Exploring spatiotemporal non-stationarity of the marine environmental impact on skipjack fishery under different climate conditions in the Western and Central Pacific

Abstract

The skipjack tuna (Katsuwonus pelamis) is a significant fishery resource within purse seine fisheries in the Western and Central Pacific Ocean. The distribution of skipjack resources is influenced by the El Niño-Southern Oscillation (ENSO), exhibiting diverse spatiotemporal impact mechanisms. To investigate the spatiotemporal non-stationarity of how marine environments affect skipjack catch rate under different climate conditions, we applied the geographically and temporally weighted regression (GTWR) model to skipjack catch rate and marine environmental data from 2004 to 2021. We then compared GTWR with geographically weighted regression and generalized additive models to evaluate its validity. The findings reveal: (1) The GTWR model significantly outperformed the other two models, achieving an adjusted R² of 0.63 and exhibiting lower MAE, RMSE, and RSS values, which demonstrates the presence of spatiotemporal non-stationarity. (2) Skipjack catch rate was significantly affected by sea surface salinity and temperature (SSS, SST) as well as net primary production (NPP). Sea level anomaly (SLA) and mixed layer depth (MLD) had a moderate impact, whereas zonal and meridional current velocities (U55, V55) had a relatively minor influence. (3) The impact of environmental factors displayed varying spatial non-stationarity. Spatial non-stationarity was most pronounced for NPP, SSS, and V55, followed by SLA and SST, and least pronounced for MLD and U55. Notably, SLA spatial non-stationarity decreased during normal periods, U55 increased during El Niño, and MLD and V55 increased during La Niña. (4) Due to spatial differences influenced by ENSO, the eastern study area displayed more significant temporal non-stationarity than the western area.