Can observer sampling validate industry catch reports from trawl fisheries?

Abstract

Stock assessment scientists and fishery managers operate under the necessary assumption that the identities of species and quantities of catch from industry landing reports are known without error. To evaluate this assumption, we compared industry-generated reports of landed catch to independent observer estimates. An observer sampling design for shore-based processing plants was developed and implemented at four plants in Kodiak, Alaska. A total of 13 deliveries from three fisheries were examined. Observers were able to track different portions of the catch and treat them as strata from which to randomly sample or completely enumerate. Differences between observer- and industry-derived species proportions were negligible when measured across the entire study, but differed by fishery. Industry weight in the shallow-water flatfish fishery exhibited a small negative bias not related to processor or species type. Weight differences in complete enumerations for big (Beringraja binoculata) and longnose skates (Raja rhina) were of similar magnitude but in opposite directions, leading to the conclusion that the identification of these species is confused since there is no detection error. Where observers needed to sample, they were able to detect most species in open access fisheries, with a resolution comparable to that of the industry, and were more likely to detect skates than industry. However, in the cooperative rockfish fishery, where tighter controls on the dockside sorting of fish by plant staff are in place, industry reports had enhanced detection of rare species relative to observer sampling. Notwithstanding, differences between data sources remained substantial in strata where the observer sampled even after considering rarity. The results here highlight the utility of using third-party verification to improve data quality of self-reported data, and identified the logistical, database, and analytical challenges to effectively monitor fishery quotas.