Exploring how land use and the management of agro-ecosystems may evolve in the future is important for advancing scientific understanding and for informing policy makers and land managers of ways to respond and adapt sustainably to future change. In this paper, we investigate the future land-use trajectories of a semi-arid Mediterranean agro-ecosystem in south-eastern Spain using two different approaches: Markovian cellular automata and an agent-based approach. Methodologically, the paper offers a systematic method for agent parameterization to facilitate the development of an empirical agent-based scenario analysis. This approach is achieved by integrating information from cadastral and recent land-use maps, agricultural statistics, and sampled survey data. Through this integration, an effective approach is provided for up-scaling an agent typology from the sampled survey to the landscape scale. The output of the up-scaling provides a basis for modeling the aggregate effect of the responses of different types of farmers to environmental and policy changes across the study region. Empirically, the paper highlights the contrasting futures that the studied agro-ecosystem could have depending on the direction and intensity of the changes in environmental and policy conditions. The Markovian-CA land-use projection indicates further decline of rain-fed agriculture and describes the scope of the expansion of irrigated agriculture. The agent-based scenario analysis shows that the future of irrigated agriculture is highly sensitive to the expected future water scarcity. The analysis also reveals that the way that the future environmental and policy changes are conceptualized and presented to farmers and the range of different farmers in the agro-ecosystem determine the extent of the resulting aggregate effects of individual farmer reactions to future changes at the landscape scale. The empirical evidence of this research emphasizes the need for policy makers to consider multiple and interacting factors, including the design of interventions and likely farmer responses, which shape future agricultural land-use trajectories.