Adjusting Fish Survival Estimates to Account for Premature Transmitter Failure in Telemetry Studies

A critical assumption of survival estimation with tagged animals is that the animal retains a functioning tag throughout the study period.  In telemetry studies, any tag that ceases to operate is equivalent to tag loss and induces negative bias in survival estimates.  Here, we discuss available methods to account for tag failure by separating the probabilities of fish survival and tag “survival”.  By using simulation and a case study from the Sacramento River, we show that current statistical methods adjust the estimate of fish survival upwards, but we also found that the adjusted estimate remains negatively biased.  Current methods estimate tag survival probability from the tag extinction curve and the observed travel time distribution of tagged fish.  However, the observed travel time distribution is truncated when tags fail prematurely, resulting in an overestimate of tag survival and an underestimate of fish survival.  We show that when an independent, unbiased estimate of the travel time distribution is incorporated into the survival model, both tag survival probabilities and fish survival probabilities are unbiased.