Designing Tagging Programs for Pelagic Longline Fisheries: Trade‐offs between the Number of Releases and Observer Coverage

Tagging experiments are becoming increasingly important for providing an alternative to catch indices in large pelagic fisheries. The design of these experiments, in terms of the resources to devote to the number of releases versus recovery activity, has generally been ad hoc. We examined the relationship between the level of observer coverage and the variance in the estimates of annual exploitation rates from a tagging experiment in which the reporting rate is estimated from comparisons of tag returns from vessels with and without observers. The results indicate that the predicted precision will be a function of the level of observer coverage, the number of tags released, and the actual fishing exploitation rate. The results suggest that a direct tradeoff exists between the number of tags released and the level of observer coverage, such that a proportional increase in either one has similar effects on the predicted coefficient of variation. However, considering the assumptions (particularly multistage aspects of observer sampling and the need to estimate catches by age), higher observer coverage would be favored. In fact, substantial benefits in terms of ensuring accuracy and precision of the estimates would be gained from having 100% observer coverage. In any case, the results emphasize the importance of ensuring both adequate numbers of tag releases and robust estimation of reporting rates.