Identification of Environmental Predictors and Their Spatial Heterogeneity for VMS-Derived Fishing Duration in the Argentine Shortfin Squid Fishery Across the Southwest Atlantic Ocean.

Argentine shortfin squid (Illex argentinus) fishing grounds in the Southwest Atlantic extend across the Argentine Exclusive Economic Zone (EEZ) and adjacent high seas, where environmental contrasts may lead to spatial heterogeneity. In this study, Vessel Monitoring System data from December 2020 to June 2025 were used to derive fishing duration as an indicator of fishing-ground use intensity. Conventional and derived environmental variables described vertical temperature and salinity differences, surface gradients, and topography. K-means clustering identified fishing-ground units, while CatBoost, SHapley Additive exPlanations, and accumulated local effects identified major environmental predictors and high-response ranges. No persistent fishing-ground center was identified within the EEZ, whereas the high seas were divided at 44.27° S into northern and southern fishing grounds with persistent centers. Seabed slope, bathymetry, and sea surface temperature were important predictors across the full area and subregions, but their rankings and high-response ranges differed among regions. Under LOSO validation, the full-study-area model yielded R2 = 0.083 ± 0.104 and Spearman's ρ = 0.4485 ± 0.0903, indicating limited absolute predictive performance but some relative ranking capability. Modeling the full area as a single unit obscures regional differences in predictor combinations and response ranges. The results demonstrate that region-specific modeling can characterize spatially varying environmental associations and support differentiated management.