Understanding changes in global landscape patterns is essential for identifying the drivers of land use and land cover (LULC) change and their broader impacts on the Earth System, as human activities continue to drive widespread habitat loss, fragmentation and environmental consequences such as carbon emissions and biodiversity decline. These changes affect both the composition (area) and configuration (spatial arrangement) of ecosystems, with effects that vary across scales, making the choice of scale and landscape metrics critical for meaningful analysis. However, existing global studies have often focused on single LULC classes or entire landscapes, using scale-sensitive metrics that limit comparability and hinder multiscale understanding.  

To address these gaps, а new ForestPaths study, authored by partners University of Edinburgh and Karlsruhe Institute of Technology, alongside University of Aberdeen and University of Southampton, quantifies global landscape patterns for multiple LULC classes and landscape extents using scale-invariant metrics, first validated in Colombia and then applied globally from 1992 to 2020. By capturing both trends and spatial variability in landscape patterns over recent decades, the study provides a comprehensive dataset that can inform future research on the drivers and ecological consequences of landscape change. 

This study used the HILDA+ v2b dataset to analyse global land use and land cover (LULC) changes from 1992 to 2020, covering six LULC classes: urban, cropland, pasture/rangeland, forest, unmanaged grass/shrubland and sparse/no vegetation - at 1 km resolution. To ensure meaningful comparisons across scales, six landscape metrics with predictable, scale-invariant behaviour were identified using Colombia as a test case: class area (CA), Landscape Shape Index (LSI), number of disjunct core area patches (NDCA), number of patches (NP), total core area (TCA) and total edge length (TE). These metrics were then calculated globally across five landscape extents (100–25,600 km²) to quantify changes in both composition and configuration of LULC classes. Analyses focused on mean and variation of metrics over time, net changes between 1992 and 2020 and spatiotemporal patterns at global and continental scales. Particular attention was given to CA and LSI to disentangle changes in area versus configuration, and landscapes were classified into nine categories based on concurrent changes in these two metrics, providing a comprehensive view of global landscape pattern dynamics over nearly three decades. 

From 1992 to 2020, global landscape patterns varied widely across regions and LULC classes, with urban and unmanaged grass/shrubland expanding in area, while forests and pasture/rangelands generally declined and became less fragmented. Changes were spatially heterogeneous, driven by factors such as urban expansion, agricultural shifts and land abandonment. Forest loss in tropical regions threatens biodiversity and carbon storage, whereas pasture-to-grassland transitions may offer ecosystem recovery opportunities. Data uncertainties, particularly for grassland classes, highlight challenges in interpreting trends. Overall, the study underscores the importance of multiple scales and LULC classes in understanding landscape pattern change and provides a global dataset of scale-invariant metrics to support research on environmental impacts and land system dynamics. 

Read the full study here.