Publication Abstract

Development of an estimation system for U.S. longline discard estimates of bluefin tuna

C.M. O'Brien, G.M. Pilling and C.Brown

Historical estimates of U.S. longline dead discards of bluefin tuna have been based on tallies of reported dead discards made by fishers on logbooks. Initiation of an observer programme allowed provisional estimates of bluefin tuna discards to be based on direct observation of the U.S. pelagic fleet for the period 1992–1999. The method applied provided a basis for characterizing uncertainty in estimates. However, sparse sampling in a number of spatial (e.g. geographical) and temporal (e.g. quarterly) strata meant pooling was required across strata to achieve a minimum number of observations per stratum. Further attention must be paid to the estimation of discards, so that its effects can be fully included in assessments of shared stocks. The paper presents an alternative approach to estimating dead bluefin tuna discards, which will improve both the estimates themselves and the calculation of uncertainty about them. Models are developed using observer data. A number of important model covariates are identified: quarter, year, grouped fishing area and set target species. The statistical concept of conditionality is applied, and a flexible mixture distribution – the negative binomial – proposed for modelling discard data. The results of applying these models to observer programme data to estimate total bluefin tuna discards are presented.

Reference:

C.M. O'Brien, G.M. Pilling and C.Brown. 2004. Development of an estimation system for U.S. longline discard estimates of bluefin tuna. Management of Shared Fish Stocks; 23-41pp