Agro-ecosystems are ecological systems subject to human interaction whose simulation is of interest to several disciplines (e.g. agronomy, ecology and sociology). The agent-based modeling approach appears as a suitable tool for modeling this kind of complex system, along with a corresponding agent-oriented software engineering (AOSE) methodology for the construction of the simulation. Nevertheless, existing AOSE methodologies are general-purpose, have not yet accomplished widespread use, and clear examples of applications to agro-ecosystems are hard to find. This article sets the ground for an AOSE methodology devised specifically for developing agro-ecosystem simulations. The methodology framework is based upon other general-purpose AOSE methodologies, and it relies on the Unified Modeling Language for an easy uptake from interdisciplinary teams. As a first proof of concept, it is applied to a real case study: the evolution of the strategies followed by cattle producers of the basalt-region of north Uruguay against severe draughts.