Publications des agents du Cirad


Modeling livestock farmers' strategies in the Uruguayan Pampa

Grosskopf H.M., Bommel P., Tourrand J.F.. 2005. In : Zerger A. (ed.), Argent R.M. (ed.). Advances and applications for management and decision making. Proceedings of the MODSIM 2005 International Congress on Modelling and Simulation, 12-15 December 2005, Melbourne, Australia. s.l. : MSSANZ, p. 2340-2345. International Congress on Modelling and Simulation, 2005-12-12/2005-12-15, Melbourne (Australie).

Rangelands are an important ecosystem as they occupy nearly fifty percent of the emerged earth. They are used mainly for cattle production. Uruguay is a country where cattle were introduced by Europeans before their settlement in 1728. High potential rangelands, with rather fertile soils and more than a meter of average annual rainfall boosted the presence of the introduced beasts. Since then, the cattle industry has been very important for the local population, especially from the point of view of international trade. There have been many proposals tending to increase productivity in rangeland areas. Nowadays, the industry faces new challenges and institutions must adapt themselves in order to facilitate learning and improving the actual situation. The difficulties are that we face complex systems, that there are many points of view and a growing group of stakeholders. We are trying to design decision supporting tools which potentially can improve collective decision making, through the construction of a shared comprehension. For this we admit that it is important to develop whole farm models. The fundamental processes ongoing within the farm household should be modeled in order to simulate and explore the consequences of the dynamics of different interacting systems components. In order to deal with this problem we have developed a multi agent system, named Arapey, which mimics the evolution of different kinds of livestock farmers in the last thirty five years. The differences among the livestock farmers are their financial and stocking rate strategies. There are some unexpected results. We have presented this model to many livestock farmers and to other persons related to the livestock industry, to test if we can accept that it proposes new interesting aspects of farm management. Our experience lets us suggest that multi agent systems are able to propose new insights, which could act supporting private or collective action. We conclude that by using multi-agent systems it is possible to construct a model taking into account different points of view and that modeling and simulating make learning faster. We propose that these models coupled with Unified Modeling Language activity diagrams' are appropriate to describe systems dynamics and to support the construction of shared comprehension as shown by the corroboration of the model's functioning and of its results by a variety of stakeholders.

Mots-clés : Élevage; modèle; uruguay; système multiagents; stratégie paysanne

Documents associés

Communication de congrès

Agents Cirad, auteurs de cette publication :