A model of salmonid planktivory: field test of a mechanistic approach to size-selection

A mechanistic model was developed to simulate the planktivory process for the pelagic lake herring (Coregonus artedi) on common crustacean zooplankton of Lake Superior. The model was constructed as a product of probabilities associated with detection, attack, capture and retention. Sensitivity analysis showed that these coefficients were robust and that initial zooplankton density affected the model behavior more than altering the coefficients. The model output provides percent composition of adult lake herring stomach contents. Zooplankton communities from several stations and seasons in Lake Superior were sampled to provide an initializing zooplankton community for the model. Model output was then contrasted with simultaneously field caught lake herring stomachs. The model predictions agreed with the observed stomach contents at an average accuracy of 80%. The overall agreement of the predictions with observed stomach contents is suggestive that the mechanisms modeled track planktivore feeding dynamics adequately.