Modelling the dispersal of riverine fish larvae: from a raster-based analysis of movement patterns within a racetrack flume to a rheoreaction-based correlated random walk (RCRW) model approach

Abstract

Recruitment of Chondrostoma nasus and similar fish species in rivers is related to spatiotemporal linkages between larval hatching and nursery habitats. Active swimming behaviour contradicts the assumption that passive particle tracing models can serve as a proxy for larval dispersal models. A racetrack flume with an inshore area of near-natural slope was created to observe individual larval trajectories. A new three-step, raster-based analysis was developed to distinguish four types of movement patterns: active upstream, active downstream, active–passive, and passive. Both larval developmental stage-specific and release site-specific occurrences of these movement patterns were experimentally found for nine flow velocity classes (≤0.225 m·s−1). These current-induced movement patterns, and evaluated durations within them, were used to develop a biased and correlated random walk model that includes rheoreaction — a key behavioural response of fish to flow within rivers. The study introduces the concept and application of a rheoreaction-based correlated random walk model, which coupled with a 3D hydrodynamic model, allows prediction of the spatiotemporal effects of various river discharges, morphologies, and restoration scenarios on larval fish dispersal.