PS01 ECOP-15

Combination of machine learning and partial differential equations on a spatially explicit model for plant invasive species: Real case scenario with Acacia dealbata in the Iberian Peninsula

Monday, July 14 at 6:00pm in

SMB2025 SMB2025 Follow

Share this

Pedro Lago-González

Department of Natural Resources and Environmental Engineering, Universidade de Vigo
"Combination of machine learning and partial differential equations on a spatially explicit model for plant invasive species: Real case scenario with Acacia dealbata in the Iberian Peninsula"
Plant species from the Australian temperate forests were introduced to Europe in the late 1700s, some for economic profit (e.g. Eucalyptus sp.) and others as ornamental plants, as is the case for some species of Acacia. These plants, have been legally recognized as invasive species in various countries including Spain and Portugal. In this work, we develop a model to predict the spread of Acacia dealbata in the northwestern part of the Iberian Peninsula, using a deterministic approach based on partial differential equations (PDEs). To account for seed spread by streamflow, we add an advection term where rivers are present. The spatially explicit carrying capacity is determined by a machine learning method (MaxEnt) using environmental factors such as land use, climatology, soil conditions and occurrence of disturbances. The simulation results compare well with overall observational data in riparian areas, but not elsewhere in the landscape. The predictive accuracy of the model is improved if observational data is segregated according to year. Our model gives us some insight into the connections between Acacia patches, and the sensitivity of intervening areas to invasion. This information can play a key role elaborating strategies for eliminating or mitigating the spread of the species. We hypothesize that our approach will be useful to researchers studying invasion processes in other areas and with other species.



SMB2025
#SMB2025 Follow
Annual Meeting for the Society for Mathematical Biology, 2025.