Improving multi-species habitat identification through species weighting assignment using joint species distribution model

Habitat conservation serves as the cornerstone for fishery resource recovery. Traditionally, multi-species habitat predictions are typically generated by stacking individual species' habitat suitability maps using the equal-weighting strategy. However, this approach neglects interspecific differences in data quality and life-history traits, potentially resulting in biased or ecologically suboptimal identification of shared habitats. In this study, we developed a novel framework that incorporates species-specific weighting schemes to enhance multi-species habitat prediction. We applied this approach to fish communities in the central and southern Yellow Sea, China, and compared shared habitat predictions generated by the Joint Species Distribution Model (JSDM) under alternative weighting strategies. The weighting factors considered included species-specific Area Under the Receiver Operating Characteristic (ROC) Curve (AUC) from JSDM, prevalence, and trophic level. Our results demonstrated that incorporating species-specific weights significantly improves the prediction accuracy of multi-species habitat distributions in the study area, as indicated by higher correlation coefficients and lower standard deviations between predicted and observed efficiency indices. Notably, the weighted approach would also narrow the extent of prioritized areas and improve conservation efficiency. Under the weighted scenarios, the area identified for protection was nearly 50 % smaller than that under the unweighted approach, while the protection effort directed toward shared habitats increased by 33.7 %. This study highlights the potential of incorporating species-specific weighting into JSDMs to improve multi-species habitat suitability predictions. This will support the design of marine protected areas that enable more accurate identification and protection of key shared habitats across species, while also promoting cost-effective conservation outcomes. • Utilizing species joint distribution models (JSDM), the shared habitats for multiple species were determined by weighting ecological and biological factors. • Compared to unweighted scenarios, using the weighted model demonstrated a 33.7 % improvement in habitat protection efficiency. • Species prevalence emerged as the most influential factor on the construction of shared habitat identification framework. • We proposed framework that provide strategic insights for the development of fisheries and the designation of marine protected areas.