Species, cultivar, and cutting maturity drive rooting success in Osmanthus

1 Introduction

The genus Osmanthus (Oleaceae) comprises approximately 30 species of evergreen shrubs and small trees native primarily to temperate and subtropical regions of Asia, with China as a center of diversity (Xiang and Liu, 2008; Yuan et al., 2025). Several species, including O. fragrans (fragrant tea-olive), O. heterophyllus (holly tea-olive), and the hybrid O. × fortunei (traditionally treated horticulturally as a species and referred to as O. fortunei hereafter) are widely cultivated as ornamentals for their glossy foliage, fragrant flowers, and adaptability to landscape use (Alexander, 2019). More than 160 named cultivars of O. fragrans alone have been described, reflecting significant variation in plant architecture, leaf morphology, and ornamental traits (Xiang and Liu, 2008). This diversity, combined with increasing demand for improved selections, underscores the need for efficient propagation methods to support nursery production and cultivar preservation.

Vegetative propagation by stem cuttings is the most common method for producing woody ornamentals because it maintains genetic fidelity and can be integrated into containerized production systems (Dirr and Heuser, 2006; Hartmann et al., 2013). Cutting type, defined by the degree of tissue lignification at the time of collection, strongly influences rooting success. Softwood cuttings taken from actively growing shoots are used for vigorously growing species such as lilac (Syringa) and azalea (Ryals et al., 2021; Sanchez et al., 2020). Hardwood cuttings, collected from dormant stems, are common for deciduous shrubs like forsythia (Forsythia) and grapes (Vitis) and narrow-leaved evergreens such as junipers and yews (Evans and Blazich, 1999; Yeager, 2020). Semi-hardwood cuttings, prepared from partially lignified shoots after a growth flush, are widely used for broadleaf evergreens such as camellia (Camellia) and holly (Ilex), and are generally preferred in commercial nurseries because they root more quickly than hardwood cuttings and fit established production schedules (Evans and Blazich, 1999; Hartmann et al., 2013).

Although Osmanthus species can be propagated by seed, this approach is rarely practical for nurseries. Seeds exhibit multiple dormancy mechanisms that require extended stratification physical scarification, or chemical treatments (Tang et al., 2013). Furthermore, the genus is androdioecious such that male cultivars produce no seed at all (Alexander, 2023; Xu et al., 2014). These constraints, combined with the need to maintain clonal fidelity, make vegetative propagation by stem cuttings the primary method for commercial production. However, reports of rooting success are inconsistent: some species root readily from semi-hardwood cuttings, while others respond poorly or require hardwood material (Blazich and Acedo, 1989a, 1989b; Lambert et al., 2011). Furthermore, cultivar-level variation in rooting ability has been noted, but systematic evaluations across species and cutting types remain limited. Similar challenges have been documented in olive (Olea europaea), a close relative in Oleaceae, where rooting success varies widely among cultivars and cutting types (de Oliveira da Silva et al., 2012; Walali and Loussert, 1990).

Despite the widespread use of stem cuttings in Osmanthus production, propagation guidelines remain inconsistent, especially regarding how species differences, cultivar identity, and the maturity of cutting wood interact to influence rooting. Prior studies have typically evaluated single species or single cutting types, leaving a critical gap in understanding how these factors jointly affect survival and root development across the genus. Furthermore, little information exists on whether stock-plant overwintering conditions meaningfully alter cutting performance in evergreen broadleaf ornamentals such as Osmanthus. To address these gaps, this study simultaneously compared three species, multiple cultivars within each species, three levels of wood maturity available at a single collection time, and two overwintering environments. We anticipated that rooting success would vary predictably with tissue maturity and genotype, and that auxin would enhance root system development. Our goal was to generate evidence-based guidelines that accommodate the physiological diversity within Osmanthus and support reliable propagation of both common and specialty selections.

2 Materials and methods

Seven Osmanthus cultivars and two unnamed accessions representing three species (n = 9 genotypes) were obtained from the U.S. National Arboretum or commercial nurseries and grown in containers at the Tennessee State University Nursery Research Center in McMinnville, TN, USA. Stem cuttings of Osmanthus fortunei (cvs ‘Fruitlandii’, ‘San Jose’, and an unnamed accession), O. fragrans (cvs ‘Fodingzhu’, ‘Beni Kin Mokusei’, and an unnamed accession), and O. heterophyllus (cvs ‘Goshiki’, ‘Head-Lee Fastigiate’, and ‘Rotundifolius’) were collected from 24 June 2016 to 8 July 2016 from 3-year old plants. Cuttings were collected during a single seasonal window, during which shoots on each plant exhibited a range of tissue maturation stages. We therefore use the terms non-lignified, semi-lignified, and lignified strictly to denote the anatomical maturity of tissues available at that time, rather than the seasonal softwood–semi-hardwood–hardwood framework commonly used for deciduous species. All cuttings were collected concurrently, and wood-type labels refer only to differential lignification at the time of harvest. Plants were maintained in 7-gallon pots under full sun and micro-irrigated using spray stakes. Growing media consisted of pine bark amended with 6.6 kg·m−3 19N-2.1P-7.4K Osmocote Pro fertilizer (Scotts-Sierra Horticultural Products Co., Maryville, OH, USA), 0.6 kg·m−3 Micromax (Scotts-Sierra Horticultural Products Co.), 0.6 kg·m-3 iron sulfate, and 0.2 kg·m−3 Epsom salts.

Plants had spent the previous winter either in a pot-in-pot (winter exposed) or in a covered polyhouse heated to -1 °C (winter protected). Each cultivar in each overwintering treatment was represented by three stock plants, although cutting quality varied among plants. To minimize physiological artifacts, cuttings were preferentially taken from the healthiest plants within each cultivar × overwintering group. When all three plants were of comparable quality, cuttings were distributed evenly; when plants differed noticeably, more cuttings were taken from vigorous individuals. Because individual cuttings were not uniquely assigned to specific mother plants, stock-plant identity could not be included as a random factor in the statistical model. As a result, the experimental unit in this study is the individual cutting, and inferences apply at the level of cutting responses rather than at the level of stock-plant or cultivar means. This approach results in partial biological replication at the stock-plant level and may introduce pseudo-replication within cultivar × wood type × overwintering combinations. We address this in the Discussion by describing limitations and by interpreting the results conservatively.

Seasonal flushing was diverse even on a single plant, such that at the time of collection some shoots were soft, floppy, and pubescent (non-lignified: NL), some were thickened but still flexible (semi-lignified: SL), and some were rigid (lignifed: L). Six cuttings per wood type (non-lignified, semi-lignified, lignified) were taken from winter-exposed and winter-protected plants of each cultivar, for a total of 324 cuttings. This resulted in a factorial structure of 3 species × 9 genotypes × 3 wood types × 2 overwintering treatments × 3 IBA concentrations. Because cuttings were sampled from multiple stock plants but not tracked to those plants, the design is properly considered a cutting-level factorial experiment using partial biological replication rather than a hierarchical (nested) stock-plant design.

Stem cuttings (approximately 10 cm long) were lightly scored and treated with 0, 1000, or 4000 ppm auxin (applied as KIBA [MilliporeSigma, Burlington, MA, USA], the potassium salt of IBA; hereafter IBA for simplicity). All auxin treatments followed the same protocol (15-second basal dip), and no solvent or surfactant differences occurred among treatments. Cuttings were stuck in 5-inch pots in fresh pine bark screened to 1.6 cm. Intermittent mist operated 6 sec every 5 min from 7:00 a.m. to 7:00 p.m. daily. A natural photoperiod was provided. After 6 weeks, cuttings were gently tugged and graded visually. Plants that removed easily or were visually dead were discarded. After 12 weeks, remaining plants were removed from containers and gently rinsed. Plants with at least one primary root were considered alive and retained for further analysis. Number of primary roots and root area were determined for each plant. Primary roots were those emerging directly from the stem or callus area. ImageJ was used to determine root area from binary-scaled photographs (Figure 1). Percentages reported in Results (e.g., survival at 6 weeks, rooting at 12 weeks) are always based on the total number of cuttings originally stuck, not only surviving cuttings.

Statistical analysis was performed using the SAS software system v.9.4 for Windows. The general linear model procedure was used to partition variation in percentage of surviving stems at six weeks, percentage of rooted stems at 12 weeks, number of primary roots per stem at 12 weeks, and root area per stem at 12 weeks into sources attributable to species, cultivar, wood type, source plant winter protection, and KIBA concentration. Model selection was performed using the PROC GLMSELECT procedure, which evaluates candidate models based on adjusted R² to account for model complexity and Akaike’s Information Criterion (AIC). The final model was selected at the step where the Akaike Information Criterion (AIC) reached a local minimum, and the adjusted R² was maximized, indicating an optimal balance between model fit and complexity. Model selection terminated when no additional effects could be added without introducing linear dependence, suggesting that the remaining candidate terms were redundant given the existing model structure. Least squared means for each treatment were separated using Tukey’s Range test with a significance level of 0.05.

3 Results

ANOVA tables for all main effects and interactions are provided in Tables 1–5. The GLM framework treats each cutting as an independent observational unit. Because stock-plant identity was not tracked, mixed-model approaches incorporating random stock-plant effects were not possible. We therefore used a fixed-effects model appropriate for factorial cutting-level inference and interpret results with explicit acknowledgment of this limitation. Proportion variables (survival and rooting) were analyzed without transformation because residual diagnostics indicated adequate model fit. Root count data were analyzed with the same GLM framework; root-area data met normality assumptions. We provide a discussion of limitations associated with cutting-level analysis in the Discussion section.

Table 1. ANOVA F-statistics and corresponding p-values for sources of variation on 6- and 12-week survival of Osmanthus fortunei, O. fragrans, and O. heterophyllus stem cuttings collected 24 June - 8 July 2016 in McMinnville, TN, USA.

Sample size: n = 324 total cuttings across all species, cultivars, wood types, overwintering treatments, and IBA concentrations.

^a6- and 12-week survival percentages are based on the total number of cuttings originally stuck; twelve-week rooting percentages represent the proportion of original cuttings producing ≥1 primary root.

Table 2. Percentage (mean ± SD) of lignified (L), semi-lignified (SL), and non-lignified (NL) Osmanthus stem cuttings surviving at six weeks and rooting at twelve weeks.

Sample size per cell: six cuttings per wood type × overwintering treatment × cultivar. Means followed by the same lowercase letter within columns are not significantly different according to Tukey’s test (α = 0.05). Percentages are calculated using the total number of cuttings originally stuck.

Table 3. Least-squared means (± SD) for cultivar, wood type, overwintering treatment, and IBA concentration effects on the percentage of Osmanthus stem cuttings producing ≥1 primary root at twelve weeks.

Sample size varies by cultivar based on the number of cuttings surviving at 12 weeks. Means sharing lowercase letters within each factor are not significantly different (Tukey’s test, α = 0.05). All percentages use the original number of cuttings stuck as the denominator.

Table 4. Mean number of primary roots (± SD) for lignified (L) and semi-lignified (SL) Osmanthus stem cuttings after twelve weeks, shown by species and IBA concentration. Sample size varies by cultivar based on the number of cuttings surviving at 12 weeks.

Means sharing lowercase letters within columns are not significantly different (Tukey’s test, α = 0.05). Primary roots are defined as roots emerging directly from the stem or callus area.

Table 5. Least-squared means (± SD) for root area (mm²) of Osmanthus stem cuttings, grouped by cultivar, wood type, overwintering treatment, and IBA concentration.

Sample size varies by cultivar based on the number of cuttings surviving at 12 weeks. Means followed by the same lowercase letter within factor columns are not significantly different (Tukey’s test, α = 0.05). Root area represents the total black pixel area (root surface) in standardized images.

3.1 Influences on rooting success

Species, cultivar, and wood type were significant sources of variation for survival at six and twelve weeks (Table 1). Overwintering location of stock plants had variable effects that depended on species, and IBA concentration had no effect on six- and twelve-week survival of cuttings. The best model for six-week survival (F = 5.7, p < 0.001, Adj-R² = 0.48) included significant main effects of species (F = 4.32, p = 0.017) and wood type (F = 7.22, p = 0.002), as well as the species × winter-protection interaction (F = 3.95, p = 0.024; Table 1). Cultivar (F = 2.22, p = 0.063) and the species × wood-type interaction (F = 1.84, p = 0.133) were not significant at α = 0.05. Survival rates after six weeks were 73%, 57%, and 37% for O. fortunei, O. fragrans, and O. heterophyllus respectively. Percentages represent the proportion of total cuttings originally stuck.

Across species, 6-week survival was highest for lignified cuttings (78.9%), intermediate for semi-lignified cuttings (57.9%), and lowest for non-lignified cuttings (30.3%; Table 2). While non-lignified cuttings were the weakest type, the pattern varied by species. For O. fortunei and O. fragrans, lignified and semi-lignified cuttings had moderately high survival, and non-lignified cuttings had moderate or fair survival. For O. heterophyllus, however, 90% of lignified cuttings were graded alive after six weeks, while 0% of non-lignified cuttings were graded as alive, indicating that six-week survival of O. heterophyllus cuttings were heavily influenced by wood type than the other two species in the study.

Across all species and treatment combinations, 52% of the cuttings originally stuck were alive at six weeks, and 22% of the original cuttings had at least one primary root at twelve weeks (Table 2). The best model for twelve-week rooting survival (F = 8.8, p < 0.001, Adj-R² = 0.63) included significant effects of species (F = 7.3, p = 0.001), cultivar (F = 6.77, p < 0.0001), and wood type (F = 6.02, p = 0.004). The species × wood-type interaction (F = 2.39, p = 0.06) was not significant at α = 0.05 and is therefore interpreted descriptively only (Table 1). Survival rates after twelve weeks were 26.3%, 32.3%, and 8.8% for O. fortunei, O. fragrans, and O. heterophyllus respectively. These percentages reflect the proportion of original cuttings that produced ≥1 primary root. Across all species, 34.2% of lignified stem cuttings had at least one root after 12 weeks as did 28.7% of semi-lignified stems, and 0% of non-lignified stems (Table 2). Rooting percentages for non-lignified cuttings were 0% across species, consistent with their lower six-week survival. Cultivar significantly affected twelve-week rooting (p < 0.0001). Within each species, cultivar-level variation was pronounced, with some cultivars exhibiting two- to six-fold differences relative to congeners (Table 3). For example, the popular O. fragrans cultivar ‘Fodingzhu’ rooted at a rate of 63% (the highest in this study), while the other cultivars averaged 11% and 21% respectively. The popular O. fortunei cultivar San Jose rooted at a much lower rate (10%) than the other cultivars tested here (Table 3). Winter-protection effects varied among species (Table 3). For O. fragrans, rooting percentages were higher for cuttings taken from winter-protected stock plants, whereas O. heterophyllus showed higher rooting from winter-exposed plants. Rooting percentages for O. fortunei were similar between overwintering treatments. KIBA concentration made a minimal impact on percentage of cuttings with roots at 12 weeks; no significant main effect of IBA concentration on twelve-week rooting percentage was detected (p = 0.323; Table 1).

3.2 Number of roots

The average number of primary roots was 2.4 across all treatments (Table 4). Cultivar, IBA concentration, and species were significant sources of variation for root number, whereas wood-type effects varied by species. Semi-lignified cuttings generally had more roots than lignified cuttings, except for O. heterophyllus. For lignified cuttings, the 4000-ppm IBA treatment produced significantly more roots (3.8 ± 4.3) than untreated controls (1.0 ± 2.3; Table 4).

The difference in number of roots among cultivars was the largest overall, regardless of rooting hormone or wood type. All species showed a two- or three-fold difference in root number between the best and worst rooting cultivar. For example, the popular O. fragrans cultivars Fodingzhu and Beni Kin Mokusei averaged 1.7 and 4.0 primary stems per cutting, respectively (Figure 2).

3.3 Root area

The average root area for Osmanthus cuttings over all species and treatments was 990.0 mm². Species was the only variable that significantly influenced root area (F = 3.7, p = 0.045); cultivar, wood type, winter exposure, and IBA concentration were not significant. Osmanthus fortunei had the largest mean root area (1234 ± 978 mm2), followed by O. fragrans (1190 ± 788 mm2), while O. heterophyllus cuttings produced significantly less root area (545.5 ± 399 mm2). The average root area across all treatments was 990 mm² (Table 5). Species was the only significant effect (F = 3.7, p = 0.045). Semi-lignified cuttings produced the largest root areas for O. fragrans and O. heterophyllus, whereas lignified cuttings of O. × fortunei produced larger root areas than semi-lignified cuttings (Figure 3).

4 Discussion

4.1 Species and wood type effects

The species effect was pronounced: O. heterophyllus exhibited the lowest six- and twelve-week survival and rooting compared to O. fragrans and O. fortunei (Table 2). Non-lignified cuttings failed to root across all species, and survival of O. heterophyllus showed a strong dependence on wood maturity. These patterns indicate that cutting lignification, as expressed at the time of collection, is closely associated with rooting outcomes in this study. Although anatomical and physiological traits (e.g., carbohydrate reserves, vascular continuity, hormonal balance) have been proposed as mechanisms influencing rooting in woody ornamentals, these factors were not measured here and therefore cannot be confirmed directly. Interpretations involving such traits are offered only as potential explanations consistent with previous literature (Hartmann et al., 2013; Sanchez et al., 2020). The strong wood-type dependence observed in O. heterophyllus may reflect species-specific anatomical or physiological constraints noted in other Oleaceae, but further study would be required to verify these mechanisms in Osmanthus. Even without mechanistic data, these results reinforce that, even within the same genus, wood maturity thresholds for successful propagation can differ by species, necessitating species-specific protocols. Comparable patterns occur in olive, where basal semi-hardwood cuttings exhibit higher rooting potential than apical or softwood material (de Oliveira da Silva et al., 2012; Walali and Loussert, 1990), reinforcing the role of tissue maturity across Oleaceae.

4.2 Cultivar-level variation

Cultivar identity was a significant determinant of twelve-week rooting, with two- to three-fold differences both in rooting percentage and root number among cultivars (Table 3). O. fragrans ‘Fodingzhu’ rooted substantially more than other accessions of the same species, whereas O. fortunei ‘San Jose’ rooted poorly. Such pronounced genotypic variation is consistent with reports in Osmanthus where rooting capacity varies widely among cultivars (Blazich and Acedo, 1989a, 1989b; Fabbri and Borselli, 1982; Lambert et al., 2011). In ornamentals, these differences have practical consequences: high-performing cultivars can be propagated reliably with standard protocols, whereas low-performing cultivars may require modified hormone regimes, altered timing, basal wounding, or alternative propagation methods (e.g., air-layering, stooling; Evans and Blazich, 1999; Dirr and Heuser, 2006). These results reinforce the practical need for cultivar-specific propagation protocols, a pattern consistent with other woody ornamentals and with reports of substantial cultivar-level variation in olive and related species (de Oliveira da Silva et al., 2012; Rafique et al., 2021).

4.3 Auxin effects

Auxin concentration had no significant effect on survival at six or twelve weeks but did increase the number of primary roots in lignified cuttings. For lignified material, the 4000-ppm treatment produced significantly more roots than untreated controls, indicating that auxin contributed to root system development in tissues already capable of rooting. In many woody species, exogenous IBA promotes root primordia formation and accelerates rooting, but its efficacy can be modulated by endogenous hormone balances, phenolic co-factors, and carbohydrate availability (Hartmann et al., 2013). The lack of survival benefits from IBA in non-lignified cuttings here is consistent with the notion that auxin application cannot overcome the intrinsic deficits of immature tissues such as limited carbohydrate reserves, whereas in mature tissues auxin may increase root number and uniformity, ultimately improving root quality even when overall rooting percentages are constrained by species/cultivar genetics. These data indicate that auxin treatments in Osmanthus should be targeted to lignified or semi-lignified cuttings, with emphasis on root system quality outcomes rather than survival gains. These findings align with previous propagation studies in Osmanthus and olive, which have documented increased root number under moderate IBA concentrations but inconsistent effects on overall rooting percentage (Nahlawi et al., 1976; Petry et al., 2012). Moreover, inhibition at very high concentrations in Osmanthus (Blazich and Acedo, 1989a, 1989b) parallels similar diminishing returns noted in olive.

4.4 Stock plant overwintering and physiology

Stock plant overwintering produced species-dependent effects: O. fragrans cuttings were favored by winter-protected conditions, whereas O. heterophyllus responded better after winter-exposed conditions, and O. fortunei showed minimal differences. Cold exposure and overwintering can alter carbohydrate allocation, bud dormancy status, hydraulic conductance, and endogenous hormone balances, thereby influencing cutting physiology at the time of collection (Hartmann et al., 2013). For some evergreen broadleaf shrubs, moderate chilling may enhance rooting competence by aligning phenology and carbohydrate reserves with the optimal cutting window; in others, cold stress may reduce tissue integrity or reserve status, diminishing early survival under mist (Dirr and Heuser, 2006; Evans and Blazich, 1999; Yeager, 2020). Although overwintering environment is known to influence stock-plant condition in many evergreen shrubs, the specific physiological bases for the species-specific responses observed here were not measured. Therefore, overwintering effects are interpreted strictly at the level of observed variation. These results highlight the importance of stock-plant management in Osmanthus propagation and indicate that optimal overwintering practice may differ among species.

4.5 Practical implications for propagation

The results herein support the following actionable guidance for Osmanthus propagation:

• Avoid non-lignified cuttings across species, as they performed poorly in survival and showed no rooting at twelve weeks.
• Prioritize lignified or semi-lignified cuttings when developing propagation schedules, particularly for O. heterophyllus, which exhibited strong dependence on wood maturity.
• Use auxin treatments (e.g., 4000 ppm IBA) to enhance root number and root system quality in lignified cuttings; do not expect auxin to rescue immature material.
• Manage stock plant overwintering by species: consider protected overwintering for O. fragrans and exposed overwintering for O. heterophyllus, while O. fortunei may be more flexible.
• Implement cultivar-specific protocols: high-performing cultivars (e.g., O. fragrans ‘Fodingzhu’) can be scheduled reliably; low-performing cultivars (e.g., O. fortunei ‘San Jose’) may require modified hormone dips, basal wounding, adjusted cutting window, or alternative propagation methods.

These recommendations apply to cutting-level responses in this factorial design and should not be interpreted as stock-plant-level effects due to the partial replication structure of the experiment. From an operational standpoint, these recommendations can be integrated into nursery scheduling by allocating early-season bench space and longer rooting windows to difficult-to-root cultivars/species, while easy-to-root selections can be scheduled later to optimize facilities and labor.

5 Future research directions

Given the pronounced species and cultivar effects, future applied propagation studies should map seasonal windows for each species/cultivar using multi-cohort trials (spring to early fall) to align physiological maturity with optimal rooting. Other avenues to increase success include comparing auxin regimes (IBA vs. KIBA vs. NAA), dip duration, and basal wounding to refine protocols for low-performing cultivars and manipulation of stock plant nutrition (e.g., controlled nitrogen-to-carbohydrate ratios) and light environment (shade vs. full sun) to enhance carbohydrate reserves prior to cutting. Because individual cuttings were not tracked to specific stock plants, future studies should consider incorporating stock-plant identity as a random factor to better partition biological variation and reduce potential pseudo-replication. Additionally, long term physiological and genetic studies could incorporate anatomical (histology) and physiological assays (e.g., endogenous hormone profiling, carbohydrate quantification) to elucidate mechanisms underlying species/cultivar differences.

6 Conclusions

Overall, these findings indicate that successful propagation in Osmanthus depends strongly on species, cultivar, and tissue maturity at the time of collection. Auxin enhanced root system development in mature tissues but did not increase survival. Given the cutting-level experimental unit and partial stock-plant replication, results should be interpreted at the level of cutting responses. Operationally, avoiding non-lignified shoots, tailoring overwintering by species, and adopting cultivar-specific schedules should improve rooting outcomes and production efficiency. These results, together with future mechanistic and seasonal studies, provide a foundation for evidence-based propagation guidelines in Osmanthus.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Sudip Kumar Dutta, Sikkim Centre, India

Reviewed by: Mariana Georgina Corigliano, CONICET Institute of Biotechnological Research (IIB-INTECH), Argentina

Sushma Ningombam, Gandhi Institute of Engineering and Technology University (GIET), India