This course is aimed at anyone who desires improving accuracy and precision of the data inputs to fishery stock assessments and/or individuals directly involved in the design and implementation of CPUE/LPUE time series data analysis or their use in stock assessments.
Objective:
Annual indices of stock abundance based on catch and effort data are central to many fisheries stock assessments. It has become more common in recent years to use advanced statistical methods to standardize catch rates against explanatory variables as a means of adjusting indices for unequal sampling over space or time or habitat. Commonly used methods include general linear models (GLMs), generalized linear mixed models (GLMMs) for non-normally distributed data, delta – lognormal or delta – gamma GLMMs, and generalized additive models (GAMs).
This course is an introduction to these model-based approaches for estimating annual CPUE indices from sampling programmes. The methods to be covered are relevant for standardization of catch and effort in both commercial catch data and fishery independent monitoring surveys. The stocks for which such methods are appropriate include some data-rich stocks where fishery-independent survey data are available to accurately track stock abundance, stocks taken in mixed fisheries (i.e. not targeted) and some data-limited stocks for which fishery-dependent abundance indices provide the main source of information on stock trends. The course will include an introduction to statistical modeling beyond the usual models based on assuming normality of the data, discussion of the effect of sampling strategies on model development and a review of advantages and limitations of standardization. Participants will learn the importance and practical application using real fisheries data examples. All examples will use R: A language and environment for statistical computing (R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/).
At the end of the course, the participants should be able to:
- understand the basic concepts related to standardization methods using model-based approaches
- identify advantages and disadvantages of model-based standardization
- identify approaches for selection of the explanatory variables on which standardization is based
- identify the role of the sample selection mechanism in a model-based approach
- identify the distinctions among the various commonly used model forms, including the effect of assuming alternative probability distributions for the catch data
- perform some model comparisons to identify the “best" fitting model given the data
Level:
Participants should be familiar with current methods for collecting and analyzing data by Member Countries for input into stock assessment models. In addition, participants should be familiar general statistical approaches for modeling such as regression, analysis of covariance, and general linear models. In addition, participants should be comfortable working with the R language. Please contact the instructors if you desire more detail about the course.