Method for Estimating Detection Probabilities of Nonmigrant Tagged Fish: Applications for Quantifying Residualization Rates

Abstract

Detection probabilities are commonly accounted for in spatial mark–recapture studies to estimate quantities of biological interest such as survival, movement, or abundance, but they generally require large numbers of tagged animals to be detected. In some studies, few tagged animals are present at a particular time; the ability to estimate detection probability despite small sample sizes during some time periods would greatly improve inferences of ecologically relevant attributes. We developed a method for estimating time-varying detection probabilities of tagged fish at acoustic or radio receiver stations during periods in which few tagged fish are present and mark–recapture methods are thus prohibitive. We quantified how an index of detection probability varies with an environmental covariate, and then calibrated this index against mark–recapture detection probability estimates derived from detection data collected when tagged fish were abundant. With a known time series of the environmental covariate, the method generates a time series of predicted detection probabilities for each receiver station in a study. We apply the method to a case study involving steelhead Oncorhynchus mykiss smolts tagged with acoustic transmitters to estimate the proportion of tagged fish residualizing in a river (i.e., remaining in freshwater instead of migrating seaward). Despite few detections of tagged fish in the months after the primary downstream migration period, we were able to estimate a residualization rate of 5% (95% confidence interval, 3–12%), which is comparable to residualization rate estimates from studies employing sampling methods that do not allow the survival and movement patterns of tagged fish to also be quantified. This method can be used in conjunction with mark–recapture survival estimation methods to better isolate probabilities of residualization and survival, which are otherwise confounded.

Received November 19, 2010; accepted June 3, 2011

ACKNOWLEDGMENTS

We thank Caroline Melville and crew members of Instream Fisheries Research for assistance in the field with fish capture and test-tag passes. We thank crew members of Kintama Research for performing fish tagging surgeries and the post project for general support. We thank Carl Walters and John Petkau for comments on a previous version of the manuscript. We thank Tom Grothues and two anonymous reviewers for comments on the current version. We thank Scott Hinch, University of British Columbia, and Steven Cooke, Carleton University, for lending acoustic telemetry equipment. Funding was provided by the Canadian National Railway Company under the Cheakamus Ecosystem Recovery Fund (2008 Monitor 2, Project F). This work is a contribution to the Census of Marine Life.