Comprehensive analysis of chemical and enantiomeric stability of terpenes in Cannabis sativa L. flowers

Abstract

Objective

Cannabis sativa L. is renowned for its medicinal and recreational uses. With the increasing global legalization of C. sativa L.‐based products for medicinal purposes, there is a growing need for well‐characterized products. While the stability of cannabinoids such as tetrahydrocannabinol and cannabidiol is well understood, information on the chemical and enantiomeric stability of terpenes remains scarce. This is despite the fact that terpenes are also thought to have pharmacological activity and may contribute to the overall effect of C. sativa L.

Methods

To address these challenges, four analytical methods based on chiral, polar, and apolar gas chromatographic separation combined with either MS or FID detection were developed and validated. These methods successfully separated and quantified a total of 29 terpenes, including 13 enantiomers and 5 diastereomers specific to C. sativa L. Furthermore, terpenes and authentic C. sativa L. flowers and extracts were subjected to UV and heat treatments to observe potential degradation reactions over time.

Results

Each terpene generates a unique pattern of degradation products resulting in a diverse array of oxidation and cyclization products. P‐cymene was identified as a major product of terpene aging. Notably, no enantiomeric conversion was detected, suggesting that the formation of (−)‐α‐pinene in cannabis extracts, for example, originates from other terpenes.

Conclusion

Terpenes have different degradation rates, even though they are structurally similar. In addition, cultivar‐ and growth‐condition‐specific enantiomeric ratios were observed in C. sativa L., confirming that enantiomer production is species‐specific and has to be considered for therapeutical applications.

Short abstract

The present work focuses on the impact of environmental stresses, such as UV light and temperature, on terpene patterns in cannabis flowers and extracts thereof. It is obvious from the GC‐FID and GC‐MS data, that terpenes have different degradation rates despite being structurally similar. In addition, different enantiomeric ratios were observed in C. sativa L. depending on cultivar and growth conditions. We conclude that the enantiomeric terpene pattern is species‐specific and must be considered for therapeutic applications.

1. INTRODUCTION

Cannabis sativa L. is an herbaceous annual plant belonging to the Cannabaceae family along with the Humulus genus (hops). ¹ , ² , ³ The plant has been used for centuries for industrial, pharmaceutical, and recreational purposes. ³ It has further attracted interest for its medicinal properties in the treatment of a variety of conditions and is considered a multicomponent phytopharmaceutical for which more than 100 cannabinoids, 200 terpenes, and dozens of flavonoids have been reported. ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ Previous studies suggest an estimate of 1200 different plant cultivars. ¹⁵ As a result, cannabis is a highly variable natural product whose therapeutic effects depend on the composition of its secondary metabolites, particularly cannabinoids and terpenes. ¹⁶ Due to their psychoactive properties, cannabidiol (CBD) and tetrahydrocannabinol (THC) have been the main focus of chemotype classification and pharmacological research on cannabis. However, some experts argue that the effects of these two cannabinoids as isolated compounds cannot fully explain the therapeutic effects of the plant. ⁴ , ⁷ It has been postulated that the therapeutic effects of cannabis do not solely arise through the action of cannabinoids but rather are a result from a complex interplay of cannabinoids, terpenes, and flavonoids referred to as the “entourage effect.” ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ Until recently, terpene profiles in cannabis were of interest primarily for their characteristic flavor and fragrance properties, which were considered key criteria for cannabis cultivation. Although these sensory properties have been linked to pharmacological effects, there is a lack of scientific evidence to support these claims. ⁵ , ¹⁰ , ²⁰ , ²¹ Today, however, terpenes are increasingly recognized as chemical entities with their own unique pharmacological properties. ²² In general, plants produce terpenes as secondary metabolites to defend against biotic and abiotic stress, attract pollinators, protect against pests, and support development during their growth cycle. ¹⁷ , ¹⁸ , ²³ , ²⁴ While the concentrations and ratios of cannabinoids in cannabis are well described and reproducible, terpene profiles can be variable and unpredictable. Terpene composition depends not only on plant genetics but also on external factors such as the environment (e.g., exposure to light leading to photooxidation) and stage of development. For instance, the terpene ratio shifts from predominantly sesquiterpenes at the beginning of flowering to predominantly monoterpenes at the end of flowering. ²² , ²⁵ , ²⁶ Studying the terpene profile can not only provide valuable information on the developmental stage and pharmacological activity but can also be used for a more detailed chemotype subdivision and as a quality marker. ⁵ , ²⁷ , ²⁸ , ²⁹ The abundance of terpenes can vary according to cultivar and plant origin, but several compounds such as myrcene, β‐caryophyllene, limonene, α‐humulene, α‐ and β‐pinene, linalool, and γ‐terpinene are generally among the most common constituents and can occur in pharmacologically relevant concentrations. ⁵ , ¹⁰ , ³⁰ Studies have described a wide range of effects for terpenes, including cannabimimetic activity for α‐humulene, geraniol, linalool, and β‐pinene. In addition, α‐pinene, myrcene, and limonene have shown anti‐inflammatory, analgesic, and sedative properties in animal studies. ⁴ , ¹⁰ , ²¹ , ³¹ One terpene of interest in cannabis has been β‐caryophyllene due to its affinity for the CB2 receptor, making it a dietary cannabinoid. ³² Terpenes are the main components of essential oils, which are sensitive to atmospheric oxygen, temperature, and UV light. ²⁶ , ³³ While the stability of cannabinoids in cannabis preparations has been extensively studied, there is a lack of knowledge regarding the stability and rearrangement reactions of terpenes. ³⁴ , ³⁵ , ³⁶ Given that terpenes can also function as antioxidants for cannabinoids, comprehending the composition of terpenes and their fluctuations over time becomes crucial and can significantly impact the shelf life of medicinal cannabis products. ³⁷ This observation is further supported by a study conducted by Bueno et al. They found that the addition of terpenes can reduce the degradation of cannabinoids such as THC by nearly 50%. ³⁵ Hence, this understanding can also be used as an advantage, as the artificial addition of terpenes could extend the shelf life of medicinal cannabis products. Concerns have been raised that radiation and drying procedures for cannabis, as well as storage conditions, can have a significant impact on the terpene pattern. Terpenes are susceptible to degradation by processes such as isomerization, (photo)oxidation, dehydrogenation, polymerization, and thermal rearrangement. ¹⁶ , ²⁸ , ³³ , ³⁸ , ³⁹ Terpenes that undergo these types of transformations may become allergenic or rearrange into another terpene, resulting in a loss of quality and/or pharmacological activity. ³³ , ³⁷ For example, studies investigating the stability of essential oil preparations identified the monoterpene p‐cymene as a marker of aging. ³⁷ In the case of cannabis, a photolytic degradation of ‐myrcene to hashishene was observed. ⁴⁰ Furthermore, the enantiomeric ratio can be altered, which is of significance for, for example, authenticity control and aging, due to enantioselective terpene synthesis in plants. ²⁹ , ⁴¹ , ⁴² , ⁴³ Changes in the enantiomeric ratio of terpenes can affect their pharmacological activity. While different receptor affinities have been observed for the enantiomers of α‐pinene and limonene, there are few studies investigating the enantiomeric ratio and its biological effects in cannabis. ³¹ , ⁴⁴ , ⁴⁵ , ⁴⁶ To the best of our knowledge, three studies have investigated enantiomers in C. sativa L. so far. The studies had partly overlapping results. Mainly (+)‐α‐pinene, (S)‐limonene, (+)‐β‐pinene, (+)‐trans‐nerolidol, and (+)‐linalool were found in C. sativa L. However, the enantiomeric excess (EE) seems to be highly cultivar‐dependent. ¹ , ⁴⁷ , ⁴⁸

In here, our investigation focuses on the impact of environmental stresses, such as UV light and temperature, on terpene patterns in cannabis. These changes can be used for the biomarker discovery of aging processes. Furthermore, we propose that the enantiomeric ratio of terpenes in cannabis is underinvestigated but could significantly influence the proposed synergy and resolve conflicting results of the entourage effect. This comprehensive understanding may improve the way cannabis pharmacology studies are currently conducted and may encourage the use of standardized cannabis extracts. Overall, we aim to improve quality control (QC) and increase patient safety and drug efficacy.

2. EXPERIMENTAL

2.1. Materials

Nerol (97%) and geraniol (99%) were purchased from Acros Organics (Geel, Belgium); eugenol and nerolidol from Essencia (Winterthur, Switzerland); α‐terpinene (95%), limonene (98%), phenylethanol (>99%), and menthone (97%) from Fluka Chemie GmbH (Buchs, Switzerland); β‐caryophyllene and γ‐terpinene from Frey & Lau (Henstedt‐Ulzberg, Germany); citronellol (97%) and a C8‐C40 n‐alkane mix (calibration standard) from Merck (Darmstadt, Germany); sabinene from Carl Roth GmbH (Karlsruhe, Germany); and methyleugenol (>98%) from SAFC (St. Louis, MO, USA). Camphene (95%), p‐cymene (99%), rose oxide (≥98%, isomer mix), cis‐hexen‐1‐ol (>98%), linalool (97%), citronellyl acetate (>95%), geranyl acetate (analytical standard), farnesol (95%), carvacrol (>98%), isobornyl acetate (>90%), myrcene (95%), α‐terpinolene (>90%), (−)‐α‐pinene (>99%), (+)‐α‐pinene (>99%), (−)‐β‐pinene (99%), and (+)‐β‐pinene (analytical standard) were obtained from Sigma Aldrich (St. Louis, MO, USA). The rose oxide standard used for analysis consisted of a 1:1 ratio for both the (+/−)‐cis‐ and (+/−)‐trans enantiomers, with the cis‐enantiomer being the predominant one. Further, citral (>98%, isomer mix), neryl acetate (>95%), and α‐humulene (>93%) were purchased from TCI Chemical (Eschborn, Germany).

2.2. Chiral GC‐FID analysis

Chiral GC‐FID analysis was performed using two validated methods published by Raeber et al. and Allenspach et al. on a BGB 178 30% CD (25 m × 0.25 mm × 0.25 μm) and BGB 176 SE (30 m × 0.25 mm × 0.25 μm) capillary columns (BGB Analytik, Boeckten, Switzerland), respectively. ²⁹ , ⁴³ Only for the measurement of authentic cannabis extract, the split ratio was lowered to 1:10. The BGB 178 30% CD column was installed on a GC Trace 1600 equipped with an AI 3000 autosampler (Thermo Fisher Scientific, Waltham, MA, USA), and the temperature program and settings were followed according to Raeber et al. ⁴³ The BGB 176 SE column was installed in a GC Trace 1300 equipped with an AI 3000 autosampler (Thermo Fisher Scientific, Waltham, MA, USA), and the settings and temperature program were followed according to Allenspach et al. ²⁹ Helium 6.0 and hydrogen 5.0 (PanGas, Dagmersellen, Switzerland) were used as the carrier gas and for the FID detector, respectively. For the FID detector, the hydrogen flow was set at 35 mL/min and the air flow at 350 mL/min. Injection volume was set to 1 μL. Data acquisition was performed using Chromeleon (Thermo Fisher Scientific, Version 7.3.1) for the GC Trace 1600 and ChromCard (Thermo Fisher Scientific, Version 2.9) for the GC Trace 1300. The EE was calculated based on the following formula: ⁴³

$$\mathrm{EE}\left[\%\right]=\frac{{\text{peak area}}_{\text{major enantiomer}}-{\text{peak area}}_{\text{minor enantiomer}}}{{\text{peak area}}_{\text{major enantiomer}}+{\text{peak area}}_{\text{minor enantiomer}}}\times 100.$$

2.3. Polar GC‐FID analysis

Polar GC‐FID analysis was carried out using a validated method described by Raeber et al. ⁴³ A BGB‐wax (30 m × 0.25 mm × 0.25 μm) capillary column (BGB Analytik, Boeckten, Switzerland) was installed in a GC Trace 1600 equipped with an AI 3000 autosampler (Thermo Fisher Scientific, Waltham, MA, USA). Temperature program and settings were followed according to Raeber et al. ⁴³ Helium 6.0 and hydrogen 5.0 (PanGas, Dagmersellen, Switzerland) were used as the carrier gas and for the FID detector, respectively. For the FID detector, the hydrogen flow was set at 35 mL/min and the air flow at 350 mL/min. Injection volume was set to 1 μL. Data acquisition was performed using Chromeleon (Thermo Fisher Scientific, Version 7.3.1).

2.4. Apolar GC–MS analysis

Apolar GC–MS analysis was carried out using a DB‐5 MS (30 m × 0.25 mm × 0.25 μm) capillary column (Agilent Technologies, Santa Clara, USA) installed in a Trace GC‐Ultra equipped with a Triplus autosampler and coupled to a DSQ II MS (Thermo Fisher Scientific, Waltham, MA, USA). The MS analysis was performed in full scan positive mode from 40 to 300 Da at an energy of 70 eV and an ion source temperature of 250°C. The MS transfer line was held at 250°C. The GC temperature program started at 60°C and was held for 3 min, then increased by 5°C/min to 220°C and held for another 5 min. The inlet temperature was set to 250°C with a split flow of 20 mL/min and a split ratio of 20. Helium 6.0 (PanGas, Dagmersellen, Switzerland) was used as the carrier gas at a constant flow of 1.0 mL/min. The injection volume was 1 μL. Data acquisition was performed using XCalibur (Thermo Fisher Scientific, Version 2.2 SP1.48). GC–MS analysis also entailed the usage of an MS spectra library consisting of the NIST MS spectra library, an in‐house databank, and the Adams library. ⁴⁹

2.5. Method validation and data analysis

Method validation was performed based on the ICH Q2(R2) guidelines for the validation of analytical procedures. ⁵⁰ Calibration standards and QC samples were prepared by directly diluting a terpene stock solution (0.1% m/V) to a working range of 1–1000 μg/mL using ethanol (EtOH). In each case, six calibration standards and three QC samples in the high, medium, and low concentration ranges were used for the validation. All samples contained a final concentration of 10 mM cis‐3‐hexen‐1‐ol as an internal standard. Individual working ranges for the corresponding GC method can be viewed in Tables S1–S4. Method validation was performed on five individual days over 3 months. QC samples were each analyzed in duplicate, and their concentration was back‐calculated using the fitted regression model acquired from the daily calibration. Accuracy was determined as a bias by calculating the relative difference in percentage between the mean experimentally determined and original QC concentrations. Imprecisions for intraday and interday performance were calculated according to Peters et al. as the relative standard deviation. ⁵¹ The limits of detection (LoD) and quantification (LoQ) were determined using the slope and the standard deviation of a linear response. ⁵⁰ In order to determine LoD and LoQ of a nonlinear equation (e.g., quadratic equation), a linear regression was fitted to the lower four calibrators, situated within the model's nonquadratic increasing region. Linear retention indices, used as an additional parameter for analyte identification, were calculated using the van Den Dool and Kratz equation after analyzing a homologues series n‐alkanes. ⁵² As a measure of the separation quality, the resolution (R_s) between peaks was calculated following the guidelines of the European Pharmacopeia 11.5. ⁵³

2.6. Preparation of authentic cannabis extracts

Three commercially available CBD‐rich cannabis flowers were purchased online in Switzerland. One sample was grown outdoors and the other two indoors. The reported THC content was less than 1% by weight. Stress treatment was carried out on both whole flowers and flower extracts. For the whole flowers, 2 g of dried flowers were placed in clear screw‐top vials. The ethanolic extracts were prepared by grinding the flowers in a mortar and dissolving them in EtOH to give an extract of 20 mg/mL. The extract was sonicated for 5 min and then centrifuged at 4000 RPM for 15 min. Lastly, the extracts were filtered through a 0.45 μm PTFE syringe filter. Twenty additional C. sativa L. flowers were extracted for EE analysis by the Swiss Drug Testing Lab (Winterthur, Switzerland) using the sample preparation described.

2.7. Stress testing of authentic cannabis samples and multianalyte and single‐analyte mixtures

A stock solution containing 29 terpenes was prepared by dilution with EtOH to a final concentration of 1 mg/mL. One milliliter of aliquots of the multimix were transferred into clear screw‐top vials. Half of the aliquots were covered with aluminum foil to protect them from UV light and cross‐evaluate heat effects. For the stress test, covered and uncovered aliquots were placed in technical triplicates in a UV chamber (SOL 2, Filter H2 Honle UV technology, Gliching, Germany) at 120,000 Lux illumination and 42°C heat treatment. A wavelength range from UVB to IR was covered by the instrument. Intact flowers were extracted before and after stress testing. Treated samples were collected at different time points over the course of 28 days and diluted 20‐fold with EtOH, and a final concentration of 10 mM internal standard was added. As a proof of concept, selected terpenes were individually diluted to 1 mg/mL in EtOH and exposed to either UV light or heat alone. Authentic samples were exposed to UV light both as whole flowers and as ethanolic extracts and collected for measurement at six different time points.

2.8. Data analysis and visualization

A mixed‐effects model was fitted using GraphPad Prism (Version 10.1.2) to examine changes in concentration over time for selected terpenes. An interaction term was included, and sphericity was assumed in the model construction. MATLAB (Version 9.13.0, Release R2022b, The Mathworks Inc.) was used for the visualization of chromatograms.

3. RESULTS AND DISCUSSION

3.1. Chromatographic profile and method validation

Chromatographic analysis was performed on three GC‐FID systems and one GC–MS system. The GC‐FID methods presented here have been previously published and further extended to the analysis of C. sativa L. As no chiral column is able to separate all the enantiomers present, two phases were chosen to extend the separation range. In addition, a GC–MS system equipped with an apolar DB‐5 column was chosen to annotate new, unknown degradation products during the stress treatment of terpenes. The GC‐FID systems were equipped with a polar column and two different chiral columns. On the GC–MS system, separation was achieved on an apolar DB‐5 MS column. Of the 29 analytes analyzed, 13 were enantiomers and five were diastereomers. Farnesol exhibited an E/Z, E/E, and Z/E configuration which was separable on the BGB‐wax and chiral BGB 178 30% CD columns. None of the studied chiral columns showed satisfactory separation for all studied enantiomers. The (S)‐enantiomer and (R)‐enantiomer of limonene showed good separation on both chiral columns; however, the (S)‐enantiomer coeluted with other terpenes such as p‐cymene or α‐terpinene. The separation power of both chiral columns is partially complementary. Coelution was observed for neryl acetate (22) and trans‐citral (23) on the BGB‐wax column, α‐terpinene (9) and (S)‐limonene (10) as well as cis‐(+/−)‐menthone (21) and (+)‐linalool (22) on the BGB 178 30% CD, and (+)‐β‐pinene (7) and cis‐hexen‐1‐ol (8) and p‐cymene (11) and (S)‐limonene (12) as well as nerol (30) and (+/−)‐citronellol (31) on the BGB 176 SE column. These analytes were excluded from method validation. In cases where enantiomer pure standards were not available, the enantiomers are referred to by their elution number on the respective column (e.g., farnesols 1–3). Baseline separation is achieved with an R_s > 1.5, whereas an R_s value of 0.7 results in an 8% overlap. ⁴³ All analytes studied on the BGB‐wax column fulfilled the requirement for optimal peak separation. On the DB‐5 MS column, nerol (18) and citronellol (19) as well as eugenol (26) exhibited reduced values for the R_s; however, method validation led to a bias, RSD_R, and RSD_T for all QC levels below 20%. This allows the target analytes to be used for reliable quantification despite the insufficient chromatographic separation. Validation for the BGB‐wax GC‐FID and DB‐5 GC–MS methods resulted in biases and RSD_R and RSD_T values below 20%. Further, sensitivity presented as LoD and LoQ was lower for GC‐FID than GC–MS as a detector. Although enantiomer pure standards for rose oxide are commercially available, they comprise a mixture of the cis‐ and trans‐isomers. QC_low samples failed during method validation for some analytes studied on both chiral columns. Specifically, these failures were observed for trans‐(+/−)‐rose oxide (18), trans‐(+/−)‐menthone (23 and 24), cis‐(+/−)‐nerolidol (42), and farnesol (46) as well as cis‐ (31) and trans‐citral (27) analyzed on the BGB 178 30% CD column. Isomers for farnesol exhibited biases and RSD_R and RSD_T values below 20% on the BGB‐wax column, and the analysis using that specific method was thus satisfactory. The same condition was true for cis‐ and trans‐citral studied on the DB‐5 MS column. Failed QC_low samples were observed for trans‐(+/−)‐rose oxide (18 and 19), trans‐(+/−)‐menthone (24), and cis‐citral (29) on the BGB 176 SE column. Obtained values in the lower calibration range for the enantiomers of trans‐rose oxide (18 and 19) and trans‐menthone (24) are therefore considered as semiquantitative. However, it can be concluded that even when studying the same analytes at identical concentrations on the respective columns, sensitivity differences may arise, despite the use of the same detector. The separation of enantiomeric mixtures introduces a further dilution step, potentially resulting in concentrations below the LoQ. However, analyzing samples on different columns can yield valuable additional information. An overview of all validation data for the studied analytical systems is made available in Tables S1–S4. In addition, corresponding chromatograms are shown in Figures S1–S4. Method validation was conducted over the course of 3 months.

3.2. Stress testing of multimix and selected terpenes

3.2.1. Monoterpenes (multimix)

The effects of the two treatments, UV and heat, were examined using a mixed‐effects model. The model consisted of two fixed effects, time and treatment, and an interaction term (time × treatment). Significance was determined with a threshold of a p‐value below 0.05. Analysis on both the chiral and the polar columns provided consistent trends (Tables S5–S7). The screw‐top vials used in the stress testing were tightly sealed, but some loss of analytes due to evaporation can be expected. Time exhibited a significant effect on the terpene concentration, which decreased for all terpenes except for p‐cymene (Figure 1F). The formation of p‐cymene was observed during the light treatment of γ‐terpinene, which is consistent with literature reports. ⁵⁴ Double‐bond isomers can exhibit strong differences in their rate of degradation. This was observed for α‐pinene, β‐pinene, camphene, and sabinene. In addition, degradation rates vary in terpene classes, such as bicyclic nonoxygenated monoterpenes, as observed for sabinene, α‐pinene, β‐pinene, and camphene. Sabinene and β‐pinene showed greater degradation under UV light, whereas the degradation of α‐pinene was mostly time‐dependent (Figure 1A–D). After 18 days, 94% of the initial α‐pinene concentration was still present in the solution, whereas for β‐pinene, only 75% was found in the UV‐treated samples. However, both analytes were found to be quite stable under the influence of heat. Further, at the beginning of the stress treatment, a slight increase of α‐ and β‐pinene was observed due to transformation of other terpenes such as myrcene into the respective analytes. It was suggested by Rhoderick and Lin that β‐pinene is transformed into α‐pinene, limonene, p‐cymene, and camphene over the course of 6 months when kept in aluminum cylinders and that high temperatures might not be necessary for this conversion. They made similar observations for the long‐term stability of α‐pinene. ⁵⁵ , ⁵⁶ Complete and rapid degradation was observed for α‐terpinene after 72 h as well as for myrcene after 96 h (Figure 1E,G). All degradation profiles are displayed in Figures S5–S7. In an independent single‐analyte analysis, UV irradiation of myrcene resulted in the formation of α‐pinene, β‐pinene, and p‐cymene. More than 40 different autoxidation products have been described for myrcene, of which we were able to verify the above; however, we did not observe the formation of limonene and α‐terpinene. ⁵⁷ The treatment of myrcene with UV light seems to trigger cyclization and polymerization reactions in particular. Analysis with GC–MS further tentatively identified the formation of pseudolimonene, 3‐carene, trans‐geranylgeraniol, geranyllinalool, and 1‐heptatriacontanol (Figures S8–S12). The formation of hashishene from myrcene, as suggested by Marchini et al., was not confirmed in control samples. ⁴⁰ Mehdizadeh et al. conducted a study on the stability of Cuminum cyminium L. under different storage temperatures. Similar to our results, they also observed a moderate decrease of α‐ and β‐pinene and a rapid decline of myrcene at room temperature. ⁵⁸ Like myrcene, α‐terpinene was found to degrade rapidly. Interestingly, UV‐treated samples exhibited a lemon‐like smell. When comparing the degradation of α‐terpinene with its double‐bond isomer γ‐terpinene, a different rate of degradation becomes apparent (Figure 1G,H). In addition, limonene showed higher stability against heat and UV radiation compared with α‐terpinene and γ‐terpinene (Figure 1I). The degradation of α‐terpinene is accelerated by UV light, but temperature appears to be the main driver. Conversely, the stability of γ‐terpinene can be improved by light protection, and degradation is generally much slower. Further, the degradation appears to plateau and stabilize after around 16 days. Accordingly, a conversion of α‐terpinene to p‐cymene, limonene, α‐terpinolene, terpinen‐4‐ol, trans‐2‐caren‐4‐ol, ascaridol, citronellol, and linalyl acetate was observed (Figures S14 and 15). The polar GC‐FID analysis resulted in the observation of mainly oxidized products for α‐terpinene. These observations are partially consistent with those previously reported in the literature. ⁵⁹ , ⁶⁰ The transformation reactions of γ‐terpinene were not as diverse as those of α‐terpinene. The formation of α‐terpinene, α‐terpinolene, and p‐cymene was observed, as well as oxygenated products such as cis‐citral and carveol (Figures S16 and 17). P‐cymene appears to have a particularly stable structure against UV light, as camphene and α‐ and β‐pinene steadily decrease after 200 h (Figure 1A–C,F).

FIGURE 1.

Concentration profiles of terpene mixtures (n = 3) analyzed on a GC‐FID system equipped with a BGB‐wax column. The blue profile corresponds to UV treatment and the red profile to heat treatment. (A) α‐Pinene, (B) β‐pinene, (C) camphene, (D) sabinene, (E) myrcene, (F) p‐cymene, (G) α‐terpinene, (H) γ‐terpinene, and (I) limonene.

P‐cymene is frequently recognized as the terpene associated with aging processes in natural products and can be responsible for off‐flavors. ³⁷ However, the stability of p‐cymene over time and storage has been the subject of conflicting results in the literature. Some studies observed a reduction of p‐cymene during storage, while others proposed an increase of, for example, p‐cymene, camphor, and caryophyllene oxide if myrcene, α‐phellandrene, β‐caryophyllene, and/or α‐terpinene were present. ⁵⁸ , ⁶¹ , ⁶² It has also been previously stated that the specific oxidation reactions that occur can depend largely on the composition of the terpenes present and the type of stressor (UV light, oxygen, and heat) to which they are exposed to. ⁶² We suspect that these contradictory results for p‐cymene concentration changes can be directly related to this phenomenon. Overall, it was observed that p‐cymene was formed as the main degradation product during UV treatment of terpenes.

When assessing stability, it is worth mentioning that a quantitative approach should be preferred over an area‐normalized approach. For example, comparing the areas of α‐ and β‐pinene and α‐humulene and β‐caryophyllene suggests a transformation reaction (Figure 2). However, β‐pinene degrades more rapidly than α‐pinene (Figure 1A,B). A similar trend can be observed between α‐humulene and β‐caryophyllene (Figures 2 and 3).

FIGURE 2.

Area ratios between two double‐bond isomers. Areas (n = 3) were determined on the BGB‐wax column using GC‐FID.

FIGURE 3.

Concentration profile (n = 3) determined on a BGB‐wax column for α‐humulene and β‐caryophyllene.

3.2.2. Sesquiterpenes

The isomers α‐humulene and β‐caryophyllene exhibited similar degradation profiles and were strongly influenced by UV treatment (Figure 3). Overall, however, α‐humulene showed extended stability under UV light compared with β‐caryophyllene, which was nearly completely degraded after 400 h. During light treatment, we observed the formation of p‐cymene as well as humulene epoxide II and caryophyllene oxide for both α‐humulene and β‐caryophyllene. GC–MS analysis also revealed the formation of isocaryophyllene, α‐copaene, and α‐cubebene for β‐caryophyllene and cis‐α‐bisabolone and aromadendrene for α‐humulene (Figures S18–S21). While caryophyllene oxide is a well‐described product of the autoxidation of β‐caryophyllene, the other oxidation products are less studied. ⁶³ , ⁶⁴ , ⁶⁵ Overall, however, as the identities have been determined using libraries and retention indices, they should be treated with caution. Reference substances or NMR studies would be necessary to make a definitive statement. However, MS spectra indicate the formation of oxidized compounds strongly related to the caryophyllene structures. The identification of conversion products is challenging: The conversion of β‐caryophyllene to α‐humulene was not observed, and oxidation products and ring opening reactions are suspected. However, the presence of α‐humulene was confirmed in stressed β‐caryophyllene samples after 192 h of UV treatment, suggesting a favored configuration of α‐humulene over β‐caryophyllene.

3.2.3. Oxygenated terpenes

In contrast to the analytes previously presented, several analytes exhibited good stability toward the stressors such as cis‐ and trans‐rose oxide, isobornyl acetate, citronellyl acetate, geranyl acetate, and phenylethanol. Especially acetylated terpenoids showed higher stability compared with their nonacetylated counterparts (geraniol vs. geranyl acetate, nerol vs. neryl acetate, and citronellol vs. citronellyl acetate). Generally, oxygenated analytes appeared more susceptible to UV light, with the exception of citronellol, nerol, and geraniol, which are closely related in structure (Figure 4). However, this effect only takes place after approximately 8 days in a multimix and progresses slowly. Cis‐trans isomerization was observed for single‐analyte samples of nerol, leading to the formation of geraniol and citronellol in GC–MS experiments, but was not detected in single‐analyte samples of geraniol. Additionally, UV‐treated nerol samples exhibited the formation of p‐cymene, linalool oxide, geraniol, and epoxides. For citronellol, dihydro‐citronellol and α‐citronellol were observed, along with traces of nerol, citronellol, and geraniol. All chromatograms including the MS spectra of the formed reaction products can be studied in Figures S22–S30. It is likely that a cis‐trans transformation occurs among citronellol, geraniol, and nerol under the influence of UV light, consequently slowing down the concentration loss compared with heat‐treated samples. Potential polymerization products were also observed for all analytes, but not further investigated. Another explanation for the slower concentration decrease of geraniol, nerol, and citronellol under UV exposure may be derived from a study by He et al. focusing on the analysis of lemon‐flavored hard tea. They observed a decrease in the concentration of linalool, citronellol, cis‐ and trans‐citral, geraniol, and nerol during storage, which we were also able to confirm. In addition to their findings, it was suggested that linalool is particularly reactive and may decompose into α‐terpineol, geraniol, and nerol during storage. ⁶¹ In this study, the conversion of linalool into traces of nerol, citronellol, and geraniol was observed, along with the formation of limonene. Only after a 2‐week UV treatment, the formation of p‐cymene was observed (Figure S31).

FIGURE 4.

Concentration profile (n = 3) determined on a BGB‐wax column for citronellol, nerol, and geraniol.

3.2.4. Enantiomeric ratios

The conversion of enantiomers and/or diastereomers into their counterparts has been hypothesized numerous times in the literature. ³⁷ , ⁶¹ However, for all compound classes discussed before, these types of transformations were not observed for any of the diastereomer and enantiomer pairs studied.

The area ratios of selected enantiomers were also further compared. The conversion of (+)‐α‐pinene to (−)‐α‐pinene was not observed, underlined by the lack of detection of (−)‐α‐pinene. In addition, the conversion of (+)‐β‐pinene to (−)‐β‐pinene was not registered. The change of area ratios over time of enantiomers can be studied in Figure 5. With the exception of citronellol, no significant change in the ratios was observed. Heat treatment of citronellol leads to an overall decrease of the ratio. The concentration of citronellol decreased steadily during the temperature treatment. It is possible that one enantiomer was degraded more rapidly than the other was. It has to be noted that enantiomeric conversion may not have occurred under our experimental conditions.

FIGURE 5.

Comparison of area ratios (n = 3) of enantiomers over time under UV treatment (blue) and heat treatment (red). Data were acquired using GC‐FID equipped with a BGB 178 30% CD column.

3.3. Stress test of authentic C. sativa L. samples

Three authentic high‐CBD C. sativa L. extracts and flowers were analyzed using both GC‐FID and GC–MS before and after stress testing. The main terpene constituents identified were α‐ and β‐pinene, sabinene, myrcene, limonene, linalool, β‐caryophyllene, and α‐humulene, which is consistent with previously described results in the literature. ²⁰ , ⁶⁶ , ⁶⁷ , ⁶⁸ Sabinene and p‐cymene were detected as degradation products exclusively in stressed ethanolic extracts, but not in whole flowers. An initial increase in α‐ and β‐pinene and cis‐citral was also observed, consistent with the controlled stress study. Myrcene was degraded rapidly, but at different rates. While extracts contained no myrcene after 1 week, myrcene in whole flowers was degraded after about 2 weeks of UV treatment. Other rapid degradations were observed for linalool, β‐caryophyllene, and α‐humulene. A slight increase in geraniol was observed in extracts but not in flowers, and nerolidol showed an initial increase in peak area after 24 h and degraded by the end of the stress study. Corresponding chromatograms of authentic sample 1 are shown in Figure 6. Further GC‐FID profiles and MS spectra are available in Figures S32–S51. GC–MS analysis and spectral library comparison confirmed the presence of hashishene, but only in two of three stressed ethanolic extracts. In general, the profiles for stressed flowers and extracts were different. Whole flowers showed improved stability and reduced oxidation reactions. The EE was determined for authentic C. sativa L. samples, and changes during the UV treatment were monitored. A comprehensive table of all EEs is available in Table 1.

FIGURE 6.

GC‐FID chromatographic profiles of UV treated Cannabis sativa L. sample 1 as flower (A) and as an extract (B). The overlay contains the time points at 0 h, 24 h, 48 h, and 2 weeks of treatment of authentic samples separated on a BGB‐wax capillary column.

TABLE 1.

Enantiomeric excess (EE; %) over time for authentic Cannabis sativa L. samples (n = 3).

EE (%) of C. sativa L. flowers
Analyte	(+)‐α‐Pinene			(+)‐β‐Pinene			(S)‐limonene			(+)‐Linalool			(−)‐Citronellol			(−)‐Camphene			cis‐Nerolidol 2			(+)‐trans‐Nerolidol			(−)‐cis‐Menthone
Sample cond.	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID
0 h	−100	90	47	−45	56	14	86	82	88	88	82	87	n.d.	100	100	100	45	100	n.d.	n.d.	100	100	100	100	100	100	100
24 h	−100	89	43	−68	53	7	77	81	85	100	79	89	−100	100	100	100	33	n.d.	100	100	100	100	n.d.	100	100	100	100
48 h	−63	88	46	−54	53	2	80	86	89	100	100	100	−100	100	100	100	27	100	100	100	100	100	n.d.	100	100	100	100
72 h	−100	88	46	−28	57	−4	79	82	88	100	100	88	100	100	100	100	22	100	100	100	100	100	n.d.	n.d.	100	100	100
1 week	−100	88	47	−66	58	18	85	82	89	100	100	100	n.d.	100	100	100	22	100	100	100	100	100	n.d.	n.d.	100	100	n.d.
2 weeks	−100	88	48	−100	59	25	84	75	87	100	100	100	n.d.	100	100	100	25	100	100	100	100	100	n.d.	n.d.	100	100	n.d.

EE (%) of C. sativa L. extracts
Analyte	(+)‐α‐Pinene			(+)‐β‐Pinene			(S)‐limonene			(+)‐Linalool			(−)‐Citronellol			(−)‐Camphene			cis‐Nerolidol 2			(+)‐trans‐Nerolidol			(−)‐cis‐Menthone
Sample cond.	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID	1 OD	2 ID	3 ID
0 h	−100	90	47	−45	56	14	86	82	88	88	82	87	n.d.	100	100	100	45	100	n.d.	n.d.	100	100	100	100	100	100	100
24 h	−100	90	50	−100	50	3	78	83	88	100	100	100	−100	100	100	100	43	100	n.d.	100	100	3	n.d.	100	100	100	100
48 h	−100	90	42	−100	77	10	67	81	89	44	100	100	−100	100	100	100	27	100	n.d.	85	100	−44	100	100	100	100	100
72 h	−100	89	39	−100	51	18	78	83	88	60	100	100	n.d.	100	100	18	4	23	n.d.	100	n.d.	−48	100	n.d.	100	100	100
1 week	−68	86	26	−41	47	8	77	79	86	100	100	100	n.d.	100	100	−36	−8	−40	n.d.	100	100	n.d.	100	n.d.	100	100	n.d.
2 weeks	−80	100	−3	−52	48	11	100	79	89	100	100	100	n.d.	100	100	−13	6	−40	n.d.	0	100	n.d.	100	n.d.	100	100	n.d.

Note: Calculated EEs for whole flower samples on top and extracts on the bottom.

Unstressed samples showed an excess for (+)‐α‐pinene, (+)‐β‐pinene, (S)‐limonene, (+)‐linalool, (−)‐citronellol, (+/−)‐camphene 1, (+/−)‐cis‐nerolidol 2, (+/−)‐trans‐nerolidol 2, and (+/−)‐cis‐menthone 2 as shown in Table 1. Enantiomeric annotation was based on pure enantiomeric standards or based on their elution order, if neither literature values nor standards were available (e.g., farnesols 1–3 or (+/−)‐cis‐nerolidols 1 and 2). ¹ , ⁴⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ , ⁷² , ⁷³ Interestingly, one sample exhibited an excess for (−)‐α‐ and (−)‐β‐pinene, and no citronellol was detected. Differences in EE could be due to growing conditions but could also be cultivar‐dependent. Previous studies have either not reported the conditions under which the plants were grown or have only included samples from one growth condition. ¹ , ⁴⁷ , ⁴⁸ Overall, the EE in flower samples for most analytes was stable throughout the entire study, and conversion reactions from one enantiomer into the other were not observed. The EE for α‐ and β‐pinene exhibited some dynamic changes. In sample 1, a slight increase of (−)‐α‐pinene was observed in the flowers, while (+)‐α‐pinene was not detected until the end of the stress study. The degradation profile for extracts was different. A steep increase was observed for the peak area of (−)‐α‐pinene, while a weak increase was registered for (+)‐α‐pinene toward the end of the study. (−)‐β‐pinene showed a similar behavior in stressed flower and extract profiles. Samples 2 and 3 showed similar profiles for stressed flowers, whereas the degradation profile of the extract for sample 2 was flat and barely any degradation reactions were observed. In contrast, the extracts of samples 1 and 3 showed a strong increase of (−)‐α‐pinene toward the end of the study. An increase and decrease caused by the conversion of one enantiomer into the other could not be deduced from the profiles. Profiles for α‐pinene and β‐pinene can be found in Figure S52. It is likely that any enantiomeric concentration increase is due to the conversion of another terpene into the corresponding analyte and not due to the transformation of, for example, (−)‐α‐pinene to (+)‐α‐pinene. Because the formation of α‐pinene from myrcene was shown above, we hypothesize that the majority of pinenes is formed from cyclization reactions of myrcene. Overall, no uniform degradation patterns were observed for both enantiomers of α‐pinene and β‐pinene. As differences in EE were observed for α‐ and β‐pinene, 20 additional ethanolic flower extracts of C. sativa L. were analyzed. A complete table of all the EE determined can be found in Table S8. Nine samples were grown outdoor, and 11 were grown indoor. Six samples showed an excess for (−)‐α‐pinene and seven for (−)‐β‐pinene. Furthermore, no citronellol was detected in these samples, and most of them were indoor grown plants. The presence of citronellol could indicate an excess of either the (+)‐ or (−)‐enantiomer, potentially making additional chiral analysis unnecessary. Figure 7 shows the EE distribution for the analyzed 20 authentic samples. Variation in the EE was only observed for both α‐ and β‐pinene. (+/−)‐cis‐Nerolidol 2 was detected in only one sample. All other enantiomers were found in similar EE across the tested samples. Compared with their (−)‐counterparts, (+)‐α‐pinene and (+)‐β‐pinene have higher antibacterial activity, while (+)‐α‐pinene also has higher antifungal activity. ³¹ , ⁷⁴ As plants produce enantiomers based on survival pressure, we hypothesize that α‐ and β‐pinene are produced in different amounts depending on growth conditions. Citronellol is also known to act as an antifungal and herbicide in plants. ⁷⁵ , ⁷⁶ The simultaneous absence of (+)‐α‐ and (+)‐β‐pinene and citronellol suggests that there is less evolutionary pressure for the plant to produce these compounds. This could be due to growth conditions such as indoor versus outdoor cultivation or the use of external antifungals and herbicides. This observation also highlights the importance of understanding the enantiomeric profile of terpenes as this can influence pharmacological activity.

FIGURE 7.

The enantiomeric excess (EE; %) for 20 authentic Cannabis sativa L. extracts observed after chiral GC‐FID analysis. The enantiomeric ratio as well as the column used to the determine the EE is presented in Table S9.

Supporting information

Table S1: Elution order of analytes studied on a BGB‐wax column with their corresponding retention time (RT), resolution (R_s), retention index (RI) and validation results. RSD_R and RSD_T are relative standard deviations for the intra‐ and inter‐day imprecision, respectively.

Table S2: Elution order of analytes studied on a DB‐5 MS column with their corresponding retention time (RT), resolution (R_s), retention index (RI) and validation results. RSD_R and RSD_T are relative standard deviations for the intra‐ and inter‐day imprecision, respectively.

Table S3: Elution order of analytes studied on a BGB 178 30% CD column with their corresponding retention time (RT), resolution (Rs), retention index (RI) and validation results. RSDR and RSDT are relative standard deviations for the intra‐ and inter‐day imprecision, respectively. Failed QC samples are marked bold.

Table S4: Elution order of analytes studied on a BGB 176 SE column with their corresponding retention time (RT), resolution (Rs), retention index (RI) and validation results. RSDR and RSDT are relative standard deviations for the intra‐ and inter‐day imprecision, respectively. Failed QC samples are marked bold.

Figure S1: Chromatogram of terpene multi mix analysed on a BGB‐wax column with GC‐FID Analytes are colour coded to see different elution orders across chromatographic platforms. A complete list of analytes can be found in table S4.

Figure S2: Chromatogram of terpenes multi mix analysed on a DB‐5 MS column with GC–MS. Analytes are colour coded to see different elution orders across chromatographic platforms. A complete list of analytes can be found in table S4.

Figure S3: Chromatogram of terpenes analysed on a BGB 178 30% CD column with GC‐FID. Analytes are colour coded to see different elution orders across chromatographic platforms. A complete list of analytes can be found in table S4.

Figure S4: Chromatogram for terpenes analysed on a BGB 176 SE column with GC‐FID. Analytes are colour coded to see different elution orders across chromatographic platforms. A complete list of analytes can be found in table S4.

Table S5: results for the mixed effects model on the BGB‐wax column. Significance levels were set at ns for P > 0.05, * for ≤ 0.5, ** for P ≤ 0.01, *** for P ≤ 0.001 and **** for P ≤ 0.0001.

Table S6: results for the mixed effects model on the BGB 178 30% CD column. Significance levels were set at ns for P > 0.05, * for ≤ 0.5, ** for P ≤ 0.01, *** for P ≤ 0.001 and **** for P ≤ 0.0001.

Table S7: results for the mixed effects model on the BGB 176 SE column. Significance levels were set at ns for P > 0.05, * for ≤ 0.5, ** for P ≤ 0.01, *** for P ≤ 0.001 and **** for P ≤ 0.0001.

Figure S5: Concentration changes of terpenes after UV light treatment (blue) and heat treatment (red). Profiles were acquired using GC‐FID equipped with a BGB wax column. Experiments were carried out in triplicates (n = 3).

Figure S6: Concentration changes of terpenes after UV light treatment (blue) and heat treatment (red). Profiles were acquired using GC‐FID equipped with a BGB 178 30% CD chiral column. Experiments were carried out in triplicates (n = 3).

Figure S7: Concentration changes of terpenes after UV light treatment (blue) and heat treatment (red). Profiles were acquired using GC‐FID equipped with a BGB 176 SE chiral column. Experiments were carried out in triplicates (n = 3).

Figure S8: GC–MS profile of a stressed 1 mg/ml myrcene solution for 48 hours. Asterisk (*) marks putatively identified compounds.

Figure S9: MS spectrum of pseudolimonene and trans‐geranylgeraniol.

Figure S10: GC–MS chromatogram and MS spectrum of α‐pinene identified in a 144 h UV‐stressed myrcene sample.

Figure S11: GC–MS chromatogram and MS spectrum of geranyllinalool and 1‐heptatriacotanol identified in a 144 h UV‐stressed myrcene sample.

Figure S12: GC–MS chromatogram and MS spectrum of 2,7‐dimethyl‐2,7‐octadiene and 3‐carene identified in a 336 h UV‐stressed myrcene sample.

Figure S13: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml myrcene samples, which was stressed for 48 hours.

Figure S14: GC–MS profile with corresponding MS spectra for a 1 mg/ml α‐terpinene solution. The solution was stressed for 48 hours with UV‐light.

Figure S15: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml α‐terpinene sample, which was stressed for 48 hours.

Figure S16: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml γ‐terpinene sample, which was stressed for 168 hours.

Figure S17: Chromatogram and MS spectrum acquired using GC–MS equipped with a DB‐5 MS column of a 1 mg/ml γ‐terpinene sample, which was stressed for 168 hours with UV light.

Figure S18: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml α‐humulene sample, which was stressed for 192 hours.

Figure 19: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml β‐caryophyllene sample, which was stressed for 192 hours.

Figure S20: Chromatogram and MS spectrum acquired using GC–MS equipped with a DB‐5 MS column of a 1 mg/ml α‐humulene sample, which was stressed for 192 hours with UV light.

Figure S21: Chromatogram and MS spectrum acquired using GC–MS equipped with a DB‐5 MS column of a 1 mg/ml β‐caryophyllene sample, which was stressed for 192 hours with UV light.

Figure S22: Chromatogram and MS spectrum acquired using GC–MS equipped with a DB‐5 MS column of a 1 mg/ml nerol sample, which was stressed for 192 hours with UV light.

Figure S23: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml nerol sample, which was stressed for 192 hours.

Figure S24: Chromatogram and MS spectrum acquired using GC–MS equipped with a DB‐5 MS column of a 1 mg/ml nerol sample, which was stressed for 336 hours with UV light.

Figure S25: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml geraniol sample, which was stressed for 192 hours.

Figure S26: Chromatogram and MS spectrum acquired using GC–MS equipped with a DB‐5 MS column of a 1 mg/ml geraniol sample, which was stressed for 192 hours with UV light.

Figure S27: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml citronellol sample, which was stressed for 192 hours.

Figure S28: Chromatogram and MS spectrum acquired using GC–MS equipped with a DB‐5 MS column of a 1 mg/ml citronellol sample, which was stressed for 192 hours with UV light.

Figure S29: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml linalool sample, which was stressed for 192 hours.

Figure S30: Chromatogram and MS spectrum acquired using GC–MS equipped with a DB‐5 MS column of a 1 mg/ml linalool sample, which was stressed for 192 hours with UV light.

Figure S31: Chromatogram acquired using GC‐FID equipped with a BGB‐wax column of a 1 mg/ml linalool sample, which was stressed for 336 hours. The above chromatogram presents the linalool sample during UV treatment and the bottom only under heat treatment.

Figure S32: UV treated C. sativa L. sample 2 as an extract. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S33: UV treated C. sativa L. sample 2 as a flower. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S34: UV treated C. sativa L. sample 3 as an extract. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S35: UV treated C. sativa L. sample 3 as a flower. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S36: UV treated C. sativa L. sample 1 as an extract. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S37: UV treated C. sativa L. sample 1 as a flower. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S38: UV treated C. sativa L. sample 2 as an extract. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S39: UV treated C. sativa L. sample 2 as a flower. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S40: UV treated C. sativa L. sample 3 as an extract. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S41: UV treated C. sativa L. sample 3 as a flower. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S42: UV treated C. sativa L. sample 1 as an extract. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S43: UV treated C. sativa L. sample 1 as a flower. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S44: UV treated C. sativa L. sample 2 as an extract. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S45: UV treated C. sativa L. sample 2 as a flower. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S46: UV treated C. sativa L. sample 3 as an extract. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S47: UV treated C. sativa L. sample 3 as a flower. The overlay contains the time points at 0, 24 h, 48 h and 2 weeks of treatment.

Figure S48: GC–MS chromatographic profile of authentic C. sativa L. samples before stress treatment.

Figure S49: GC–MS chromatographic profile of authentic C. sativa L. flower samples two weeks after stress treatment.

Figure S50: GC–MS chromatographic profile of authentic C. sativa L. extract samples two weeks after stress treatment.

Figure S51: MS spectra of degradation profiles after two weeks after stress treatment.

Figure S52: Change of peak areas against time of (−)‐α‐pinene (orange), (+)‐α‐pinene (green), (−)‐β‐pinene (blue) and (+)‐β‐pinene (pink) after UV treatment. Data was acquired using GC‐FID equipped with a BGB 176 SE chiral column. The degradation profile for flowers are presented on top row and for extract on the bottom row.

Table S8: EE [%] for 20 authentic C. sativa L. samples analysed using chiral GC‐FID.

Table S9: Enantiomeric ratio for 3 authentic C. sativa L. samples analysed using chiral GC‐FID. Undefined is abbreviated as undef and not detected as n.d. The numerator was the analyte in the analyte row.

Raeber J, Bajor B, Poetzsch M, Steuer C. Comprehensive analysis of chemical and enantiomeric stability of terpenes in Cannabis sativa L. flowers. Phytochemical Analysis. 2025;36(1):205‐217. doi: 10.1002/pca.3432

DATA AVAILABILITY STATEMENT

The data supporting this article have been included as part of the supporting information.