Mixed agro-silvo-pastoral landscapes are typically of significant natural, economic and cultural value. The montado (wood pasture) dominated systems of southern Portugal are a prime example, and experience high rates of land cover change. Tracking these changes is of importance for landscape level conservation and management. Whilst satellite remote sensing is the most realistic approach at large spatial scales, heterogeneous landscapes can present challenges to the accurate classification of land cover and its change.

In this investigation we demonstrate a novel approach to the investigation of land cover change over a 25 year period (1984-2009) in an area of importance for the conservation of the Iberian lynx, Black vulture, and other threatened biodiversity in south-east Alentejo, Portugal. We apply a Tasselled Cap Transformation (TCT) to Landsat imagery from these two years, and then Change Vector Analysis (CVA) on the transformed data to highlight areas of vegetation gain and loss during the intervening period. Using a 2009 land cover classification, and a set of rules based on these vegetation changes, we then predict the land change over the 25 year period focussing on predominant classes of vegetation physiognomy. The results are discussed in terms of probable drivers, as well as implications for biodiversity and other landscape values.