Tests of the robustness of habitat-standardized abundance indices using simulated blue marlin catch–effort data

Abstract

This paper describes a computer model designed to simulate longline catch–effort data, and to test the robustness of methods to standardize the relative catch per unit effort (CPUE) abundance index. The model considers abundance and catch by month, area and depth based on the temperature habitat of the species. Characteristics of fishing gears are used in a submodel to predict catch rate by set and depth. An application to an Atlantic blue marlin bycatch fishery from 1956 to 1999 is presented. The depth distributions of sets, the propensity of marlin to bite stationary moving baits, and the assumed habitat preferences were each varied. The simulated CPUE series were standardized using both habitat-based and General Linear Models (GLM). Unstandardized CPUE trends were strongly biased. The GLM standardizations accurately reflected trends in the simulated population. The habitat standardizations proved accurate when the assumptions used in the analysis were precisely correct, but were often strongly biased when the assumptions were inaccurate. These results suggest that habitat standardization of CPUE time series is potentially useful if there is accurate knowledge of the distributions of the population, the actual fishing depths of the gears and factors that may affect the fish's propensity to take bait. Absent certain knowledge of these factors, the habitat approach to standardization should be used with great caution.