Horizon 1 1 17 ACLN : Articles dans des revues avec comité de lecture non répertoriées par l'AERES Computational Management Science 281-298 GESTION DE L'ENVIRONNEMENT EXPLOITATION DES RESSOURCES NATURELLES DIVERSITE SPECIFIQUE MODELISATION ANALYSE MATHEMATIQUE INCERTITUDE VIABILITE 2013 fdi:010070556 ENG Computational Management Science 1619-697X 4 10.1007/s10287-013-0192-4 https://www.documentation.ird.fr/hor/fdi:010070556 https://www.documentation.ird.fr/intranet/publi/depot/2017-09-12/010070556.pdf 10 Horizon (IRD) This study proposes a new framework to tackle the uncertainty that prevails in the exploitation of renewable resources. It deals with the question of how to guarantee both a minimum multi-species harvest and the renewal of resources when their evolutions are uncertain. The problem is twofold: to decentralize a global constraint (on a multi-species harvest) into local constraints (on the resources of the different species) and, then, to use a "tychastic" approach necessitating only the forecasts of the lowers bounds of the resource growth rates. This study, formulated as a "tychastic" regulated system with viability constraints, departs from stochastic approaches generally used to deal with uncertain situations. It provides the time dependent harvesting rule allowing to always comply with a minimum harvest objective and resources replenishment thresholds whatever happens and a tychastic measure of risk viability in terms of minimum resources initially required. To solve this problem involving global and local constraints a new method that decentralizes the constraints has been devised. An example is presented whose numerical results are obtained thanks to a dedicated software using mathematical and algorithmic tools of viability theory. 082ECOSYS ; 020MATH01