한국 연근해 살오징어(Todarodes pacificus)의 CPUE 표준화와 베이지안 상태공간 잉여생산량 모델에 의한 자원평가

Abstract

The common squid (Todarodes pacificus) is one of the most commercially important species caught in Korean waters. The catch increased throughout the 1990s, peaking at 250 thousand tons in 1996 and remained above 200 thousand tons until the early 2000s. However it has been steadily declining since then, falling below 15 thousand tons in 2024. As the common squid is a short-lived migratory species and is sensitive to the ocean environment, it is difficult to quantitatively assess the stock. For that reason, it relies heavily on fishery-dependent data when assessing the stock. Catch Per Unit Effort (CPUE), widely used as a proxy for relative abundance, can be affected by spatial and temporal variability in fishing activity. Accordingly, standardization of CPUE is essential for its use in the stock assessment. In this study, the catch and effort data of the offshore jigging fishery (2003–2024) were used to standardize CPUE of the common squid. The CPUE standardization was implemented using Generalized Linear Models (GLMs) and Generalized Linear Mixed Models (GLMMs), incorporating temporal and spatial factors such as year, quarter (3-month intervals), area (latitude-longitude) and environmental factor (sea surface temperature). Among the models, GLMM, which included year, quarter, area, and temperature as fixed effects, and the interaction between year and area, and vessel id as random effects, showed the best performance. Accordingly the abundance indicates from this model were used as an input for a Bayesian State-space Surplus production model (BSS model) based on the Schaefer model to estimate the stock status. Informative priors were assigned to the intrinsic growth rate(r) and carrying capacity (K), while non-informative priors were assigned to the initial biomass level (B₁/K=b), catchability coefficient (q) and variance (σ²) of process error (eᵘ ͭ), variance (τ²) of observation error (eᵛ ͭ). These error terms were assumed to follow normal distribution with zero means and their variances (uₜ ~ N(0, σ²), vₜ ~ N(0, τ²)). From the BSS model, K and r were estimated to be 1,105,371 tons and 0.564 respectively. The initial biomass level (b) was 0.530, the catchability coefficient (q) was 2.4 x 10⁻⁶, the variance of process error (σ²) was 0.039 and the observation error variance (τ²) was 0.026. Based on the Kobe plot analysis for the common squid, it showed that overfishing is not currently occurring (F/F_(MSY) < 1), but the stock has been overfished (B/B_(MSY) < 1).