Lebanese Cannabis: Agronomic and Essential Oil Characteristics as Affected by Sowing Date and Irrigation Practice

: A ﬁeld experiment was carried out in Lebanon to assess the agronomic and essential oil characteristics of cannabis as affected by sowing date and irrigation practice. The experiment consisted of a split-plot design with the water regime being the main factor (I opt -irrigated when the readily available soil water is depleted; I 50 - receiving 50% of the irrigation amounts in I opt treatments) and sowing date as the sub-plot factor (mid-April; end of April; mid-May). Biometric and seed quality parameters of the cannabis crop were determined. The essential oils (EO) of the inﬂorescence were subjected to a multivariate analysis such as principal component analysis (PCA) and hierarchical cluster analysis (HCA). The obtained results revealed that the aboveground fresh biomass, the dry matter, and the plant height were 55.08%, 59.62%, and 43.11% higher in I opt than in I 50 , respectively. However, the EO content was neither statistically affected by the irrigation regime nor by the sowing date. Under early sowing, both the water-use efﬁciency (WUE) for biomass and the EO production reached their highest values. All treatments presented a similar seed composition except that the crude fat and crude protein content were more elevated in I opt than in I 50 treatments. The main extracted essential oils in cannabis inﬂorescence corresponded to twenty-six identiﬁed compounds representing 79.34% of the monoterpenes and 81.25% of the sesquiterpenes. The monoterpenes were highly correlated with the irrigation treatment and early-April sowing while the sesquiterpenes were better enhanced under I 50 and end of April to mid-May sowing. The study reveals that agronomic practices lead to differential responses of pharmacologically useful plant compounds for improved health beneﬁts. Further research is required to clarify the potential for cannabis cultivation in Lebanon.


Introduction
Cannabis (Cannabis sativa L.) is an important herbaceous species native to Central Asia [1] and its use by humans as a food source [2], bioenergy [3], fiber production [4], cosmetics [5], and medicinal preparations [6] has spread around the world for millennia. Cannabis is an annual, dioecious plant [7]. Three hemp species are widely recognized: The soil has a loam texture, a mean pH of 7.39 ± 0.15, an EC of 0.12 ± 0.02 dS.m −1 , and an OM content of 2.14 ± 0.30%. The average soil water holding capacity is 127 mm/m.
The climate of the study area is characterized by a hot and dry season from April to October. The main weather parameters were collected from a standard agrometeorological station located at the experimental station of the Lebanese Agriculture Research Institute (LARI). Figure 1 shows the meteorological regimes for the 2020 season's baseline (ETo), (P), (Tmax), and (Tmin).
1049 m above sea level), specifically in Kferden village, during the summer season of 20 The soil has a loam texture, a mean pH of 7.39 ± 0.15, an EC of 0.12 ± 0.02 dS.m −1 , and OM content of 2.14 ± 0.30%. The average soil water holding capacity is 127 mm/m. The climate of the study area is characterized by a hot and dry season from April October. The main weather parameters were collected from a standard agrometeorolo cal station located at the experimental station of the Lebanese Agriculture Research In tute (LARI). Figure 1 shows the meteorological regimes for the 2020 season's basel (ETo), (P), (Tmax), and (Tmin).
In general, the average Tmax and Tmin from April to October 2020 were 31.9 a 13.4 °C, respectively. The total precipitation amount was 42.40 mm. These data are agreement with the historical weather data for the study area as shown in the Supplem tary Materials (Table S1).

Management of Crop and Experimental Design
The experiment was established as a split-plot design with the water regime as main factor (Iopt-irrigated when the readily available soil water is depleted; I50-receiv 50% of the irrigation amounts in Iopt treatments) and sowing date (SD) as the sub-plot f tor (mid-April -SD1-; end of April -SD2-; mid-May -SD3). In total, there were six tre ments and three replicates per treatment, which gave 18 plots. Each plot had a size of 2 × 2m, and cannabis seeds were sown in situ, 3 to 4 cm deep in rows 20 cm apart. Bef sowing, the soil was prepared by following the local farming practices in the region. particular, the soil was plowed at a depth of 30 cm in the autumn season, while in spring, it was leveled once with a double disc halo and once with a cultivator just bef planting. Then, triple superphosphate was applied at 60 kg ha −1 . Herbicides and pesticid were not applied during the growing season. Harvest took place in September.
All the plots were equipped with a drip irrigation system. The driplines were ma using low polyethylene surface laterals with external diameters of 16 mm having inl drippers with a discharge rate of 4 L h −1 . The distance between the inline drippers was cm. The spacing between laterals was 40 cm. Each plot had a separate shut-off valve.
Irrigation management was applied by checking the conditions of the weather. T soil moisture balance of the active root zone (40 cm) was considered. Therefore, an Exc  In general, the average Tmax and Tmin from April to October 2020 were 31.9 and 13.4 • C, respectively. The total precipitation amount was 42.40 mm. These data are in agreement with the historical weather data for the study area as shown in the Supplementary Materials (Table S1).

Management of Crop and Experimental Design
The experiment was established as a split-plot design with the water regime as the main factor (I opt -irrigated when the readily available soil water is depleted; I 50 -receiving 50% of the irrigation amounts in I opt treatments) and sowing date (SD) as the sub-plot factor (mid-April -SD1-; end of April -SD2-; mid-May -SD3). In total, there were six treatments and three replicates per treatment, which gave 18 plots. Each plot had a size of 2 m × 2 m, and cannabis seeds were sown in situ, 3 to 4 cm deep in rows 20 cm apart. Before sowing, the soil was prepared by following the local farming practices in the region. In particular, the soil was plowed at a depth of 30 cm in the autumn season, while in the spring, it was leveled once with a double disc halo and once with a cultivator just before planting. Then, triple superphosphate was applied at 60 kg ha −1 . Herbicides and pesticides were not applied during the growing season. Harvest took place in September.
All the plots were equipped with a drip irrigation system. The driplines were made using low polyethylene surface laterals with external diameters of 16 mm having inline drippers with a discharge rate of 4 L h −1 . The distance between the inline drippers was 20 cm. The spacing between laterals was 40 cm. Each plot had a separate shut-off valve.
Irrigation management was applied by checking the conditions of the weather. The soil moisture balance of the active root zone (40 cm) was considered. Therefore, an Excel-based irrigation tool was used to calculate irrigation volumes [41]. The tool considers weather, soil, and crop data for a daily estimation of the soil water balance. First, it calculates the reference evapotranspiration daily from weather data using the Penman-Monteith Equation (ETo) [42]. Then, the crop evapotranspiration (ET) was calculated on daily basis by multiplying the ETo with the crop coefficient (Kc) values of 0.5 from sowing to 3-4 pairs of leaves, 0.9 from 3-4 pairs of leaves to the appearance of male flowers, and 1.1 from the appearance of male flowers to the fruit ripening stage, as also adopted by [27][28][29][30][31][32][33][34][35][36][37][38][39][40][41] with hemp plants. The depletion factor of the readily available water was set up as 0.66 [31]. Irrigation began when the readily available water in the 40 cm soil layer was completely depleted. I opt plots were replenished to the field capacity level while I 50 plots received half of the water supplied to I opt .
Irrigation was stopped at the end of the fruit ripening stage. The total net irrigation amounts for I opt mid-April , I opt end of April , and I opt mid-May were 623, 644, and 681 mm, respectively. I 50 treatments received 50% of those quantities.

Field Measurements and Sample Collection
In the field, the plants inside a 1 m × 1 m frame were manually cut at the base of the stem from each plot and placed in paper bags. In the laboratory, the plant height and fresh weight were determined. Then, some plants were oven-dried at 70 • C to a constant weight to measure dry weight, while the inflorescences of other plants were dried in the shade at room temperature (25 • C) for 3 days and powdered before hydrodistillation for essential oil extraction. Seeds were also separated from some inflorescence and then dried (at 25 • C) and cleaned.
The water-use efficiency (WUE) in terms of biomass and also essential oil production of the flowers were calculated by considering the ratio of those parameters over the total net irrigation amount.

2.
Mass spectrometric conditions The MS (Agilent 5975B) was set from 40 to 450 amu; the ionization energy was set to 70 eV; the ion source temperature was set to 230 • C; the quadrupole temperature was set to 150 • C; the solvent delay was set to 4 min and the transfer line temperature was set to 280 • C. Software (NIST 2.0d; National Institute of Standards and Technology Standard Reference Data Program distributed by Agilent Technologies Germany) was used to help identify compounds.

Identification of constituents
The identification of constituents was performed based on retention indices (RI) determined from a homologous series of n-alkanes (C4-C30) under identical experimental conditions, with co-injection with either standard (Sigma-Aldrich, St. Louis, MO, USA) or known essential oil constituents, MS library search (NIST 05), and by comparing with MS literature data [45]. The used retention indices are shown in the Supplementary Materials (Table S2).

Seed Composition
Seed samples collected from each plot were used for the determination of the dry matter content, the ash content, the crude fat, the crude protein, and the crude fiber. All the analysis methods were according to AOAC [46].

1.
The dry matter and ash content The dry matter was determined by drying 1 g of the ground sample in the oven at 105 • C until reaching a constant weight. The ash content was analyzed in a 3 g seed sample that was dried in the oven at 500 • C for 3 h.

2.
The crude protein The crude protein was determined by weighing 0.5 g of the seed sample. 3.5 g of the catalytic mix and 8.5 mL of H 2 SO 4 were added and heated for about 90 min. 30 mL of H 3 BO 3 and 2 drops of the indicator were also added. The sample was transferred into the distiller and NaOH was poured into the boiling chamber. The beaker was held under the distiller and collected not less than 20 mL. Then, 0.1 N HCl was poured into the sample. The spent amount of 0.1 N HCl was noted.

3.
The crude fat The crude fat was determined by weighing 5 g of the sample. 70 mL of petroleum ether was added, then put into the digester extractor and heated up until the sample started boiling. The beaker was left to evaporate at room temperature, then placed in the oven. Finally, it was left in a desiccator until a constant weight was achieved.

The Crude fiber
Two grams of the defatted sample were treated sequentially with boiling 0.26 N H 2 SO 4 solution and 0.23 N KOH. The residue was then filtered off, washed, and transferred to a crucible, then posed in a controlled oven at 105 • C for 24 h. The crucible with the sample was weighed and calcined in a muffle at 500 • C and weighed.

Statistical Analysis
According to the Kolmogorov-Smirnov test, all dependent variables were provisionally evaluated for a normal distribution. The field experimental plot was set up in a whole-plot factor split-plot design with three replicates in a randomized complete block.
The least significant difference (LSD) was calculated to test the significance of the difference between the means. Analysis of variance was performed using SAS University Edition (Cary, NC, USA). Essential oil compositions (monoterpenes and sesquiterpenes) underwent principal component analysis (PCA) to examine relationships between variables and treatments. The PCA was based on the Pearson correlation matrix. The PCA results included the factor loading of a variable and a given principal component (PC). This analysis was performed using the software package FactoMineR [47] in the software R Studio [48].
The hierarchical cluster analysis (HCA) with a single linkage method that used Euclidean distances for the essential oil compositions (monoterpenes and sesquiterpenes) was performed using the clValid package [49] in the R studio software [47]. All packages used in the statistical analysis are available through the Comprehensive R Archive Network (CRAN, https://cran.r-project.org, accessed on 1 October 2020). Table 1 shows the means of the biometric, quality parameters, and WUE of cannabis as affected by water regime and sowing date during the growing season. If we consider the source of variance is the water regime, there was a significant difference among the treatments in terms of aboveground biomass production and plant height, with higher values in the fully irrigated treatments (I opt ) compared to those that received 50% levels of irrigation water (I 50 ). The aboveground fresh biomass, the dry biomass, and the plant height were 55.08%, 59.62%, and 43.11% higher in I opt than in I 50 , respectively. Biomass-WUE was the highest under I 50 . However, there was no statistical significance between I opt and I 50 for the essential oil content, although the treatments under deficit irrigation showed a 22.20% higher oil yield than those under full irrigation. The WUE for oil production was also not dependent upon the water regime.

Biometric, Productive Parameters and WUE
If we consider the source of variance as the sowing date, there was a significant difference among the treatments in terms of aboveground biomass production and plant height, with the highest values in the treatments established in mid-April and the lowest values in those sown in mid-May. The essential oil content was not statistically influenced by the sowing date but the WUE for oil production was significantly higher under early sowing. Table 2 shows the means of the main essential oils extracted from the cannabis inflorescence as affected by the water regime and sowing date during the trial period in season 2020.

Essential Oil Composition of the Inflorescence
The main extracted essential oils in cannabis inflorescence corresponded to twentysix identified compounds representing 79.34% of the monoterpenes and 81.25% of the sesquiterpenes. The main monoterpenes were the β-pinene, (8.87% in I opt and 14.2% in I 50 ), β-myrcene (9.96% in I opt and 14.81% in I 50 ), and D-limonene (8.46% in I opt and 10.86% in I 50 ). It should be noted that most of the identified compounds were not affected by the irrigation regime. Only the B-pinene and the β-caryophyllene resulted in significantly higher in the I 50 treatment than in the I opt , while the 1.8-cineol and the cis-α-bergamotene were significantly higher in the I opt treatment than in the I 50. Considering the sowing date as the source of variance, results revealed that B-pinene, D-limonene, β-caryophyllene levels were significantly higher in the treatments sown by the end of April (12.19, 10.86 and 16.45%, respectively), while the β-myrcene, α-phelandrene, borneol, and γ-cadinene were significantly higher in the treatments sown by mid-April (14.83, 2.69, 1.65 and 0.7%, respectively), and the β-ocimene, γ-terpinene,α-caryophyllene were significantly higher in the mid-May-sown treatments (7.45, 1.6 and 13.25%, respectively). Finally, a multivariate statistical analysis was conducted to better understand the combined effect of the water regime and sowing date.

Multivariate Statistical Analysis
The entire data set was analyzed by multivariate statistical analysis (PCA and HCA) to provide a thorough overview of the essential oil composition of the cannabis, notably the monoterpenes and sesquiterpenes, in response to the water regime and sowing time. The first three principal components (PCs) had eigenvalues greater than one which explained 81.25% and 79.34% of the cumulative variance for monoterpenes and sesquiterpenes compositions of the essential oils extracted, respectively (Tables 3 and 4). PC1 (first component) accounted for 45.84% and 36.70%, while PC2 (second component) accounted for 21.13% and 24.65% of the cumulative variance of the monoterpenes and sesquiterpenes compositions, respectively (Figures 2b and 3b). For monoterpenes compositions, PC1 was positively and strongly correlated (>0.6) with increased α-thujene %, α-fenchol, camphene %, borneol %, 1,8-cineol %, and terpinolene %. PC2 was positively correlated with increased δ-3-carene% and β-ocimene % (Table 3). In sesquiterpenes, PC1 was positively correlated with aromadandrene %, α-bisabolol %, caryophyllene oxide %, and γ-selinene %, whereas PC2 was significantly correlated with only α-farnescene % and γ-cadinene % (Table 4).  The factor load in bold shows the characters that are most relevant to each principal component. The factor load in bold shows the characters that are most relevant to each principal component.
The three combined treatments (I 50 -SD1, I 50 -SD2 and I opt -SD2) caused similar responses in the monoterpene's quality parameters, and were, therefore, grouped into one cluster: Group 1. Group 2 included two combined treatments (I 50 -SD3 and I opt -SD3). The combined treatment I opt -SD1 was the most distinctive, creating a group of its own, i.e., Group 3.
The positive side of PC1 for sesquiterpenes (Figure 3b), in particular the upper right quadrant (A), included the combined treatment water supply regime (I 50 ) with the sowing date SD2 (end of April) and was characterized by high aromadandrene %, α-bisabolol %, caryophyllene oxide %, γ-selinene %, α-fremescence %, and γ-cadinene %. The combined treatments I 50 -SD1, I 50 -SD3, and I opt -SD3 from the lower right quadrant (D) were characterized by high aromadandrene %, α-bisabolol %, caryophyllene oxide %, and γ-selinene %. The water supply regime (I opt ) with the sowing date SD1 (mid-April) treatment from the upper left quadrant (B) was characterized by high β-caryophyllene %, α-fremescence %, and γ-cadinene %. Finally, the combined treatment I opt -SD2 (I opt grown at the end of April) from the lower left quadrant (C) was characterized by high β-caryophyllene %. Figure 2a shows the dendrogram generated by the HCA of dissimilarities among the combined treatment on their Euclidean distances for the sesquiterpenes compositions of the extracted essential oils. The HCA showed three major treatment groups (Figure 3a). The three combined treatments (I 50 -SD1, I 50 -SD2, and I opt -SD3) caused similar responses in the sesquiterpenes quality parameters and were, therefore, grouped into one cluster: Group 1. The combined treatment I 50 -SD3 was the most distinctive, creating a group of its own, i.e., Group 2. Group 3 included two combined treatments (I opt -SD1 and I opt -SD2). Table 5 shows the means of the seed quality of cannabis as affected by the water regime and sowing date during the trial period. If we consider the source of the variance as the water regime, all treatments presented a similar seed composition except that the crude fat and crude protein content were higher in I opt than in I 50 treatments. If we consider the source of variance as the sowing date, only the crude fat was significantly higher in the treatments that were planted in mid-April.

Discussion
The study showed that the biometric and oil characteristics of cannabis could be influenced by agronomic practices.
Irrigation practice, in particular, affected cannabis biomass, plant height, and WUE. The results agree with the findings of [31] who reported that water stress could reduce the aboveground biomass production of cannabis, and of [24] who showed that plant height is positively correlated to plant density. Other studies reported that irrigation significantly affected the yield of fresh stems, fresh leaves, flowers, and plant height [41].
Sowing timing also affected the biomass and essential oil WUEs that were the highest under early sowing. These results agree with the findings of [27] who evidenced that cannabis cultivars grown in April resulted in better biometrics, productive characteristics, and WUE. Other works similarly observed that the optimal sowing date was between the last week of April and mid-May [31]. The main compounds of the essential oil found in our study agreed with the findings of other authors in the literature. Specifically, β-pinene was the main extracted oil from cannabis [50]. For the sesquiterpenes, the main extracted compounds were β-caryophyllene (9.53% in Iopt and 19.01% in I50), α-caryophyllene (8.01% in Iopt and 13.25% in I50) and caryophyllene oxide (4.42% in Iopt and 5.32% in I50). Other studies [25] found also that β-caryophyllene was the main compound in cannabis. The results confirm that the terpenoids in the plant vary according to numerous parameters including the variety of cannabis, the plant part, the environmental conditions, and the maturity stage of the plant [51,52].
It is important to highlight that the considered agronomic practices, irrigation, and sowing time had mainly affected the terpenoid type rather than the essential oil yield. The monoterpenes were highly correlated with the irrigation treatment I opt and early-April sowing, while the sesquiterpenes were better enhanced under I 50 and end of April to mid-May sowing. In the literature, there is some evidence of the accumulation of terpenoids in response to drought conditions (e.g., [53]) in several medicinal plants. In fact, both the quantity and the quality of some specialized metabolites, such as terpenoids, can be strongly affected by environmental stresses [35]. Some studies showed that the selection of the appropriate irrigation water regime or drought stress could influence the levels of different plant compounds for improved health benefits. For example, the increase in drought stress duration was reported to enhance the concentrations of plant phenolic and flavonoid compounds [54]. The study of [33] showed that the total phenolic and flavonoid contents were elevated under drought stress treatment, while antioxidants responded inconsistently to stress. Moderate water stress, coupled with the use of biostimulants was reported to enhance specialized metabolites like EO components, but also yield [34]. Imposed waterlimited stress has led to differential responses of pharmacologically useful diterpenoids for the obtention of the desired composition [35]. Finally, the results confirm that the interfaces in the production and synthesis of fatty acids in plants are influenced by variations in temperature, light, moisture amount, and farming conditions, as reported by [50][51][52][53][54][55].

Conclusions
This work presented the first evidence of the agronomic and essential oil characteristics of Lebanese cannabis as affected by the sowing date and irrigation regime. According to the findings, agronomic practices are closely linked to the quality characteristics of the main cannabis products. Within this context, further studies are needed to recommend the appropriate planting practices for the legalization framework of this crop in Lebanon.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/w14233842/s1, Table S1: The monthly maximum temperature (T max), minimum temperature (T min), rain and evapo-transpiration (ETo) for the Kferden region for the 2020 season (April to October) compared to the historical monthly data (2010-2020).; Table S2: The retention indices used for the identification of the essential oil components.