Modelling the dispersal and mortality of Browns Bank egg and larval haddock (Melanogrammus aeglefinus)

An early life stage model is described with constant mortality for egg and larval stages. The model is used to simulate the stage 4 egg and larval data for haddock (Melanogrammus aeglefinus) in southwest Nova Scotia for the years 1983-1985. The model is initialized using published abundance and mortality estimates for these years, and its output is compared with field data. We find that the model does a systematically poor job of reproducing both the spatial-temporal and area-integrated versions of the data. To understand the discrepancy, we derive an integrated version of the model (integral model) and analyze its properties. This leads to a general method for determining whether sequential stage abundance data is consistent with a stage-dependent constant-mortality model. We use this method to show that a constant-mortality early life stage model is not consistent with the data. The integral model allows for year-day dependent mortality functions, which results in almost perfect fits to the abundance data. These functions can be transferred to the early life stage model with significantly improved model performance, although spatial differences remain. The implications of the integral-model analysis for sequential stage mortality estimation are discussed.