- Email: [email protected]
An approach based on spatial multicriteria analysis to map the nature conservation value of agricultural land
ARTICLE IN PRESS
Journal of Environmental Management 83 (2007) 228–235 www.elsevier.com/locate/jenvman
An approach based on spatial multicriteria analysis to map the nature conservation value of agricultural land Davide Geneletti Department of Civil and Environmental Engineering, University of Trento, Via Mesiano, 77 38050 Trento Italy Received 14 April 2005; received in revised form 14 March 2006; accepted 15 March 2006 Available online 27 June 2006
Abstract Knowledge of the nature conservation value of agricultural land provides a useful input to land-use planning. However, the scarcity of suitable data causes this component to rarely play a role. The paper proposes a methodology based on commonly available data to assess the nature conservation value of agricultural landscapes, and to generate cartographic results to be used as decision variables in planning. The approach relies on landscape ecological indicators and on the application of multicriteria analysis in a Geographical Information System (GIS) context. Four criteria were selected: the agricultural landscape type, the cover of vegetation remnants and marginal features, the length of forest–agriculture ecotones, and the proximity to nature reserves. These criteria were assessed directly or by means of specific indicators, generating maps that were subsequently aggregated through spatial multicriteria analysis. The approach was tested in an alpine area located in Trentino (northern Italy). r 2006 Elsevier Ltd. All rights reserved. Keywords: Spatial indicators; Landscape ecology; GIS; Alpine areas
1. Introduction European rural landscapes and their biodiversity are currently threatened by the intensification of farming, as well as by the marginalisation and abandonment of traditional land uses due to economic forces (CEC, 2000). For this reason, the European Union (EU) has recently promoted the Agricultural Action Plan on Biodiversity, as part of the activities needed to fulfil its commitments under the Convention on Biological Diversity (Hoffmann, 2000). The Plan aims at enhancing the potential role of rural areas for biodiversity protection and nature conservation. In order to achieve the objective of the Plan, it is fundamental to assess and map the nature conservation value of rural areas (Bu¨chs, 2003). On top of this, a spatially explicit mapping of the ecological relevance of the countryside can contribute to put into practice the multifunctional model of agriculture. Agriculture is multifunctional when it has one or several roles in addition to its primary role of producing food and Tel.: +39 0461 882685; fax: +39 0461 882672.
E-mail address: [email protected]. 0301-4797/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2006.03.002
fibre (OECD, 1998). These additional functions might include rural viability, cultural heritage, sanitary health, and nature conservation. The idea that agriculture is capable of delivering multiple benefits is not particularly novel, but it is still poorly conceptualised and not always consistently realised by EU policies (Potter and Burney, 2002). Most efforts in linking agriculture and biodiversity protection are being directed toward the development of evaluation schemes oriented to provide either broad-scale overviews, i.e. with a spatial scale of 1:250,000 or coarser (EEA, 2001; OECD, 2002; Wascher, 2000), or farmlevel assessments, i.e. with a scale on the order of 1:1000 (MacNaeidhe and Culleton, 2000; Stobbelaar and Mansvelt, 2000). Less attention is being paid to intermediatescale analyses (1:10,000–1:25,000), which are most suited for local and regional planning. The situation is worsened by the common lack of data addressing the ecological relevance of rural land at a suitable scale. The present paper aims at bridging this gap by proposing a methodology to assess the nature conservation value of agricultural landscapes, and to generate cartographic results that can be used as decision variables in land-use
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
planning. This will overcome a traditional shortcoming in spatial planning: considering agricultural land only as a productive unit or as a cultural landscape component, and not as an ecosystem (Diamantini and Zanon, 2000). The approach relies on landscape ecological indicators and on the application of multicriteria analysis in a Geographical Information System (GIS) context (Geneletti, 2005, 2004; Malczewski, 1999). Quality elements within the agricultural landscape, such as vegetation remnants and ecotones, were first identified and assessed, and then aggregated into synthetic value maps. This was mainly performed through aerial photos interpretation and GIS operations. The choice of limiting the analysis to these elements was suggested by the common lack of further ecological data that affect rural areas. Typically, data on biodiversity distribution or on farming pressures (e.g., use of fertilisers) are aggregated for whole physical or administrative units (watersheds, communities, etc.), and therefore they are not available with the level of detail required by planning procedures (CEC, 1999). The approach was tested in an alpine area located in northern Italy: the Avisio River basin, which lies in the north-eastern part of the Trentino region. Fig. 1 shows the location of Trentino in Italy and a simplified land-cover map of the Avisio basin. Within the basin, forests and shrubs are the predominant land covers, especially where geomorphology severely constrains the use of land. The basin covers about 1000 km2 and has an elevation range of over 2000 m. Agricultural land is mostly found between 500 and 1300 m, within the valley floors and the most favourably oriented slopes. The Avisio basin was selected because it features a wide range of agricultural practices: from the intensive farming of the valley floors and the favourably oriented hills, to the traditional forms of mountain agriculture of the higher slopes.
2. Methods In multicriteria analysis, a criterion can be defined as a standard of judging, i.e., a way to express the degree of achievement of an objective. Its evaluation can be supported by resorting to indicators, i.e. to measurable parameters. Four criteria were selected to assess the nature conservation value of agricultural areas: the agricultural landscape type, the cover of vegetation remnants and marginal features, the length of open area-forest ecotones, and the proximity to nature reserves. These criteria were assessed directly or by means of specific indicators, generating maps that were subsequently aggregated through multicriteria analysis. A cell size of 100 m was used in the analysis, as proposed in Osinski (2003). This size was selected after a set of tests conducted using cell sizes ranging from 250 to 25 m. The smallest fields in the study region have an area of about 1 ha. This is because the 100-m size proved to be the most effective in terms of capturing the diversity of the agricultural landscape, and avoiding redundancy in the data analysis. As described in the following sub-sections, all input data have a spatial resolution higher than the selected cell size. All GIS operations were conducted using ILWIS version 3.2 (ITC, 2001). 2.1. Agricultural landscape type The first criterion relates to the farming practice and the agricultural landscape type that characterise each elementary unit. The intensity of farming and the methods of production influence the spatial heterogeneity and composition of farmland, the use of chemicals, the presence of disturbance activities (ploughing, mowing, weeding, etc.), and therefore they are deeply related to the ecological relevance of rural areas (Stobbelaar and Mansvelt, 2000; MacNaeidhe and Culleton, 2000). In CEC (2000), the following classification of agricultural landscape types according to the intensity of farming practice was proposed:
Fig. 1. Location of the study region and simplified land-cover map of the Avisio basin.
229
Type 1: landscapes characterised by overexploitation, pollution and resource depletion; Type 2: landscapes characterised by intensive or extensive good farming practices in a balanced relationship with the land; Type 3: landscapes characterised by low-input farming, low pollution and resource depletion; Type 4: farming-dependent landscapes where agriculture has a particular role in creating environmental quality.
This classification is rather general and it obviously does not account for the specific characteristics of each field. However, it represents a guideline to distinguish main agricultural types, and it is suitable to rapid appraisal approaches, due to the limited information requirement.
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
230
Table 1 Grouping of agricultural land covers (classified according to EEA, 2000) into landscape types Landscape Landscape Landscape Landscape
type type type type
1 2 3 4
Arable land; vineyards; orchards Pastures Annual crops associated with permanent crops; complex cultivation patterns Land principally occupied by agriculture with significant areas of natural vegetation; agro-forestry areas.
Furthermore, having being proposed at a European level, it allows to generate results that can be replicated and compared within different study areas and contexts. Following a proposal by Lazzerini (2001), the agricultural land covers present in the study area were grouped into the four classes above, as shown in Table 1. The landcover data used in this analysis were extracted by the official land-cover map of the Province of Trento, which was derived from aerial surveys and field validations carried out at a of 1:10,000 scale (PAT, 2003). Data aggregation to a 100-m cell size was performed using the dominant class method. 2.2. Cover of vegetation remnants and marginal features In addition to the agricultural landscape type, the nature conservation relevance of farmland depends on the extent to which management practices retain non-farmed marginal features, such as hedges and trees, which provide crucial habitat for wildlife (OECD, 2001). Even intensively farmed land can be important for biodiversity where hedges are maintained. Woodlots, hedgerows, field verges, remnant vegetation along streams and canals constitute essential elements of the landscape’s biodiversity, playing an important role for the dispersal of species and colonisation of semi-natural habitats (Marshall, 2002; Le Coeur et al., 2002). The conservation of these structural elements is threatened by land consolidation, increase in average field size, and mechanisation. For this reason, promoting and maintaining hedges and trees, as well as the natural and semi-natural elements created by agricultural practice, is explicitly mentioned among the objectives of the EU ‘‘Strategy on the environment integration and sustainable development in common agriculture policy’’ (Agriculture Council, 1999). The vegetation remnants and the marginal features within rural areas were mapped using colour aerial photos acquired in July 2000 with a spatial resolution of 1 m. The photos, already ortho-corrected and mosaiced, were made available by the Autonomous Province of Trento. The high spatial resolution allowed to identify linear features (hedgerow, stream vegetation), as well as small woodlots and isolated trees. First, all non-agricultural areas were masked out of the photo mosaic. A training set of spectral signatures was then collected for the different cultivation types and for the forest class. Subsequently, a supervised classification based on a likelihood algorithm was per-
formed using ILWIS 3.2. The classification results were improved through post-classification operations (e.g., spatial filtering), as well as visual interpretation. The latter was especially required within the shadowed areas of the images. The percentage cover of vegetation remnants and marginal features within each 100-m elementary cell was selected as an indicator and computed throughout the agricultural landscape. In landscape ecology, variables related to the spatial patterning of natural patches are frequently used to assess nature conservation value. For this reason, few of the most popular indices (maximum patch size, number of patches, adjacency) were computed to measure the spatial distribution of the remnant vegetation within each agricultural cell (Giles and Trani, 1999). The resulting maps were tested for correlation using the product moment correlation coefficient (see details in Section 2.5). All the indices were strongly correlated to the total cover of vegetation remnants, and therefore they were not included in the analysis.
2.3. Length of forest–agriculture ecotones An ecotone is a zone of transition between different ecosystems. Ecotones are generally species rich and characterised by properties that do not exist in either of the adjacent ecosystems (Odum, 1993). Ecotones between open areas and forest are particularly relevant as habitat for species and for the ecological processes they host. In the study region, this type of ecotones have been decreasing in the last decades, due to the abandonment of traditional agriculture practices, and the subsequent bush encroachment into pastures and open areas. This emerged as a deep concern in the recently drawn project for the sustainable development of the region (Diamantini and Zanon, 2000). In this project, one of the key indicators used to monitor biodiversity depletion is the abundance of Alpine rock partridge (Alectoris graeca saxatilis), which proved to be positively correlated to the presence of forest–agriculture ecotones (De Marchi and Amato, 2005). For these reasons, it was decided to use the length of open area/forest ecotones as a criterion to assess the ecological relevance of farmland. Ecotones between agricultural areas and forests were extracted from the l and-cover map through GIS operations. First of all, the land-cover map was classified into two broad classes: agriculture and forest. All other cover types were masked out of the map. Then, an edge-enhancement spatial filter
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
231
Fig. 2. Detail of the ecotone map (left) and the ecotone length map (right).
(Richards, 1993) was run to detect the boundaries between agricultural and forest areas. Finally, these boundaries were automatically extracted and mapped in a separate layer (see Fig. 2). Only transition areas connected to forest patches with a size of at least 5 ha were considered, so as to exclude the ecotones originated by small woodlots and tree rows. This is because the presence of these elements was accounted for by the second criterion (see Section 2.2). The 5-ha threshold was selected after a process of trial and error and proved to be effective in the light of the average size of the landscape elements within the study area. The total length of ecotones present within each elementary cell of the agricultural landscape was selected as an indicator and computed, generating the map shown in Fig. 2. 2.4. Proximity to nature reserves The last criterion relates to the proximity of agricultural land to sites designated for nature conservation, as in particular the Nature 2000 sites, established by the EU Directive 92/43 (Habitat Directive). This criterion accounts for the buffering action of rural areas that can shield nature reserves from harmful activities and disturbances. Especially in the case of small reserves, buffer zones serve a number of ecological functions, such as ameliorating edge effects, filtering out chemicals and noise, reducing weed invasion and providing connectivity between natural areas (Noss and Cooperrider, 1994). In order to maintain efficient buffering action, agricultural areas need to be properly managed. Conversely, they can actually induce additional stress and disturbance on protected areas (e.g., pollutant run-off). Therefore, the buffer action is to be interpreted as a potential role of rural areas located around nature conservation sites. Nevertheless, this potential role is quite relevant in terms of orienting land-use decisions. For this reason, this criterion was included in the evaluation scheme, and assessed by using as an indicator the distance of rural areas from the nearest protected area. Several Nature 2000 sites, among which is a portion of a larger regional park, are present within the study area. Distance operators in a GIS were applied to assign to each
elementary agricultural cell a value corresponding to the distance from the closest conservation site. 2.5. Multicriteria analysis In order to generate a map of the nature conservation value of rural areas, the four criteria have to be aggregated through multicriteria analysis. Multicriteria analysis requires the use of criteria that are independent from each other. Correlated criteria introduce redundancy and double counting, and generate inconsistent results. For this reason, prior to their aggregation, the four criterion maps were tested for correlation. For simplicity, only linear correlation was considered in the test. Correlation analysis aims at understanding to what extent the patterns of pairs of maps are spatially associated. If the spatial correlation between two maps is significantly greater than might occur due to chance, then the two maps are not independent. When maps are measured using continuous interval or ratio scale variables, their linear correlation can be expressed by the product moment correlation coefficient (Bonham-Carter, 1994): n P
ðxi xÞðy ¯ i y¯ Þ r ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , n n P P ðxi xÞ ¯ 2 ðyi y¯ Þ2 i¼1
i¼1
(1)
i¼1
where x and y are the values of the two maps, x¯ and y¯ are their respective means and i are the map cells. The coefficient varies between 1 (perfect correlation) through 0 (no correlation or independence) to 1 (perfect negative correlation). The results of the correlation analysis did not show significant correlations between the criteria, as presented in Table 2. Following the typical steps of multicriteria analysis (Geneletti, 2005, 2004), the four criteria were normalised, prioritised, and then aggregated. Through normalisation, criterion scores lose their dimension and become an expression of the degree of achievement of the evaluation objective. A conventional value range between zero (minimum desirability) and one (maximum desirability) was adopted. The criteria based on measurable indicators were normalised through piece-wise linear functions, as
ARTICLE IN PRESS 232
D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
shown in Table 3. Given that no evidence was available to allow elicitation of curves, linear relationships between the measured criterion scores and their values were assumed. The presence of ecotones was considered optimal for length over 100 m, corresponding to one side of the elementary cell (i.e., to sharing at least one boundary with forest). As to the presence of vegetation remnants, the optimal value was assigned to cover over 20%. The distance from nature conservation areas was considered optimal if below 100 m and not relevant if over 500 m. The agricultural landscape type is a class criterion that was not measured through indicators. Its normalisation was performed using equal intervals, and assigning the maximum value to landscape Type 4, as shown in Table 3. Prioritisation was used to express the relative importance of the different criteria. The agricultural landscape type was considered more important than the presence of vegetation remnants and ecotones, which in turn were considered more important than the proximity to conservation sites (see last column in Table 3). This is to stress the influence on the nature conservation value of rural areas determined by farming practices, and consequently resource consumption. On the other hand, the proximity to protected areas is less relevant because it refers to a potential role played by rural land, rather than to an actual one.
Table 2 Correlation coefficients calculated between the four criteria used in the analysis Criterion 1
Criterion 2
r
Agricultural type Agricultural type Agricultural type Ecotones Ecotones Vegetation remnants
Ecotones Vegetation remnants Prox. to nature reserves Vegetation remnants Prox. to nature reserves Prox. to nature reserves
0.116 0.017 0.001 0.234 0.027 0.001
A weighted summation of the normalised criterion layers was performed, according to the formula: V¼
4 X
ai wi ,
(2)
i¼1
where V is the nature conservation value of a given cell, a the normalised criterion score, w the criterion weight, and i the evaluation criteria. The resulting map of the nature conservation value of agricultural areas is shown in Fig. 3, together with an enlarged sub-window to better appreciate the spatial distribution of values. In order to understand how geomorphologic patterns influence the index values, the correlation between nature conservation value and topography was studied. This analysis aimed also at verifying if topographic variables could be used to meaningfully represent ecological values. Using a 1:10,000 digital elevation model, three topographical variables were mapped: elevation, slope (in percentages) and aspect (in radians). The correlation between nature conservation values and topography was performed for the whole study area applying formula (1). The analysis was then repeated for three sub-regions, corresponding to the main valleys that form the Avisio basin (see Fig. 3). The results are presented in Table 4. No significant correlation between the distribution of nature conservation values and topography was found. Finally, sensitivity analysis was performed to test the robustness of the results with respect to changes in the selected weights and normalisation functions. Weights were changed by 720% and the break points of the linear functions shifted by 720%. As a result, new classifications were obtained and compared with the original one. In these new classifications, significant value changes (i.e., greater than 20% of the original value) occurred in less than 9% of the elementary cells. Sensitive cells were randomly distributed in the study region. Even in this case, no correlation between cell sensitivity and topography was found.
Table 3 Normalisation and weighting of the four criteria Criterion
Indicator (unit)
Normalisation
Weight
Agricultural type
—
0.40
Vegetation remnants and marginal features
Total cover (%)
Ecotones
Total length (m)
Proximity to nature reserves
Distance from closest reserve (m)
Type 1: 0.25 Type 2: 0.5 Type 3: 0.75 Type 4: 1 X20%: 1 0%: 0 In between: linear variation X100 m: 1 0 m: 0 In between: linear variation p100 m: 1 X500 m: 0 In between: linear variation
0.25
0.25
0.10
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
233
Fig. 3. Nature conservation value of agricultural areas in the Avisio basin.
Table 4 Correlation coefficients calculated between nature conservation values and topographical variables Cembra valley
Fiemme valley
Fassa valley
0.040 0.093 0.045
0.028 0.049 0.097
0.032 0.124 0.027
0.114 0.141 0.126
High
3000
Medium
2500
Cells
Elevation Slope Aspect
Avisio basin
3500
Low
2000 1500 1000 500 0
3. Results The map of Fig. 3 shows that rural areas with a low nature conservation value are found especially in the southwestern tip of the Avisio basin. This is the most urbanised part of the study area and it is characterised by the degradation of the landscape and the loss of naturalness. In this area, the only ecologically valuable farmland is found where agricultural practices allowed the conservation of marginal features. In order to gain a concise understanding of the results, nature conservation values were aggregated into three classes: low (below 0.58), medium (0.58–0.76), and high (over 0.76). The class thresholds were selected by subdividing the range of values that occur in the study region into equal intervals. The classification was then performed separately for the three main sub-regions that form the Avisio basin: the Cembra, Fiemme and Fassa valleys. The results are presented in Fig. 4 and aim at facilitating the use
Avisio basin (939 Km2)
Cembra valley Fiemme valley (176 Km2) (483 Km2)
Fassa valley (280 Km2)
Fig. 4. Classified nature conservation values for the Avisio basin and its sub-regions.
of the nature conservation value map to orient land managers. In the Cembra and Fassa valleys, only about 10% of agricultural land is classified as highly valuable for nature conservation. In the Cembra area, this is due to the predominance of agricultural landscapes of type 1 (see Table 1), whereas in the Fassa area the scores are influenced by the lack of marginal features and by the remoteness of nature reserves. In both valleys, ecologically valuable farmland is mainly found within the less accessible slopes, where agricultural and mountain ecosystems interact. The Fiemme valley shows a far better performance, and over 30% of rural areas have a high nature conservation value. The agricultural landscape is very similar to the one
ARTICLE IN PRESS 234
D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
of the Fassa valley, but there is a stronger presence of vegetation remnants, hedgerows, and ecotones.
a proposed project), and during the selection of the most suitable location for new projects. Acknowledgement
4. Discussion and conclusion The nature conservation relevance of agricultural land is seldom used as a decision variable in land-use planning. Farmlands are generally viewed solely for production purposes and not as agro-ecosystems that provide ecological services. Moreover, agro-biodiversity data are scarce and most evaluation schemes are not tailored to generate output at a suitable spatial scale. This paper presents an approach aimed at providing an operational appraisal of the nature conservation relevance of rural areas. The selected evaluation criteria and indicators are intended to serve as surrogates for more detailed biodiversity data and offer the advantage of being easy to assess and applicable at a defined spatial scale. For this reason, the indicators were largely based on information extracted from aerial photographs, rather than field survey. On the other hand, the methodology is limited by the lack of detailed information on the intensity in the management of each field: farming practices were grouped into broad classes, regardless of parameters such as fertiliser input, pesticides used, crop rotation, size of fields, etc. Analogously, the quality of vegetation remnants (e.g. naturalness, age of trees, etc.) was not assessed, and all remnant habitats were considered as equally relevant. The extension of buffer zones needed to shield nature reserves is also a variable that could be studied more in detail, as it depends on the habitat types. However, habitat mapping (despite the impulse provided by the EU Habitat Directive), as well as data on field management, hedgerow species, and marginal features, are still largely unavailable. In the light of these constraints, the resulting map provided a useful input to land-use planning, allowing to understand what rural areas need to be protected from new developments (e.g., urban encroachment) in order to conserve natural resources. Being based on explicitly formulated criteria and indicators, the evaluation scheme is open to comments and improvements. In particular, the approach will benefit from in-depth data collections targeted on the biotic and abiotic conditions (bird atlas, plant species inventories, etc.). Moreover, the value assessment performed by experts (i.e., the selection of weights and normalisation functions) can be refined by enlarging the panel to include further knowledge and perspectives. The results generated by this study are currently being used by the technical offices of the local administration to support the procedure of Environmental Impact Assessment (EIA). In particular, the map of the nature conservation value of rural areas represents one of the thematic layers that support decision-making during the screening phase (i.e., to decide whether EIA is required for
The comments of three anonymous reviewers contributed to the quality of this paper. References Agriculture Council, 1999. Council strategy on the environment integration and sustainable development in common agriculture policy established by the Agricultural Council. Council of the European Union, 17 November 1999, Brussels. Bonham-Carter, G.F., 1994. Geographic Information Systems for Geoscientist. Modelling with GIS, Pergamon, Ontario. Bu¨chs, W., 2003. Biotic indicators for biodiversity and sustainable agriculture—introduction and back ground. Agriculture, Ecosystems and Environment 98, 1–16. Commission of the European Community (CEC), 1999. Communication from the Commission to the Council; the European Parliament; the economic and social committee and the committee of the regions, Directions towards sustainable agriculture. Comm (1999) 22 final, Brussels. Commission of the European Community (CEC), 2000. Communication from the Commission to the Council and the European Parliament. Indicators for the Integration of environmental concerns into the Common Agricultural Policy. Comm (2000) 20 final, Brussels. De Marchi, M., Amato, S., 2005. La libellula della sostenibilita`: indicatori per la gestione sostenibile della biodiversita`. In: Diamantini, C. (Ed.), Temi e indicatori di sostenibilita` ambientale in una regione alpina. Temi, Trento. Diamantini, C., Zanon, B., 2000. Planning the urban sustainable development. The case of the plan for the Province of Trento. Environmental impact assessment review 20 (3), 299–310. European Environmental Agency (EEA), 2000. CORINE Land cover technical guide-addendum 2000. Technical Report No. 40, Copenhagen. European Environmental Agency (EEA), 2001. Toward agri-environmental indicators. Integrating statistical and administrative data with land cover information. Topic Report No. 6, Copenhagen. Geneletti, D., 2004. A GIS-based decision support system to identify nature conservation priorities in an alpine valley. Land Use Policy 21, 149–160. Geneletti, D., 2005. Multicriteria analysis to compare the impact of alternative road corridors. A case study in northern Italy. Impact Assessment and Project Appraisal 22 (4), 135–146. Giles, R.H., Trani, M.K., 1999. Key elements of landscape pattern measures. Environmental Management 23 (4), 477–481. Hoffmann, L.B. (Ed), 2000. Stimulating positive linkages between agriculture and biodiversity. Recommendations for the EC-Agricultural Action Plan on Biodiversity. European Centre for Nature Conservation, Tilburg. ITC, 2001. ILWIS 3.0 Academic—User’s Guide. IT Department, ITC, Enschede. Lazzerini, G., 2001. Gli ecosistemi agricoli. In: Diamantini, C. (Ed.), Progetto per lo sviluppo sostenibile del Trentino. Department of Civil and Environmental Engineering, University of Trento. Le Coeur, D., Baudry, J., Burel, F., Thenail, C., 2002. Why and how we should study field boundary biodiversity in an agrarian landscape context. Agriculture, Ecosystems and Environment 89, 23–40. MacNaeidhe, F.S., Culleton, N., 2000. The application of parameters designed to measure nature conservation and landscape development on Irish farms. Agriculture, Ecosystems and Environment 77, 65–78. Malczewski, J., 1999. GIS and multicriteria decision analysis. Wiley, New York.
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235 Marshall, E.J.P., 2002. Introducing field margin ecology in Europe. Agriculture, Ecosystems and Environment 89, 1–4. Noss, R.F., Cooperrider, A.Y., 1994. Saving nature’s legacy. Protecting and restoring biodiversity. Island Press, Washington, DC. Odum, E.P., 1993. Ecology and Our Endangered Life Support Systems. Sinauer, Sunderland. Organisation for Economic Co-operation and Development (OECD), 1998. Agriculture and the environment: issues and policies. Directorate for Food, Agriculture and Fisheries, Paris. Organisation for Economic Co-operation and Development (OECD), 2001. Environmental Indicators for Agriculture. Methods and Results, vol. 3, Paris. Organisation for Economic Co-operation and Development (OECD), 2002. The development of agri-environmental indicators in the EU: The Irena Project. STD/NA/AGR(2002)16, Paris. Osinski, E., 2003. Operationalisation of landscape-oriented indicator. Agriculture, Ecosystems and Environment 98, 371–386.
235
Potter, C., Burney, J., 2002. Agricultural multifunctionality in the WTO— legitimate non-trade concern or disguised protectionism? Journal of Rural Studies 18, 35–47. Provincia Autonoma di Trento (PAT), 2003. Carta dell’uso del suolo reale—Aggiornamento. Servizio Urbanistica, Trento. Richards, J.A., 1993. Remote Sensing Digital Image Analysis. An Introduction. Springer, Berlin. Stobbelaar, D.J., van Mansvelt, J.D., 2000. The process of landscape evaluation Introduction to the 2nd special AGEE issue if the concerted action: ‘‘the landscape and nature production capacity of organic/ sustainable types of agriculture’’. Agriculture, Ecosystems and Environment 77, 1–15. Wascher, D.M., 2000. ELISA—environmental indicators for sustainable agriculture. Final Project Report. European Centre for Nature Conservation, Tilburg.
Journal of Environmental Management 83 (2007) 228–235 www.elsevier.com/locate/jenvman
An approach based on spatial multicriteria analysis to map the nature conservation value of agricultural land Davide Geneletti Department of Civil and Environmental Engineering, University of Trento, Via Mesiano, 77 38050 Trento Italy Received 14 April 2005; received in revised form 14 March 2006; accepted 15 March 2006 Available online 27 June 2006
Abstract Knowledge of the nature conservation value of agricultural land provides a useful input to land-use planning. However, the scarcity of suitable data causes this component to rarely play a role. The paper proposes a methodology based on commonly available data to assess the nature conservation value of agricultural landscapes, and to generate cartographic results to be used as decision variables in planning. The approach relies on landscape ecological indicators and on the application of multicriteria analysis in a Geographical Information System (GIS) context. Four criteria were selected: the agricultural landscape type, the cover of vegetation remnants and marginal features, the length of forest–agriculture ecotones, and the proximity to nature reserves. These criteria were assessed directly or by means of specific indicators, generating maps that were subsequently aggregated through spatial multicriteria analysis. The approach was tested in an alpine area located in Trentino (northern Italy). r 2006 Elsevier Ltd. All rights reserved. Keywords: Spatial indicators; Landscape ecology; GIS; Alpine areas
1. Introduction European rural landscapes and their biodiversity are currently threatened by the intensification of farming, as well as by the marginalisation and abandonment of traditional land uses due to economic forces (CEC, 2000). For this reason, the European Union (EU) has recently promoted the Agricultural Action Plan on Biodiversity, as part of the activities needed to fulfil its commitments under the Convention on Biological Diversity (Hoffmann, 2000). The Plan aims at enhancing the potential role of rural areas for biodiversity protection and nature conservation. In order to achieve the objective of the Plan, it is fundamental to assess and map the nature conservation value of rural areas (Bu¨chs, 2003). On top of this, a spatially explicit mapping of the ecological relevance of the countryside can contribute to put into practice the multifunctional model of agriculture. Agriculture is multifunctional when it has one or several roles in addition to its primary role of producing food and Tel.: +39 0461 882685; fax: +39 0461 882672.
E-mail address: [email protected]. 0301-4797/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2006.03.002
fibre (OECD, 1998). These additional functions might include rural viability, cultural heritage, sanitary health, and nature conservation. The idea that agriculture is capable of delivering multiple benefits is not particularly novel, but it is still poorly conceptualised and not always consistently realised by EU policies (Potter and Burney, 2002). Most efforts in linking agriculture and biodiversity protection are being directed toward the development of evaluation schemes oriented to provide either broad-scale overviews, i.e. with a spatial scale of 1:250,000 or coarser (EEA, 2001; OECD, 2002; Wascher, 2000), or farmlevel assessments, i.e. with a scale on the order of 1:1000 (MacNaeidhe and Culleton, 2000; Stobbelaar and Mansvelt, 2000). Less attention is being paid to intermediatescale analyses (1:10,000–1:25,000), which are most suited for local and regional planning. The situation is worsened by the common lack of data addressing the ecological relevance of rural land at a suitable scale. The present paper aims at bridging this gap by proposing a methodology to assess the nature conservation value of agricultural landscapes, and to generate cartographic results that can be used as decision variables in land-use
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
planning. This will overcome a traditional shortcoming in spatial planning: considering agricultural land only as a productive unit or as a cultural landscape component, and not as an ecosystem (Diamantini and Zanon, 2000). The approach relies on landscape ecological indicators and on the application of multicriteria analysis in a Geographical Information System (GIS) context (Geneletti, 2005, 2004; Malczewski, 1999). Quality elements within the agricultural landscape, such as vegetation remnants and ecotones, were first identified and assessed, and then aggregated into synthetic value maps. This was mainly performed through aerial photos interpretation and GIS operations. The choice of limiting the analysis to these elements was suggested by the common lack of further ecological data that affect rural areas. Typically, data on biodiversity distribution or on farming pressures (e.g., use of fertilisers) are aggregated for whole physical or administrative units (watersheds, communities, etc.), and therefore they are not available with the level of detail required by planning procedures (CEC, 1999). The approach was tested in an alpine area located in northern Italy: the Avisio River basin, which lies in the north-eastern part of the Trentino region. Fig. 1 shows the location of Trentino in Italy and a simplified land-cover map of the Avisio basin. Within the basin, forests and shrubs are the predominant land covers, especially where geomorphology severely constrains the use of land. The basin covers about 1000 km2 and has an elevation range of over 2000 m. Agricultural land is mostly found between 500 and 1300 m, within the valley floors and the most favourably oriented slopes. The Avisio basin was selected because it features a wide range of agricultural practices: from the intensive farming of the valley floors and the favourably oriented hills, to the traditional forms of mountain agriculture of the higher slopes.
2. Methods In multicriteria analysis, a criterion can be defined as a standard of judging, i.e., a way to express the degree of achievement of an objective. Its evaluation can be supported by resorting to indicators, i.e. to measurable parameters. Four criteria were selected to assess the nature conservation value of agricultural areas: the agricultural landscape type, the cover of vegetation remnants and marginal features, the length of open area-forest ecotones, and the proximity to nature reserves. These criteria were assessed directly or by means of specific indicators, generating maps that were subsequently aggregated through multicriteria analysis. A cell size of 100 m was used in the analysis, as proposed in Osinski (2003). This size was selected after a set of tests conducted using cell sizes ranging from 250 to 25 m. The smallest fields in the study region have an area of about 1 ha. This is because the 100-m size proved to be the most effective in terms of capturing the diversity of the agricultural landscape, and avoiding redundancy in the data analysis. As described in the following sub-sections, all input data have a spatial resolution higher than the selected cell size. All GIS operations were conducted using ILWIS version 3.2 (ITC, 2001). 2.1. Agricultural landscape type The first criterion relates to the farming practice and the agricultural landscape type that characterise each elementary unit. The intensity of farming and the methods of production influence the spatial heterogeneity and composition of farmland, the use of chemicals, the presence of disturbance activities (ploughing, mowing, weeding, etc.), and therefore they are deeply related to the ecological relevance of rural areas (Stobbelaar and Mansvelt, 2000; MacNaeidhe and Culleton, 2000). In CEC (2000), the following classification of agricultural landscape types according to the intensity of farming practice was proposed:
Fig. 1. Location of the study region and simplified land-cover map of the Avisio basin.
229
Type 1: landscapes characterised by overexploitation, pollution and resource depletion; Type 2: landscapes characterised by intensive or extensive good farming practices in a balanced relationship with the land; Type 3: landscapes characterised by low-input farming, low pollution and resource depletion; Type 4: farming-dependent landscapes where agriculture has a particular role in creating environmental quality.
This classification is rather general and it obviously does not account for the specific characteristics of each field. However, it represents a guideline to distinguish main agricultural types, and it is suitable to rapid appraisal approaches, due to the limited information requirement.
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
230
Table 1 Grouping of agricultural land covers (classified according to EEA, 2000) into landscape types Landscape Landscape Landscape Landscape
type type type type
1 2 3 4
Arable land; vineyards; orchards Pastures Annual crops associated with permanent crops; complex cultivation patterns Land principally occupied by agriculture with significant areas of natural vegetation; agro-forestry areas.
Furthermore, having being proposed at a European level, it allows to generate results that can be replicated and compared within different study areas and contexts. Following a proposal by Lazzerini (2001), the agricultural land covers present in the study area were grouped into the four classes above, as shown in Table 1. The landcover data used in this analysis were extracted by the official land-cover map of the Province of Trento, which was derived from aerial surveys and field validations carried out at a of 1:10,000 scale (PAT, 2003). Data aggregation to a 100-m cell size was performed using the dominant class method. 2.2. Cover of vegetation remnants and marginal features In addition to the agricultural landscape type, the nature conservation relevance of farmland depends on the extent to which management practices retain non-farmed marginal features, such as hedges and trees, which provide crucial habitat for wildlife (OECD, 2001). Even intensively farmed land can be important for biodiversity where hedges are maintained. Woodlots, hedgerows, field verges, remnant vegetation along streams and canals constitute essential elements of the landscape’s biodiversity, playing an important role for the dispersal of species and colonisation of semi-natural habitats (Marshall, 2002; Le Coeur et al., 2002). The conservation of these structural elements is threatened by land consolidation, increase in average field size, and mechanisation. For this reason, promoting and maintaining hedges and trees, as well as the natural and semi-natural elements created by agricultural practice, is explicitly mentioned among the objectives of the EU ‘‘Strategy on the environment integration and sustainable development in common agriculture policy’’ (Agriculture Council, 1999). The vegetation remnants and the marginal features within rural areas were mapped using colour aerial photos acquired in July 2000 with a spatial resolution of 1 m. The photos, already ortho-corrected and mosaiced, were made available by the Autonomous Province of Trento. The high spatial resolution allowed to identify linear features (hedgerow, stream vegetation), as well as small woodlots and isolated trees. First, all non-agricultural areas were masked out of the photo mosaic. A training set of spectral signatures was then collected for the different cultivation types and for the forest class. Subsequently, a supervised classification based on a likelihood algorithm was per-
formed using ILWIS 3.2. The classification results were improved through post-classification operations (e.g., spatial filtering), as well as visual interpretation. The latter was especially required within the shadowed areas of the images. The percentage cover of vegetation remnants and marginal features within each 100-m elementary cell was selected as an indicator and computed throughout the agricultural landscape. In landscape ecology, variables related to the spatial patterning of natural patches are frequently used to assess nature conservation value. For this reason, few of the most popular indices (maximum patch size, number of patches, adjacency) were computed to measure the spatial distribution of the remnant vegetation within each agricultural cell (Giles and Trani, 1999). The resulting maps were tested for correlation using the product moment correlation coefficient (see details in Section 2.5). All the indices were strongly correlated to the total cover of vegetation remnants, and therefore they were not included in the analysis.
2.3. Length of forest–agriculture ecotones An ecotone is a zone of transition between different ecosystems. Ecotones are generally species rich and characterised by properties that do not exist in either of the adjacent ecosystems (Odum, 1993). Ecotones between open areas and forest are particularly relevant as habitat for species and for the ecological processes they host. In the study region, this type of ecotones have been decreasing in the last decades, due to the abandonment of traditional agriculture practices, and the subsequent bush encroachment into pastures and open areas. This emerged as a deep concern in the recently drawn project for the sustainable development of the region (Diamantini and Zanon, 2000). In this project, one of the key indicators used to monitor biodiversity depletion is the abundance of Alpine rock partridge (Alectoris graeca saxatilis), which proved to be positively correlated to the presence of forest–agriculture ecotones (De Marchi and Amato, 2005). For these reasons, it was decided to use the length of open area/forest ecotones as a criterion to assess the ecological relevance of farmland. Ecotones between agricultural areas and forests were extracted from the l and-cover map through GIS operations. First of all, the land-cover map was classified into two broad classes: agriculture and forest. All other cover types were masked out of the map. Then, an edge-enhancement spatial filter
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
231
Fig. 2. Detail of the ecotone map (left) and the ecotone length map (right).
(Richards, 1993) was run to detect the boundaries between agricultural and forest areas. Finally, these boundaries were automatically extracted and mapped in a separate layer (see Fig. 2). Only transition areas connected to forest patches with a size of at least 5 ha were considered, so as to exclude the ecotones originated by small woodlots and tree rows. This is because the presence of these elements was accounted for by the second criterion (see Section 2.2). The 5-ha threshold was selected after a process of trial and error and proved to be effective in the light of the average size of the landscape elements within the study area. The total length of ecotones present within each elementary cell of the agricultural landscape was selected as an indicator and computed, generating the map shown in Fig. 2. 2.4. Proximity to nature reserves The last criterion relates to the proximity of agricultural land to sites designated for nature conservation, as in particular the Nature 2000 sites, established by the EU Directive 92/43 (Habitat Directive). This criterion accounts for the buffering action of rural areas that can shield nature reserves from harmful activities and disturbances. Especially in the case of small reserves, buffer zones serve a number of ecological functions, such as ameliorating edge effects, filtering out chemicals and noise, reducing weed invasion and providing connectivity between natural areas (Noss and Cooperrider, 1994). In order to maintain efficient buffering action, agricultural areas need to be properly managed. Conversely, they can actually induce additional stress and disturbance on protected areas (e.g., pollutant run-off). Therefore, the buffer action is to be interpreted as a potential role of rural areas located around nature conservation sites. Nevertheless, this potential role is quite relevant in terms of orienting land-use decisions. For this reason, this criterion was included in the evaluation scheme, and assessed by using as an indicator the distance of rural areas from the nearest protected area. Several Nature 2000 sites, among which is a portion of a larger regional park, are present within the study area. Distance operators in a GIS were applied to assign to each
elementary agricultural cell a value corresponding to the distance from the closest conservation site. 2.5. Multicriteria analysis In order to generate a map of the nature conservation value of rural areas, the four criteria have to be aggregated through multicriteria analysis. Multicriteria analysis requires the use of criteria that are independent from each other. Correlated criteria introduce redundancy and double counting, and generate inconsistent results. For this reason, prior to their aggregation, the four criterion maps were tested for correlation. For simplicity, only linear correlation was considered in the test. Correlation analysis aims at understanding to what extent the patterns of pairs of maps are spatially associated. If the spatial correlation between two maps is significantly greater than might occur due to chance, then the two maps are not independent. When maps are measured using continuous interval or ratio scale variables, their linear correlation can be expressed by the product moment correlation coefficient (Bonham-Carter, 1994): n P
ðxi xÞðy ¯ i y¯ Þ r ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , n n P P ðxi xÞ ¯ 2 ðyi y¯ Þ2 i¼1
i¼1
(1)
i¼1
where x and y are the values of the two maps, x¯ and y¯ are their respective means and i are the map cells. The coefficient varies between 1 (perfect correlation) through 0 (no correlation or independence) to 1 (perfect negative correlation). The results of the correlation analysis did not show significant correlations between the criteria, as presented in Table 2. Following the typical steps of multicriteria analysis (Geneletti, 2005, 2004), the four criteria were normalised, prioritised, and then aggregated. Through normalisation, criterion scores lose their dimension and become an expression of the degree of achievement of the evaluation objective. A conventional value range between zero (minimum desirability) and one (maximum desirability) was adopted. The criteria based on measurable indicators were normalised through piece-wise linear functions, as
ARTICLE IN PRESS 232
D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
shown in Table 3. Given that no evidence was available to allow elicitation of curves, linear relationships between the measured criterion scores and their values were assumed. The presence of ecotones was considered optimal for length over 100 m, corresponding to one side of the elementary cell (i.e., to sharing at least one boundary with forest). As to the presence of vegetation remnants, the optimal value was assigned to cover over 20%. The distance from nature conservation areas was considered optimal if below 100 m and not relevant if over 500 m. The agricultural landscape type is a class criterion that was not measured through indicators. Its normalisation was performed using equal intervals, and assigning the maximum value to landscape Type 4, as shown in Table 3. Prioritisation was used to express the relative importance of the different criteria. The agricultural landscape type was considered more important than the presence of vegetation remnants and ecotones, which in turn were considered more important than the proximity to conservation sites (see last column in Table 3). This is to stress the influence on the nature conservation value of rural areas determined by farming practices, and consequently resource consumption. On the other hand, the proximity to protected areas is less relevant because it refers to a potential role played by rural land, rather than to an actual one.
Table 2 Correlation coefficients calculated between the four criteria used in the analysis Criterion 1
Criterion 2
r
Agricultural type Agricultural type Agricultural type Ecotones Ecotones Vegetation remnants
Ecotones Vegetation remnants Prox. to nature reserves Vegetation remnants Prox. to nature reserves Prox. to nature reserves
0.116 0.017 0.001 0.234 0.027 0.001
A weighted summation of the normalised criterion layers was performed, according to the formula: V¼
4 X
ai wi ,
(2)
i¼1
where V is the nature conservation value of a given cell, a the normalised criterion score, w the criterion weight, and i the evaluation criteria. The resulting map of the nature conservation value of agricultural areas is shown in Fig. 3, together with an enlarged sub-window to better appreciate the spatial distribution of values. In order to understand how geomorphologic patterns influence the index values, the correlation between nature conservation value and topography was studied. This analysis aimed also at verifying if topographic variables could be used to meaningfully represent ecological values. Using a 1:10,000 digital elevation model, three topographical variables were mapped: elevation, slope (in percentages) and aspect (in radians). The correlation between nature conservation values and topography was performed for the whole study area applying formula (1). The analysis was then repeated for three sub-regions, corresponding to the main valleys that form the Avisio basin (see Fig. 3). The results are presented in Table 4. No significant correlation between the distribution of nature conservation values and topography was found. Finally, sensitivity analysis was performed to test the robustness of the results with respect to changes in the selected weights and normalisation functions. Weights were changed by 720% and the break points of the linear functions shifted by 720%. As a result, new classifications were obtained and compared with the original one. In these new classifications, significant value changes (i.e., greater than 20% of the original value) occurred in less than 9% of the elementary cells. Sensitive cells were randomly distributed in the study region. Even in this case, no correlation between cell sensitivity and topography was found.
Table 3 Normalisation and weighting of the four criteria Criterion
Indicator (unit)
Normalisation
Weight
Agricultural type
—
0.40
Vegetation remnants and marginal features
Total cover (%)
Ecotones
Total length (m)
Proximity to nature reserves
Distance from closest reserve (m)
Type 1: 0.25 Type 2: 0.5 Type 3: 0.75 Type 4: 1 X20%: 1 0%: 0 In between: linear variation X100 m: 1 0 m: 0 In between: linear variation p100 m: 1 X500 m: 0 In between: linear variation
0.25
0.25
0.10
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
233
Fig. 3. Nature conservation value of agricultural areas in the Avisio basin.
Table 4 Correlation coefficients calculated between nature conservation values and topographical variables Cembra valley
Fiemme valley
Fassa valley
0.040 0.093 0.045
0.028 0.049 0.097
0.032 0.124 0.027
0.114 0.141 0.126
High
3000
Medium
2500
Cells
Elevation Slope Aspect
Avisio basin
3500
Low
2000 1500 1000 500 0
3. Results The map of Fig. 3 shows that rural areas with a low nature conservation value are found especially in the southwestern tip of the Avisio basin. This is the most urbanised part of the study area and it is characterised by the degradation of the landscape and the loss of naturalness. In this area, the only ecologically valuable farmland is found where agricultural practices allowed the conservation of marginal features. In order to gain a concise understanding of the results, nature conservation values were aggregated into three classes: low (below 0.58), medium (0.58–0.76), and high (over 0.76). The class thresholds were selected by subdividing the range of values that occur in the study region into equal intervals. The classification was then performed separately for the three main sub-regions that form the Avisio basin: the Cembra, Fiemme and Fassa valleys. The results are presented in Fig. 4 and aim at facilitating the use
Avisio basin (939 Km2)
Cembra valley Fiemme valley (176 Km2) (483 Km2)
Fassa valley (280 Km2)
Fig. 4. Classified nature conservation values for the Avisio basin and its sub-regions.
of the nature conservation value map to orient land managers. In the Cembra and Fassa valleys, only about 10% of agricultural land is classified as highly valuable for nature conservation. In the Cembra area, this is due to the predominance of agricultural landscapes of type 1 (see Table 1), whereas in the Fassa area the scores are influenced by the lack of marginal features and by the remoteness of nature reserves. In both valleys, ecologically valuable farmland is mainly found within the less accessible slopes, where agricultural and mountain ecosystems interact. The Fiemme valley shows a far better performance, and over 30% of rural areas have a high nature conservation value. The agricultural landscape is very similar to the one
ARTICLE IN PRESS 234
D. Geneletti / Journal of Environmental Management 83 (2007) 228–235
of the Fassa valley, but there is a stronger presence of vegetation remnants, hedgerows, and ecotones.
a proposed project), and during the selection of the most suitable location for new projects. Acknowledgement
4. Discussion and conclusion The nature conservation relevance of agricultural land is seldom used as a decision variable in land-use planning. Farmlands are generally viewed solely for production purposes and not as agro-ecosystems that provide ecological services. Moreover, agro-biodiversity data are scarce and most evaluation schemes are not tailored to generate output at a suitable spatial scale. This paper presents an approach aimed at providing an operational appraisal of the nature conservation relevance of rural areas. The selected evaluation criteria and indicators are intended to serve as surrogates for more detailed biodiversity data and offer the advantage of being easy to assess and applicable at a defined spatial scale. For this reason, the indicators were largely based on information extracted from aerial photographs, rather than field survey. On the other hand, the methodology is limited by the lack of detailed information on the intensity in the management of each field: farming practices were grouped into broad classes, regardless of parameters such as fertiliser input, pesticides used, crop rotation, size of fields, etc. Analogously, the quality of vegetation remnants (e.g. naturalness, age of trees, etc.) was not assessed, and all remnant habitats were considered as equally relevant. The extension of buffer zones needed to shield nature reserves is also a variable that could be studied more in detail, as it depends on the habitat types. However, habitat mapping (despite the impulse provided by the EU Habitat Directive), as well as data on field management, hedgerow species, and marginal features, are still largely unavailable. In the light of these constraints, the resulting map provided a useful input to land-use planning, allowing to understand what rural areas need to be protected from new developments (e.g., urban encroachment) in order to conserve natural resources. Being based on explicitly formulated criteria and indicators, the evaluation scheme is open to comments and improvements. In particular, the approach will benefit from in-depth data collections targeted on the biotic and abiotic conditions (bird atlas, plant species inventories, etc.). Moreover, the value assessment performed by experts (i.e., the selection of weights and normalisation functions) can be refined by enlarging the panel to include further knowledge and perspectives. The results generated by this study are currently being used by the technical offices of the local administration to support the procedure of Environmental Impact Assessment (EIA). In particular, the map of the nature conservation value of rural areas represents one of the thematic layers that support decision-making during the screening phase (i.e., to decide whether EIA is required for
The comments of three anonymous reviewers contributed to the quality of this paper. References Agriculture Council, 1999. Council strategy on the environment integration and sustainable development in common agriculture policy established by the Agricultural Council. Council of the European Union, 17 November 1999, Brussels. Bonham-Carter, G.F., 1994. Geographic Information Systems for Geoscientist. Modelling with GIS, Pergamon, Ontario. Bu¨chs, W., 2003. Biotic indicators for biodiversity and sustainable agriculture—introduction and back ground. Agriculture, Ecosystems and Environment 98, 1–16. Commission of the European Community (CEC), 1999. Communication from the Commission to the Council; the European Parliament; the economic and social committee and the committee of the regions, Directions towards sustainable agriculture. Comm (1999) 22 final, Brussels. Commission of the European Community (CEC), 2000. Communication from the Commission to the Council and the European Parliament. Indicators for the Integration of environmental concerns into the Common Agricultural Policy. Comm (2000) 20 final, Brussels. De Marchi, M., Amato, S., 2005. La libellula della sostenibilita`: indicatori per la gestione sostenibile della biodiversita`. In: Diamantini, C. (Ed.), Temi e indicatori di sostenibilita` ambientale in una regione alpina. Temi, Trento. Diamantini, C., Zanon, B., 2000. Planning the urban sustainable development. The case of the plan for the Province of Trento. Environmental impact assessment review 20 (3), 299–310. European Environmental Agency (EEA), 2000. CORINE Land cover technical guide-addendum 2000. Technical Report No. 40, Copenhagen. European Environmental Agency (EEA), 2001. Toward agri-environmental indicators. Integrating statistical and administrative data with land cover information. Topic Report No. 6, Copenhagen. Geneletti, D., 2004. A GIS-based decision support system to identify nature conservation priorities in an alpine valley. Land Use Policy 21, 149–160. Geneletti, D., 2005. Multicriteria analysis to compare the impact of alternative road corridors. A case study in northern Italy. Impact Assessment and Project Appraisal 22 (4), 135–146. Giles, R.H., Trani, M.K., 1999. Key elements of landscape pattern measures. Environmental Management 23 (4), 477–481. Hoffmann, L.B. (Ed), 2000. Stimulating positive linkages between agriculture and biodiversity. Recommendations for the EC-Agricultural Action Plan on Biodiversity. European Centre for Nature Conservation, Tilburg. ITC, 2001. ILWIS 3.0 Academic—User’s Guide. IT Department, ITC, Enschede. Lazzerini, G., 2001. Gli ecosistemi agricoli. In: Diamantini, C. (Ed.), Progetto per lo sviluppo sostenibile del Trentino. Department of Civil and Environmental Engineering, University of Trento. Le Coeur, D., Baudry, J., Burel, F., Thenail, C., 2002. Why and how we should study field boundary biodiversity in an agrarian landscape context. Agriculture, Ecosystems and Environment 89, 23–40. MacNaeidhe, F.S., Culleton, N., 2000. The application of parameters designed to measure nature conservation and landscape development on Irish farms. Agriculture, Ecosystems and Environment 77, 65–78. Malczewski, J., 1999. GIS and multicriteria decision analysis. Wiley, New York.
ARTICLE IN PRESS D. Geneletti / Journal of Environmental Management 83 (2007) 228–235 Marshall, E.J.P., 2002. Introducing field margin ecology in Europe. Agriculture, Ecosystems and Environment 89, 1–4. Noss, R.F., Cooperrider, A.Y., 1994. Saving nature’s legacy. Protecting and restoring biodiversity. Island Press, Washington, DC. Odum, E.P., 1993. Ecology and Our Endangered Life Support Systems. Sinauer, Sunderland. Organisation for Economic Co-operation and Development (OECD), 1998. Agriculture and the environment: issues and policies. Directorate for Food, Agriculture and Fisheries, Paris. Organisation for Economic Co-operation and Development (OECD), 2001. Environmental Indicators for Agriculture. Methods and Results, vol. 3, Paris. Organisation for Economic Co-operation and Development (OECD), 2002. The development of agri-environmental indicators in the EU: The Irena Project. STD/NA/AGR(2002)16, Paris. Osinski, E., 2003. Operationalisation of landscape-oriented indicator. Agriculture, Ecosystems and Environment 98, 371–386.
235
Potter, C., Burney, J., 2002. Agricultural multifunctionality in the WTO— legitimate non-trade concern or disguised protectionism? Journal of Rural Studies 18, 35–47. Provincia Autonoma di Trento (PAT), 2003. Carta dell’uso del suolo reale—Aggiornamento. Servizio Urbanistica, Trento. Richards, J.A., 1993. Remote Sensing Digital Image Analysis. An Introduction. Springer, Berlin. Stobbelaar, D.J., van Mansvelt, J.D., 2000. The process of landscape evaluation Introduction to the 2nd special AGEE issue if the concerted action: ‘‘the landscape and nature production capacity of organic/ sustainable types of agriculture’’. Agriculture, Ecosystems and Environment 77, 1–15. Wascher, D.M., 2000. ELISA—environmental indicators for sustainable agriculture. Final Project Report. European Centre for Nature Conservation, Tilburg.