Differences in lipid composition among fruit parts and comprehensive evaluation of Cornus wilsoniana accessions

Background Lipid composition is a critical indicator for assessing oil-bearing plants, where mapping component distribution across parts aids in screening promising accessions. Exploring these spatial patterns and bioactive micro-constituents is essential for effective germplasm evaluation. Cornus wilsoniana , an emerging oil-producing tree species, yields fruits rich in lipid-soluble bioactives that modulate lipid metabolism and reduce oxidative stress. Profiling part-specific metabolites and establishing thorough quality assessment frameworks are thus crucial for identifying promising individual trees. Methods Twelve C. wilsoniana accessions were evaluated for 19 lipid components, including four tocopherols, three phytosterols, squalene, and 11 fatty acids, in whole fruits, pulp-peel composites, and seeds. Multivariate analyses and entropy weight-TOPSIS were used to characterize part-specific variation and comprehensively evaluate whole-fruit lipid quality. Results Boxplot analysis indicated significant differences in lipid bioactives and fatty acid concentrations among whole fruits, pulp-pulp composites, and seeds ( P < 0.05). Specifically, α-tocopherol peaked in pulp-peel composites (mean: 463.26 mg/kg). Linoleic acid (C18:2n6c) exhibited strong part specificity, reaching 66.90% in seeds—substantially higher than in whole fruits (47.80%) or pulp-pulp composites (43.73%) ( P < 0.05). Two-way ANOVA verified that accession, part type, and their interactions significantly influenced lipid profiles. Redundancy analysis (RDA) showed that seeds were primarily enriched in linoleic acid (C18:2n6c) and cis - 11-eicosenoic acid (C20:1), whereas palmitic acid (C16:0), α-linolenic acid (C18:3n3), palmitoleic acid (C16:1), and α-tocopherol accumulated mainly in whole fruits and pulp-pulp composites. Correlation analysis demonstrated strong synergistic relationships between α-tocopherol, β-sitosterol, and major unsaturated fatty acids within pulp-pulp lipids, whereas oleic acid negatively correlated with most key constituents. Furthermore, seed correlation networks were more decentralized than those of whole fruits and pulp-pulp composites. Principal component analysis (PCA) extracted five principal components explaining 87.00% of the total variance. Entropy weight-TOPSIS ranked GZCW-6, GZCW-2, and GZCW-12 highest in comprehensive whole-fruit lipid quality. HCA grouped the 12 accessions into three clusters, and post hoc OPLS-DA identified cis - 11-eicosenoic acid, heptadecenoic acid, palmitic acid, oleic acid, β-sitosterol, and α-tocopherol as major contributors to differentiation among the HCA-derived groups. Whole-fruit UFA/SFA and ω-3/ω-6 ratios ranged from 3.028–4.898 and 0.056–0.170, respectively, indicating marked variation in fatty acid nutritional profiles among accessions. Conclusion C. wilsoniana showed clear part-specific lipid distribution and substantial accession-associated variation. GZCW-6, GZCW-2, and GZCW-12 exhibited relatively favorable comprehensive lipid-quality profiles and may represent promising materials for further evaluation and targeted utilization.