Abstract:

Objective: This study aims to develop an efficient prediction system that does not require genotypic data through multi-dimensional phenomics analysis, reveal the key phenotypic characteristics of early-fruiting Juglans regia (walnut), and provide clues for deciphering their genetic linkage relationships, so as to offer a methodological reference for investigating the formation mechanisms of other special groups in woody plants. Method: A total of 156 typical and closely related mature walnut samples from mature trees preserved at the National Long-term Scientific Research Base in Aksu, Xinjiang were targeted, and 31 early-fruiting and 8 late-fruiting germplasm resources were selected for phenotype identification. According to international standardized measurement protocols, a total of 31 phenotypic traits were systematically determined, covering four categories: stem/branch morphology, leaf characteristics, bud development, and fruit external quality. A multi-dimensional integrative strategy was used for data analysis. Firstly, orthogonal partial least squares-discriminant analysis (OPLS-DA) was applied to reduce the dimensionality of high-dimensional phenotypic data and identify discriminatory variables. Then, analytic hierarchy process (AHP) was used to assign weight coefficients to each variable. Finally, grey relational analysis (GRA) was integrated to construct a synergistic model incorporating multi-parameter interactions. The entire workflow, from data processing to model construction, was automated via custom Python scripts. This analytical framework operated entirely on phenotypic data, achieving classification and prediction of complex traits without reliance on genotypic information. Result: The multivariate analysis was used to identify 13 core phenotypic parameters significantly associated (P<0.01) with early-fruiting traits in walnut, covering fruit traits (nut weight, nut transverse diameter, and shell thickness), leaf traits (second leaflet pair-petiole angle below terminal leaf, leaflet width, basal leaf length, basal leaf area, basal leaf-petiole angle, terminal leaf area, leaflet area, and basal leaf width), stem/branch trait (short-shoot internode length), and mixed bud trait (mixed bud insertion position). The model constructed based on these parameters achieved 90.81% classification accuracy on an independent validation set, demonstrating strong predictive power of the selected phenotypic markers for early-fruiting traits and confirming the reliability of this genotype-free prediction system. Furthermore, the directionality of correlations with early-fruiting traits was clarified: nut transverse diameter, basal leaf area, basal leaf-petiole angle, leaflet area, and mixed bud insertion position showed positive correlations with the formation of early fruiting traits, while nut weight, shell thickness, and internode length showed negative correlations. Conclusion: This study has identified the core phenotypic traits of early-fruiting walnuts and elucidated the basis for their trait formation. The established phenotypic marker screening system provides a standardized method for early selection of early-fruiting germplasm without requiring genotypic data. Furthermore, the proposed multi-dimensional analytical framework is applicable not only to deciphering genetic mechanisms in walnuts but also extensible to characterizing other special groups in woody plants, thereby providing phenotypic-level theoretical support for germplasm utilization.

Key words: Juglans regia, early-fruiting, phenotypic marker, prediction model, orthogonal partial least squares-discriminant analysis (OPLS-DA), analytic hierarchy process (AHP), grey relational analysis (GRA)