Research outline for CPUE data used in WCPO stock assessments Simon Hoyle SPC
Introduction Longline CPUE data drive abundance trends for bigeye, yellowfin, and albacore assessments There are conflicts between CPUE and size data trends We have all been aware for some time of the need to improve methods for analyzing CPUE data Highlighted by the recent YFT reviews Progress has been made – ‘more work is needed’
Background Long history of analyzing CPUE data to estimate abundance Various approaches have been used, including habitat- based standardization (HBS), statistical HBS, and standardization that includes oceanographic effects The current approach focuses on standardizing out spatial and HBF effects, to extract a year-quarter effect as the index of abundance Developments are under way towards using operational data instead of aggregated data
Recommendations from past work Numerous recommendations in many areas, summarized in the research outline document Regional weighting Modeling approaches Spatial effects and data weighting Targeting, fishing behavior and catchability Covariates and data resolution Data Model selection
Proposed work areas 1 Weighting given to different strata Weights given to individual records affect abundance trends Aggregated analyses give each stratum the same weight, while operational data analyses give each set the same weight This is a statistical problem that can have a surprisingly large effect on the indices Related to issue of error distributions better approaches are available than the current delta lognormal
Proposed work areas 2 Hidden assumptions about abundance trends in areas not fished early or late Classic problem – what is the abundance in areas not being fished Approaches are available – spatial backfilling to infer abundance Related to issues of spatial stratification, consistency of abundance trends among areas, and range expansion / contraction
Proposed work areas 3 Fishing practices changing through time Fishing fleets change their behaviour through time Clear evidence for changes in interactions among vessels, changes in technology, changes in incentives from fish prices and marketing strategies Need for analyses using fine-scale location data, to better understand the changes in fishing strategies and their effects on CPUE indices Fishing fleets are not homogeneous, but CPUE analyses should work with homogeneous data Need for clustering approaches to separate subfleets
Proposed work areas 4 Regional weighting factors Very influential for the assessments, but assumptions not particularly robust Need for alternative perspectives and alternative datasets Should also try assessment approaches that do not use regional weighting
Conclusions The research outline is intended as advice from the science provider to the SC. To be considered for inclusion in the overall research plan This is a work in progress – early stages of development Seeking advice about priorities, and how this document might be improved.