Woody landscape features on agricultural land in Europe
Indicator
Preparing download
The removal of landscape features on agricultural land in Europe is one of the main agricultural pressures for biodiversity and ecosystem services. The EU Green Deal aims to bring back high-diversity landscape features on at least 10% of agricultural land by 2030, including woody features like tree lines, tree groups and hedges among other small habitats. Copernicus Earth observation data shows, woody landscape features covered 5% of the EU’s agricultural land in 2018.
Figure 1. Share of woody landscape features on agricultural area in the EU member states and non-EU EEA member and cooperating countries (%)
Country
Woody Vegetation Mask (WVM) on agricultural area (AA)(%)
The agricultural landscape features are small fragments of habitats on agricultural landscapes, including woody features, field margins, ponds, ditches, stone walls and other features. They support biodiversity and provide ecosystem services such as soil protection and pollination .
Copernicus Land Monitoring Service earth observation data estimated that woody landscape features covered 5% of the EU’s agricultural land in 2018, ranging from 2.6% in Cyprus to 9.3% in Ireland. Four EU Member States have more than 7% of their agricultural areas covered by woody landscape features (Ireland, Slovenia, Portugal and Croatia). Eleven Member States have between 5% and 7%, followed by a range of 3% to 5% in another eight Member States. The estimation is lower than 3% in Cyprus, Romania, Luxembourg and Spain.
The estimated share of woody landscape features for agricultural areas in EEA member countries outside the EU and for cooperating countries range from 0.2% in Iceland to 7.7% in Liechtenstein.
Traditionally, landscape features existed due to their agricultural functions, such as acting as windbreaks or as remnants of previously larger natural or semi-natural habitats on land unfavourable for agricultural production.
With the modernisation of agriculture in Europe, the disappearance of landscape features has been one of the most important pressures on biodiversity, as reported by Member States under the nature directives. However, comprehensive EU-level trend information is not available, therefore establishing the monitoring of landscape features is crucial to track changes. This indicator provides the first comprehensive estimation for the share of woody landscape features in Europe in 2018.
When assessing the prospects based on current agricultural policydevelopments in the EU, without increased ambition and strengthened implementation, the share of woody landscape features is not expected to increase considerably by 2030. More incentives and knowledge transfer are needed to overcome barriers like high costs, technical knowledge needs and insufficient awareness about environmental and economic benefits.
The new EU Nature restoration law may aid the increasing trend in the share of woody landscape features as it intends to increase the agricultural land covered with high-diversity landscape features.
Figure 2. Share of woody landscape features on agricultural areas in NUTS3 regions
The estimated proportion of woody landscape features on agricultural land highly varies within countries. Reasons can include agricultural management, geographical, climatic and cultural factors.
Some NUTS3 regions — mainly in Albania, Austria, Croatia, France, Greece, Ireland, Montenegro, Portugal, Sweden, and Türkiye — have reached levels close to or exceeding 10%. Conversely, many regions have values below 5%. The lowest proportions can be observed in regions in France, Hungary, Iceland, Romania, Spain and Türkiye. The highest proportions (>20%) appear in southern France, Greece, central Italy, northern Spain and northern Türkiye.
Clear distinctions between regions with high and low woody landscape feature shares within a country are evident in various cases such as the Île-de-France region compared to the rest of France, between southern and northern Portugal and the region of Türkiye along the Black Sea in comparison to the rest of Türkiye.
Some urban NUTS3 regions with relatively high values compared to the surrounding regions can be observed, which results from a lower extent of agricultural areas with a relatively higher share of woody features in the urban areas.
Supporting information
Landscape features include woody, grassy, wet and stony features, such as tree lines, hedges, field margins, ponds, ditches, stone walls and terraces and other types. They support biodiversity and deliver ecosystem services, providing multiple benefits to agro-ecosystems, the wider environment and to agricultural production.
This indicator estimates the share of area covered by woody landscape features on agricultural land in Europe using data from the Copernicus Land Monitoring Services.
The methodology is described in detail in an ETC report. The indicator is based on two components: the reference area (i.e. the agricultural area) and the target class to be analysed (i.e. the woody landscape features).
a/ Agricultural area
The delineation of the agricultural area aimed at providing the best possible spatial approximation for the indicator is based on Earth observation data using Copernicus Land Monitoring products. This area is different from the statistical data of the Utilised agricultural area (UAA) that is used in agricultural statistics. The agricultural area is higher than UAA by 5-20% in most countries.
The indicator values are also calculated for sub-classes of agricultural areas as the importance and biodiversity benefits of woody landscape features can vary according to the type of agricultural landscape.
The agricultural area mask (AA mask) is derived from the Corine Land Cover (CLC) data for 2018 by including the CLC class agricultural areas (Classes 2xx) and natural grasslands (Class 3.2.1) and it was refined using Copernicus High Resolution Layers (HRL). The AA mask is a raster layer with 100m spatial resolution. It includes those 100m x100m cells that are dominated by arable land, permanent crops, pastures, grasslands and/or heterogeneous agricultural areas.
As CLC uses a 25ha minimum mapping unit, HRLs (HRL Imperviousness 2018, HRL Forest Type 2018, HRL Water and Wetness 2018) were used to remove non-agricultural surfaces (artificial areas, forest and water bodies, respectively) that had been included due to generalisation. On the other hand, areas covered by greenhouses included in the imperviousness layer have been added back to the mask from existing Copernicus local component data (Riparian Zones, N2K and Coastal Zones) and visual interpretation.
b/ Woody landscape features
The landscape features are assessed via the Woody Vegetation Mask (WVM) – one of the layers of the high-resolution Small Woody Features (SWF) 2018 product portfolio.
The WVM is a product depicting all woody features (i.e. trees and scrubs) detected from the images of 2-4 m resolution without filtering by vegetation height or by the size or shape of the detected features. Artefacts including tree rows such as olive tree plantations, vineyards and orchards have been removed after the classification/segmentation process through a manual thematic enhancement. The product allows the user to flexibly apply their own rules to derive any subtype of woody features they specifically require for their topic of interest. To separate the WVM from forested areas, the Forest mask 2018 was applied. To limit overlaps with large and densely tree covered areas (> 0.5ha), the production workflow included a masking approach using the HRL Tree cover density 2018 layer.
Methodology to derive woody landscape features within agricultural area
The original 5m x 5m information is aggregated at 100m x 100m, i.e. as a percentage share inside a 1ha cell of the agricultural area mask to match the spatial resolution of that mask.
The share of the area of the woody features was aggregated to different administrative units (NUTS). The actual percentage within each 1ha cell is used for the calculation, i.e. if the percentage within one cell is 50%, then only 0.5ha is taken into account for the sum for the NUTS region.
The surface covered by landscape features is compared to the surface covered by the agricultural area (per administrative unit).
The results are expressed as a percentage.
Landscape features, including woody landscape features, provide habitats and can act as corridors in the agricultural landscape for wildlife, including farmland birds, wild pollinators and the natural enemies of pests. In general, landscape features are elements of structural diversity. Higher structural diversity extends the number of different ecological niches and habitats available for species. Compared to their relatively small size, they have important agricultural, environmental, cultural and other functions, which has been recognised in recent EU policies.
They contribute to soil and water protection, pest control, pollination and climate change mitigation and adaptation, benefitting agricultural production. They help to reduce water pollution and to increase water availability for crops. Woody landscape features provide shade for livestock, cool the microclimate and provide cultural services such as recreation and ecotourism.
The EU Biodiversity Strategy for 2030, part of the European Green Deal, addresses the main drivers of biodiversity loss. One of the key targets of the strategy demands that at least 10% of agricultural land is composed of high-diversity landscape features. EU policies are expected to support this objective, including the habitats directive, the Common Agricultural Policy (CAP) and the upcoming nature restoration law.
The habitats directive specifies that Member States, if they consider it necessary, shall maintain landscape features that are of major importance for wild fauna and flora to improve the connectivity between Natura 2000 sites.
The CAP has included policy tools to support the maintenance of landscape features since 1992. The current CAP supports the maintenance and creation of landscape features as part of its key policy objective ‘contribute to the protection of biodiversity, enhance ecosystem services and preserve habitats and landscapes’. It requires beneficiaries of the CAP subsidies to maintain landscape features and to devote a minimum share of agricultural area to non-productive areas or features (as set under good agricultural and environmental conditions standard 8). However, the possibility to derogate from the latter requirement for farmers has been provided to Member States in both first two years of the policy, in 2023 and 2024.
The CAP offers possibilities for Member States to support the maintenance and establishment of landscape features through the eco-schemes and through interventions for rural development with voluntary participation for farmers. Types of interventions for rural development relevant for landscape features include payments for non-productive investments and environmental, climate and other management commitments.
The Performance Monitoring and Evaluation Framework (PMEF) of the CAP introduced a new impact indicator, the I. 21 Share of landscape features within agricultural area. The indicator is calculated based on the Land Use/Cover Area frame Survey LUCAS Landscape features module that provides the first comprehensive statistical dataset for the share of landscape features in agricultural land for 2022, covering woody, grassy, wet and stony features. The data from the present indicator is used for comparison with the LUCAS data in relation to woody landscape features. The consistencies and deviations revealed by the comparison can help to understand the strengths and limitations of each dataset.
The provisionally agreed text of the Nature Restoration Law includes the indicator ‘share of agricultural land with high-diversity landscape features (HDLF)’. It is included in a list of three indicators for agricultural ecosystems, out of which Member States shall choose two and put in place measures which aim at achieving an increasing trend. The text describes what constitutes HDLF that provide ecosystem services and support for biodiversity, including a set of pre-conditions that landscape features must comply with.
The present indicator has the potential to contribute to measuring progress towards the EU objective of bringing back nature to agricultural land with regards to woody landscape features. Tracking changes at European level might be challenging in the near future due to the relatively low expected annual changes compared to the uncertainty level of the data source. However, it will depend on the scale and distribution of changes and on further technological and conceptual improvements.
The indicator is based on the only available wall-to-wall geospatial data of woody landscape features in Europe. The data has the potential to improve in quality with time, and to be used for assessing key areas or connectivity of habitats within the agricultural landscapes, beyond the overall quantification of woody landscape features.
Methodology Uncertainties
The reference area of the agricultural area mask provides the best available approximation of agricultural area within the covered geographic area. Nonetheless, an agricultural area itself is subject to uncertainties and differs depending on the data source or definitions under consideration. A consolidated reference area is a key prerequisite for the indicator calculation.
Considering the higher uncertainty in detecting woody landscape features within agroforestry areas, the share of woody landscape features in an agricultural area has been calculated both by including or excluding the agroforestry CLC class (2.4.4.). This alters the share of landscape features only slightly in a few countries: Spain from 2.77% to 2.90%, Italy from 5.39% to 5.36%, and Portugal from 8.24% to 9.02%. No significant impact is confirmed at both NUTS0 and NUTS3 levels.
Data sets uncertainty
The agricultural area (AA) is necessarily different from the Utilised Agricultural Area (UAA) as it is based on land cover/land use information in contrast with the UAA that is a statistical/administrative data. The differences are reported in the ETC report.
The Copernicus Woody Vegetation Mask (WVM) 2018 product derived from very high-resolution images between 2 m and 4 m spatial resolution. The minimum mapping unit of the WVM product is larger than the minimum size of woody landscape features. Therefore, it can exclude relevant features such as individual trees and narrow tree lines.
The data has varying accuracy in different regions of Europe. Reliable trend information is not available yet.
Quantitative assessment of thematic accuracy is available for the Small woody features (SWF) 2018 product and not for the WVM 2018 product because the WVM 2018 was considered previously as an intermediate product. However, as the SWF 2018 is derived from the WVM 2018, its assessment is relevant for the indicator.
The internal validation of the SWF 2018 for the entire area of coverage (EE38 and the UK, including France’s overseas departments and regions) shows high overall accuracy (OA), indicating generally high reliability of the product.
The weighted OA is above 90% at both primary- (PSU) and secondary sampling unit (SSU) levels, 92.80% and 91.79%, respectively, with a confidence interval of +/- 0.10 at 95% confidence level. PSUs are 100m x 100m quadrats and SSUs are 20 randomly distributed points within the PSUs, considering a 5 x 5m area (pixel size) for the validation analysis.
The non-SWF class in the dataset shows very high accuracy (above 90%) and low overestimation and underestimation errors. The user’s and producer’s accuracies of the SWF-class are above 80% (90.9% and 88.2%, respectively) at primary sampling unit level but do not reach 80% at SSU level.
The validation is based on photo-interpretation using reference data from the images used in the production complemented by other ancillary data and on-line Earth observation platforms.
The working definition of SWF has been less suitable for reflecting the wide range of woody landscape feature types across Europe. The differentiation between woody features and the different types of forest is conceptually and technically complex. This makes capturing of woody landscape features challenging, especially by a semi-automated production workflow.
The accuracy of the SWF products is heterogeneous across different areas of Europe. Landscapes where the identification of woody landscape features is difficult include heterogeneous landscapes with complex mosaics of agricultural land and semi-natural vegetation (e.g. in the mediterranean and Scandinavian areas). In open forest areas manual enhancements were applied to remove woody features within those areas considered forest.
Qualitative ‘look and feel’ verification was performed to analyse the usability of the SWF and the WVM products and to identify weaknesses and possibilities to improve fitness for purpose. It was done on six 200x200 km tiles in different regions of Europe. The checks focused on areas where it is highly difficult to identify woody landscape features so high uncertainty was expected. Therefore, the result of the assessment does not represent the overall accuracy of the entire dataset but rather provides recommendations to improve the usability of the product portfolio.
The WVM data was compared with other available pan-European datasets from date 2018, such as CLC+ Backbone (CLC+ BB) raster and CORINE Land Cover (CLC) raster. The accuracies in the studied difficult landscapes were either estimated to reach 85% in most of the verified areas with minor errors or not expected to reach the minimum 85% with several errors in different regions (but not much below 85% and the majority of the verified areas were mapped correctly).
The verification supported the selection of the used WVM dataset for the indicator through its conclusion that the complex geometry rules used to produce the SWF cause inconsistencies in the data.
Overall, besides the limitations, the WVM data can be considered a valuable data source for a European level estimation of a baseline for the share of woody landscape features on agricultural land. The data quality is expected to further improve in the coming years.
Rationale uncertainty
The following uncertainties are related to the indicator rationale:
The biodiversity supporting functions and delivery of ecosystem services by landscape features need to be viewed depending on specific organisms: some groups need specific species composition (e.g. specialised butterfly-species with their need of certain food-plant species); others need certain structures (raised stands for birds of prey) or physical/chemical conditions (soil insects). In this respect, the benefits are always dependent on the target organisms or nature protection or restoration goals.
The nature protection and restoration targets and goals differ across Europe. In general, many different situations, conditions and combinations in a landscape can provide a high number of habitat types. Hence, it is difficult and may be even counterproductive to define general characteristics of landscape features that ensure benefits for biodiversity.
The question can be asked whether trees are landscape features in the case of agro-forestry areas or heterogeneous agricultural areas with transitions between open land and closed woody vegetation, with woody vegetation not appearing as distinct patches or linear features. The differentiation of landscape features is conceptually difficult in these types of agricultural areas.
Descriptive indicator (Type A - What is happening to the environment and to humans?)Responsible consumption and production, Life on land
Percentage of agricultural area (%)
Every 3 years
References and footnotes
Czúcz, B., Baruth, B., Terres, J. M., Gallego, J., Hagyo, A., Nocita, M., Perez Soba, M., Angileri, V., Koeble, R. and Paracchini, M.-L., 2022, Classification and quantification of landscape features in agricultural land across the EU: a brief review of existing definitions, typologies, and data sources for quantification, Publications Office of the European Union, Luxembourg.
Marshall, E. J. P. and Moonen, A. C., 2002, 'Field margins in northern Europe: Their functions and interactions with agriculture', Agriculture, Ecosystems & Environment 89(1), pp. 5–21.
Henderson, I. G., Holland, J. M., Storkey, J., Lutman, P., Orson, J. and Simper, J., 2012, 'Effects of the proportion and spatial arrangement of un-cropped land on breeding bird abundance in arable rotations', Journal of Applied Ecology 49(4), pp. 883–891 (https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1365-2664.2012.02166.x) accessed January 26, 2024.
Henderson, I. G., Holland, J. M., Storkey, J., Lutman, P. J. W., Orson, J. and Simper, J., 2012, 'Effects of the proportion and spatial arrangement of un‐cropped land on breeding bird abundance in arable rotations', Wiley.
England, J. R., O’Grady, A. P., Fleming, A., Marais, Z. and Mendham, D., 2020, 'Trees on farms to support natural capital: An evidence-based review for grazed dairy systems', The Science of the Total Environment 704, pp. 135345.
