Modulation of cannabis flower characteristics and THC through a biostimulant complex of molasses, Aloe vera extract, and fish-hydrolysate

Experiment 1

The plants (n = 8) were cut at the base, measured for height, and stem width (Digital Vernier Calliper 150 mm, Kincrome, Scoresby, VIC, Australia), and then all flowers were removed by hand and weighed (perplant fresh weight yield, FW). Five flowers were selected randomly from each plant, weighed individually, and then dried in an oven (ED 53, Binder GmbH, Tuttlingen, Germany) for 72 hr at 55°C and re-weighed (dry weight [DW]) to calculate flower water content.

Experiment 2

Before felling of the plants, one random representative flower was removed from each plant (n = 3), and the sugar leaves were immediately removed with a scalpel and sampled separately in liquid nitrogen and then transported on dry ice to a −80°C freezer for storage before analyses.

The plants (n = 3) were then cut at the base and hung upside down within the grow room for 2 weeks to dry in the dark, at 18°C, and 50% relative humidity (RH). All inflorescences (flowers) were removed using an automated bucker (Trimpro Bucker, Trimpro Manufacturing Inc., Montreal, Canada), trimmed to remove excess leaves using an automated trimmer (BatchOne Go, Twister Trimmer, Keirton Inc., Surry, Canada), and placed into curing bags (TerpLoc, Grove Bags LLC, Bedford Heights, OH, USA) for 1 week in the dark at 25°C, to allow the partially dried flowers to mature as per industry standard practise. After 1 week in the curing bags, three cured flowers of consistent size, shape, and appearance were selected from each plant set of flowers (n = 9) for analysis consisting of consecutively: near-infrared (N-IR) analysis, sensory analysis, and headspace (HS) analysis. Each flower sample was weighed prior to analysis (cure weight [CW]), and then following analysis were dried in an oven (ED 53, Binder GMBH, Tuttlingen, Germany) for 72 hr at 55°C and re-weighed (DW). Cured flower water content was calculated as a percentage of the weight loss from drying.

Flower volatiles

Profile analysis (PCA) of cannabis flower volatile profiles indicated that they were not significantly different between treatments (Figure S1 in Supplementary Materials). When comparing the changes to individual peaks of the HS spectra, BC treatment was associated with significant (p < 0.05) or marginally significant (p < 0.1) changes to 24 peaks (out of 44 peaks total), which were all reduced as compared to the control. Of the identified peaks (Table S2 in Supplementary Materials), 10 volatiles were reduced with marginal significance (p < 0.1, Table 1), which were reduced between 0.64-fold and 0.86-fold. This reduction in volatiles from BC treatment is surprising in light of similar studies, such as Malík et al. (2022), which observed significant increases to the volatile terpenes limonene and β-myrcene from the growth of cannabis following supplementation with an amino acid biostimulant. These same volatiles were not significantly changed in this study (Table S2 in Supplementary Materials). However, the trends of decreased volatiles indicate an opposite directional effect has occurred on cannabis from the application of BC, which contained fish hydrolysate (an amino acidrich complex).

Human OR affinity model

Volatiles that bind with ORs within the olfactory system elicit neuronal signals which have the potential to result in perception of an associated odour (Bak et al., 2018). The strength of the interaction between the volatile and the OR (the affinity) is impactful to the likelihood of successful binding and then triggering of the associated neuronal signal (Persuy et al., 2015). Prediction of OR affinity (model 1: Table S3 in Supplementary Materials) was achieved with a 13-term model comprised of the physicochemical measures of LogP, MW, and Nrot with an R² = 0.9158, and a 10-fold R² = 0.8804 (Figure S2 in Supplementary Materials). The OR affinities of the volatiles impacted by BC (Table 1) were found to be between −7.0 kcal · mol⁻¹ (ylangene and β-caryophyllene) and −5.0 kcal · mol⁻¹ (2-norbornanol).

The use of the physicochemical measures MW, LogP, and Nrot to predict OR affinity is consistent with the previous study (Castro et al., 2021), which explored relationships between a range of physicochemical measures with OR affinity and found that MW, LogP, and vapour pressure had the strongest relationships. The OR affinity model is an advancement on this prior work, by combining these relationships into a predictive model with validation, thereby allowing for application to new volatiles. The model predicts binding affinity, which is a negative value wherein larger negative values indicate higher binding affinity (Henrich et al., 2010). An experimentally determined affinity value of −3.39 kcal · mol⁻¹ was considered a very weak binding (Boyce et al., 2009), while some high-affinity interactions have been measured between −9.0 kcal · mol⁻¹ and −12.7 kcal · mol⁻¹ (Velazquez-Campoy and Freire, 2006). Accordingly, given the range of OR affinities predicted for the volatiles changed from the BC treatment was between −7.0 kcal · mol⁻¹ and −5.0 kcal · mol⁻¹, these represent moderatestrength affinities and so would likely bind with the ORs. Therefore, based on the OR affinity prediction of these changed volatiles, BC is altering volatiles that are potentially impactful to odour perception.

Flower sensory analysis – Strength, desirability, quality, and profile changes

Noting that the BC-altered flower volatiles were predicted to have moderate-strength OR affinity, sensory testing was performed to explore changes to perceived odour. The sensory assay participant demographics were of diverse ages and had an approximately even split in ethnicity and gender (Table S4 in Supplementary Materials). The ages of the participants ranged from 19 years to 63 years (mean 34.9 years), 43% were White or Caucasian ethnicity, 57% were Asian or Pacific Islander ethnicity, 50% were male, and 50% were female.

Profile analysis of the sensory data indicated that the profiles were not significantly different between treatments (Figure S3 in Supplementary Materials). Comparison of individual sensory measures indicated that the flower odour strength was significantly reduced (p = 0.007) from an average of 2.1 for control to 1.6 for BC treatment (Figure 1, Table S5 in Supplementary Materials), which represented an on-average 0.5 change on a 5-point Likert scale. No significant differences were identified between the treatments for their desirability (p = 0.547) or quality (p = 0.103), which had means for both treatment groups close to 0 (neither desirable nor undesirable and average quality, respectively).

Figure 1.

Application of the same complex (BC) to other plants has been associated with mixed effects on odour, such as when applied to tomato there was no change to fruit odour measures (Wise and Selby-Pham, 2024a), when applied to strawberry under controlled conditions BC increased fruit odour (Wise et al., 2024b), and when BC was applied to strawberry under field conditions, it reduced odour intensity (Wise and Selby-Pham, 2024b) – which is similar to the reduction in odour strength observed in cannabis flowers in this study. These results indicate there are differences observed from biostimulant applications between species (tomato, strawberry, and cannabis), tissue types (fruit vs flowers), and growth conditions (indoor vs field). These varied biostimulant effects are consistent with Baghdadi et al. (2022), who observed differences in biostimulant effects from kelp when applied to tomatoes under lab, greenhouse, or field conditions. One of the hypothesised modes of action of these plant-derived biostimulants and their associated complexes is the indirect plant response to the stimulation of microbial species from the addition of carbohydrates to the substrate (Wise et al., 2024d). However, substrate microbial populations have been shown to be highly variable even following inoculation (Sheridan et al., 2017), and microbe stimulation by carbohydrate provision would be relatively nondiscriminant. Therefore, variation in either substrate or rhizosphere microbiome compositions may account for across-study differences in observed BC effects on plant odour.

The perceived strength of an odour is dependent on the odour intensity of the volatile compounds and their concentration (Rincón et al., 2019). Therefore, the reduction in odour strength of the flowers is likely driven by the reduction in volatile contents of flowers (Table 1). However, cannabis odour pleasantness (i.e. desirability), but not terpene contents, has been associated with pleasant subjective effects as an indicator of quality (Plumb et al., 2022). This is consistent with the results in this study, where despite the changes in volatiles, neither desirability nor quality odour assessments were significantly impacted by the BC treatment. Within the context of medicinal cannabis patients, the reduction in odour strength may be desirable for certain patient groups, such as ASD patients who use cannabis for managing anxiety and sleep-related symptoms (Erridge et al., 2022) but tend to prefer reduced-odour or unscented medicinal products (Hrdlicka et al., 2011; Schroder et al., 2021; Chitta, 2024), due to having enhanced sensory sensitivity. This is impactful for achieving acceptance by a broader patient base, as optimising medicinal product odour properties promotes patient acceptability, which thereby achieves greater compliance and associated product efficacy.

Modelling of cannabis flower odour properties

The association of volatile changes with the odour measures of strength, desirability, and quality were further explored by LR analysis. No significant volatile-based relationship was identified for odour desirability or quality (data not shown); however, a significant relationship was identified between odour strength and volatiles (model 2: Table S6 and Table S7 in Supplementary Materials). This model included 12 variables (volatiles), including α-humulene, p-cymen-8-ol, α-pinene, γ-terpinene, and ipsdienol which were positively correlated and α-terpinene, β-caryophyllene, β-eudesmene, 2-pinanol, camphene, dehydrosabinene, and fenchol which were negatively correlated. Noting that overall strength was reduced from the BC treatment (Figure 1), and that only significant reductions in volatiles were observed (Table 1), this would suggest that the positively correlating variables (volatiles) are likely driving the observed treatment-associated change in odour strength. Of the positively correlating variables, α-humulene and p-cymen-8-ol were the only volatiles that were reduced by the BC treatment (0.66-fold reduced, p = 0.052; and 0.70-fold reduced, p = 0.005, respectively), suggesting these reductions were key drivers of the BC treatment-associated reductions to perceived cannabis odour strength.

Impact of BC on cannabis flower ODs

Although cannabis has a distinct and strong odour, selective breeding has produced many different strains with diverse volatile profiles that produce subtly different odours (Gilbert and DiVerdi, 2018). The sub-odours that comprise the overall cannabis odour are known as ‘odour descriptors’ (Wise et al., 2023), which were compared between the treatment flowers by sensory odour assay. Out of the 30 ODs included in the questionnaire, 21 were detected at least once in the control flowers, 20 were detected at least once in the BC flowers, and 18 were detected at least once in both (Table S8 in Supplementary Materials). The most common descriptor detected was ‘herbal’, which was detected 50% of the time for control and 52% of the time for BC treatment; however, the difference in these proportions was not significantly different (p = 0.827). ‘Fruity’ was detected 14% of the time for control, which was significantly (p = 0.043) more than BC treatment, where it was detected 2% of the time. There was also a marginally significant (p = 0.081) difference in ‘minty’ detection, which was 36% for control and 19% for BC treatment.

When comparing the average strength of the detected odours (Figure 2, Table S9 in Supplementary Materials), all odours were between strength 0 (none) and 1 (weak), except ‘herbal’ which was slightly >1 for both conditions and not significantly different between treatments (p = 0.818). Two descriptors were identified as having significantly different strengths, which were both significantly reduced by the BC treatment (Figure 2). These were ‘minty’, which had an average strength of 0.690 for the control and 0.262 for the BC treatment (p = 0.036), and ‘fruity’, which had an average strength of 0.191 for the control and 0.024 for the BC treatment (p = 0.046). ODs within an odour profile can be attributed to a subset of volatiles (Sánchez et al., 2020), therefore, these results indicate a likely change in volatiles associated with minty and fruity odours. Of the volatiles which were altered from the BC treatment and their associated individual ODs (Table S10 in Supplementary Materials), borneol and terpineol are described as having partly minty odours, while p-cymen-8-ol is described as having a partly fruity odour. Accordingly, the reduction in these volatiles may explain the reductions in the minty and fruity strength of BC flowers within the sensory assay (Figure S4 in Supplementary Materials). These reductions to cannabis flower odour may be desirable changes for paediatric patients, who find minty odours and taste unpleasant (Hoffman et al., 2016) and for whom improving palatability is associated with improved adherence to dosing regimens and therefore therapeutic outcomes (Walsh et al., 2014).

Figure 2.

Cannabinoid changes

Cannabinoids such as CBD and THC are the major sources of therapeutic potential within cannabis flowers (Andre et al., 2016), and THC specifically is the primary quality attribute of recreational cannabis due to its psychoactive properties (Cash et al., 2020). Therefore, both medicinal and recreational cannabis cultivators tend to optimise their practices to maximise THC content of flowers (Trancoso et al., 2022). Quantitative N-IR models for the prediction of flower cannabinoid content were produced for CBD, which achieved an R² = 0.9619, and for THC, which achieved an R² = 0.9851. Application of these models to experimental samples (Table S11 in Supplementary Materials) indicated that the CBD content was unchanged between treatments (p = 0.434); however, THC content was significantly (p = 0.016) increased by 3-fold from 0.32% in control plants to 0.99% in BC treatment plants. Medicinal and recreational varieties tend to have much higher THC content >15% (Pennypacker et al., 2022), as compared to the industrial variety utilised within this work. However, this trend of increased THC content (3-fold increase) from the BC treatment is promising for cultivators of varieties where increasing THC content is a primary objective.

Changes to volatile pharmacokinetics

Prediction of T_½ (model 3: Table S12 in Supplementary Materials) was achieved with an 88-term model comprised of the physicochemical measures number_ of_aliphatic_OH_groups, fraction_of_rotatable_bonds, number_of_heterocycles, molarRefractivity, number_ of_atoms, number_of_bonds, geometrical_radius, zagreb_group_index_1, zagreb_group_index_2, abonds, logP_1, R2NH, ROH, RCOR, and RCCH, with an R² = 0.7980, and a 10-fold R² = 0.6067 (Figure S5 in Supplementary Materials). In the previous study, Wise et al. (2019) presents a model for T_½ prediction based on 14 compounds and the physicochemical measures molecular weight, volume, and number of rotatable bonds. Accordingly, the expansion of this to incorporate a larger dataset of compounds along with validation is a substantial improvement in the predictive capabilities. Application of the T_½ predictive model to the volatiles altered from the BC treatment (Table 2) predicts that these compounds have half-lives between 9.2 hr for p-Cymen-8-ol and 43.9 hr for α-humulene. For drug design, a T_½ of between 12 hr and 48 hr is considered ideal for daily dosing (Smith et al., 2017), and except for 2-norbornanol (T_½ = 9.4 hr) and p-Cymen-8-ol (T_½ = 9.2 hr), all the volatiles that were changed from the BC treatment had predicted T_½ within this daily dosing range (Table 2). Accordingly, the compounds impacted by the BC treatment are predicted to remain in circulation for a sufficient duration to be potentially pharmacologically active.

Following absorption into the circulatory system, neurologically active compounds must traverse the BBB to elicit effects within the brain (Nagpal et al., 2013). Although the majority of drugs do not cross the BBB, the potential for a drug to have neurological pharmacological potential can be screened based on its permeability index of the BBB (Pardridge, 2003; Nicolazzo et al., 2006). Prediction of BBB indices (model 4: Table S13 in Supplementary Materials) was achieved with an 18-term model comprised of the physicochemical measures of logP, MW, Nrot, and volume, with an R² = 0.8716, and a 10-fold R² = 0.7689 (Figure S6 in Supplementary Materials). Other studies have also explored the prediction of the BBB index by QSAR, such as Gupta et al. (2019), which identified a predictive model based on 270 CNS and 720 non-CNS oral drugs, and Ma et al. (2005), which developed a model from 35 organic compounds that was applied to a small training set of 8 compounds. Interestingly, both these previous models utilised different physiochemical properties to each other and the volatile BBB model presented herein (model 4). The BBB model presented in this study is an improvement on the previous studies, as it is specific to essential oil compounds (organic volatiles) and utilises a k-fold crossvalidation approach to improve the predictive reliability for application to new compounds.

Application of the BBB index model to the volatiles altered from the BC treatment (Table 2) indicated that these compounds had predicted BBB values between 0 for 2-norbornanol and 14.2 for α-humulene. According to the classification in Ma et al. (2005), compounds with a BBB index value >2 cross the BBB readily. Further, a comparison of CNS and non-CNS pharmaceutical drugs reported by Gupta et al. (2019) concluded that the CNS drugs have an average BBB index of 4.79 and the non-CNS drugs have an average BBB index of 3.29. Therefore, with the exception of 2-norbornanol (BBB index = 0), all the volatiles changed by the BC treatment are predicted to have the potential to cross the BBB, with seven compounds having very high BBB index scores (≥5), similar to those of many CNS-active drugs. Accordingly, the compounds impacted by the BC treatment are predicted to cross the BBB with the potential to impart neuropharmacological activity.

Of the volatiles which were significantly changed following the treatment, linalool, borneol, terpineole, β-caryophyllene, and α-humulene are the most studied for their medicinal potential, which share a range of demonstrated in vitro benefits relating to inflammation, cardiovascular health, cancer, neuroprotection and more (Machado et al., 2018; Pereira et al., 2018; An et al., 2021; de Lacerda Leite et al., 2021; Chen et al., 2023; Mei et al., 2023). The association with neuroprotection is consistent with the high BBB index scores predicted for these compounds, and borneol has been shown to function as an adjuvant for further increasing BBB permeability (Mei et al., 2023). Accordingly, the reductions in these compounds may represent a reduction in the neuroprotective or neuropharmacological benefits of the BC-grown flowers.

Sugar leaf metabolites

Cannabis sugar leaves are the small leaves surrounding the female flowers within the inflorescence (Spitzer-Rimon et al., 2019; Das et al., 2022). The protruding portion of the sugar leaf is partially removed during cannabis flower processing; however, due to their dense clustering within the inflorescence, a portion of the sugar leaf remains attached and so is utilised as a part of the ‘flower’. Sugar leaves are also one of the tissues throughout the inflorescence where glandular trichomes are present, which are the storage sites for many cannabis metabolites (Tanney et al., 2021). Accordingly, changes in sugar leaf metabolites represent changes to metabolites in the utilised portion of the inflorescence, as well as indicative changes that may have occurred within the adjacent flower tissue and trichomes. Sugar-leaf metabolite profiles were found to be significantly different between control and BC-grown plants, based on the non-overlapping regions of 95% confidence within the PCA (Figure 3). Comparison of individual metabolites in sugar leaves indicated that BC treatment was associated with significant reductions to 25 peaks, and a further 29 peaks were varied with marginal significance (all reduced between 0.45-fold and 0.91-fold). This corresponded with 11 identified metabolites (Table S14 in Supplementary Materials) that were significantly reduced (between 0.52-fold and 0.85-fold), and 14 identified metabolites that were reduced with marginal significance (between 0.47-fold and 0.85-fold). Of these reduced metabolites, several have been associated with beneficial bioactivities and therefore, reductions in their presence within the flowers are not considered beneficial for the therapeutic use of the associated flower. Most notable are the reductions to cannabisin B, orientin, and apigetrin, which are associated with both antioxidant (Chen et al., 2012; Ku et al., 2014; Ali et al., 2017) and neuroprotective potentials (Law et al., 2014; Lim et al., 2016; Di Palo et al., 2022). Furthermore, orientin specifically is associated with anxiolytic activities (Ku et al., 2014), and so reduction in this metabolite may affect the utility of these cannabis flowers for anxiety treatment. Therefore, based on the metabolite changes within the sugar leaves that make up a portion of the utilised flower, the growth of cannabis with BC produced flowers with reduced presence of several potentially beneficial therapeutic metabolites.

Figure 3.

Sugar leaves were also analysed by N-IR, wherein profile analysis (sPLS-DA) indicated that the IR spectra were not significantly different between treatments (Figure S7 in Supplementary Materials). A comparison of individual wavenumbers identified a region where the IR spectra were considerably reduced from the BC treatment (Figure 4). The broadest section of this region was between wavenumbers 6337–7628 cm⁻¹ (1311–1578 nm), wherein spectra were reduced with marginal significance (p < 0.1), while a narrower portion of the spectra between 6734 cm⁻¹ and 7623 cm⁻¹ (1312–1485 nm) were significantly (p < 0.05) reduced between treatments. The region of the spectra between 1400 nm and 1450 nm corresponds with water due to the hydroxy groups (O-H), however, this region is also attributed to hydrocarbons (C-H) and alcohols (R-OH), resulting in the association of 1450 nm with cannabinoids (THC, CBD, and cannabigerol), and terpenes (β-myrcene, and D-limonene) (Birenboim et al., 2022; Tran et al., 2024). Noting that the water content of the analysed cured flowers was not significantly different between treatments (p = 0.865, Table S11 in Supplementary Materials), and cannabinoid content (CBD and THC) was either unchanged or increased (Table S11 in Supplementary Materials), it seems unlikely that these reductions in the spectra would be attributed to changes in moisture or cannabinoids. The specific terpenes attributed to the 1450 nm band are β-myrcene, and D-limonene (Birenboim et al., 2022), and whilst these weren’t significantly changed, they were both on average reduced (0.85-fold p = 0.117 and 0.96-fold p = 0.401, respectively), and so may have contributed to this change in the spectra. Furthermore, for the volatiles that were significantly changed from the BC treatment (Table 1), they all contain either the hydrocarbon (C-H) or alcohol (R-OH) group, and so would likely affect the spectra within this region. Accordingly, the N-IR spectral reduction across the region of 6337–7628 cm⁻¹ (1311-1578 nm) may reflect the general reduction in terpenes and associated odour strength from the BC treatment.

Figure 4.

Modelling the proportion of reviewers indicating that a strain helps with anxiety or depression

Noting that cannabis therapeutic potential is associated with various phytochemicals (as discussed above), it was hypothesised that statistical models could be generated from flower’s volatile and THC profiles to predict the proportion of reviewers that find it helpful for anxiety and depression treatment. Prediction of reviewer proportions for being helpful for anxiety (model 5: Table S15 in Supplementary Materials) was achieved with a 19-term model comprising the variables α-humulene, β-limonene, β-myrcene, β-ocimene, camphene, γ-terpinene, terpinolene, THC, and transnerolidol, with an R² = 0.6831, and a 10-fold R² = 0.5615 (Figure S8 in Supplementary Materials). Prediction of reviewer proportions for being helpful for depression (model 6: Table S16 in Supplementary Materials) was achieved with a 10-term model comprising of the variables eucalyptol, linalool, THC, trans-ocimene and trans-nerolidol, with an R² = 0.6667, and a 10-fold R² = 0.4201 (Figure S9 in Supplementary Materials).

The models demonstrate associations between terpenes and THC, with reviewer proportions indicating that a strain helps with mood-related disorders, anxiety and depression. THC is the primary psychoactive compound in cannabis, which imparts recreational enjoyment (Cash et al., 2020), but also imparts medicinal benefits (Szejko et al., 2024). THC was included in both models within various terms of both positive and negative coefficients, indicating a complex relationship between THC concentration and predicted reviewer proportions for being helpful with anxiety and depression. Clinical studies with cannabis and THC seem to reflect this complex relationship, with both positive and negative effects being reported for mood-related disorders (Mammen et al., 2018; Stanciu et al., 2021). Similarly, various volatiles and their crossterms are included within both models, indicating that a relationship is present between their relative contents and how helpful that strain is predicted to be for the indicated conditions. This is consistent with Gulluni et al. (2018), who found that low THC cannabis essential oil had brain activity relating to both anxiety and depression that was promising for the treatment of these conditions. Accordingly, the predictive models suggest that both cannabis volatiles and THC impact utility for the management of anxiety and depression.

Within the context of the predictive model for anxiety, both camphene and terpinolene were present only as individual terms with positive coefficients, indicating that increases in these terpenes correlate with increased reviewer proportions for being helpful for anxiety (Table S15 in Supplementary Materials). De Sousa et al. (2015) conducted a systematic review of the anxiolytic properties of various terpenes, while Nuutinen (2018) reviewed the medicinal effects of cannabis terpenes specifically, both of which do not include anxiolytic properties for camphene or terpinolene. Interestingly, Kamal et al. (2018) explored correlations between terpenes and anxiolytic efficacy, and whilst not statistically significant terpinolene was associated with reduced efficacy against anxiety, which is the opposite direction to the positive association observed in model 5. Accordingly, the positive association of these terpenes (camphene and terpinolene) with the predicted proportion of reviewers that find a strain helpful for anxiety is relatively novel and thus future work should explore these terpenes within the context of cannabis use for the treatment of anxiety.

For the model for depression, β-limonene and eucalyptol were present in terms with negative coefficients (Table S16 in Supplementary Materials), whilst β-ocimene had a positive association (coefficient). This negative association between β-limonene and eucalyptol is contrary to other studies wherein rodent models indicated reduced depression from limonene (Lorigooini et al., 2021), limonenedominant essential oil (Zhang et al., 2019) and eucalyptol-dominant essential oil (Kim et al., 2022). Similarly, as noted above, the positive correlation of the minimally studied β-ocimene with depression use may be a positive indicator for further research.

Predicted changes to the proportion of reviewers indicating that a strain helps with anxiety or depression for the BC flowers

Changes to the BC-grown flowers were further explored by application of the predictive models (5 and 6) to the changed volatile and THC profiles. The model prediction results (Table 3) indicated that BC treatment was associated with a reduction in the proportion of reviewers that are predicted to find the BC flowers helpful for anxiety (0.91-fold, p = 0.069), and an increase in the proportion finding BC flowers helpful for depression (1.10-fold, p = 0.031). THC is sometimes associated with negative anxiogenic effects (Tambaro and Bortolato, 2012; Niesink and van Laar, 2013; Lichenstein, 2022), and so the predicted reduction in being helpful for anxiety (less anxiolytic) along with the measured increase in THC contents (Table S11 in Supplementary Materials) may reflect this relationship between THC and anxiety. For cannabis use and depression, evidence suggests there is a bidirectional relationship between increased cannabis use and increased depression, however, the interaction of cannabinoids with the endocannabinoid receptor CB1 has prompted interest in the potential of cannabis as a therapeutic option for depression (Feingold and Weinstein, 2021; Langlois et al., 2021). The predicted increase in being helpful for depression for the BC-treatment flowers with increased THC, supports this potential relationship between cannabis use (having CB1 interaction) and depression treatment. Accordingly, growth with BC is associated with reduced utility of cannabis for the management of anxiety, but increased utility for the management of depression.

Impacts to roots

Noting that the biostimulant is applied to plant roots within the fertigation solution, the molecular changes within the root tissues represent the immediate response to the physical interaction with the biostimulant. Principle component analysis indicated that the root molecular profiles were significantly different between treatments for both their metabolites (Figure 5A) and phytohormones (Figure 5B), indicating that BC treatment had a relatively large impact on the plant-root biochemistry. Comparison of individual metabolites within the roots identified 71 peaks that were significantly changed between treatments, and a further 27 peaks that differed with marginal significance (with fold-changes ranging between 0.16-fold and 12.67-fold). Of the identified metabolites (Table S17 in Supplementary Materials), 6 metabolites significantly increased, 2 metabolites increased with marginal significance, 25 metabolites significantly reduced, and 14 metabolites reduced with marginal significance. Among the metabolites that were significantly altered by the BC treatment were the phenylpropanoid-associated antioxidant compounds caffeic acid 3-glucoside, 1-caffeoyl-beta-D-glucose, rutin, and SA, which indicate a change in secondary phenolic compound metabolism likely associated with induction of a defence response (Fürstenberg-Hägg et al., 2013; Jiao et al., 2018). Comparison of individual root phytohormones identified that GA4 was reduced with marginal significance by 0.86-fold (p = 0.061, Figure 5C), while methyl-IAA and SA were both significantly increased, by 2.82-fold (p < 0.001, Figure 5D) and 3.51-fold (p < 0.001, Figure 5E), respectively, while all other phytohormones were unchanged (Table S18 in Supplementary Materials). For root bioassay measures, a marginally significant 1.19-fold increase in chitinase activity was detected (p = 0.054, Figure 5F), while protein content, hydrogen peroxide content, and peroxidase activity were all unchanged (Table S19 in Supplementary Materials). The increase in the stress-hormone SA, along with the increased chitinase activity, which is a defence-related enzyme (Wise et al., 2020a), indicates the likely induction of a stress response (Zhong et al., 2021; Castillo-Sánchez et al., 2024). Combined with this, the unchanged jasmonate hormones (JA, JA-Ile, and OPDA) indicate that this is an SA-derived stress response (Kachroo and Kachroo, 2012). The prevalence of SA over JA induction within a stress response is similar to that of a biotrophic pathogen and symbiotic microbial responses (Gutjahr and Paszkowski, 2009). Furthermore, as auxins and jasmonates are known to inhibit one another (Saniewski et al., 2002), the increase in the auxin methyl-IAA may represent a dampening of JA to prioritise the SA-response. These molecular changes suggest that BC application to cuttings has induced an SA-driven defence response within the roots.

Figure 5.

Impacts on leaves

As compared to the roots, the leaf tissues are physically separate from the site of BC application, and so molecular changes within leaf tissues represent signal transduction from the root tissues following exposure. By contrast to the relatively substantial molecular changes that were observed in the roots, the leaves of the same cuttings exhibited fewer changes. No significant changes were identified in leaf metabolite profiles (Figure S10 in Supplementary Materials), however, comparison of individual leaf metabolites identified 37 peaks that were significantly different between treatment groups, and a further 53 peaks that were different with marginal significance (all increased between 2.15-fold and 7.25-fold). Of the metabolites that could be identified, there were 9 that were significantly increased from BC treatment, a further 21 that were increased with marginal significance, and no metabolites that were reduced with any significance (Table S20 in Supplementary Materials). As with the roots, leaf phenylpropanoid-related antioxidant compounds were significantly affected including 1-caffeoyl-beta-D-glucose, caffeic acid 3-glucoside, p-coumaric acid glucoside, genistein, p-coumaric acid glucoside, and N-feruloylglycine, as well as the flavonoids luteolin and luteolin 7-O-malonylglucoside, which indicate a change to phenolic metabolism which may be associated with the defence response (Falcone Ferreyra et al., 2012; Yadav et al., 2020). A comparison of individual phytohormones did not identify any significant differences between BC and control (Table S21 in Supplementary Materials). However, leaf phytohormone profiles were found to be significantly different (Figure S11 in Supplementary Materials), likely attributed to the on-average large differences of ABA (1.26-fold increased, p = 0.237), ICA (2.17-fold increased, p = 0.473), JA (1.52-fold increased, p = 0.177), JA-Ile (2.35-fold increased, p = 0.333), Methyl-JA (0.55-fold decreased, p = 0.244), and SA (1.19-fold increased, p = 0.447). Additionally, no significant differences were observed for any of the leaf measures of protein content, hydrogen peroxide content, peroxidase activity (not detected), or chitinase activity (Table S22 in Supplementary Materials). The lack of significant changes in individual phytohormones and bioactivities (including chitinase) of the leaves indicates that the defence response to the BC treatment was largely localised to the root tissues rather than systemic. However, as noted in Gamir et al. (2014), plant defence is hypothesised to occur in stages wherein the early stages relate to altered metabolism and priming for a stronger response involving phytohormones upon pathogen detection. Following priming and induction of a mild defence response, subsequent exposure results in a stronger response with associated improvements in stress tolerance and growth (Suwanchaikasem et al., 2023a, 2023b, 2024). Accordingly, the increase in leaf antioxidant metabolites – without changes to phytohormones – may reflect the priming of a defence response, such that BC-treated plants would be more resistant to subsequent exposure to a stressor.

Impacts on mature plants

As cannabis flower harvest occurs after several weeks of BC application, the plant’s physical and molecular state at the time of harvest represents the effects of sustained BC exposure and the underlying effects that resulted in the flowers that were produced. Growth of cannabis with BC did not significantly affect plant growth or yield (Table S23 in Supplementary Materials), as indicated by height (p = 0.510), stem thickness (p = 0.839), flower yield (p = 0.631), and flower water % (p = 0.947). Accordingly, BC treatment does not appear to be associated with changes to the primary metabolism, which is the driver of growth and yield (Coleman et al., 2010).

Sugar leafphytohormones were found to be unchanged from BC treatment, as indicated by no significant change to the profile comparison (Figure S12 in Supplementary Materials), and no significant change to any individual phytohormones (Table S24 in Supplementary Materials). This is consistent with results from BC application to cuttings, wherein phytohormone changes were not observed within the aerial portion of the plant. Sugar leaf bioassay results (Table S25 in Supplementary Materials) indicated that protein content was increased by 1.01-fold with marginal significance (p = 0.052), and peroxidase activity increased from 0 in control to 0.027 △Abs₄₇₀ · min⁻¹ · mg⁻¹ total protein in BC plants (p = 0.016). As peroxidase enzymes are associated with defence (Bolwell and Wojtaszek, 1997; Almagro et al., 2009; Dieng et al., 2011), this increase in peroxidase activity further supports the hypothesis that BC application induced a defence response within the cannabis plants.

In addition to impacting therapeutic potential, the sugar leaf metabolite changes (Table S14 in Supplementary Materials), may also provide insight into molecular drivers of induced mechanisms. Wise et al. (2024d) explored sugar leaf metabolite expression following growth with a biostimulant molasses, which is a component of the BC applied herein. Within both studies, sugar leaf metabolites were observed to reduce from the biostimulant treatment, which included four metabolites – cannabisin B, SA, trans-p-Sinapoyl-β-D-glucopyranoside, and 1-Hydroxy-3,5,6-trimethoxy-10-methyl-9(10H)-acridinone (Wise et al., 2024d). The reductions to these metabolites from molasses application were part of a larger set of biochemical and metabolite changes, which indicated induction of a defence response. However, the shared reductions in the phenolic-like compound cannabisin B (Di Palo et al., 2022), the phenolic antioxidant trans-p-Sinapoyl-Î2-D-glucopyranoside (Materska and Perucka, 2005), and the defence-related phytohormone SA (Kachroo and Kachroo, 2012) indicate that molasses and BC (which contains molasses) may have shared modes of action relating to defence induction, attributed to the provision of molasses.

The BC treatment also significantly reduced the sugar leaf metabolite 2-C-methylerythritol 4-phosphate (aka methyl-D-erythritol phosphate, MEP), which is part of the MEP pathway and is involved in a range of processes including abiotic stress signalling (Banerjee and Sharkey, 2014). An output of the MEP pathway in plants is the 5-carbon isoprenoid isopentenyl pyrophosphate (IPP), which is the precursor to carotenoids (Saladié et al., 2014), terpenes (Frank and Groll, 2017), and in cannabis, cannabinoids (Desaulniers Brousseau et al., 2021). Of the 10 volatiles which were changed following the BC treatment (Table 1), 9 were terpenes (including mono- and sesqui-terpenes) and 1 was a non-terpene (p-Cymen-8-ol). The decrease in MEP and terpenes, along with the increase in THC (a cannabinoid) in the aerial tissues may therefore represent a stress-related diversion of resources within the MEP pathway towards increased THC production at the cost of terpene synthesis. Accordingly, based on the molecular changes within the different tissues, BC application appears to induce an initially localised defence response within the roots, with minimal translocation of this signal to the leaves, but achieves systemic defence induction after several weeks of application within a solid substrate, as evidenced by translocation of this defence signal to the aerial parts of mature cannabis plants.