Multivariate analysis of wild Prunus × keredjensis (Browicz) A.E.Murray based on morphological and pomological characterizations.

Prunus × keredjensis (Browicz) A.E. Murray is a rare natural hybrid between P. scoparia and P. lycioides. Despite its potential importance for almond breeding and genetic resource conservation, it has never been systematically characterized. Morphological and pomological evaluations are essential for elucidating the genetic potential, taxonomic position, and adaptive strategies of this taxon. This study represents the first comprehensive investigation of P. × keredjensis, integrating multivariate statistical approaches to examine trait relationships and phenotypic structure among wild accessions collected from three distinct ecological regions of Tehran Province, Iran. A total of 75 wild P. × keredjensis accessions were evaluated based on 76 morphological and pomological traits. One-way analysis of variance (ANOVA) revealed highly significant differences among accessions (p < 0.05), with coefficients of variation (CV) ranging from 10.28% (exocarp thickness) to 291.36% (petal apex notching). Of the examined traits, 68 (89.47%) exhibited CV values exceeding 20%, highlighting substantial intraspecific phenotypic diversity. Correlation analysis indicated strong positive associations among key nut and kernel traits, including nut length–width (r = 0.72**), nut length–diameter (r = 0.73**), and kernel length–weight (r = 0.82**). Multiple regression analysis (MRA) identified kernel length (β = 0.49, p < 0.001), kernel width (β = 0.31, p < 0.001), and stigma length (β = 0.22, p < 0.001) as the most influential predictors of kernel weight. Principal component analysis (PCA) extracted 19 components with eigenvalues > 1, explaining 84.31% of the total variance. The first three components accounted for 28.74% of the total variation, with nut, kernel, and floral characteristics contributing most to genotype differentiation. Hierarchical cluster analysis (HCA) grouped the accessions into two main and four subclusters, revealing clear morphological differentiation patterns associated with geographic origin. Across these analyses, the accessions ‘Khojir-27’, ‘Khojir-40’, ‘Khojir-15’, ‘Khojir-46’, ‘Khojir-26’, ‘Khojir-36’, ‘Khojir-49’, ‘Khojir-54’, ‘Khojir-10’, ‘Azgi-2’, ‘Azgi-4’, ‘Khojir-8’, ‘Azgi-1’, ‘Kalak-e-Bala-9’, ‘Khojir-33’, and ‘Kalak-e-Bala-7’ consistently combined larger nuts and kernels, higher kernel weight, favorable shell thickness, and desirable tree vigor and fruit set compared with the remaining accessions, and were therefore classified as superior genotypes. These superior accessions represent concrete candidates for further multi-environment testing and potential use as parents in almond improvement programs. From a breeding perspective, the strong correlations and regression models indicate that simple nut and kernel measurements can be used as efficient proxies for kernel weight in early selection. From a conservation standpoint, the existence of distinct, region-specific clusters underlines the need to conserve multiple populations to capture the full range of phenotypic diversity present in this rare hybrid. Overall, the findings provide concrete, trait-based evidence for the taxonomic distinctiveness and breeding relevance of P. × keredjensis and contribute to a broader understanding of phenotypic diversity and adaptive evolution within the Prunus genus.