This study examines the landscape transformations in the Monti Prenestini area of Lazio, Italy, from 1990 to 2023. It analyses changes in land cover dynamics and spatial structure using a landscape ecology approach. The study area covers approximately 229 km² and is ecologically significant due to its diverse habitats, which include forests, shrublands, grasslands, and agricultural lands. In recent decades, anthropogenic pressures such as urban expansion and agricultural abandonment have altered the landscape's composition and configuration, affecting biodiversity and ecosystem services.

Land cover data were obtained from the Copernicus CORINE Land Cover datasets covering the years 1990 to 2018. For the year 2023, land cover was reconstructed using the Random Forest machine learning algorithm, which was applied to Sentinel-2 satellite imagery, topographic data, and ground truth points. All land cover maps were reclassified into seven thematic classes to ensure consistency and comparability.

Spatial metrics were calculated using FRAGSTATS 4.2 to quantify fragmentation, connectivity, and landscape heterogeneity at both class and landscape levels. The results reveal a progressive transformation of land cover types, with significant forest expansion and the disappearance of natural grasslands by 2018. Between 2018 and 2023, notable increases in the Largest Patch Index and reductions in diversity metrics (e.g., Modified Simpson’s Diversity Index) were observed, suggesting increasing dominance of fewer land cover types. Artificial surfaces and permanent crops have shown a marked spatial expansion in recent years, while agricultural mosaics and transitional shrublands experienced reductions in area and connectivity.

Overall, the study highlights how land cover change patterns in the Monti Prenestini reflect broader socio-environmental trends, such as land abandonment and urbanisation. The integrated use of remote sensing, landscape metrics, and machine learning provides a robust framework for monitoring ecological resilience and informing sustainable landscape management strategies.