摘要:Co-adaptive management of biodiversity is largely based on a collective learning process. This collective learning concerns “instrumental policy learning,” “social policy learning,” and “political learning.” This paper focuses on instrumental policy learning that has been launched in four West African biosphere reserves. It is based on a MAB-UNESCO/UNEP-GEF programme concerning the co-construction of interaction indicators (between development and conservation), inspired by the Integrated Natural Resource Management (INRM) methodology. Using this process, we were able to test conventional Pressure-State-Response indicators, highlight their limitations, and develop new indicators starting from stakeholders’ stories and perceptions. These new indicators can also be tested through collective restitutions and simulations. We also discuss: a proposed framework for producing interaction indicators that are relevant to all stakeholders and enjoy a certain legitimacy; the importance of an ecosystem services approach to support discussions on biodiversity conservation; opportunities for using the indicators in an interactive, decentralized way at the ecosystem scale through simulation models; the costs of collecting, processing, and maintaining these interaction indicators, and how these costs may be offset using local knowledge.