Cluster-based identification of fishing strategies with robustness assessment: A case study of Taiwanese albacore longline fisheries in the Indian Ocean

Abstract

Understanding fishing strategies is essential for evaluating the dynamics of multi-species fisheries and supporting effective management. Fishing strategies in the Taiwanese albacore longline fishery in the Indian Ocean were analysed based on species composition data collected from 2005 to 2023. Hierarchical cluster analysis identified distinct operational patterns across four regions defined for the albacore stock assessment. The optimal number of clusters was determined using the elbow method, while principal component analysis and non-parametric tests were used to validate cluster differentiation. Results revealed spatial and temporal shifts in targeting, including transitions between tropical and temperate tunas and increased reliance on non-tuna species in some areas. Robustness was assessed via 100 split-sample iterations using Bray-Curtis similarity. Matched clusters showed consistently high similarity (median >0.89) across regions, whereas the rate of unmatched clusters remained low (similar to 0-15 %, depending on region). Compared with traditional indicators such as the number of hooks between floats, the multivariate approach proved more effective in characterising fishing strategies under complex or data-limited conditions. Across the four regions, we identified three to four distinct strategy clusters per region, characterised by dominant species mixes including albacore dominant, mixed albacore and other species, and tropical tuna oriented, with the first two principal components visualising and explaining 72-96 % of the variance. This approach enhances catch per unit effort (CPUE) standardisation and provides valuable insights to support regionally adaptive fisheries management.