Estimating agroforestry area and greenhouse gas mitigation potential in Europe

Conference abstract for oral presentation World Congress on Agroforestry (Kigali, Rwanda, 20-24 October 2025) & IALE 2025 (Bratislava, Slovakia, 2-5 September 2025)

Michael den Herder¹, Gerry Lawson², Davor Deranja³, Sonja Kay⁴, Manuel Bertomeu Garcia⁵, Arina Machine⁶, Kristoffer Rønn-Andersen⁷, Miska Nevalainen¹, João Palma⁸, Ana Tomás^8
1European Forest Institute, Yliopistokatu 6 B, FI-80100 Joensuu, Finland
²European Agroforestry Federation, Montpellier, France
³Planet Labs, Ljubljana, Slovenia
⁴Agroscope, Zurich, Switzerland
⁵University of Extremadura, Plasencia, Spain
⁶University of Leicester, United Kingdom
⁷Regen Farmer, Odense, Denmark
⁸Moinhos de Vento, Agroecology Research Centre, Mértola, Portugal

Reliable estimates on the extent of agroforestry are important for designing policies impacting this land use practice, particularly in the context of the EU Deforestation Regulation where agroforestry land must be clearly distinguished from forest. Until 2009, the extent of agroforestry had not been quantified leading to widely varied interpretations on the importance of this practice. In a first attempt, using remote sensing derived global datasets¹, it was estimated that agroforestry would cover about 46% of the agricultural land globally. For Europe, the authors estimated an impressive 40% (91 million ha) of agricultural land covered by agroforestry. Later studies using a pan-European database called LUCAS estimated that about 15.4 – 20.3 million hectares would be covered by agroforestry^2,3, but when longer time series became available the updated database suggested a 47% decline during the period 2009-2018⁴. The most recent study noted a significant underreporting of grazed areas due to LUCAS becoming increasingly reliant on photo-interpretation⁵. These widely varying estimates demonstrate the difficulty in making reliable estimates on the extent of agroforestry.

Agroforestry has a significant climate change mitigation potential because of its capacity to store carbon in biomass and soil. Temperate agroforestry systems store less carbon compared to tropical systems, but as a large part of Europe is covered by agriculture (about 40% of the total land area), converting agricultural land to agroforestry would have enormous potential of storing additional carbon. Even in the boreal region, where carbon storage is relatively limited due to slow growth, the potential is large due to the vastness of this area. Therefore, it is important to develop a more harmonized, standardized, and reliable estimate of the current extent of agroforestry to more precisely assess the amount of carbon currently stored in agroforestry systems and its future potential.

In this study, we will focus on existing methods which could be used in estimating agroforestry areas with a focus on Europe in particular. These include: 1) Forest Inventory methods to report the area of trees outside forests, 2) EU Agricultural Policy monitoring methods to report the area of agroforestry and landscape features, 3) geospatial Land Parcel Identification System (LPIS) information on “hectares of agroforestry”, 4) Landscape features (stored in the Ecological Focus Area layer of some EU member state’s LPIS systems, 5) current best assessments of percent cent crown cover from a combination of COPERNICUS and CORINE data (Figure 1) and LPIS orthoimages, and 6) combined methods using aerial photographs, satellite pictures, LPIS data, and identification of trees based on machine learning. The pros and cons of the examined methods will be evaluated with the aim of giving guidance aiming at more harmonised estimates on the extent of agroforestry in Europe, as well as thinking towards compatibility with global estimates on the extent of agroforestry.

¹ Zomer, R.J., et al. 2009. ICRAF Working Paper no. 89.
² den Herder et al. 2017, https://doi.org/10.1016/j.agee.2017.03.005
³ Plieninger et al. 2015, https://doi.org/10.1016/j.biocon.2015.05.014
⁴ Rubio-Delgado et al. 2023, https://doi.org/10.3389/fenvs.2023.1258697
⁵ Machine et al. 2025. https://doi.org/10.1088/1748-9326/adc8bf