Elsevier

Land Use Policy

Volume 83, April 2019, Pages 488-504
Land Use Policy

Combining LULC data and agricultural statistics for A better identification and mapping of High nature value farmland: A case study in the veneto Plain, Italy

https://doi.org/10.1016/j.landusepol.2019.02.034Get rights and content

Abstract

High Nature Value farmland (HNVf) are characterised by high naturalness of land cover and low intensity of agricultural practices. They are essential for biodiversity conservation in rural environments, and their presence is an important indicator of the effectiveness of different EU policies that aim to support biodiversity in agricultural areas. Consequently, their identification, protection and implementation is of strategic importance. Previous studies have analysed the landscape along an urban-rural-natural gradient, beginning with Land Use and Land Cover (LULC) maps. Building up from this baseline, this paper integrates spatial analysis methods with statistical data related to agricultural practices and their intensity, with the aim of mapping and assessing HNVf in a portion of the Veneto Plain, north-east Italy. In particular, this paper presents a methodology for the identification of HNVf applied to two datasets: (i) the first encompassing only LULC data and (ii) the second encompassing also statistical data on agricultural practices. The aim is to demonstrate how this additional information improves the identification of HNVf.

In the first step, a Kernel Density Estimation (KDE) technique is applied to a reclassified LULC map, in order to calculate continuous intensity indicators. A Principal Components Analysis and an ISODATA Cluster Analysis are then performed respectively to remove redundant information and to identify the different landscape structures of the study area. The second analysis follows the same steps, with the difference that LULC intensity indicators are analysed in combination with data on crop rotations, irrigation and livestock from a census survey.

The first analysis returns a map of the landscape driven only by different intensities of land use. The second returns a map where the statistics on agricultural practices allow for a better characterisation of the natural value of the landscape. Agricultural statistics improved the results, since they allow the discrimination of lower intensity clusters within the cultivated areas, which are traditionally excluded from HNVf by considering only the land cover. The comparison between the results of the two analyses shows that the combined use of the agricultural statistics determines a more detailed representation of the study area, that allows a better differentiation of the agricultural areas between HNVf and non-HNVf, leading to an improvement of the HNVf identification methodology. The benefit of using additional information can be therefore of interest for territorial planning, with the ultimate aim of promoting biodiversity conservation.

Introduction

Loss of biodiversity, due to different natural and anthropic impacts, is a global trend that also affects agroecosystems. The decrease of biodiversity in rural environments is mainly due to habitat alteration and loss, linked to the intensification of agricultural practices and to the abandonment of isolated and unfavourable rural areas (Bignal and McCracken, 2000; Fondazione Lombardia per l’ambiente, 2013). These processes affect also High Nature Value farmlands (HNVf) (Plieninger and Bieling, 2013).

HNVf are rural areas characterised by high proportion of natural vegetation and low intensity farming systems. These features have shaped landscapes that are able to support a highly habitat and species biodiversity (Bernues et al., 2016; Bignal and McCracken, 2000; Plieninger and Bieling, 2013).

HNVf are therefore essential for biodiversity conservation in rural environments. Their presence is an important indicator of the effectiveness of different European Union (EU) policies that aim to support biodiversity in agricultural areas, such as the third target of EU Biodiversity Strategy to 2020 (European Commission, 2011) and the Rural Development Programmes (Servadei, 2015) of the Common Agricultural Policy (CAP). For this reason, their identification, protection and implementation is of strategic importance.

In line with these EU directives, this study was conducted on the central part of the Veneto Plain, north-east Italy, with the aim of testing and further developing recently proposed methods for identifying and mapping the HNVf in support of rural planning.

One of the first studies that tried to identify HNVf was carried out by Andersen et al. (2003). They proposed two main approaches: (i) a method based on the CORINE Land Cover (CLC) database; and (ii) a method based on agronomic and economic farm data, retrieved from the Farm Accountancy Data Network (FADN). The first approach identified land cover classes where HNVf may be expected, the list of which changes depending on the environmental zone within Europe. This discrimination is important because the different types of HNVf have different characteristics depending on the country where they are located and on the different situations in terms of climate and elevation. The first approach, also defined as Land Cover Approach (LCA), is useful for identifying the potential location of HNVf, but it cannot infer information about the intensity of the farming systems or management practices occurring in those areas. The second approach proposed by Andersen et al. (2003) is a Farming Systems Approach (FSA). It gives a general indication of the presence and characteristics of farming systems that are more likely to be HNVf, by analysing the intensity of the farming practices, (i.e. farm area, production method, livestock, input levels). The FSA can be considered as a complementary approach to the Land Cover one. In fact, as specified by Cooper et al. (2007): “it is the combination of suitable land covers and features, the ‘state’, with appropriate management, the ‘driving force’, that creates the conditions for a farming system to be HNVf”. The FADN dataset is useful for this purpose because it contains a broad set of data that enables links to environmental aspects, it contains data at the individual farm level, and is updated regularly. However, the FADN dataset does not allow a sufficiently precise localisation of HNVf because data is made available only for large scale units and because the FADN sample does not represent the totality of the farms or an adequate statistical coverage. FADN, in fact, does not include in the census those farms that are too small in economic terms, which are at the same time extremely important for HNVf areas, being mostly under extensive management.

To increase accuracy, the HNVf approach proposed by Andersen et al. (2003) by means of a Land Cover Approach was updated and refined by Paracchini et al. (2006,2008). They used updated land cover data, additional expert rules (altitude, soil quality, steepness of slope) and additional biodiversity datasets (Natura 2000 network, IBAs, PBAs, National biodiversity dataset). Similarly to Andersen et al. (2003), they proposed a list of CLC classes where HNVf may be expected.

Pointereau et al. (2007) tested a new FSA on a national scale based on the Farming Structure Survey (FSS) conducted by Eurostat. In contrast to the FADN dataset, FSS provides data on farming systems and practices for each farm, which can be used as indicators of the intensity of the agriculture. Furthermore, it provides the possibility of characterising HNVf with a better resolution than the one provided by FADN. The importance of using a higher resolution is largely suggested by Carvalho-Santos et al. (2010) and Lomba et al. (2014), because different agro-ecological processes operate at different scales. A method similar to Pointereau et al. (2007) was used by Samoy et al. (2007), Almeida and Pinto-Correia (2012) and Sutkowska et al. (2013), who applied it at a regional scale.

Recently Kikas et al. (2018), Lomba et al. (2017), Sagris et al. (2016) and Strohbach et al. (2015) underlined the potentials of using Integrate Administrative and Control System (IACS) data for HNVf identification and mapping. IACS is a database created to manage and control payments to farmers made by the EU Member States in application of the CAP. It contains farm-level (e.g. number of animals, farm type) and parcel level information (e.g. land use, size) and it is connected to a system for the identification of all agricultural parcels in Member States, named Land Parcel Information System (LPIS). It provides, in this way, high resolution, annually-updated farm and parcel level information that are available across EU Member States. A database based on farmers’ applications for CAP payments was used also by Ribeiro et al. (2014) and Ribeiro et al. (2016) to analyse patterns of farming systems and dynamics of areas already designated as HNVf. In order to protect the farmers privacy, these data are characterized by a highly restricted access that make them of difficult use.

With the aim of further increasing the accuracy of HNVf mapping by other authors, in recent regional scale studies, the land cover and farming systems approaches were combined with habitats and species datasets, that give additional information on biodiversity (Boyle et al., 2015; Lomba et al., 2015; Matin et al., 2016; O’Rourke et al., 2016; Overmars et al., 2014; Sullivan et al., 2011).

The first efforts to assess HNVf on the Italian national level were carried out by Trisorio (2006) and Povellato and Trisorio (2007): the approach proposed by Andersen et al. (2003) was adapted to the Italian context in terms of CLC classes, and overlaid with a national biodiversity dataset related to the richness of species of vertebrates (National Ecologic Network). A farming systems approach was proposed for Italy by Trisorio and Borlizzi (2011) and Trisorio et al. (2012): the intensity of farming was described by individual farm indicators collected from the FSS dataset, such as absence of irrigation, minimum or no-tillage, crop rotation, green manure, grass covering and livestock density. An economic and structural specification of HNV farms has also been undertaken, based on the Italian FADN dataset. A combined approach for HNVf mapping was recently proposed on a regional scale in Italy by Lazzerini et al. (2015), who has used the CLC dataset together with the ISTAT survey on farming systems, particularly the data on extensive breeding, permanent grassland and nitrogen surplus. This overview of publications focused on High Nature Value farmland (HNVf) identification and mapping is reported in Table 1.

These approaches follow a conventional definition of the landscape, usually considered in landscape ecology as an arrangement of relatively homogeneous CLC patches, which are repeated across the space (Bridges et al., 2007; Forman, 1995). This model does not accurately represent the real landscape, which is not made of clearly distinct uniform patches. Rather, it is usually characterized by gradual changes from one type of land use patches to another that creates “transitional” environments (McGarigal and Cushman, 2005). For this reason, new views suggest that the landscape can be described as a gradient across a vast spatial scope.

Gradients are formed as a result of the variability of environmental features from one site to another (Bridges et al., 2007) and the degree of the environmental changes in space determines the steepness of the gradient in the system’s structure and function (McDonnell and Pickett, 1990). The increasing anthropic intensity of Land Use/Land Cover (LULC) produces by itself an anthropogenic gradient characterised by the succession in space of natural-managed-cultivated-suburban-urban landscapes (Forman and Godron, 1986). These urban-rural gradients are particularly suitable for the description of spatial land use patterns, ecosystem structures and functions in urban-rural regions (Kroll et al., 2012; Luck and Wu, 2002). HNVf can be considered as areas characterised by an interaction of multiple features, such as agricultural fields and natural and semi-natural elements; thus, a landscape model based upon gradients permits not only a more effective representation of the territory, but also a better identification of the HNVf, because this interaction of features are captured by changes of the gradients.

Building upon the existing literature briefly introduced above and summarized in Table 1, this study aims to identify and map the HNVf within a study area characterised by scattered urbanisation patterns and intensive agriculture, intersected by natural environments of high naturalistic interest for biodiversity conservation (rivers, volcanic hills, etc.). Moreover, it aims to show that the use of statistical data on agricultural practices in combination with LULC data allows for a better classification of HNVf in comparison with the sole use of LULC data. We also aimed to use only open source data, as they allow easy reproducibility of the analyses. These aims have been pursued by:

  • Performing two spatial analyses of the urban-rural gradient: (i) based only on LULC data, which is related to the naturalness of the study area, and (ii) based on both LULC data and census data, which in addition include information on the intensity of agricultural practices, refining the mapping obtained by (i);

  • Identifying, for each of these two analyses, landscape typologies potentially belonging to HNVf, with HNVf characterised by high naturalness in the first analysis, while in the second one, they are characterised by both high naturalness and low intensity of agricultural practices;

  • Comparing the results of the two analyses, to demonstrate whether or not additional information on agricultural practices permits a better characterization of HNVf.

Section 2 presents a detailed account of the methodology, the case study, and the datasets used in this study, while the results are described in Section 3. The discussion is presented in Section 4 and conclusion in Section 5.

Section snippets

Methodology for spatial analysis of HNVf

The construction of a gradient can be based on intensity indicators that express the density with which features are present at each point location. In this case, the indicators can be computed by using a gridding technique, the Kernel Density Estimation (KDE), which transforms values measured at specific locations in continuous surfaces (Bailey and Gatrell, 1995) by taking into consideration the surrounding context in which they are placed. This method was used in various territorial studies

Kernel bandwidth analysis

The clusters resulting from the ISODATA Cluster Analysis provided the classification of landscape typologies within the study area. The landscape typologies partially reflect the LC map, but they correspond to a smoother representation of urban, agricultural and natural clusters and to the transition among them. The number and the typologies of these clusters depends on the characteristics of the study area (represented by the LC map) and on the bandwidth of the kernel selected during the KDE

Discussions

Several studies have already utilized KDE techniques to characterise the urban-rural-natural structure of the landscape and their gradients, with the aim, for example, of detecting and quantifying the landscape pattern (Diti et al., 2015; Modica et al., 2012; Vizzari, 2011b; Vizzari and Sigura, 2013; Wu et al., 2006), or of analysing the provision of ecosystem services (Vizzari et al., 2015a,b; Vizzari et al., 2015a,b). In our case the analysis of urban-rural gradients was aimed at identifying

Conclusions

HNVf are key areas for biodiversity conservation in the rural environment, and the first step for their protection and implementation is their identification. The aim of this work was the identification and mapping of the HNVf in a portion of the Veneto Plain. It also aimed to show that the use of statistical data on agricultural practices in combination with LULC data allows for a better classification of HNVf in comparison with the sole use of LULC data. This study has contributed to the

References (55)

  • P.F. Ribeiro et al.

    Modelling farming system dynamics in High Nature Value Farmland under policy change

    Agric. Ecosyst. Environ.

    (2014)
  • P.F. Ribeiro et al.

    An applied farming systems approach to infer conservation-relevant agricultural practices for agri-environment policy design

    Land Use Policy

    (2016)
  • M.W. Strohbach et al.

    High nature value farming: from indication to conservation

    Ecol. Indic.

    (2015)
  • C.A. Sullivan

    Modelling semi-natural habitat area on lowland farms in western Ireland

    Biol. Conserv.

    (2011)
  • M. Vizzari

    Spatial modelling of potential landscape quality

    Appl. Geogr.

    (2011)
  • M. Almeida et al.

    Exploring the use of landscape as the basis for the identification of High nature value farmland: a case-study in the Portuguese montado

    Proceedings of the 10th European IFSA Symposium on Producing and Reproducing Farming Systems: New Models of Organisation for Sustainable Food Systems of Tomorrow

    (2012)
  • E. Andersen et al.

    Developing a High Nature Value Farming Area Indicator

    (2003)
  • T. Bailey et al.

    Interactive Spatial Data Analysis

    (1995)
  • E.M. Bignal et al.

    The nature conservation value of European traditional farming systems

    Environ. Rev.

    (2000)
  • X. Cai et al.

    Using kernel density estimation to assess the spatial pattern of road density and its impact on landscape fragmentation

    Int. J. Geogr. Inf. Sci.

    (2013)
  • C. Carvalho-Santos et al.

    Fine-scale mapping of High nature value farmlands: novel approaches to improve the management of rural biodiversity and ecosystem services

    Proceedings of the IUFRO Landscape Ecology Working Group International Conference

    (2010)
  • T. Cooper

    Final Report for the Study on HNV Indicators for Evaluation

    (2007)
  • European Commission

    The EU Biodiversity Strategy to 2020

    (2011)
  • Fondazione Lombardia per l’Ambiente

    Aree Agricole Ad Alto Valore Naturalistico in Lombardia (2011-2013)

    (2013)
  • R.T.T. Forman

    Some general principles of landscape and regional ecology

    Landsc. Ecol.

    (1995)
  • R.T.T. Forman et al.

    Landscape ecology

    Landsc. Ecol.

    (1986)
  • A. Gatrell et al.

    Spatial point pattern analysis and its application in geographical epidemiology

    Trans. Inst. Br. Geogr.

    (1996)
  • Cited by (11)

    • Assessment of LULC change dynamics and its relationship with LST and spectral indices in a rural area of Bengaluru district, Karnataka India

      2023, Remote Sensing Applications: Society and Environment
      Citation Excerpt :

      Some of the supervised classification algorithms are: Maximum Likelihood Classifier (MLC) (Owen et al., 1998; Benzer, 2010; Fichera et al., 2012; Shi et al., 2015; Rwanga and Ndambuki, 2017; Chamling and Bera, 2020; Yonaba et al., 2021), Parallelepiped, Minimum Distance, Mahalanobis Distance, Support Vector Machine (Foody et al., 2006; Bhagyanagar et al., 2012; Lin et al., 2015; Dibs et al., 2020). Unsupervised ISODATA and K-Mean algorithms were also adopted for LULC change detection (Saadat et al., 2011; Gadrani et al., 2018; Bonato et al., 2019; Talukdar et al., 2020). Other techniques include, spectral indices (Chen et al., 2006; Guha and Govil, 2020; Yang et al., 2020), Naïve Bayes (NB), K-Nearest Neighbour (KNN), Bootstrap-aggregation ensemble of decision trees (BagTE), Artificial and Deep Neural Networks, and Machine Learning techniques (Heydari and Mountrakis, 2018; Tolentino and Galo, 2021).

    • Spatio-temporal evaluation of the urban agglomeration expansion in the middle reaches of the Yangtze River and its impact on ecological lands

      2021, Science of the Total Environment
      Citation Excerpt :

      Considering the focalization of the study, the tree and shrubland classes and their subordinate land-use types in the LULC were merged into the forest class. The classes for urban, grassland, agricultural land, bare land, and wetland and water areas followed the original LULC classification (Bonato et al., 2019). According to the land space planning classification standard of the Ministry of Natural Resources of RPC (http://vod.mnr.gov.cn/spxw/), agricultural area, forest land, grassland, bare area, water area, and wetland are all classified as ecological lands (Liu et al., 2015).

    View all citing articles on Scopus
    View full text