Genetic parameters of essential-oil traits for Eucalyptus bosistoana

ABSTRACT A Eucalyptus bosistoana breeding trial established in New Zealand to select plants with improved growth and wood properties was assessed for essential-oil traits. Mature leaves of 8-year-old E. bosistoana were collected from 1901 trees representing 85 families. Twenty compounds were quantified in these samples. Heritability estimates (h2) of the quantified essential-oil compounds ranged from 0.06 to 1.14, with the most abundant compounds 1,8-cineole, aromadendrene and the unidentified compound 8 showing the highest h2 of 0.78, 1.14 and 0.59, respectively. Total oil content of the leaves had moderate (0.25) heritability. The estimated negative correlation between total oil content and 1,8-cineole concentration at the phenotypic and genetic levels (rp = –0.44 and rg = –0.70, respectively) implies that families with higher-quality oil had less oil in the leaves. 1,8-cineole was genetically negatively correlated with myrcene (rg = –0.74), α-pinene (rg = –0.71), linalool (rg = –0.90), aromadendrene (rg = –0.94), trans-pinocarveol (rg = –0.75) and the unknown compounds 3 (rg = –0.91), 6 (rg = –0.83), 8 (rg = –0.88) and 9 (rg = –0.75). Seven of the 85 families had breeding values consistent with the standard commercial oil-quality requirement of over 70% 1,8-cineole. The results indicate that a breeding program could aid essential-oil production from E. bosistoana.


Introduction
Eucalyptus bosistoana F.Muell.features ground-durable heartwood, fast growth and straight stems, combined with drought-, frost-and pest-tolerance.The tall-growing species occurs naturally in coastal areas of eastern Victoria and southern New South Wales, Australia, where it grows on betterquality soils near rivers (Boland et al. 2009).The New Zealand Dryland Forests Initiative (NZDFI) has prioritised this species for developing a durable hardwood industry in Aotearoa New Zealand (van Ballekom and Millen 2017).Improved plants from a breeding program are now planted commercially (Millen et al. 2020).
Eucalyptus bosistoana foliage was reported to yield similar quantities and qualities of essential oil as E. globulus Labill., the major source of essential eucalyptus oil (Boland et al. 1991;Coppen 2002).Therefore, it is of interest to explore the potential of essential-oil production as a byproduct of timber plantations to improve their economic viability.
Eucalypt essential-oil yield and composition are inherited in part from parent plants.Most research on genetic parameters in eucalypt leaf essential oils over the last two decades has centred on the mallee E. polybractea F.Muell.ex R.T.Baker, a crop grown solely for essential-oil production in Australia (Kainer et al. 2017;Mazanec et al. 2017Mazanec et al. , 2020Mazanec et al. , 2021)).Trials indicated that yields in E. polybractea could be improved by more than 50% compared with wild genotypes through a breeding program (Mazanec et al. 2021) and that oil composition was under strong genetic control (Goodger and Woodrow 2008).Oil quality is mainly dependent on the 1,8-cineole content (ISO770 2002;British Pharmacopoeia 2020;Doran et al. 2021), and micropropagations of selected E. polybractea genotypes were shown to yield an oil with an 80% increase in 1,8-cineole content (Goodger et al. 2008).
In Myrtaceae, phenotypic variations of terpene biosynthesis in leaves are strongly linked to gene expression (Webb et al. 2014;Külheim et al. 2015).Hence, the genotypic correlations between individual terpenes provide information about monoterpene biosynthesis (Wilkinson et al. 1971;Keszei et al. 2010;Webb et al. 2014).Although a single gene is responsible for the occurrence of a component, the overall quantity of phytochemicals is linked to the activity of many genes (White and Nilsson 1984).
The development of trees with superior oil traits through selection and breeding is underway for E. polybractea under separate commercial programs in New South Wales and Victoria (Goodger and Woodrow 2008;Bush et al. 2022) and has commenced for other species (e.g.see Mazanec et al. 2017Mazanec et al. , 2020)).
The aim of this study was to quantify the variation of essential-oil traits between families in an E. bosistoana breeding population and determine genetic parameters.These can then be used by industry to select planting stock with superior essential oil.

Site and sample collection
Eucalyptus bosistoana leaves were collected between 8 and 21 March 2020 from a breeding trial located on a valley slope in Marlborough, New Zealand, at 41°65′35.5″Slatitude and 173°66′42.0″Elongitude.NZDFI's E. bosistoana breeding population consists of 186 families (a group of individuals with one or two parents in common, derived from seeds collected of a single mother tree) collected between 2008 and 2011 throughout the known natural distribution.Seeds were collected from trees west of Melbourne, from southern Gippsland in Victoria through to coastal southeastern New South Wales, and from a smaller form occurring more inland between Goulburn and Bungonia Gorge on the New South Wales Southern Tablelands.Breeding trials were established in triplicate throughout the warmer dry regions of New Zealand in 2009, 2010 and 2012, using the seeds available at the time.The trial assessed in this study had been planted in 2012 and included 87 E. bosistoana families and up to 70 individuals each, originating from mother trees located in Victoria, east of Melbourne.The trial contained 140 incomplete blocks, with each block consisting of 36 randomly selected trees planted at 2.4 m × 1.8 m spacing.At the date of sampling, 12−70 trees per family had survived.
Mature leaf samples of 8-year-old E. bosistoana were collected from 1901 trees representing 85 families.A total of 25 trees were randomly selected in each family and, within each family, five leaf extracts were prepared by pooling the leaves (five leaves from each tree) of five random trees.For 15 families, only 10 or 15 trees were selected to prepare five samples by pooling leaves from two or three trees, respectively, because fewer than 25 individuals and in some cases fewer than 15 individuals were available.

Essential-oil extraction
Essential oils were extracted using solvent extraction.Leaves of each pooled sample were cut into small (approx.25 mm 2 ) pieces and 4.00 g of leaves was placed in a 25 ml sealable plastic vial on the day of collection.8.00 ml of absolute ethanol (≥99%) was added to each vial and kept in a cooler box (4°C) for at least 1 week until transport to the laboratory.Each extract was then filtered through a 0.45 µm PTFE syringe filter.

Analysis
Oil extracts were analysed by gas chromatography (GC, Agilent, model 7820A) equipped with an Agilent DB-wax polyethylene glycol column (30 m × 0.250 mm × 0.25 μm).GC settings were: injector temperature 250°C; initial oven temperature 35°C for 3 min, increasing to 70°C at 3°C min −1 , increasing to 110°C at 5°C min −1 , increasing to 240°C at 50°C min −1 and finally held for 3 min; and detector temperature 300°C.Each identified compound was quantified using response factors obtained with the pure reference compounds.The mean response factor of all identified compounds was used for the unidentified compounds.Total oil yield was defined as the sum of all 20 quantified compounds per green leaf mass.

Data analysis
All data were analysed using R statistical software (R Core Team 2021).The (co)variance components needed for the genetic parameters were obtained from univariate and bivariate analyses.Univariate analyses were conducted using a linear mixed model, as below: where y ij represents the phenotypic observation of a single variable from the j th composite sample of the i th family, μ the fixed effect of the intercept, f i , the random effects of the i th family, and e ij the random residual of the i th family's j th composite sample.The expected value for the phenotype is μ, while the variances (V) for each of the random terms are We used a weighted analysis because the samples for chemical composition combined leaves from two to five trees (weight option in ASReml-R).
Coefficients of variation (CV) for the composite samples were adjusted according to Rohde (1976).
Estimating the covariance between the family effects required bivariate analyses, 'stacking up' two traits in a vector y and expressing the model in matrix notation: where, b represents the vector with the fixed intercepts for each trait, f the vector of random family effects for each trait and e the random vector of residuals for each trait.X and Z are incidence matrices joining the observations to their respective traits and effects.The expected value and variances changed to � represents the Kronecker matrix product, while F 0 and R 0 are as below.
Variance components and breeding values were estimated using the fitted model in ASReml-R 4.0 (Butler et al. 2017).Narrow-sense heritability (h 2 ) for each trait was estimated from the phenotypic and additive genetic variability using: where, σ 2 f is the variance between families and σ 2 e is the variance of residuals.The heritability calculation assumed that families were true half-siblings, with the relationship coefficient among trees in a family being ¼.
Genetic correlation estimates the association between the additive genetic values of two variables (Falconer 1989).It was calculated as the family covariance between both traits, divided by the product of the family standard deviations for each of the traits, as shown in the equation below.
where r g 1;2 ð Þ is the additive genetic correlation between traits 1 and 2, σ f 1;2 represents the family covariance between 1 and  2, and σ 2  f 1 and σ 2 f 2 represent the family variance of traits 1 and 2, respectively.
Of all quantified compounds, 1,8-cineole was the least variable, with a coefficient of variation (CV) of 25%.However, the high variability for minor compounds was at least partially caused by a larger relative experimental error arising because concentrations were below the detection threshold in some samples.Total oil yields varied ten-fold between families, from 3.9 µl g −1 to 41.2 µl g −1 .

Genetic control of essential-oil traits in Eucalyptus bosistoana
The total oil content of the leaves was both heritable (h 2 = 0.25) and variable (CV = 65%) (Table 1), indicating that it could be altered though a breeding program.However, oil yields of plantations are also controlled by factors affecting biomass (Coppen 2002;Mazanec et al. 2021;Spencer et al. 2021), and were reported to be influenced by genes responsible for secretory cavity development, tree growth, precursor availability and leaf development (Thavamanikumar et al. 2013).Eucalyptus essential-oil quality is related to the relative amounts of individual compounds, particularly the concentration of 1,8-cineole.Breeding values of seven E. bosistoana families met the minimum 70% 1,8-cineole concentration required by the British Pharmacopoeia (2020) standard for eucalyptus oil, with two even exceeding 80%.Limonene and αpinene concentrations also fell within the specified ranges.
Heritability estimates for the concentration of the quantified compounds in E. bosistoana essential oil ranged from 0.06 to 1.14 (Table 1).The most abundant compounds 1,8-cineole, aromadendrene and unidentified compound 8 showed the highest genetic control, at h 2 = 0.78, 1.14 and 0.59, respectively.The exception was limonene, the thirdmost abundant compound in the E. bosistoana leaf extracts, which showed a very low heritability of 0.07 (0.01, 0.15) in this dataset, suggesting it is more subject to non-additive rather than additive genetic control.However, it was reported that the E. urophylla S.T.Blake × E. grandis W. Hill ex Maiden hybrid clone with the highest limonene content was most resistant against myrtle rust (Austropuccinia psidii) (Silva et al. 2020).
The detection limit of GC and the associated experimental error likely reduced heritability estimates for minor compounds.No or low heritability values were found for βmyrcene, α-pinene, caryophyllene and the unidentified compounds 4, 5 and 7.However, genetic control was still observed for the minor compounds, trans-pinocarveol and linalool, and the unidentified compounds 1, 2, 3, 6, 9 and 10, suggesting moderate to strong genetic control.
The true relatedness of the trees in the sampled E. bosistoana breeding population was unknown.Each family -that is, derived from seeds from a single mother tree -was considered to comprise true half-siblings from unrelated E. bosistoana parents.In reality, however, unrelated mating, relative inbreeding (mating with relatives) and selfing occur in such open-pollinated families and hence families are not true half-siblings (Burgess et al. 1996).Therefore, in the absence of molecular genetic information revealing the true relatedness within the breeding population (Bush et al. 2011), heritability cannot be exactly estimated, explaining the implausibly high heritability of h 2 = 1.14 for aromadendrene in this study (Table 1).It also implies that heritabilities cannot be compared in absolute terms with other populations.Still, essential-oil composition is typically recorded to be under strong genetic control (Doran 2002) (Barton et al. 1991).

Genetic and phenotypic correlations between essential-oil traits
Significant genetic and phenotypic correlations were obtained between phytochemicals extracted from E. bosistoana foliage (Table 2).Oil quality, determined by 1,8-cineole concentration, was negatively correlated with most other quantified compounds in the E. bosistoana oils (Table 2).Only the unidentified compound 2 was positively correlated with 1,8-cineole concentration, suggesting that it is a product of the same biosynthetic pathway.Little or negative genetic and phenotypic correlations were reported between 1,8-cineole, α-pinene, βpinene, limonene and p-cymene in E. camaldulensis (Doran and Matheson 1994).
The negative correlation between total oil content and 1,8-cineole concentration at the phenotypic and genetic levels (r p = −0.44 and r g = −0.70,respectively) implies that families with higher-quality oil had less oil in the leaves.This was in agreement with the observation for E. polybractea that the selection of trees with higher 1,8-cineole concentrations led to reduced oil yields (Mazanec et al. 2021).Although more leaf matter needs to be extracted for a given volume of oil, this is not proof that these families will produce less oil in plantations because it has been shown for other species that oil production is dependent on foliage mass rather than the oil content of the leaves (Doran and Matheson 1994;Coppen 2002).Significant positive phenotypic (r p = 0.81) and genetic (r g = 0.80) correlations were reported between total monoterpenes and 1,8-cineole yields in 19 families of E. camaldulensis (Doran and Matheson 1994).

Superior families
In the global market, 1,8-cineole percentages range between 45% and 52% for E. globulus oil from China (IFEAT 2017) and 80 −95% for E. polybractea oil from Australia (Goodger and Woodrow 2008;Kainer et al. 2017).A potential breeding target for oil quality could be 70% 1,8-cineole content, as specified in British Pharmacopoeia (2020), which was met by seven of the 85 E. bosistoana families, assuming a half-sibling relationship.All breeding values can be found in the Appendix.Family 876 yielded the purest oil, with a 1,8-cineole content of 85.3%, 28% higher than the population mean.Breeding values of 1,8-cineole percentages were plotted in Figure 1 against those for total oil contents.Ten families with above-average breeding values for both traits are in the top-right quadrant, but only one family (877) had above-average oil yield and a 1,8-cineole content above 70%.It should be kept in mind that oil yield in commercial plantations is also controlled by biomass (Coppen 2002;Mazanec et al. 2021;Spencer et al. 2021) and that breeding values would change if the individuals of a family deviated from the assumed half-sibling relationship.

Conclusion
Eucalyptus bosistoana leaf-oil traits are under genetic control.The strongest genetic control (h 2 ) of 0.78, 1.14 and 0.59 were found for 1,8-cineole, aromadendrene and unidentified 8, respectively.Total oil yield was moderately heritable (h 2 = 0.25).Significant genetic and phenotypic correlations were found between individual essential-oil compounds; most notably, 1,8-cineole concentration was negatively correlated with most other quantified compounds (myrcene, α-pinene, linalool, aromadendrene, trans-pinocarveol, terpineole and unidentified compounds 3, 6, 8 and 9) in E. bosistoana oil.Of the 85 tested E. bosistoana families, seven families possessed breeding values indicating that their oil would meet the British Pharmacopoeia standard (2020) specification of a minimum of 70% 1,8-cineole.The negative correlation between total oil content and 1,8-cineole concentration at the phenotypic and genetic levels (r p = −0.44 and r g = −0.70,respectively) indicates that families with higher-quality oil have less oil in their leaves.

Table 1 .
Minimum and maximum values, mean, coefficient of variation (CV) and estimated heritability (h 2 with 95% confidence intervals within brackets) of total oil and oil compounds for Eucalyptus bosistoana families at 8 years old.

Table 2 .
Estimated genetic (above diagonal) and phenotypic (below diagonal) correlations between total oil content and individually quantified compounds in Eucalyptus bosistoana essential oil at 8 years old.Correlations with 95% confidence intervals (in brackets) excluding 0 in bold and negative correlations in red.NA = not available; Un = unidentified.