Comparing Default Movement Algorithms for Individual Fish Avoidance of Hypoxia in the Gulf of Mexico

Abstract

The northern Gulf of Mexico is the site of one of the largest areas of seasonal, coastal hypoxia (up to 22,000 km$^{2}$). Hypoxia can have both direct and indirect effects on fish. Atlantic croaker (Micropogonias undulatus) is a good model organism for studying the effects of hypoxia on fish in the Gulf of Mexico because it is a demersal species that lives in the area where hypoxia occurs and has been studied extensively. Virtual croaker movement was examined for three algorithm groups on a two-dimensional grid encompassing the Gulf hypoxia region. The model was run for seven days using four static dissolved oxygen maps reflecting progressively increasing hypoxia severity. Individual fish movement was modeled using a particle-tracking module with outputs from a three-dimensional hydrodynamic-water quality model for the 2002 hypoxia season. The three algorithm groups included the neighborhood search for hypoxia avoidance and the random walk, Cauchy correlated random walk, or kinesis for the default behavior. The results show that the default algorithms have little effect on hypoxia exposure of individual fish, but affect sinuosity (wiggle in fish path). The variables to consider when choosing between the three default algorithms are time step, dispersal, and the effects of stressors other than hypoxia. This study emphasizes the need to acquire suitable data for calibration of fish movement models that are presently not available for the northern Gulf of Mexico.