Cardinale, Massimiliano
- Department of Aquatic Resources (SLU Aqua), Swedish University of Agricultural Sciences
Conference paper2019
Winkler, Henning; Walter, John; Cardinale, Massimiliano; Fu, Dan
This paper presents a multivariate lognormal (MVLN) Monte-Carlo approach to produce Kobe phase plots and Kobe II projection matrices for range of fixed catch scenarios from the 2018 Indian Ocean yellowfin tuna reference grid of Stock Synthesis models. First, we present Kobe-phase plots for the current stock status that compare within-model uncertainty estimates for a single reference case model to the structural uncertainty estimates from a reference grid of 24 models. The Kobe phase plot results portrait a more pessimistic stock status for the reference case model (94.3% overfished) compared to the uncertainty grid of 24 Stock Synthesis model configurations (83.9% overfished), which captures a wider range of plausible outcomes along SSB/SSBMSY axis. Projections were conducted based on the 2018 reference grid models for fixed catch scenarios, ranging from 60-120% of the 2017 catch. These projections predict that a 20% reduction of current catches would is required to achieve MSY-based targets by 2027 and a reduction by at least 15% is required to prevent a severe stock collapse by 2024. Our results generally support previous findings that structural uncertainty across models is often more important to capture than the often narrower within model uncertainty, given that the grid comprise an adequate range of plausible alternative configurations of the reference case model. A potential advantage of the MVLN approach over the bootstrap and MCMC routines is that it reduces the computing time, thereby enabling rapid generation Kobe phase plots for advice during typically time constraint assessment meetings.
Stock Synthesis, inverted Hessian, model grid, delta-method, covariance matrix
Working Party on Tropical Tunas (WPTT); IOTC-2019-WPTT21-51; October 2019; https://iotc.org/documents/WPTT/21/51
Fish and Aquacultural Science
https://res.slu.se/id/publ/103257