Cannabis Varieties Can Be Distinguished by Achene Shape Using Geometric Morphometrics

Abstract

Introduction:

Cannabis plant uses are widespread across human cultures. The current tendency is to classify Cannabis varieties into chemovars upon their chemical fingerprint, mainly cannabinoids and terpenoids content. The identification of chemovars has important medical implications; however, their pharmacological characterization is costly and time consuming. The goal of this study was to assess whether achene shape variation could be related to Cannabis varieties with contrasting cannabinoid concentrations, as a first approach to chemovar identification.

Methods:

We used two-dimensional geometric morphometrics (GM) of the achenes and multivariate statistical analysis. We used achenes from five varieties, two from Type II chemotype (expressing both tetrahydrocannabinol [THC] and cannabidiol [CBD]), two Type I (THC-only), and one Type III (CBD-only).

Results:

The achenes from the different chemotypes were clearly distinguishable. No significant differences between varieties from the same chemotype were observed. The varieties with high THC concentration (Type I) were rounded and bigger, whereas achene from varieties containing only CBD (Type III) had a slender shape with smaller size.

Conclusion:

Achene shape variation is a potential biomarker of cannabinoid content in the plant flowers. Further studies are needed to confirm the suitability of GM methods for high-throughput screening of Cannabis cultivars, including larger diversity of varieties, and taking into account growth conditions, which can also influence plant chemical fingerprint.

Introduction

Cannabis plants have been used by humankind for thousands of years, and proof of its medicinal use are widespread across cultures.¹ Its domestication/use can be traced back thanks to archaeological sites in Central Asia dating at least 10,000 years ago. Owing to its adoption by humans, it subsequently spread through Eurasia and all over the world.¹ The utility of this plant exceeds that of therapeutics, being widely used as a source of fiber and food. Of special interest is the use of Cannabis “seeds” or achenes (one-seeded dry fruits in which the pericarp is not as tightly joined to the seed) as a source of protein and unsaturated fatty acids, compounds for which they are considered of high nutritional value.²

There is now scientific consensus on the fact that Cannabis sativa L. is one single species that can be classified into two subspecies (Cannabis sativa ssp. sativa and ssp. indica); however, many of the Cannabis plants used for medical and recreational uses are hybrids. Another classification used is “drug-type” versus “fiber-type” Cannabis; however, this grouping does not relate to any scientific criterion, being this classification mostly useful in a legal context.³ Under this approach, the strains with the highest amount of fiber and low content of psychoactive cannabinoids are distinguished from the medicinal types, and these are in turn classified as varieties with high tetrahydrocannabinol (THC) content, or containing also other cannabinoids such as cannabidiol (CBD).^3,4 Although the most commonly known are the THC-predominant varieties (named as Type I), the other emergent varieties (Type II, with similar concentrations of THC:CBD; Type III, CBD-predominant) are also considered as of high therapeutic value.³

The current tendency is to classify Cannabis into chemovars, according to the complex fingerprint of potentially active chemical constituents, mainly cannabinoids and terpenoids.⁵ The classification into chemovars has important practical implications, as not only the cannabinoid but also terpenoid profiles have correlation with synergistic therapeutic effects.⁵ Although pharmacological chemovars characterization has proven useful for classification of Cannabis plants, it is costly and time consuming as it implies analysis of many compounds (e.g., by high-pressure liquid chromatography) and a sufficient number of samples. For this reason, any alternative method to infer and/or predict chemovar and its related cannabinoid content departing from external features deserves careful attention, as it could save considerable analysis time and cost in certain conditions.

Up to now, it is not clear whether chemical characteristics can be related to the plant morphology. As early as in 1974, it was recognized that Cannabis subspecies had different achene shapes.⁶ Other studies have been made on traits such as cotyledon asymmetry.⁷ More recently, achenes were analyzed in relation to their general biochemical characteristics, which allowed the authors to distinguish ruderal from domesticated plants, as well as psychoactive Cannabis (high-THC) versus hemp (very low-THC, for fiber and oilseed).⁸ In this context, it would be helpful to have the possibility to rapidly predict varieties or even chemovars by using the morphology of the achenes in the ever-growing Cannabis industry, before the effort of growing plants has been invested.

Geometric morphometrics (GM) is a multivariate approach that allows the exploration of shape and size variations in objects with a high level of detail.^9,10 GM, unlike the conventional linear-distance based morphometrics, enables a proper separation between size and shape variation.¹¹ Another advantage is the preservation of the geometric information throughout the statistical analyses, which allows the detailed and graphic visualization of both the magnitude and the direction of the morphometric changes.¹²

In the past two decades, the use of GM has increased in biological studies,^13–15 enhancing the general understanding of shape aspects related to genetic and nongenetic variation causes. Previous studies such as Van der Molen et al.¹⁶ and Trovant et al.¹⁷ were successful in using shell shape, as represented by contour deviations captured after landmark-based methods, to separate different mussels species. In contrast, Pollicelli et al.¹⁸ and Idaszkin et al.¹⁹ determined the use of leaf shape, in the field and laboratory experiments, respectively, as a biomarker of sediment pollution in halophyte plants.

The aim of this study was to assess whether achene shape variation can be a predictive factor related to classification in Cannabis varieties. For this, we used plants belonging to varieties with contrasting cannabinoid concentrations and analyzed their achene shape by GM.

Materials and Methods

Samples

The sample included a total of 91 achenes from five Cannabis varieties donated by local experienced cultivators from Argentina. Two varieties from THC-only (Type I) chemotype (CCK and PATAG, 24 and 23 achenes, respectively), two belonging to Type II chemotype expressing both THC and CBD (YCBDG and YCBDM, 12 and 10 achenes, respectively), and one from a Type III chemotype (Hemp, 22 achenes), were used.

Chemical analyses

The presence of cannabinoids in mother plants was detected as a rapid test by thin layer chromatography, using ethanol extractions of plant material and chloroform as mobile phase, and further revealed with FastBlue B dye. Quantification of cannabinoids was performed by high-pressure liquid chromatography of oil extracts prepared from the same specimens (Supplementary Fig. S1).

GM analysis

Each achene was carefully oriented with its abscission zone upward, with the fruit wall suture on the radicle side at the left. This step is critical, as the orientation allows the achene shape to be comparable across individuals. The plane of scanning was parallel to the achenes longitudinal section using an Epson Perfection v350 scanner with a 600 dpi resolution (Fig. 1).

FIG. 1.

External view (perianth) of the longitudinal plane of the Cannabis sp. achene showing the two-dimensional landmark configurations used to capture the outline shape. These landmarks are (1) inflexion point between abscission zone and fruit wall suture; (2) micropyle zone; (3) inflexion point between abscission zone and outline near the outer cotyledon; (4–8) semilandmarks placed equidistantly between landmarks 1 and 2, coping the suture outline; and (9–13) semilandmarks on the outline, near the outer cotyledon, between landmarks 2 and 3.

The digitization process of the landmark configurations on the achenes images was made by using the computerized software Thin Plate Spline (TPS).²⁰ Samples were digitized by one observer (F.M.) using two-dimensional (2D) landmarks and semilandmarks configuration (3 anatomical landmarks and 10 semilandmarks; Fig. 1), in the module TpsDig2.²¹ In brief, a landmark is a “fixed point” as it defines a discrete biological form, whereas a semilandmark is a special type of landmark employed to discretize variation in outlines. To homologate semilandmarks, we used a mathematical algorithm (“sliding” iterative process) that minimizes the bending energy of the TPS function,²² using the software TpsRelw.²³ The effects of rotation, translation, and scale were eliminated by Procrustes analysis²⁴ so that the pure shape information was preserved. Aligned specimens containing shape information were then exported to Morphoj software²⁵ to carry out the statistical analysis. The data set containing the co-ordinates of the aligned individuals was further subdivided using chemotype (I, II, or III) and variety as classifiers. The centroid size (CS)¹⁰ was calculated, and used as a proxy for achene size. To evaluate differences in achene size, Kruskal–Wallis test²⁶ was used since the assumptions for parametric tests were not achieved. For significant results (p<0.05) pairwise multiple comparisons test of subgroups was applied. To evaluate the putative pooled-within group allometric effect (change in the shape associated with size increment) a multivariate regression between shape scores (dependent variables) and size (CS, independent variable) was performed.^10,27 A principal component analysis based on the variance-covariance matrix was carried out to summarize the principal achene shape variation in a smaller set of variables (principal components). To describe the shape components that maximized the separation among Cannabis varieties, a canonical discriminant analysis was performed. Finally, we used a multiple comparison method based on cluster analysis generated using an unweighted pair group method with arithmetic mean, the Multivariate Di Rienzo, Guzmán, Casanoves Method (UPGMA-MDGC).^28,29 This method is a hybrid technique that combines a hierarchical clustering method based on Mahalanobis distances with the principle of hypothesis testing for multivariate cases. Based on inferential statistics, this method is successful in addressing the problem of determining the number of groups in hierarchical dendrogram analysis. The graphical output of the MDGC test is a useful tool since it shows a clear distinction between statistically different achene shapes from each chemovar as well as their relationships.

Results and Discussion

This research represents one of the few studies on the achene of Cannabis and is a pioneer in the application of GM for addressing changes in achene shape related to cannabinoids content. We were able to capture and describe shape changes on the Cannabis achene in the different varieties studied. Although harvesting time was not standardized in this study, CS of the achene differed significantly among chemovars (Kruskal–Wallis test: H=20.42, p<0.0004; Supplementary Fig. S2). The achenes from the chemovars with high THC concentration (Type I, CCK) were rounded and bigger, whereas achene containing only CBD (Type III, Hemp) had a slender shape with smaller size. The qualitative achene shape variation described by Emboden⁶ for C. sativa being large and round and C. indica small and laterally compressed was in agreement with our quantitative results since typically C. sativa is related to high THC levels and C. indica with high CBD levels.

We also report in this study an absence of allometric patterns: the pooled-within group multivariate regression being nonsignificant (p=0.12). Overall, 99% of the achene shape variations were summarized by the first 10 principal component scores. We used only the first 10 Principal Components (PCs) to avoid a bias from higher PCs representing only minor shape variation and to accomplish the rule of thumb of inferential statistical analysis, which indicates that the sample size of the smallest group (in this case CBDM variety, n=10) needs to exceed or at least equal the number of predictor variables. The Mahalanobis distances were nonsignificant within both Type I and Type II chemotypes, involving two different varieties each, and showed that each chemotype group was distinct. A higher similarity was observed between Type I and Type II chemotypes, whereas Type III was more divergent (Fig. 2).

FIG. 2.

UPGMA cluster showing the relationships among Cannabis chemotypes and their varieties and the diagrams of the reconstructed consensus configurations of each sampled group. The cutoff criterion (p=0.05) obtained with the MDGC test is indicated with a dashed horizontal line. Three statistically different groups were identified by this method.

Concerning the canonical discriminant analysis, the most conspicuous traits separating achene shapes along the CV1 (61.04% of the variance) were the suture expansion, along the radicle side, shortening the achene longitudinal plane, and the abscission zone expansion toward the negative values (rounded shape in CCK; Fig. 3).

FIG. 3.

Achene shapes differences among Cannabis chemotypes varieties. Canonical variate analysis of the outline shape variation along the first 2 canonical axes. Wireframe diagrams show shape changes from consensus shape (grey polygon) to the positive and negative extreme (scale factor ±6) in both axes. The 95% equal frequency ellipses are shown for each variety, and represent a probability of 0.95 for each point of falling within the ellipse.

Some plant studies using GM are focused on determining patterns of variation in the form at the inter- or intraspecific level, on a variety of anatomical structures. To the best of our knowledge, only Chemisquy et al.³⁰ used variations in seed shape to classify and discriminate varieties of Orchids, on their 2D landmark and semilandmark-based GM study.

Although more studies are necessary to demonstrate that our findings are applicable in general for the analysis of chemovars, the results shown in this study are promising. This method clearly distinguished hemp from the other chemotypes, and could be, therefore, used for detecting achenes with very low THC content, for example, for compliance with regulations regarding maximum THC content. Also, it is important to distinguish CBD-expressing chemovars in the context of developing cultivars for some medicinal use such as the treatment of epilepsy and other neurological disorders.^31,32 Moreover, other cannabinoids contributing to the chemovar classification, such as cannabigerol (CBG), cannabichromene (CBC), tetrahydrocannabivarin (THCV), as well as terpenoid profile could be also shaping these differences. Further studies involving complete chemical fingerprints should be carried out to explore this possibility.

In this study, we could observe variations in achene size (e.g., among the two Type I cultivars) that did not affect the grouping pattern. The use of GM as a robust, useful, practical, and low-cost tool allows the detection of small shape changes independently of the variations in size, strengthening the use of achene shape variations as a potential biomarker of cannabinoid content in the plant flowers.

In conclusion, we report in this study a consistent grouping of varieties of Cannabis and a clear distinction among chemotypes based on achene shape variation. We have to note, however, that only five achene groups were analyzed in this study, and that the cannabinoid content was only measured in mother plants. Further quantitative multivariate studies should be performed to assess whether the observed morpho-geometric variations are indeed correlated with cannabinoid and/or terpenoid content, and/or to other genetic or biochemical traits. Scanning C. sativa achenes by a morpho-geometric software represents a rapid high-throughput technique that could be applied when deciding the chemovar types to obtain or preserve, during development and stabilization of novel strains. This kind of technology could be used in the cannabis-related industry, whether medicinal or industrial, to select achenes with different cannabinoids composition before the massive production.

Abbreviations Used

2D

two-dimensional

CBD

cannabidiol

CS

centroid size

GM

geometric morphometrics

THC

tetrahydrocannabinol

TPS

thin plate spline

Author Disclosure Statement

No competing financial interests exist.

Funding Information

F.M., M.L., Y.L.I., R.G.J., and G.B. are staff members of CONICET, Argentina.

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2

Cite this article as: Márquez F, Lozada M, Idaszkin YL, González-José R, Bigatti G (2022) Cannabis varieties can be distinguished by achene shape using geometric morphometrics, Cannabis and Cannabinoid Research 7:4, 409–414, DOI: 10.1089/can.2020.0172.