Comparative analysis of modeling methods and prediction accuracy for Japanese sardine habitat under three climate scenarios with differing greenhouse emission pathways

Abstract

Climate change threatens to marine ecosystems, making it essential to understand habitat shifts for ecologically and economically valuable species like the Japanese sardine (Sardinops melanostictus). This study employs Biomod2 ensemble algorithm and MaxEnt model, integrating different spatial resolutions and varying numbers of species presence data points at the same resolution, to select the most accurate method for assessing the habitat suitability of the Japanese sardine under both current and future climate scenarios. Biomod2 outperforms MaxEnt across all conditions, particularly with high-resolution data and increased presence data points, achieving its highest accuracy (AUC = 0.965) with the EMwmean at 0.083° resolution and 10,462 presence points. Results suggest that suitable habitats for Japanese sardine will contract at low latitudes and expand marginally at higher latitudes under future climate scenarios, with Chla and SST as key environmental drivers. We identified a precise method for simulating Japanese sardine habitats, providing insights for similar marine species.