Conventional “spaghetti” tags and tags
originally designed for “sport fishing” and called Betyp tags, were used
during a tuna tagging program conducted on board Dakar baitboats in 1999.
With the aim of comparing the recapture rate of both types of tags,
additional information obtained from previous tagging trips are used in a
Bayesian context to set up an informative prior for conventional tags. We
show in this study (1) how to account for the sampling uncertainty in the construction
of the Beta prior with a likelihood method, and (2) how a simulation-based
alternative can be useful for performing the probability density function of
the difference between two posterior recapture rates. On the light of the
resulting simulated difference we found that Betyp tags have a very strong
negative effect on the return rate of bigeye tuna (−19.6% on average).
For skipjack the strength of evidence concerning the decrease in recapture
rate due to the implementation of Betyp tags (−3.2%) was supported only
at a 10% level. The Bayesian approach is compared with the conventional
“frequentist” approach and with a likelihood method allowing for the
integration of previous information. The results obtained from different
approaches indicate that the choice of the method, as well as the choice of
the prior, does not modify the conclusion of the study. Potential causes for
explaining the lowest efficiency of Betyp tags are discussed.