Introduction
The genus Hydrangea (family Hydrangeaceae) comprises important ornamental plants that are widely cultivated in the global floriculture industry (Cai et al. 2015). Native mainly to temperate regions of eastern Asia and eastern North America, Hydrangea species are valued for their large, showy inflorescences that display diverse morphologies and colors (McClintock 1957). Although H. macrophylla has traditionally dominated commercial cultivation, recent cultivar development has produced an increasing number of registrations in alternative species, particularly H. paniculata and H. arborescens (UPOV 2025), which offer distinct ornamental characteristics and cultivation advantages.
Pruning is a key horticultural practice applied to ornamental woody shrubs to improve inflorescence quality, regulate plant architecture, and enhance ornamental value. Previous studies have shown that pruning influences shoot development, branching, and floral characteristics in a range of woody ornamentals (Kovaleski et al. 2015;Zhang et al. 2022). These responses are mediated physiologically by the disruption of apical dominance and the reallocation of carbohydrates and growth regulators to the remaining buds, which can increase shoot proliferation and inflorescence size (Thimann et al. 1971;Cline and Harrington 2007). Despite the widespread use of pruning in ornamental production, systematic comparative studies of pruning effects across Hydrangea species remain limited.
Botanical and physiological differences among Hydrangea species suggest the potential for species-specific responses to pruning. Hydrangea arborescens is a deciduous shrub that flowers on current-season growth, forming flower buds on new shoots during the growing season (Huang et al. 2023). Hydrangea paniculata is also classified as a current-season (new-wood) flowering shrub (Dirr 2004) but is a more strongly woody, treelike shrub that differs markedly from H. arborescens in growth habit and shoot development. These differences suggest that the efficacy of pruning and the optimal pruning strategy may vary considerably among species and cultivars, yet data to inform cultivar-specific recommendations remain scarce.
The present study examined the effects of early-spring hard pruning on inflorescence morphology in three Hydrangea cultivars representing two species: H. arborescens ‘NCHA1’ and H. paniculata ‘Limelight’ and ‘Couharie.’ The objectives were to (1) quantify the effects of hard pruning on inflorescence dimensions across species and cultivars, (2) evaluate the effect of pruning on the uniformity of inflorescence size, and (3) provide data-driven, cultivar-specific pruning recommendations for commercial Hydrangea production.
Materials and Methods
Plant materials and pruning treatment
Three Hydrangea cultivars—H. paniculata ‘Limelight’ and ‘Couharie,’ and H. arborescens ‘NCHA1’—maintained at the National Forest Seed and Variety Center (NFSV), Chungju, Republic of Korea (36.88°N, 127.98°E; approximately 150 m elevation) were used. Plants were grown under open-field conditions and spaced 1.0 m within and between rows. The original calcareous clay soil was amended with a 50 cm layer of decomposed granite soil (masato) mixed with sand to improve drainage. Fertilization and irrigation followed standard horticultural management and were applied uniformly to all plants. ‘Limelight’ and ‘NCHA1’ had been field-established in 2023 and ‘Couharie’ in 2024; thus, at the time of pruning (March 2025), plants corresponded to approximately their second and first growing seasons after field establishment, respectively. For each cultivar, 20 plants of comparable size were randomly assigned to a pruning treatment (n = 10) or a non-pruned control (n = 10) in a completely randomized design. For ‘NCHA1,’ plant mortality during the growing season reduced the number of analyzable plants to 6 (pruned) and 7 (non-pruned).
Pruning was performed in early March 2025, before the onset of active spring growth. Plants in the pruning treatment were cut back to approximately 5 cm above ground level. This severe cut was applied deliberately as a standardized experimental treatment: it removed essentially all existing above-ground growth and imposed a uniform, high pruning intensity across plants and cultivars, forcing regrowth from basal dormant buds. Plants in the non-pruned control were left unpruned, retaining their existing above-ground structure.
Inflorescence sampling and measurement
Inflorescences were sampled at full bloom (peak inflorescence development) for each cultivar, on 20 June 2025 (H. arborescens ‘NCHA1’), 10 July 2025 (H. paniculata ‘Couharie’), and 30 July 2025 (H. paniculata ‘Limelight’). On each plant, all fully open, undamaged inflorescences present at full bloom were measured (1–7 per plant, depending on the number borne); no subsampling was applied. Because the number of inflorescences differed among plants, all inflorescence measurements within a plant were averaged to a single plant-level mean, which served as the experimental unit (see Data processing and statistical analysis).
Each inflorescence was photographed against a white background under standardized, diffuse lighting, with a ruler placed in the same plane as the inflorescence for scale calibration (Figs. 1, 2). To minimize projection distortion, photographs were taken with a Nikon Z5 digital camera (Nikon, Tokyo, Japan) mounted on a tripod at a fixed working distance, with the optical axis oriented perpendicular to the plane in which width and height were measured. Images were analyzed with ImageJ v1.54g (Schneider et al. 2012; National Institutes of Health, Bethesda, MD, USA). Inflorescence width was defined as the maximum horizontal extent perpendicular to the inflorescence axis, and inflorescence height as the maximum extent along the axis from base to apex; all measurements were calibrated to millimeters (mm) using the ruler.
Data processing and statistical analysis
To avoid pseudoreplication arising from multiple inflorescences measured per plant (Hurlbert 1984), the plant-level means described above were used as the experimental units in all analyses (n = 6–10 plants per treatment group). All analyses were performed in R v4.5.1 (R Core Team 2025). Descriptive statistics (mean, standard deviation, and coefficient of variation []) were calculated for each cultivar × treatment combination. Because equal variances could not be assumed, pruned and non-pruned plant-level means were compared using Welch’s t-test, performed separately for inflorescence width and height within each cultivar. Homogeneity of variances between pruned and non-pruned plants was assessed with the Fligner–Killeen test. Data were visualized with ggplot2 (Wickham 2016), using boxplots overlaid with individual plant-level means.
Results
Inflorescence dimensions
Descriptive statistics for plant-level mean inflorescence width and height are presented in Table 1. Hard pruning increased the mean inflorescence width and height of all three cultivars (Table 1, Fig. 3). Pruned H. paniculata ‘Limelight’ plants had a mean inflorescence width of 169.2 ± 7.7 mm and height of 174.9 ± 9.8 mm, compared with 143.8 ± 7.6 mm and 143.3 ± 10.8 mm in non-pruned plants. Pruned H. paniculata ‘Couharie’ plants had a mean width of 146.6 ± 19.3 mm and height of 147.9 ± 24.3 mm, compared with 120.3 ± 14.6 mm and 120.5 ± 23.8 mm in non-pruned plants. Pruned H. arborescens ‘NCHA1’ plants had a mean width of 156.5 ± 10.5 mm and height of 63.5 ± 6.6 mm, compared with 137.9 ± 30.6 mm and 56.1 ± 9.5 mm in non-pruned plants.
Statistical comparisons
Welch’s t-test on plant-level means revealed significant pruning effects on both inflorescence width and height in the two H. paniculata cultivars (Table 2). ‘Limelight’ showed increases of 25.4 mm in width (95% CI 18.2–32.6 mm; p < 0.001) and 31.6 mm in height (95% CI 21.9–41.3 mm; p < 0.001), and ‘Couharie’ showed increases of 26.4 mm in width (95% CI 10.2–42.5 mm; p = 0.003) and 27.4 mm in height (95% CI 4.8–50.0 mm; p = 0.020). In H. arborescens ‘NCHA1,’ inflorescence dimensions showed the same increasing trend with pruning (+18.5 mm in width, +7.4 mm in height); however, these differences were not statistically significant (width: 95% CI −10.2 to 47.3 mm, p = 0.174; height: 95% CI −2.5 to 17.3 mm, p = 0.129). The lack of significance in ‘NCHA1’ is likely attributable to the limited number of surviving plants available for analysis (n = 6 pruned, n = 7 non-pruned).
Variability of inflorescence size
In H. arborescens ‘NCHA1,’ pruned plants showed a lower coefficient of variation (CV) of inflorescence width than non-pruned plants (6.7% vs. 22.2%), with a smaller reduction for height (10.4% vs. 16.9%; Table 1). However, the Fligner–Killeen test did not indicate a significant difference in variance between treatments for either trait (p > 0.05), and the two H. paniculata cultivars showed no consistent change in CV (Table 1). Thus, although pruning may reduce the variability of inflorescence width in H. arborescens, this trend was not statistically supported at the present sample size.
Discussion
Hard pruning significantly increased inflorescence width and height in both H. paniculata cultivars, with gains of 25–26 mm in width and 27–32 mm in height (Table 2). These increases are consistent with well-documented responses to pruning in woody ornamentals, in which the removal of competing shoots and the associated disruption of apical dominance redirect carbohydrate and nutrient resources to the remaining developing shoots, supporting larger inflorescences (Thimann et al. 1971;Cline and Harrington 2007).
In H. arborescens ‘NCHA1,’ the increases in inflorescence size were not statistically significant. This most likely reflects reduced statistical power: plant mortality during the growing season lowered the number of analyzable plants to six pruned and seven non-pruned, rather than the absence of a biological response, given that the direction and magnitude of the change paralleled those in the H. paniculata cultivars. Pruned ‘NCHA1’ plants also showed a lower CV for inflorescence width (6.7% vs. 22.2%; Table 1); however, this apparent gain in uniformity was not statistically supported (Fligner–Killeen test, p > 0.05) and, given the small sample, should be regarded as a preliminary observation. If confirmed in larger trials, such an effect could reflect more synchronous regrowth from basal buds after hard pruning (Huang et al. 2023).
Taken together, these results indicate species- and cultivar-specific responses to hard pruning. In the two H. paniculata cultivars, pruning clearly enhanced inflorescence size, whereas in H. arborescens ‘NCHA1’ the size response was not statistically significant, owing to the limited number of surviving plants. For commercial production, early-spring hard pruning can therefore be recommended for the H. paniculata cultivars to increase inflorescence size; for H. arborescens, larger trials are needed to confirm both the size response and the suggested improvement in size consistency.
This study has several limitations. It was conducted at a single site during one growing season, and the number of H. arborescens plants was reduced by mortality, limiting the statistical power for this species. Inflorescence size was the only floral trait quantified; effects on flower number, flowering duration, and overall floral development were not measured and should be examined in future work. Further studies incorporating multiple years, sites, and a broader range of cultivars, as well as different pruning intensities and timings, are needed to confirm and generalize these findings and to optimize cultivar-specific pruning strategies.








