Evaluating the Effects of Observer Sampling on Estimates of Fishing Mortality in a U.S. Pacific Groundfish Fishery

Fishery-dependent data from monitoring programs are critical for fisheries management, e.g., for estimating fishing mortality. The percent of fishing vessels monitored by on-board observers (i.e., observer coverage) will determine the proportion of catch sampled which is used to estimate fishing mortality.  Thus, observer coverage has a direct effect on the accuracy and precision of mortality estimates. Using bootstrap resampling, we investigate how variation in observer coverage influences estimates of fishing mortality.  Using data from a U.S. west coast groundfish fishery (2011-13) when observer coverage was 100% (complete census) we resampled these data at 5-90% coverage rates.  We then measured accuracy and precision in our estimates by comparing the estimated mortality (5-95% coverage) with the known mortality (100% coverage) across a variety of species types (e.g., target, bycatch, protected).  Increases in coverage should produce more accurate estimates with less uncertainty.  We show that variability in the probability of catch can lead to biased estimates with relatively low uncertainty, whereas variability in the magnitude of catch can lead to relatively unbiased estimates with high uncertainty around the estimate.  This work provides guidance to both monitoring programs and fisheries managers for identifying minimum necessary observer coverage to meet scientific standards and goals.