Spatial variation in determinants of agricultural land abandonment in Europe. Levers, C., Schneider, M., Prishchepov, A. V., Estel, S., & Kuemmerle, T. Science of the Total Environment, 644:95 – 111, Elsevier B.V., 2018. Cited by: 175![Spatial variation in determinants of agricultural land abandonment in Europe [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Agricultural abandonment is widespread and growing in many regions worldwide, often because of agricultural intensification on productive lands, conservation policies, or the spatial decoupling of agricultural production from consumption. Abandonment has major environmental and social impacts, which differ starkly depending on the geographical context, as does its potential to serve as a land reservoir for recultivation. Understanding determinants of abandonment patterns, and especially how their influence varies across broad geographic extents, is therefore important. Using a pan-European map of agricultural abandonment derived from MODIS NDVI time series between 2001 and 2012, we quantified the importance of farm management, climatic, environmental, and socio-economic variables in explaining abandonment patterns. We chose a machine learning modelling framework that accounts for spatial variation in the relationship between abandonment and its determinants. We predicted abandonment probability as well as determinant coefficients for the entire study area and summarised them for regions under selected EU support schemes. Our results highlight that agricultural abandonment was mainly explained by climate conditions suboptimal for agriculture (i.e., low/high growing degrees days). Determinants related to farm management (smaller field size, lower yields) and socio-economic conditions (high unemployment, negative migration balance) also contributed to describing agricultural abandonment patterns in Europe. Several determinants influenced abandonment in strongly non-linear ways and we found substantial spatial non-stationarity effects, although abandonment patterns were equally well-explained by predictors specified with spatially constant and varying effects. Predicted abandonment probability was similar inside and outside EU support or conservation zones, whereas observed MODIS-based abandonment was generally higher outside these zones, suggesting that schemes such as Natura 2000 or High Nature Value Farmland likely influence abandonment patterns. Our work highlights the potential value of spatial boosting for gaining insights into land-use change processes and their outcomes, which should increase the ability of such models to inform context-specific, regionalised decision making. © 2018 Elsevier B.V.
@ARTICLE{Levers201895,
author = {Levers, Christian and Schneider, Max and Prishchepov, Alexander V. and Estel, Stephan and Kuemmerle, Tobias},
title = {Spatial variation in determinants of agricultural land abandonment in Europe},
year = {2018},
journal = {Science of the Total Environment},
volume = {644},
pages = {95 – 111},
doi = {10.1016/j.scitotenv.2018.06.326},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85049352376&doi=10.1016%2fj.scitotenv.2018.06.326&partnerID=40&md5=d59d21a4df123eab8c5ad89e637621fd},
affiliations = {Geography Department, Humboldt-Universität zu Berlin, Unter den Linden 6, Berlin, 10099, Germany; Department of Statistics, University of Washington, Box 354322, Seattle, 98195-4322, WA, United States; Section of Geography, Department of Geosciences and Natural Resource Management, University of Copenhagen, Øster Voldgade 10, Copenhagen K, DK-1350, Denmark; Institute of Environmental Sciences, Kazan Federal University, Tovarisheskaya str. 5, Kazan, 420097, Russian Federation; Institute of Steppe of the Ural Branch of the Russian Academy of Sciences, Pionerskaya str.11, Orenburg, 460000, Russian Federation; Department of Earth & Environment, Boston University, 685 Commonwealth Avenue, Boston, 02215, MA, United States; Integrative Research Institute on Transformations of Human-Environment Systems (IRI THESys), Humboldt-Universität zu Berlin, Unter den Linden 6, Berlin, 10099, Germany},
abstract = {Agricultural abandonment is widespread and growing in many regions worldwide, often because of agricultural intensification on productive lands, conservation policies, or the spatial decoupling of agricultural production from consumption. Abandonment has major environmental and social impacts, which differ starkly depending on the geographical context, as does its potential to serve as a land reservoir for recultivation. Understanding determinants of abandonment patterns, and especially how their influence varies across broad geographic extents, is therefore important. Using a pan-European map of agricultural abandonment derived from MODIS NDVI time series between 2001 and 2012, we quantified the importance of farm management, climatic, environmental, and socio-economic variables in explaining abandonment patterns. We chose a machine learning modelling framework that accounts for spatial variation in the relationship between abandonment and its determinants. We predicted abandonment probability as well as determinant coefficients for the entire study area and summarised them for regions under selected EU support schemes. Our results highlight that agricultural abandonment was mainly explained by climate conditions suboptimal for agriculture (i.e., low/high growing degrees days). Determinants related to farm management (smaller field size, lower yields) and socio-economic conditions (high unemployment, negative migration balance) also contributed to describing agricultural abandonment patterns in Europe. Several determinants influenced abandonment in strongly non-linear ways and we found substantial spatial non-stationarity effects, although abandonment patterns were equally well-explained by predictors specified with spatially constant and varying effects. Predicted abandonment probability was similar inside and outside EU support or conservation zones, whereas observed MODIS-based abandonment was generally higher outside these zones, suggesting that schemes such as Natura 2000 or High Nature Value Farmland likely influence abandonment patterns. Our work highlights the potential value of spatial boosting for gaining insights into land-use change processes and their outcomes, which should increase the ability of such models to inform context-specific, regionalised decision making. © 2018 Elsevier B.V.},
author_keywords = {De-intensification; Drivers; Land-use change; Machine learning; Model-based boosting; Non-stationarity},
keywords = {Europe; Adaptive boosting; Economics; Farms; Land use; Machine learning; Radiometers; Agricultural abandonments; De-intensification; Driver; Farm management; Landuse change; Machine-learning; Model-based boosting; Model-based OPC; Non-stationarities; Spatial variations; abandoned land; agricultural intensification; agricultural land; land use change; machine learning; modeling; MODIS; NDVI; spatial variation; agricultural abandonment; agricultural land; agricultural management; algorithm; Article; climate; environmental factor; environmental protection; Europe; farm management; geographic distribution; land use; machine learning; priority journal; probability; socioeconomics; spatial analysis; trend study; article; decision making; Europe; human; machine learning; time series analysis; unemployment; Decision making},
correspondence_address = {C. Levers; Geography Department, Humboldt-Universität zu Berlin, Berlin, Unter den Linden 6, 10099, Germany; email: christian.levers@geo.hu-berlin.de},
publisher = {Elsevier B.V.},
issn = {00489697},
coden = {STEVA},
pmid = {29981521},
language = {English},
abbrev_source_title = {Sci. Total Environ.},
type = {Article},
publication_stage = {Final},
source = {Scopus},
note = {Cited by: 175}
}
Downloads: 0
{"_id":"FiLppHWXuGB5zSGH4","bibbaseid":"levers-schneider-prishchepov-estel-kuemmerle-spatialvariationindeterminantsofagriculturallandabandonmentineurope-2018","author_short":["Levers, C.","Schneider, M.","Prishchepov, A. V.","Estel, S.","Kuemmerle, T."],"bibdata":{"bibtype":"article","type":"Article","author":[{"propositions":[],"lastnames":["Levers"],"firstnames":["Christian"],"suffixes":[]},{"propositions":[],"lastnames":["Schneider"],"firstnames":["Max"],"suffixes":[]},{"propositions":[],"lastnames":["Prishchepov"],"firstnames":["Alexander","V."],"suffixes":[]},{"propositions":[],"lastnames":["Estel"],"firstnames":["Stephan"],"suffixes":[]},{"propositions":[],"lastnames":["Kuemmerle"],"firstnames":["Tobias"],"suffixes":[]}],"title":"Spatial variation in determinants of agricultural land abandonment in Europe","year":"2018","journal":"Science of the Total Environment","volume":"644","pages":"95 – 111","doi":"10.1016/j.scitotenv.2018.06.326","url":"https://www.scopus.com/inward/record.uri?eid=2-s2.0-85049352376&doi=10.1016%2fj.scitotenv.2018.06.326&partnerID=40&md5=d59d21a4df123eab8c5ad89e637621fd","affiliations":"Geography Department, Humboldt-Universität zu Berlin, Unter den Linden 6, Berlin, 10099, Germany; Department of Statistics, University of Washington, Box 354322, Seattle, 98195-4322, WA, United States; Section of Geography, Department of Geosciences and Natural Resource Management, University of Copenhagen, Øster Voldgade 10, Copenhagen K, DK-1350, Denmark; Institute of Environmental Sciences, Kazan Federal University, Tovarisheskaya str. 5, Kazan, 420097, Russian Federation; Institute of Steppe of the Ural Branch of the Russian Academy of Sciences, Pionerskaya str.11, Orenburg, 460000, Russian Federation; Department of Earth & Environment, Boston University, 685 Commonwealth Avenue, Boston, 02215, MA, United States; Integrative Research Institute on Transformations of Human-Environment Systems (IRI THESys), Humboldt-Universität zu Berlin, Unter den Linden 6, Berlin, 10099, Germany","abstract":"Agricultural abandonment is widespread and growing in many regions worldwide, often because of agricultural intensification on productive lands, conservation policies, or the spatial decoupling of agricultural production from consumption. Abandonment has major environmental and social impacts, which differ starkly depending on the geographical context, as does its potential to serve as a land reservoir for recultivation. Understanding determinants of abandonment patterns, and especially how their influence varies across broad geographic extents, is therefore important. Using a pan-European map of agricultural abandonment derived from MODIS NDVI time series between 2001 and 2012, we quantified the importance of farm management, climatic, environmental, and socio-economic variables in explaining abandonment patterns. We chose a machine learning modelling framework that accounts for spatial variation in the relationship between abandonment and its determinants. We predicted abandonment probability as well as determinant coefficients for the entire study area and summarised them for regions under selected EU support schemes. Our results highlight that agricultural abandonment was mainly explained by climate conditions suboptimal for agriculture (i.e., low/high growing degrees days). Determinants related to farm management (smaller field size, lower yields) and socio-economic conditions (high unemployment, negative migration balance) also contributed to describing agricultural abandonment patterns in Europe. Several determinants influenced abandonment in strongly non-linear ways and we found substantial spatial non-stationarity effects, although abandonment patterns were equally well-explained by predictors specified with spatially constant and varying effects. Predicted abandonment probability was similar inside and outside EU support or conservation zones, whereas observed MODIS-based abandonment was generally higher outside these zones, suggesting that schemes such as Natura 2000 or High Nature Value Farmland likely influence abandonment patterns. Our work highlights the potential value of spatial boosting for gaining insights into land-use change processes and their outcomes, which should increase the ability of such models to inform context-specific, regionalised decision making. © 2018 Elsevier B.V.","author_keywords":"De-intensification; Drivers; Land-use change; Machine learning; Model-based boosting; Non-stationarity","keywords":"Europe; Adaptive boosting; Economics; Farms; Land use; Machine learning; Radiometers; Agricultural abandonments; De-intensification; Driver; Farm management; Landuse change; Machine-learning; Model-based boosting; Model-based OPC; Non-stationarities; Spatial variations; abandoned land; agricultural intensification; agricultural land; land use change; machine learning; modeling; MODIS; NDVI; spatial variation; agricultural abandonment; agricultural land; agricultural management; algorithm; Article; climate; environmental factor; environmental protection; Europe; farm management; geographic distribution; land use; machine learning; priority journal; probability; socioeconomics; spatial analysis; trend study; article; decision making; Europe; human; machine learning; time series analysis; unemployment; Decision making","correspondence_address":"C. Levers; Geography Department, Humboldt-Universität zu Berlin, Berlin, Unter den Linden 6, 10099, Germany; email: christian.levers@geo.hu-berlin.de","publisher":"Elsevier B.V.","issn":"00489697","coden":"STEVA","pmid":"29981521","language":"English","abbrev_source_title":"Sci. Total Environ.","publication_stage":"Final","source":"Scopus","note":"Cited by: 175","bibtex":"@ARTICLE{Levers201895,\n\tauthor = {Levers, Christian and Schneider, Max and Prishchepov, Alexander V. and Estel, Stephan and Kuemmerle, Tobias},\n\ttitle = {Spatial variation in determinants of agricultural land abandonment in Europe},\n\tyear = {2018},\n\tjournal = {Science of the Total Environment},\n\tvolume = {644},\n\tpages = {95 – 111},\n\tdoi = {10.1016/j.scitotenv.2018.06.326},\n\turl = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85049352376&doi=10.1016%2fj.scitotenv.2018.06.326&partnerID=40&md5=d59d21a4df123eab8c5ad89e637621fd},\n\taffiliations = {Geography Department, Humboldt-Universität zu Berlin, Unter den Linden 6, Berlin, 10099, Germany; Department of Statistics, University of Washington, Box 354322, Seattle, 98195-4322, WA, United States; Section of Geography, Department of Geosciences and Natural Resource Management, University of Copenhagen, Øster Voldgade 10, Copenhagen K, DK-1350, Denmark; Institute of Environmental Sciences, Kazan Federal University, Tovarisheskaya str. 5, Kazan, 420097, Russian Federation; Institute of Steppe of the Ural Branch of the Russian Academy of Sciences, Pionerskaya str.11, Orenburg, 460000, Russian Federation; Department of Earth & Environment, Boston University, 685 Commonwealth Avenue, Boston, 02215, MA, United States; Integrative Research Institute on Transformations of Human-Environment Systems (IRI THESys), Humboldt-Universität zu Berlin, Unter den Linden 6, Berlin, 10099, Germany},\n\tabstract = {Agricultural abandonment is widespread and growing in many regions worldwide, often because of agricultural intensification on productive lands, conservation policies, or the spatial decoupling of agricultural production from consumption. Abandonment has major environmental and social impacts, which differ starkly depending on the geographical context, as does its potential to serve as a land reservoir for recultivation. Understanding determinants of abandonment patterns, and especially how their influence varies across broad geographic extents, is therefore important. Using a pan-European map of agricultural abandonment derived from MODIS NDVI time series between 2001 and 2012, we quantified the importance of farm management, climatic, environmental, and socio-economic variables in explaining abandonment patterns. We chose a machine learning modelling framework that accounts for spatial variation in the relationship between abandonment and its determinants. We predicted abandonment probability as well as determinant coefficients for the entire study area and summarised them for regions under selected EU support schemes. Our results highlight that agricultural abandonment was mainly explained by climate conditions suboptimal for agriculture (i.e., low/high growing degrees days). Determinants related to farm management (smaller field size, lower yields) and socio-economic conditions (high unemployment, negative migration balance) also contributed to describing agricultural abandonment patterns in Europe. Several determinants influenced abandonment in strongly non-linear ways and we found substantial spatial non-stationarity effects, although abandonment patterns were equally well-explained by predictors specified with spatially constant and varying effects. Predicted abandonment probability was similar inside and outside EU support or conservation zones, whereas observed MODIS-based abandonment was generally higher outside these zones, suggesting that schemes such as Natura 2000 or High Nature Value Farmland likely influence abandonment patterns. Our work highlights the potential value of spatial boosting for gaining insights into land-use change processes and their outcomes, which should increase the ability of such models to inform context-specific, regionalised decision making. © 2018 Elsevier B.V.},\n\tauthor_keywords = {De-intensification; Drivers; Land-use change; Machine learning; Model-based boosting; Non-stationarity},\n\tkeywords = {Europe; Adaptive boosting; Economics; Farms; Land use; Machine learning; Radiometers; Agricultural abandonments; De-intensification; Driver; Farm management; Landuse change; Machine-learning; Model-based boosting; Model-based OPC; Non-stationarities; Spatial variations; abandoned land; agricultural intensification; agricultural land; land use change; machine learning; modeling; MODIS; NDVI; spatial variation; agricultural abandonment; agricultural land; agricultural management; algorithm; Article; climate; environmental factor; environmental protection; Europe; farm management; geographic distribution; land use; machine learning; priority journal; probability; socioeconomics; spatial analysis; trend study; article; decision making; Europe; human; machine learning; time series analysis; unemployment; Decision making},\n\tcorrespondence_address = {C. Levers; Geography Department, Humboldt-Universität zu Berlin, Berlin, Unter den Linden 6, 10099, Germany; email: christian.levers@geo.hu-berlin.de},\n\tpublisher = {Elsevier B.V.},\n\tissn = {00489697},\n\tcoden = {STEVA},\n\tpmid = {29981521},\n\tlanguage = {English},\n\tabbrev_source_title = {Sci. Total Environ.},\n\ttype = {Article},\n\tpublication_stage = {Final},\n\tsource = {Scopus},\n\tnote = {Cited by: 175}\n}\n\n","author_short":["Levers, C.","Schneider, M.","Prishchepov, A. V.","Estel, S.","Kuemmerle, T."],"bibbaseid":"levers-schneider-prishchepov-estel-kuemmerle-spatialvariationindeterminantsofagriculturallandabandonmentineurope-2018","role":"author","urls":{"Paper":"https://www.scopus.com/inward/record.uri?eid=2-s2.0-85049352376&doi=10.1016%2fj.scitotenv.2018.06.326&partnerID=40&md5=d59d21a4df123eab8c5ad89e637621fd"},"keyword":["Europe; Adaptive boosting; Economics; Farms; Land use; Machine learning; Radiometers; Agricultural abandonments; De-intensification; Driver; Farm management; Landuse change; Machine-learning; Model-based boosting; Model-based OPC; Non-stationarities; Spatial variations; abandoned land; agricultural intensification; agricultural land; land use change; machine learning; modeling; MODIS; NDVI; spatial variation; agricultural abandonment; agricultural land; agricultural management; algorithm; Article; climate; environmental factor; environmental protection; Europe; farm management; geographic distribution; land use; machine learning; priority journal; probability; socioeconomics; spatial analysis; trend study; article; decision making; Europe; human; machine learning; time series analysis; unemployment; Decision making"],"metadata":{"authorlinks":{}}},"bibtype":"article","biburl":"https://bibbase.org/f/akvPCQYE74MCSypqX/scopus natura 2000.bib","dataSources":["jEMMcdj2v9At9Q5md"],"keywords":["europe; adaptive boosting; economics; farms; land use; machine learning; radiometers; agricultural abandonments; de-intensification; driver; farm management; landuse change; machine-learning; model-based boosting; model-based opc; non-stationarities; spatial variations; abandoned land; agricultural intensification; agricultural land; land use change; machine learning; modeling; modis; ndvi; spatial variation; agricultural abandonment; agricultural land; agricultural management; algorithm; article; climate; environmental factor; environmental protection; europe; farm management; geographic distribution; land use; machine learning; priority journal; probability; socioeconomics; spatial analysis; trend study; article; decision making; europe; human; machine learning; time series analysis; unemployment; decision making"],"search_terms":["spatial","variation","determinants","agricultural","land","abandonment","europe","levers","schneider","prishchepov","estel","kuemmerle"],"title":"Spatial variation in determinants of agricultural land abandonment in Europe","year":2018}