Exploratory analysis of spatio-temporal patterns in Pacific saury size composition data

This study analyzed Pacific saury size composition data from Japan and Chinese Taipei using multivariate regression tree analysis to identify spatial patterns and propose candidate fleet definitions for stock assessment models. Length frequency and commercial category analyses revealed distinct spatial boundaries at 155.5°E for Japanese fleets and 160.3°E for Chinese Taipei fleets. Japanese data showed a mixed length distribution west of 155.5°E and unimodal patterns to the east, while Chinese Taipei data exhibited uniform length frequency patterns across space but heterogeneous commercial size distributions. Both fleets demonstrated consistent seasonal patterns, with smaller fish distributed in southwestern areas in the fourth season (October-December), and between 163.3°E-170°E. Although additional spatial splits increased explained variance, we recommend using primary boundaries (155.5°E and 160.3°E for Japanese and Chinese Taipei fleets, respectively) to maintain analytical simplicity while effectively capturing size composition heterogeneity for both fleets. These findings suggested the candidate spatially distinct fleet definitions may better account for heterogeneity in catch-at-size data in stock assessment modeling.