A Fishery‐Dependent CPUE‐Based Proxy Index for Juvenile Japanese Anchovy ( Engraulis japonicus ) Recruitment Variability in Southeastern Korea: Asso

ABSTRACT

We developed and evaluated a fishery‐dependent catch‐per‐unit‐effort (CPUE)‐based proxy index for juvenile Japanese anchovy ( Engraulis japonicus ) recruitment‐related variability in the southeastern coastal waters of Korea. The proxy was constructed from bimonthly CPUE of the smallest juvenile size class (20–40 mm) in the anchovy drag‐net fishery during 1992–2018. We reassessed lagged associations with large‐scale climate indices and local coastal variables using wavelet coherence, autocorrelation diagnostics, structural‐equation/path‐regression diagnostics, and generalized linear/additive modeling sensitivity analyses. Wavelet analysis used a bimonthly time grid (dt = 2 months), 1000 Monte Carlo simulations, false‐discovery‐rate correction, and robustness checks based on Fourier‐randomized surrogates, STL‐residual series, and AR(1)‐prewhitened series. Raw white‐noise wavelet analyses reproduced localized seasonal‐to‐annual coherence, particularly for local temperature, salinity, log10‐transformed precipitation, and the vertical temperature‐difference index. However, multiyear PDO/MEI coherence was not stable after FDR correction or under autocorrelation‐sensitive robustness checks. We therefore interpret multiyear climate associations as exploratory low‐frequency associations rather than direct cohort‐level biological delays. Phase‐dependent path‐regression diagnostics indicated heterogeneous association structures across PDO phases, with the most consistent short‐lag signal involving temperature. Seasonal log‐linear sensitivity models provided moderate explanatory power on the log scale, with adjusted R ² values of 0.558, 0.539, and 0.523 for the August, October, and August–October seasonal proxy models, respectively. Overall, the analysis supports using the CPUE series as an operational fishery‐dependent proxy for recruitment‐related variability, while emphasizing that the inferred lagged relationships are statistical associations requiring cautious interpretation and future validation with fishery‐independent surveys.