Invasions by alien species are among the most important threats to biodiversity, ecosystems and human well-being. The extent of their negative impacts demands a considerable effort to monitor the invasion status and territorial susceptibility to implement prevention and mitigation measures. Remote sensing (RS) is an important tool for large-scale ecological studies, gathering information in a constant, consistent and repeatable manner over large areas. Therefore, in recent years RS has been applied as an efficient approach to assess
and monitor the dynamics of invasive plant species, including their expansion along the roadside. This study aims to describe the spatiotemporal distribution patterns of several invasive plant species (Acacia dealbata, A. melanoxylon, Robinia pseudoacacia, Ailanthus altissima, and Arundo donax) along the roadside, in one of the main transport/energy corridors that links Portugal to Spain. Our goal is to develop a methodology that can support preventive protocols and invasion control measures to be adopted in this kind of infrastructures. We analysed a set of multi-temporal aerial photos from 1995, 2010 and 2016. Aerial photos for 2016 were acquired in the framework of the project Life LINES - “Linear Infrastructure Networks with Ecological Solutions”. We conducted a field survey to
obtain training data of the invasive plant species along the roadsides of the study area with the aid of a real-time kinematic (RTK) GPS receiver. The aerial photos were segmented using the multiresolution algorithm and an object-oriented classification (Nearest Neighbour classifier) in eCognition Developer software. We conducted two sequential classifications. The first classification was used to eliminate the objects of the image without interest for the study of the invasive plants. The second classification was performed to identify invasive plant species. The classification accuracy was assessed using a confusion matrix and the overall accuracy (OA) and Kappa Index of Agreement (KIA) metrics. To analyse the expansion of the invasive species along the roadside, we built a model based on
the maximum entropy approach (MAXENT). Finally, we identified areas that are more susceptible to invasion to assist prevention, detection and early intervention directed to control/elimination, as well as to produce an updated distribution of these invasive species. In general, the cover of invasive species increased in the study sites between 1995 and 2016. In the last six years (2010-2016), A. donax expanded more than the other invasive species. During this period, some invasive trees were cut along the roadside, suggesting localised management. The probability of expansion of invasive species along the roadside is reduced when there is control of the ditches, with an exception for A. donax. In conclusion, in this study, we could observe how invasive species expanded along the roadside between
1995 and 2016. The use of species distribution modelling will help to assess the susceptibility of a territory to invasion. In the light of maintenance effectiveness, RS is a good approach to help in the development of well-planned management of invasive species that spans from anticipation/prevention to control/confinement and even local eradication.
Spain
