Multivariate analysis of Quercus castaneifolia C.A. Mey based on morphological characterizations

Quercus castaneifolia C.A. Mey. is an ecologically important oak species in the Hyrcanian forests of Iran, yet comprehensive studies integrating large-scale morphological datasets with multivariate analyses remain limited. The present study aimed to evaluate the phenotypic diversity of 51 Q. castaneifolia accessions collected from three natural populations in Golestan and Semnan provinces of Iran, and to identify the most informative traits for characterization and selection. A total of 52 morphological and pomological traits related to tree growth, leaf, nut, cupule, and kernel characteristics were assessed. One-way ANOVA, Pearson correlation analysis, hierarchical cluster analysis (HCA), and multiple regression analysis (MRA) were applied to elucidate trait variability and interrelationships. The results revealed substantial phenotypic diversity among the studied accessions, with 75% of the evaluated traits exhibiting coefficients of variation above 20%, indicating high discriminatory capacity. Reproductive traits, particularly nut and kernel dimensions and weights, were identified as the most informative and biologically meaningful descriptors. Strong and highly significant positive correlations were observed among nut size, nut weight, kernel size, and kernel weight, whereas the nut cover/kernel ratio exhibited negative associations with most size-related traits. Overall, the combined statistical analyses demonstrated clear patterns of phenotypic differentiation among accessions, and identified the key morphological traits contributing to the observed variability. HCA confirmed a structured yet heterogeneous phenotypic pattern among accessions. Overall, the integration of multivariate approaches demonstrated that fruit- and kernel-related traits are the most useful criteria for germplasm characterization, indirect selection, and future breeding programs in Q. castaneifolia. Based on a comprehensive assessment combining multivariate analyses and trait performance, 'Olang-5', 'Olang-11', 'Tooskestan-1', 'Tooskestan-15', and 'Tooskestan-10' emerged as the most promising accessions.