A Bayesian Analysis of a Closed Population with Assumption Testing and Multimodel Inference Applied to Capture-Mark-Recapture Studies of Steelhead Smolts in the Wind River, WA

With the mass production of the screw trap in 1990’s estimating the abundance of juvenile outmigrants has become widespread especially in the Pacific Northwest.  However, a large gap still remains between operating a screw trap, and obtaining credible unbiased estimates of juvenile outmigrants.  Stratified Petersen Estimators of smolt abundance are often necessary due to the variable flow and outmigration patterns that lead to the violation of complete mixing or equal proportions assumption required for the use of the pooled Petersen estimator.  Therefore, biologists rely on variations of Darroch’s temporally stratified estimator, flow/trap efficiency relationships, or the Schwarz and Dempson travel time model.   However, other violations which typically occur including missing trap days, not trapping over the entire period, tagging effects, and equal catchability are sometimes analyzed but seldom is their uncertainty accounted for in the abundance estimator.  A smolt trapping study design with assumption testing presented along with a Bayesian framework for data analysis incorporating the uncertainty from marking effects, tag loss, missed marks, and lack of closure for Wind River steelhead smolts and along with methods to combine independent estimates of abundance from trap efficiency trials and the back-calculation method based on returning tagged and untagged adults.