How effective are European agri-environment schemes in conserving and promoting biodiversity?

Authors

  • David Kleijn,

    Corresponding author
    1. Nature Conservation and Plant Ecology Group, Wageningen University, Bornsesteeg 69, 6708 PD Wageningen, The Netherlands; and
    Search for more papers by this author
  • William J. Sutherland

    1. Centre for Ecology, Evolution and Conservation, School of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, UK
    Search for more papers by this author

*Correspondence: David Kleijn. fax +31 317 484845. E-mail david.kleijn@wur.nl

Summary

  • 1Increasing concern over the environmental impact of agriculture in Europe has led to the introduction of agri-environment schemes. These schemes compensate farmers financially for any loss of income associated with measures that aim to benefit the environment or biodiversity. There are currently agri-environment schemes in 26 out of 44 European countries.
  • 2Agri-environment schemes vary markedly between countries even within the European Union. The main objectives include reducing nutrient and pesticide emissions, protecting biodiversity, restoring landscapes and preventing rural depopulation. In virtually all countries the uptake of schemes is highest in areas of extensive agriculture where biodiversity is still relatively high and lowest in intensively farmed areas where biodiversity is low.
  • 3Approximately €24·3 billion has been spent on agri-environment schemes in the European Union (EU) since 1994, an unknown proportion of it on schemes with biodiversity conservation aims. We carried out a comprehensive search for studies that test the effectiveness of agri-environment schemes in published papers or reports. Only 62 evaluation studies were found originating from just five EU countries and Switzerland (5). Indeed 76% of the studies were from the Netherlands and the United Kingdom, where until now only c. 6% of the EU agri-environmental budget has been spent. Other studies were from Germany (6), Ireland (3) and Portugal (1).
  • 4In the majority of studies, the research design was inadequate to assess reliably the effectiveness of the schemes. Thirty-one percent did not contain a statistical analysis. Where an experimental approach was used, designs were usually weak and biased towards giving a favourable result. The commonest experimental design (37% of the studies) was a comparison of biodiversity in agri-environment schemes and control areas. However, there is a risk of bias if either farmers or scheme co-ordinators select the sites for agri-environment schemes. In such cases the sites are likely to have a higher biodiversity at the outset compared to the controls. This problem may be addressed by collecting baseline data (34% of studies), comparing trends (32%) or changes (26%) in biodiversity between areas with and without schemes or by pairing scheme and control sites that experience similar environmental conditions (16%).
  • 5Overall, 54% of the examined species (groups) demonstrated increases and 6% decreases in species richness or abundance compared with controls. Seventeen percent showed increases for some species and decreases for other species, while 23% showed no change at all in response to agri-environment schemes. The response varied between taxa. Of 19 studies examining the response of birds that included a statistical analysis, four showed significant increases in species richness or abundance, two showed decreases and nine showed both increases and decreases. Comparative figures for 20 arthropod studies yielded 11 studies that showed an increase in species richness or abundance, no study showed a decrease and three showed both increases and decreases. Fourteen plant studies yielded six studies that showed increases in species richness or abundance, two showed decreases and no study showed both increases and decreases.
  • 6Synthesis and applications. The lack of robust evaluation studies does not allow a general judgement of the effectiveness of European agri-environment schemes. We suggest that in the future, ecological evaluations must become an integral part of any scheme, including the collection of baseline data, the random placement of scheme and control sites in areas with similar initial conditions, and sufficient replication. Results of these studies should be collected and disseminated more widely, in order to identify the approaches and prescriptions that best deliver biodiversity enhancement and value for money from community support.

Introduction

Post-war European agriculture can be considered a success in that it has resulted in increased yields and an enhanced capacity for self-sufficiency. For example, in the UK the yields per hectare of wheat, barley, potatoes and sugar beet have tripled since 1950, while over the same time milk yields have more than doubled (Pretty et al. 2000). However, it is widely accepted that increased agricultural productivity has associated costs in economic, consumer perception and environmental terms.

More recently, there has been a global shift towards reducing subsidies. For example, in the UK, manufacturing subsidies have been virtually eliminated, yet agriculture remains heavily subsidized at about 40% of the income. The free trade talks of the World Trade Organization have repeatedly identified agricultural subsidies as an area badly needing reform, especially the European Union (EU) Common Agricultural Policy (Yu, Sutherland & Clark 2002). The €16 900 million annual cost of the European Union Common Agricultural Policy largely comprises direct payments to farmers, price support, taxing imports from non-EU countries, subsidizing exports and paying for storage when no market is available. As a result, prices in the European Union exceed those on the international market. The external costs of agriculture were estimated by Pretty et al. (2001) to be about €180 per hectare of grassland and arable, with external benefits equivalent to €17 to €50 per hectare. It is widely accepted that the expansion of the European Union in 2004 to include Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovakia and Slovenia will make the current agricultural support mechanisms financially unviable (Donald et al. 2002).

Consumers are currently questioning the benefits of intensive agriculture. While the concerns may not necessarily always be rational (Beringer 2000), there is clear public mistrust and distaste for some aspects of modern agriculture.

The intensification of agriculture has resulted in major environmental problems in recent decades, notably declines in bird populations together with their associated food resources (Donald, Green & Heath 2000; Benton et al. 2002; Robinson & Sutherland 2002) and this is likely to continue (Tilman et al. 2001). Future intensification, such as the use of genetically modified crops, is likely to have further detrimental consequences for biodiversity (Watkinson et al. 2000). There are also implications for wider environmental issues, such as flood risk and effects on water quality (Sutherland 2002).

One response to concerns over biodiversity loss has been the introduction of agri-environment schemes, in which farmers are paid to modify their farming practice to provide environmental benefits. The EU agricultural policy first explicitly addressed the impact of agriculture on the environment in a Green Paper published in 1985 (CEC 1985). The reform of the EU agricultural policy in that year (EEC Regulation 797/85) included a novel set of measures for environmental protection and Article 19 allowed Member States to pay national aid in environmentally sensitive areas (ESAs). In 1992 EEC Regulation 2078/92 was introduced, requiring all EU member states to apply agri-environment measures according to environmental needs and potential. Between 50% and 75% of the costs of approved agri-environment schemes are co-funded by the EU, making this regulation a financially attractive form of environmental protection. Concurrently, extensive agri-environment programmes were developed in Norway and Switzerland (both non-EU Member States) and in Austria and Sweden before their entry into the EU in 1995. Besides their intended positive effects on biodiversity and the environment, agri-environment schemes decouple payments from agricultural output. Thus they continue to provide income transfers to farmers, but in a way that does not distort world markets (Potter & Goodwin 1998; Matthews 2002).

More than a decade after the introduction of regulation 2078/92, little information is available on the effects of agri-environment schemes on biodiversity conservation. The limited number of studies that have been published present contrasting results (e.g. Kleijn et al. 2001; Peach et al. 2001). Most EU countries are currently implementing their second 5-year agri-environment programme. National schemes have been initiated in three, and there are plans for pilot incentive schemes in another six Central and Eastern European countries (Petersen & Feehan 2003). There is an obvious need for an overview that shows exactly what agri-environment schemes achieve in terms of biodiversity conservation. We attempt such a review here.

First, we briefly describe the differences in design and implementation of agri-environment programmes between countries in Europe. Subsequently, we review the effectiveness of agri-environment schemes by surveying all available literature, with the aim of integrating the findings of various studies to produce recommendations for improvement. We have restricted ourselves to the effects of schemes on biodiversity. We only consider schemes implemented until 2000, as the new modified programmes are too recent for proper evaluation. We do not consider set-aside schemes, as these are not formally agri-environment schemes but a means of reducing production, and their ecological merits have been discussed elsewhere (Clarke 1992; Buckingham et al. 1999). Likewise, although organic farming is an agri-environment scheme and support is co-funded by the EU under Regulation 2078/92, we do not consider the effects of organic farming as this has been discussed extensively elsewhere and the objectives are not necessarily biodiversity conservation (Weibull, Bengtsson & Nohlgren 2000; Mäder et al. 2002).

Design of agri-environment programmes across Europe

For clarity, in this review we distinguish between agri-environment programmes, schemes and measures. We consider an agri-environment programme to be the collection of schemes implemented in a country. Individual schemes have different objectives (e.g. grassland extensification or conservation of endangered livestock breeds) and regularly consist of a set of measures. For example, in the case of a grassland extensification scheme, measures (also called prescriptions) may consist of a reduction in stocking densities or a cessation of fertilizer inputs.

Agri-environment programmes vary markedly between countries in Europe (Table 1). The objectives of these programmes usually reflect a combination of the main environmental, ecological and socio-economic problems associated with agriculture, as well as the political situation in each country. In Switzerland, the Netherlands and the United Kingdom, schemes available to farmers concentrate on wildlife and habitat conservation. In Denmark and Germany most schemes offered to farmers aim to reduce agrochemical emissions, while in France the programme is geared towards the prevention of land abandonment in agriculturally marginal areas. In Ireland and Austria, the objectives of programmes are balanced between environmental protection, biodiversity conservation and landscape maintenance (Table 1).

Table 1.  Characteristics of agri-environment programmes in European countries until the year 2000. Pilot agri-environment schemes currently applied in CEE countries are not included. UAA, Utilized Agricultural Area; AEP, agri-environment programme; AES, agri-environment scheme; ECA, ecological compensation area
Austria. (UAA’95 3 425 100 ha; area with AES’97 2 500 000 ha; AEP since 1995, previous programme outside the EU-context since 1972). The Austrian programme (ÖPUL) consists of 25 schemes. Eight horizontal schemes address extensification and reduction of the negative impact of agriculture on the environment, the other zonal schemes address specific farming practices, biodiversity conservation and the creation or conservation of landscape elements. ÖPUL aims to promote farming with reduced environmental impact, maintain farming in agriculturally marginal areas (Alps) and conserve biodiversity and landscape. However, in 1996 83% of the budget was spent on the horizontal schemes and only 17% on schemes aimed at biodiversity and landscape conservation. Schemes with the highest uptake: crop rotation stabilization (18% of AEP budget) and the basic subsidy (17%). Source: Groier & Loibl (2000).
Belgium. (UAA’95 1 354 400 ha; area with AES’97 17 000 ha; AEP since 1994). In Flanders no AEP existed before 2000 (Reheul & van Huylenbroeck 2000). The Walloon programme consists of five horizontal schemes and six zonal schemes. The programme addresses environmental and biodiversity aspects more or less equally but in 1997 only 25% of the AEP area was under some scheme addressing biodiversity or landscape conservation issues. Highest uptake: planting a cover-crop between two crops (41%) and restricting stocking densities to between 0·6 and 1·4 lifestock units (26% of AEP area). Source: Walot (2002).
Denmark. (UAA’95 2 726 600 ha; area with AES’97 94 000 ha; AEP since 1992, previous schemes under regulation 797/85 since 1990). The majority of the schemes of the Danish AEP are applied zonally (ESA approach). Schemes aimed at the reduction of nitrogen use, promotion of rygrass as ground cover and organic farming can be implemented throughout the country. The main objective of the Danish AEP is to achieve a reduction in nitrogen inputs. Landscape and nature protection has been of minor importance so far. Highest uptake: maintenance of extensive grasland (52% of AEP area) and organic farming (37%). Source: Andersen, Henningsen & Primdahl (2000).
Finland. (UAA’95 2 191 700 ha; area with AES’97 2 000 000 ha; AEP since 1995). Finland has a strictly horizontal ‘General Protection Scheme’ (GPS) with six compulsory basic measures and five additional measures of which one has to be selected. Furthermore, a ‘Special Protection Scheme’ (SPS, 12 measures) exists that is optional but participation is available only in combination with the GPS. The emphasis of the Finnish programme is on environmental aspects: one of six compulsory measures and one of five additional measures of the GPS address biodiversity and landscape maintenance. Three of the 12 measures of the SPS address promotion of biodiversity and landscape. Source: M. Kaljonen (unpublished paper).
France. (UAA’95 28 267 200 ha; area with AES’97 5 725 000 ha; AEP since 1992, previous schemes under regulation 797/85 since 1989). In France, national and regional schemes exist alongside ‘local operations’. As regional schemes are the same in each region, both the national and the regional schemes can be considered horizontal whereas the local operations are zonal. Main goal of the AEP is to maintain agricultural activities in areas with a high risk of agricultural land abandonment and rural depopulation. Highest uptake: the national scheme – maintenance of extensive animal husbandry (70% of the total AEP budget) and local operations (c. 15% of AEP budget). By 1997 some 67% of the local operations addressed wildlife and ecosystem protection. Source: Buller & Brives (2000).
Germany. (UAA’95 17 156 900 ha; area with AES’97 6 353 000 ha; AEP since 1992, previous schemes under regulation 797/85 since 1985). The German AEP is difficult to summarize as each federal state (‘Land’) has its own AEP. Almost all schemes are horizontal within each federal state with the exception of schemes aimed at the protection of environment, natural resources, countryside and landscape, which are zonal in some of the states. German agri-environment schemes can be divided in two main types. First, schemes aimed at changing farming practices and second, schemes aimed at the preservation of specific environmentally vulnerable areas, biotopes or species. The latter schemes contribute only 9% of the total AEP area (Osterburg 2001), however, in some federal states these schemes operate outside the framework of regulation 2078/92 and are therefore not co-funded by the EU. c. 70% of the German AEP budget between 1993 and 1996 was spent by the agriculturally extensive German states Bayern, Baden-Würtemberg and Sachsen. Highest uptake: environmentally orientated basic payment – only in Bayern and Sachsen (57% of total German AEP budget) and grassland schemes – extensification, conversion to arable land, preservation of specific biotopes (23%). Source: Grafen & Schramek (2000).
Greece. (UAA’95 3 464 800 ha; area with AES’00c. 49 500 ha; AEP since 1995, previous schemes under regulation 797/85 since 1986). So far, five of a projected 13 schemes have been implemented. The schemes address organic plant production, organic livestock production, 20-year set aside, reduction of nitrogen pollution and conservation of endangered breeds. Highest uptake: reduction of nitrogen pollution (29·500 ha). Source: Louloudis, Beopoulos & Vlahos (2000), Louloudis & Dimopoulos (2001).
Ireland. (UAA’95 4 324 500 ha; area with AES’99 1 575 000 ha; AEP since 1994). The Irish Rural Environmental Protection Scheme (REPS) consists of one scheme only with 11 compulsory measures and a further six ‘Supplementary Measures’. The basic scheme is very comprehensive and addresses biodiversity and environmental protection, training courses and keeping of farm and environmental records. The REPS aims to conserve wildlife habitats and endangered species of flora and fauna as well as to address environmental problems. Five compulsory measures are particularly relevant to biodiversity conservation. All Supplementary Measures are primarily aimed at conservation aspects and only apply in designated areas. Source: Emerson & Gillmor (1999).
Italy. (UAA’95 14 685 500 ha; area with AES’97 1 608 000 ha; AEP since 1994/1995). Italy is divided into 21 regions, each having their own agri-environmental programme. Within regions most schemes are implemented horizontally. The AEP is primarily used as an instrument to reduce the negative impact of agriculture on the environment. Biodiversity conservation is only addressed indirectly through the maintenance of the countryside and the landscape scheme. However, 94% of this scheme is implemented in the provinces of Bolzano, Trento and Valle d’Aosta, and is therefore virtually restricted to the alpine region. Highest uptake: reduction of fertilizer and pesticides inputs (37% of AEP area) and maintenance of countryside and landscape (32%). Source: INEA (1999).
Luxembourg. (UAA’95 126 900 ha; area with AES’97 97 000 ha; AEP since 1996). Only one scheme, available to all farmers in Luxembourg, had been implemented in 1997. This scheme addressed maintenance of the countryside and landscape. Source: Anonymous (1998).
Norway. (UAA 980 000 ha; area with AES unknown). Norway has two major agri-environment schemes. The Acreage and Cultural Landscape Scheme is mainly aimed at maintaining agricultural practices in marginal areas and has general prescriptions that are easy to adapt to. The Special Measures for the Cultural Landscape Scheme consists of much more detailed prescriptions, many having objectives aimed at nature conservation. Highest uptake: unknown. Source: Rønningen (2001).
Portugal. (UAA’95 3 924 600 ha; area with AES’97 606 000 ha; AEP since 1994). Only schemes addressing the reduction of agricultural pollution and training courses and demonstration projects are applied horizontally, all other schemes are zonal and most of them address specific farming systems. Emphasis of the Portugese AEP is on the maintenance of extensive farming systems. The schemes with the expected highest uptake rates are those aimed at the maintenance of extensive grazing systems and Holm Oak landscapes (‘montados’). Highest uptake: not available yet. Source: Eden & Vieira (2000).
Spain. (UAA’95 25 230 300 ha; area with AES’97 532 000 ha; AEP since 1993). The Spanish AEP is implemented by the individual regions but a set of mandatory horizontal and zonal schemes is prescribed by the national government. The implementation of the Spanish scheme has met with considerable delay and data on uptake are only preliminary. Estimated budget allocation suggests that the emphasis of the Spanish AEP lies on landscape protection (48% of AEP budget) and extensification (30%). Highest uptake: preliminary data indicate that landscape conservation and fire prevention in extensive grasslands are the two schemes with the highest uptake rates followed by schemes aimed at wildlife protection in extensive croplands. Source: Peco et al. (2000).
Sweden. (UAA’95 3 059 700 ha; area with AES’97 2 449 990 ha; AEP since 1995, previous schemes outside the EU-context since 1986). The Swedish AEP consists of four clusters of schemes each having a different objective. The ‘environmentally sensitive area’ cluster is zonal, the others are basically horizontal. The AEP objectives are to maintain a naturally and culturally valuable and varied landscape, to conserve biodiversity and to minimize nutrient leaching and pesticide use. Uptake figures indicate that schemes aimed at the maintenance of open landscapes and conservation of cultural-historical remains are very popular, whereas uptake of schemes aimed at biodiversity conservation remain far below the targeted areas. Highest uptake: maintenance of open landscape in forest and northern regions (30% of AEP area) and perennial ley farming (29%). Source: Carlsen & Hasund (2000).
Switzerland. (UAA’99 985 000 ha; area with ECA’99 82 700 ha; ECA since 1993). The Swiss AEP differs considerably from that of EU-member countries. Farmers throughout Switzerland may manage at least 7% of their UAA as so-called Ecological Compensation Areas (ECAs) in order to obtain a basic direct payment. The 7% ECA may consist of a variety of biotopes such as extensive grasslands, traditional orchards, hedges, field margin strips, conservation headlands, ditches, stone walls or unpaved roads. Farmers can receive additional management subsidies for some of these biotopes, such as extensive grasslands. Some types of biotopes, such as again extensive grasslands, that meet a certain quality level and/or are located in ecological corridors between important habitats qualify for additional subsidies. The overall aim of ECAs is halting the agriculturally induced loss of biodiversity by conserving valuable biotopes, restoring degraded biotopes and creating new biotopes. Highest uptake: low-intensity meadows (49% of ECA area) and extensively used meadows (41%). Source: Günter et al. (2002).
The Netherlands. (UAA’95 1 998 900 ha; area with AES’99c. 70 000 ha; AEP since 1992, previous schemes partly under regulation 797/85 and partly outside the EU-context since 1981). The Dutch AEP consists of seven schemes. One scheme (management agreements) specifically addresses the maintenance and conservation of biodiversity and landscape and is applied zonally. All other schemes address a variety of topics including demonstration projects, training courses and public access to farmland. In budgetary terms the zonal scheme is by far the most important. Highest uptake: management agreements (90% of AEP area). Source: Anonymous (2000).
The United Kingdom. (UAA’95 16 446 600 ha; area with AES’97 1 322 000 ha; AEP since 1992, previous schemes under regulation 797/85 since 1987). The AEP varies somewhat between England, Wales, Scotland and Northern Ireland but the basic outline is the same. For the whole of the UK nine different schemes exist of which only one, the ‘Organic Aid Scheme’ is truly horizontal. Others can either be applied in certain regions or address certain biotopes. There is a strong emphasis in the UK AEP on wildlife conservation. The concept of Environmentally Sensitive Areas (ESA) was originally developed in the UK and first implemented here under regulation 797/85 and still forms the backbone of the UK AEP. Wildlife conservation in the wider countryside is addressed by the Countryside Stewardship Scheme. Environmental issues play a minor role (Nitrate Sensitive Areas scheme and Organic Aid Scheme). Highest uptake: ESA scheme (58% of AEP budget and 74% of area) and Countryside Stewardship Scheme (21% of budget and 7% of area). Source: Hart & Wilson (2000).

Schemes can be implemented either horizontally throughout the country or zonally (also known as ‘targeted’ or ‘vertically’) in certain areas that have been identified as being particularly vulnerable or a local biodiversity hotspot (e.g. environmentally sensitive areas (ESAs)). The designation of areas where zonal measures can be implemented is usually carried out by governmental organizations. Most countries have a combination of both approaches because a limited set of zonal schemes exist that aim to conserve vulnerable ecosystems. Switzerland and Finland are the only countries that have entirely horizontal programmes, although most schemes in the German, Irish and Swedish programmes are applied horizontally. By contrast, most schemes in the United Kingdom and Spain are implemented in a zonal manner. A more extensive discussion of the history and lay-out of the agri-environment programmes in a range of European countries is given in Buller, Wilson & Höll (2000).

Patterns of implementation of agri-environment programmes

Differences in uptake rate of individual schemes largely determine whether and where the overall objectives of agri-environment programmes can be met. In most countries uptake is very unequally divided over the available schemes, with a single scheme usually comprising more than 40% of the total area covered by agri-environment schemes (Table 1). Furthermore, schemes are often unequally distributed geographically across countries, with high uptake rates in areas with extensive agriculture and low uptake rates in areas where agriculture is more intensive (Emerson & Gillmor 1999; Buller & Brives 2000; Grafen & Schramek 2000). The mechanism resulting in this pattern is illustrated in Fig. 1(a), which shows that for extensive farmers participation in an agri-environment programme is associated with comparatively low costs of adaptation. Few changes are required to meet the requirements of the schemes (Osterburg 2001). Thus, when uniform payments per hectare (calculated on an average base) are offered for voluntary measures, most uptake will occur in less favoured areas. The same mechanism probably explains why in most countries (especially France and Austria) the low impact/low compensation schemes are those with the highest uptake.

Figure 1.

Conceptual models describing (a) the relationship between farming intensity and the impact of schemes on a farmer's activities (solid line) as well as the uptake of those schemes (dashed line), and illustrating (b) the potential effects of schemes addressing ‘improvement effects’ and ‘protection effects’ (sensuPrimdahl et al. 2003). An equal shift in land-use intensity may result in a more pronounced effect on biodiversity (shaded area) in extensive areas compared with intensive areas.

The effects of agri-environment schemes on biodiversity

EU members are obliged to evaluate their agri-environment programme with respect to their socio-economic, agricultural and environmental aspects (Article 16, EC Regulation 746/96). Currently, most evaluation studies simply examine uptake patterns of different schemes within programmes. However, implementation of schemes does not guarantee that the stated objectives of the scheme will actually be met. Furthermore, the biodiversity and environmental objectives are rarely defined clearly at the outset, which hampers proper evaluation in a number of countries (Schramek 2001).

Table 2 summarizes all those studies that we have been able to locate that evaluate the effects of agri-environment schemes on the abundance or species richness of organisms. Initially, we performed an extensive literature review. However, as most evaluation studies are published outside the mainstream scientific journals, we also searched the internet and approached some 40 key people outside the Netherlands and the United Kingdom to ascertain whether they knew of any evaluation studies in their country or of any person who might have more information. Many studies claimed to evaluate the effects of schemes but simply described the status or trends of species of interest in the scheme site without any reference or control data. These studies cannot be used to infer effects of the changes in management due to the agri-environment schemes, hence we did not consider them further in this review. Although we may have missed some studies, we are confident that we have conducted a thorough search for studies throughout Europe. We located 62 studies from just six countries, of which 76% were from just two countries (18 from the Netherlands and 29 from the United Kingdom). Only 27% (17) of the studies were published in international peer-reviewed journals. Excluding the United Kingdom and Ireland, 83% of the studies were published in the national language and remain therefore largely inaccessible to people outside that country (Table 2, Table 3).

Table 2.  Summary of characteristics of studies that evaluate the effectiveness of agri-environment schemes. Number of replicates and controls in numbers unless other units are given. For abbreviations see Table 1. Studies that just report status or changes within schemes were excluded
CountrySchemeInvestigated species (group)DesignNumber of replicatesNumber of controlsStatistical analysisBase-line dataDuration studyResultsNotesReference
  • *

    Published in the national language.

  • Numbers of replicates differ depending on type of scheme or sampling date.

CHECA – wildflower stripsSkylark1998Relative biotope use within skylark territories24 territoriesYesNo1995Skylarks foraged more frequently and longer in wildflower strips than in any other biotope Weibel (1998)
CHECA – extensive grasslandsCarabid beetlesComparison ECA and control sites 16, 7  7NoNo1997Higher number of species and red list species on extensive and low-intensity grasslands compared to control Pfiffner et al. (2000)*
CHECAGrass-hoppersSpecies richness and abundance on target sites and wider countryside before and after schemes 62398YesYes1990 & 2000Proportion of ECA area relative to total area occupied by grasshoppers increased significantly for seven species from 1990 to 2000ECA sites were perennial biotopes only whereas controls included arable fieldsHunziker (2001)
CHECA – extensive grasslandsGrass-hoppersSpecies richness and abundance on target and control sites before and after schemes152152YesYes1990 & 2000Species richness and abundance of individual species increased more on fields with ECA Peter & Walter (2001)*
CHECABirdsSpatial distribution of territories relative to that of ECA sites 23YesNo1998 & 1999Five species (mostly hedgerow species) more abundant, one species less abundant on/near ECAs than expectedSpatial autocorrelation between ECA and vertical structures. Explains part of the observed effectsHofer et al. (2002)*, 2002 Spiess, Marfurt & Birrer (2002)*
DConservation headlands for arable weedsHoverflies and carabid beetlesComparison AES and control sites  2  2NoNo1988Species richness and abundance of hoverflies and carabid beetles higher on AES sites Raskin (1994)*
DConservation of wet meadowsBlack-tailed godwit, curlew, snipePopulation trends inside/outside AES area  2  2NoNo1989–98Number of pairs inside stable and outside declining or inside declining less rapidly than outside AES areaScheme areas include fields of nature conservation organizationWeiss et al. (1999)*
DConservation of wet meadowsWadersPopulation trends inside/outside AES area2292 ha437 haNoNo1988–98Number of pairs inside stable and outside declining or inside declining less rapidly than outside AES areaScheme areas include fields of nature conservation organizationIkemeyer & Krüger (1999)*
D‘Mittelgebirgs- programm’– grassland extensificationPlantsChanges in species richness on fields with and without AES 29 53NoYes1986 & 1997Plant species richness increases on fields with AES and remains stable on control fields Weis (2001)*
D‘Mittelgebirgs- programm’– Resumed grazing on abandoned pasturesPlantsTrends in species richness on grazed AES fields and exclosures that serve as controls  8  6NoYes1987–90, 1994, 1996 & 1999Plant species richness increases slightly in grazed plots on AES fields and decreases sharply in exclosures Weis (2001)*
DGrazing extensificationPlants, various insect groupsSpecies richness and abundance in a randomized block design  6  6YesNo1996Plant diversity not different, insect richness and abundance significantly higher on scheme sites relative to control sites Kruess & Tscharntke (2002a,b)
EIREPS schemePlants and carabid beetles in grasslandsSpecies richness in field boundaries on farms with and without REPS 15 15YesNo1999Plant species richness lower; carabid beetle richness similar to control farms Feehan, Gillmor & Culleton (2002)
EIREPS schemePlants and carabid beetles in tillage landSpecies richness in field boundaries on farms with and without REPS 15 15YesNo2000Species richness of plants and carabid beetles similar on REPS and control farms Feehan et al. (2002)
EIREPS schemeFarmland birdsSpecies richness on farms with and without REPS 5  5YesNo2000Bird species richness similar on REPS and control farms Flynn et al. (2002)
NLBotanical management agreementsPlantsComparison of changes on fields with and without AES35 9NoNo1984/85 & 1990Changes in species richness/cover similar on AES fields and controlsMost of the control fields located outside the ESAAltenburg & Wymenga (1991)*
NLMeadow bird agreementsMeadow birdsComparison of changes on fields with and without AES23 ha81 haNoYes1988 & 1991Trends in settlement densities similar on fields with and without AES Terlouw (1992)*
NLMeadow bird agreementsMeadow birdsComparison of trends (1) in ESAs and control areas and (2) inside ESAs on fields with and without schemes1 : 11 2 : 901 :7 2 : 276YesPartially1986−901. Trends of two species more positive and one species more negative in ESAs relative to outside ESAs
2. Trends of lapwing more positive on AES fields than control fields
1. ESAs include reserves 2. Prior to the scheme higher densities of three and lower densities of two species were present on AES fields relative to control fieldsVan den Brink & Fijn (1992)*
NLBotanical management agreementsPlantsComparison of trends (1) in ESAs and control areas and (2) inside ESAs on fields with and without schemes45–16929–35YesPartially1986−901. In ESAs more positive vegetation development than outside ESAs in both ditch banks and grasslands
2. Trends more positive on AES fields than control fields in both ditch banks and grasslands
1. ESAs include reserves 2. Prior to scheme ditch banks contain less and grasslands more species on AES fields relative to controlsVan den Brink & Fijn (1992)*
NLMeadow bird agreementsMeadow birdsComparison of population trends on fields with and without AES119 ha144 haNoNo1987−91Population trends more positive on AES fields for three species Brandsma (1993)*
NLMeadow bird agreementsMeadow birdsComparison of population trends on fields with and without AES122 ha702 haNoNo1983, 1986, 1989, 1992 & 1995Population trends more positive on AES fields for six species Altenburg, Rebergen & Wymenga (1993)*, Uilhoorn (1996)*
NLBotanical management agreementsVegetationComparison of shifts in vegetation classes on fields with and without AES255 ha117 haNoNo1987 & 1993Shift towards qualitatively better vegetation classes between 1987 and 1993 more pronounced on fields with AESVegetation broadly classified, significance of results difficult to interpretWymenga, Jalving & Jansen (1994)*
NLMeadow bird agreementsMeadow birdsComparison of population trends on fields with and without AES388 ha420 haNoNo1985, 1987, 1990 & 1993Population trends less negative on AES fields for two speciesMost control fields outside ESA in area with woodlotsAltenburg & Griffioen (1994)*
NLBotanical management agreementsVegetationComparison of changes in ‘Nature Value Index’s’ in edges of fields with and without AES 26161YesYes1990 & 1994Nature Value Index decreases significantly in edges of fields without but stays stable in edges of fields with AES Dijkstra (1994)*
NLBotanical management agreementsVegetationComparison of shifts in vegetation classes on fields with and without AES 86 ha500 haNoNo1988 & 1994Shift towards qualitatively better vegetation classes between 1988 and 1994 more pronounced on fields with AESVegetation broadly classified, significance of results difficult to interpretTer Stege, Jalving & Wymenga (1995)*
NLMeadow bird agreementsMeadow birdsComparison of population trends on fields with and without AES115 ha 49 haYesNo1987, 1990 & 1993No significant differences between fields with and without AES Van Buel & Vergeer (1995)*
NLBotanical management agreementsPlantsComparison of changes on fields with and without AES 14 14NoYes1989 & 1995Trends in species richness/ cover of (hay meadow) plant species more positive on fields with AES Brongers & Kolkman (1996)*
NLMeadow bird agreementsMeadow birdsComparison of densities on fields with and without AES189 ha462 haNoNo1995Higher settlement densities of five species on AES fields Van Buel (1996)*
NLField margin strips and conservation headlandsInsectsComparison of paired field margin strip/conservation headland with conventional crop edge 12, 13 12, 13YesNo1995Higher number of insect taxa, and higher abundance of lady bugs (Coccinellidae), dragon flies (Odonata), bumblebees (Bombus spp.) and hover flies (Syrphidae) on AES stripsAnalysis makes no distinction between conservation headlands and field margin stripsCanters (1996)*
NLBotanical management agreementsFritillary (Fritillariameleagris)Trends in abundance on fields with and without AES 71 32YesYes1990, 1994 & 1998Significant increase in juvenile plants on AES fields relative to controls Brongers (1999)*
NLMeadow bird agreements and botanical management agreementsBirds, plants, bees, hover flies, butterflies, carabidsAbundance and species richness on paired AES and control fields  7  7YesNo1998One carabid beetle species more abundant on fields with AES relative to control sitesWithin ESAs two fields within a pair in environmentally similar areasKleijn et al. (1999)*
NLMeadow bird agreements and botanical management agreementsBirds, plants, bees, hover fliesAbundance and species richness on paired AES and control fields 39 39YesNo2000Diversity and abundance of plants equal, that of insects higher on fields with AES. One bird species less abundant on AES fieldsWithin ESAs two fields within a pair in environmentally similar areasKleijn et al. (2001, in press)
NLMeadow bird agreementsMeadow birdsPopulation trends on paired AES and control fields 17 17Yespartially1989, 1992 & 1995Population trends similar on AES and control fieldsWithin ESAs two fields within a pair in environmentally similar areasKleijn & Van Zuijlen (in press)
PCastro Verde Zonal PlanSteppe birdsChanges in abundance of species in target and control sites 16 17YesYes1995 & 1997Higher numbers of great bustard, lesser kestrel and little bustard in fields with AES Borralho et al. (1999)*
UKNorth Peaks ESABirdsComparison of AES and control sites 1 1YesNo1994–1996Similar densities for eight species but twite and lapwing much lower in ESAESA & control in different regions and surveyed in different yearsADAS (1997a)
UKBreckland ESA – conservation headlandsInvertebrates, plantsComparison of AES and control sites27 9YesNo1993No significant differences for a range of variables ADAS (1997b)
UKRadnor ESA – hay meadowsPlantsChanges in target and control plots1619YesYes1994 & 1997Significant increase in species richness in higher tier sites but not in control or lower tier ADAS (1999b)
UKRadnor ESA – wetlandsPlantsChanges in target and control plots1520YesYes1994 & 1997Significant increase in species richness in higher tier sites but not in control or lower tier ADAS (1999b)
UKYnys Môn ESA – coastal habitatsPlantsChanges in target and control plots2125YesYes1994–1997Significant increases in species suited to grazing in AES stands contradicts target but increase in maritime species is as required ADAS (1999c)
UKYnys Môn ESABirdsComparison of population trends with those in wider countryside20YesNo1995–199813 out of 15 wintering waders and waterfowl decreased. Five of five ‘target’ passerines increased. Two of six breeding waders and waterfowl increasedSample sizes small for breeding wader and waterfowl (mean 2·5 territories in total)ADAS (1999a)
UKLleyn Peninsula ESA – coastal grasslandsPlantsChanges in target and control plots16 4YesYes1995 & 1998Significant increase in species richness in controls but not in AES ADAS (2001b)
UKClwydian Range ESA – calcareous grasslandsPlantsChanges in target and control plots82 2YesYes1995 & 1998No significant difference in species richness between years or treatments ADAS (2001a)
UKClwydian Range ESAButterflies in calcareous grasslandChanges in target and control plots 4 1NoYes1995 & 1998Numbers decreased by 58% on sole control transect but increased by 13% on AES sites ADAS (2001a)
UKESA – arable reversionGrey partridgePopulation trends in target and control areas 1 1YesYes1970–1995Greater declines on area with AES Aebischer & Potts (1998)
UKCountryside Stewardship SchemeStone curlewPopulation trends before and after AES scheme 1 0NoYes1991–1999Increase from 150 pairs in 1991 to 233 in 1999 after AES introduced. Rapid decline between 1940s and 1980sWardens also find nests of and ensure they are not damaged by farming operationsAebischer et al. (2000)
UKESA – corncrake initiativeCorncrakePopulation trends before and after AES scheme 1 0NoYes1993–19984.2% annual increase after introduction scheme (1992–98) compared to 3·5% annual decline in reference period (1988–93)Includes purchase of nature reserves mainly for corncrakeAebischer et al. (2000)
UKPilot Arable StewardshipBumblebeesComparison of paired sites and controls. Carried out for various schemes8484YesNo1999–2000For four schemes higher numbers in AES than controls. For one scheme none on AES. Numbers generally low Allen, Gundrey & Gardner (2001)
UKRegenerating heather moorsMoorland birdsAbundance and trends in areas with and without AES12 (1176 ha) 12 (1032 ha)YesNo1996–2000Black grouse increased 4·6% p.a. with AES but declined 1·7% p.a. in controls. Significantly more females retained broods in AES. Eight of 11 species rarer in AES (two significantly) including black grouse. Waders and other gamebirds declined faster in areas with AES Baines et al. (2002), Calladine Baines & Warren (2002)
UKPilot Arable stewardshipWinter birdsFarms with AES and controls54 48YesNo1998–2000Of 56 tests of groups and areas four significant positive effects and five negative Bradbury (2001), Bradbury & Allen (2003)
UKPilot Arable StewardshipBreeding birdsFarms with AES and controls25 24YesNo1999–2000Of 16 comparisons seven showed positive effect of AES (one, lapwing significant) and nine negative effects (three, woodpigeon, sedge warbler and rook significant) Bradbury & Allen (2003)
UKESA and Countryside Stewardship SchemeButterfliesAbundance and trends in AES and control sites85160YesNo1994–2000Equl numbers increased and decreased. Lower, but non significant decline (12% v 15·5%) on AES sites. 10 of 13 specialist species increased (five significantly)Over 50% sites owned by conservation organizationsBrereton, Stewart & Warren (2002)
UKBarnacle Goose Management SchemeBarnacle gooseTrends in abundance on reserve and areas without disturbance or limited disturbance before and after start scheme 16  0YesYes1990–2000Numbers increased at a proportionately higher rate on AES sites than on reserve. No difference in change between undisturbed and limited disturbance sitesAuthors suggests numbers on reserve reached capacity thus increases elsewhere could be due to buffer effectCope et al. (2003)
UKPilot Arable StewardshipTrue bugsComparison AES sites and paired controls 93 44YesNo1999–2000Higher numbers on six region/treatment combinations (4 significant). Lower (not significant) in remaining combination Gardener et al. (2001b)
UKPilot Arable stewardshipCarabid beetlesComparison various AES options and controls 82–103 31–34YesNo1999–2000Of 29 region/treatment/date combinations higher numbers in AES for 14 (nine significant) and lower in five (two significant). For carabid larvae of 24 region/higher treatment/date comparisons 15 higher in AES and nine lower but none were significant Gardener et al. (2001a)
UKESA – cereal headlandsCarabid beetlesComparison of paired cereal headlands with or without AES 2  2YesNo1991Of three carabids, two more abundant in AES, one more abundant in control Hawthorne, Hassall & Sotherton (1998)
UKCountryside Stewardship SchemeCirl buntingTrends in bird numbers inside scheme or outside within 47 tetrads 47 47YesYes1992–1998Increased by 82% on land in scheme but by 2% on controls Peach et al. (2001)
UKPilot Arable StewardshipSawfliesComparison of sites with AES and adjacent controls224188YesNo1999–2000No obvious effect on sawfly abundance but diversity higher on AESNo distinction made between seven different scheme optionsReynolds (2001)
UKESAs – raised water levelsWadersComparison of trends in AES and control sites 8 4NoYes1992–1997Wader numbers increased in three AES sites, stable in four AES sites, decreased in three AES sites and in four controlsMonitored since 1989, schemes started 1992Chown (1998)
UKBreckland ESA – cereal headland managementCarabid beetles, spiders, HeteropteraComparison of uncropped headlands and cereal headlands with reduced pesticide input with controls 3 2YesNo1988For all groups higher abundance and more species than control in uncropped AES headlands but only for Heteroptera in AES with reduced pesticide inputs Cardwell, Hassall & White (1994); Hassall et al. (1992); White & Hassall (1994)
UKBreckland ESA – cereal headland managementCarabid beetlesComparison of paired cereal headlands with or without AES 4 2YesNo1991More species and diversity than control in uncropped AES headlands but not in AES with reduced pesticide inputs  Hawthorne & Hassall (1995)
UKPilot Arable StewardshipBrown hareComparison of farms with AES and controls4138YesNo1999–2000No difference detected in numbers Tapper (2001)
UKESABreeding skylarkComparison of AES and controls in two ESAs25–22741–49YesNo1994–1996For downland reversion scheme 3–6 times as many skylarks on AES as on controls. For permanent grassland reversion AES had significantly fewer skylarks for one but significantly more for another time period Wakeham- Dawsonet al. (1998)
UKESAWintering skylarksComparison on AES and controls in two ESAs113–11740–47YesNo1994/5– 1996/7Highest number on cereal stubbles, then AES of reverted downland, then far fewer on AES of permanent grassland reversion and fewest on winter wheat Wakeham- Dawson & Aebischer (1998)
Table 3.  Summary of all studies that were published, in congress proceedings or in reports. Percentages are given in relation to the total of 62 studies
Total studies100%
  • *

    Excluding 32 studies from UK and Ireland.

  • Including four studies with just two replicates.

  • Including three studies with just two replicates.

  • §

    Bias resulting from scheme sites likely to be placed in better habitat reduced by use of baseline data, comparing trends/changes in time or pairing of scheme and control sites.

Published in peer reviewed journals27%
In national language*83%
Have control sites90%
Have replication77%
Use statistical tests of significance69%
Analyse changes between two points in time26%
Analyse trends in time32%
Have paired scheme and control sites16%
Have baseline data34%
Controls, replication and statistical analysis58%
Controls, replication, statistical analysis and reduced bias§39%

approaches used to evaluate biodiversity effects of schemes

The approaches to evaluation varied enormously, even within individual countries, making it very difficult to ascribe a specific study design (Table 2). For example, the most common approach (37% of the studies) compared biodiversity in the agri-environment scheme and control areas at one point in time. However, some studies compared entire areas with a mosaic of schemes, nature reserves and conventional management with areas that were managed conventionally throughout and usually were located outside ESAs. Other studies compared the pooled species diversity of all fields with agri-environment schemes with the pooled species diversity of all conventionally managed fields in a single area that consisted of a mosaic of scheme and conventional fields. The same difficulties apply to the two other common study design, examining changes in biodiversity (26% of the studies) or trends in time in areas with and without schemes (32%). Only 34% of the studies included baseline data, and 16% used a paired study approach to reduce environmental noise (Table 3).

The number of replicates varied from 1 to 398. The number of controls was often similar to the number of replicates but in some cases far larger or smaller (161 controls for 26 experimental replicates and, of greater concern, 2 controls for 82 experimental replicates). Two Swiss studies compared the spatial distribution of birds over the landscape and analysed whether sites with schemes were used by birds more than would be expected based on a random distribution. These studies did not contain formal control areas. The data from 31% of the studies were not analysed statistically. Some reports divided the analysis into a number of groups, such as common vs. Red List plant species. To avoid replication and information overload we selected the measure (usually species richness) that seemed to best represent the results. We checked that this was not distorting the conclusions.

Twenty studies (32%) assessed the effects of schemes on plants, 20 (32%) on various insect groups and spiders, one (2%) on mammals (brown hare Lepus europaeus Pallas) while 29 (47%) studies investigated the response of birds.

results of studies evaluating biodiversity effects of schemes

Our results show that plant diversity may be difficult to enhance with agri-environment schemes (Table 2). Eleven of the 20 studies addressing botanical diversity found positive effects of schemes whereas two studies reported negative effects. Considering the subsample of 14 studies that subjected the data to some form of statistical analysis, six studies demonstrated positive and two studies demonstrated negative effects of schemes, the remaining seven studies finding no effect at all. The poor performance of the evaluated agri-environment schemes with botanical objectives is in accordance with results of experimental studies. These generally show that it is extremely difficult to enhance the botanical diversity of intensively farmed agricultural fields (Berendse et al. 1992), particularly when the period of intensive use has been long enough to deplete the local seed bank (Bekker et al. 1997).

The diversity of arthropods appears to be much easier to enhance through implementation of agri-environment schemes than other groups. Fourteen of 20 studies reported significant increases in the number of species and three reported significant increases for some and decreases for other species in response to agri-environment schemes. Considering only those studies that included statistical tests yielded similar results. Of 17 studies, 11 found positive effects, three both positive and negative effects, and the remaining three studies did not find any effects of schemes. Kleijn et al. (2001) and Kruess & Tscharntke (2002a,b) found no increase in plant species richness, but nevertheless reported significant increases in insect diversity on fields with agri-environment schemes. This positive effect may be due to reduced levels of disturbance on less intensively used fields, allowing organisms to complete their life cycle before the vegetation is removed by mowing or grazing (Kruess & Tscharntke 2002a). As with plants, increased diversity is usually due to more common species. However, Hunziker (2001) and Peter & Walter (2001) found that Ecological Compensation Areas in Switzerland significantly enhanced the number and abundance of endangered grasshopper species. Their studies furthermore indicated the importance of nearby source populations, for instance in nature conservation areas, for achieving positive effects of conservation management on farmland (see also Duelli & Obrist 2003).

The studies investigating the effects of agri-environment schemes on birds show no consistent pattern. Thirteen of 29 studies reported positive effects of agri-environment schemes on bird species richness or abundance, two reported negative effects and nine reported both positive and negative effects. Taking the subsample of 19 studies with statistical tests, only four reported positive effects, two reported negative effects and nine reported both positive and negative effects of agri-environment schemes on birds.

The best known agri-environment scheme success is the cirl bunting Emberiza cirlus (Peach et al. 2001). This species declined massively in abundance and range in the 20th century, so that it became restricted to a small region in Devon and Jersey, UK (Wooton et al. 2000). The species became the target of an intensive research and management programme by the Royal Society for the Protection of Birds, English Nature and the National Trust in the UK. The Country Stewardship Scheme offered a standard payment for maintaining low intensity grassland and, in Devon, a special cirl bunting project was set up to promote weedy spring sown barley stubble in the species range. Between 1992 and 1998, cirl buntings increased by 83% on land entering the Countryside Stewardship Scheme, but only by 2% on adjacent land not in the scheme. The population increased from 118 pairs in 1989 to 450 pairs in 1998.

Similar successful programmes in the United Kingdom for the black grouse Tetrao tetrix (Baines, Warren & Calladine 2002), stone curlew Burhinus oedicnemus (Aebischer, Green & Evans 2000) and corncrake Crex crex (Aebischer et al. 2000) suggest that agri-environment schemes can work well as part of a closely supervised, integrated programme. However, it is unreasonable to extrapolate from such studies to those without intensive support and additional management activities.

Our impression from the literature, discussions with researchers, extension officers and farmers, and from visiting a wide range of schemes is that agri-environment schemes are most effective when they provide the finances that enable farmers or conservationists to carry out measures that they feel positive about. Schemes that are considered financially beneficial but an inconvenience and with little support, feedback, encouragement or inspection are much less likely to provide gains. Thus, we have observed many situations where the land managers care about the outcome and tune their management of the agri-environment scheme to benefit biodiversity. Conversely, we have observed many other situations where an agri-environment scheme is clearly considered a financially beneficial inconvenience and carried out in the minimal manner possible, without regard to the outcome. It would be useful to test whether these impressions are correct.

Discussion

how effective are european agri-environment schemes in protecting biodiversity?

We are unable to say how effective agri-environment schemes are in protecting and promoting biodiversity on farmland. A limited number of well-designed and thoroughly analysed studies demonstrate convincing positive effects measured in terms of increased species diversity or abundance, while other studies show no effects, negative effects, or positive effects on some species and negative effects on others. A number of schemes do not achieve the expected effect or even have negative side-effects. This suggests that the prescribed management may require modification. However, modifications and improvements can only result from a regular evaluation of all agri-environment schemes.

The most striking conclusion from this review is that there is a lack of research examining whether agri-environment schemes are effective. Only the Netherlands and the United Kingdom have made any significant effort to evaluate the effects of agri-environment programmes on biodiversity (Table 2). Nevertheless, in the Netherlands, the usefulness of these studies in evaluating the effectiveness of the main agri-environment scheme is limited. The studies were contracted out to a range of different ecological consulting agencies and the methodology differed between most of the studies, thereby making an integrated analysis impossible (Wymenga, Jalving & Ter Stege 1996). Currently, agri-environment incentive schemes are being initiated in the Central and Eastern European (CEE) countries that will join the EU in 2004, but, as far as we know, no evaluation studies are integrated into these programmes. The implementation of nation-wide schemes, without learning from the mistakes made by their predecessors in other parts of Europe, represents a missed opportunity to make agri-environment programmes as effective as possible from the outset.

This review has revealed a considerable bias towards studies in intensively farmed areas. Uptake of schemes is higher in areas farmed under extensive systems, but we found very few evaluation studies in extensive areas. Currently, biodiversity levels are low in most intensively farmed areas (Kleijn & Van der Voort 1997; Kleijn et al. 2001). Agri-environment schemes targeted at these areas are expected to enhance species diversity over time (Fig. 1b; improvement effects). Generally, biodiversity levels are higher in extensively farmed areas (Wolff et al. 2001; Dullinger et al. in press). Agri-environment schemes are expected to maintain this diversity by protecting areas from intensification or abandonment (Fig. 1b; protection effects). Changes in land-use intensity will have a greater impact on biodiversity in extensively farmed land than on intensively used farmland (Fig. 1b, see also Potter & Goodwin 1998). Agri-environment schemes that aim to protect biodiversity in extensively farmed areas may therefore be more effective than those aiming to improve biodiversity in intensively farmed areas. Most studies detailed in Table 2 are from intensively farmed areas in Western Europe; studies are lacking completely from the Mediterranean countries. It is unlikely that results from the studies carried out so far (Table 2) can be extrapolated at all to southern European countries, hence there is a need for more research in these countries.

problems with the design of evaluation studies

We conclude that the experimental designs of a large proportion of the evaluation studies are weak. The main approach was to compare areas of land under existing agri-environment schemes with control areas not covered by schemes. If sites qualifying for agri-environment schemes are located preferentially in the most diverse areas, comparisons between these and control sites will be biased towards giving favourable results. Bias is unavoidable if study sites in designated areas (for instance Environmentally Sensitive Areas selected on the basis of their conservation interest) are compared with control areas outside designated areas. Furthermore, farmers that volunteer to enter agri-environment schemes may already farm in a more environmentally sensitive manner. In the Netherlands this is compounded when farmers locate schemes on the more inaccessible or marginal fields within a farm (Kleijn et al., in press), and in the UK agri-environment schemes are often located in habitats of greater conservation interest (Carey et al. 2002). Kleijn & Van Zuijlen (in press) reanalysed data of Van Buel (1996) and found significantly higher densities of meadow birds on fields managed under agri-environment schemes relative to conventionally managed fields. They showed that the higher meadow bird densities were primarily due to the higher quality of fields within schemes (higher groundwater table). Between 1989 and 1995 population trends were similar on fields within schemes and on control fields. It was shown that the difference observed by Van Buel (1996) in 1995 was caused primarily by differences in initial site conditions.

Bias can be avoided by randomly assigning half of a subset of farmers that sign up for schemes to the control treatment (continue farming as they had done prior to the scheme) and the other half to the scheme treatment. For small-scale measures, it may be more practical to ask farmers to identify a pair of sites and then allocate one at random to be managed under the agri-environment scheme and the other to be managed conventionally. This would neutralize any influence of farmer or agri-environment scheme officer on the initial quality of scheme and control sites. Baseline data should be collected prior to the start of the scheme and repeated biodiversity surveys should be carried out in subsequent years, or in the final year of the scheme. This would then give a fair estimate of the effects of the scheme. This method has been adopted in the UK study of the environmental consequences of genetically modified (GM) herbicide tolerant crops. Individual fields were divided in two and one half was randomly allocated to the GM treatment while the other half contained the control with conventional crops (Firbank et al. 2003; Perry et al. 2003). Clearly, it is essential that allocation of a site to experimental or control must be random and cannot be influenced by local decisions.

Where the ideal situation is not possible, for example when a scheme that is already in place needs to be evaluated (which will be the rule rather then the exception), the best possible alternatives are (i) to collect baseline data, (ii) to examine trends in time, and (iii) to try to reduce systematic differences in initial conditions between scheme and control sites as far as possible. Great care should be taken to pair scheme areas with nearby control areas that are similar in most respects (e.g. soil type, groundwater table and landscape structure) so that these are eliminated as confounding factors. However, it will remain difficult to interpret positive results with confidence. None of the studies reviewed here met standards set in the previous section and only a few complied with the ‘best alternatives’ described above. For instance, only 36 studies had controls, sufficient replication and rigorous statistical analysis (Table 3). Just 24 studies from five countries additionally made use of baseline data, and/or trend analysis, and/or pairing of control and scheme sites.

costs and benefits of agri-environment schemes

What is the efficiency of agri-environment schemes (the benefits per unit costs)? None of the studies listed in Table 2 addressed this aspect. Up to 2003, €24 300 million has been spent in the 15 EU countries alone (EEA 2002). Most of this money was spent on measures whose objectives were not biodiversity conservation. It is impossible to estimate the amount spent on biodiversity conservation schemes and no estimates of benefits are available either. Hanley, Whitby & Simpson (1999) compared costs and benefits of the United Kindom ESA scheme and found that benefits greatly outweighed the costs. Benefits were estimated by means of the Contingent Valuation Method, a survey-based approach that directly elicits preferences for environmental goods from individuals. However, it is important to point out that ‘respondents were shown pictures of the landscape “with” and “without” the ESA, but were not given any information on the probability of successful restoration of the “with scenario”, nor how long it would take’ (Hanley et al. 1999). Thus, individuals valued agri-environment schemes on the assumption that they would result in increased diversity. Our review shows that this is not always the case. Thus, there is a need for studies that directly link the costs of schemes with their biodiversity benefits. To our knowledge, the only study that has attempted this correlated the amount of agri-environmental subsidy received for the management of grassland fields in Austria with plant species richness of those fields (Zechmeister et al. 2003). They found no positive relationship between amount of subsidy and botanical diversity.

Some agri-environment schemes have other objectives besides biodiversity conservation. The environmental and economic benefits of these other objectives may be substantial, particularly when their impact extends outside the area targeted for the schemes (Daily & Ellison 2002). For example, in Germany the conversion of 6% of permanent grassland to arable land resulted in the release of 10 tonnes nitrogen ha−1 (as NO3) and 100 tonnes of carbon (as CO2) as well as enhanced winter water runoff. This has been suggested as a contribution towards the greater flooding frequencies along the major German rivers (Van der Ploeg, Ehlers & Sieker 1999). Agri-environment schemes that revert arable land to permanent grassland should result in reduced emissions and flooding frequencies in the wider countryside, as well as enhancing biodiversity. In the UK, creating a 80-m wide salt marsh along an eroding coast can result in the annual cost of coastal defence dropping from c.€7200 m−1 to c. €600 m−1 (House of Commons Select Committee on Agriculture 1998). Economic analysis showed that ecological protection and restoration of the Catskill-Delaware watershed was a more cost effective means of protecting the water quality for New York than improving the technology for water treatment (Ashendorff et al. 1997).

synthesis and applications

The outcome of this review does not allow for a general judgement of the effectiveness of agri-environment schemes because of a lack of sufficiently rigorous studies. There is a particularly urgent need for studies evaluating the effects of schemes in extensively farmed areas and in Mediterranean countries. The fact that a number of studies found no change or even negative effects of agri-environment schemes on biodiversity highlights the importance of regular evaluations of all major agri-environment schemes. To do this, it is necessary to formulate clear and unambiguous biodiversity objectives for each scheme, if they have not been formulated already. Furthermore, the design of evaluation studies deserves more attention. Studies should include the collection of baseline data, should incorporate control sites that are similar to scheme sites in every respect but the change in management, and both control sites and scheme sites should be sufficiently replicated. This can be achieved most effectively by making evaluation programmes an integral part of each agri-environment programme. The results of these studies need to be disseminated to the international scientific community, preferably through publication in international peer-reviewed journals, or by making an institution responsible for collating and distributing this type of research. Only then will we be able to (i) find out why some schemes are effective and others not, (ii) determine how existing schemes can be made more effective, and (iii) decide what schemes may be abandoned and how new schemes should be formulated.

Acknowledgements

We thank Jon Marshall for suggesting this review and the following for providing useful information: Jacques Baudry, Harriet Bennett, Nigel Boatman, Anne Bonis, Pierre-Yves Bontemps, Richard Bradbury, Val Brown, Pierre Burghart, Nigel Critchley, Nicola Crockford, Mario Diaz, Jane Feehan, John Feehan, Maeve Flynn, Aldina Franco, Gary Fry, Des Gillmor, Phil Grice, Knut Per Hasund, Johnny Kahlert, Mikko Kuussaari, Anders Larsson, Leonidas Louloudis, Bernhard Osterburg, Jørgen Primdahl, Katrina Rønningen, Norbert Sauberer, Wolfgang Schumacher, Riccardo Simoncini, Franz Sinabell, Patrick Steyaert and Anki Weibull. The constructive comments of three anonymous referees and Gillian Kerby are much appreciated. The work of D.K. was done within the framework of the EU-funded project ‘EASY’ (QLK5-CT-2002–01495).

Ancillary