Dr. Gwladys Lambert^1
1AFSC
May 31, 2016 9:00 (PST): FSH 105

Optimizing otoliths sampling design in fishery-independent surveys for stock assessments

The sampling design of composition data can have a major impact on our perception of the stock status and the estimated reference points for management. Composition data as sampled from bottom trawl surveys are famously clustered which creates major challenges when it comes to estimating length and age composition of the population of interest and in particular with respect to effective sample size/data weighting in stock assessment models. Here I present the approach I have been exploring in order to prepare a simulation framework to compare otoliths sampling strategies and their impact on stock assessment outputs. First I will talk about the questionnaires I have designed as part of the Otolith Sampling Size Working Group that aimed at informing the development of an operating model. Then I will discuss the implications of removing age data in data rich stock assessments based on a couple of examples and I will show some early simulation work using Stock Synthesis and developments from the R package ss3sim. Ss3sim developments include the addition of an option to test for the effect of ageing error on stock assessment outputs. The main focus will be on another option I have been experimenting with which aimed at simulating the sampling of clustered length data (that resemble real survey data) from random or length-stratified sampling in order to investigate the effect of different sampling strategies on the effective sample size estimated in stock assessments. I will also briefly talk about the Stock Synthesis model for GOA Pacific Ocean Perch I have been working on in order to use in the simulations described above.