Summary: | 碩士 === 國立臺灣大學 === 海洋研究所 === 107 === Catch-per-unit-effort (CPUE) is usually assumed to be an index of the relative abundance of a fish stock. However, it is necessary to standardize the CPUE to remove the effects of factors such as the fishing time, location, and change of fishing tactic before to be used as a reliable index of abundance in the stock assessment. Taiwanese distant water longline fishery logbook comprises multiple species CPUE data. Due to the development of the ultra-low temperature tuna longline vessels in the 1970s, Taiwanese distant water longline fisheries have included various fishing tactics in the Pacific Ocean. Therefore, it is necessary to consider the change in fishing tactics when conducting the CPUE standardization. In this study, I used the GLM model with explanatory variable of the targeting tactic derived from the logbook’s targeting group for estimating the reference index of relative abundance of South Pacific albacore based on the Taiwanese distant water longline fisheries logbook data during 2007-2017. I evaluated four approaches: categorical hooks per basket, categorical clusters and continuous principal component scores as GLM explanatory variables, and the finite mixture modeling, by comparing the estimated standardized indices with the reference index. Results showed that all approaches could remove a substantial amount of variation in the CPUE data. However, the principal component analysis and the finite mixture model were identified performing better for estimating the standardized index compared to the reference index based on the bias metric and mean absolute error. This study recommend using the principal component analysis or finite mixture modeling approach for constructing a historical standardized index for the stock assessment and fishery management of South Pacific albacore.
|