Object-Oriented Design of MAS, a Metapopulation Assessment System

In this talk, we introduce a new object-oriented metapopulation assessment system (MAS) that incorporates up-to-date modeling capabilities, such as management strategy evaluation analyses, and is readily extensible for specific stock assessment applications using either frequentist or Bayesian estimation approaches. The object-oriented design allows for adaptation and reuse of existing code. A key design feature is the inherent spatial structuring of the population dynamics and fishery system. Spatial structure can have substantial impacts on the interpretation of fishery selectivity patterns, relative abundance indices, and age or size composition information. In MAS, each instance of a population model consists of one or more structured subpopulations with movement dynamics among areas where each subpopulation has a characteristic habitat area with geo-referencing capability to account for spatial scale and pattern. Information on age or size structure, life history parameters, and habitat is part of the primary subpopulation class object. Other primary classes are the observation, environment, and information analysis classes, which characterize assessment input data, biotic and abiotic environment forcing, and model construction and evaluation processes. Another key element is a standard GUI-based model development interface tool in which model selection, validation, and uncertainty quantification can be investigated and addressed within MAS.