||Politikos Dimitrios1, Huret Martin1, Petitgas Pierre2
||1 : IFREMER, STH LBH, F-29280 Plouzane, France.
2 : IFREMER, EMH, F-44311 Nantes 03, France.
||Ecological Modelling (0304-3800) (Elsevier Science Bv), 2015-10 , Vol. 313 , P. 212-222
|WOS© Times Cited
||Individual-based model, Behavioral movement, Dynamic Energy Budget, Anchovy, Bay of Biscay
||Adult anchovies in the Bay of Biscay perform north to south migration from late winter to early summer for spawning. However, what triggers and drives the geographic shift of the population remains unclear and poorly understood. An individual-based fish model has been implemented to explore the potential mechanisms that control anchovy's movement routes toward its spawning habitats. To achieve this goal, two fish movement behaviors – gradient detection through restricted area search and kinesis – simulated fish response to its dynamic environment. A bioenergetics model was used to represent individual growth and reproduction along the fish trajectory. The environmental forcing (food, temperature) of the model was provided by a coupled physical–biogeochemical model. We followed a hypothesis-testing strategy to actualize a series of simulations using different cues and computational assumptions. The gradient detection behavior was found as the most suitable mechanism to recreate the observed shift of anchovy distribution under the combined effect of sea-surface temperature and zooplankton. In addition, our results suggested that southward movement occurred more actively from early April to middle May following favorably the spatio-temporal evolution of zooplankton and temperature. In terms of fish bioenergetics, individuals who ended up in the southern part of the bay presented better condition based on energy content, proposing the resulting energy gain as an ecological explanation for this migration. The kinesis approach resulted in a moderate performance, producing distribution pattern with the highest spread. Finally, model performance was not significantly affected by changes on the starting date, initial fish distribution and number of particles used in the simulations, whereas it was drastically influenced by the adopted cues.
||Access on demand
|Author's final draft