Objective To analyze the phenotypic and physiological characteristics of different groups of P. deltoides, and to compare the effectiveness and representativeness of the phenotypic core collection constructed by different sampling strategies.

Method Based on the results of population structure analysis of phenotypic and physiological traits analyzed in previous studies, the P. deltoides resources were grouped, and 21 phenotypic and physiological characters of different groups were analyzed by multiple comparison. The Euclidean distance between clones was calculated, and the individuals of each group were clustered by the method of intergroup connection. The clones were screened by random sampling, deviation degree sampling, locus priority sampling and trait frequency sampling strategies, respectively. The best sampling strategy was selected by comparing and analyzing the diversity of core collection constructed by different sampling strategies. On the basis of the phenotypic core collection, the optimized core collection was constructed by making the phenotypic retention ratio (RPR) reaching 100.00%. The representativeness and validity of core collections were evaluated by comparing the mean, variance, range, coefficient of variation and RPR between the core and the original collections.

Result The height growth, leaf number and leaf nutrient characters of the individuals of the provenance ‘Was’ were higher. The roots of the clones of provenances ‘Iow’ and ‘Que’ were developed. The stem growth, leaf morphology and single leaf growth of the clones of provenances ‘Mis’, ‘Lou’ and ‘Ten’ were prominent. The phenotypic core collection constructed by the strategy of character frequency sampling represented the diversity information of the original collection with the least collection. The effectiveness of the core collection was better than that constructed by the strategy of random sampling, deviation degree sampling and site priority sampling. Finally, the phenotypic core collection consisting of 27 clones was constructed, with the sampling ratio of 10.47%. The mean difference percentage (MD), variance difference percentage (VD), range coincidence rate (CR), coefficient of variation coincidence rate (VR) and RPR were 0.00%, 80.95%, 100.00%, 148.94% and 90.85%, respectively. On the basis of core collection, 5 individuals were added to construct phenotype optimized core collection, so that the RPR reached 100.00%, and MD, VD, CR and VR were 0.00%, 71.43%, 100.00% and 142.19%, respectively. The results of PCoA analysis showed that the phenotypic core collection and the optimized core collection were well representative.

Conclusion There are significant differences in root, stem and leaf growth traits among different groups of P. deltoides germplasm resources. Through the strategy of character frequency sampling, the core collection of phenotype and the optimization of core collection of P. deltoides could retain the diversity of the original collection, greatly reduce the size and redundancy of the collection, and improve the variation level. The results could provide scientific basis for the conservation, management and utilization of P. deltoides germplasm resources.