Collective transitions of schooling fish in bi-chamber domains

Fish often congregate into cohesive groups to navigate environments efficiently and safely. Most computational models focus on the collective patterns exhibited by cohesive schools in unbounded domains, such as polarized schooling and rotational milling. In contrast, less is known about how fish schools navigate complex environments. Yet, geometric constraints are common in natural habitats, making it essential to incorporate confinement into mathematical models of collective behavior. Here, we investigate how fish schools behave in a bi-chamber domain connected via a narrow channel. We combine boundary element methods for flow interactions with behavioral rules inferred from empirical data of fish swimming in shallow-water tanks. For a single swimmer, the interplay between wall-avoidance and noise enables transition through the small gap. At the collective level, social behavior rules enhance these collective transitions. Collective hydrodynamic interactions enhance these transition rates further with increasing number of swimmers. Our models and results pave the way towards understanding survival strategies and collective behavior in challenging environments.