and summer to achieve both water pollution control and higher levels of biodiversity under different environmental circumstances. Winter cover crops are a particular area of promise that deserves greater attention. Pest control was diffi cult to manipulate in a way that would allow ‘acceptable’ populations of weeds or invertebrates to thrive without damag-ing crops and reducing profi tability. Intermediate levels of pest control were not tested, however, and this should attract serious further investigation. Future development of more selective herbicides and non-chemical ‘push–pull’ technologies may also help. For example, if pernicious weeds can be selectively removed, the outcomes for wider biodi-versity can be positive.

The project also highlighted the practical diffi -culties associated with growing profi table crops whilst achieving the 50 mg l�1 nitrate limit set out in EU legislation. Reducing N fertilizer applica-tions alone does not achieve this goal and in some circumstances has jeopardized crop growth to the

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 59

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

point of commercial failure. Using other approaches in combination offers the potential for substan-tially reduced nitrate losses. These include use of cover crops, cultivation timing, careful control of N fertilizer, and changes to crop rotations. There were encouraging indications that pesticide leaching losses can be substantially reduced by the use of band spraying in combination with mecha-nical weeding.

Clearly there are practical and economic issues to consider when changing any management prac-tices. The effects of experimental scenarios on crop yields varied widely, and at worst led to 60% reductions. At other times yields were only slightly reduced and gross margins were actually higher for experimental plots with low inputs of fertilizers and pesticides. Reducing pesticides had the greatest effect on yield, especially in winter-sown crops. Reducing N inputs had less of an effect, but despite reduced input costs, gross margins were still adversely affected. It should be noted that compar-ing gross margins assumes similar fi xed costs across the same farming system, whereas these project scenarios effectively represent different farming systems. Gross margin data, whilst offering an empirical indication of major differences in fi nan-cial viability of the imposed treatments, should not be relied upon as the sole measure. A detailed fi nancial analysis would take into account either fi xed costs (e.g. labour/machinery), externality costs (e.g. nitrate or pesticide leaching to drinking water, eutrophication effects, etc.), and cash fl ow.

The cost–benefi t differential for sustainable prac-tices depends not only on market, legislative and practical factors, but also on capital and capability investment. These issues need to be considered carefully alongside technical factors. The Colworth farm project has achieved and exceeded its original aims. It has proven the potential for clear benefi t, and highlighted practices that need more work before they can be implemented. It shows the need to consider the whole farming system when look-ing at sustainability measures. It has also provided Unilever with helpful management techniques that can be introduced into its supply chains, and others that can be developed further.

The project has also provided a platform for further sustainability research. One example is controlled traffi c farming, where trials are under-way for improving soil health. The approach has

generated much interest in the international farm-ing community, and the Colworth farm is pioneer-ing the use of the technology in Europe. Other parts of the experimental programme have been adopted and continued by other organizations – nitrate monitoring, for example, has led to a government-funded initiative to measure nitrate losses at catchment level to support the Nitrates Directive monitoring requirements. Bird data from Colworth will also contribute to the government debate on how best to achieve its target for national bird population recovery on farmland by 2020. The intention is that the Colworth farm will continue to explore opportunities for farming innovation, and seek to contribute to the con-tinuing development of innovative sustainable farming practices.

Acknowledgements

The authors are very grateful for help and support from a wide range of people in the production of this papers: Jeroen Bordewijk, Robert Borrill, Liz Chadd, Prof. Olaf Christen, Erich Dumelin, Dimitris Efthymiopoulos, Richard Fairburn, Fabrizio Fontana, Billy Ghansah, Claudia Guelke, D.G. Hegde, Kostas Konstantopoulos, Willem-Jan Laan, Shaohong Ma, Neville Martin, Anniek Mauser, Innes McEwen, Sikke Meerman, Evert-Jan Mink, Lettemieke Mulder, James Onsando, A. Ramesh, Rogerio Rangel, Hans Reiterer, Randy Rickert, Sylvia Rutatina, Volker Schick, Gabriel Tuei, Gerrit van Duijn, and Colin Wright. Some of the authors and contributors are or were employees of Unilever: this paper is an independent review led by scientists in independent institutions, and has not been infl u-enced or shaped by any specifi c requirements of the company itself. Two of the authors (Pretty, Goulding) are members of the Unilever Sustainable Agriculture Steering Group and the Sustainable Agriculture Advisory Board.

References

ACRE (2007) Managing the Footprint of Agriculture: Towards a Comparative Assessment of Risks and Benefi ts for Novel Agricultural Systems. London: Advisory Committee on Releases to the Environment, Defra.

60 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

Altieri, M.A. (1995) Agroecology: The Science of Sustainable Agriculture. Boulder: Westview Press.

Balfour, E.B. (1943) The Living Soil. London: Faber and Faber.

Baumol, W.J. and Oates, W.E. (1988) The Theory of Environmental Policy. Cambridge: Cambridge University Press.

Bawden, R. (2007) A paradigm for persistence: A vital challenge for the agricultural academy. Int Journal of Agric. Sust. 5 (1), 17–24.

Bunch, R. and Lopez, G. (1999) Soil recuperation in Central America. In F. Hinchcliffe, J. Thompson, J.N. Pretty, I. Guijt and P. Shah (eds) (1999) Fertile Ground: The Impact of Participatory Watershed Management (pp. 32–41). London: Intermediate Technology Publ.

Buttel, F.H. (2003) Internalising the societal costs of agricultural production. Plant Physiol. 133, 1656–1665.

Chambers, R., Pacey, A. and Thrupp, L.A. (eds) (1989) Farmer First: Farmer Innovation and Agricultural Research. London: IT Publ.

Champion, G.T., May, M.J., Bennett, S., Brooks, D.R., Clark, S-J., Daniels, R.E., Firbank, L.G., Haughton, A.J., Hawes, C., Heard, M.S., Perry, J.N., Randle, Z., Rossall, M.J., Rothery, P., Skellern, M.P., Scott, R.J., Squire, G.R. and Thomas, M.R. (2003) Crop management and agronomic context of the Farm Scale Evaluations of genetically modifi ed herbicide-tolerant crops. Phil. Trans. Roy. Soc. Lond. 358, 1801–1818.

Clements, D. and Shrestha, A. (2004) New Dimensions in Agroecology. Binghampton, NY: Food Products Press.

Conway, G.R. (1997) The Doubly Green Revolution. London: Penguin.

Conway, G.R. and Pretty, J.N. (1991) Unwelcome Harvest: Agriculture and Pollution. London: Earthscan.

Cormack, B. and Metcalfe, P. (2000) Energy Use in Organic Farming Systems. Terrington: ADAS.

Costanza, R., d’Arge, R., de Groot, R., Farber, S., Grasso, M., Hannon, B., Limburg, K., Naeem, S., O’Neil, R.V., Paruelo, J., Raskin, R.G., Sutton, P. and van den Belt, M. (1997) The value of the world’s ecosystem services and natural capital. Nature 387, 253–260.

Costanza, R., Graumlich, L. and Steffen, W. (2007) Sustainability or Collapse? Cambridge, MA: MIT Press.

Cox, T.S., Picone, C. and Jackson, W. (2004) Research priorities in natural systems agriculture. In D. Clements, and A. Shrestha (eds) (2004) New Dimensions in Agroecology. Binghampton, NY: Food Products Press.

Daily, G. (ed.) (1997) Nature’s Services: Societal Dependence on Natural Ecosystems. Washington DC: Island Press.

DETR (1998) Management of Harm to the Environment. London: DETR.

Dobbs, T. and Pretty, J.N. (2004) Agri-environmental stewardship schemes and multifunctionality. Review of Agricultural Economics 26 (2), 220–237.

Dobson, A. (ed.) (1999) Fairness and Futurity. Oxford: Oxford University Press.

FAO (2006) FAOSTAT database. Rome.Ferro Luzzi, A. and James, P. (2000) European Diet and

Public Health: The Continuing Challenge. Eurodiet Final Report.

Firbank, L.G., Rothery, P., May, M.J., Clark, S.J., Scott, R.J., Stuart, R.C., Boffey, C.W.H., Brooks, D.R., Champion, G.T., Haughton, A.J., Hawes, C., Heard, M.S., Dewar, A.M., Perry, J.N. and Squire, G.R. (2005) Effects of genetically modifi ed herbicide- tolerant cropping systems on weed seedbanks in two years of following crops. Biol Lett 42, 1–5.

Frumkin, H. (ed.) (2005) Environmental Health: From Global to Local. San Francisco: Jossey-Bass.

Gallagher, K., Ooi, P., Mew, T., Borromeo, E., Kenmore, P. and Ketelaar, J-W. (2005) Ecological basis for low-toxicity integrated pest management (IPM) in rice and vegetables. In J. Pretty, (ed.) (2004) The Pesticide Detox. London: Earthscan.

Giles, J. (2005) Nitrogen study fertilizes fears of pollu-tion. Nature 433, 791.

Gliessman, S.R. (2004) Integrating agroecological pro-cesses into cropping systems research. In D. Clements, and A. Shrestha (eds) New Dimensions in Agroecology. Binghampton, NY: Food Products Press.

Gliessman, S.R. (2005) Agroecology and agroecosys-tems. In J. Pretty (ed.) The Earthscan Reader in Sustainable Agriculture. London: Earthscan.

Goodman, D. and Watts, M.J. (eds) (1997) Globalising Food: Agrarian Questions and Global Restructuring. London and New York: Routledge.

Hanley, N., MacMillan, D., Wright, R.E., Bullock, C., Simpson, I., Parrison, D. and Crabtree, R. (1998) Contingent valuation versus choice experiments: Esti-mating the benefi ts of environmentally sensitive areas in Scotland. Journal of Agricultural Economics 49 (1), 1–15.

Hazell, P. and Wood, S. (2008) Drivers of change in agri-culture. In C. Pollock, I. Crute, H. Dalton, C. Leaver and J. Pretty (eds) Philosophical Transactions of the Royal Society of London (series B) 363 (1491), 495–516, Special Issue on Sustainable Agriculture.

Henderson. I., Clark, N., Bodey, T., Holloway, S.J. and Armitage, M. (2004) The use of pea crops by farmland birds: Evidence for extended breeding and enhanced feeding opportunities in crop mosaics. British Trust for Ornithology Research Report No. 358.

Henderson, I., Clark, N. and Holloway, S. (2006) BTO Research Report No. 457. Changes in Farmland Bird Abundance at Colworth 2000–2006. A report by the British Trust for Ornithology under contract to Unilever Research. November 2006.

Hinchcliffe, F., Thompson, J., Pretty, J., Guijt, I. and Shah, P. (eds) (1999) Fertile Ground: The Impacts of Participatory Watershed Development. London: IT Publications.

MULTI-YEAR ASSESSMENT OF UNILEVER’S PROGRESS TOWARDS AGRICULTURAL SUSTAINABILITY I 61

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

Jackson, D.L. and Jackson, D.L. (2002) The Farm as Natural Habitat. Washington DC: Island Press.

Lampkin, N.H. and Padel, S. (eds) (1994) The Economics of Organic Farming. An International Perspective. Wallingford: CAB International.

Leach, G. (1976) Energy and Food Production. London: IPC Science and Technology Press, Guildford and IIED.

Leake, A. (2000) The development of integrated crop management in agricultural crops: Comparisons with conventional methods. Pest Manag Sci. 56, 950–953.

Lewis, W.J., van Lenteren, J.C., Phatak, S.C. and Tumlinson, J.H. (1997) A total system approach to sustainable pest management. Proc Nat Acad Sci 94, 12243–12248.

Li Wenhua (2001) Agro-Ecological Farming Systems in China. Man and the Biosphere Series Volume 26. Paris: UNESCO.

MAFF (2000) Towards Sustainable Agriculture: A Pilot Set of Indicators. London: MAFF.

MEA (Millennium Ecosystem Assessment) (2005) Ecosystems and Well-Being. Washington, DC: Island Press.

McNeely, J.A. and Scherr, S.J. (2003) Ecoagriculture. Washington, DC: Island Press.

Moss, B. (2007) Water pollution by agriculture. In C. Pollock, I. Crute, H. Dalton, C. Leaver and J. Pretty (eds) Philosophical Transactions of the Royal Society of London (series B). Special Issue on Sustainable Agriculture.

Naess, A. (1992) Deep ecology and ultimate premises. Society and Nature 1 (2), 108–119.

Norse, D., Li, Ji, Jin Leshan and Zhang Zheng. (2001) Environmental Costs of Rice Production in China. Bethesda: Aileen Press.

NRC. (2000) Our Common Journey: Transition Towards Sustainability. Board on Sustainable development, Policy Division, National Research Council. Washington DC: National Academy Press.

OECD (1998) Report on the, OECD. Workshop in Agri-Environmental Indicators. COM/AGR/CA/ENV/EPOC (98), Paris 136. Olsson, P. and Folke, P. (2001) Local ecological knowledge and institutional dynamics for ecosystem management: A study of Lake Racken watershed, Sweden. Ecosystems 4, 85–104.

Pretty, J. (1995) Regenerating Agriculture: Policies and Practice for Sustainability and Self-Reliance (320 pp.). London: Earthscan. Washington: National Academy Press.

Pretty, J. (ed.) (2005) The Pesticide Detox. London: Earthscan.

Pretty, J. (2007) Agricultural sustainability: Concepts, principles and evidence. In C. Pollock, I. Crute, H. Dalton, C. Leaver and J. Pretty (eds) Philosophical Transactions of the Royal Society of London (series B). Special Issue on Sustainable Agriculture.

Pretty, J. and Ward, H. (2001) Social capital and the environment. World Development 29 (2), 209–227.

Pretty, J., Brett, C., Gee, D., Hine, R., Mason, C.F., Morison, J.I.L., Raven, H., Rayment, M. and van der Bijl, G. (2000) An assessment of the total external costs of, UK agriculture. Agricultural Systems 65 (2), 113–136.

Pretty, J.N., Ball, A.S., Li, Xiaoyun and Ravindranath, N.H. (2002) The role of sustainable agriculture and renewable resource management in reducing green-house gas emissions and increasing sinks in China and India. Phil. Trans. Roy. Soc. Series A 360, 1741–1761.

Pretty, J., Mason, C.F., Nedwell, D.B. and Hine, R.E. (2003a) Environmental costs of freshwater eutrophi-cation in England and Wales. Environmental Science and Technology 37 (2), 201–208.

Pretty, J., Noble, A., Bossio, D., Dixon, J., Hine, R.E., Penning de Vries, P. and Morison, J.I.L. (2006) Resource conserving agriculture increases yields in developing countries. Environmental Science and Technology 40 (4), 1114–1119.

Pretty, J., Smith, G., Goulding, K.W.T., Groves, S.J., Henderson, I., Hine, R.E., King, V., van Oostrum, J., Pendlington, D.J., Vis, J.K. and Walter, C. (2008) Multi-year assessment of Unilever’s progress towards agricultural sustainability: outcomes for peas (UK), spinach (Germany, Italy), tomatoes (Australia, Brazil, Greece, USA), tea (Kenya, Tanzania, India) and oil palm (Ghana). International Journal of Agricultural Sustainability 6 (1), 64–89.

Royal Commission on Environmental Pollution (RCEP) (1998) Setting Environmental Standards, Cmnd 4053 21st (Report. HMSO).

Ruttan, V. (1999) The transition to agricultural sustain-ability. Proc Natl Acad Sci 96, 5960–5967.

Smil, V. (2001) Enriching the Earth. Cambridge, MA: MIT Press.

Smith, P., Powlson, D.S., Smith, J.U., Falloon, P. and Coleman, K. (2000) Meeting Europe’s climate change commitments: Quantitative estimates of the potential for carbon mitigation by agriculture. Global Change Biology 6, 525–539.

Smith, P. and Trines, E. (2007) Agricultural measures for mitigating climate change: Will the barriers pre-vent any benefi ts to developing countries? IJAS 4 (3), 173–175.

Sommerville, L. and Walker, C.H. (ed.) (1990) Pesticide Effects on Terrestrial Wildlife. London: Taylor and Francis.

Stirling, A. and Mayer, S. (2000) Rethinking Risk. Sussex: SPRU.

Stout, B.A. (1998) Energy for agriculture in the 21st cen-tury. In J.C. Waterlow, D.G. Armstrong, L. Fowden, and R. Riley, (eds) Feeding the World Population of More Than Eight Billion People. New York and Oxford: Oxford University Press.

Swift, M.J., Izac, A-M.N. and van Noordwijk, M. (2004) Biodiversity and ecosystem services in agricultural landscapes – are we asking the right questions. Agric Ecossys and Environment 104, 113–134.

Swingland, I. (ed.) (2003) Capturing Carbon and Conserving Biodiversity. London: Earthscan.

62 J. PRETTY ET AL.

INTERNATIONAL JOURNAL OF AGRICULTURAL SUSTAINABILITY 6(1) 2008, PAGES 37–62

Tegtmeier, E.M. and Duffy, M.D. (2004) External costs of agricultural production in the United States. Int. J. Agric. Sust. 2 (1), 1–20.

Tilman, D. (1999) Global environmental impacts of agricultural expansion: the need for sustainable and effi cient practices. Proc Natl Acad Sci 96, 5995–6000.

Tilman, D., Cassman, K.G., Matson, P.A., Naylor, R. and Polasky, S. (2002) Agricultural sustainability and intensive production practices. Nature 418, 671–677.

Tomich, T.P., Chomitz, K., Francisco, H., Izac, A-M.N., Murdiyarso, D., Ratner, B.D., Thomas, D.E. and van Noordwijk, M. (2004) Policy analysis and environ-mental problems at different scales: Asking the right questions. Agric. Ecosys and Environment 104, 5–18.

Townsend, A.R., Howarth, R.W., Bazzaz, F.A., Booth, M.S., Cleveland, C.C., Collinge, S.K., Dobson, A.P., Epstein, P.R., Holland, E.A., Keeney, D.R., Mallin,

M.A., Rogers, C.A., Wayne, P. and Wolfe, A.H. (2003) Human health effects of a changing global nitrogen cycle. Front Ecol Environ 1 (5), 240–246.

Trewevas, A. (2002) Malthus foiled again and again. Nature 418, 668–670.

Unilever (2003) Sustainable Vining Peas – Good Agricultural Practice Guidelines (28 pp.). Unilever.

Uphoff, N. (1998) Understanding social capital: Learning from the analysis and experience of participation. In P. Dasgupta, and I. Serageldin (eds) Social Capital: A Multiperspective Approach. Washington, DC: World Bank.

Uphoff, N. (ed.) (2002) Agroecological Innovations. London: Earthscan.

Victor, T.J. and Reuben, R. (2002) Effects of organic and inorganic fertilizers on mosquito populations in rice fi elds of southern India. Med Vet Entomol. 14, 361–368.