An XGBoost-Based Morphometric Classification System for Automatic Subspecies Identification of Apis mellifera

Area of Science:

• Entomology
• Bioinformatics
• Machine Learning

Background:

• Accurate subspecies assignment for western honey bees (Apis mellifera) is vital for conservation and breeding.
• Traditional methods like morphometrics and molecular assays are time-consuming and expensive.

Purpose of the Study:

• To develop an efficient and accurate classification framework for Apis mellifera subspecies assignment.
• To utilize machine learning with routinely measurable characters for subspecies identification.

Main Methods:

• Developed an XGBoost-based classification framework using a compact set of measurable characters.
• Curated a labeled dataset of Apis mellifera workers and screened features by importance.
• Assessed model performance using five-fold cross-validation and SHAP analyses.

Main Results:

• The XGBoost model achieved high performance (accuracy = 0.98, F1 = 0.99, AUC = 0.99) using only the top 10 characters (forewing venation angles, abdominal plate metrics).
• The model outperformed standard machine learning baselines.
• Misclassifications were primarily observed in morphologically similar lineages.

Conclusions:

• The developed model serves as a rapid triage tool, complementing genetic testing for Apis mellifera subspecies assignment.
• The framework is scalable, interpretable, and portable to other insect taxa for custom morphometric model development.