Resumo

GUTIERRES, Francisco et al. Modelling the Potential natural vegetation in the Loures Municipality (Lisbon Metropolitan Area). Finisterra [online]. 2015, n.99, pp.31-62. ISSN 0430-5027.  https://doi.org/10.18055/Finis3146.

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 Loures based on the vegetation series and the main environmental variables. The modelling approach is based on the distribution of communities referred to as classification-then­modelling. Subsequently, several statistical model-fitting techniques, such as regression models, machine learning and rule-based, were successfully applied to the survey data (vegetation series; and 8 environmental/predictor variables). The spatial database was orga­nized as a geo graphic 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 Loures, and presented an overall accuracy of 88%. 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

Palavras-chave : Potential natural Vegetation; vegetation series; spatial modeling and land use planning; Loures; Portugal.

        · resumo em Português | Francês     · texto em Português     · Português ( pdf )