Variability of volatiles in Pinus cembra L. within and between trees from a stand in the Salzburg Alps (Austria) as assessed by essential oil and SPME analysis

Pinus cembra L. is a coniferous European mountain range tree rich in oleoresins and essential oils. Twig tips with needles are the most common tree parts used to obtain essential oils. As the whole tree contains volatiles, the essential oil composition from different parts was studied, including twig tips with needles from the bottom, the top and the cone bearing branches, cones, twigs without needles from the lower and upper part of the crown as well as wood and bark from the trunk. The variability in essential oil composition between these plant parts and between individual trees has been studied using multivariate statistical analyses. α-Pinene was present in all samples, being highest in cones (49.3%) and lowest in sapwood (0.7%). Twig tips with needles from different parts of the crown had similar essential oil composition with α-pinene (43.9–48.3%), β-phellandrene (13.1–17.2%), β-pinene (6.6–9.3%), germacrene D (5.1–6.8%) and limonene (4.1–6.1%) as main compounds. Twig essential oils had usually more β-phellandrene (23.9–29.8%) than α-pinene (23.3–24.3%) and also appreciable amounts of β-pinene (13.5–15.1%) and limonene (11.9–17.9%). Cone essential oils contained mainly α-pinene, β-pinene (20.1%) and limonene (13.9%). The essential oils from wood and bark were rich in diterpenes as cembratrienol (4.9–21.4%), cembrene (4.8–14.3%) and methyl daniellate (2.7–6.8%). Sapwood distillates contained also notable amounts of alkane derivatives. Finally, the solvent free SPME technique has been employed to analyse the volatiles in the plant parts. For needles and cones the observed patterns were in good accordance with the compositions of the respective essential oils.


Introduction
Pinus cembra L. (Pinaceae), also known as swiss stone pine, cembran pine, ceder pine, or arrola pine (Apetrei et al. 2011), naturally occurs in two disjunct regions: in the European Alpine chain (Austria, Switzerland, Italy, France) from 1500 to 2500 m a.s.l., and the Carpathian Mountains (Poland, Slovakia, Ukraine, Romania) from 1300 to 1700 m a.s.l. (Wieser et al. 2006;Wojnicka-Półtorak et al. 2015). Isolated trees can be found up to 2850 m a.s.l. in the Alps and up to 1985 m a.s.l. in the Carpathian Mountains (Wojnicka-Półtorak et al. 2015). Pinus cembra occurs in pure or mixed stands with Picea abies L. Karst, Larix decidua Miller, Pinus sylvestris L. and Pinus mugo Turra, and often reaches the tree-line. Natural competition with associated tree species as well as human activities, like shepherding, alpine farming and timber extraction strongly affects the distribution of P. cembra (Boden et al. 2009;Zong et al. 2010) and has led to displacement to marginal habitats (Höhn et al. 2009).
P. cembra is a five-needled evergreen conifer tree (Boden et al. 2009), growing up to 25 m in height and 1.7 m in trunk diameter (Schütt 2014). Synzoochory dispersion of the wingless seeds is mostly done by the European nutcracker (Nucifraga caryocatactes L). The bird caches seeds for later consumption in winter and, if failing to recover, the seeds can germinate. P. cembra trees are well adapted to subalpine climate conditions (Boden et al. 2009;Caudullo and de Rigo 2016;Zong et al. 2010) and are highly resistant to abiotic stress factors at high altitudes (Apetrei et al. 2013). They are frost-hardy and can persist winter temperatures down to -43°C (Caudullo and de Rigo 2016;Schütt 2014). P. cembra is supposed to be reaching the age of 1000 years (Schütt 2014;Schweingriber and Wirth 2009).
Regionally P. cembra gained economic significance. It is listed as economically important plant (Smith 2017), in alpine northern Italy seeds are sold as food (Ducci et al. 2014). In the Austria federal land Tyrol, P. cembra accounts for about 2.3% of the usable trees in forestry (Landwirtschaftskammer Tirol 2014) and the price for its wood showed considerable fluctuations in the last years (Land 2018). So, in alpine regions, P. cembra wood is used for construction and furniture industry while the essential oil has various applications in medicine and aromatherapy, in local liquor production and as a fragrance in cosmetic (Thaler-Rizzoli and Gasteiger 2017). Furthermore, the essential oil is said to have positive effects on indoor climate although scientific evidence is lacking. Recently, efforts have been undertaken to support rural regions in promoting uses of P. cembra. The interest in non-timber products of this tree is mainly based on aromatic properties of its parts (Weiss et al. 2017).
Analysis of essential oil composition has been focused only on a few plant parts. Yet, other parts could also contain substantial amounts of essential oils and natural aromes. Therefore, we focussed our study (1) to investigate the variability in essential oil composition from different parts of P. cembra, (2) to obtain information to which extent trees from a stand differ in essential oil composition, and (3) to investigate the pattern of volatiles obtained with the solvent free SPME technique as it is close to the fraction perceived by the nose and can be considered as responsible for the fragrance. A comparison to distilled oils is attempted.
The findings may support further exploitation of P. cembra as a source of essential oils and natural aromes.

Plant material
Samples from seven individual P. cembra trees were collected in September 2015 in Austria (Salzburg, Thomatal, N 47°02 0 35.8 00 , E 13°45 0 26.6 00 ) at an altitude of 2003 m. The age of the individuals ranged from 120 to 175 years as estimated by counting the annual growth rings. Due to the very slow growth of P. cembra an uncertainity of ± 10 years was not avoidable. The distance between the sampled individuals ranged from 5 m (tree N o . 1 to tree N o . 2) to 700 m (tree N o . 1 to tree N o . 7). A voucher specimen was deposited in the Herbarium of the University of Vienna (WU Generale, http://herbarium.univie.ac.at/). From each individual tree the following parts were collected and denoted as follows: shoots with needles (NT) and twigs without needles (max. 2 cm in diameter) (TT) from treetop, shoots with needles (NB) and twigs without needles (TB) from the lower third of the tree, shoot with needles from cone bearing twigs (NC), one year-old cones from treetop (CO), bark from the trunk (BA), heartwood (HW) and sapwood (SW) from the lower part of the trunk. Heartand sapwood were cut in pieces of 2-5 cm. Only healthy appearing plant parts were taken. The plant parts were dried in the ambient air and kept at room temperature until analysis.

Essential oils extraction by hydrodistillation
The materials subjected to distillation were about 10 g for CO, NB, NC, NT and TT, 20 g for TB and 50 g for BA, HW and SW. These plant materials were further chopped and hydrodistillated with 400 mL of double distilled water in a Clevenger type apparatus for 3 h. The amount of essential oil obtained was read from the capillary of the distillation unit. The collected volatile fractions were stored at -18°C until further GC and GC-MS analyses. Prior to analysis, 5 ll of the essential oil samples were diluted with 495 ll hexane containing hexadecane (0.107 mg/mL) as internal standard.

GC/MS and GC/FID
The analyses were carried out on an Agilent Technologies 7890A gas chromatograph equipped with a 5975 C quadrupole mass selective detector, a flame ionization detector (FID) and a CTC-PAL autosampler (Agilent Technolgies, Santa Clara, CA, USA) as described in Chizzola et al. (2018) with slight modifications. The separation was done on a 30 m 9 0.25 mm fused silica column coated with 0.25 lm HP5-MS. The compounds eluting from the column were distributed with a Deans switch at equal proportions to the detector of the mass spectrometer (MSD) and FID. The temperature program of the oven was: isothermal at 50°C for 1 min, then increasing to 220°C at a rate of 5°C/min and increasing further to 280°C to a rate of 15°C/min. The split of the injector was set at 1:20. The injection volume was 1 ll. The total ion current (m/z 40 to 500) from the MSD was used to identify the compounds according their mass spectra and their retention indices (Adams 2007;McLafferty 1989). As the analytical column did not separate limonene and b-phellandrene well, these two compounds were deconvoluted using the AMDISsoftware within the NIST spectral library. The FID operated at 250°C with 400 mL/min air flow and 40 mL/min hydrogen flow. The peak areas of the FID signal were used to calculate the percental composition of the oils without any correction. SPME From three individual trees the woody fractions (BA, HW, SW, TT) as well as the cones (CO) and needles only from the top twigs (NT-N) were also anlysed by SPME. Small amounts of the samples (about 120 mg for HW and SW and about 30 mg for other parts) were finely chopped and put into a 10 mL vial together with a filter paper disc (6 mm in diameter) soaked with 10 ll of cyclododecanone as internal standard (0.4 mg/mL in methanol). The vial was tightly closed with a septum and further processed in the CTC-PAL autosampler with the mounting for SPME fibers. The SPME fiber (PDMS-DVB, polydimethylsiloxanedivinylbenzene, Supelco, Bellefonte, PA, USA) was exposed for 30 min at 50°C while stirring to the headspace of the sample. Afterwards the fiber was introduced into the injection port of the GC system and desorbed for 5 min at 250°C at a split ratio of 10:1. The conditions for GC/MS and FID were the same as described above. Each sample was analysed thrice with SPME.

Statistical analysis
The statistical analyses were done with the package SPSS for Windows, version 24.0 (IBM Corporation, Armonk, NY, USA). A hierarchical cluster analysis (HCA) using the squared Euclidian distance with linkage between groups was carried out to group the plant parts according to the composition of their essential oils. Furthermore, the complex interplay of major essential oil compounds was studied by principal component analysis (PCA) using the samples as cases and the essential oil compounds as listed in Table 1 as variables. The data were Z-transformated by the software before calculating the loadings of the variables and scores of the samples. Canonical discriminant analysis (CDA) tested whether the sampled could be correctly assigned to the originating tree or plant part.

Essential oil yield
The hydrodistillation of the various plant parts gave different oil yields as presented in Table 1. Essential oil yield was highest in the unripe cones (CO), reaching 3.0% v/w. Twig tips with needles, which are commonly used for commercial essential oil production, were collected from the upper (NT) and lower part (NB) of the trees as well as from cone bearing twigs (NC). In those parts oil yield ranged between 2.3 and 2.5% v/w. Twigs from the tree top (TT, 2.2% v/w) were richer in essential oils than those from the bottom (TB, 1.1% v/w) while bark (BA) and wood (HW, SW) had very low oil content. Varying essential oil contents were reported in this species: High oil content in unripe cones (2.4% v/w) was also reported from trees in an arboretum in Rogow, Poland. These plants had about 0.46-0.66% v/w oil in needles and twigs and 0.94% v/w in the bark while the wood was also low in volatiles (Lis et al. 2017). The higher oil content in the bark in comparison to the present results may arise from the circumstance that the authors analysed bark from branches. High oil content was also reported in needles (1.96% v/w) and twigs (2.45% v/w) from trees in Romania (Apetrei et al. 2013), while needles from trees grown near Trento in Italy gave 0.61% v/w essential oil (Domrachev et al. 2012).

Essential oil composition
To compare the essential oil compositions from the different plant parts, the similarities among the oils have been first studied with a hierarchical cluster analysis (Fig. 1). The highest similarities were found between the three twig tips with needles (NT, NB and NC). These samples contained a-pinene as major compound, averaging 43.9-48.3% of the total compounds quantified (Table 1). Available literature reported a-pinene contents in oil from needles or twigs with needles ranging from 21.1% (Ioannou et al. 2014) to 69.1% (Apetrei et al. 2013). In the present essential oils, b-phellandrene, b-pinene, limonene and germacrene D were next major compounds in twig tips with needles. While most previous reports mention up to 20% germacrene D in needle oils (Ioannou et al. 2014), this compound occurred only in trace in oils from Romania (Apetrei et al. 2013).
Together with the cones (CO) and the twigs from the top (TT) and the bottom (TB) of the trees, twig tips with needles were grouped in a cluster clearly distinct from the cluster formed by bark (BA) and wood (HW, SW) samples. In cones, a-pinene was also found to be the main oil compound (49.3%) followed by b-pinene (20.1%) and limonene (13.9%). In twig oils (TT, TB), the relative amount of the components decreased in the order b-phellandrene, a-pinene, limonene, b-pinene. Contrary to the cones, twig and needle oils contained more b-phellandrene than limonene. This result is in some contrast to the finding of Lis et al. (2017) who reported more limonene than b-phellandrene in various parts (needles, twigs, cones). In sum in needles, monoterpenes prevailed in the essential oils of needles, cones and twigs.
Wood (HW, SW) and bark (BA) from the trunk were low in essential oils. The bark oil was rich in monoterpenes again with a-pinene at the top, followed by b-phellandrene, limonene and b-pinene, similarly as in needles. There were also low amounts of d-3carene and a-terpineol. However, the bark essential oil contained also sesquiterpenes as a-cadinol and abisabolol, and diterpenes. Diterpenes, finally, formed the main part of the wood oils (HW, SW). The main diterpenes identified were cembrene (= thunbergene), cembratrienol (= thunbergol) and methyl daniellate (= methyl lambertianate). To compare, the essential oil from Pinus heldreichii heartwood contained 23.8% cembrene, 5.6% cembrene A, 28.7% limonene and 6.4% a-pinene (Graikou et al. 2012). In our study, the main diterpene in cones was abieta-6,8(14)-dien-18-al (2.7%). Likewise, cone oils from plants grown in a Polish arboretum contained 5.7-10.0% of this compound (Lis et al. 2017) that occurred also in oils from Pinus mugo (Venditti et al. 2013). In contrast to all other plant parts, sap wood oil contained various alkane derivatives participating for about 30% of the Table 1 Essential oil yield and composition (%) obtained from the different parts of Pinus cembra  (7), 121 (5), 165 (4) Means are presented in ordinary fonts while standard deviations (SD) are given in italics essential oil content. The composition of this fraction is presented in Table 2 with nonanoic acid, nonanal and 2E-octenal as the dominant compounds. Lis et al.
(2017) analysed oils from bark and wood from 2 to 3 years old branches and found that the wood oil contained about 16% diterpenes and the bark oil 1.4%, but no alkane derivatives were reported.
This differentiation between the parts of the trees has further been visualized calculating a principal component analysis (PCA). From 23 essential oil compounds as variables, the PCA calculated seven components having eigenvalues greater than one and representing together 86.1% of the total variance. The first axis accounted for 40.5% and the second for 12.1% of the variance. Figure 3 represents the scoring of the samples on the components. Wood and bark samples could well be separated from cones, twigs and twigs with needles by the first component where the latter scored with positive values (Fig. 2A). This differentiation is based mainly on monoterpenes with negative factor 1 loading occurring mainly in needles, twigs and cones while sesquiterpene and diterpene compounds showing positive factor 1 loadings were characteristic for the woody parts of the stem. In addition, bark and heartwood samples had the highest d-3-carene contents, so this compound had a similar positive component 1 loading as sesquiterpenes and diterpenes. Similarly, bark and some heartwood samples are represented in the lower right quadrant on the factor 3 and 4 plottings (Fig. 3a, b). Twig tips with needles (NC, NT, NB) had a negative component 2 score according to their high a-pinene content but also with their germacrene D and d-3-cadinene levels as these compounds had a negative component 2 loading. Twig (TT, TB) and cone (CO) oils scored positively on factor 2 according to their relative content of b-phellandrene, b-pinene and limonene. Especially for wood and bark samples, the spread of the individual samples from a plant part on the plane of the factor scores shows that there is some diversity between the seven analysed trees. This diversity is also the cause of some high standard deviations in respect to the means in Table 1. In cone oils, a-pinene varied from 26.4 to 71.7%, in bark oils from 6.8 to 50.5%. Three samples from the heart wood had less than 0.5% a-pinene while in the remainig four samples this compound ranged from 14.4 to 60.3% (data with the supplementary material). Sapwood showed a considerable variability in alkane derivatives (Table 1) and the main diterpene cembratrienol (9.4-29.1%).
We used canonical discriminant analysis (CDA) for a further characterization of the essential oil variability. In this multivariate analysis, objects have to be assigned to predefined groups. In the present analysis, we tested to relate the essential oil samples to the plant parts (Fig. 4a) or to the individual trees (Fig. 4b).
Sapwood essential oils had strong negative function 1 scores and were differentiated clearly from all other oils (Fig. 4a). Function 2 distinguished cone, twig and needle essential oils from heartwood and bark oils. The latter two were also distinguished. Concerning the differentiation of the individual trees, there were two outstanding trees, number 1 and 7, and some overlapping between other trees. Altogether, 54 of 62 samples could be correctly classified in CDA. Tree 1 had the highest limonene contents in all plant parts and was rather low in a-pinene. Tree 6 appeared to have higher b-phellandrene contents in most parts. Tree 7 is characterised by the highest myrcene contents (18.7%) in the cones. Trees 2 and 4 appeared to be most similar. High diversity of terpenes within a population might be seen in the context to increase resistance to herbivory (Keefover-Ring and Linhart 2010). Monoterpene variability occurred also within Pinus sylvestris where a-pinene and d-3-carene rich chemotypes could be distinguished within the same populations (Kännaste et al. 2013). Accordingly, within a population, there were trees emitting mainly a-pinene Fig. 1 Dendrogram showing the similarities between the plant parts of Pinus cembra according to their oil composition or d-3-carene or a mixture of both compounds (Bäck et al. 2012).
Twig tips with needles were most commonly used to obtain the P. cembra essential oil in practice. In the present study, they came from the lower part of the tree (NB) and from the treetop (NT and NC). Their essential oil profiles were similar (Table 1; Fig. 2) with 43.9-48.3% a-pinene, 13.1-17.2% b-phellandrene and 6.6-9.3% b-pinene as main compounds. From the view of plant morphology these samples consisted of two different plant organs, leaves and shoot tips. Secretory ducts are present in both parts but the amount and composition of the stored essential oil may be different. Therefore, the seven NT samples were further separated into needles (NT-N) and twig tips (NT-T) and analysed separately. The compositions of the resulting essential oils are presented in Table 2. Twig tips contained more essential oil (2.3%) than needles alone (1.0%). Needle essential oil contained considerably more a-pinene (62.9%) than Mean and standard deviation (SD) from 7 trees Means are presented in ordinary fonts while standard deviations (SD) are given in italics twig tip essential oil (22.3%). A high percentage of apinene in P. cembra needle essential oil has also been reported by Apetrei et al. (2013) (69.1%) and Kubeczka and Schultze (1987) (64.7%). Currently, also the sesquiterpenes, in particular germacrene D, were at a higher level in the needles as compared to the twig tips. To counterbalance this, twig tip essential oil had higher b-phellandrene, limonene and b-pinene contents than the essential oils from the needles. Finally, more diterpenes were found in twig tips than in needles. In sum, the essential oil obtained from the twig tips had a very similar composition to that obtained from the twigs up to 2 cm in diameter without needles (TT and TB). These results are in good accordance with data obtained from parts of P. cembra grown in an arboretum (Lis et al. 2017).

Volatiles from HS-SPME
Head space solid phase microextraction (HS-SPME) has been developed as fast solvent free extraction technique to isolate volatile compounds from a range of different plant matrices where the volatiles can be directly transferred into a GC-system. The headspace-SPME method has been applied to compare the volatiles from six Pinus species growing in Canada and the major compounds were suitable to be employed as chemotaxonomical markers (Kilic and Kocak 2014). Similarly, relative abundance of apinene, d-3-carene and bornyl acetate in the headspace-SPME fraction from needles has been used to differentiate between the closely related species Pinus mugo, P. uncinata and P. uliginosa (Celinski et al. 2015).
In the present research, as shown in Table 3, apinene was the major volatile compound in all SPME fractions. High levels of this compound were also found in heartwood and sapwood, contrasting the distilled fractions where a-pinene was very low in some samples. Also Pinus sylvestris showed high apinene levels in SPME analyses of sapwood and heartwood (Wajs et al. 2007). Actually, in contrast to hydrodistillates, SPME fractions contained very few diterpenes. Therefore, SPME and distillates from wood (HW, SW) were clearly different. Alkane derivatives occurred also in sapwood SPME fractions with hexanoic acid and nonanal as major compounds. Furthermore, in contrast to distillates, SPME from twigs (TT) showed also an appreciable level (7.5%) of hexanoic acid (Table 3).
Currently, bark (BA) and wood (HW, SW) SPME fractions contained appreciable d-3-carene amounts that were higher than in the respective essential oils. The high germacrene D levels in needles as compared to other tree parts could also be obtained in SPME. A comparison between essential oil composition and SPME is shown in the supplementary material (Figure S5) for the different parts of three trees. At least for needles (NT), cones (CO) and twigs (TT) the individual essential oil pattern of a tree could be observed as well in the corresponding SPME. So, in sum, SPME as an user-friendly, solvent-free and fast technique gave comparable results for the main volatile compounds in the case of cone and needle samples as compared to hydrodistillation. To a limited extent, this was also true for twigs but not for wood and bark samples.

Conclusions
The essential oils of Pinus cembra showed remarkable differences in their compositions between the different plant parts. Twig tips with needles and cones are rich in a-pinene, while twigs had also appreciable share of b-phellandrene, b-pinene and limonene in their essential oils. Twig tips with needles that are mostly used to obtain essential oil gave comparable essential oil compositions when harvested in different parts of the  Needles from the top twigs (NT-N), one year old cones from treetop (CO), twigs without needles from the treetop (TT), bark (BA), heartwood (HW) and sapwood (SW) Means are presented in ordinary fonts while standard deviations (SD) are given in italics crown. Material from the trunk is low in essential oil and contained mainly diterpenes, some monoterpenes and alkane derivatives. A future evaluation of the biological activities of the essential oils must consider these differences in composition of the volatile fractions.

Supplementary materials
The following are available online: Table S1: Essential oil yield and composition (%) obtained from the different parts of Pinus cembra; Figure S1: Mean essential oil composition in the different plant parts of Pinus cembra; Figure S2: Essential oil composition of needles with twigs in the crown from the top (NT), the bottom (NB) and cone bearing twigs (NC) of seven individual trees; Figure S3: Essential oil composition of cones (CO), twigs from the top (TT) and the bottom (TB) of the crown of seven individual trees; Figure S4: Essential oil composition of bark (BA), heartwood (HW) and sapwood (SW) of seven individual trees; Figure S5: Comparison of volatile compounds composition in distillates and SPME-fractions obtained from different parts of Pinus cembra.