| description abstract | Section contractions in rivers often induce accelerated flows in nature, exerting substantial influence on the swimming behaviors of various fish species. This study focuses on the undulatory swimming strategies of fishes and the effects of hydrodynamics on the swimming behavior in a flume with constriction. By using a combination of computational fluid dynamics (CFD) modeling, convolutional neural network (CNN), and observation of fish behavior with kinematic and morpho-kinematic parameters, we investigate how fish select their swimming strategies when encountering complex flow conditions. The results reveal that fish exhibit distinct swimming behaviors in response to accelerating flows, including avoidance behavior, drifting with the flow, and escaping upstream. Fish demonstrate a preference for swimming on one side of the flume, with the side bank of the adaptation region being the most favored location due to the lower flow velocity and thigmotactic behavior. Additionally, we identify three swimming gaits employed by the fish to navigate complex river environments, adjusting their swimming behavior based on environmental cues. This study highlights the significance of both head and tail characteristics in governing fish swimming behavior, with head features crucial for perceiving the environment and tail movements directly influencing swimming speed and maneuverability. | |
