Accurate Predictive Modeling of Conservation Status in Animal Species Using Supervised Learning

Area of Science:

• Conservation biology
• Population genetics
• Machine learning

Background:

• Global anthropogenic activities and climate change threaten wildlife extinction.
• The International Union for Conservation of Nature (IUCN) Red List currently does not incorporate genetic data for species status assessments.
• Molecular data is increasingly available for conservation applications.

Purpose of the Study:

• To investigate the utility of genetic diversity and differentiation in predicting species endangerment.
• To develop machine learning models for predicting IUCN threat levels using genetic data.
• To assess the accuracy of genetic markers in classifying species at risk of extinction.

Main Methods:

• Utilized data from over 7300 animal studies from the MacroPopGen database and 450 articles from DataDryad.
• Analyzed genetic diversity and differentiation across various invertebrate and vertebrate taxa.
• Applied machine learning algorithms to predict species endangerment based on genetic data and IUCN classifications.

Main Results:

• Found significant decreases in genetic diversity and increases in genetic differentiation in bird and fish taxa with higher endangerment levels (p < 0.05).
• Developed machine learning models that accurately predicted IUCN threat levels with 93.16% overall accuracy.
• Demonstrated the strong correlation between genetic metrics and species conservation status.

Conclusions:

• Genetic diversity and differentiation are valuable predictors of wildlife endangerment.
• Integrating genomic data into conservation assessments can significantly enhance the accuracy of species threat level predictions.
• Future conservation status assessments should incorporate genetic data alongside demographic, phenotypic, and census data for a comprehensive evaluation.