Abstract
Keywords: building materials, Cannabis sativa ,
cannabinoids, crop breeding, flower development, flowering time, hemp,
marijuana, sex determination, sustainability
Introduction – Cannabis in a nutshell
Cannabis sativa is a highly versatile crop with dozens of
different uses (Figure 1). There are a multitude of medical applications
for Cannabis secondary compounds, which have been shown to
reduce pain, nausea and neurological conditions like seizures
(Whiting et al.,
2015), and research on effects on inflammation, depression and cancer
is also being conducted (Atalay et al., 2019; Fraguas-Sánchez and
Torres-Suárez, 2018; Poleszak et al., 2018; Śledziński et al., 2018,
Russo, 2011). Beyond
that, fibre varieties of Cannabis have high carbon sequestering
potential because of their rapid growth. They are therefore utilized for
carbon storage in building materials or as biofuel
(Finnan and Styles,
2013). For those different reasons, the Cannabis industry is
gaining more traction and the need for specialized varieties, adapted to
local climatic conditions, or suited for specific applications, is
steadily increasing.
Cannabis is probably best known for one secondary compound, the
psychotropic substance tetrahydrocannabinol (THC). Depending on the THC
content of the plant, or more specifically the dried inflorescence,Cannabis is either classified as marijuana (or drug-type, plants
above 0.3% THC) or hemp (fibre-type, below 0.3% THC), which is mainly
a legal and not a strict taxonomic classification. A more refined
classification of Cannabis according to the phytocannabinoid
profile into distinct ’chemotypes’ can also be useful, with chemotype I
and II being marijuana while chemotypes III, IV and V can be seen as
hemp (see chapter 3).
Many countries have been easing the ban on medical and even recreational
use of THC during the past decade. However, because of the prohibition
of Cannabis in many countries throughout the last century, it was
not bred to the same extent as other high-value crops. Hence, hemp and
marijuana lines retain a high level of genetic variability and
heterozygosity, that is not found in other crops
(Sawler et al.,
2015).
Here, we review the biology as well as the applications and future
perspectives of Cannabis research and breeding. We discussCannabis taxonomy and cannabinoid synthesis as well as flower
development and flowering time control with an emphasis on sex
determination in this predominantly dioecious species. We also summarize
the currently available genomics resources. Since Cannabis is so
versatile, we discuss its applications in medicine as well as in the
building industry. Cannabis ’ future role in a sustainable society
is summarized as well as the future of cannabinoid production via cell
suspension cultures.
Cannabis systematics
Cannabis is the botanical name of a genus that historically
includes three species, C. sativa , C. ruderalis andC. indica . However, since the three species can intercross, they
are also often considered one single species, C. sativa(Small, 2015).Recent genetic data support the single species concept and recommend
that three subspecies should be recognized: Cannabis sativa
s ubsp. sativa , subsp. indica and subsp. ruderalis(Q. Zhang et al.,
2018).
Cannabis is a dioecious species, meaning there are male and
female individuals (Figure 2a-c). However, through breeding, monoecious
lines with male and female flowers on the same plant have also been
generated (Figure 2d)
(Moliterni et al.,
2004).
The genus Cannabis is part of the Cannabaceae, a small family of
flowering plants with 10 genera and some 120 species
(Jin et al., 2020;
Yang et al., 2013). The Cannabaceae have been estimated to have
originated ca. 70 to 90 million years ago, and are distributed in
temperate and tropical regions throughout the world (Figure 3)
(Jin et al., 2020;
Magallón et al., 2015). Most species of the Cannabaceae are trees or
shrubs, Cannabis as a herb is therefore the exception rather than
the rule in the family. However, a trait Cannabis shares with
many other species in the family is the inconspicuous unisexual flowers
(Yang et al., 2013).
The closest relative of Cannabis is the genus Humulus(Yang et al., 2013),
which consists of three species, among which Humulus lupulus(hop) is economically important for the beer brewing industry. Both hop
and Cannabis produce separate male and female flowers, and the
trichomes in the female inflorescences are the site of secondary
compound production that make both of those plants economically valuable
(Page and Nagel,
2006).
Within the angiosperm phylogeny, Cannabaceae are most closely related to
the Moraceae (mulberry or fig family) and Urticaceae (nettle family).
Together with the Ulmaceae (elms and relatives) they form a group known
as the urticalean rosids (Figure
3) (Sytsma et al.,
2002). It is interesting to note that unisexual flowers appear to be
prevalent in the urticalean rosids, whereas bisexual flowers are by far
the dominant system in angiosperms in general
(Renner, 2014; Sytsma
et al., 2002). The evolution of sex expression and sex determination in
this group is an interesting area of future research.
The urticalean rosids belong to the order Rosales, which are eudicots
(The Angiosperm
Phylogeny Group, 2016). Though the Rosales comprise some 7700 species
(Zhang et al., 2011),
they contain relatively few well characterized model plants. The
flowering plant super-models Arabidopsis thaliana (thale cress,
Brassicales) and Oryza sativa (rice, monocots) are only distantly
related to Cannabis , the lineages leading to Arabidopsisand Cannabis separated some 120 million years ago, those leading
to rice and Cannabis some 130 to 140 million years ago (Figure 3)
(Magallón et al.,
2015). Among the relatively well characterized plants that are more
closely related to Cannabis are many Rosaceae species (rose
family, apple, peach and relatives), for which several well assembled
and annotated genomes exist
(Aranzana et al.,
2019; Zhang et al., 2019), the Cucurbitaceae (cucumber, pumpkin and
relatives), which serve as an important model for sex determination and
sex expression (Li et
al., 2019; Schilling et al., 2020a; Zheng et al., 2019) and Fabaceae
(bean family) for flowering time regulation
(Cao et al., 2017;
Schmutz et al., 2010) .
Cannabis sativa itself is phenotypically extremely diverse.Cannabis plants vary in numerous traits including height, leaf
shape, photoperiod response, tetrahydrocannabinol (THC) and cannabidiol
(CBD) content, plant architecture and sex expression
(Clarke and Merlin,
2016; Grassi and McPartland, 2017; Raman et al., 2017; Schilling et al.,
2020b). The dioecy of many Cannabis lines and thus the
relatively high levels of heterozygosity further contribute to the fact
that even within one cultivar the phenotypic diversity can be
substantial (our unpublished observations).
For breeders and farmers, the high level of genetic and phenotypic
diversity can be problematic, as a crop is usually best to handle when
it possesses a high degree of uniformity in the field. However, at the
same time, the existing diversity can be harnessed by breeders to
produce new lines for a multitude of different purposes. For plant
genetics research, the phenotypic and genetic diversity is a gold mine,
as it provides the possibility to study the genetic basis of many traits
in Cannabis . Some developments in this arena are outlined in the
subsequent chapters, but many more are sure to come.
Ever more complex: The genetics of phytocannabinoid
biosynthesis
One of the commercially most interesting and valuable products that can
be generated from Cannabis plants are phytocannabinoids. We use
the term phytocannabinoids here for plant derived cannabinoids, and to
distinguish them from synthetic cannabinoids or those produced by the
human endocannabinoid system. Phytocannabinoids are of great interest
for medical applications (see chapter 8 for a detailed discussion) as
well as commercial exploitations for recreational use. Hence, one of the
major breeding goals involves the accurate prediction and targeted
manipulation of phytocannabinoid profiles to ensure the optimal
combination of active components in plant extracts (see entourage effect
chapter 8) or legal compliance for non-psychoactive products.
While there are over 100 different phytocannabinoids described
(Pertwee, 2014),
three phytocannabinoids are usually at the centre of attention from a
medical and commercial perspective: cannabigerol (CBG), cannabidiol
(CBD) and tetrahydrocannabinol acid (THC) (Figure 4). Cannabisitself synthesizes phytocannabinoids in the carboxylated form with a
carboxylic acid group, i.e. as CBGA, CBDA and THCA. However, to be
active in the human endocannabinoid system, phytocannabinoids need to be
consumed in their decarboxylated forms, which are usually generated by
high temperature treatment (for example during smoking)
(Moreno-Sanz, 2016).
Phytocannabinoids are predominantly produced in female inflorescences,
more precisely they are secreted from trichomes of perigonal bracts,
subtending flowers, and leaves (‘sugar leaves’) within inflorescences.
However, in lower concentrations, phytocannabinoids can also be detected
in vegetative leaves at certain times during the growth period
(Aizpurua-Olaizola et
al., 2016).
Among all phytocannabinoids, THC is the major psychotropic one. However,
chemically all molecules mentioned above are very similar in structure
and are produced from the same precursor molecules (Figure 4). CBDA and
THCA are biochemically synthesized by two closely related enzymes, CBDA
and THCA synthase
(Shoyama et al., 2012;
Taura et al., 1996). CBDA and THCA are both synthesized from CBGA,
while CBGA is synthesized from two non-cannabinoids, olivetolic acid and
geranyl pyrophosphate by a prenyltransferase
(Fellermeier and Zenk,
1998)(Figure 4). Cannabichromenic acid (CBCA) synthase converts CBGA to
CBCA (Morimoto et al.,
1997) and is closely related to THCA and CBDA synthase (Figure 5), but
the CBCA content of most mature Cannabis flowers is low
(de Meijer et al.,
2009a). Interestingly, CBDA synthase-like genes have been found in
other plants and fungi
(Aryal et al., 2019;
Vergara et al., 2019).
Cannabis plants can have very high levels of phytocannabinoids or
close to no phytocannabinoids at all, or anything in between
(Aizpurua-Olaizola et
al., 2016; de Meijer et al., 2009a). This has stipulated the
description of different chemotypes that are characterized by their
distinct phytocannabinoid profiles. The chemotypes are a very useful
concept for chemical classifications and for breeding programmes. It
should be kept in mind, however, that they do not necessarily constitute
a phylogenetic classification based on evolutionary relationships
(de Meijer et al.,
2009b; Small and Beckstead, 1973). Cannabis plants can roughly
be categorized into five different ‘chemotypes’ (Figure 4). Plants of
chemotype I (short ‘type I’) produce high levels of THCA and only low
levels of CBDA and CBGA
(Small and Beckstead,
1973). This means the ratio of THCA/CBDA is much larger than 1. In type
II Cannabis plants THCA and CBDA are both produced in
approximately equal amounts
(Small and Beckstead,
1973). Both, type I and type II plants, are usually classified as
‘marijuana’ and can underlie strong regulations, depending on the
country or jurisdiction. These plants are bred to produce up to 20 % of
their dry mass as phytocannabinoids.
In contrast, type III plants have high CBDA levels and low to very low
amounts of THCA.
Chemotype IV and V refer to Cannabis plants which have CBGA as
their dominant phytocannabinoid or very low levels of phytocannabinoids
overall, respectively
(de Meijer et al.,
2009a; de Meijer and Hammond, 2005)(Figure 4).
In addition to the five different chemotypes, also the hemp-marijuana
distinction is used to characterize different Cannabis plants
(Figure 4). If the THC/THCA content in the dry flower mass is below
0.2-1 %, these plants are usually categorized as hemp, above that as
marijuana (depending on the jurisdiction this threshold can
vary) (Brunetti et
al., 2020; Mead, 2017). The differentiation between hemp and marijuana
can typically also be drawn genetically, with hemp and marijuana
varieties forming two genetically distinct populations
(Sawler et al.,
2015). Further, hemp and marijuana can be phenotypically quite distinct
with marijuana plants generally being bushier and with a dense set of
inflorescences while hemp plants tend to be taller, less branched and
with less dense flower structures. However, there are also plants with
low THC/THCA content (type III) which strongly resemble marijuana in
overall plant and inflorescence architecture
(Grassa et al.,
2018). Hence, the terms hemp and marijuana do not necessarily always
refer to distinct genetic populations or phylogenetic categories. As the
critical distinction between hemp and marijuana is the THC/THCA content,
they can also be considered broader categories of chemotypes.
The underlying genetics of the different chemotypes have been studied in
quite some detail in the last two decades
(de Meijer et al.,
2009a, 2009b, 2003; de Meijer and Hammond, 2005; Pacifico et al., 2006;
Toth et al., 2020; Weiblen et al., 2015; Welling et al., 2016).
However, the complex nature of the Cannabis genome with its many
transposable elements, low complexity regions and high heterozygosity
have made a conclusive analysis of the loci controlling phytocannabinoid
production challenging
(Grassa et al., 2018;
Laverty et al., 2019; McKernan et al., 2018).
Different genetic loci had been postulated which determine a plant’s
chemotype, they are encoding for the different types of synthases: at
locus B two codominant alleles were hypothesized to exist, the allele
BT encodes for the THCA synthase, BD for
the CBDA synthase (Figure
4)(de Meijer et al.,
2003). Depending on the presence of either or both loci, the plant will
be chemotype I (BT/BT), chemotype II
(BT/BD) or chemotype III
(BD/BD)
(de Meijer et al.,
2003; Toth et al., 2020; Welling et al., 2016). Additionally,
non-functional alleles of the synthase gene (B0) are
predicted to be associated with chemotype IV, where neither CBDA nor
THCA are produced and the precursor, CBGA, accumulates (Figure 4)
(de Meijer and
Hammond, 2005; Onofri et al., 2015; Welling et al., 2016).
Further, according to this model, CBCA synthase is encoded by an
independent locus (C) while another independent locus (O) is relevant
for precursor production, with a knockout resulting in overall minimal
phytocannabinoid levels (Figure 4)
(de Meijer et al.,
2009a, 2009b).
The genetic basis of the chemotypes was analysed in detail by producing
a cross between high-THC Purple Kush (chemotype I) and low-THC Finola
(chemotype III). This resulted in an F1 generation of mainly type II
plants, producing both, THCA as well as CBDA
(Weiblen et al.,
2015). This confirmed earlier findings of crosses between type I and
type II plants, resulting in intermediate type II individuals
(de Meijer et al.,
2003). The segregation pattern of phytocannabinoid profiles in the F2
generation pointed towards a Mendelian inheritance pattern: type I, type
II and type III plants were all observed in the F2 generation with the
expected distribution of 1:2:1
(de Meijer et al.,
2003; Weiblen et al., 2015). A correlation of the expression of either
THCA or CDBA synthase with the respective chemotype was also observed
and the THCAS/CBDAS locus could be mapped
(Weiblen et al.,
2015).
However, although these findings were consistent with the idea of
codominant alleles at one single locus, it became apparent that the
situation is more complex
(Grassa et al., 2018;
Laverty et al., 2019; Weiblen et al., 2015). New draft genomes
generated with third generation sequencing technology indicated that the
THCA and CBDA synthases do not seem to be encoded by alleles of one and
the same gene, but rather by distinct loci in marijuana and hemp,
respectively, without a clear counterpart in the other genome
(Grassa et al., 2018;
Laverty et al., 2019). Sequencing of the hemp cultivar ‘Finola’ and the
marijuana cultivar ‘Purple Kush’ indicates that a functional CBDA
synthase gene is present only in in the ‘Finola’ genome while the
‘Purple Kush’ genome only encodes for a functional THCA synthase
(Laverty et al.,
2019). While mapping to approximately the same region in both genomes,
the DNA sequences surrounding the respective synthase genes are
drastically different from each other. Further, a low albeit still
detectable recombination rate between the two loci supports the notion
that they are genetically distinct
(Laverty et al.,
2019). The sequencing of a different Cannabis variety (‘CBDRx’),
which is a chemotype III hemp-marijuana hybrid revealed an even more
complex genomic arrangement with a number of pseudo- and functional
synthase genes in three different cassettes on the same chromosome
(Figure 5) (Grassa et
al., 2018).
The CBDA and THCA synthase genes themselves seem to be embedded in
cassettes of multiple tandem duplications of putatively non-functional
synthase genes, which are regularly interspersed with long terminal
repeat (LTR) retrotransposons, making the assembly and analysis of these
loci even more challenging (Figure 5)
(Grassa et al., 2018;
Laverty et al., 2019). This is also the reason why these complex loci
could not be resolved in the first published Cannabis genome,
which relied on short-read sequencing data
(van Bakel et al.,
2011). This genomic constitution, where the difference between
marijuana and hemp comes down to a large structural variation is, if
true, very unusual. Hence, the aforementioned locus “B” with its
different alleles might look very different from what was previously
assumed to be simple isoforms of a single gene.
The complexity of phytocannabinoid synthases does not end there, though.
Copy number variation of CBDA and THCA synthase genes might be involved
in phytocannabinoid level and composition
(Vergara et al.,
2019) and most likely, the number of synthase (pseudo)genes might be
different for each cultivar sequenced
(Grassa et al., 2018;
Laverty et al., 2019;
McKernan et al.,
2020).
High throughput assays for BT and BDmarkers have been developed and show that many plants actually contain
both loci (Cascini et
al., 2019; McKernan et al., 2020; Toth et al., 2020). Moreover, many
BD/BD plants, especially those with
higher CBDA levels, have THCA levels of above 0.3 % of dry flower mass,
despite the absence of a functional BT allele
(Toth et al., 2020).
This residual THCA is probably at least to some extent a by-product of
the CBDA synthase itself. The THCA and CBDA synthase have a relatively
high sequence similarity (83.85 %, Figure 5) and process the same
precursor molecule, CBGA (Figure 4). In vitro studies have shown
that the CBDA synthase produced CBDA and THCA at roughly a ratio of 20:1
(Zirpel et al.,
2018). This is similar to ratios observed in planta in high-CBD
hemp varieties as well
(Toth et al., 2020;
Weiblen et al.,
2015). This potentially results in the problem that, if CBDA production
is increased, THCA also increases as a by-product, even if plants do not
express a functional THCA synthase. Cannabis varieties with very
high CBD levels may thus be at risk of exceeding legal THC thresholds.
Understanding the exact genetics underlying the different chemotypes
will be important for future targeted breeding approaches. Tight
restrictions across the world make it difficult for farmers to grow
chemotype III, IV and V varieties, because the presence of residual THC
creates regulatory problems and uncertainties. Especially type III
plants often have THCA/THC levels slightly above the legal THC limit
(Aizpurua-Olaizola et
al., 2016; Toth et al., 2020). Hence, one important breeding goal is
going to be the generation of zero-THC lines which still produce high
levels of CBD in the range of 15 to 20 % of dry flower mass. Whether
this is possible to achieve is difficult to say, since even in the
absence of a THCA synthase, CBDA synthases produce THCA as a by-product
(Toth et al., 2020;
Zirpel et al., 2018). This will, therefore, require identification of a
CBDA synthase that does produce only very low or no amounts of THCA.In vitro experiments show that point mutations can alter the
amount of by-products
(Zirpel et al.,
2018). Natural variation in synthase genes exists and have been linked
to altered phytocannabinoid compositions
(Onofri et al.,
2015). Hence, naturally occurring or artificially generated CBDA
synthase varieties could be used for targeted breeding in this
direction.
In addition, Cannabis varieties used for fibre or seed production
could be selectively bred and genotyped to have 0 % overall
phytocannabinoids (chemotype V), as currently even the farming of these
kinds of varieties is heavily restricted in many countries.
Other phytocannabinoids like CBG(A) and CBC(A) as well as the manifold
variants of terpenes produced in Cannabis flowers are
increasingly coming into focus in the medical research fields (reviewed
in Booth and Bohlmann,
2019; Deiana, 2017; Pollastro et al., 2018), hence generating lines
with specific phytocannabinoid profiles might be of interest in further
research.
A hairy topic: Flower development and morphology inCannabis
The flower is the reproductive structure of flowering plants
(angiosperms), which represent one of the most successful and diverse
groups of organisms on this planet
(Krizek and Fletcher,
2005). While the characteristic shape of the Cannabis leaf is
often used as a symbol for the whole plant, Cannabis female
flowers are of particular interest because they are the main site of
production of pharmacologically active compounds (phytocannabinoids)
(Spitzer-Rimon et al.,
2019). Understanding the morphology of Cannabis flowers and
their developmental genetics is therefore especially important.
The typical angiosperm flower consists of four different organ types,
which are organized in concentric whorls: sepals, petals, stamens and
carpels (Endress,
1992; Krizek and Fletcher, 2005). Sepals are in the outermost whorl and
usually green and leaflike in appearance. Petals are in the second whorl
and often coloured to attract pollinators. Petals together with sepals
are termed the perianth and constitute the non-reproductive part of a
flower. Stamens are typically located in the third floral whorl. They
are the male reproductive organs and are composed of an anther and a
filament. The anthers grow on top of the stalk-like filaments and are
the site of pollen production. Finally, carpels develop in the fourth
and central whorl of a typical flower. Carpels are the reproductive
organs that contain an ovary inside which ovules develop. The tip of the
carpel, the stigma, receives the pollen. The style connects the stigma
to the ovary (Becker,
2020; Endress, 1992; Krizek and Fletcher, 2005).
Notably, the number, arrangement, and morphology of the floral organs
varies substantially between different species of flowering plants
(Endress, 2011;
Theissen and Melzer, 2007). Most flowers contain, as described above,
both carpels and stamens, and are therefore termed bisexual flowers
(Renner, 2014).
However some 15 % of flowering plant species are monoecious or
dioecious and have unisexual flowers that develop only stamens or
carpels (Renner,
2014). In dioecious plants, female and male flowers develop on separate
individuals. In contrast, in monoecious plants male and female flowers
develop on the same individual
(Renner, 2014).
Cannabis is primarily dioecious
(Moliterni et al.,
2004). The male Cannabis flower is green-yellow in appearance
and has a perianth of five sepals, while petals are completely absent.
Further, an individual male flower contains five free stamens, and no
female reproductive organs (Figure 6a and b)
(Leme et al., 2020;
Spitzer-Rimon et al., 2019).
On the other hand, the female flower is enclosed within a green leaflike
perigonal bract. The perigonal bract is sometimes also described as a
sepal, but morphological studies agree that it is a bract
(Leme et al., 2020;
Spitzer-Rimon et al., 2019). As such, it is not strictly a part of the
flower. Between the perigonal bract and the carpel is a membranous and
hyaline perianth which tightly embraces the ovary
(Leme et al., 2020;
Reed, 1914; Spitzer-Rimon et al., 2019). It is worth noting that this
inconspicuous perianth sometimes is not mentioned in the structure of
female Cannabis flowers or is considered missing as it is not
visible from the outside of the flower. Most likely, these
membranous structures are homologous to sepals
(Leme et al., 2020).
At the top of the ovary are two filamentous styles. The stigma is
brush-like and has epidermal cells elongated into hair-like projections
(Reed, 1914; Lemeet al., 2020) (Figure 6c and d).
The commercially interesting phytocannabinoids and terpenes are
predominantly produced on the perigonal bracts of female flowers, more
specifically in glandular trichomes that cover those bracts. Glandular
trichomes can be categorized into sessile, stalked and bulbous trichomes
(Hammond and Mahlberg,
1973), with bulbous trichomes being metabolically less active
(Livingston et al.,
2020). Cannabis plants also have non-glandular trichomes:
hair-like uni- or multicellular trichomes which protect them from biotic
and abiotic stresses
(Andre et al., 2016;
Dayanandan and Kaufman, 1976). However, glandular trichomes are the
main site of phytocannabinoid synthesis
(Furr and Mahlberg,
1981).
Because phytocannabinoids are cytotoxic in higher concentrations, they
have to be secreted and are not stored within cellular compartments.
Phytocannabinoids along with other secondary metabolites are secreted
from glandular trichomes with a globose head-like structure (Figure 7).
This head is formed by an enlarged secretory cavity which is surrounded
by a culticule that encapsulates the secreted secondary metabolites
(Hammond and Mahlberg,
1973). At the base of the head is a layer of secretory cells
(Kim and Mahlberg,
1991; Livingston et al., 2020). The head can be sessile, directly on
the epidermis and often be found on vegetative leaves (sessile
trichomes), or pre-stalked or stalked with the head being elevated above
the epidermis (pre-stalked and stalked trichomes), which are mainly
found on female inflorescences
(Kim and Mahlberg,
1991; Livingston et al., 2020). Additionally, these structures can be
distinguished by different levels of autofluorescence, cell numbers as
well as phytocannabinoid and terpene profiles
(Livingston et al.,
2020; Turner et al., 1978). Stalked trichomes seem to be developing
from pre-stalked trichomes and contain a terpene profile distinct from
true sessile trichomes
(Livingston et al.,
2020). Transcriptome analysis of floral trichomes of a CBD hemp
(‘Finola’) confirmed high expression levels of genes involved in the
synthesis of phytocannabinoids, terpenes and their respective precursor
molecules in glandular trichomes, with expression differences between
bulbous, sessile, and (pre-)stalked trichomes
(Livingston et al.,
2020).
It is not clear why predominantly female plants produce glandular
trichomes within their inflorescence structures. Illuminating the
genetic underpinnings of this sexual dimorphism remains a challenge for
further research. Glandular trichomes also develop on male flowers
(Leme et al., 2020),
albeit at lower density and probably with less phytocannabinoids.
Understanding which genetic factors restrict the development of
glandular trichomes largely to female inflorescences during flower
development would provide a valuable resource for an increase of
phytocannabinoid production.
The battle of the sexes: Sex determination in Cannabis