Habitat suitability of two benthopelagic fish species in the Caspian Sea using maximum entropy model

Abstract

Protecting valuable aquatic reserves requires understanding marine species' geographical locations and selected habitats. The current research uses the maximum entropy model (MaxEnt) and 21 environmental parameters to model the distribution of Chelon saliens, and Vimba persa, in the Iranian coast waters of the Caspian Sea (CS). The study encompassed 302 sampling efforts, where presence percentages for C. saliens, and V. persa were 203 (50.87 %), and 99 (24.81 %), respectively. The optimal algorithm for C. saliens was random forest (rf), achieving 83 % accuracy. For V. persa, the best-performing algorithm was the support vector machine (svm), with an accuracy of 80 %. The area under the system performance characteristic curve (AUC) learning scores for each species was above 0.9, indicating the high predictive power of the MaxEnt model in determining species' actual distribution. For C. saliens, key environmental factors such as oxygen percentage saturation (DO%, 34 %), photosynthetically active radiation (PAR, 11 %), and sea surface temperature (SST, 7 %) played a significant role in predicting the species distribution. Similarly, for V. persa, environmental variables such as PAR (34 %), SST (19 %), and ammonium (7 %) emerged as the most influential predictors. Based on this model, the percentage of high-desirability habitats in V. persa and C. saliens varied from 38 % to 73 %, respectively. In conclusion, because of the ecological importance of these benthopelagic species, it is imperative to implement appropriate management strategies. These should include protecting desirable habitats, enforcing fishing bans in certain areas, and attempting to develop suitable areas to mitigate potential harm to their populations.