Mixed-scale multivariate analysis reveals phenotypic structure in wood apple (Feronia limonia L.).

Wood apple (Feronia limonia L.) is an underutilized perennial fruit tree with substantial ecological, nutritional, and economic potential, yet its phenotypic diversity and trait organization remain poorly characterized. Here, we applied a mixed-scale multivariate framework to resolve phenotypic structure in 62 wood apple genotypes using 31 ordinal and categorical vegetative, leaf, floral, fruit, and seed descriptors. Trait interrelationships were examined through the complementary use of Spearman’s rank correlation and Cramér’s V association analyses, capturing both directional rank-based dependencies and scale-independent categorical linkages. Hierarchical clustering based on Gower distance separated the genotypes into three distinct phenotypic clusters, with inter-cluster dissimilarities (0.92–1.18) consistently exceeding intra-cluster variation (0.42–0.55), indicating well-supported phenotypic stratification based on cluster validation. Multiple Correspondence Analysis (MCA) explained 23.30% of total inertia across the first two dimensions, with tree growth habit, branch angle, tree shape, and fruit color emerging as the principal drivers of phenotypic differentiation. Vegetative and leaf traits formed a tightly integrated module, whereas fruit-related traits displayed weaker monotonic but persistent categorical associations, reflecting partial phenotypic independence. The strong concordance among association analyses, clustering, and MCA indicates structured patterns of coordinated and partially independent trait associations in wood apple. Overall, this study demonstrates the effectiveness of mixed-scale multivariate approaches for resolving complex trait architecture in underutilized perennial fruit crops and provides a quantitative phenotypic framework to support germplasm conservation, parent selection, and ideotype-oriented improvement of wood apple.