A novel hybrid model for species distribution prediction using neural networks and Grey Wolf Optimizer algorithm

Area of Science:

• Ecology
• Computational Biology
• Machine Learning

Background:

• Backpropagation neural networks (BPNN) are used for species distribution modeling but face parameter setting challenges.
• Optimizing connection weights in BPNN is crucial for accurate species distribution simulations.
• Existing models like GBM, GLM, MaxEnt, and RF have varying performance in species distribution prediction.

Purpose of the Study:

• To develop a novel hybrid algorithm, GNNA, by integrating the Grey Wolf Optimizer (GWO) with BPNN for enhanced species distribution prediction.
• To evaluate the performance of GNNA against established species distribution models (SDMs) using multiple evaluation metrics.
• To assess the efficacy of GNNA in forecasting potential non-native distributions of invasive plant species.

Main Methods:

• Developed the Grey Wolf Optimizer algorithm optimized backpropagation neural networks (GNNA) by incorporating GWO's global search capabilities into BPNN.
• Compared GNNA with Generalized Boosting Model (GBM), Generalized Linear Model (GLM), Maximum Entropy (MaxEnt), and Random Forest (RF).
• Utilized Area Under the Receiver Operating Characteristic Curve (AUC), Cohen's Kappa, and True Skill Statistic (TSS) for model evaluation across 23 species.

Main Results:

• GNNA demonstrated significantly improved predictive performance compared to the standard BPNN.
• GNNA outperformed GLM and GBM and showed comparable results to MaxEnt and RF, particularly with small sample sizes.
• GNNA proved highly effective in predicting the potential distribution of invasive plant species.

Conclusions:

• The integration of GWO with BPNN results in a powerful and accurate tool (GNNA) for species distribution modeling.
• GNNA offers a robust alternative to existing SDMs, especially when dealing with limited data.
• GNNA shows significant potential for ecological applications, including the management of invasive species.