Detection of bias in observer-based estimates of retained and discarded catches from a multi species trawl fishery

Abstract Observer-based estimates of quantities, size and age distributions of by-catches and discarded catches may be biased by nonrepresentative selection of sampling units (fisher-days or trips) or by changes in fishing practices onboard trawlers when observers are present. In this study, we examined the accuracy of estimates of catch derived from an observer survey of retained and discarded catches in a multispecies fish trawl fishery off the coast of NSW, Australia. Observer-based estimates of magnitudes and size-distributions of retained catches were compared with independent, unbiased estimates that were available for a subset of species (species managed by catch quotas) caught in the fishery. Conclusions about bias in estimates of other components of catch (especially discards) are based on the premise that bias is unlikely to affect these estimates without also affecting estimates of retained catches of quota species. We conclude that estimates of catch, based on the 3-year period 1993–1995, were unaffected by significant bias. Observer-based estimates of magnitudes of retained catches did not differ significantly from reported landings for 6 out of 7 species and the combined catch of quota species (CQS) for the Ulladulla fleet, 11 out of 11 species and CQS for the Eden fleet and 10 out of 11 species and CQS for the 2 fleets combined. There was, however, some evidence of bias in estimates of catch for each fleet in 1 of the 3 years. Observer-based size-distributions were not significantly biased. We conclude that our approach to validating observer-based estimates of catch would also be of use in observer surveys of other fisheries.