Applied and theoretical considerations for constructing spatially explicitindividual-based models of marine larval fish that include multiple trophic levels

Individual-based modelling (IBM) techniques offer many advantages for spati ally explicit modelling of marine fish early life history. However, computa tionally efficient methods are needed for incorporating spatially explicit circulation and prey dynamics into IBMs. Models of nutrient-phytoplankton-z ooplankton (NPZ) dynamics have traditionally been formulated in an Eulerian (fixed spatial grid) framework, as opposed to the pseudo-Lagrangian (indiv idual-following) framework of some IBMs. We describe our recent linkage of three models for the western Gulf of Alaska: (1) a three-dimensional, eddy- resolving, wind- and runoff-driven circulation model, (2) a probabilistic I BM of growth and mortality for egg and larval stages of walleye pollock (Th eragra chalcogramma), and (3) an Eulerian, stage-structured NPZ model which specifies production of larval Pollock prey items. Individual fish in the IBM are tracked through space using daily velocity fields generated from th e hydrodynamic model, along with self-directed vertical migrations of Pollo ck appropriate to each life stage. The NPZ dynamics are driven by the same velocity, temperature. and salinity fields as the pollock IBM, and provide spatially and temporally varying prey fields to that model. The resulting p rey fields yield greater variance of individual fish attributes (e.g. lengt h), relative to models with spatially uniform prey, Practical issues addres sed include the proper time filtering and storage of circulation model outp ut for subsequent use by biological models, and use of different spatial gr ids for physical and biological dynamics. We demonstrate the feasibility an d computational costs of our coupled approach using specific examples from the western Gulf of Alaska.