Which factors determine the invasion of plant species? Machine learning based habitat modelling integrating environmental factors and climate scenarios
The increase in the spread of invasive plant species (IPS) causes major disturbances to ecosystem functions. Monitoring systems are considered necessary to implement effective measures against their spread. We created species distribution models that identify the potentially suitable habitat under p...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-02-01
|
| Series: | International Journal of Applied Earth Observations and Geoinformation |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1569843222003466 |
| _version_ | 1797950725453512704 |
|---|---|
| author | Fabian Sittaro Christopher Hutengs Michael Vohland |
| author_facet | Fabian Sittaro Christopher Hutengs Michael Vohland |
| author_sort | Fabian Sittaro |
| collection | DOAJ |
| description | The increase in the spread of invasive plant species (IPS) causes major disturbances to ecosystem functions. Monitoring systems are considered necessary to implement effective measures against their spread. We created species distribution models that identify the potentially suitable habitat under present and future climatic conditions for 46 IPS in Germany and incorporated habitat types obtained through remote sensing methods to assess their influence on habitat suitability.We included 18 environmental variables that describe habitat characteristics, including soil type, altitude, land use, transport infrastructure, temperature and precipitation. Models were based on two machine learning techniques: Support Vector Machines (SVM) and Boosted Regression Trees (BRT). SVM classification of Natura2000 habitat types using MODIS reflectance data was included to provide a vegetation type-based approach to interspecific competition. We integrated predicted climate variables to determine changes in habitat suitability for two forecast periods (2041–2060 and 2061–2080) and three Representative Concentration Pathways.Averaging over all species, the models showed good predictive power, with the quality of BRT (AUC 0.861; RMSE 0.225) surpassing that of SVM (AUC 0.804; RMSE 0.285). We observe that the majority of the species have not yet filled their potentially suitable habitat. An increase in habitat suitability for predicted climatic conditions is implied for most species.Our results indicate that the dynamics of biological invasions will intensify with anticipated climatic changes. Climate factors, soil type and transport infrastructure are of great relevance for the distribution of IPS, while interspecific competition, indirectly assessed through the distribution of habitat types, is only relevant for some species. |
| first_indexed | 2024-04-10T22:20:47Z |
| format | Article |
| id | doaj.art-4e12bec1ec9f4dd1be4cbbf094da30b2 |
| institution | Directory Open Access Journal |
| issn | 1569-8432 |
| language | English |
| last_indexed | 2024-04-10T22:20:47Z |
| publishDate | 2023-02-01 |
| publisher | Elsevier |
| record_format | Article |
| series | International Journal of Applied Earth Observations and Geoinformation |
| spelling | doaj.art-4e12bec1ec9f4dd1be4cbbf094da30b22023-01-18T04:30:08ZengElsevierInternational Journal of Applied Earth Observations and Geoinformation1569-84322023-02-01116103158Which factors determine the invasion of plant species? Machine learning based habitat modelling integrating environmental factors and climate scenariosFabian Sittaro0Christopher Hutengs1Michael Vohland2Leipzig University, Geoinformatics and Remote Sensing, Institute for Geography, Johannisallee 19a, 04103 Leipzig, Germany; German Biomass Research Centre gemeinnützige GmbH (DBFZ), Department of Bioenergy Systems, Torgauer Straße 116, 04347 Leipzig, Germany; Corresponding author.Leipzig University, Geoinformatics and Remote Sensing, Institute for Geography, Johannisallee 19a, 04103 Leipzig, Germany; Leipzig University, Remote Sensing Centre for Earth System Research, Talstraße 35, 04103 Leipzig, Germany; German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Puschstr. 4, 04103 Leipzig, GermanyLeipzig University, Geoinformatics and Remote Sensing, Institute for Geography, Johannisallee 19a, 04103 Leipzig, Germany; Leipzig University, Remote Sensing Centre for Earth System Research, Talstraße 35, 04103 Leipzig, Germany; German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Puschstr. 4, 04103 Leipzig, GermanyThe increase in the spread of invasive plant species (IPS) causes major disturbances to ecosystem functions. Monitoring systems are considered necessary to implement effective measures against their spread. We created species distribution models that identify the potentially suitable habitat under present and future climatic conditions for 46 IPS in Germany and incorporated habitat types obtained through remote sensing methods to assess their influence on habitat suitability.We included 18 environmental variables that describe habitat characteristics, including soil type, altitude, land use, transport infrastructure, temperature and precipitation. Models were based on two machine learning techniques: Support Vector Machines (SVM) and Boosted Regression Trees (BRT). SVM classification of Natura2000 habitat types using MODIS reflectance data was included to provide a vegetation type-based approach to interspecific competition. We integrated predicted climate variables to determine changes in habitat suitability for two forecast periods (2041–2060 and 2061–2080) and three Representative Concentration Pathways.Averaging over all species, the models showed good predictive power, with the quality of BRT (AUC 0.861; RMSE 0.225) surpassing that of SVM (AUC 0.804; RMSE 0.285). We observe that the majority of the species have not yet filled their potentially suitable habitat. An increase in habitat suitability for predicted climatic conditions is implied for most species.Our results indicate that the dynamics of biological invasions will intensify with anticipated climatic changes. Climate factors, soil type and transport infrastructure are of great relevance for the distribution of IPS, while interspecific competition, indirectly assessed through the distribution of habitat types, is only relevant for some species.http://www.sciencedirect.com/science/article/pii/S1569843222003466Boosted regression treesClimate changeInvasive plant speciesRemote sensingSpecies distribution modelSupport vector machines |
| spellingShingle | Fabian Sittaro Christopher Hutengs Michael Vohland Which factors determine the invasion of plant species? Machine learning based habitat modelling integrating environmental factors and climate scenarios International Journal of Applied Earth Observations and Geoinformation Boosted regression trees Climate change Invasive plant species Remote sensing Species distribution model Support vector machines |
| title | Which factors determine the invasion of plant species? Machine learning based habitat modelling integrating environmental factors and climate scenarios |
| title_full | Which factors determine the invasion of plant species? Machine learning based habitat modelling integrating environmental factors and climate scenarios |
| title_fullStr | Which factors determine the invasion of plant species? Machine learning based habitat modelling integrating environmental factors and climate scenarios |
| title_full_unstemmed | Which factors determine the invasion of plant species? Machine learning based habitat modelling integrating environmental factors and climate scenarios |
| title_short | Which factors determine the invasion of plant species? Machine learning based habitat modelling integrating environmental factors and climate scenarios |
| title_sort | which factors determine the invasion of plant species machine learning based habitat modelling integrating environmental factors and climate scenarios |
| topic | Boosted regression trees Climate change Invasive plant species Remote sensing Species distribution model Support vector machines |
| url | http://www.sciencedirect.com/science/article/pii/S1569843222003466 |
| work_keys_str_mv | AT fabiansittaro whichfactorsdeterminetheinvasionofplantspeciesmachinelearningbasedhabitatmodellingintegratingenvironmentalfactorsandclimatescenarios AT christopherhutengs whichfactorsdeterminetheinvasionofplantspeciesmachinelearningbasedhabitatmodellingintegratingenvironmentalfactorsandclimatescenarios AT michaelvohland whichfactorsdeterminetheinvasionofplantspeciesmachinelearningbasedhabitatmodellingintegratingenvironmentalfactorsandclimatescenarios |