Deriving a Statistically Reliable Abundance Index from Landings Data: An Application to the Taiwanese Coastal Dolphinfish Fishery with a Multispecies Feature

Abstract

Data from coastal fisheries are often incomplete, as these fisheries are usually small in scale, rendering them exempt from logbook submission requirements. The catch of Dolphinfish Coryphaena hippurus by Taiwanese fisheries once ranked second in the world but has dramatically declined to very low levels in recent years. To address this decline, assessment of a Dolphinfish abundance index was necessary. However, due to the small scale of the fisheries, logbook data were not available to calculate CPUE. This study aimed to estimate a statistically reliable index by (1) assigning effort matrices to landings data using coastal surveillance radar data; (2) standardizing the 2001-2015 CPUEs while using four approaches (classifying fishing tactics by multivariate techniques and principal components analysis) to differentiate the fisheries' effort toward catching Dolphinfish from the effort toward other target species; and (3) evaluating performance of the standardization models by using an R-2 estimated by cross-validation and bootstrap procedures. The approach that used a delta-generalized additive model with a direct principal components procedure demonstrated the best fit. This study presents an example of deriving a statistically reliable abundance index from the data-incomplete situations common for coastal fisheries, which allows for follow-up population dynamics studies. The resulting index for Dolphinfish in the Taiwanese region showed two 7-year cycles, with a prominent decline in 2015. Reasons for the fluctuation are unknown but may be due to environmental factors, the fast-growing nature of the fish, and heavy exploitation of the stock by Taiwanese fisheries.