bfisher is a set of functions for estimating the run sizes of anadromous fish populations, from mark-recapture data gathered using a T-JAMM design (Tagging Juveniles And Monitoring Migrants). This sort of design is suitable for fish that can be tagged while in freshwater (usually using PIT tags), and then detected during their sea-ward and return migrations (usually by a system of fixed instream tag-detectors). The bfisher code uses a system of Bayesian hierarchical models to estimate run size from such datasets, and is written in the statistical programming language R. The estimation algorithms use standard Markov-Chain Monte Carlo (MCMC) techniques, as implemented in the R package BRugs. The reporting of results uses the conventions of the package R2WinBUGS. Please note that some proficiency with R will be necessary to use these functions, and familiarity with the strengths and limitations of MCMC algorithms will also be helpful. In particular, convergence of chains may be impractically slow under some circumstances, such as when the model is overparameterized due to poor choice of covariates, etc. That said, experience thus far indicates the model converges well under most reasonable datasets. MCMC is a numerically intensive procedure and benefits from a fast computer with ample memory.
Author: David Boughton
The zip file includes the source code, instructions for running the code, and a copy of the BRugs package. Please note that BRugs requires the Windows operating system.
This estimation model is more thoroughly described in following journal article:
Boughton, D. A. In Press. Estimating the size of steelhead runs by tagging juveniles and monitoring migrants. North American Journal of Fisheries Management.