Journal of Environmental Management

Transcription

1 Journal of Environmental Management 92 (2011) 902e909 Contents lists available at ScienceDirect Journal of Environmental Management journal homepage: Assessing the environmental performance of English arable and livestock holdings using data from the Farm Accountancy Data Network (FADN) D.B. Westbury *, J.R. Park, A.L. Mauchline, R.T. Crane, S.R. Mortimer School of Agriculture, Policy & Development, University of Reading, Earley Gate, Reading RG6 6AR, UK article info abstract Article history: Received 13 January 2010 Received in revised form 23 July 2010 Accepted 21 October 2010 Available online 13 November 2010 Keywords: Agri-environmental footprint index Agri-environment schemes Policy evaluation Environmental assessment Farm business survey Agri-environment schemes (AESs) have been implemented across EU member states in an attempt to reconcile agricultural production methods with protection of the environment and maintenance of the countryside. To determine the extent to which such policy objectives are being fulfilled, participating countries are obliged to monitor and evaluate the environmental, agricultural and socio-economic impacts of their AESs. However, few evaluations measure precise environmental outcomes and critically, there are no agreed methodologies to evaluate the benefits of particular agri-environmental measures, or to track the environmental consequences of changing agricultural practices. In response to these issues, the Agri-Environmental Footprint project developed a common methodology for assessing the environmental impact of European AES. The Agri-Environmental Footprint Index (AFI) is a farm-level, adaptable methodology that aggregates measurements of agri-environmental indicators based on Multi- Criteria Analysis (MCA) techniques. The method was developed specifically to allow assessment of differences in the environmental performance of farms according to participation in agri-environment schemes. The AFI methodology is constructed so that high values represent good environmental performance. This paper explores the use of the AFI methodology in combination with Farm Business Survey data collected in England for the Farm Accountancy Data Network (FADN), to test whether its use could be extended for the routine surveillance of environmental performance of farming systems using established data sources. Overall, the aim was to measure the environmental impact of three different types of agriculture (arable, lowland livestock and upland livestock) in England and to identify differences in AFI due to participation in agri-environment schemes. However, because farm size, farmer age, level of education and region are also likely to influence the environmental performance of a holding, these factors were also considered. Application of the methodology revealed that only arable holdings participating in agri-environment schemes had a greater environmental performance, although responses differed between regions. Of the other explanatory variables explored, the key factors determining the environmental performance for lowland livestock holdings were farm size, farmer age and level of education. In contrast, the AFI value of upland livestock holdings differed only between regions. The paper demonstrates that the AFI methodology can be used readily with English FADN data and therefore has the potential to be applied more widely to similar data sources routinely collected across the EU-27 in a standardised manner. Ó 2010 Elsevier Ltd. All rights reserved. 1. Introduction When the Common Agricultural Policy (CAP) was established in Article 39 of the Treaty of Rome (1957) food security across Europe was paramount and as a consequence, policies focused primarily on the optimum utilisation of the factors of production. However, since * Corresponding author. Tel.: þ ; fax: þ address: d.b.westbury@reading.ac.uk (D.B. Westbury). the 1980s, CAP measures have increasingly supported methods of agricultural production that protect the environment and maintain the countryside. In 1992 this led to Council Regulation (European Economic Community (EEC)) No 2078/92. The Regulation directs that countries should implement schemes for the protection of the European countryside; in particular, agricultural production methods should be compatible with the requirements of the protection of the environment and the maintenance of the countryside (Commission of the European Communities, 1992). Further environmental support was provided by the Agenda 2000 programme, which included /$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi: /j.jenvman

2 D.B. Westbury et al. / Journal of Environmental Management 92 (2011) 902e a priority area to enhance and extend the adoption of agri-environmental measures within the European Union (EU) under the Rural Development Regulation (Council Regulation (EC) 1257/1999). Agrienvironmental measures became the only compulsory component of the Member States rural development programmes submitted to the Commission and by 2009, some 18% of the EU-27 s utilisableagriculture area was managed under agri-environment schemes (AESs) (European Commission, 2010). To determine the extent to which such policy objectives are being fulfilled and to identify changes necessary to bridge the gap between policy aims and outcomes, EU Member States are obliged to monitor and evaluate the environmental, agricultural and socio-economic impacts of their AESs. However, very few evaluations attempt to measure precise environmental outcomes and critically, there are no agreed methodologies to evaluate the benefits of particular agrienvironmental measures, or to track the environmental consequences of changing agricultural practices. Kleijn and Sutherland (2003) identified 62 studies that assessed the impact of European AESs on biodiversity, and for the majority of studies the methodologies used were inadequate to reliably assess the efficacy of AESs in terms of environmental performance. Measures of farmer participation i.e. the number of participating farmers or area of land under agreement, have been widely used to document the extent of progress made towards the achievement of particular policy objectives. However, although this approach is frequently adopted owing to the ease of recording such information, it is evident that participation per se does not guarantee delivery of environmental protection or improvement (Kleijn and Sutherland, 2003). In response to these issues, the Agri-Environmental Footprint project developed a common methodology for assessing the environmental impact of European AES. The Agri-Environmental Footprint Index (AFI) is a farm-level, adaptable approach that aggregates measurements of agri-environmental indicators based on Multi- Criteria Analysis (MCA) techniques (Mortimer et al., 2009, 2010; Purvis et al., 2009). It is a step-wise process based on the adaptation of the basic stages of MCA (see DCLG, 2009) allowing the scoring of farms against weighted indicators. The AFI methodology is constructed so that high values represent good environmental performance. A similar approach has been used previously as a structure for evaluating landscape and habitat enhancement mechanisms (Park et al., 2004). The method was developed specifically to allow assessment of differences in the environmental performance of farms according to AES participation. However, it also has the potential to be used for the wider surveillance of differences in environmental performance with respect to agricultural system and underlying factors such as farm size, farmer age, level of education and region. The AFI methodology has been successfully applied with primary data to investigate the influence of agri-environmental scheme participation on environmental performance (Knickel and Kasperczyk, 2009). An aim of the current paper was therefore to test whether its use could be extended for the routine surveillance of environmental performance of agri-environment schemes using established data sources. This paper explores the use of the AFI methodology in combination with data from the Farm Accountancy Data Network (FADN). FADN was launched in 1965 following the introduction of Council Regulation 79/65, with the overall objective of determining the income of agricultural holdings and the impacts of the CAP for all Member States of the European Union. As such, it is the only source of micro-economic data that is harmonised across the EU. The survey only covers agricultural holdings in the Union that due to their size could be considered commercial. On average, 2000 farm holdings are surveyed annually by each Member State and in the UK the Department for the Environment, Food and Rural Affairs (Defra) assumes responsibility, via the Farm Business Survey (FBS). Overall, the aim of the paper was to measure the environmental impact of three different types of agriculture (arable, lowland livestock and upland livestock) in England and to identify differences in AFI due to participation in agri-environment schemes. Responses were also considered in relation to farm size, farmer age, level of education and Government Office because of their potential to influence environmental performance. 2. Method Data from the FBS (England) were obtained from Defra for three broad farm types derived from FADN farm descriptors: arable, lowland livestock, and upland livestock (Table 1). Individual holdings within each of these three broad types were included in the analyses if consecutively surveyed in 1995, 2000 and 2005, enabling temporal changes to be analysed across a tenyear time series. A farm holding was excluded from analyses if the broad farm type changed between years. Holdings were also omitted if changes in farm size resulted in a shift in size category or changes in farm ownership resulted in a change in farmer age category during the 10 year period. In total, 64 different arable holdings, 43 lowland livestock and 23 upland livestock holdings met the criteria for further analyses. Farm size was based on land area, under three categories: small, medium and large. For arable holdings small farms were classified as being less than 150 ha; medium, 150e300 ha, and large, greater than 300 ha. For lowland and upland livestock holdings, small farms were classified as being less than 80 ha; medium, 80e120 ha, and large, greater than 120 ha. The farmer age category was based on ages in 1995 to produce two groups, those under 50 years old and those aged 50 and over. Farmer education was based on four levels of qualification, 1) school only, 2) GCSE or equivalent (taken by students aged 15e16 years old), 3) A level or equivalent, or college/national Diploma Certificate (typically taken by students aged 17e18 years old), and 4) Degree and/or postgraduate qualification. As FBS data on farmer education was available only for years 2000 and 2005, the level of education attained by 2005 was used for analyses. Holdings were also grouped according to their location within England using Government Office s (hereafter simply referred to as regions). If a region was represented by fewer than four holdings, it was omitted from analyses. Participation in agri-environment schemes was defined as those holdings that were registered under a scheme for at least one of the sample times during the 10-year period. It was therefore not possible to determine whether environmental performance of holdings increased with time under agri-environment scheme participation. However, it was possible to explore whether agri-environment scheme participation in general was associated with an increasing environmental performance with time Assessment Criteria Matrices Following the procedure adopted in the AFI methodology (Purvis et al., 2009), assessment criteria relating to the environmental performance of farms were identified. Two different Assessment Table 1 Broad farm types investigated according to FADN descriptors. Broad farm type Constituent FADN farm types Arable Cereals General cropping Lowland livestock Dairy Grazing livestock Mixed Upland livestock Dairy (less favoured areas) Specialist sheep (LFA - Severely disadvantaged area) Other cattle and sheep (LFA - severely disadvantaged area) Grazing livestock (LFA - disadvantaged area)

3 904 D.B. Westbury et al. / Journal of Environmental Management 92 (2011) 902e909 Criteria Matrices (ACMs) were created for the arable and livestock (both lowland and upland) broad farm types (Tables 2 and 3) and populated using indicators derivable from FBS data. Indicator selection for the ACMs was based on the availability of FBS data combined with knowledge of the environmental impacts of farming practices relevant to each of the three issues: natural resource protection (NR), biodiversity conservation (B), and protection of landscape character (L). Indicators derivable from FBS data for the three years used in this study relate primarily to animal and crop husbandry. In the AFI methodology, indicators relating to management of uncropped areas of the farm, such as hedgerows and semi-natural vegetation, can be included (Purvis et al., 2009). However, this was not possible here due to lack of data. In some cases indicators were derived by approximation, for instance amounts of fertiliser used were derived from information on the total expenditure on fertiliser and dividing this by a standard fertiliser cost (Nix, 1995, 2000, 2005) to estimate an overall quantity used. To avoid bias with respect to farm size, certain indicators were expressed per unit area (hectare), based on the Utilisable Agricultural Area (UAA). All indicators and assessment criteria within the ACM were given equal weighting because in this study the process did not involve stakeholder engagement (see Purvis et al., 2009). In several cases assessment criteria related to complex environmental issues (e.g. intensity of crop husbandry) and as a consequence, multi-metric indicator functions were used to aggregate multiple values relating to various aspects of the indicator. Single values for multi-metric indicators were calculated by taking an average value of its components. was based on Shannon diversity values (Magurran, 1988) determined using areas of woodland, rough grazing, total grassland and total arable. All indicators were scored on a scale of 0e10 where 0 represented the lowest farm score for environmental performance and 10 was the maximum. Consequently the index was constructed in the same way, with high AFI values representing good environmental performance Data analysis Differences in AFI values according to farm size, farmer age, farmer education, region and participation in agri-environment schemes were analysed for each broad farm type using mixed models in SAS (Version 9.1, 2003). The model considered the effects of these factors individually and as interactions; year was also specified as a fixed effect. A full factorial model was not considered as some interactions were not of interest, for example the interaction with farmer age and region. Instead, eleven key interactions were specified: farmer age agri-environment scheme participation, farmer age farm size, region x agri-environment scheme participation, region x farm size, farm size agri-environment scheme participation, farmer education x farmer age, farmer education x farm size, farmer education x agri-environment scheme participation, year region, year farm size and year agri-environment scheme participation. In all instances, model simplification was by deletion of non-significant factors, except where a factor was part of a significant interaction. For all models, holding number and the interaction between holding and year were specified as random effects. Year was also specified as a repeated measure with an autoregressive covariance structure. Degrees of freedom were calculated using the iterative Satterthwaite s method. When a factor was shown to have a significant effect, post-hoc pairwise comparisons (P ¼ 0.05) were made to investigate differences. When a factor or interaction between factors was associated with a significant effect on the AFI value, the underlying responses of the indicators were investigated individually using mixed models in SAS. If an underlying indicator (e.g. fertiliser units) had a significant response to the factor being investigated (e.g. farm size), post-hoc (Tukey) pairwise comparisons (P ¼ 0.05) were made to determine differences. 3. Results 3.1. Agri-environment Scheme participation Agri-environment scheme participation as a factor in the mixed model analyses was included in five of the 11 interactions specified for each farm type in addition to being included separately. It was determined that for lowland and upland livestock holdings agrienvironment scheme participation had no significant influence on AFI values. However, a significant interaction between agri-environment scheme participation and region was found for arable holdings (see Section 3.2.) AFI scores for arable holdings did not differ between regions, although a significant interaction between region and agri-environment scheme participation was found (F 4,134 ¼ 3.6, P < 0.01). Holdings participating in agri-environment schemes in the East and West Midlands were associated with greater AFI values compared with non-participating holdings (Fig. 1). The main driver for the significant interaction between region and agri-environment scheme participation was the percentage of uncropped land, as this was the only significant interaction found for the indicators (F 4,174 ¼ 5.9, P < 0.001). There was a tendency for arable holdings in Table 2 Assessment Criteria Matrix used for arable farms. Environmental issue Assessment criterion Indicator Natural resources protection Protection of groundwater quality - Fertiliser units (tonnes) - Crop protection costs Protection of groundwater quantity - % of UAA that is irrigated Energy consumption - Electricity costs and machinery, heating and vehicle fuels and oil per hectare UAA Biodiversity conservation Intensity of crop husbandry - Fertiliser units (tonnes) - Crop protection costs - Crop diversity (Shannon Diversity) - % of spring crops - (Shannon diversity) Provision of woodland habitats - % of total farm area that is woodland Landscape protection Provision of woodland habitats - % of total farm area that is woodland Evidence of uncultivated land - % of total farm as uncropped land (including fallow and set-aside) - (Shannon diversity)

4 D.B. Westbury et al. / Journal of Environmental Management 92 (2011) 902e Table 3 Assessment criteria matrix used for livestock farms. Environmental issue Assessment criterion Indicator Natural resources protection Protection of groundwater quality - Fertiliser units (tonnes) - Average number of grazing livestock units per hectare of forage Protection of groundwater quantity - Water units per hectare UAA Energy consumption - Electricity costs and machinery, heating and vehicle fuels and oil per hectare UAA Biodiversity conservation Intensity of livestock production - Fertiliser units (tonnes) per hectare of UAA - Average number of grazing livestock units per hectare of forage - Percentage of grassland area that is temporary grassland Provision of semi-natural grassland habitats - Percentage of UAA that is classified as rough grazing - (Shannon Diversity) Provision of woodland habitats - % of total farm area that is woodland Landscape protection Provision of woodland habitats - % of total farm area that is woodland Provision of semi-natural grassland habitats - Percentage of UAA that is classified as rough grazing - (Shannon diversity) the West Midlands to have a much greater percentage of total farm area as uncropped land (including fallow and set-aside) when participating in an agri-environment scheme. In the other regions, differences were less distinct. AFI values for lowland livestock holdings were also influenced by region (F 4,29.1 ¼ 3.0, P < 0.05). Holdings in South West England were associated with a greater AFI value than North West England and the West Midlands (Tukey, P < 0.05). Exploration of the underlying indicators revealed differences according to fertiliser use, number of livestock grazing units, area of rough grassland and land use diversity (Table 4). A significant effect of region was also found for upland livestock holdings (F 3,21 ¼ 4.9, P < 0.01), with greater AFI values associated with farms in the Yorkshire and the Humber region compared with North West England, West Midlands and South West England (Tukey test, P < 0.05). Exploration of the underlying indicators revealed differences according to fertiliser use, the number of livestock grazing units and area of rough grassland (Table 5) Farm size Farm size had no significant effect on AFI values calculated for arable and upland livestock holdings. In contrast, a significant effect was found for lowland livestock holdings (F 2,29.1 ¼ 9.5, P < 0.001), AFI Value Increasing environmental performance Yorkshire & the Humber East Midlands West Midlands Participating Non-participating East of England South East England Fig. 1. Arable AFI values (SE) according to region and agri-environment scheme participation. with a significant increase in environmental performance with farm size (Tukey test, P < 0.05) (Fig. 2). Interrogation of the underlying data revealed that large holdings used significantly less energy (Tukey test, P < 0.05) than small sized farms. The proportion of woodland and values of land use diversity were also significantly greater for large farms compared with small and medium sized farms (P < 0.05), whilst water use was significantly greater on small farms (Table 6) Farmer age and level of education The influence of farmer age or level of education and their interaction had no significant effect on AFI values for arable or upland livestock holdings. In contrast, a significant interaction between farmer age and level of education was found for lowland livestock holdings (F 3,29.1 ¼ 4.4, P < 0.05), indicating that responses between farmer age and level of education were not consistent. As a consequence, the underlying indicators were not investigated individually for farmer age and level of education. Overall, the interaction inferred a tendency for holdings managed by farmers that were over 50 years old and educated to Level 3 or Level 4 to have greater AFI values than holdings managed by similarly educated farmers aged under 50 (Fig. 3). The main drivers for the significant interaction between farmer age and level of educationwere values of rough grassland (F 3,33.8 ¼ 3.9, P < 0.05) expressed as a percentage of UAA, and values of woodland cover (F 3,34.8 ¼ 8.1, P < 0.001) expressed as a percentage of total farm area. No significant effects were determined for the other underlying indicators. Overall, irrespective of the level of education there was a tendency for farmers over 50 years old to have a greater proportion of rough grassland on their holdings, although the opposite was found for holdings managed by farmers under 50 and educated to Level 2. Woodland cover also differed with farmer age and level of education, with a tendency for farmers over 50 years old to have a greater proportion of woodland on their holding, unless educated to Level 1 only Year When year was considered as a single explanatory factor or as an interaction with other factors, no significant effects were found for the three broad farm types investigated. 4. Discussion The application of the AFI methodology to FBS data revealed that, of the explanatory variables explored, the key factors determining the environmental performance of lowland livestock holdings were

5 906 D.B. Westbury et al. / Journal of Environmental Management 92 (2011) 902e909 Table 4 Values (SE) of the lowland livestock indicators according to region. Indicators Fertiliser Units (tonnes) per ha UAA Livestock grazing units per ha forage Energy use e units Water usage e m 3 Rough grassland - % of UAA Temporary grassland - % of total grassland Woodland cover - % of total farm area (Shannon) North West England (n ¼ 24) East midlands West midlands South East England (n ¼ 39) South West England (n ¼ 39) F Value 0.7 (0.1) c 0.5 (0.1) abc 0.3 (0.1) ab 0.3 (0.1) a 0.6 (0.1) bc F 4,37.7 ¼ 2.7* 2.0 (0.1) c 2.0 (0.2) bc 2.2 (0.2) c 1.2 (0.1) a 1.5 (0.1) b F 4,38.0 ¼ 8.5*** 98.8 (7.7) 78.7 (8.6) 78.0 (9.7) 73.2 (15.3) 79.1 (7.3) ns 39.0 (5.3) 24.6 (3.2) 20.2 (3.6) 26.1 (4.8) 22.6 (3.6) ns 5.9 (1.9) b 0.0 (0.0) a 0.0 (0.0) a 0.0 (0.0) a 4.9 (2.0) ab F 4,38.0 ¼ 2.8* 31.3 (7.5) 26.1 (7.4) 23.9 (5.9) 22.7 (5.7) 31.6 (3.5) ns 0.1 (0.1) 0.0 (0.0) 0.0 (0.0) 1.5 (0.8) 2.1 (0.4) ns 0.21 (0.05) a 0.45 (0.07) ab 0.30 (0.07) a 0.35 (0.05) a 0.62 (0.04) b F 4,38.0 ¼ 3.1* Values with the same superscript in each row do not differ significantly (P > 0.05). ns ¼ P > 0.05; * ¼ P < 0.05; ** ¼ P < 0.01; *** ¼ P < farm size, farmer age, region and level of education. In contrast, AFI values for arable holdings were strongly influenced by region and agri-environment scheme participation, whilst upland livestock holdings were influenced by region only. The fact that year had no significant influence indicates that AFI scores remained relatively constant during the 10-year period that was analysed. A limitation of using FBS data to specifically evaluate the efficacy of agri-environment schemes was the absence of information on farm environmental features (e.g. the presence of hedgerows). Such data were collected from 2006, but since 2008, much has been omitted, only to be collected on an ad hoc basis. In the absence of such data, the AFI methodology can still be used to detect broad environmental benefits, but the value of the AFI approach would be enhanced if a range of additional environmental variables were collected via FADN. The AFI methodology would also benefit from the inclusion of additional non-environmental data, for example information on the intensity of arable cultivation (i.e. minimum tillage versus ploughing) and whether grassland was managed for silage or cut for hay. Specific estimates of fertiliser and pesticide use per hectare rather than deriving from cost estimates would also improve precision. Table 5 Values (SE) of the upland livestock indicators according to region. Indicators Fertiliser units (tonnes) Livestock grazing units per ha forage Energy use e units Water usage e m 3 Rough grassland - % of UAA Temporary grassland - % of total grassland Woodland cover - % of total farm area (Shannon) North West England Yorkshire & the Humber (n ¼ 15) West Midlands (n ¼ 24) South West England F Value 0.43 (0.05) b 0.15 (0.04) a 0.34 (0.05) b 0.44 (0.06) b F 3,20.5 ¼ 3.6* 1.3 (0.1) b 0.6 (0.1) a 1.3 (0.1) b 1.1 (0.1) b F 3,21.4 ¼ 5.9** 74.8 (8.0) 36.4 (7.4) 76.5 (10.6) 60.2 (7.2) ns 4.6 (1.9) 4.2 (1.1) 12.0 (4.3) 2.3 (0.9) ns 16.1 (4.8) a 55.0 (5.1) b 13.2 (5.4) a 11.9 (3.0) a F 3,19.4 ¼ 5.1** 5.4 (2.7) 1.8 (1.3) 7.3 (4.8) 12.2 (2.9) ns 1.2 (0.5) 2.3 (1.0) 1.1 (0.5) 1.0 (0.5) ns 0.37 (0.08) 0.69 (0.04) 0.29 (0.05) 0.37 (0.08) ns Values with the same superscript in each row do not differ significantly (P > 0.05). ns ¼ P > 0.05; * ¼ P < 0.05; ** ¼ P < 0.01; *** ¼ P <