Application of a fish habitat model considering mesoscale oceanographic features in evaluating climatic impact on distribution and abundance of Pacific saury (Cololabis saira)

Abstract

The continuing development of appropriate environmental predictors is important to improving the performance of fish habitat models. Mesoscale oceanographic features (MOFs) such as fronts, eddies, or upwelling zones, sometimes appearing as “oases in a fluid desert” occupying >50 km ocean surface, have customarily been regarded as key factors driving fish distributions. However, previous algorithms quantifying oceanographic features have not provided greatly enhanced predictive power to the fish habitat models which are characterized by a low variable importance due to inadequate attention having been paid to its scale-specific aspects. Here, a new predictor representing MOFs is introduced and a reappraisal by common fish habitat models of the role of the MOFs in fish distribution within a case study on Pacific saury (Cololabis saira) in the northwestern Pacific is reported. When the new predictor with MOFs was introduced, the performances of three commonly used fish habitat models were all improved significantly (26%–30%), together with the paramount importance of MOFs among all predictive variables, suggesting that MOFs may exert significant effects on Pacific saury distribution. Using the optimal model (Random Forest) selected among the three, the habitat distribution of Pacific saury in Oyashio water was then reconstructed for the period of 1993–2020 to explore its spatio-temporal variations and its relationship with the abundance variation. The “suitable habitats” (areas of reconstructed catch > 12 tons) during major fishing seasons (August–November) estimated by the optimal habitat model for Pacific saury showed a clear northward shift of 0.045°/year for the period of 1993–2020, associated with the poleward movement of both oceanic isotherms and the Oyashio extension due to northward movement of the whole wind field in concert with global warming. The estimated annual suitable habitat area (SHA) during the early fishing seasons (June–September) showed large inter-annual variations with a peak around 2010 and a valley in 2015, and correlated significantly with the abundance index of Pacific saury, implying that the SHA can be regarded as a practical indicator of abundance. Moreover, it was also found that a marked decline in the SHA was involved in the dramatic decrease in the abundance and catch of Pacific saury after 2010, while eastward movements of SHA resulted from MOFs possibly aggravated the decrease in the catch of Japanese waters by changing its southern migration route. This study illustrates the significance of MOFs in predicting fish distribution by means of empirical habitat models and may provide new insights for understanding fish habitat variability in relation to physical-biological interactions in the ocean.