Spatially explicit models in local dynamics analysis: The potential natural vegetation (PNV) as a tool for beach and coastal management

Abstract

The concept of Potential Natural Vegetation (PNV) and its mapping have become extremely important within the scope of habitat restoration in almost every European country. The aim of this study is to predict the PNV in the sites of Natura 2000 Network ‘Sado Estuary’ and ‘Comporta-Galé’ based on the vegetation series and the main environmental variables. The modelling approach is based on the distribution of communities referred to as classification then modelling. Subsequently, several statistical model-fitting techniques, such as regression models, machine learning and rule-based, were successfully applied to the survey data (9 vegetation series; and 7 environmental/predictor variables). The spatial database was organized as a Geographic Information System (GIS) and was also used to perform the Species Distribution Models (SDM) at community level. The results show a high correspondence between the vegetation series and the environmental gradients. The predicted PNV maps based on the Maximum Entropy Model were validated with the official map of the PNV of the sites of Natura 2000 Network ‘Sado Estuary’ and ‘Comporta-Galé’, and presented an overall accuracy of 86%. Often, conservation planning and biodiversity resource management is carried out at more detailed scales, where SDM allows integration of community direct observations and improve our interpretation of PNV local distributions along environmental gradients VNP in beach and coastal sand dunes environments.