In this paper we present our implementation of a Genetic Algorithm on the BOINC volunteer computing platform. Our main objective is to construct a computational framework that applies to the optimum design problem of prairies. This ecology problem is characterized by a large parameter set, noisy multi-objective functions, and the presence of multiple local optima that reflects biodiversity. Our approach consists in enhancing the iterative (synchronous) master-worker genetic algorithm to overcome the limitations of volatile and unreliable distributed computing resources considering a sufficiently large number of volunteer computers. Though volunteer computing is known to be much less performing than parallel environments such as clusters and grids, our GA solution turns to exhibit competitive performance.