Developing a Forest Naturalness map for Europe: a case study for Beech dominated forests

European forests represent around 5% of world’s forests and are the result of complex interactions between man and environment over the last thousands of years. Over more than 200 years, the forest cover has steadily increased in Europe. Intensive afforestation and reforestation projects were started in many countries since the beginning of the 1800’ies as a response to the lack of timber resources. In more recent times, urban areas have increased determining a stronger pressure on natural and semi-natural environments. And at the same time, natural reforestation is increasing as a consequence of the abandonment of rural mountain pasture and agricultural lands. Over the last decade a significant effort has been made to estimate the presence of High Nature Value (HNV) farmland in Europe. The concept of HNV farmland ties together the biodiversity to the continuation of farming on certain types of land and the maintenance of specific farming systems. Typical examples include semi-natural grassland systems, traditional olive, vine and fruit production, Dehesa, Montado and extensive farming in bocage landscapes. High Nature Value farmland was adopted as a specific indicator in the SEBI process (SEBI 020: Agriculture: area under management practices potentially supporting biodiversity). So far no similar concept has been developed for assessing the area of High Nature Value forests in Europe. Whatever the definition and the monitoring method adopted, the assessment of forest naturalness is essential to support European environmental protection policy implementation. This development is mirrored in the policy agenda of the EU (Europe 2020, Biodiversity Strategy 2020, 7th EAP). The present work focuses on identifying areas of natural forests, or semi-natural forests that approximate to naturalness through the application of a model based on fuzzy Multicriteria Analysis applied wall-to-wall in Europe with a geographical resolution of 1 km to Beech (Fagus sylvatica) dominated forest. Five variables were tested: i) naturalness of tree species composition, ii) hemeroby, iii) growing stock volume, iv) accessibility, v) connectivity. Different weighted linear combinations were recursively tested using different Monte Carlo simulations and model results were compared with the real locations of old-growth beech forests. This accuracy assessment was carried out applying a Relative Operating Characteristic (ROC) and calculating the Area Under the Curve (AUC) accuracy indicator in order to find the best model able to better predict the presence/absence of old-growth forests. We found that the best combination was obtained with three variables. The AUC for naturalness alone was 0.706, naturalness and accessibility was 0.787 and naturalness, accessibility and connectivity together was 0.809. The study carried out for beech forests demonstrated that some pan-European datasets exist for developing a wall-to-wall spatially explicit multi-criteria analysis of forest naturalness.