We present a model that simulates the foraging behaviour of tunas in the vicinity of ocean fronts. Stochastic dynamic programming is used to determine optimal habitat choice and swimming speed in relation to environmental variables (water temperature and clarity) and prey characteristics (abundance and energy density). By incorporating submodels for obligate physiological processes (gastric evacuation, standard and active metabolic costs) and sensory systems (visual feeding efficiency), we have integrated into a single fitness‐based model many of the factors that might explain the aggregation of tunas at ocean fronts. The modelling technique describes fitness landscapes for all combinations of states, and makes explicit, testable predictions about time‐ and state‐dependent behaviour. Enhanced levels of searching activity when hungry and towards the end of the day are an important feature of the optimal behaviour predicted. We consider the model to be particularly representative of the behaviour of the warm‐water tunas or Neothunnus (e.g. skipjack, Katsuwonus pelamis, and yellowfin, Thunnus albacares) and for surface‐dwelling temperate tunas (e.g. young albacore, Thunnus alalunga), which are often observed to aggregate near fronts. For the bluefin group (i.e. older albacore; northern and southern bluefin, Thunnus thynnus and Thunnus maccoyii), for which extended vertical migrations are a significant and as yet unexplained component of behaviour, the model is able to reproduce observed behaviour by adopting the lower optimal temperature and standard metabolic rate of albacore. The model cannot explain why physiological differences exist between and within the different tuna species, but it does show how differences in susceptibility to thermal stress will permit different behaviour.

No abstract

Seamounts are habitats of considerable interest in terms of conservation and biodiversity, and in terms of fisheries for bentho-pelagic and pelagic species. Twenty datasets on seamounts and bathymetry from different sources and scales (from individual cruise to worldwide satellite data) have been gathered to compile a detailed list of seamount features for the Western and Central Pacific Ocean. None of the datasets is complete and errors exist in most of them. The Kitchingman and Lai (2004) dataset (KL04) from satellite altimetry data provided the baseline of this study because it covered the entire region of interest and includes depth information. All KL04 potential seamounts were cross-checked with other datasets to remove any atolls and islands incorrectly classified as seamounts, to add seamounts previously undetected by KL04, to update the overall database (geolocation, depth) and provide a 12-classes typology of the different types of underwater features. Of the 4,132 KL04 potential seamounts identified, 835 (20%) were actually atolls and islands and 268 were multiple identifications of the same feature (e.g. multiple peak seamounts) and 2 were removed, leaving 3,027 actual underwater features. Conversely, 541 seamounts documented in other datasets but not registered in KL04 were added. The screening of all the potential WCPO seamounts produces a list of 3,568 features with accurate position and information that should have many applications in fisheries and oceanography.

No abstract