Care and Feeding of Virtual Marine Fish Larvae

Mathematical models of foraging behavior and growth physiology are useful tools for the study of marine fish larvae. Our model “Quirks”, for example, predicted over half of the variability among 53 growth rates published in 17 empirical studies of young European anchovy, Atlantic cod, Atlantic herring, and European sprat larvae. A major challenge in applying such models to study larval growth and survival is a lack of detailed information regarding in situ prey fields. Empirical plankton datasets offer limited spatial and temporal coverage and plankton estimates obtained from biogeochemical models generally lack realism and detail regarding the size distribution of potential prey. Working with both empirical (e.g., continuous plankton recorder) and modeled (e.g., ECOHAM4) North Sea data, we developed novel statistical methods to a) objectively compare haphazard survey data with gridded model output b) provide complete (gridded) estimates of prey biomass from incomplete field observations, via imputation, and c) generate dynamic prey-size distributions from modeled bulk carbon estimates of phytoplankton and zooplankton groups, based on size spectrum theory. We apply these tools to hindcast larval fish prey fields (and physical features of the environment), simulate larval fish growth using Quirks, and test whether the results are predictive of recruitment indices.