Department of Statistics Seminar
North Carolina State University

presents

Dr. William Kendall

USGS Patuxent Wildlife Research Center

"Estimation of Movement To/From an Unobservable or Partially Observable State Using Capture-Recapture Data"

ABSTRACT

Work begun by Arnason (1972, 1973) and expanded by Schwarz et al. (1993) and Brownie et al. (1993) extended the traditional Cormack-Jolly-Seber capture-recapture model to the multi-state case. When animals are released/recaptured in multiple states (e.g., geographic areas) these models allow for the estimation of state-specific survival and detection probabilities, as well as the probability of transitions between states. In many cases multiple states are acknowledged or suspected, but at least one of those states is unobservable or observed with some chance of misclassification. I will present methods to estimate parameters of interest, including movement, in the face of an unobservable or partially observable state. I will focus on the need for additional information (e.g., sub-samples in the context of Pollock's robust design) or additional assumptions about movement (e.g., requisite transition from the observable to unobservable state). I will also give some attention to the usefulness of combining band recoveries or ancillary observations with recapture information to estimate movement or tag loss. Finally, I will illustrate these methods using several wildlife examples. These include mice going in and out of torpor, nesting sea turtles where only the breeding females are observable, breeding manatees where the calf is not always seen with the mother, and neck collar loss in geese.

Friday, April, 27, 2001

3:35 - 4:35 pm

206 Cox Hall

Refreshments will be served on the second floor of Dabney Hall (left of Room 222) at 3:00 pm.