Correcting Bias in Survival Estimation Resulting From Tag Failure in Acoustic and Radiotelemetry Studies

The high detection rates of acoustic- and radio-tagged fish greatly improve the ability of an investigator to obtain information on survival and movement of fish with fewer tags. The trade-off, though, is a greater dependence on the individual tag performance, as each tagged fish in a smaller study represents a greater proportion of the outcome. This reduction in release size, due to the increase in detection capability, places a greater emphasis on the need to accurately gauge the status of the tagged fish. Should a tag fail while a smolt is migrating through the study area, the release-recapture model cannot discern the difference between smolt death and tag failure. If the release-recapture models are not adjusted for the probability of tag failure, the estimates of smolt survival will therefore be negatively biased. This article presents a semiparametric approach for adjusting survival estimates from release-recapture studies for tag failure, and provides subsequent estimation of sampling variance and its contributing components.