Using Correlation Patterns in Schooling Behavior to Distinguish Tetra Species

Fish species are usually identified morphologically or genetically, but group-level kinematics may also provide diagnostic signatures. The advancement in automated tracking technology provides an opportunity to distinguish between collective behavioral signatures and to use those signatures to classify distinct species of schooling fish. In this work, we analyzed videos of schooling behaviors of three tetra species: Black neons, Buenos Aires, and Pristella tetras. We found that Black neon tetras have a higher polarization order parameter, shorter distance to their neighbors, and higher swimming speed than the other two species. We further found that the velocity correlation function of Buenos Aires tetras decays near exponentially up to very large distances while those of the other two species decay near linearly. Our findings show that simple correlation-based metrics can separate species and provide parameters to model these fish species.