No abstract

No abstract

An object‐orientated, two‐dimensional, cellular automata (CA) model is developed to describe and predict the schooling behaviour of fish in general, with Norwegian spring‐spawning herring, Clupea harengus L., being used as a case study. The CA model is applied to visualize internal school dynamics based on individual decision rules. Several antipredator strategies, such as split, join and vacuole, performed by schools during predator attack, are visualized in the model. The primary driving force of individual fish is based on simple attraction rules. The model includes stochastic elements which assume that individual herring do not have perfect information about their surroundings. Isolation of individual fish from a school during predator attack is also predicted by the model. The disruption of highly organized fish schools, followed by an attack on solitary herring individuals, may be an important tactic for predators feeding on schooling prey. The conceptual CA model identifies patterns and mechanisms both within and between schools that may be important in all schooling fish. Model simulations are compared with observed predator–prey interactions between killer whales, Orcinus orca L., and herring in northern Norway.

Search by olfaction is common in many aquatic animals, and this feature is exploited by the fishing industry, which has a long tradition in the use of longlines, pots, and other kinds of baited gear. Here we discuss a range of possible search strategies that fish might apply when searching for prey; this in order to improve our understanding of fish movement dynamics towards baited gear. Various search strategies were investigated using an individual-based behavioural model. The search phase was divided into plume-search (search for relevant stimuli) and bait-search (search when an olfactory stimulus has been encountered). The search strategies were evaluated based on their efficiency in providing guidance to the goal (plume or bait). The model was developed based on previous tagging studies of cod (Gadus morhua L.). The results for plume-search show that when the landscape is considered to be continuous, strategies based on moving at an angle against the current performed better than strategies moving straight into the current, or ''random walk''-based strategies. When it is assumed that the fish is constrained to a home range, the results are reversed so that ''random walk''-based strategies perform better than the ''counter current'' strategies. For bait-search the ''counter current'' strategies performed much better than strategies based on gradient-search, which rarely resulted in contact with the bait.