The management of multiple-use reserves is challenging due to trade-offs between the conservation of natural capital, the provision of different ecosystem services (ES) and the capture of its benefits, as well as a poor governance. In this context, the potential of serious games for simultaneous training of decision-makers and informing researchers is of special interest. Here we present MARCHI, a serious computer game inspired by a MAB Biosphere Reserve from Argentina, through which we evaluated player's preferences for investing annual budgets in different management instruments, and player's ability to feedback their investing decisions on the outcomes from previous games. The main objective for MARCHI players is to maximize the sustainable capture of benefits from natural capital under unpredictable changes in the rate of natural capital loss. Different social actors played the game in their simulated role of members of a management committee. Each game comprises 15 consecutive runs, and each run is an opportunity for players to allocate limited annual funds to Monitoring and Prospective, Control, Payment for ES (PES), Access to ES, and Land Use Planning. MARCHI was able to induce significant and relevant changes of initial preferences for conventional conservation instruments (Control) towards instruments that are little known and practically not applied in the country, such as the PES, or still poorly prioritized in the context of protected natural areas like the access to the ES. Mean Learning Index, an indicator of players ability to improve their game scores along successive games, was positive and significantly different from zero (18.29% ± SE = 4.46%). Final players' performance was not related to their allocation of time to reviewing tutorials, but to their time spent with a review screen after each game. This study illustrates the utility of serious games as a research-action tool for the participatory governance of ES.