Review and assess stock assessment methods used in Australia

Project number: 2014-039

Project Status:

Completed

Budget expenditure: $152,339.00

Principal Investigator: Richard Little

Organisation: CSIRO Oceans and Atmosphere Hobart

Project start/end date: 4 Jan 2015 - 29 Jun 2016

Contact:

FRDC

Need

Stock assessment is a set of tools and methods generally used to assess the status of wild capture fisheries stocks. They range from complex statistical and mathematical models, to simple, almost back of the envelope, methods. They are used to predict population size, quantify the impact of fisheries on the population and in some jurisdictions, provide key outputs needed in harvest strategies. There is a diverse range of methods in a field where practitioners have tended to produce home-grown tools in their favourite code languages (R, Fortran, C++, Visual Basic, ADMB etc.). The use of a specific model or method is often historical rather based on an objective evaluation of options e.g. the risk-cost-catch framework (see Method references). In recent years changes have occurred allowing some shift away from previous approaches:
• More off-the-shelf methods, with a diverse range of flexible features, have become available and some uptake has occurred e.g. Stock Synthesis (SS) (http://nft.nefsc.noaa.gov/SS3.html)
• There has been some convergence of language tools using the open source model (e.g. ADMB, Gnu and R)
• Stock assessment tool kits have become freely available e.g. the NOAA fisheries toolbox (http://nft.nefsc.noaa.gov/index.html)

However, in many cases it is still standard practice in Australia to develop home-grown models. Although this is not in itself an issue, it does not always allow for synergies and more cost effective practices. For example, it has become standard practice in the USA to have a model developed and maintained by a team, have it independently tested and then made available as an off-the-shelf GUI driven tool. Many stock assessment scientists now use these tools. In Europe, ICES also tends to use standard approaches.

There is a real need for a more strategic view of which framework Australia should adopt in the present climate of:
• Fewer finance and capability resources
• Data rich to data poor fisheries
• Small and large fisheries.

This review does not preclude the use of specific modelling.

Objectives

1. Review existing stock assessment methods used in Australia.

2. Review Australian stock assessment needs, and model developer and user capacity.

3. Review methods used and reviews undertaken elsewhere in the world.

4. Assess the relative merits of off-the-shelf versus case-specific assessments.

5. With input from the different jurisdictions, provide recommendations for a possible set of investment models.

Final report

ISBN: 978-1-4863-0997-9

Authors: Cathy Dichmont Roy Deng Andre Punt Rich Little

Final Report • 2018-06-25 • 1.31 MB

2014-039-DLD.pdf

Summary

Stock assessments provide scientific advice in support of fisheries decision making. They involve fitting population dynamics models to fishery and monitoring data to provide estimates of time-trajectories of biomass and fishing mortality in absolute terms and relative to biological reference points such as BMSY (the biomass corresponding to Maximum Sustainability Yield, MSY) and FMSY (the fishing mortality rate corresponding to MSY), along with measures of uncertainty. Some stock assessments are conducted using software developed for a specific stock or group of stocks. However, increasingly, stock assessments are being conducted using packages developed for application to several taxa and across multiple regions. We reviewed the range of packages used to conduct assessments of fish and invertebrate stocks in the United States because these assessments tend to have common goals, and need to provide similar outputs for decision making. Sixteen packages were considered, five based on surplus production models (“A Stock Production Model Incorporating Covariates”; “Bayesian Surplus Production Model-1”; “Bayesian Surplus Production Model-2”; “Depletion-Based Stock Reduction Analysis”; “Extended Depletion-Based Stock Reduction Analysis”), one based on a delay-difference model (“Collie-Sissenwine Analysis”), and the remainder based on age-structured models (“Assessment Method for Alaska”, “Age Structured Assessment Program”, “Beaufort Assessment Model”, “MULRIFAN-CL”, “Statistical catch-at-length”, “Stock Synthesis”, ”Simple Stock Synthesis”, “Extended Stock Synthesis”, “Virtual Population Analysis”, “VPA-2BOX”, “).

This report highlights the benefits and disadvantages of stock assessment packages in terms of allowing analysts to explore many assessment configurations and facilitating the peer-review of assessments. It also highlights the disadvantages associated with the use of packages for conducting assessments. Packages with the most options and greatest flexibility are the most difficult to use, and see the greatest development of auxiliary tools to facilitate their use.