Modelling the Present and Future Distribution
In all models techniques for H. leucisculus AUC (range of 0.90e1.00), KAPPA (range of 0.78e0.92) and TSS (range of 0.70e0.89) were high predictive capacity. However, for H. leucisculus GLM, ANN, MaxEnt and RF were the best performing models, with AUC, KAPPA and TSS > 0.80 (Table. 2).
The greatest contribution to model performance were BIO8 = mean temperature of wettest quarter (38.68%), BIO 1 = annual mean temperature (23.14%), BIO 7 = temperature annual range (14.26%) and BIO 13= Precipitation of Wettest Month (10.89) (Table. 3).
Based on the future climate change scenarios (SSP126 and 585), suitable habitats will sharply increase over time across most of the H. leucisculus range (Fig. 4), and it tend to shift a wide range where human population density is more than other areas.