Objective To investigate growth rhythm using nine Eucalyptus pellita provenances in southern Papua New Guinea and northeastern Australia, provide a scientific basis for the breeding and management of E. pellita.

Method A randomized block design experiment was conducted, and nested analysis of variance was utilized to analyze the growth rhythm variation patterns of 98 families from nine E. pellita provenances. The information of climate and environment factors in various sources were obtained, and Pearson method was used to reveal the correlation between growth traits and geography/climate factors. The fitting results of Logistic, Gompertz and Von Bertalanffy models were compared to select the optimal model, and calculate the growth rhythm of E. pellita. The excellent family selection was carried out by systematic clustering method.

Result Significant differences were observed in the tree height and DBH of 2.5-year-old E. pellita at provenance and family levels (P<0.05). Furthermore, a significant and positive correlation was found between isothermality and tree height of the provenances, exhibiting typical zonal variation patterns. The Logistic model had the best fitting effect on tree height and DBH of E. pellita. The Logistic model calculated the growth rhythms, with the average values of the maximum acceleration period (T₁), maximum deceleration period (T₂), and linear growth period (L) for tree height being 96, 636 and 540 d respectively, as well as for DBH being 165, 681 and 516 d respectively. A systematic cluster analysis was conducted using the growth traits of 2.5-year-old E. pellita, and 29 general families, 30 medium families and 39 outstanding families were selected from 98 families. All the families of the three grades were further classified into three categories of I, II and III according to the cluster of growth rhythm T₁.

Conclusion The growth traits and growth rhythm of 2.5-year-old E. pellita exhibited considerable variation at provenance and family levels. The growth rhythm calculated using Logistic model can provide a theoretical basis for optimizing seedling rearing and management, as well as improved variety breeding of E. pellita.