ARSF project WM06/16; led by Dr. Patrick Osborne (Centre for Environmental Sciences, School of Civil Engineering and the Environment, University of Southampton); Site: San Marcos / Portugal

Portugal's cereal steppes hold many bird species of conservation concern. Their protection status is precarious, often relying on compensation payments to farmers for maintaining traditional practises. Although management prescriptions are based on information on habitat selection, it is often not appreciated that habitat selection and our knowledge of it are scale dependent. Policy is applied at large spatial scales whereas ecologists think and birds select habitats at (different) smaller scales. The aim of this project was therefore to study habitat selection by three key bird species of different sizes at multi-spatial scales to determine which features are important and how they link hierarchically from fine to coarse resolutions. Accurately located bird data were modelled against habitat data derived from remotely sensed imagery and field studies on 1000m, 100m and 10m grids. AVHRR imagery provided the 1000m data, Landsat TM the 100m data, and CASI co-registered with LiDAR the 10m data. Models were built both separately for each spatial scale and hierarchically to resolve mis-classifications of species occurrences at coarser resolutions and to estimate abundance within presence-absence models. The study was meant to reveal which habitat features are important at different spatial scales, how the spatial resolution of habitat data limits our understanding of habitat selection, and the implications for conservation management.