全球变化知识资源中心

Aims High mountain pastures are hotspots of biodiversity, but grazing cessation and climate change are causing tall-grass encroachment and expansion of scrublands and forests. As part of biodiversity conservation efforts, grassland variation needs to be investigated at different spatial scales. We aimed to assess the landscape mosaic variation that occurred between 1988 and 2015 in the higher Mediterranean mountains. We investigated the recovery or land-degradation processes related to land use change, the effects of site condition, the impacts on grassland mosaic heterogeneity, and the threats to biodiversity. Location Sibillini Mountains (central Italy), over 1,650 m a.s.l. Methods We used two-step object-based supervised classification on Landsat 5 and 8 satellite images to analyze changes in landscape patterns and vegetation cover on formerly low-intensity pastures, by assessing the Normalized Difference Vegetation Index variation between 1988 and 2015. Twenty percent of the polygons obtained from segmentation were visually interpreted and assigned to five land cover classes. We generated a land use transition matrix and used Fourier Transforms to detect trends in variation of landscape mosaics and fragmentation. Results We observed prominent dynamics of the grassland mosaic leading to the homogenization of its structure through decreasing patch heterogeneity, especially on south-facing slopes. Grasslands shifted from open communities to dense pastures, with a reduction of scree and spread of tall grasses. The former trend could be understood as a recovery process reverting screes to conditions in equilibrium with local landform and climatic features, while the invasion of tall grasses is a land-degradation process that might lead to local species extinction and loss of habitat connectivity. Conclusions Pronounced changes in the large-scale landscape characteristics, mainly due to land use changes, of which scientists and managers of protected areas are not fully aware, are underway in the top mountain sectors of the study area.