Recommendations for Pooling Annual Bycatch Estimates when Events are Rare

摘要

Fishery bycatch estimates reported in U.S. marine mammal stock assessment reports (SARs) are typically pooled over multiple years to increase precision and reduce small-sample biases in resulting mean annual estimates. Such pooling can remedy biases in single-year estimates resulting from interannual variability in marine mammal abundance, oceanographic conditions, or observer coverage. Bias is defined as systematic over- or underestimation of true bycatch levels over any particular pooling period. Current SAR guidelines suggests that pooling bycatch estimates for periods of five years is typically adequate to obtain an acceptable level of precision, generally defined as a coefficient of variation (CV) of . 0.30. However, this guidance is based on an assumption that bycatch is a relatively common event with adequate sample sizes and sufficient observer coverage. Pooling over longer periods is also acceptable if additional years accurately represent the current state of the fisheries and their inclusion reduces estimation error. For rare events, estimation error can be severe, with problems that include 1) an inability to document bycatch that is occurring because observer coverage is too low to detect rare events (negative bias) and 2) exaggerated estimates of annual bycatch when observer coverage is low and observed bycatch events are extrapolated by a large multiplicative factor (positive bias). Rare bycatch events often involve species that pose management concerns because their abundance and potential biological removal (PBR) levels are low to begin with. As such, it is important to reduce estimation error for these species. We report the results of a simulation trial for rare bycatch events in a hypothetical fishery with 20% observer coverage, showing how estimation error is reduced as a function of the number of years that annual estimates are pooled beyond the 5-year time period most typically used in marine mammal stock assessments.