Efficacy of a bycatch estimation tool

The bycatch estimation tool developed by Babcock (2022) was subjected to simulation testing using the species distribution model and longline simulator (LLSIM) developed by Goodyear (2021). To evaluate the efficacy of the bycatch estimation tool, generalized representations of ICCAT CPC longline fisheries were created using LLSIM and were coupled with alternative representations of observer programs to produce simulated logbook and observer databases for a range of observer coverage levels and allocation methods. Using a semi-automated model selection process, linear predictors based on negative binomial and delta lognormal models were used to predict total annual bycatch of blue marlin from the simulated datasets. A stratified ratio estimator was also used for comparison. Across representations of observer programs, bycatch estimates were reasonably unbiased, with diminishing variation in bias estimates as observer coverage increased. The use of simulated data sets provides a demonstration of the utility of the bycatch estimation tool as well as evaluation of its reliability.