Résumé
This paper focuses on understanding human impact on landscape. Both ecological and human practices are analysed as interacting processes. An agent-based model integrating biological and historical knowledge is used to analyse the pattern of Scots Pine encroachment in a French Mediterranean upland. In the STIPA model, pine trees are autonomous agents and a cellular automaton simulates land-use. We test the effects of shifting cultivation on tree establishment at the landscape scale. This allows us to understand how agro-pastoral practices patterned this area from the 17th to 19th century: simulations show the importance of shifting cultivation in limiting woodland progression. Fallow duration linked to environmental heterogeneity is a significant factor for explaining pine dynamics and landscape patterning at the scale of the study region. We put this result in perspective with current rangeland management policies that often consider grazing as the most relevant tool for open landscape maintenance. Our results also show the importance of taking into account time-scale effects when linking landscape patterns to agricultural systems.