Research paperNonparametric tests of double-tagging assumptions

•Method to analyze tagging experiments for dependence between tags on the same fish.•Can be applied by field staff to monitor an experiment that is in progress.•Exact Bayesian method uses Monte Carlo simulation.•Easily programmed in R, code provided.•Applied to previously published cod tagging data, shows need for reanalysis.

Shedding rates of tags on fish are commonly estimated from double-tagging experiments, for which an assumption of independence between the two tags on a fish is required. For tags of qualitatively different types, a nonparametric test for this assumption was proposed by Myhre (1966), making use of concurrent double- and single-tagging of fish. We extend Myhre's test by developing a nonparametric Bayesian test that is also applicable to the common situation where the two tags attached to a fish are identical and assumed to shed at the same rate; the validity of this assumption can be checked by an extra test that we supply in the case that each tag is identified uniquely. In addition to dependence between tags, the dependence test can also be triggered by departures from other experimental assumptions, such as marked variation in the expertise of taggers. We recommend the dependence test for monitoring tag-return data on an ongoing basis during an experiment. We apply our test to Atlantic cod tagging data listed by Barrowman and Myers (1996). Frequentist tests based on Fisher's Exact Test are also presented.

Graphical abstractDownload high-res image (140KB)Download full-size image