Machine Learning Mapping of Animal Species Vulnerability Under Future Climate Scenarios

Abstract

Introduction: The changing climate is impacting the distribution and viability of animals through changes in temperature, rainfall, vegetation, and other environmental conditions. Thus, identifying locations that may become vulnerable in future climatic scenarios becomes necessary for conserving the environment and for developing adaptation strategies. This research proposes a machine learning algorithm to map the vulnerability of animal species based on different future climatic scenarios.

Methodology: The presence of species was linked with climatic and environmental features like temperature, precipitation, elevation, vegetation features, and land cover. The data were preprocessed and split into training and test data sets. Three machine learning models namely support vector machine, random forest, and XGBoost were assessed in terms of accuracy, precision, recall, F1 score, and area under receiver operating characteristic curves. The model with the highest performance was used for predictions in current as well as future SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios. Predicted changes in habitat suitability were converted into spatial vulnerability classes.

Results: XGBoost model showed better predictive power with 83.2% accuracy, 82.6% precision, 82.1% recall, 82.3% F1-score, and AUC of 0.88. Vulnerability assessment showed a progressive rise in regions categorized as high and very high vulnerability to the impacts of climate change with increasing climate forcing scenarios. High and very high vulnerabilities combined increased from 27% with the present-day situation to 49% with the SSP5-8.5 climate forcing scenario. The results demonstrate that machine learning can effectively capture complex relationships between species vulnerability and environmental predictors while providing spatially interpretable predictions.

Conclusion: Such a framework would present an effective method for determining vulnerable areas due to their sensitivity to the climate and predicting any future changes regarding species vulnerability. It may be beneficial in conservation, habitat preservation, ecological monitoring, and adaptation to climate change.