Modeling biotope distributions is of paramount importance to monitor species habitats and guide conservation and restoration actions to decrease population extinction rates. However, modeling biotopes as independent landscape units, as is current practice, has some limitations. Vegetation communities that define biotopes evolve through different stages and associations until they reach an equilibrium. To consider these temporal dynamics, we developed a modeling approach based on potential natural vegetation (PNV) corresponding to ecological contexts supporting vegetation succession. The assumption made is that modeling PNV better distinguishes biotope ecological niches, improving prediction accuracy. Results of the final prediction map were excellent, with an overall accuracy of 0.95 and a kappa coefficient of 0.91. The proposed method was also compared with a classic biotope model and our approach showed 29% mean improvement in accuracy. Our results produced a good distinction between the different ecological niches of potential natural vegetation. However, some areas of confusion were identified but these are mainly explained by imprecision and incompleteness of the reference biotope dataset and long-term human management. Using potential natural vegetation is therefore recommended for further studies on biotope mapping.
Consultez la notice complète de l’article sur ORBi