Precautions and Potential for Bias When Using Auxiliary Tagging Data for Stock Assessments of Highly Migratory Tuna Species

Information from tagging experiments can have an influential role in stock assessment by way of direct estimation of population dynamic parameters such as fish abundance, movement, recruitment, and mortality or as one of several information sources in an ‘integrated’ analysis.  The estimation of movement (or transition) probabilities from tagging data is particularly important for the assessment of highly migratory species because of the dynamic spatial stock structure inherent with these stocks.  However, results from tagging analyses can be sensitive to assumptions associated with the tagged population (e.g., random selection, independence, and complete mixing) and if not accounted for properly may result in biased management advice.  Bias can also result from non-random measurement errors within the experimental procedure itself (e.g., those due to tag shedding, tag-related mortality, and tag reporting).  We begin by discussing general precautions and potential pitfalls when using tagging data for highly migratory pelagic species.  Results from integrated catch-at-age stock assessments are then compared among different tagging assumptions for tuna in the Western and Central Pacific Ocean to identify model sensitivities.  We conclude by presenting a general analytical framework to quantitatively evaluate the degree of estimator bias under alternative experimental tagging designs (or realities).