The purpose of this study was to investigate the affordances of participatory simulations by comparing students' models of engagement and science learning outcomes in a multi-team participatory simulated game (MPSG) and a single-team participatory simulated game (SPSG). Two versions of a mobile-based game about marine fishery management were created. Participants were 105 seventh-grade students in Taiwan. Research instruments included a Science Game Engagement questionnaire and a Marine Ecosystem and Sustainability Test. Students' interviews and classroom videos were also collected. Students' models of engagement were analyzed by using Partial Least Squares Structural Equation Modeling, and comparisons of subscales of engagement and learning outcomes were made by Analysis of Covariance statistics. The results showed that the students in the MPSG group had a higher level of behavioral engagement and better learning achievement than the students in the SPSG group did. Also, in exploring the interrelationships among the subscales of students' engagement, we found that, in the SPSG group, students' behavioral engagement was positively predicted by their emotional engagement, while in the MPSG group, it was positively predicted by their social engagement. In both groups, behavioral engagement predicted cognitive engagement. This finding was supported by the video data in that more interactions among the group members were observed in the MPSG group. This indicated that the multi-agent, dynamic modeling in the MPSG may have shifted the quality of the learners' social interactions. The affordances of participatory simulated games are discussed, and future research directions are provided.