Gmacs: A Generalized Size-Based Stock Assessment Modeling Framework

Wednesday, August 20, 2014: 9:00 AM
301A (Centre des congrès de Québec // Québec City Convention Centre)
Athol Whitten , School of Aquatic and Fishery Sciences, University of Washington, Seattle, WA
James Ianelli , Alaska Fisheries Science Center, National Oceanic and Atmospheric Administration, Seattle, WA
André Punt , School of Aquatic and Fishery Sciences, University of Washington, Seattle, WA
Gmacs is a statistical size-structured population modeling framework designed to be flexible, scalable, and useful for data-limited to data-rich situations. Gmacs can incorporate multiple data types from a variety of fisheries or surveys by combing all data in the form of an integrated analysis.

Gmacs has initially been designed for application to red king crab stocks of Alaska. The population sub-model is typical of those represented in most statistical size-structured models. However, the observation sub-model is designed to utilise the wide variety of data types typical of Alaskan crab stocks. Observation data types can include fishery or survey indices of abundance or effort; fishery discard data; size-composition and weight-at-size data; and tag-recapture data. The current framework allows for time-varying selectivity, time-varying natural mortality and growth, and time-varying molting probabilities, with more to come.

Gmacs is coded using Auto-Differentiation Model Builder (ADMB), so inherits its capability to efficiently estimate hundreds of parameters using maximum likelihood. The source code is heavily commented and formally structured, allowing for easy expansion. Gmacs output processing is handled by a purpose-built R package. The Gmacs model code, example model files, and associated R code is publicly available at https://github.com/awhitten/gmacs; new users and developers are welcomed.