Morphometric trait analysis and machine learning-based yield modeling in wood apple (Feronia limonia L.)

Area of Science:

• Horticulture
• Plant Breeding
• Agricultural Science

Background:

• Wood apple is an underutilized yet significant fruit tree in the Indian subcontinent.
• Its potential as a climate-resilient crop is high, but yield determinants are not well understood.
• This study addresses the knowledge gap in wood apple yield variability.

Purpose of the Study:

• To quantify how morphometric descriptors of canopy architecture, floral, and fruit traits explain yield variation in wood apple.
• To develop a data-driven framework for identifying trait combinations governing productivity.
• To provide insights for ideotype selection and precision orchard design.

Main Methods:

• Utilized multivariate statistics and explainable machine-learning models (Random Forest + SHAP).
• Analyzed morphometric variability across 62 wood apple genotypes.
• Integrated Principal Component Analysis (PCA), correlation analysis, and hierarchical clustering.

Main Results:

• Significant morphometric variability was observed across genotypes.
• Random Forest model achieved high predictive performance (R² = 0.84).
• Key predictors of yield included tree shape, open flower color, and pulp color; compact canopies and superior reproductive traits led to higher yields.

Conclusions:

• Wood apple yield is governed by an integrated set of architectural and reproductive traits.
• Genotypes with compact canopies and preferred pulp characteristics are promising for high productivity.
• Explainable machine-learning tools offer a robust framework for trait-based breeding and climate-smart orchard design.