Mark-recapture and tag loss

Laura Cowen successfully defended her Ph.D. thesis entitled "Mark-recapture and tag-loss" on 15 April 2005.

Mark-recapture studies are used to estimate population parameters such as abundance, survival and recruitment. Briefly, animals are captured, marked with an individually identifiable tag and released. First capture provides information about abundance. Subsequent recaptures provided survival information about the individual. One of the fundamental assumptions in mark-recapture studies is that tags are not lost. If this assumption is violated, parameter and standard error estimates are biased. This thesis deals with the analysis of 3 mark-recapture experiments under tag-loss.

The first paper looks at premature radio failure in radio-telemetry studies. Radio-tags, because of their high detectability, are often used in capture-recapture studies. A key assumption is that radio-tags do not cease functioning during the study. Radio-tag failure before the end of a study can lead to underestimates of survival rates. We develop a model to incorporate secondary radio-tag failure data. This model was applied to chinook smolts (Oncorhynchus tshawytscha) on the Columbia River, WA.

The second paper incorporates tag loss into the Jolly-Seber model. Tag loss in the Jolly-Seber model has only been dealt with in an ad hoc manner. We develop methodology to estimate population sizes and tag-retention in double-tagging mark-recapture experiments. We apply this methodology to walleyes (Stizostedion vitreum) in Mille Lacs, Minnesota.

Finally, in the third paper, we develop a Poisson migration model incorporating tag loss. This model is applied to Yellowtail Flounder (Limanda ferruginea) on the Grand Bank of Newfoundland.

This type of interdisciplinary work is a hallmark of our program in Applied Statistics at Simon Fraser University. For more information, please contact Laura Cowen (lcowenstat.sfu.ca) or her supervisor Carl Schwarz (cschwarzstat.sfu.ca), Department of Statistics and Actuarial Science.