Mapping and modelling plant species distribution (natives and aliens) on coastal ecosystems using remote sensing data

The preservation and sustainable management of Nature is one of the major challenges of the humanity in the next future. Remote Sensing (hereafter RS) being effective on reporting earth surface status and changes at different scales has a great potential for supporting innovative monitoring tools and methodologies. This research thesis explored the potential of free RS data (e.g. Sentinel-2, PlanetScope LiDAR) for monitoring coastal dune landscapes in the Mediterranean. The thesis, organized in 4 main chapters, addressed several issues related with RS applications on ecological monitoring. The first chapter stressed the usefulness of free RS, specifically LiDAR data, for modelling and mapping the invasion of a woody alien plant (Acacia saligna (Labill.) H. L. Wendl.) on coastal landscapes. Invasive Species Distribution Models (iSDM) are crucial for dealing with biological invasions, and the performed research gave new evidences of RS data usefulness for improving SDMs procedures and results. The second and third chapters tested the potential of RS for mapping coastal dune vegetation through semi-automatic and automatic classification procedures stressing the RS vegetation phenology (second chapter) and the RS yearly variability of three basic elements of coastal dunes landscapes (vegetation, water and bare soil; third chapter). The good temporal, spatial and spectral resolution of Sentinel-2 RS data, allowed to effectively map a dynamic and heterogeneous mosaics like those of coastal dunes . The fourth chapter explored the relationship between field measured and remote sensed biodiversity and tested such relation on coastal dunes. This chapter, gave new evidences concerning the potential of the “Spectral Variability Hypothesis” (SVH) for depicting floristic biodiversity on well preserved and disturbed (e.g. Carpobrotus acinaciformis (L.) L. Bolus, Carpobrotus edulis. (L.) N. E. Br. invaded) coastal dune landscapes. The results presented in this thesis have highlighted several strengths and limits of using RS data for modelling and mapping coastal dune ecosystems.