Dissecting Phenotypic Architecture and Trait Trade-Offs in Thai Aromatic Coconuts by Integrating Multivariate Phenomics and Machine Learning for Precision Breeding

Thai aromatic coconut faces persistent breeding challenges arising from limited genetic diversity, complex trait trade-offs, and increasing climate vulnerability. These constraints highlight the need for comprehensive phenotypic characterization to improve understanding of trait variation and support the identification of key traits associated with yield and quality improvement. This study aimed to dissect trait architecture and associations in Thai aromatic coconut using an integrated multivariate and machine learning framework. Two populations of Thai aromatic coconut, Ratchaburi (RB) and Pak Phanang (PP), were evaluated through comprehensive phenotypic characterization. Thirty-seven morphological, reproductive, and soil-influenced traits were evaluated using analysis of variance, broad-sense heritability estimates, Pearson correlation analysis, principal component analysis (PCA), hierarchical clustering, and machine learning models. The PP population exhibited superior water yield, indicated by a strong positive correlation between water content and TWW, and larger fruit size, but showed a pronounced trade-off with kernel weight. High phenotypic variability was observed for key traits (CV > 39%), accompanied by moderate to high heritability estimates. Principal component analysis revealed that PC1, PC2, and PC3 explained 32.2%, 13.0%, and 11.1% of the total phenotypic variation, respectively, accounting for a cumulative 56.3% of the observed variation among accessions. Random Forest models achieved high predictive accuracy for total water weight (R2 = 0.942), with water content (WC), fruit weight (FW), fruit diameter (FD), fruit length (FL), and hole spacing (HS) identified as the most influential predictors. Overall, the findings provide a non-destructive phenotypic framework for germplasm evaluation and trait-based selection in Thai aromatic coconut.