A Multi-Resolution Approach to Bridge Scales in Biodiversity Monitoring using Unmanned Aerial Vehicles (UAVs)

Land-use change and the expansion of socio-ecological barrier networks are major drivers of biodiversity loss in agricultural landscapes. Because these processes operate across spatial scales, biodiversity monitoring requires methods capable of integrating indicators across scales. This study aimed to fill scales gaps in biodiversity monitoring using Unmanned Aerial Vehicles (UAVs), which quantitatively provides a pathway to map and model vegetation ecophysiological traits as well as to assess habitat structure, corridors, and barriers, thereby supporting ecosystem integrity and functioning, offering insights into the complex interactions shaping agricultural landscapes.

Developing a methodological framework grounded in landscape ecology principles, this study integrates multiple spatial scales into biodiversity monitoring. As part of this framework, a multi-scale UAV-based approach (RGB and multispectral imagery) was applied to link fine-scale vegetation dynamics with landscape-level biodiversity patterns in an extensively managed farmland landscape in the East Havel region (Brandenburg, Germany). UAV imagery was processed to derive ecological indicators including vegetation indices (NDVI, NDSI), canopy height model (CHM), Shannon diversity index (SHDI) , patch density, patch connectivity, texture metrics, and habitat maps across nested spatial scales (6m, 50m, and 120m UAV flight altitudes) to assess vegetation structure, habitat heterogeneity, and ecological complexity vary with scale.

UAV-derived metrics revealed clear scale-dependent patterns. At 6m altitude, individual vegetation species were identified, while at 50m and 120m, structural metrics were quantified to link habitat features to vegetation composition. Fine-scale analyses captured small elements, such as flowering patches and micro-topography, important for insects, whereas landscape-scale metrics highlighted fragmentation and semi-natural elements as biodiversity reservoirs. Some indicators, like CHM and NDVI variability, were consistent across scales, while others, such as patch density and texture entropy, shifted with resolution.

These multi-scale relationships demonstrate how UAV-based remote sensing can fill critical gaps in understanding vegetation and wildlife interactions. Overall, this multi-scale approach bridges theoretical principles of landscape ecology with practical biodiversity assessment. By uncovering structural linkages between vegetation traits, habitat configuration, and scale-dependent ecological patterns, the framework provides a transferable tool for landscape ecologists and conservation planners. Our methodological framework can also help to identify thresholds between land-sharing and land-sparing approaches in the landscape planning.