Response of the Northwest Indian Ocean purpleback flying squid (Sthenoteuthis oualaniensis) fishing grounds to marine environmental changes and its prediction model construction based on multi-models and multi-spatial and temporal scales

• Published in: Ecological indicators Vol. 154; p. 110809
• Main Authors: Han, Haibin, Jiang, Bohui, Shi, Yongchuang, Jiang, Peiwen, Zhang, Heng, Shang, Chen, Sun, Yuyan, Li, Yang, Xiang, Delong
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-10-2023; Elsevier
• Subjects: Environment variables; Northwest Indian Ocean; Sthenoteuthis oualaniensis; Temporal and spatial scales; Environment variables; Models; Temporal and spatial scales; Northwest Indian Ocean; Sthenoteuthis oualaniensis
• ISSN: 1470-160X
• DOI: 10.1016/j.ecolind.2023.110809

[Display omitted]
•High inter-annual variability in catches and gravity center of the fishing ground.•3DCNN model achieves highest accuracy in predicting fishing grounds.•The best prediction performance was at 3 days and 0.25° × 0.25°.•Visualization of the results of the optimal models (GBT and 3DCNN models).
Climate change has posed a great challenge to global fisheries harvesting. Purpleback flying squid (Sthenoteuthis oualaniensis) is a major economic cephalopod in the northwestern Indian Ocean waters, but how to choose the optimal spatiotemporal scales and models for constructing a prediction model of its fishery remains to be researched. Reasonable modeling is essential for predicting and managing the fishery of purpleback flying squid. Based on the statistical data of purpleback flying squid fishery and multi-variable marine environment data in the northwest Indian Ocean from 2015 to 2021, this paper used the gravity center of fishing grounds method, GAM, GBT, 2DCNN, and 3DCNN models to analyze. Results: 1) The catches of purpleback flying squid reached their peak in 2019, and the higher catch areas were concentrated at 15° ∼ 19°N and 61° ∼ 65°E. The annual gravity center of the fishing grounds as a whole moved northeast. 2) Different models have advantages and disadvantages on different scales; Overall, the optimal models were GBT and 3DCNN models; The Optimal spatiotemporal scale was three days (0.25° × 0.25°). 3)In the GBT model, the most critical variables were SST, Chla, 97 m-DO, and SSS; In the 3DCNN model, the relative importance ranking of variables depended on the training data set. The results of the study help us to select the optimal model and optimal spatial and temporal scales to construct a prediction model for the Northwest Indian Ocean purpleback flying squid fishing grounds, which will provide a scientific basis for the sustainable development and management of the purpleback flying squid fishery in the region.