Effective use of ecosystem and biological knowledge in fisheries

Abstract

North Atlantic (NA) stock assessments address the marine phase, estimating returns to home waters, with Pre-Fishery Abundance (PFA) estimated through raising of national (or regional) annual catches by exploitation rates and attributing unreported catch and natural mortality ranges in Monte Carlo simulations. Baltic stocks in contrast, are estimated through integrated Bayesian life cycle state-space models including riverine and sea phases (Michielsens et al., 2008). There is presently no interaction between the two methodologies.We detail the two approaches specifying similarities in biology, as a prerequisite to their harmonization for parallel inference and risk analysis, independent of scales, available data and management objectives. Through aggregations of scale and availability, assimilations of data differ. For the Baltic much is performed within the forecasting framework, and while aggregations in the NA case are disparate, finer scale details are available. In the Baltic a scale of “river” is used as the geographical unit, while in the NA, 3 geo-regions are treated independently, each operating at arbitrary regional scales. To harmonize NA and Baltic approaches, a multi-scale integrated life cycle model in a Hierarchical Bayesian Modelling (HBM) framework is proposed for the NA to capture inherent complexities from mixing of life cycle age and stage cohorts, which is currently not addressed. A stage-structured life cycle approach is proposed, incorporating freshwater and marine phase variability of life histories (survival and life history choices) and auto-regenerated cohort dynamics. This represents a large change in both the modelling and statistical inference framework.Key structural hypotheses and common informative prior distributions for modelling demographic processes, for both NA and Baltic models are developed. Together with the Bayesian methodology these form the core of the harmonization process. To harmonize modelling of the demographic process the following items are necessary:  State-space representation of all life stages including those not directly observed to explicitly separate out modeling of the demographic and observation processes, so as the harmonization of the models for the core ecological process can be thought independently from the data availability.  Age/stage-based demographic models to integrate biological and ecological knowledge of population dynamics, characterized by seaward migrations of smolts and spawning migration of adults back to freshwater, accommodating intra- and inter-population variability in life history traits.  Probabilistic demographic transitions and between-years variability of certain parameters to capture both environmental and demographic stochasticity.  Variable egg to juvenile density-dependent average survival, of classical survival functions.  Common approach to forecast yearly variations of marine post-smolts survival.