Artificial Intelligence Prediction of Coastal Invertebrate Distribution Under Ocean Warming

Abstract

Introduction: Climate change is influencing the coastal marine ecosystem through changes in the environment, habitat suitability, and species distribution. Invertebrates form an integral part of the ecosystem, yet the response of these organisms to environmental changes cannot be easily predicted using conventional methods due to the non-linear nature of these responses. This paper examines the use of artificial intelligence in prediction of coastal invertebrate distribution. Materials and Methods: Information on the distribution of coastal invertebrates was combined with environmental variables such as sea surface temperature, temperature anomaly, salinity, dissolved oxygen, depth, chlorophyll-a concentration, and habitat information. The information was preprocessed and utilized to build Random Forest, XGBoost, and Artificial Neural Network models. Performance metrics that were considered during model development include accuracy, precision, recall, F1-score, and ROC-AUC score. The best-performing model was employed to evaluate habitat suitability in different warming scenarios. Results: The XGBoost algorithm performed better than the others in terms of predicting accuracy of 82.6%, precision of 81.9%, recall of 80.8%, F1 score of 81.3% and a ROC-AUC of 0.88. The suitability of the predicted habitat declined with rising temperatures of the ocean, falling from 82% under present conditions to 74% under moderate warming and 63% under high warming. The results indicate that increasing temperature and associated environmental changes may substantially influence the spatial suitability of coastal habitats for invertebrate communities. Conclusion: From the results obtained, it is clear that artificial intelligence is capable of accurately recognizing environmental relationships and can provide reliable predictions regarding the distribution of coastal invertebrates under climate change. The reduction in the habitat suitability observed indicates the possible threats posed to coastal invertebrates due to rising temperatures in the oceans. Thus, AI can be used for ecological forecasting of marine biodiversity.