Majella National Park is located in the central Apennines (Italy), in a vast mountainous area of about 740 km 2. Owing to the complex geological history of the Majella Massif and surrounding areas, it features many different landforms. Woodlands rich in water characterize the wide tectonic depression of Caramanico, which separates the rounded gentle profile of the Majella to the east from the impervious steep slope of the Morrone Mt. to the west. Bare pitted highlands, like lunar landforms, characterize the top of the Majella (i.e., Femmina Morta Valley) shaped by flowing ice that long ago covered the higher parts of the massif. Sedimentary structures and fossil content recovered in carbonates attest to a long period of sedimentation in warm, shallow-marine environments, revealing that the Majella and the surrounding carbonate mountains looked, approximately from 140 to 7 Ma, like the present-day Bahamas and Persian Gulf. The Park hosts at least 95 geosites, some of which (22) are well-known in the international literature because of their scientific relevance. In addition, its natural and cultural wealth, deeply fused together with geoheritage, preserves several features of national and world rarity. Therefore, the Park Authority decided to put forward its territory as a candidate to become part of the United Nations Educational, Scientific and Cultural Organization (UNESCO) Global Geoparks Network.

Question: Can the pattern and pace of spontaneous Fagus forest expansion from 1975 to 2003 be accurately detected with mid‐resolution satellite imagery? Can the historical Fagus expansion be modelled on the basis of environmental predictors? If so, where are the highest probabilities for future Fagus expansion? What are the implications for park management? Location: Majella National Park, Italy, > 1000 m a.s.l.; municipalities of S. Eufemia and Pacentro. Methods: Fagus cover change was detected by overlaying three classified sequential satellite images. Historical Fagus expansion was related to environmental variables using ordinary logistic and autologistic regression models. Fagus expansion probabilities were generated with the best predictive model. Results: From 1975 to 2003 Fagus advanced into abandoned farmland and subalpine pastures from the contiguous, midaltitudinal Fagus forest and from Fagus outliers, at a rate of 1.2 % per year. Substantial spatial and temporal variations in expansion rates were detected. The ordinary and autologistic models based on the single predictor Distance‐from‐Fagus‐1975 forecasted the Fagus expansion well (AUC 0.81 resp. 0.88). Multiple logistic models, including the topo‐climatic and substrate predictors, improved prediction insignificantly. The strong predictive power of proximity to historical Fagus presence is explained by the dispersal biology of Fagus combined with the shading impact of the Fagus canopy at the forest fringe. Conclusion: Decade‐long Fagus expansion patterns might be reliably forecasted by proximity to historical Fagus distribution. Consequences for park management options are outlined.