Uncertainty associated with total bycatch estimates for rarely-encountered species varies substantially with observer coverage levels: Informing minimum requirements for statutory logbook validation

Concerns regarding quantifying fishing impacts on rarely-encountered bycatch species in a demersal fish trawl fishery in north-western Australia have raised issues with, 1) uncertainty associated with extrapolated estimates from low levels of coverage by human observers, and 2) appropriate minimum levels of observer coverage. This study used data collected from a very high level of independent observations (85% coverage, n = 2127 trawls) using electronic monitoring (EM) to determine the uncertainty of extrapolated estimates of chondrichthyan, cetacean and reptile bycatch at different levels of observer coverage. The distribution of extrapolated estimates of total numbers for rarely-encountered bycatch (i.e. dolphins and hammerhead sharks) was asymmetric, with a higher tendency for overestimation at lower levels of observer coverage (≤ 25%). Conversely, the minimum level of observer coverage required to achieve a reasonable level of precision (± 30% or lower) for rarely-encountered species would need to be very high (≥90%). Combining the interaction rates of rarely-encountered species (i.e. via statutory logbook validation) reduces the minimum level of observer coverage required to an achievable level (~64%), whilst maintaining relatively accurate (± 25%) quantitative bycatch estimates for management purposes. Future programs need to consider that, in contrast to the infrequent presence of human observers, EM with cameras installed to monitor all fishing activity will effectively negate the need to assume that fishing practices are equivalent between observed and unobserved events. Hence, analysis of data collected by human observers may need to account for additional biases and/or imprecision along with the marked increase in uncertainty associated with lower levels of observer coverage.