Quercus pyrenaica Willd. is one of the main marcescent oak species of the Iberian Peninsula, forming submediterranean forests of high ecological, biogeographical, and landscape value. In Portugal, many populations occur near the southern limit of the species’ distribution, where increasing climatic stress, land-use change, and rural abandonment may reshape forest structure, connectivity, and long-term persistence. Understanding recent spatial dynamics is therefore essential to support conservation planning, adaptive forest management, and the design of functional ecological corridors.

This study assesses recent spatial changes in the distribution, structure, and fragmentation of closed-canopy Q. pyrenaica forests in north-central Portugal between 2019 and 2024. Forest distribution maps were produced using a Random Forest classifier based on Sentinel-2 time-series imagery, vegetation indices, topographic variables, and recent satellite embedding features. Spatial and landscape metrics were then used to evaluate changes in forest area, patch configuration, core habitat, and connectivity-related attributes.

The classification achieved high predictive performance (Overall Accuracy = 95.1%, Kappa = 83.7%, F1-score = 86.9%), supporting the use of Earth Observation data for operational monitoring of submediterranean oak forests. Results indicate an apparent increase in mapped closed-canopy Q. pyrenaica forest area during the study period. However, this expansion was accompanied by a marked increase in the number of forest patches and a reduction in median patch size, revealing a trend towards landscape subdivision. Although total core forest area increased, the proportion of patches containing core habitat declined, suggesting a growing dominance of small, edge-affected stands. Forest patches remained concentrated at mid-elevations and on predominantly north-west-facing slopes, indicating a relatively stable ecological profile over time.

These patterns suggest that recent forest expansion is occurring mainly through the emergence or canopy closure of small and spatially isolated patches, probably linked to secondary succession after agricultural abandonment. Consequently, gains in forest cover should not be interpreted automatically as gains in ecological integrity, habitat quality, or functional connectivity. The results highlight the need to protect regenerating stands, promote links between fragmented nuclei, and guide management towards the recovery of structurally mature and connected oak woodlands.

Finally, by integrating remote sensing, machine learning, and landscape ecology, this study provides a spatially explicit basis for identifying priority areas for conservation corridors and for supporting decision-making in Mediterranean forest landscapes under ongoing environmental change.

The FCT funded the PhD grant (reference UI/BD/152853/2022 and DOI identifier https://doi.org/10.54499/UI/BD/152853/2022), and the programs CERNAS (UID/681/2025, DOI https://doi.org/10.54499/UID/00681/2025), CEF (UID/00239/2025, DOI: 10.54499/UID/00239/2025, and UID/PRR/00239/2025, DOI: 10.54499/UID/PRR/00239/2025), Associate Laboratory TERRA (LA/P/0092/2020, DOI: 10.54499/LA/P/0092/2020), CEGOT (UIDB/04084/2025), and UID500272025UIDB/50027/2020.